Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Bioaccumulation of Cry1Ab Protein from an Herbivore Reduces Anti-Oxidant Enzyme Activities in Two Spider Species

  • Ji Zhou ,

    Contributed equally to this work with: Ji Zhou, Kaifu Xiao, Baoyang Wei

    Affiliation College of Bioscience & Biotechnology, Hunan Agriculture University, Changsha, China

  • Kaifu Xiao ,

    Contributed equally to this work with: Ji Zhou, Kaifu Xiao, Baoyang Wei

    Affiliation College of Bioscience & Biotechnology, Hunan Agriculture University, Changsha, China

  • Baoyang Wei ,

    Contributed equally to this work with: Ji Zhou, Kaifu Xiao, Baoyang Wei

    Affiliation College of Bioscience & Biotechnology, Hunan Agriculture University, Changsha, China

  • Zhi Wang ,

    wangzhispider@hotmail.com (ZW); SongQ@missouri.edu (QSS)

    Affiliation College of Bioscience & Biotechnology, Hunan Agriculture University, Changsha, China

  • Yun Tian,

    Affiliation College of Bioscience & Biotechnology, Hunan Agriculture University, Changsha, China

  • Yixing Tian,

    Affiliation College of Bioscience & Biotechnology, Hunan Agriculture University, Changsha, China

  • Qisheng Song

    wangzhispider@hotmail.com (ZW); SongQ@missouri.edu (QSS)

    Affiliations College of Bioscience & Biotechnology, Hunan Agriculture University, Changsha, China, Division of Plant Sciences, University of Missouri, Columbia, Missouri, United States of America

Abstract

Cry proteins are expressed in rice lines for lepidopteran pest control. These proteins can be transferred from transgenic rice plants to non-target arthropods, including planthoppers and then to a predatory spider. Movement of Cry proteins through food webs may reduce fitness of non-target arthropods, although recent publications indicated no serious changes in non-target populations. Nonetheless, Cry protein intoxication influences gene expression in Cry-sensitive insects. We posed the hypothesis that Cry protein intoxication influences enzyme activities in spiders acting in tri-trophic food webs. Here we report on the outcomes of experiments designed to test our hypothesis with two spider species. We demonstrated that the movement of CryAb protein from Drosophila culture medium into fruit flies maintained on the CryAb containing medium and from the flies to the spiders Ummeliata insecticeps and Pardosa pseudoannulata. We also show that the activities of three key metabolic enzymes, acetylcholine esterase (AchE), glutathione peroxidase (GSH-Px), and superoxide dismutase (SOD) were significantly influenced in the spiders after feeding on Cry1Ab-containing fruit flies. We infer from these data that Cry proteins originating in transgenic crops impacts non-target arthropods at the physiological and biochemical levels, which may be one mechanism of Cry protein-related reductions in fitness of non-target beneficial predators.

Introduction

Transgenic rice lines expressing Cry protein-encoding genes from the bacterium Bacillus thuringiensis (Bt) have recently been developed to control the major lepidopteran pests, the stem borers and leaffolders in China [1], in wake of environmental contamination and pest outbreak due to the continuous use and overuse of broad spectrum pesticides over the past several decades [2]. Despite the development of at least 8 transgenic rice lines expressing toxin [3], [4], [5], [6], [7], [8], [9], no Bt rice lines have yet been approved for commercial release in China. However, one cultivar (Huahui 1) and its hybrid line (Shanyou 63), expressing Cry1Ab and Cry1Ac for resistance to lepidoperans, have been granted a limited trial in farmlands in Hubei Province for the five-year period 2009 through 2014 [10].

While transgenic rice offers tremendous benefits to agricultural production and the environment, substantial risks are associated with the deployment of Bt rice lines. One concern is the potential movements of Bt toxins through natural food webs, including non-target species, in rice cropping systems. Bernal et al. (2002) investigated five transgenic rice lines for movement of Cry proteins from rice to brown planthoppers (non-target herbivore) and to its predator, the plant bug, Cyrtorhinus lividipennis [11]. While they detected the Cry proteins in the planthopper honeydew and in the predator, they concluded that exposure to Bt proteins might not influence the fitness of either insect species. Li et al. (2007) came to a similar conclusion based on the outcomes of a three-year study of arthropod communities in rice paddy fields [12]. They reported substantial similarity between arthropod communities in experimental Bt rice crops and control non-Bt rice crops. Another field study, using three sites in Zhejiang Province, considered three planthopper species and their predator, C. lividipennis [13]. Again, their two-year program recorded no differences in plant hopper populations, with Sogatella furcifea remaining the predominant planthopper in experimental and control plots. The Bt rice was not deleterious to the predator populations and the authors inferred that Bt rice may control the targeted lepidopteran pests while not influencing an important non-target species, the planthopper biocontrol agent, C. lividipennis. With respect to planthoppers, Chen et al. (2003a) reported that the transgenic lines TT9-3 and TT9-4, expressing a fused Cry1Ab-Cry1Ac gene, reduced brown planthopper feeding and oviposition behaviors, relative to the non-transgenic parental line, IR72 [14]. They inferred that the transgenic rice lines reduced planthopper fitness. Similarly, they reported that transgenic rice lines influenced dispersal of planthoppers and leafhoppers in field conditions [15].

Cry proteins are certainly transferred from rice plants to planthoppers, Nilaparvata lugens, and on to one of its predators, the wolf spider, Pirata subpiraticus [16]. Spiders are generalist predators and the idea that spiders can reduce populations of pest insects is well established [17], [18]. Pardosa pseudoannulata and Ummeliata insecticeps are the dominant spider species in Chinese farmlands. Tian et al. (2012) demonstrated in field trials the movement of a Bt protoxin, Cry1Ab, from the Bt rice lines KMD1 and KMD2 through the brown planthopper, N. lugens to the ground spider P. pseudoannulata and showed that the Cry1Ab expressing rice did not influence the spider's fitness as determined by several parameters [19]. Alternatively, Chen et al. (2003b) suggested that Bt rice lines reduced the fitness of at least some non-target species, raising questions on the sub-lethal impacts of Bt rice at the level of physiology and biochemistry [15]. Oppert et al. (2012) reported on changes in transcriptome expression during Bt intoxication in larvae of Cry3Aa-sensitive beetles, Tenebrio molitor [20]. Among many changes in transcript expression, they reported down-regulation of transcripts encoding antioxidant enzymes, including superoxide dismutase, glutathione S transferase and catalase. We posed the hypothesis that the dietary ingestion of Bt protein could also influence enzyme activities in spiders acting in tri-trophic food webs. Here we report on the outcomes of experiments designed to test our hypothesis, using the spiders P. pseudoannulata and U. insecticeps as the test animals.

Materials and Methods

Drosophila processing

Routine Drosophila culture medium consists of cane sugar (124 g), corn starch (165 g), agar (124 g), and yeast powder (2 g) dissolved in 1 liter distilled water. The routine (or control) medium is prepared by mixing the components, boiling the mixture, cooling it down to 60°C and finally transferring aliquots of selected size to culture bottles. Experimental medium containing a final concentration Cry1Ab at 100 ng/ml (a predetermined effective dose to ensure sufficient Bt protein in medium for development of each fly from larva to adult in the bottle with about 300–600 flies), was prepared by adding the Cry1Ab protein (purchased from Shanghai YouLong Biotechnology Co.) into the routine medium after the temperature of the newly prepared medium was below 50°C.

Drosophila melanogaster (obtained from the genetics laboratory, Hunan Agricultural University) was reared on control or experimental media for the indicated times as noted in Results. To analyze movement of Cry1Ab protein from flies to spiders, separate groups of control and experimental flies were maintained on control or experimental media for 10 days, and then used in spider feeding experiments.

Spider cultures

P. pseudoannulata and U. insecticeps were collected from the experimental farmland in Hunan Academy of Agricultural Sciences. No specific permissions were required for a limited number of spiders collected from this location because the sample collection did not involve endangered or protected species. Individual spiders (both species) were held in glass tubes (12×100 mm) maintained in the laboratory at room temperature and natural photoperiod. The spiders were provided moist cotton balls for water and supplied with 14 control or experimental flies per day.

Estimating quantities of Cry1Ab protein in flies and spiders

Cry1Ab quantities in flies and spiders were determined using the Bt-Cry1Ab/Ac ELISA kits (American EnviroLogix, Inc., Portland, ME USA, distributed by Shanghai YouLong Biotech Co.). After recording wet weights, spiders (13) and flies (26) were homogenized in 1 ml PBS buffer using glass homogenizers (2 ml). The homogenates were centrifuged at 16,000g for 10 min at 4°C. One aliquot (0.25 ml) of each supernatant was used to determine protein concentrations via the Bradford assay (Bradford, 1976) [21] against BSA as a quantitative standard and a second aliquot (0.1 ml) was used to estimate Cry1Ab protein concentrations. Each assay used 26 flies and 13 mature spiders. Three biological replicated were performed for each assay.

Enzyme assays

Spiders were maintained on experimental or control flies for 3 to 11 days. Spiders (13/treatment) were homogenized in 1 ml PBS then centrifuged at 16,000 g for 10 min at 4°C. Several aliquots were taken from the resulting supernatant. One aliquot (0.25 ml) was used for protein determination, another (0.1 ml) for Cry1Ab determination, a third (0.1 ml) to assay acetylcholine esterase (AChE; EC 3.1.1.7) activity following the method of Gorun (1978) [22], a fourth to assay glutathioine peroxidase (GSH-Px; EC 1.11.1.9) activity, and a final aliquot to assay superoxide dismutase (SOD; EC 1.15.1.1) activity. GSH-Px and SOD activities were determined using the commercial enzyme activities kits (Chinese Nanjing Jiancheng Bioengineering Institute, Nanjing, China). Three biological replicated were performed for each assay.

Data analysis

The data were analyzed using the SPSS 19.0 version of Wilcoxon signed-rank test. Significant differences at p<0.05 or p<0.01 were designated with * or ** respectively.

Results

Cry1Ab protoxin moved from the fruit fly culture medium to fruit flies and spiders

Figure 1 displays the accumulation and clearance of the Cry1Ab protein by fruit fly tissues. The Cry1Ab protein was not detected in fruit flies reared on control media. Flies reared on Cry1Ab-supplemented diets accumulated the Cry1Ab protein over the experimental feeding regimen, from about 0.6 ng Cry1Ab/mg protein on day 1 to almost 5 ng Cry1Ab/mg protein on day 10. Cry1Ab protein concentrations declined to just over 1 ng/mg by day 15.

thumbnail
Figure 1. Cry1Ab protein concentrations in D. melanogaster at the indicated feeding times.

The data points represent the mean ± SE of three independent biological replicates. All points representing experimental fruit flies are significantly different from points representing controls.

https://doi.org/10.1371/journal.pone.0084724.g001

Figure 2 shows that both spider species accumulated the Cry1Ab protein from fruit flies maintained on experimental culture media, although the patterns of accumulation differed between the two species. U. insecticeps accumulated Cry1Ab protein at about 7 ng/mg on day 3, then declined steadily over the following six days to about 4 ng/mg and remained constant over the next two days. The pattern of Cry1Ab accumulation was different in P. pseudoannulata (Fig. 2B) and these spiders accumulated about 10 ng Cry1Ab/mg protein after three days feeding on experimental fruit flies, increased to about 18 ng/mg protein by day 7, and then declined over the next four days to about the day 3 value.

thumbnail
Figure 2. Cry1Ab protein concentrations in, Panel A, U. insecticeps and, Panel B, P. pseudoannulata at the indicated times.

The data points represent the mean ± SE of three biological replicates. All points representing experimental spiders are significantly different from points representing controls.

https://doi.org/10.1371/journal.pone.0084724.g002

Dietary Cry1Ab influenced enzyme activities in spiders fed on experimental fruit flies

Fruit fly-derived Cry1Ab protein did not influence AChE activity in U. insecticeps until day 7 of the experimental feeding regimen. For days 7, 9 and 11, AChE activity was reduced, relative to controls, in the experimental spiders (Fig. 3A). Again, the pattern differed for P. pseudoannulata, in which AChE activity was approximate 2-fold higher in experimental spiders compared to controls on day 3, then significantly lower in the experimentals by day 7. AChE activities did not differ between the experimentals and controls on the other days.

thumbnail
Figure 3. AChE activity in, Panel A, U. insecticeps and, Panel B, P. pseudoannulata at the indicated time periods.

Each data point reflects the mean ± SE of three biological replicates. The significant differences (p<0.05) are indicated by an asterisk. The most significant differences (p<0.01) are indicated by two asterisks.

https://doi.org/10.1371/journal.pone.0084724.g003

Dietary Cry1Ab protein increased GSH-Px activities in both spider species (Fig. 4). For U. insecticeps, GSH-Px activities were about 5-fold higher than control spiders by day 3 of the experimental feeding regimen, then declined to control values by day 9 (Fig. 4A). GSH-Px activities were higher than controls throughout the feeding period in P. pseudoannulata (Fig. 4B).

thumbnail
Figure 4. GSH-Px activity in, Panel A, U. insecticeps and, Panel B, P. pseudoannulata at the indicated times.

Each data point reflects the mean ± SE of three biological replicates. The significant differences (p<0.05) are indicated by an asterisk. The most significant differences (p<0.01) are indicated by two asterisks.

https://doi.org/10.1371/journal.pone.0084724.g004

SOD activities were lower, compared to controls, in experimental spiders throughout the feeding experiment, with the exception of day 11 in U. insecticeps (Fig. 5). The Y-axis scales differ between Fig. 5A and 5B: P. pseudoannulata expressed higher SOD activities in experimentals and controls, compared to U. insecticeps.

thumbnail
Figure 5. SOD activity in, Panel A, U. insecticeps and, Panel B, P. pseudoannulata at the indicated times.

Each data point reflects the mean ± SE of three biological replicates. The significant differences (p<0.05) are indicated by an asterisk. The most significant differences (p<0.01) are indicated by two asterisks.

https://doi.org/10.1371/journal.pone.0084724.g005

Discussion

The data presented in this paper strongly support our hypothesis that intoxication with Bt protoxins influences enzyme activities in spiders acting in tri-trophic food webs. Our data show that fruit flies reared on experimental media amended with Cry1Ab protein accumulated Cry1Ab protein from the culture media and that the predatory spiders accumulated Cry1Ab from their experimentally-reared prey. The activities of three enzymes, AChE, GSH-Px and SOD, were influenced, relative to controls, in all experimental treatments. Taken together, these findings bolster the idea that Bt protoxins can move through multi-trophic food webs and can influence the biochemistry and physiology of non-target species.

We selected AChE, GSH-Px and SOD for this study on the basis of their physiological significance in arthropod biology. Acetylcholine is a neurotransmitter that expresses its actions through a family of receptors, all ligand-gated ion channels [23]. Some insecticides, the organophosphates and carbamates, act by inhibiting AChE activity, which is necessary for normal physiology of nervous systems. Dietary Cry1Ab protein altered AChE activity in both spider species in this study and may reduce their fitness by interfering with any physiological system based on normal nervous system function, including behavioral functions such as reproductive behavior.

SOD and GSH-Px are anti-oxidant enzymes, responsible for protecting cells and cell components from oxidative damage. Our data show that the presence of Cry1Ab protein in the food stream led to decreased SOD activity in both spider species. This is contrary to the findings of Liu et al. (2010) [24]. These authors maintained rice leafrollers, Cnaphalocrocis medinalis, on Cry1Ab-expressing rice for 24 h, then allowed spiders P. pseudoannulata, to consume the leafrollers. This led to increased SOD activity in the spiders. Again, alterations in the activities of these enzymes could be deleterious to long-term fitness of the spiders. In particular, experiments based on the oxidative stress model of aging revealed that overexpressing SOD led to extended lifespans in Drosophila [25]. The influence of SOD deficiency, however, may be subtler than a simple effect of accumulating oxidative damage. Rogina and Helfand (2000) showed that SOD-deficient fruit flies experienced acceleration in the normal age-related pattern of expression of a single marker gene, wingless [26]. They concluded that the attenuated life span associated with SOD deficiency was due to increased rate of aging, rather than a pathological oxidative process. Nonetheless, SOD function appears to impact lifespans in animals. We note dietary Cry1Ab exerted opposite effects on GSH-Px and SOD, increasing activity of the former and decreasing activity of the latter. In Drosophila, an increase in GSH-Px activity is detrimental when combined with a decrease in Sod1 activity, suggesting the two enzymes have opposing functions in vivo [27]. These differences may relate to the separate cellular functions of SOD and GSH-Px. Although superoxide is formed by many reactions within cells, the mitochondrial electron transport chain is a major source of superoxide, that is, it is a normal product of oxygen metabolism and SOD is highly expressed in mitochondria [28]. GSH-Px generally acts to reduce lipid hydroperoxides, most of which form in the lipid fractions of cells. Hence, these two anti-oxidant enzymes largely occur in different fractions of cells. This may relate to the opposite effects of dietary Cry1Ab on SOD and GSH-Px.

It is not clear how dietary Cry1Ab protein influenced the activity of the three enzymes in this study. The changes were seen at the time scale of several days. At this lengthy scale, Cry1Ab protein could act at any level of protein functioning, including expression of genes, post-translation modifications or direct interactions with the proteins. It is also unclear where Cry1Ab exerted its actions. All experiments were performed by homogenizing whole animals, leaving important questions open. Did the fruit flies and spiders actually accumulate Cry1Ab protein, or was the material simply present in the alimentary canals? How much, if any, of Cry1Ab protein moved from the alimentary canals into the body proper? Figures 1 and 2 show that the Cry1Ab is at least partially cleared from the fly and spider bodies after 10 day exposure for the fly and 6 or 7 day exposure for spiders. Is this due to protein metabolism or to simple excretion? As often happens in newly-opened research corridors, our data on the effects of Bt protoxins on the biochemistry and physiology of non-target arthropods prompts meaningful new research directions. One of which is to identify other biochemical and physiological functions that are impacted by Bt proteins and another is to reconcile the observed effects of dietary Cry1Ab on three enzymes and a large body of literature showing little effects of Bt toxin-expressing rice on rice agroecosystems.

Author Contributions

Conceived and designed the experiments: ZW QSS. Performed the experiments: JZ KFX BYW. Analyzed the data: JZ KFX BYW. Contributed reagents/materials/analysis tools: YT JZ. Wrote the paper: ZW YXT QSS.

References

  1. 1. Chen M, Shelton A, Ye GY (2011) Insect-resistant genetically modified rice in China: from research to commercialization. Annual Review of Entomology 56: 81–101.
  2. 2. Sogawa K, Zhang H, Yang XJ, Liu GJ (2003) Whitebacked planthopper resistance in China rice varieties. Chinese Journal of Rice Science 17(17): 47–52.
  3. 3. Bashir K, Husnain T, Fatima T, Latif Z, Mehdi SA, et al. (2004) Field evaluation and risk assessment of transgenic Indica basmati rice. Molecular Breeding 13(4): 301–312.
  4. 4. Breitler JC, Vassal JM, Catala MD, Meynard D, Marfa V, et al. (2004) Bt rice harbouring Cry genes controlled by a constitutive or wound-inducible promoter:protection and transgene expression under Mediterranean field conditions. Plant Biotechnology Journa l2(5): 417–430.
  5. 5. Chen H, Mang G, Zhang QF, Lin YJ (2008) Effect of transgenic Bacillus thuringiensis rice lines on mortality and feeding behavior of rice stem borers (Lepidoptera:Crambidae). Journal of Economic Entomology 101(1): 182–189.
  6. 6. Riaz N, Husnain T, Fatima T, Makhdoom R, Bashir K, et al. (2006) Development of Indica Basmati rice harboring two insecticidal genes for sustainable resistance against lepidopteran insects. South African Journal of Botany 72(2): 217–223.
  7. 7. Ye GY, Shu QY, Yao HW, Cui HR, Cheng XY, et al. (2001) Field evaluation of resistance of transgenic rice containing a synthetic cry1Ab gene from Bacillus thuringiensis Berliner to two stem borers. Journal of Economic Entomology 94(1): 271–276.
  8. 8. Ye RJ, Huang HQ, Yang Z, Chen TY, Liu L, et al. (2009) Development of insect-resistant transgenic rice with Cry1C*-free endosperm. Pest Management Science 65(9): 1015–1020.
  9. 9. Zaidi MA, Ye GY, Yao HW, You TH, Loit E, et al. (2009) Transgenic rice plants expressing a modified cry1Ca1 gene are resistant to Spodoptera litura and Chilo suppressalis. Molecular Biotechnology 43(3): 232–242.
  10. 10. Ministry of Agriculture of the People's Republic of China (MAPRC) (2009) The second list of approval agricultural genetically modified organisms' safety certificates in 2009.
  11. 11. Bernal CC, Aguda RM, Cohen MB (2002) Effect of rice lines transformed with Bacillus thuringiensis toxin genes on the brown planthopper and its predator Cyrtorhinus lividipennis. Entomologia Experimentalis et Applicata 102(1): 21–28.
  12. 12. Li FF, Ye GY, Wu Q, Peng YF, Chen XX (2007) Arthropod abundance and diversity in Bt and non-Bt rice fields. Environmental Entomology 36(3): 646–654.
  13. 13. Chen M, Liu ZC, Ye GY, Shen ZC, Hu C, et al. (2007) Impacts of transgenic cry1Ab rice on non-target planthoppers and their main predator Cyrtorhinus lividipennis (Hemiptera: Miridae)—A case study of the compatibility of Bt rice with biological control. Biological Control 42(2): 242–250.
  14. 14. Chen M, Ye GY, Hu C, Datta SK (2003a) Effects of transgenic Bt Indica rice on the feeding and oviposition behavior of the brown planthopper, Nilaparvata lugens. Acta Phytophylacica Sinica 30(4): 365–370.
  15. 15. Chen M, Ye GY, Hu C, Tu J, Datta SK (2003b) Effect of transgenic Bt rice on dispersal of planthoppers and leafhoppers as well as their egg parasitic wasp. Journal of Zhejiang University (Agriculture & Life Sciences) 29(1): 29–33.
  16. 16. Chen M, Ye GY, Lu XM, Hu C, Peng YF, et al. (2005) Biotransfer and bioaccumulation of Cry1Ab insecticidal protein in rice plant—brown planthopper—wolf spider food chain. Acta Entomologica Sinica 48(2): 208–213.
  17. 17. Hoefler CD, Chen A, Jakob EM (2006) The potential of a jumping spider, Phidippus clarus, as a biological control agent. Journal of Economic Entomology 99(2): 432–436.
  18. 18. Sigsgaaard L (2007) Early season natural control of the brown planthopper, Nilaparvata lugens: the contribution and interaction of two spider species and a predatory bug. Bulletin Entomological Research 97: 533–544.
  19. 19. Tian JC, Chen Y, Li ZL, Li K, Chen M, et al. (2012) Transgenic Cry1Ab rice does not impact ecological fitness and predation of a generalist spider. PLOS One (4): e35164.
  20. 20. Oppert B, Dowd SE, Bouffard P, Li L, Conesa A, et al. (2012) Transcriptome profiling of the intoxication response of Tenebrio molitor larvae to Bacillus thuringiensis Cry3Aa protoxin. PLOS One (7)4: e34624.
  21. 21. Bradford M (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Analytical Biochemistry 72(7): 248–254.
  22. 22. Gorun V, Proinov I, Baltescu V, Balaban G, Barzu O (1978) Modified Ellman procedure for assay of cholinesterases in crude enzymatic preparations. Analytical Biochemistry 86(1): 324–6.
  23. 23. Jones AK, Sattelle DB (2010) Diversity of insect nicotinic acetylcholine receptor subunits. Advances in Experimental Medicine and Biology 683: 25–43.
  24. 24. Liu LJ, Yan HM (2010) Effects of Bt protein expressed by transgenic rice on the activities of protective enzymes in Wolf spider (Pardosa pseudoannulata). Acta Arachnologica Sinica 19(1): 47–51.
  25. 25. Orr WC, Sohal RS (1994) Extension of life-span by overexpression of superoxide dismutase and catalase in Drosophila melanogaster. Science 263: 1128–1130.
  26. 26. Rogina B, Helfand SL (2000) Cu, Zn superoxide dismutase deficiency accelerates the time course of an age-related marker in Drosophila melanogaster. Biogerontology 1(2): 163–169.
  27. 27. Missrirlis F, Phillips JP, Jackle H, Rouault TA (2003) Drosophila and antioxidant therapy design. In: Galaris D, editor. Free Radicals and Oxidative Stress: Chemistry, Biochemistry and Pathophysiological Implications.Ioannina, Greece. pp. 147–151.
  28. 28. Nishikawa T, Edelstein D, Du XL, Yamagishi SI, Matsumura T, et al. (2000) Normalizing mitochondrial superoxide production blocks three pathways of hyperglycaemic damage. Nature 404: 787–790.