Insect Biochemistry and Molecular Biology 71 (2016) 49e57

Contents lists available at ScienceDirect

Insect Biochemistry and Molecular Biology journal homepage: www.elsevier.com/locate/ibmb

Potential detoxification of gossypol by UDP-glycosyltransferases in the two Heliothine moth species Helicoverpa armigera and Heliothis virescens Corinna Krempl a, Theresa Sporer a, Michael Reichelt b, Seung-Joon Ahn a, Hanna Heidel-Fischer a, Heiko Vogel a, David G. Heckel a, Nicole Joußen a, * a b

Department of Entomology, Max Planck Institute for Chemical Ecology, Hans-Knoell-Str. 8, 07745 Jena, Germany Department of Biochemistry, Max Planck Institute for Chemical Ecology, Hans-Knoell-Str. 8, 07745 Jena, Germany

a r t i c l e i n f o

a b s t r a c t

Article history: Received 4 December 2015 Received in revised form 8 February 2016 Accepted 8 February 2016 Available online 10 February 2016

The cotton bollworm Helicoverpa armigera and the tobacco budworm Heliothis virescens are closely related generalist insect herbivores and serious pest species on a number of economically important crop plants including cotton. Even though cotton is well defended by its major defensive compound gossypol, a toxic sesquiterpene dimer, larvae of both species are capable of developing on cotton plants. In spite of severe damage larvae cause on cotton plants, little is known about gossypol detoxification mechanisms in cotton-feeding insects. Here, we detected three monoglycosylated and up to five diglycosylated gossypol isomers in the feces of H. armigera and H. virescens larvae fed on gossypol-supplemented diet. Candidate UDP-glycosyltransferase (UGT) genes of H. armigera were selected by microarray studies and in silico analyses and were functionally expressed in insect cells. In enzymatic assays, we show that UGT41B3 and UGT40D1 are capable of glycosylating gossypol mainly to the diglycosylated gossypol isomer 5 that is characteristic for H. armigera and is absent in H. virescens feces. In conclusion, our results demonstrate that gossypol is partially metabolized by UGTs via glycosylation, which might be a crucial step in gossypol detoxification in generalist herbivores utilizing cotton as host plant. © 2016 Elsevier Ltd. All rights reserved.

Keywords: Host plant adaptation Helicoverpa armigera Heliothis virescens Gossypol detoxification UDP-Glycosyltransferase

1. Introduction Generalist insect herbivores can cope with a large diversity of toxic secondary metabolites produced by their host plants by developing different strategies like behavioral avoidance, rapid excretion, target site mutation, or sequestration (Heckel, 2014; Heidel-Fischer and Vogel, 2015). They also utilize diverse mechanisms of metabolic detoxification to circumvent the toxicity of plant secondary metabolites. In many cases metabolic detoxification can be divided into three phases. In phase I, enzymes such as cytochrome P450 monooxygenases (P450s) or carboxylesterases act directly on the toxin molecule, introducing or releasing functional

Abbreviations: DMSO, dimethyl sulfoxide; P450, cytochrome P450 monooxygenase; UGT, UDP-glycosyltransferase. * Corresponding author. E-mail addresses: [email protected] (C. Krempl), [email protected] (T. Sporer), [email protected] (M. Reichelt), [email protected] (S.-J. Ahn), hfi[email protected] (H. Heidel-Fischer), [email protected] (H. Vogel), heckel@ ice.mpg.de (D.G. Heckel), [email protected] (N. Joußen). http://dx.doi.org/10.1016/j.ibmb.2016.02.005 0965-1748/© 2016 Elsevier Ltd. All rights reserved.

groups and thus increasing the reactivity and the hydrophilicity of the toxin. In phase II, enzymes like glutathione S-transferases, UDPglycosyltransferases, and methyl-, acetyl-, phospho- and sulfotransferases conjugate endogenous molecules to the toxins (Robertson et al., 1999; Wilkinson, 1986). The resulting conjugates are less reactive and more water-soluble than the original toxins and thus lose the ability to diffuse through membranes. Phase III enzymes such as ATP-binding cassette transporters facilitate the active transport of toxins across membranes, thus enabling their excretion. However, the detoxification process is much more flexible than the classification into the three phases implies (Rowland et al., 2013). Glycosylation of toxins is a particularly important detoxification mechanism, in which a lipophilic aglycone is converted into a more hydrophilic and readily excretable compound (Wilkinson, 1986). The underlying mechanism is a second order nucleophilic substitution (Radominska-Pandya et al., 2010) catalyzed by glycosyltransferases, which are found in animals, plants, fungi, bacteria, and viruses (Bock, 2015; Paquette et al., 2003). Insects use UDP-glucose

50

C. Krempl et al. / Insect Biochemistry and Molecular Biology 71 (2016) 49e57

as an activated sugar donor (Ahn et al., 2012) that is transferred by UDP-glycosyltransferases (UGTs), which are anchored in the endoplasmic reticulum (Fig. 1). While UGTs also fulfill important endogenous functions in insect olfaction (Bozzolan et al., 2014; Robertson et al., 1999), pigmentation (Hopkins and Kramer, 1992), UV-shielding (Daimon et al., 2010), and cuticular tanning (Kramer and Hopkins, 1987), several studies describe the involvement of UGTs in detoxification of plant secondary metabolites in insects (Ahmad and Hopkins, 1992; Ahn et al., 2011b; Kojima et al., 2010; Wouters et al., 2014). However, compared to other detoxification enzymes such as P450s (Feyereisen, 2012) knowledge about UGTs involved in detoxification of plant toxins in insects is rare (HeidelFischer and Vogel, 2015). Helicoverpa armigera (Hübner), the cotton bollworm, and Heliothis virescens (Fabricius), the tobacco budworm, (Lepidoptera: Noctuidae) are important agricultural pests on many different crop plants. A favorite host plant of both noctuid moth species is cotton (Gossypium spp.), which contains a number of secondary metabolites to defend itself against herbivores and pathogens. A major defensive compound of cotton is gossypol, a yellow colored sesquiterpene dimer stored in subepidermal glands that is toxic to many insect species. The toxicity of gossypol can be mostly attributed to its highly reactive aldehyde groups that interact with amino acids of proteins and to the presence of six phenolic hydroxyl groups (Dodou, 2005). The rather hydrophobic nature of gossypol allows the molecule to diffuse across membranes (Laughton et al., 1989) and thus to harm the organism. While generally very little is known about the detoxification of gossypol in lepidopteran larvae that feed successfully on cotton plants, a particular P450 enzyme, CYP6AE14, has been associated with gossypol detoxification, due to its induction in H. armigera larvae after gossypol ingestion (Celorio-Mancera et al., 2011; Mao et al., 2007). However, evidence for a direct involvement of this particular P450 enzyme in gossypol metabolism is still missing.

Rojas et al. (1992) reported gossypol glycosides in feces of H. virescens after ingestion of a gossypol diet, but the identity of the conjugates and the enzymes involved was not clearly described. In this study we detected and characterized several isomers of gossypol glycosides in the feces from H. armigera and H. virescens that fed on gossypol-supplemented diet. We found species-specific differences in the pattern of glycosylated gossypol isomers. Heterologous expression of H. armigera UGTs revealed two UGTs that showed specific gossypol glycosylation activity. With these results we shed some light upon the mechanism that enables two Heliothine pest species to utilize cotton as a host plant without being poisoned by gossypol.

2. Materials and methods 2.1. Insects and bioassay H. armigera larvae were collected from Toowoomba, Queensland, Australia, in 2003 (TWB strain). H. virescens larvae were provided by North Carolina State University and had been originally collected in Clayton, North Carolina, USA, in 1988 (JEN strain). H. armigera larvae were reared on Bio-Serv diet (General Purpose Lepidoptera) and H. virescens larvae on pinto bean diet (Joyner and Gould, 1985) under laboratory conditions (26
C, 55% relative humidity, 16:8 h light:dark photoperiod) in Jena, Germany. Pinto bean diet was used for feeding assays with both species. Two gossypol-supplemented diets were prepared with a racemic mixture of gossypol isolated from cotton seeds (TimTec) as described previously by Stipanovic et al. (2006) with slight modifications. Gossypol (800 mg or 1600 mg) was dissolved in ethyl acetate (35 mL), added to alphacel (15 g; MP Biomedicals), a nonnutritive cellulose bulk, and dried at room temperature for 24 h. Hexane (50 mL) was added and evaporated for 36 h. Freshly prepared pinto bean diet (490 g) was cooled down to about 47
C and

Fig. 1. Schematic drawing of the general structure of a membrane bound animal UDP-glycosyltransferase (UGT). After cleavage of the N-terminal signal peptide that directs the protein to the ER, only a short cytoplasmic tail remains in the cytosol. A short transmembrane domain anchors the protein in the ER membrane flanked by a highly conserved, negatively charged residue [glutamic acid (E) or aspartic acid (D)] at the intersection to the globular part of the enzyme located in the ER lumen. This part consists of two major domains, the C-terminal sugar-donor binding domain and the N-terminal substrate-binding domain (Magdalou et al., 2010). Both domains are connected via an interdomain linker. The signature motif (blue box) that is located in the C-terminal domain is conserved across all organisms (Ahn et al., 2012). It overlaps with the sugar-donor binding region 1 (DBR1, orange box), which acts in concert with the sugar-donor binding region 2 (DBR2, orange box) and several (a) nucleotide-, (b) phosphate- and (c) glucoside-interacting residues. All putative b-sheets in the C-terminal domain are indicated in red font color. In the N-terminal domain, two catalytic residues [histidine (H) and aspartic acid (D); red arrows] are located which interact with the substrate.

C. Krempl et al. / Insect Biochemistry and Molecular Biology 71 (2016) 49e57

mixed with the gossypol-alphacel mixture to final concentrations of 0.16% (w/w) or 0.32% (w/w) gossypol, respectively. Newly molted 5th instar larvae of H. armigera and H. virescens were individually reared in 24-well plates for 2 days on 1 g fresh weight of pinto bean diet supplemented with different gossypol concentrations (0.16% and 0.32%) and of control diet lacking gossypol, respectively. Forty larvae were used in each treatment. From larvae that consumed the complete 1 g of diet in 2 days, three larvae or the feces produced by these three larvae were pooled for one biological replicate. The feces and larvae were freeze-dried for 3 days, homogenized, and weighed. As larvae feeding on diet with 0.32% gossypol failed to consume the entire diet within 2 days, they were excluded from the quantification analysis of gossypol. 2.2. Quantification of free gossypol in larvae and feces and detection of gossypol metabolites For the analysis of feces ten biological replicates and for analysis of larvae four biological replicates were used for both species. The dried samples were mixed with acetonitrile:water [50:50 (v:v)] followed by a centrifugation step at 3000  g for 5 min at room temperature. This extraction procedure was repeated three times. Supernatants were pooled into glass vials and analyzed with an Agilent HPLC 1100 Series on a Nucleodur Sphinx RP HPLC column (250  4.6 mm, 5 mm particle size; MachereyeNagel). For a detailed description of HPLC conditions see Supplementary Material S1.1. Gossypol was quantified based on an external standard curve of an authentic standard of gossypol (racemic mixture of gossypol, TimTec; 0.0625, 0.125, 0.25, 0.5, 1.0, and 2.0 total mg per injected volume) dissolved in acetonitrile:water (50:50). Samples were further analyzed by liquid chromatography coupled with mass spectrometry (LC-MS; Esquire 6000 ESI-iontrap, Bruker Daltonics). If necessary, samples were analyzed by LC-MS/ MS (Agilent 1200 HPLC system coupled with an API 5000 tandem mass spectrometer, Applied Biosystems) equipped with a turbospray ion source in negative ionization mode. For a detailed description of LC-MS/MS conditions see Supplementary Material S1.1. For relative quantification, peak areas of the respective extracted ion traces for different compounds were extracted as follows: gossypol, m/z 517; gossypol monoglycosides, m/z 679; gossypol diglycosides, m/z 841. Accurate masses of gossypol metabolites were determined by high resolution mass spectrometry using a LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific). In order to obtain enough gossypol glycosides for an accurate characterization, 20 samples, each consisting of 1 g of freeze-dried feces from 5th instar H. armigera larvae which fed on a 0.32% gossypol diet for 3 days were extracted five times with 2 mL acetonitrile:water (50:50), as described above. Pooled extracts were purified using a C18 SPE cartridge (Chromabond, 5000 mg; MachereyeNagel) and were eluted with water, followed by 30% of methanol, 60% of methanol, and 100% of methanol. The collected fractions were analyzed using LC-MS, and MS4 data of the glycosylated gossypol metabolites were obtained from the 100% methanol fraction. 2.3. Total gossypol quantification in larvae and feces using aniline method For determination of both, bound and free gossypol, the aniline method described by Rojas et al. (1992) was used with slight modifications. Using an excess of aniline, free gossypol and gossypol that is covalently bound to proteins can react via a Schiff base formation with aniline forming dianilinogossypol. Eight biological replicates of feces samples and four biological replicates of

51

larvae fed on gossypol diet were used for each species. Furthermore, two biological replicates of larvae and feces of larvae fed on control diet were extracted. Additionally, the conversion of gossypol to dianilinogossypol out of diet containing 0.16% gossypol was tested. The samples were mixed with solution A, consisting of 125 mL of 70% ethanol, 250 mL of ether and 0.25 mL of glacial acetic acid. Aniline was added in 2.4-fold excess according to the maximal expected gossypol amount of the material extracted. Therefore, 133 mL of aniline (1.426 mmol) were added to both feces and larval material from H. armigera and 122 mL of aniline (1.314 mmol) were added to feces and larval material of H. virescens, respectively. An excess of 74.5 mL of aniline (0.800 mmol) was used for the diet sample containing 0.16% gossypol. Samples were mixed and incubated at 90
C for 40 min. Afterwards, 2 mL of glass beads (ColiRollers, Merck Millipore) and 2 mL of hexane were added. Samples were mixed for 30 min and centrifuged at 2000  g for 5 min at room temperature. The supernatant was diluted in methanol for HPLC measurements. For preparation of a dianilinogossypol standard, 0.5 mg of gossypol in solution A (1 mL) was incubated with 215.5 mL of aniline at 90
C for 40 min. The dianilinogossypol standard and the samples were analyzed by HPLC (Agilent HPLC 1100 Series) and separated on a Nucleodur Sphinx RP column (250  4.6 mm, 5 mm particle size). For detailed HPLC conditions see Supplementary Material S1.2. LC-UV chromatograms of the standard and the feces and tissue samples showed three peaks with similar UV spectra. The most dominant peak was dianilinogossypol, the second peak was monoanilinogossypol, and the third one was an unidentified anilinogossypol derivative. All three compounds were separately quantified using calibration curves (0.11, 0.42, 1.47, 2.52, 3.36, and 3.92 total mg per injected volume). 2.4. Bioassay and sample preparation for microarray analysis Newly molted 5th instar larvae of H. armigera and H. virescens were reared on artificial control diet and diet supplemented with gossypol, using following concentrations: 0.16% gossypol for H. armigera and 0.32% gossypol for H. virescens. After 3 days, larvae were anesthetized on ice and dissected into gut and the rest of the body. Tissues from 5 to 8 larvae were pooled as one replicate, and four biological replicates were prepared and stored at 80
C until RNA isolation. Tissues were ground to fine powder under liquid nitrogen. Total RNA was extracted by using the innuPREP RNA Mini Kit (Analytik Jena). RNA quality and integrity were assessed using RNA Nano chips on an Agilent 2100 Bioanalyzer (Agilent Technologies). RNA quantity was determined using the NanoDrop ND-1000 spectrophotometer (Peqlab). 2.5. Microarray design, labeling, hybridization, and data acquisition The H. armigera microarray design was previously described by Celorio-Mancera et al. (2011) and Kuwar et al. (2015). The final condensed Agilent H. armigera 4  44 K array design based on the eArray platform (Agilent Technologies, https://earray.chem.agilent. com/earray/) contained 43,803 non-control probes and 1417 Agilent Technologies built-in controls (structural and spike-in). H. virescens sequences obtained from Sanger sequencing of clonal cDNA libraries as well as from Illumina RNAseq of different larval tissues (e.g. midgut, fat body, integument) for all larval instars and developmental stages (larvae, pupae, adults) were jointly assembled, resulting in a total of 103,904 contigs. For each of these contigs either a single (in case the 50 -3orientation was known) or two 60 mer oligo probes (one each for the plus and minus strand, respectively) were designed using eArray tools (Agilent Technologies). The microarray was designed using the 4  180 K format with a final number of 167,728 non-control probes and 1417 Agilent

52

C. Krempl et al. / Insect Biochemistry and Molecular Biology 71 (2016) 49e57

Technologies built in controls (structural and spike in). For microarray hybridization, RNA from the gut and the rest of the body of H. armigera larvae fed on 0.16% gossypol and control diet, respectively, as well as of H. virescens larvae fed on 0.32% gossypol and control diet, respectively, (see 2.4) was prepared following the one-color microarray-based gene expression analysis protocol (Low Input Quick Amp Labeling, Agilent Technologies). For each replicate, 100 ng of total RNA was spiked using the Agilent One-Color Spike Mix for Cyanine 3-labeling and labeled with cyanine 3-CTP. For purification of the labeled and amplified RNA the RNeasy Mini Kit (Qiagen) was used. cRNA quantity was determined on a NanoDrop ND-1000 spectrophotometer using the microarray function. Labelling efficiency was estimated and only samples with a specific activity higher than 6 pmol Cy3 per mg cRNA was used for microarray hybridization. The Cy3 labeled cRNA was hybridized to H. armigera and to H. virescens microarray slides, respectively, for 17 h at 65
C in a hybridization oven. For each treatment and tissue four biological replicates were used. The microarray slides were washed and analyzed as described in Kuwar et al. (2015). Data was extracted from TIFF images with Agilent Feature Extraction software version 9.1. Feature extracted data were analyzed using GeneSpring GX version 12.5 (Agilent Technologies) software. H. armigera (control vs 0.16% gossypol) and H. virescens (control vs 0.32% gossypol) expression profiling was generated by normalizing fluorescence signals to quantile (to the median intensity and log base 2-transformation of the normalized data). Oneway ANOVA was performed for the pairwise comparisons between the treatments. Only genes with p < 0.05 after BenjaminiHochberg statistical correction (B&H FDR) were considered statistically differentially expressed (fold change  2.0, fold change cutoff: 2.0. pairing option, all against single condition, post hoc tukey). The microarray data reported in this paper have been deposited in the Gene Expression Omnibus (GEO) database (www.ncbi.nlm. nih.gov/geo, accession no.: GSE77619 and GSE77620). GSE77619 and GSE77620 series records can be directly viewed at http://www. ncbi.nlm.nih.gov/geo/query/acc.cgi?acc¼GSE77619 and http:// www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc¼GSE77620, respectively. 2.6. Selection of candidate UGTs from microarray data Candidate UDP-glycosyltransferase (UGT) genes significantly upregulated in the gut and the rest of the body of H. armigera and H. virescens larvae were extracted from the microarray data. Based on the microarray probe sequence, individual UGTs were identified by blasting them to NCBI GenBank and the in-house H. armigera database. As little information is publicly available for the H. virescens genome, H. virescens contigs derived from the microarray data set were individually blasted against NCBI GenBank to identify putative H. armigera orthologues. In addition, another UGT data set from a previous microarray study from our group of gossypol-fed H. armigera larvae (Celorio-Mancera et al., 2011) was included in the gene analysis. UGTs found to be upregulated in this study were added to the list of candidates. 2.7. Cloning of H. armigera candidate UGTs, multiple alignment and protein structure prediction of H. armigera candidate UGTs cDNA was synthesized from RNA extracted from the H. armigera larval guts (0.16% gossypol diet) using the PrimeScript First Strand cDNA Synthesis Kit (Clontech). Genes of interest (UGT40D1, UGT40D2, UGT40Q1, UGT41B1, UGT41B2, UGT42B2, and UGT33B4) were amplified from the cDNA by PCR using the high fidelity Phusion polymerase (Thermo Fisher Scientific) and gene-specific primers with forward primers including a 50 Kozak sequence and

reverse primers either containing or lacking the stop codon for V5 epitope and His-tag fusion expression (Table S1). PCR products were separated via gel electrophoresis, excised from the gel and purified. PCR products were ligated into pIB/V5-His TOPO TA expression vector (Thermo Fisher Scientific). Correct orientation of the insert was checked by colony PCR after transformation into TOP10 Escherichia coli cells (Thermo Fisher Scientific). Plasmids were isolated from positive clones using the Gene Jet Plasmid Miniprep Kit (Thermo Fisher Scientific) and further analyzed by Sanger sequencing. Sequences were aligned to their respective annotated gene and cDNA full-length sequences were deduced. Full length sequences of new variants were uploaded on GenBank (accession numbers: KU214506, KU214507, KU214508, KU214509, KU214510). For protein structure prediction, a multiple alignment with the amino acid sequences of UGT40D1, UGT40D2, UGT40Q1, UGT41B1, UGT41B2, UGT42B2, and UGT33B4 was performed using the ClustalW multiple alignment function of the BioEdit software. Signal peptides and cleavage sites were predicted using SignalP 4.1 (http://www.cbs.dtu.dk/services/SignalP). Further protein features were predicted by comparison with these of UGT40Q1 and UGT41B1 in the alignment created by Ahn et al. (2012). 2.8. Expression of candidate UGTs in insect cells and crude microsomal extraction Candidate UGT genes were transiently expressed in Spodoptera frugiperda cells (Sf9) (Thermo Fisher Scientific) using the transfection reagent FuGENE HD (Promega). After 72 h, cells were harvested and centrifuged at 500  g for 5 min at 4
C. The cells were washed twice with PBS (pH 7.4) and centrifuged at 500  g for 10 min at 4
C. The cell pellet was resuspended in hypotonic buffer [20 mM Tris, 5 mM EDTA, 1 mM DTT, 20% (v:v) glycerol at pH 7.5, 0.1% (v:v) of Benzonase nuclease (Merck Millipore) and 1 x Protease Inhibitor Cocktail Set III (Merck)] and incubated for 20 min on ice. The cells were homogenized using the Potter-Elvehjem tissue grinder (Wheaton) and subsequently mixed with an equal volume of sucrose buffer (20 mM Tris, 5 mM EDTA, 1 mM DTT, 500 mM sucrose, and 20% (v:v) of glycerol at pH 7.5). After centrifugation at 1200  g for 10 min at 4
C, the supernatant was centrifuged at 100,000  g for 1 h at 4
C. The resulting supernatant containing the cytosolic fraction was aliquoted. The cell pellet containing microsomes, plasma membrane and cell organelles was resuspended in sodium phosphate buffer (0.1 M; pH 6.4). Samples were snap frozen in liquid nitrogen and stored at 80
C. 2.9. Western blot and estimation of the amounts of heterologously expressed UGT proteins The total protein amount in crude microsomal extracts was estimated using the Quick Start Bradford Protein Assay (Bio-Rad). Successful expression of UGT proteins was verified by western blot using the V5-HRP antibody (Life Technologies). In order to estimate the relative amounts of the heterologously expressed UGTs, western blots with four increasing amounts of total protein for each UGT extract were performed. Using the Fiji software, the intensities of the bands on scanned x-ray films were determined and compared to each other to calculate the total protein amount that has to be used in enzymatic assays in order to achieve equal amounts of heterologous UGTs (Fig. S1 and Table S2). 2.10. Enzyme assays with a general substrate and gossypol In order to test the glycosylation activity of the heterologously expressed UGTs, 1-naphthol (Sigma Aldrich) was used as general substrate. Twenty-five mg of total protein of the crude microsomal

C. Krempl et al. / Insect Biochemistry and Molecular Biology 71 (2016) 49e57

A

53

B

40

Relative intensity

35

841.3 679.0

a

30

3 25 20

1 OH

OH 3ꞌ

2 HO OH 2ꞌ 3 HO

OH 1ꞌ O

c b

5

15

O

517.0 Relative intensity of m/z

45

Relative intensity of m/z

x10

4

Relative intensity of m/z

50

2 4

1

10

7 6

8

5 0 7

9

11

13 15 Retention time [min]

17

19

21

Fig. 2. Detection of glycosylated gossypol isomers in the feces of H. armigera and gossypol structure. (A) Extracted LC-MS ion chromatogram (m/z 517, m/z 679, and m/z 823) of the larval feces extract measured in negative ion mode and representative mass spectra of (a) diglycosylated gossypol isomers of m/z 841, (b) monoglycosylated gossypol isomers of m/z 679, and (c) gossypol of m/z 517. Isomer peaks are numbered from 1 to 8, starting with the compound possessing the shortest retention time. The relative intensities of the different isomer peaks do not reflect quantitative ratios between them. (B) Gossypol structure and potential position of the sugar molecules in the diglycosylated gossypol isomers. Hydroxyl groups are numbered from 1 to 3 and 1' to 3', respectively. Each of the six hydroxyl groups is a potential glycosylation site. Pairs of numerals represent putative diglycosylated gossypol isomers: 1/2, 1/3, 2/3, 1/1', 1/2', 1/3', 2/2', 2/3', and 3/3'.

extract from the reference enzyme UGT40Q1 (factor 1.0) were used. The total protein amounts for the other UGT extracts were adjusted according to their calculated factors determined by western blot analysis (Table S2). For each of the UGT extracts a control consisting of a non-transfected crude microsomal extract with the same total protein amount was prepared. UDP-glucose (110 mg in water), 1naphthol (500 ng in DMSO), MnCl2 (20 mM) and sodiumphosphate buffer (0.1 M; pH 6.4) were added to the proteins resulting in a total volume of 25 mL and incubated in glass microinserts for 30 min at 30
C. The reaction was stopped with 25 mL of methanol followed by a centrifugation step at 3000  g for 3 min at room temperature. For enzyme assays with gossypol as substrate, 50 ng in DMSO was added to the reaction mix as described above and incubated for 1 h at 30
C. Factors used for equalizing the amounts of the different heterologous UGTs are listed in Table S2. The total protein amount used per assay was adjusted to 25 mg of total protein by adding the respective amount of non-transfected Sf9 microsomes. Supernatants of the samples from both the 1-naphthol and the gossypol enzyme assay were analyzed via LC-MS/MS (API 5000 tandem mass spectrometer equipped with a turbospray ion source in negative ionization mode using multiple reaction monitoring (MRM); for detailed conditions see Table S3 and Table S4). All assays were performed in duplicates. 3. Results 3.1. Detection of unmodified and glycosylated gossypol in feces of H. armigera In order to identify potential gossypol metabolites, feces extracts of H. armigera larvae, which fed on gossypol-supplemented diet (0.32%), and of control larvae, whose diet lacked gossypol, were analyzed by LC-MS and screened for differences. Several compounds were identified that were present only in feces extracts of gossypol-fed larvae (Fig. 2A). According to the mass spectra, compounds 1 to 5 with retention times of 8.7, 9.6, 10.1, 10.6, and

11.3 min respectively, shared the same molecular ion of m/z 841 in negative ionization mode. The accurate mass was estimated by high resolution mass spectrometry as m/z 841.2930 [M-H]- resulting in the chemical formula of C42H49O18 with the theoretical mass of 841.2913. Compounds 6 to 8 with retention times of 14.2, 14.5, and 15.6 min, respectively, shared the same molecular ion of m/z 679 in negative ionization mode with the accurate mass of m/z 679.2397 [M-H]- leading to the chemical formula of C36H39O13 with the theoretical mass of 679.2385. Also, unmodified gossypol was detected in the feces extract with a retention time of 20.5 min possessing the molecular ion of m/z 517 in negative ionization mode. Fragmentation patterns of the compounds possessing the molecular ions of m/z 841 and m/z 679, respectively, were analyzed based on a MS4 data set derived from LC-MS measurement (Fig. S2 and Table S5), to gain more information about the chemical structures. According to the data of the MS2, MS3, and MS4 spectra, compounds 1 to 5 were identified as diglycosylated gossypol (m/z 841) and the compounds 6 to 8 as monoglycosylated gossypol (m/z 679). The hexose moieties that are bound to the hydroxyl groups of the gossypol molecule are most likely glucose. As one molecule of gossypol possesses six hydroxyl groups, there are several positions possible for the binding of the hexose moiety (Fig. 2B). Therefore, we assume that the compounds sharing the same mass but possessing different retention times represent structural isomers of mono- or diglycosylated gossypol differing only in the position of the hexose moiety or moieties bound to the hydroxyl groups of the gossypol molecule. 3.2. Species specific differences in glycosylated gossypol isomers in feces of H. armigera and H. virescens In order to test whether other Heliothine moth species produce glycosylated gossypol metabolites, we also analyzed the feces of gossypol-fed H. virescens larvae. We found that gossypol glycosides are also present in H. virescens feces. Interestingly, the comparison of the relative amounts of glycosylated gossypol isomers of feces extract of both species revealed differences in the isomer pattern

54

C. Krempl et al. / Insect Biochemistry and Molecular Biology 71 (2016) 49e57

(Fig. 3). The feces extracts of H. armigera larvae contained 5 isomers sharing the molecular ion of m/z 841 (diglycosylated gossypol), whereas in H. virescens only 4 of these isomers were detected. The feces extract of H. armigera contained higher amounts of the diglycosylated gossypol isomer 4, compared to H. virescens feces. The diglycosylated gossypol isomer 5 was abundant in H. armigera feces but very rare in the feces of H. virescens. H. virescens larvae predominantly produced the isomers 1 and 3, of which isomer 1 was less abundant in H. armigera feces. Isomers 2 and 3 were detected in feces of both species in similar relative amounts. Less monoglycosylated gossypol was excreted by both H. armigera and H. virescens compared to the amount of diglycosylated gossypol metabolites. Glycosylated gossypol isomers 3 to 5 can also be detected in relatively low amounts in the larval bodies of both species. The feces and larval extracts of both species were analyzed with a more sensitive LC-MS/MS (API 5000) for further detailed analysis. Relative areas of integrated mass peaks in negative ionization mode of extracted ion chromatograms of multiple reaction monitoring (m/z 679.000/517.000 and m/z 841.000/679.000) are shown in Fig. S3. 3.3. Recovery of bound vs. free gossypol Besides the glycosylated gossypol metabolites, unmodified gossypol was also detected in acetonitrile:water (50:50) extracts of feces and larvae of both species and was quantified using an UV detector coupled to the HPLC system. The recovery of ingested gossypol (1.59 mg gossypol/g fresh weight of diet) was on average 3.8% and 4.2% from the feces of one larva and 0.1% from one larval body of H. armigera and H. virescens, respectively. This indicates that free gossypol was preferentially excreted rather than kept in the larval body, as we found 39.9-fold more gossypol in feces than in larvae. The low recovery of gossypol using acetonitrile:water (50:50) as the extraction solvent indicated that a high portion of gossypol was covalently bound to proteins that is not extractable by the solvent mixture used. Therefore, the aniline extraction method described previously by Rojas et al. (1992) and Smith (1967) was used. With this method both free gossypol and gossypol covalently bound to proteins can be extracted. Feces and larvae were heated with an excess of aniline that led to a relieving of gossypol from proteins by

the binding to aniline. Using HPLC-UV detection of dianilinogossypol and intermediates, we recovered on average 45.9% and 53.8% of the ingested gossypol from feces and 2.0% and 2.3% from one larva of H. armigera and H. virescens, respectively. On average 52.2% and 44.0% of the ingested gossypol was not extracted from H. armigera and H. virescens, respectively. From diet containing 0.16% gossypol, 65.6% of gossypol was recovered. Therefore, most of the gossypol in larvae and feces is bound rather than free. 3.4. Identification of candidate genes induced by gossypol feeding Microarray data of gossypol-fed H. armigera and H. virescens larvae were screened for potential candidate genes like UGTs that were significantly upregulated after gossypol feeding. Extracted UGT sequences (contigs) from H. virescens were blasted and aligned to H. armigera UGTs to detect putative orthologues (Table S6). We found 11 UGT genes upregulated in the larval gut and the rest of the body of H. armigera and 13 UGTs upregulated in H. virescens. Eight UGT orthologues were upregulated in both species and combined with UGTs found to be up regulated in a previous microarray study of H. armigera larvae (Celorio-Mancera et al., 2011) (Fig. 4). Due to their upregulation in all three microarray studies, 6 UGTs were chosen as candidate genes: UGT40D1, UGT40D2, UGT40Q1, UGT41B1, UGT41B2, and UGT42B2. Two genes were only upregulated in H. armigera larvae: UGT41D1 about four times and UGT33B4 about 21 times higher than in the control. Due to its high upregulation, UGT33B4 was chosen as an additional candidate gene. In total, seven H. armigera UGTs, UGT40D1, UGT40D2, UGT40Q1, UGT41B1, UGT41B2, UGT42B2, and UGT33B4, were further analyzed. Comparison of sequence information of the cloned candidate UGTs to published H. armigera UGTs and further unpublished sequences revealed high single nucleotide polymorphisms (Table S7). The sequence obtained from primers originally designed for UGT41B2 turned out to be more similar to UGT41B3. As UGT41B2 could not be amplified from cDNA, UGT41B2 was replaced by UGT41B3 in the following experiments. A multiple alignment of the UGT amino acid sequences and in silico analyses provided further structure information (Fig. S4). H. armigera with 0.16% gossypol (present work)

11

13 UGT40L1

x 104

35

UGT40R1

UGT44A2

UGT43A1

UGT40F2

30 H. armigera feces

25

Relative intensity

H. virescens with 0.32% gossypol (present work)

H. virescens feces

H. virescens larvae

UGT48A1

UGT40D2

UGT50A2

UGT40Q1 UGT41B1

H. armigera larvae

20

UGT40D1

UGT33B4

UGT41B2

UGT41D1

UGT42B2

UGT46A4

15 UGT33B2 UGT33B12

10

UGT33T1 UGT46A3

5

13

8 Is om er

7 m Is o

er om Is

er

6

5 er Is om

er om Is

Is o

m

er

4

3

2 m er Is o

Is o

m

er

1

0

Fig. 3. Species specific differences in the pattern of glycosylated gossypol isomers in the feces and the larval tissues of H. armigera and H. virescens. Depicted are relative intensities (peak areas) of extracted ion chromatograms of the diglycosylated gossypol isomers 1 to 5 and the monoglycosylated gossypol isomers 6 to 8 from feces and larval extracts of H. armigera and H. virescens. Standard deviation is presented by error bars.

H. armigera with 0.16% gossypol (Celorio-Mancera et al., 2011)

Fig. 4. Venn diagram of up-regulated UGT genes in H. armigera and H. virescens based on microarray analysis. The UGTs were found to be up-regulated in the gut and in the rest of the body of larvae after ingestion of 0.16% gossypol diet in H. armigera and of 0.32% gossypol diet in H. virescens, respectively (fold change > 2). Data is derived from three independent microarray studies. UGTs selected as candidates for heterologous expression are marked in white font color. Detailed information about the microarray data including fold change values can be found in Table S6.

C. Krempl et al. / Insect Biochemistry and Molecular Biology 71 (2016) 49e57

3.5. The heterologous UGT41B3 and UGT40D1 are capable of glycosylating gossypol Seven candidates for gossypol glycosylation, UGT40D1, UGT40D2, UGT40Q1, UGT41B1, UGT41B3, UGT42B2, and UGT33B4, were transiently expressed in Sf9 cells with a carboxyl-terminal tag allowing their detection by western blot. We were able to successfully express six out of the seven candidate genes; UGT42B2 could not be expressed in Sf9 cells. This candidate was therefore excluded from further studies. As analysis of the signals from the western blots by the Fiji software revealed that the UGTs showed different expression levels (Fig. S5), factors were calculated to equalize the amounts of heterologous UGTs for subsequent enzymatic assays (Table S2). In order to test the functional expression of the heterologous UGTs, 1-naphthol was used as a general substrate. Using UDPglucose as a donor, all expressed UGTs showed 1-naphthol glycosylation activity resulting in the formation of 1-naphthyl glucoside (MRM) [m/z 305 / 143]- confirming that they were enzymatically active. All crude microsomal extracts containing one of the six heterologous UGTs produced about two times more 1-naphthol glucoside than the respective control microsomes (Fig. S6). Using gossypol as substrate, out of the six successfully heterologously expressed candidate UGTs, only UGT41B3 and to a minor extent UGT40D1 showed specific gossypol glycosylation activity by exclusively forming diglycosylated gossypol isomer 5 (Fig. 5). Due to the formation of all other di- and monoglycosylated isomers by endogenous Sf9 UGTs, which were detected in control samples, it is unclear whether the heterologous UGTs are also capable of forming these isomers themselves. 4. Discussion H. armigera and H. virescens are important polyphagous pest species that feed on a broad range of host plants and thus are exposed to a variety of plant secondary metabolites. Cotton is a

55

favorite host plant of both species and produces the sesquiterpene dimer gossypol, which is toxic to many organisms. Knowledge about the metabolic fate of gossypol in insects as well as the identification of candidate detoxification enzymes is still lacking. Here, we detected and characterized novel glycosylated gossypol metabolites in the feces of H. armigera and H. virescens that fed on gossypol-supplemented diet. To our knowledge, this is the first time that gossypol mono- and diglycosides have been identified in insects. Although Rojas et al. (1992) reported that gossypol was conjugated to six glucose moieties in feces of H. virescens after ingestion of gossypol diet, we could only detect the mono- and diglycosylated gossypol isomers that we have described. To our knowledge no other gossypol metabolites formed by insects are known so far. In mammals, however, gossypol metabolites were detected in which glucuronic acid is conjugated to gossypol or its metabolites. Three isomers of gossypol monoglucuronide were detected in bile duct cannulated rats, which was proposed by Liu et al. (2014) to be the main mechanism of gossypol clearance. Abou-Donia (1975) identified gossypol and various gossypol metabolites such as gossypolone, gossypolonic acid, demethylated gossic acid, and putative apogossypol that might be conjugated to glucuronic acid or sulfate from pig liver. There are other systems where glycosylation of toxins is considered to be a detoxification strategy. For example Wouters et al. (2014), revealed a novel detoxification mechanism in larvae of the fall armyworm, S. frugiperda, when feeding on maize. Maize contains the benzoxazinoid (2R)-DIMBOA-glucoside, which is converted to the toxic aglucone DIMBOA by plant glucosidases during herbivore feeding. S. frugiperda deactivates the plant defensive compound by stereoselectively reglucosylating DIMBOA to (2S)-DIMBOA-glucoside that can no longer be hydrolyzed by plant glucosidases. Larvae of the autumnal moth, Epirrita autumnata, detoxify toxic flavonoid aglycones present in birch leaves via glycosylation (Salminen et al., 2004). Adults and larvae of the alkaloid-adapted leaf beetle, Oreina cacaliae, convert tertiary alkaloids into more polar alkaloid glycosides (Hartmann et al., 1999). Especially in vertebrates, many more

Fig. 5. UGT mediated metabolism of gossypol. (A) Diglycosylated gossypol isomers 1 to 5 and (B) monoglycosylated gossypol isomers 6 to 8 formed by heterologously expressed UGT enzymes: (a) UGT33B4, (b) UGT40D1, (c) UGT40D2, (d) UGT40Q1, (e) UGT41B1, (f) UGT41B3 or (g) endogenous UGTs of the Sf9 cells used as a control. Depicted are extracted ion chromatograms of multiple reaction monitoring (MRM) of m/z 841.000/679.000 and m/z 679.000/517.000. Arrows highlight the diglycosylated gossypol isomer 5, solely produced by (b) UGT40D1 and (f) UGT41B3.

56

C. Krempl et al. / Insect Biochemistry and Molecular Biology 71 (2016) 49e57

examples for the detoxification of xenobiotics via glucuronidation exist (Miners and Mackenzie, 1991; Rowland et al., 2013). Our comparative analysis of the feces of H. armigera and H. virescens revealed differences in the identity of mono- and diglycosylated gossypol isomers. In the feces of H. armigera, we found five diglycosylated gossypol isomers of which four were also detected in H. virescens feces. Three monoglycosylated gossypol isomers were identified in the feces of both species. It is interesting to note that, as one molecule of gossypol possesses six hydroxyl groups, a total of nine isomeric diglycosides are theoretically possible (Fig. 2B) but only 5 isomers were detected. Possibly some of the isomers elute at the same retention time. Our results indicate species-specific differences in the isomer pattern of glycosylated gossypol in H. armigera and H. virescens. H. armigera showed a preference for the formation of the diglycosylated gossypol isomers 4 and 5, whereas H. virescens produced less diglycosylated gossypol compared to H. armigera. These differences in the isomer pattern between related species may be attributed to differences in the activity of UGT enzymes, which are candidates for the glycosylation of toxins in insects and which can differ in their regioselectivity. We show for the first time that insects form different gossypol isomers and that preferences for single isomers exist in different insect species. In order to test for specific gossypol glycosylation activity, we heterologously expressed H. armigera candidate UGTs in Sf9 insect cells and determined their gossypol glycosylation activity using in vitro assays. Out of the six candidate UGTs only UGT41B3 and to a lesser extent UGT40D1 were capable of glycosylating gossypol. When screening the heterologously expressed UGTs for activity, we found that the microsomal extracts of non-transfected Sf9 cells also possessed gossypol glycosylation activity. This complicated the analysis of the enzymatic assays but indicates that endogenous Sf9 UGTs are able to form almost all mono- and diglycosylated gossypol isomers that were also detected in H. armigera. Sf9 cells originate from S. frugiperda (Lepidoptera: Noctuidae), a generalist herbivore and a relative of H. virescens and H. armigera. A blast search with the tested H. armigera UGTs identified homologs in the transcriptome of the Sf9 cell line (data not shown). The diglycosylated gossypol isomer 5, which was only detected in enzyme assays incubated with heterologous UGT41B3 and UGT40D1, however, was not formed by endogenous Sf9 UGTs, confirming the formation solely by the heterologous UGTs. The diglycosylated gossypol isomer 5, detected in the enzyme assays, was also found in considerable amounts in H. armigera feces and not in H. virescens, indicating that H. virescens lacks isomer 5specific UGT activity. Species-specific differences in UGT activity were previously described by Ahn et al. (2011a) showing that the homogenate of entire larvae of the two generalist moth species H. armigera and Helicoverpa zea showed higher glycosylation activity towards the plant toxin capsaicin as compared to Helicoverpa assulta, a specialist of the genus Capsicum (Solanaceae). Our data show a species-specific glycosylation pattern of gossypol in two generalist moth species suggesting differences in the glycosylation capacity of both species most likely due in part to the lack of the UGT41B3 homolog in H. virescens. Our low recovery of free gossypol from H. armigera and H. virescens feces (4% of free gossypol) and larvae (0.1% of free gossypol) confirms the proposed strong binding of gossypol to proteins (Lyman et al., 1959). Indeed, higher amounts of gossypol were detected using the aniline method described by Smith (1967). Bound and free gossypol reacts with an excess of aniline under high temperature and forms anilinogossypol products. We could thus extract about 50% of the total ingested gossypol from feces and about 2% from entire larval bodies of both species. Previous studies also showed that invertebrates (Rojas et al., 1992) as well as

vertebrates (Abou-Donia and Lyman, 1970) excrete a high amount of the ingested gossypol in the feces. In conclusion, our data show that larvae of H. armigera and H. virescens predominantly excrete gossypol rather than store it in their bodies. The remaining percentage of the ingested gossypol which could not be recovered might be glycosylated gossypol that cannot be quantified because we lack the respective standard, or strongly bound gossypol which cannot be extracted. In addition other, currently unknown detoxification pathways of gossypol might account for some of the undetectable gossypol. Previous transcriptional and RNAi studies suggest the involvement of a specific cytochrome P450 monooxygenase, CYP6AE14, in gossypol detoxification due to its high up-regulation after larval feeding on gossypol diet (Celorio-Mancera et al., 2011) and the reduction of larval growth observed after feeding on plants with knocked-down CYP6AE14 via RNAi (Mao et al., 2007, 2011). However, a direct involvement of this enzyme in gossypol metabolism was not shown. So far, the fate of gossypol bound to proteins present in the diet after ingestion in the gut lumen of larvae is unknown and it remains unclear how much of the bound gossypol is released from diet proteins and actually binds to proteins present in the gut lumen or the surrounding tissue, thus damaging the organism. Free gossypol (i.e. that is not bound via a Schiff base to proteins) is considered to be especially toxic. It might be possible that the low amount of free gossypol we detected in the larval body is already below the toxic threshold. Our experiments clearly show that at least a proportion of gossypol is metabolized by glycosylation after ingestion and subsequently excreted in the feces by H. armigera and H. virescens. In addition to the direct effects of glycosylation on gossypol, such as reduced reactivity and enhanced excretion, another important effect might be a sterical hindrance of the reactive aldehyde groups, thus preventing the formation of Schiff bases with proteins. However, to prove this assumed lower toxicity of glycosylated metabolites, feeding studies with isolated and purified glycosylated gossypol are required. Here, we identified for the first time UDPglycosyltransferases of H. armigera that are involved in the glycosylation of gossypol and thus might contribute to its detoxification. Acknowledgements We thank Domenica Schnabelrauch, Henriette RingysBeckstein, Stephanie Ley, and Sheila Milker for technical assistance, Regina Seibt for insect rearing, Riya Christina Menezes and Marco Kai for high resolution MS measurements. Additionally, we thank Yannick Pauchet and Roy Kirsch for troubleshoot support and Matan Shelomi for editorial assistance. Financial support was provided by the Max-Planck-Gesellschaft. N.J. was supported by a grant from the Deutsche Forschungsgemeinschaft (DFG, Germany; JO 855/1-1). Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.ibmb.2016.02.005. References Abou-Donia, M., 1975. Metabolic fate of gossypol: the metabolism of [14 C] Gossypol in swine. Toxicol. Appl. Pharmacol. 31, 32e46. Abou-Donia, M., Lyman, C.M., 1970. Metabolic fate of gossypol: the metabolism of gossypol-14 C in laying hens. Toxicol. Appl. Pharmacol. 17, 160e173. Ahmad, S.A., Hopkins, T.L., 1992. Phenol b-glucosyltransferase and b-glucosidase activities in the tobacco hornworm larva Manduca sexta (L.): properties and tissue localization. Arch. Insect Biochem. Physiol. 21, 207e224.

C. Krempl et al. / Insect Biochemistry and Molecular Biology 71 (2016) 49e57 rez, F.R., Heckel, D.G., 2011a. A host-plant specialist, HeliAhn, S.-J., Badenes-Pe coverpa assulta, is more tolerant to capsaicin from Capsicum annuum than other noctuid species. J. Insect Physiol. 57, 1212e1219. rez, F.R., Reichelt, M., Svatos, A., Schneider, B., Gershenzon, J., Ahn, S.-J., Badenes-Pe Heckel, D.G., 2011b. Metabolic detoxification of capsaicin by UDPglycosyltransferase in three Helicoverpa species. Arch. Insect Biochem. Physiol. 78, 104e118. Ahn, S.-J., Vogel, H., Heckel, D.G., 2012. Comparative analysis of the UDPglycosyltransferase multigene family in insects. Insect biochem. Mol. Biol. 42, 133e147. Bock, K.W., 2015. The UDP-glycosyltransferase (UGT) superfamily expressed in humans, insects and plants: animal-plant arms-race and co-evolution. Biochem. Pharmacol. http://dx.doi.org/10.1016/j.bcp.2015.10.001. Bozzolan, F., Siaussat, D., Maria, A., Durand, N., Pottier, M.A., Chertemps, T., che-Coisne, M., 2014. Antennal uridine diphosphate (UDP)-glycosylMaïbe transferases in a pest insect: diversity and putative function in odorant and xenobiotics clearance. Insect Mol. Biol. 23, 539e549. Celorio-Mancera, M., Ahn, S.-J., Vogel, H., Heckel, D.G., 2011. Transcriptional responses underlying the hormetic and detrimental effects of the plant secondary metabolite gossypol on the generalist herbivore Helicoverpa armigera. BMC Genomics 12, 575. Daimon, T., Hirayama, C., Kanai, M., Ruike, Y., Meng, Y., Kosegawa, E., Nakamura, M., Tsujimoto, G., Katsuma, S., Shimada, T., 2010. The silkworm Green b locus encodes a quercetin 5-O-glucosyltransferase that produces green cocoons with UV-shielding properties. Proc. Natl. Acad. Sci. USA 107, 11471e11476. Dodou, K., 2005. Investigations on gossypol: past and present developments. Expert Opin. Investig. Drugs 14, 1419e1434. Feyereisen, R., 2012. Insect CYP genes and P450 enzymes. In: Gilbert, L.I. (Ed.), Insect Molecular Biology and Biochemistry. Elsevier, London, pp. 236e316. Hartmann, T., Theuring, C., Schmidt, J., Rahier, M., Pasteels, J.M., 1999. Biochemical strategy of sequestration of pyrrolizidine alkaloids by adults and larvae of chrysomelid leaf beetles. J. Insect Physiol. 45, 1085e1095. Heckel, D.G., 2014. Insect detoxification and sequestration strategies. Ann. Plant Rev. 47, 77e114. Heidel-Fischer, H.M., Vogel, H., 2015. Molecular mechanisms of insect adaptation to plant secondary compounds. Curr. Opin. Insect Sci. 8, 8e14. Hopkins, T.L., Kramer, K.J., 1992. Insect cuticle sclerotization. Annu. Rev. Entomol. 37, 273e302. Joyner, K., Gould, F., 1985. Developmental consequences of cannibalism in Heliothis zea (Lepidoptera: Noctuidae). Ann. Entomol. Soc. Am. 78, 24e28. Kojima, W., Fujii, T., Suwa, M., Miyazawa, M., Ishikawa, Y., 2010. Physiological adaptation of the Asian corn borer Ostrinia furnacalis to chemical defenses of its host plant, maize. J. Insect Physiol. 56, 1349e1355. Kramer, K.J., Hopkins, T.L., 1987. Tyrosine metabolism for insect cuticle tanning. Arch. Insect Biochem. Physiol. 6, 279e301. Kuwar, S.S., Pauchet, Y., Vogel, H., Heckel, D.G., 2015. Adaptive regulation of digestive serine proteases in the larval midgut of Helicoverpa armigera in response to a plant protease inhibitor. Insect biochem. Mol. Biol. 59, 18e29. Laughton, M.J., Halliwell, B., Evans, P.J., Robin, J., Hoult, S., 1989. Antioxidant and pro-oxidant actions of the plant phenolics quercetin, gossypol and myricetin:

57

effects on lipid peroxidation, hydroxyl radical generation and bleomycindependent damage to DNA. Biochem. Pharmacol. 38, 2859e2865. Liu, H., Sun, H., Lu, D., Zhang, Y., Zhang, X., Ma, Z., Wu, B., 2014. Identification of glucuronidation and biliary excretion as the main mechanisms for gossypol clearance: in vivo and in vitro evidence. Xenobiotica 44, 696e707. Lyman, C.M., Baliga, B., Slay, M.W., 1959. Reactions of proteins with gossypol. Arch. Biochem. Biophys. 84, 486e497. Magdalou, J., Fournel-Gigleux, S., Ouzzine, M., 2010. Insights on membrane topology and structure/function of UDP-glucuronosyltransferases. Drug Metab. Rev. 42, 159e166. Mao, Y.B., Cai, W.J., Wang, J.W., Hong, G.J., Tao, X.Y., Wang, L.J., Huang, Y.P., Chen, X.Y., 2007. Silencing a cotton bollworm P450 monooxygenase gene by plantmediated RNAi impairs larval tolerance of gossypol. Nat. Biotechnol. 25, 1307e1313. Mao, Y.B., Tao, X.Y., Xue, X.Y., Wang, L.J., Chen, X.Y., 2011. Cotton plants expressing CYP6AE14 double-stranded RNA show enhanced resistance to bollworms. Transgenic Res. 20, 665e673. Miners, J.O., Mackenzie, P.I., 1991. Drug glucuronidation in humans. Pharmacol. Ther. 51, 347e369. Paquette, S., Møller, B.L., Bak, S., 2003. On the origin of family 1 plant glycosyltransferases. Phytochemistry 62, 399e413. Radominska-Pandya, A., Bratton, S.M., Redinbo, M.R., Miley, M.J., 2010. The crystal structure of human UDP-glucuronosyltransferase 2B7 C-terminal end is the first mammalian UGT target to be revealed: the significance for human UGTs from both the 1A and 2B families. Drug Metab. Rev. 42, 133e144. Robertson, H., Martos, R., Sears, C., Todres, E., Walden, K., Nardi, J., 1999. Diversity of odourant binding proteins revealed by an expressed sequence tag project on male Manduca sexta moth antennae. Insect Mol. Biol. 8, 501e518. Rojas, M.G., Stipanovic, R., Williams, H., Vinson, S., 1992. Metabolism of gossypol by Heliothis virescens (F.) (Lepidoptera: noctuidae). Environ. Entomol. 21, 518e526. Rowland, A., Miners, J.O., Mackenzie, P.I., 2013. The UDP-glucuronosyltransferases: their role in drug metabolism and detoxification. Int. J. Biochem. Cell Biol. 45, 1121e1132. Salminen, J.-P., Lahtinen, M., Lempa, K., Kapari, L., Haukioja, E., Pihlaja, K., 2004. Metabolic modifications of birch leaf phenolics by an herbivorous insect: detoxification of flavonoid aglycones via glycosylation. Z. Naturforsch 59, 437e444. Smith, F., 1967. Determination of gossypol in leaves and flower buds of Gossypium. J. Am. Oil Chem. Soc. 44, 267e269. Stipanovic, R.D., Lopez Jr., J.D., Dowd, M.K., Puckhaber, L.S., Duke, S.E., 2006. Effect of racemic and (þ)- and (-)-gossypol on the survival and development of Helicoverpa zea larvae. J. Chem. Ecol. 32, 959e968. Wilkinson, C.F., 1986. Xenobiotic conjugation in insects. In: Paulson, G.D., Caldwell, J., Hutson, D.H., Menn, J.J. (Eds.), Xenobiotic Conjugation Chemistry. American Chemical Society, Washington, DC, pp. 48e61. Wouters, F.C., Reichelt, M., Glauser, G., Bauer, E., Erb, M., Gershenzon, J., Vassao, D.G., 2014. Reglucosylation of the benzoxazinoid DIMBOA with inversion of stereochemical configuration is a detoxification strategy in lepidopteran herbivores. Angew. Chem. 53, 11320e11324.