Introduction

To control insect pests of stored agricultural products, phosphine (PH3) has been widely used for fumigation as an alternative to methyl bromide, which depletes the ozone layer in the atmosphere1. However, frequent worldwide emergence of PH3 resistant (PH3–R) insect pests including rice weevil (Sitophilus oryzae)2, Cryptolestes ferrugineus3,4, Rhyzopertha dominica5, and Tribolium castaneum6,7,8 in stored grains, because of intensive PH3 use and global warming, threatens global grain reserves9. Although the frequency of PH3–R insects in the field is low, weak PH3 resistance has been observed in S. oryzae, which may be ascribed to a major autosomal and incompletely recessive gene, without any evidence of a fitness cost in the absence of PH3 selection2. Recently, PH3–R insects have been frequently reported in several countries, including China, Vietnam, Bangladesh, Brazil, Turkey, the United States, Korea, and Australia4,5,7,8,10,11,12,13,14,15, and several PH3–R strains of S. oryzae have been identified in Australia and Brazil2,16 as well as in Korea17.

PH3–R insect pests harbor an identical amino acid substitution in a core metabolic gene encoding dihydrolipoamide dehydrogenase (DLD), consistent with the presence of a genetic resistance factor identified in R. dominica and T. castaneum7. Variants of DLD represent a PH3 resistance factor because metabolite profiles of resistant insect pests differ from those of susceptible pests under PH3 exposure18. A fitness cost associated with the strongly resistant (SR) allele of the dld gene appears to be segregating in insect populations in the absence of selection, implying that the prevalence of resistance is a potential threat to resistant species under considerable selection pressure11. Nevertheless, the genetic and physiological mechanisms of PH3 resistance in insect pests remain unclear. Understanding the PH3 resistance mechanism in PH3–R strains is urgently needed not only to screen new pesticides but also to develop a more biologically safe and sustainable method of controlling the global population of pesticide resistant pests of stored agricultural products19,20.

Intriguingly, the emergence of strong PH3 resistance in R. dominica and T. castaneum in India, the United States, and Australia, presumably because of the substitution of proline with serine at amino acid position 49 or 45 (P49S or P45S, respectively) in DLD, may be ascribed to a strong selection by the excessive use of PH3 fumigation7,10. On the other hand, we previously showed that the expression of several key metabolic genes, including those encoding glyceraldehyde-3-phosphate dehydrogenase (GAPDH), triosephosphate isomerase (TPI), and DLD was significantly reduced in the PH3–R R. dominica strain (CRD343) compared with the PH3 susceptible R. dominica strain21. In addition, genes encoding sodium channel proteins, glutamate racemase, enolase, and vitellogenin were highly expressed in the PH3–R strain21. These findings may be ascribed to a possible trade-off between survival and proliferation plays a key role in the acquisition of pesticide resistance. However, this result does not support the previous observation that PH3 resistance is highly species-specific21. Recently, expression profiling of four mitochondrial genes, including cox1, nad3, atp6, and cob, in C. ferrugineus by quantitative real-time PCR (qRT-PCR) revealed a strong negative correlation between PH3 resistance and respiratory chain function3, suggesting that the emergence of fumigant resistance is associated with the modulation of energy production in mitochondria.

In this study, we investigated differences in protein profiles of the susceptible strain and two distinct PH3–R strains of S. oryzae by performing comparative proteome analysis and ethyl formate (EF) inhibition kinetics. In addition, expression levels of several putative PH3–R marker genes were validated by qRT-PCR. Subsequently, we aimed to identify single nucleotide polymorphisms (SNPs) or mutations in complex I (ND) subunit encoding genes through mitochondrial DNA (mtDNA) sequencing of both PH3–R S. oryzae strains.

Results

Differential acute PH3 toxicity of the susceptible and resistant S. oryzae strains

To assess the extent of PH3 resistance in S. oryzae, a susceptible wild-type (WT) strain, obtained from Australia, was compared with two resistant strains (R1 and R2) obtained from two geographically distinct Provinces in Korea (Table 1). The disinfectant concentration-time (Ct) values (mg·h/L) of the WT and PH3–R strains demonstrated that despite the similar level of PH3 toxicity (Ct50, Ct value to achieve 50% mortality) between WT and R1 strains, the Ct99 (Ct value to achieve 99% mortality) value of the R2 strain at 20 °C was 29- and 10-fold higher than that of WT and PH3–R1 strains, respectively. In addition, the R2 strain was more resistant to PH3 than the R1 and WT strains, presumably because of a mutation in the dld gene encoding DLD (Table 1 and Supplementary Table S1). Thus, acute PH3 toxicity data indicate that the PH3 resistance mechanism of the R2 strain is considerably different from that of the R1 strain. Therefore, to discriminate between these strains, R1 and R2, hereafter we refer to them as moderately resistant (MR) and strongly resistant (SR) S. oryzae strains, respectively.

Table 1 Analysis of phosphine (PH3) resistance in susceptible wild-type (WT) and resistance strains (R1 and R2).
Full size table

To investigate the effect of PH3 fumigation on the respiratory efficiency of S. oryzae strains, we measured the activities of several enzymes as reference proteins in the PH3 susceptible WT and PH3–R strains (Fig. 1a). The cytochrome c oxidase (COX) activity of both PH3–R strains was higher than that of the WT strain, suggesting as positive correlation between COX activity and the degree of PH3 resistance. However, there was little difference in the activities of acetylcholinesterase (AChE) and carboxylesterase (CE) between the WT and PH3–R strains, although glutathione S-transferase (GST) activity was slightly lower in the SR strain than in MR and WT strains. In addition, these differential enzyme activity profiles demonstrated that the mode of action of PH3 in the PH3–R strains differs from that of organophosphates and carbamates, which affect the function of the nervous system of insects22,23.

Figure 1
figure 1

Effect of ethyl formate (EF) on cytochrome c oxidase (COX) activity in phosphine (PH3) susceptible and resistant strains. (a) Enzyme activities of COX, acetylcholinesterase (AChE), glutathione S-transferase (GST), carboxylesterase (CE), expressed as unit/mg protein of rice weevils. Significant differences among PH3 susceptible (wild-type [WT]) and resistant strains (moderately resistant [MR] and strongly resistant [SR]) were determined using one-way ANOVA (p < 0.05), followed by Tukey’s post-hoc tests and are indicated using different letters. (b) Dose-response curves constructed by non-linear regression analysis and half maximal inhibitory concentration (IC50) of EF on COX activity in all three strains. See also Supplementary Table S2. (c) EF-mediated inhibition kinetics of COX in WT, MR, and SR strains using Lineweaver–Burk plots. (d) Depiction of proposed inhibitory targets of fumigants (EF and PH3), according to the degree of PH3 resistance in S. oryzae strains. The symbol ‘?’ indicates the unknown binding sites of the electron transfer chain (ETC) or other redox enzymes. In all experiments, protein samples were isolated from three independent replicates (n = 100 rice weevils) in each strain and each assay, and three biological replicates were performed for each experiment. All data represent mean ± standard deviation (SD).

Full size image

To further investigate the extent of PH3 resistance in the WT and PH3–R strains, we performed the EF-mediated inhibition kinetics of these strains24. The inhibitory effect of EF as another fumigant on COX activity in each strain revealed that the half maximal inhibitory concentration (IC50) of the SR strain was not significantly different from that of the WT strain (Fig. 1b). In addition, the inhibition constant (Ki) of the SR strain was lower than that of the WT strain under the same concentration of cytochrome used as the substrate (Supplementary Table S2). However, the MR strain did not show any significant change in IC50 and Ki values, compared with the WT strain. Intriguingly, the EF-mediated inhibition of COX activity in the WT strain was negligible, whereas that of PH3–R strains was pronounced in a concentration-dependent manner (Fig. 1c). Therefore, the discrepancy between the weak EF-mediated inhibition of COX activity and high Ct99 value of the disinfectant PH3 in the SR strain suggests that the acquisition of PH3 resistance in S. oryzae strains is not directly correlated with the degree of EF-mediated inhibition of COX activity in vitro, implying that resistance to PH3 in S. oryzae is partially gained through the perturbation of other redox proteins or metabolic sites in mitochondria (Fig. 1d).

Distinct reorganization of cellular and mitochondrial metabolisms in PH3–R strains

To understand the metabolic consequences of PH3 exposure in both PH3–R strains of S. oryzae, we analyzed the difference in their proteome profiles using nLC-ESI-MS/MS (Fig. 2). Both PH3–R strains exhibited similar abundance patterns of several proteins such as the 40 S ribosomal protein, kinases, and glycoside hydrolase (Supplementary Table S3). Remarkably, however, the MR and SR strains showed differential abundance profiles of proteins affecting their mitochondrial metabolism and energy production including core metabolic enzymes involved in central metabolism, biosynthesis, cell signaling, and enzyme regulation (Fig. 2 and Supplementary Table S3). In the MR strain, 37 proteins showed differential abundance compared with the susceptible WT strain. Among these proteins, proteasome subunits, protease regulatory proteins, and stress responsive proteins (groEL, heat shock 70 kDa protein cognate 3, and stress-induced phosphoprotein 1) were highly abundant, whereas muscle specific proteins, ATP synthase subunits, and several mitochondrial proteins (including inorganic transport and ATP-ADP antiporter proteins) were less abundant in the MR strain than in the WT strain, indicating that an elevation in the stress response would be sufficient for the acquisition of moderate PH3 resistance (Fig. 2a and Supplementary Table S3). On the other hand, the SR strain exhibited a different protein abundance pattern, revealing that proteins involved in muscle contraction such as Ca2+-dependent troponin proteins and heat shock protein 90 (Hsp 90) were highly abundant (Fig. 2b and Supplementary Table S3). Moreover, the SR strain contained >2-fold higher levels of ND, a major component of the electron transport chain (ETC), troponin C, myosin light chain, and tropomyosin than the MR and WT strains. Remarkably, the mitochondrial ATP synthase and Ca2+-transporting ATPase were less abundant in the MR strain, whereas the level of metabolic enzymes involved in glycolysis and actin depolymerization was significantly lower in the SR strain than in the MR and WT strains (Fig. 2 and Supplementary Table S3). Overall, in S. oryzae, a substantial partitioning of the energy transduction system in the mitochondrial ETC and core metabolism occurs to induce PH3 resistance. In both PH3–R strains, the level of metabolic enzymes, involved in glycolysis and the oxidative tricarboxylic acid (TCA) cycle, was significantly suppressed (Fig. 2a). In particular, the level of ND was highly up-regulated in the SR strain (Fig. 2b).

Figure 2
figure 2

Volcano plots showing the fold-change and significance level of proteomes in the PH3–R strains. (a,b) Comparison of changes in gene expression (Mann-Whitney Test) as a function of protein accumulation in the WT (control) vs. MR strain (a) and WT vs. SR strains (b). The X-axis shows the fold-change, and the Y-axis shows the significance level. Red and green dots represent up-regulated and down-regulated genes, respectively. The horizontal brown dashed line marks the significance threshold (p < 0.1), and the vertical brown dashed line displays the value of 1.5-fold-change. Full description of the abbreviated protein names are listed in Supplementary Table S3.

Full size image

To further investigate whether either mitochondrial or core metabolism is correlated with PH3 resistance in S. oryzae, we analyzed the expression of 18 genes involved in PH3 resistance including several known reference genes by qRT-PCR (Fig. 3); these genes were selected because the nucleotide sequences of these genes or their homologs were available (Supplementary Table S4). The expression levels of the genes ndufs1 and chm encoding ND and Rab protein geranylgeranyltransferase component A (GGTase), respectively, in both PH3–R strains were substantially higher than those in the WT strain. Indeed, both PH3–R strains showed relatively greater abundance of chm related to cellular signaling pathways regulating than the WT strain; this protein regulates normal cellular proliferation. By contrast, the expression level of the genes encoding ATP synthase subunit gamma (ATP5F1A), ADP, ATP carrier protein 1, Hsp90, alpha-amylase (AA), profilin (PFN), myosin heavy chain-1 (NMMHC), and fructose-bisphosphate aldolase (FBA) was down-regulated in both PH3–R strains (Fig. 3), implying that PH3 fumigation acts as a major selective pressure to change the core cellular metabolism that affects cellular energy production. The expression levels of hsp90 and amy1 involved in protein folding and carbohydrate metabolism, respectively, were down-regulated in both PH3–R strains. In addition, the expression levels of acta2 and myh1 in relation to muscle contraction were lower in the SR strain than in the MR and WT strains.

Figure 3
figure 3

Differential expression patterns of a few selected genes in the WT, MR, and SR strains. Quantitative real-time (qRT)-PCR were performed in duplicate for every three independent biological replicates (n = 30). Gene expression levels were normalized by the expression of ribosomal protein L29 (rpl29) and were calculated using the 2−∆∆Ct method. Heat map was constructed using Log2 gene expression ratio between the WT (control) and PH3–R strains. Each gene is represented by a gene name (protein name). dld, dihydrolipoamide dehydrogenase E3 subunit; ndufs1, NADH-ubiquinone oxidoreductase; cox2, cytochrome oxidase subunit II; atp5f1a, ATP synthase subunit alpha; atp5f1b, ATP synthase subunit beta; aac1, ADP, ATP carrier protein 1; gpi, glucose-6-phosphate isomerase; fba, fructose-bisphosphate aldolase; hsp90, heat shock protein 90; gh48, glycoside hydrolase family protein 48; amy1, alpha-amylase; pfn, Profilin; acta2, actin, muscle; myh1, myosin heavy chain 1; tpm2, tropomyosin-2; chm, Rab proteins geranylgeranyltransferase component A; prkar1a, cAMP-dependent protein kinase regulatory subunit; pme, pectin methylesterase; The primers used in this study are listed in Supplementary Table S4.

Full size image

Mitochondrial mutations in the SR strain

Our proteomic and qRT-PCR data suggest that both PH3–R strains share a common PH3 resistance mechanism to minimize the ATP-consuming metabolism by repressing core metabolism and modulating the magnitude of respiration. However, unlike the MR strain, the SR strain exhibited a distinct way of lowering substrate-level phosphorylation and mitochondrial respiration. To understand the metabolic discrepancy between the MR and SR strains with respect to cellular metabolism and bioenergetics, we sequenced the mtDNAs of the WT and PH3–R strains using next generation sequencing technology, and investigated whether the metabolic differences between the two PH3–R strains may be ascribed to the occurrence of SNP-like genetic mutations. A total of 15 point mutations were identified in nine genes including those encoding COX subunits 1 and 2 (cox1 and cox2), ND subunits 1, 2, 4, 4 l, 5, and 6 (nad1, nad2, nad4, nad4l, nad5, and nad6), and ATP synthase subunit 6 (atp6) (Table 2). Marginal EF-mediated inhibition patterns of COX activity between the WT and PH3–R strains imply the occurrence of mutations in other genes encoding oxidative phosphorylation (OXPHOS) system (Fig. 1c). Although two SNPs were identified in the cox1 and cox2 genes in the SR strain, these mutations were silent (Table 2). Notably, a number of mtDNA point mutations (m.) were identified in the nad4, nad5, and nad6 genes encoding protein subunits responsible for the assembly of ND in the respiratory chain: m.9082 C > A, m.8297 G > A, m.8231 G > A, m.8102 T > G in nad4; m.7551 C > T, m.7405 G > A, m.7353 C > G, m.6678 C > T in nad5, and m.10018 A > G in nad6 (the base of the WT and MR strains > that of the SR strain). Among these, m.9082 C > A in nad4, m.7405 G > A and m.7353 C > G in nad5, and m.10018 A > G in nad6 caused amino acid substitutions in the SR strain (Table 2 and Supplementary Fig. S1); serine (Ser; polar) to asparagine (Asn; polar), aspartic acid (Asp; acidic) to glutamic acid (Glu; acidic), and Asn (polar) to Ser (polar), respectively, while that in nad4 caused a drastic change from alanine (Ala; non-polar) to the negatively charged Glu (Table 2 and Supplementary Table S5 and Fig. S1). In addition, the deduced amino acid sequences of the nad genes in S. oryzae were aligned to those of their counterparts in several model organisms (Gallus gallus, Mus musculus, Homo sapiens, and Bos taurus) and Sitophilus zeamais to identify conserved regions among different organisms, demonstrating that the m.9082 C > A in nad4 is highly conserved with all identical sequence (Ala). The m.7405 G > A mutation in nad5 showed similar sequences while other sites (m.7353 C > G in nad5 and m.10018 A > G in nad6) are variable (Supplementary Fig. S2).

Table 2 Non-synonymous mutations in mitochondrial protein-coding genes of WT, MR, and SR strains of Sitophilus oryzae.
Full size table

Structural analysis of amino acid substitutions in the ND subunits

To assess the potential impact of the identified missense mutations in nad4, nad5, and nad6 genes on mitochondrial function, we constructed three dimensional (3D) models of the corresponding ND subunits, based on the protein structures of ND4 and ND5 subunits in Mus musculus (PDB No. 6G2J) and ND6 in Thermus thermophilus (PDB No. 4HEA) (Fig. 4a). We analyzed the predictive effects of mutations on the structure and function of respiratory chain complexes using the amino acid substitution algorithms (Table 3). These were combined to investigate the effect of mutations on the biological functions of proteins, with an improved prediction accuracy of >69%, when analyzed with SNPs associated with mitochondrial dysfunction25. Results using these three servers indicated that the amino acid substitution in ND4 was deleterious, while that in ND6 was relatively neutral. Subsequently, three substitutions (Ala73Glu in ND4, Ser169Asn in ND5, and Asn104Glu in ND6) identified in the SR strain were mapped onto their model structures (Fig. 4a,b). The membrane-bound domain of ND is involved in proton translocation across the membrane26. All three mutation sites in the model structures of S. oryzae ND were superimposed with polar amino acid residues, potentially involved in proton translocation chains of Escherichia coli NuoJ, NuoM, and NuoL subunits with the root-mean-square-deviation values of 4.875 Å, 1.367 Å, and 1.191 Å, respectively (Fig. 4b). Remarkably, the Ala73Glu substitution in ND4 was only 3.5 Å from a residue in E. coli ND acting as a proton entrance, whereas the Asn104Glu substitution in ND6 was 7.8 Å away from the proton exit of ND (Fig. 4b).

Figure 4
figure 4

Schematic representation of the S. oryzae mitochondrial respiratory chain and putative mutation sites in mitochondrial (mt) genes associated with PH3 resistance in NADH dehydrogenase (ND). (a) Structure of ND (NADH–ubiquinone oxidoreductase) with three mt genes encoding ND4, ND5, and ND6 subunits (colored) modeled based on swine (PDB code 5GUP), bovine (5LC5), murine (6G2J), and bacterial (4HEA) complex I. Three mutation sites indicated by spheres are mapped onto their corresponding 3D model structures depicted as transparent ribbons. The iron-sulfur (Fe-S) cofactors of ND are depicted as gray spheres, and FMN cofactor as gray sticks. (b) Close-up views of the ND subunits (PDB 3RKO): nuoJ, nuoM, and nuoL (green, purple, and red, respectively) from E. coli BL21 (DE3). Polar residues along the channel are shown as sticks, with each arrow indicating the approximate proton translocation paths in E. coli ND. Distances from the mutation sites to adjacent polar residues are marked with yellow dotted lines.

Full size image
Table 3 Structural classification and pathogenic prediction of mutations in mitochondrial protein-coding genes of S. oryzae.
Full size table

Discussion

The response of redox-active DLD, a core metabolic enzyme, to PH3 differed between the PH3–R strains and WT strain in such a way that the P49S mutation in the rph2 locus contributes to PH3 resistance in association with a synergistic rph1 locus18. This coincides with the observation that the P49S substitution in DLD is frequently found in PH3–R strains of R. dominica and T. castaneum from India and Australia in the absence of PH3 fumigation10, indicating that PH3 fumigation acts as a selection pressure that generates PH3–R insect pests by altering the activity of core metabolic enzymes. This is further supported by the recent finding that rphl variants identified in the PH3–R strains of R. dominica, S. oryzae, C. ferrugineus, and T. castaneum share common mutations in an orthologous gene encoding a cytochrome b5 fatty acid desaturase (Cyt-b5-r)13. Mutations in Cyt-b5-r in PH3–R insects limit the potential for lipid peroxidation through reactive oxygen species generated by DLD. Moreover, a proteomic study revealed that a PH3–R strain of R. dominica exhibited PH3 resistance by altering the expression of 21 proteins involved in the TCA cycle and glycolysis21.

In this study, we found two distinct PH3–R S. oryzae strains (MR and SR) that employed different strategies to develop PH3 resistance (Table 1 and Fig. 1). The potential target of PH3 seems to be the energy transduction system including the ETC located in the mitochondrial membranes of eukaryotic cells27. The PH3-mediated inhibition of COX activity interferes with energy production, rendering insects incapable of performing various functions because of the shortage of ATP28,29. In this study, both PH3–R strains showed slightly higher COX activity than the susceptible WT strain. However, the IC50 values of EF in these strains was not proportional to the extent of PH3 resistance (Fig. 1a,b), which is consistent with EF-mediated inhibition of COX activity30. However, the different magnitude of EF-mediation inhibition of COX activity in S. oryzae strains (i.e., WT and both R strains) may be caused by additional mtDNA mutations as well as changes in expression of genes encoding other energy transducing proteins (Fig. 1c and Table 2). Our data suggest that moderate PH3 resistance may be acquired by the modulation of expression levels of core metabolic enzymes, without mutations in the dld gene (Fig. 1a and Table 1 and Supplementary Table S1). The SR strain in the DLD mutation background exhibited extraordinary PH3 resistance when compared with the MR and WT strains (Table 1 and Supplementary Table S1). Overall, the energy transduction system in mitochondria and core metabolism seem to undergo a substantial metabolic partitioning, suggesting that reprogramming activities involved in metabolism and respiratory chains, such as altered bioenergetics, suppressed biosynthesis, and redox balance improves cellular fitness and provides a selective advantage during PH3 fumigation. Recent studies on C. ferrugineus and R. dominica suggest that specific core metabolisms and the mitochondrial ETC are highly associated with PH3 resistance3,5,6. Although the general PH3 resistance mechanisms in insects are related to target site insensitivity, increases in detoxifying enzyme levels, behavioral modifications, and physiological alterations (Fig. 1)31,32,33,34, the basis of the induction of PH3 resistance in insects, and the molecular and genetic bases of energy modulation by insect pests under life-threatening conditions remain unclear.

Proteome profiles of PH3–R strains indicate that reprogrammed core metabolism and mitochondrial function including the respiratory chains are the basis of PH3 resistance (Fig. 2). The MR and SR strains exhibited a reduction in glycolytic flux as well as in the level of glycolytic intermediates to suppress subsidiary pathways, resulting in reduced metabolic demands (Fig. 5). In addition, the level of TCA cycle intermediates, which serve as precursors for macromolecule biosynthesis, was also reduced, which is consistent with a classical example of a reprogrammed metabolic pathway in cancer cells such as the Warburg effect or aerobic glycolysis35. Intriguingly, the SR strain shares several common metabolic features with cancer cells. The primary characteristic of cancer cell is the avoidance of OXPHOS. Instead, cancer cells utilize substrate-level phosphorylation through aerobic glycolysis and lactate production, regardless of oxygen availability, resulting in acidosis within cells36,37. Similarly, ND was impaired in the SR strain because of missense mutations in genes encoding ND subunits 4–6 (Table 3 and Fig. 4), presumably resulting in the modulation of energy production during aerobic glycolysis as well as mitochondrial metabolism for lactate fermentation with substrate-level phosphorylation. This indicates that the inhibition by PH3 exposure could be compensated by activating other proteins for alternative respiration, which might be more favorable for cellular survival under PH3 exposure (Figs 2b and 5). Despite the incredible genetic and histological heterogeneity of PH3–R strains, resistance to PH3 might be ascribed to the common suppression of a finite set of pathways to support core functions such as anabolism, catabolism, and redox balance (Figs 2 and 5)38. The general repression of these pathways may reflect their regulation by signaling pathways and cellular metabolism, which were perturbed in both PH3–R strains (Fig. 5). This conjecture is further supported by the proteome profiles of PH3–R strains, which indicated that highly abundant proteins such as troponin, Hsp90, and mutated ND4 in the PH3–R strains of S. oryzae are closely associated with the capacity to transition between aerobic and anaerobic respiration, in relation to the modification of OXPHOS (Fig. 2 and Tables 2 and 3). An anaplerotic flux in muscle tissues appears to be activated in the MR strain, whereas the SR strain clearly exhibited a distinct means to induce Ca2+-mediated muscle contraction by actively pumping Ca2+ back into the sarcoplasmic reticulum (Fig. 5). It is known that troponin C plays a key role in cardiac muscle contraction39 by controlling Ca2+-mediated regulation of interactions between actin and myosin. In this regard, our proteomic data suggest that the SR strain possesses a mechanism for muscle contraction regulated by troponin C, which is a highly selective marker for myocardial infarction and heart muscle cell death in human40. Taken together, the MR strain of S. oryzae minimized the metabolic burden by bypassing ATP consuming pathway, whereas the SR strain reorganized the energy transduction system to seemingly undertake anaerobic respiration, regardless of the active COX, during PH3 fumigation. This change in energy production confers a respiratory advantage against PH3 fumigation. It is likely that mutations in ND subunits abolish OXPHOS to avoid the acute toxicity of PH3. Thus, if impaired mitochondrial energy transducing activities benefit the PH3 resistance of S. oryzae, some of them may be suitable as therapeutic targets. Indeed, mitochondrial mutations affecting OXPHOS efficacy confers drug resistance in malaria parasites41,42.

Figure 5
figure 5

Reprogrammed metabolic pathways in the PH3–R strains of S. oryzae. GPI, glucose-6-phosphate isomerase; PFK1, 6-phosphofructokinase; FBA, fructose-bisphosphate aldolase; MCAD, putative medium-chain specific acyl-CoA dehydrogenase; ACC, acetyl-CoA carboxylase; ACS, ATP-citrate synthase; SDH, succinate dehydrogenase; OGDH, 2-oxoglutarate dehydrogenase E1; Hsp 60 and 90, heat shock protein 60 and 90; groES, chaperonin groES; groEL chaperonin groEL; AAC, ADP/ATP carrier protein; SERCA, the sarcoplasmic/endoplasmic reticulum Ca2+-ATPase; RyR, ryanodine receptor; AK, adenylate kinase; NDPK, nucleoside-diphosphate kinase; PME, pectin methylesterase; ACP1, actin-interacting protein; MSP20, Muscle-specific protein 20.

Full size image

This plausible PH3 resistance mechanism was further reinforced by mtDNA sequencing. In the SR strain, several missense mutations in nad4, nad5, and nad6 genes encoding ND subunits and silent mutations in cox1 and cox2 genes were identified (Table 2). The structural analysis of these mutation sites suggests that Ala73 in ND4 of S. oryzae plays an important role in proton uptake, indicating that the Ala73Glu substitution is responsible for altering the proton translocation efficiency (Fig. 4b). Therefore, the m.9082 C > A mutation has a drastic effect on structure and function of the mitochondrial ND, suggesting that the mitochondrial ND of the SR strain is strongly associated with PH3 resistance through muscle contraction by Ca2+ pumping in response to PH3 fumigation. Moreover, proteins involved in stress response, biosynthesis, transport, and signaling were also differentially expressed in the MR and SR strains (Fig. 2), implying that PH3 fumigation functions as a major selection pressure to change the cellular metabolism in rice weevil. This is also supported by the proteomic results of COX2 expression, which was approximately 2-fold lower in the MR strain than in the susceptible WT strain, which is consistent with the qRT-PCR data (Fig. 3 and Supplementary Table S3). However, these data suggest that the inhibition of COX activity observed in the PH3–R strains is not well correlated with the acquisition of PH3 resistance (Fig. 1c,d), implying either the presence of various COX isozymes or an impaired OXPHOS system in PH3–R strains caused by mtDNA mutations. Furthermore, the high abundance of Hsp90 and groEL/ES, another characteristic marker of cancer cells, likely explains the development of PH3 resistance in S. oryzae strains (Fig. 2 and Supplementary Table S3). To our knowledge, this is the first report of mutations in ND subunits of insect pests that can be a resistant factor to the redox-active gas, phosphine. Therefore, to effectively manage PH3–R insect pests, fumigants that target proteins other than those involved in mitochondrial energy production should be considered.

Conclusions

Both PH3–R strains exhibited higher resistance to EF-mediated inhibition of COX than the WT, whereas the disinfectant Ct of the SR strain was much longer than that of the MR strain. Unlike the MR strain, which primarily showed changes in the expression levels of genes encoding metabolic enzymes involved in catabolic pathways that minimize metabolic burden, the SR strain showed changes in the mitochondrial respiratory chain. We found that the acquisition of strong PH3 resistance necessitates the avoidance of OXPHOS via the introduction of a few non-synonymous mutations in mitochondrial genes encoding ND as well as nuclear genes encoding DLD, concomitant with metabolic reprogramming, a recognized hallmark of cancer metabolism. These results suggest that anaerobic respiration is the survival strategy that SR insect pests use to compensate for minimized energy transduction under anoxic conditions. Taken together, PH3 toxicity acts as a selection pressure that not only alters cellular metabolism but also modulates energy transduction via mitochondrial mutations; this explains the mechanism of PH3 resistance in S. oryzae.

Methods

Insect strains and growth conditions

The susceptible WT strain of S. oryzae was obtained from Murdoch University (Perth, Australia) and maintained under pesticide-free conditions. The MR strain was obtained from the central grain storage (JaeHee RPC, Gunsan, Korea) in 2016. The SR strain was collected from Chungbuk National University (Cheongju, Korea) in 2015. All stock colonies of S. oryzae were successively cultured on rice grains at the Plant Quarantine Technology Center in Korea under controlled conditions (25 ± 1 °C temperature, 80% relative humidity, and 16 h light/8 h dark photoperiod).

Fumigation assay

Adults of S. oryzae were placed on brown rice in a plastic dish (Φ 10 cm × 4 cm; SPL Life Science, Pocheon, Korea) with a center-opened cap covered by nylon net. The toxicity of PH3 against S. oryzae was tested using a series of concentrations from 0.01 to 1.0 mg/L in 12 L desiccators (Bibby Scientific, Staffordshire, UK) sealed with glass stoppers for 20 h at 20 °C. PH3 (ECO2Fume™; 2% PH3 + 98% CO2) was obtained from Cytec (Sydney, Australia). All experiments were performed with 30 insects in triplicate. The desiccator was furnished with a lid fitted with a septum injection system (Alltech Crop Science, Nicholasville, KY). Each desiccator was measured for its volume prior to fumigation bioassay by weighing the amount of water at 20 °C. A magnetic bar placed at the bottom of the desiccator was used to properly stir the gas for even distribution. To determine residual concentrations of PH3 in the desiccator, a gas was sampled at 10 min, 1 h, 3 h, 6 h, and 20 h post PH3 fumigation and stored in a gas sampling bag (1-L Tedlar®, SKC, Dorset, United Kingdom). Gas chromatography (GC) analysis performed using an Agilent GC 7890 A coupled with a flame photometric detector (FPD) and a HP-PLOT/Q column (30 m length × 530 µm internal diameter × 40 µm film) (Agilent, Santa Clara, CA). Detailed condition of GC analysis can be found in Supplementary Methods.

Determination of the Ct value for PH3

The concentrations of PH3 measured during the exposure periods were used to determine the Ct values, as described in43. Detailed equation of Ct values can be found in Supplementary Methods.

Measurement of enzyme activities

Protein samples were isolated from three independent replicates (n = 100) and were performed according to the method described in Supplementary Methods. Activities of COX, AchE, CE, and GST were determined using the methods reported previously by Nathanailides and Tyler44, Ellman et al.45, Mackness et al.46, and Habig and Jakoby47, respectively. Each assay was performed in triplicate. Enzyme activities were expressed as units/mg protein. Data were expressed as mean ± standard deviation. The results were analyzed by one-way analysis of variance (ANOVA) and Tukey’s post-hoc test using SPSS statistics version 23.0.

EF inhibition kinetics of COX

To study the inhibitory effect of EF on COX in different PH3–R strains of S. oryzae, the mitochondrial fraction of each insect was exposed to 0, 1, 5, 10, 50, and 75 mM EF. The activity of COX was measured as described above. The IC50 value and the inhibition constant (Ki) of EF was calculated by least-squares fit dose-response curves and enzyme kinetics-inhibition modes using GraphPad Prism version 8.0.1 for Windows (La Jolla, CA). Reduced cytochrome c was used as a substrate for COX in each strain at different concentration (0.011, 0.015, 0.022, 0.044 and 0.055 mM) in the presence of 0, 1, and 10 mM EF.

Proteomic analysis using nLC-ESI-MS/MS

Proteomic analysis was performed using a Thermo Scientific Q Exactive Hybrid Quadrupole-Orbitrap instrument (Thermo Fisher Scientific Inc., Waltham, MA) with a Dionex U 3000 RSLC nano high performance liquid chromatography (HPLC) system. An ESI source fitted with a fused silica emitter tip (New Objective, Woburn, MA) was employed with a mobile phase consisting of the water:acetonitrile (98:2 [v/v]) containing 0.1% formic acid. The trypsin-treated samples were trapped on an Acclaim PepMap 100 trap column (100 μm × 2 cm, nanoViper C18, 5 μm, 100 Å) and washed for 6 min at a flow rate of 4 μL/min, and then separated on an Acclaim PepMap 100 capillary column (75 μm × 15 cm, nanoViper C18, 3 μm, 100 Å) at a flow rate of 300 μL/min. The resulting peptides were electrosprayed through a coated silica tip with ion spray voltage of 2,000 eV. Mass data were collected and analyzed using Proteome Discoverer 1.4, MaxQuant 1.6, and Scaffold 4.8.4 against the protein databases of S. oryzae and T. castaneum. Detailed information including analysis conditions and data processing can be found in Supplementary Methods.

The expression of 18 genes expected to be associated with PH3 resistance from proteome analysis were validated by qRT-PCR. The method of qRT-PCR was in Supplementary Methods.

mtDNA sequencing and annotation

Total DNA, including mtDNA, was extracted from 30 individuals of each strain using the QIAamp DNA Mini Kit (Qiagen, Dusseldorf, Germany). Short-read assembly was performed using SOAPdenovo48, and scaffolding was performed with a minimum size of 100 bp. The assembled scaffolds and contigs (≥100 bp) were mapped to the sequences of Sitophilus oryzae using the National Center for Biotechnology Information (NCBI) BLASTN tool with default parameters. Contigs and scaffolds with query coverage greater than 40% were retrieved and used to search the non-redundant nucleotide and protein databases using BLASTN (http://blast.ncbi.nlm.nih.gov/). All raw reads were realigned with the assembled sequences using the Burrows-Wheeler Aligner (BWA) software. The aligned paired-end reads were used to determine the sequencing depth. A second round of assembly was carried out using the initially assembled contigs and scaffolds. Contigs and scaffolds with an overlap of ≥12 bp were assembled into larger scaffolds based on synteny between the assembled and reference genomes, we further joined them into larger scaffolds.

The mtDNA sequence was annotated using the MITOS web server49. Nucleotide sequences of protein-coding genes were translated into amino acid sequences using the genetic code for invertebrate mitogenomes. The predictions from MITOS were manually curated using other published skipper mitogenomes as references, and the starts and ends of genes were modified, if necessary, to be consistent with other species. The new open reading frames of the protein-coding genes (after modification) were validated.

Structural mapping and analysis

Three non-synonymous mutations in mt genes encoding the ND subunits were mapped onto their corresponding rice weevil and bacterial model structures. Each subunit model, comprising S. oryzae ND, was constructed using the SWISS-MODEL server with the available high-resolution ND subunit structures of the swine (PDB accession code 5GUP for ND1 and ND3), bovine (5LC5 for ND2 and ND4L), murine (6G2J for ND4 and ND5), and bacterial (4HEA for ND6) complex I as template structures.

To determine the effect gene mutations on the structure and function of ND subunits, amino acid substitutions (AAS) analysis was performed using predictive approaches such as SIFT (http://sift.jcvi.org), PROVEAN (http://provean.jcvi.org), and Polyphen-2 (http://genetics.bwh.harvard.edu/pph/) web servers.