Energy Use Efficiency Is Characterized by an Epigenetic Component That Can Be Directed Through Artificial Selection to Increase Yield

Energy use efficiency is characterized by an epigenetic component that can be directed through artificial selection to increase yield

Miriam Haubena, Boris Haesendonckxa, Evi Standaerta, Katrien Van Der Kelenb,c, Abdelkrim Azmid, Herve´ Akpoa, Frank Van Breusegemb,c, Yves Guisezd, Marc Botsa, Bart Lamberta, Benjamin Lagaa, and Marc De Blocka,1

aBayer BioScience N.V., 9052 Gent, Belgium; bDepartment of Plant Systems Biology, Flanders Institute for Biotechnology (VIB), 9052 Gent, Belgium; cDepartment of Plant Biotechnology and Genetics, Ghent University, 9052 Gent, Belgium; and dLaboratory for Plant Biochemistry and Physiology, Department of Biology, University of Antwerp, 2020 Antwerpen, Belgium

Edited by Marc C. E. Van Montagu, Ghent University, Ghent, Belgium, and approved September 29, 2009 (received for review August 4, 2009) Quantitative traits, such as size and weight in animals and seed For a long time, the epigenome has been believed to reset yield in plants, are distributed normally, even within a population during meiosis. However, in numerous cases, epigenetic changes of genetically identical individuals. For example, in plants, various can be maintained during many generations. In plants, the factors, such as local soil quality, microclimate, and sowing depth, flower-specific hypermethylated superman and agamous epial- affect growth differences among individual plants of isogenic leles and the disease resistance hypomethylated late-flowering populations. Besides these physical factors, also epigenetic com- and ball epialleles are well-known (5–7). In fact, the transgen- ponents contribute to differences in growth and yield. The net- erational epigenetic inheritance holds true for many taxa, in- work that regulates crop yield is still not well understood. Al- cluding animals, fungi, and bacteria (8). Recently, in Arabidopsis, though this network is expected to have epigenetic elements, it is epialleles across the genome were shown to be stably transmitted completely unclear whether it would be possible to shape the over consecutive generations (9). epigenome to increase crop yield. Here we show that energy use One of the most complex quantitative traits in plants is yield efficiency is an important factor in determining seed yield in canola because it is determined by many interacting components. (Brassica napus) and that it can be selected artificially through an Although a major subject in plant research for many decennia, epigenetic feature. From an isogenic canola population of which the understanding of the biological network controlling yield is the individual plants and their self-fertilized progenies were re- only poorly understood. Here we show that in canola (Brassica cursively selected for respiration intensity, populations with dis- napus), energy use efficiency (EUE) is a major yield component tinct physiological and agronomical characteristics could be gen- with an epigenetic feature that can be directed by recursive erated. These populations were found to be genetically identical, selection. The shaping of the epigenome allows the increase of but epigenetically different. Furthermore, both the DNA methyl- seed yield in both open-pollinating lines and hybrids. ation patterns as well as the agronomical and physiological characteristics of the selected lines were heritable. Hybrids derived Results and Discussion from parent lines selected for high energy use efficiencies had a 5% Selection of Sublines from an Isogenic Line Based on Respiration. A yield increase on top of heterosis. Our results demonstrate that positive correlation between yield and low cellular respiration artificial selection allows the increase of the yield potential by rates has been demonstrated in canola, Lolium perenne, tomato selecting populations with particular epigenomic states. (Solanum esculentum), and cucumber (Cucumis sativus) (10–16). Based on the published experimental concept (2–4), an artificial Brassica napus ͉ epigenome bidirectional selection procedure was established in canola for high and low cellular respiration rates. We started from an he variability of a quantitative trait depends on genetic, isogenic doubled haploid line to eliminate variation due to Tepigenetic, and environmental factors as well as on their differences at the genetic level. As schematically presented (Fig. mutual interactions. Particularly, plants that cannot escape their 1), out of 200 seedlings from the doubled haploid population environment have to adapt quickly. The importance of the ‘‘Simon,’’ seedlings with the lowest and highest cellular respira- epigenome in both short-term and long-term adaptations to tion were identified and retained. The two plants with the highest stress was demonstrated by the elevated recombination frequen- and lowest respiration were withheld for seed production by cies in the offsprings of Arabidopsis thaliana populations that had self-fertilization. These two populations were the starting ma- been stressed with UV-B radiation or flagellin (1). These terial for four additional rounds of selection for sublines with increased levels of recombination could be transmitted to suc- high and low respiration rates. Seven lines were generated: four cessive generations that had not been exposed to the stress. lines with lower (LR76, LR76-crashed, LR77, and LR79) and Although there is no direct proof that recurrent selection can three with higher (HR80, HR82, and HR83) respiration than direct the epigenetic component of a quantitative trait, previous that of the control from which the selection had been initiated data suggest that it might be possible (2, 3). From a population (Fig. 2). The ‘‘LR76-crashed’’ line was derived from a further obtained by a cross between two inbred flour beetles (Tribolium ␮ casteneum) with an average pupa weight of 2,400 g, a popula- Author contributions: M.D.B. designed research; M.H., B.H., E.S., K.V.D.K., A.A., and H.A. tion with an average pupa weight of 5,800 ␮g could be selected performed research; F.V.B., Y.G., M.B., B. Lambert, and B. Laga analyzed data; and M.D.B. by continuous inbreeding for 120 generations (2, 3). Surprisingly, wrote the paper. such a genetic variability was maintained over so many gener- The authors declare no conflict of interest. ations of inbreeding. Most probably, at a certain stage of the This article is a PNAS Direct Submission. selection process, not genetic, but epigenetic variability, was Freely available online through the PNAS open access option. dealt with, as further supported by identical experiments starting 1To whom correspondence should be addressed. E-mail: [email protected].

with one of the inbred lines homozygous for the genes affecting This article contains supporting information online at www.pnas.org/cgi/content/full/ PLANT BIOLOGY pupa weight (4). 0908755106/DCSupplemental.

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0908755106 PNAS ͉ November 24, 2009 ͉ vol. 106 ͉ no. 47 ͉ 20109–20114 Downloaded by guest on September 28, 2021 The Selected Sublines Have a Distinct EUE. High cellular respiration was expected to be linked to high energy production and vice versa. However, when the total amount of NAD(P)H, reflecting the energy content, was measured (Table 1), lines with a high and low respiration had a low and high NAD(P)H content, respectively, except for line ‘‘LR76-crashed.’’ When the ratio NAD(P)H versus respiration, defined as EUE, was calculated, respiration and EUE correlated inversely, except for line LR76- crashed (Table 1). Probably, the respiration in line LR76-crashed dropped below a critical threshold to maintain sufficient energy production.

Complex I Activity and Ascorbate Content Are High and Photorespi- ration Is Low in the Low-Respiration Sublines. During the selection procedure, respiration was quantified by measuring the reduction of 2,3,5-triphenyltetrazolium chloride (TTC) of hypocotyl explants. TTC reduction occurs at the end of the mitochondrial respiratory chain at complex IV (17) and, therefore, reflects the total electron flow through the mitochondrial respiratory chain, including the alternative oxidative respiratory pathway. The inhibitor of the alternative oxidase SHAM inhibits the reduction of TTC. The electrons enter the mitochondrial electron trans- port chain through complex I, complex II, and the internal and external alternative NAD(P)H dehydrogenases. The activity of complex I, the main dehydrogenase, was measured in leaf 3 of control and selected lines (Table 1), and an inverse correlation with respiration and complex I activity was observed. Recently, l-galactono-1,4-lactone dehydrogenase, the last enzyme of the plant ascorbate biosynthesis pathway, has been found to asso- ciate with an 800-kDa subcomplex of complex I and to have an important function in the accumulation of complex I, because complex I does not accumulate in Arabidopsis null mutants (18, 19). This apparent interrelation between ascorbate biosynthesis and complex I prompted us to measure the ascorbate content in the leaves of the different lines (Table 1). The lines with the highest respiration and the lowest complex I activity had the Fig. 1. Artificial selection for respiration and EUE in canola. The selection was lowest ascorbate content and inversely. In summary, these results initiated from an isogenic doubled haploid line. Approximately five selected suggest that lines with a high and low respiratory rate have a plants with the lowest and highest respiration were self-fertilized, and the reduced and increased complex I content, respectively. The progenies were tested for respiration and EUE. The next rounds of selections high-respiration lines resemble the cytoplasmic male-sterile II were done in one direction for low or high respiration with the two lines with the lowest and highest respiration, respectively. Three to five rounds of (cmsII) mutant of Nicotiana sylvestris that lacks the NAD7 selection were sufficient to generate lines with distinct respiration and EUE. subunit of complex I (20–21). The csmII mutant has no complex I activity, but its respiration is still high due to the compensatory enhanced activity of complex IV (cytochrome oxidase) and, selection for low respiration starting from the line LR76. The probably, an increased activity of the alternative NAD(P)H respiration of line LR76-crashed dropped below a critical thresh- dehydrogenases to meet the ATP needs of the cells. Lines with old, resulting in reduced vigor (see below). At the individual a high complex I content are more efficient in producing ATP, plant level, the variation was high in the high-respiration lines resulting in a low respiratory rate. [for instance, line HR82 with 125 Ϯ 74% (mean Ϯ standard In tobacco (Nicotiana tabacum) and cucumber (Cucumus deviation) compared to the control], and minimal in the low- sativus), the decreased complex I capacity resulted in an increase respiration lines (for instance, line LR76 with 90 Ϯ 18% versus in photorespiration (16, 22). Similarly, we found that the high- the control). The high variation in respiration found between respiration lines had a significantly increased photorespiration, individual plants in the high-respiration lines is mainly due to the while that of the low-respiration lines was strongly reduced exponential increase of respiration once a certain threshold is (Table 1), implying that low respiration, but not too low (as in the passed (1.5 times the average respiration of the control line). LR76-crashed line), correlates with a favorable physiological This results in plants having up to six times higher respiration state and high respiration with a less fitness. than the average respiration of the control line (supporting information (SI) Fig. S1). The high efficiency with which these Yield Is Correlated with EUE. The specific characteristics of the lines were generated (from only 200 seedlings per selection physiological states hint at a high or low stress tolerance (Fig. S2) cycle) suggests that no mutations are involved. Extensive am- and field performance. The selected and control lines were plified fragment-length polymorphism (AFLP) analyses did not tested for seed yield in field trials (five locations with three to six identify differences between these lines and the control, except replications of 10 m2/line) for 3 years (Table 1). Low-respiration for the very high-respiration line HR83. The highest respiring lines with a high EUE had on average up to 8% higher seed yield plants of line HR83 have an abnormal development, are both than that of the control, while on average the seed yield was male and female sterile, and have often chromosome deletions reduced by up to 10% in high-respiration lines with a low EUE. and inversions. Therefore, HR83 was not retained. Also the line In fields with moderate drought stress, the line with the highest LR79 was omitted in further analyses because its respiration was EUE had a 20% higher yield than that of the control, while the close to that of the control (97%). seed yield of the line with the highest respiration and lowest EUE

20110 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0908755106 Hauben et al. Downloaded by guest on September 28, 2021 Fig. 2. Respiration of six selected populations versus the original starting population. The values were percent-normalized versus the average of the starting line ‘‘Simon.’’ Each node represents an individual plant. Each population contains approximately 200 plants. The bar in the boxes represents the median of the population and the other bars, the standard deviation with indication of the mean.

dropped by 20%. With a Pearson correlation of 0.96, EUE and activity corresponds to a decrease in global methylation. In yield were very strongly positively correlated. general, the genomic DNA of the selected lines was hypomethylated (Fig. 3B), but the hypomethylation was more prominent in The Selected Sublines Have Altered Global DNA Methylation and the low-respiration lines, with line LR76-crashed, being the most Histone Modification. The fact that the lines with high and low hypomethylated. respiration had been generated efficiently from an isogenic line The distinct physiological state of the lines was also reflected and that, based on the AFLP results, the selected lines were at the histone level of the chromatin (Fig. 3A). In general, in the indistinguishable, suggest that the distinct physiological charac- selected lines, except for HR82, the methylation at lysine 9 teristics of the lines have an epigenetic basis. Hence, differential (Lys-9) and Lys-27 of histone 3 and acetylation at Lys-8, Lys-12, cytosine methylation and overall histone acetylation and meth- and Lys-16 of histone 4 was lower than that of the control line. ylation were studied to explore the epigenetic changes at the The low-respiration line LR76 and its progeny LR76-crashed molecular level. First, a methylation-sensitive AFLP (ms-AFLP) had a similarly low histone methylation and acetylation, whereas analysis with HpaII and MspI as methylation-sensitive restriction in the high-respiration line HR80 and line HR82, which is less enzymes on genomic DNA of leaf 3 revealed distinct differences vigorous than line HR80, the histone methylation and acetyla- (Ϸ0.3%) between the selected lines and the control ‘‘Simon’’ tion was the lowest and similar to that of the control, respectively. (Fig. S3). Remarkably, all differentially methylated fragments Hence, each of the selected lines is characterized by a specific that were sequenced (110 in total) corresponded to coding degree of methylation and acetylation of histones 3 and 4. sequences. The ms-AFLP patterns were specific for each line and However, this overall picture does not reflect the physiological were maintained throughout different self-fertilizations for at properties of the lines. More in-depth studies, such as chromatin least eight generations. Secondly, to evaluate differences in immunoprecipitation (ChIP) analyses, might reveal a possible global methylation, cytosine extension assays were carried out on relation between histone modifications, gene expression, and genomic DNA of leaf 3 in the different lines. In this assay, by vigor. Artificial selection for altered respiration and EUE clearly means of the methylation-sensitive HpaII and MspI enzymes, it results in overall changes in DNA methylation and histone is possible to incorporate a single [3H]-labeled dCTP, after which modifications through the plasticity of the epigenome (23). the incorporated dCTP activity can be measured. An increase in Extensive ms-AFLP analyses showed that none of the DNA

Table 1. Physiological and agronomical properties of selected canola lines (% normalized versus control†) Seed yield

NAD(P)H Complex I Ascorbate Field with Optimal Line Respiration content EUE‡ activity content Photorespiration Average§ stress¶ ﬁeldʈ

LR76-crashed 74.5*** 73** 98 140** 200*** 80** 97 106 97 LR76 90*** 98 109 127** 208*** 81** 104* 114** 97 LR77 92.5*** 103* 111 120** 185*** 91* 108** 120** 103 HR80 125*** 99 79 100 79* 105* 92** 81** 93* HR82 139*** 98 71 95* 74** 116** 90** 80** 88**

†Original line (variety ‘‘Simon’’) from which the selected lines were derived. ‡Energy use efﬁciency ϭ NAD(P)H content/respiration. §Five locations, six repetitions/location, 3 years. ¶One location with sandy soil and moderate drought stress, 2 years. ʈ One location with loamy soil, 3 years. PLANT BIOLOGY *, P Ͻ 0.05; **, P Ͻ 0.01; and ***, P Ͻ 0.001 versus control.

Hauben et al. PNAS ͉ November 24, 2009 ͉ vol. 106 ͉ no. 47 ͉ 20111 Downloaded by guest on September 28, 2021 Arabidopsis (1), the epigenetic respiration component of canola can be transmitted in reciprocal backcrosses with a nonselected control line. For example, in the high-respiration line HR82 and similarly in the low-respiration line LR77, the self-fertilized progeny and the backcrosses to the original ‘‘Simon’’ line with HR82 and LR77 as females and males resulted in progenies with 127%, 124%, and 128% respiration and with 89%, 91%, and 93% respiration versus the control, respectively. The epigenetic inheritance was also reflected in the largely conserved DNA methylation patterns in the crosses compared to those of the self-fertilized selected lines. The methylation sites that were not conserved had a strong tendency toward hypomethylation, independently from the direction of the cross.

The Epigenetic EUE Component Can Be Added on Top of Heterosis. Although ‘‘Simon’’ is a rather good performing canola line, we wanted to investigate whether the epigenetic EUE component could increase yield in elite varieties and hybrids. To this end, we selected high EUE in the parental lines of a top-yielding canola hybrid. Both the female and male lines are doubled haploids, and for this, each parental line was isogenic. After three to four recurrent selections for high EUE, hybrids were made with the selected parental lines (Fig. S4). The yield in these hybrids was higher (102–106%) than that of the control hybrid of which the parental lines had not been selected. The physiological characteristics were similar to the high-EUE ‘‘Simon’’ sublines: decreased mitochondrial complex I activity, reduced photorespiration, and increased ascorbate, ATP, and NAD(P)H contents (Table S1). Also at the molecular level, the selected hybrids were distinct from the control. Besides the specific differences in global DNA methylation with hypermethylation and hypomethylation in the cotyledons and leaves of the selected hybrids, respectively (Fig. 3C), the transcription profiles as analyzed by cDNA-AFLP were very distinct from the control hybrid (Fig. S5).

Concluding Remarks. Our results show that EUE is a distinct feature of plant vigor and yield and that it possesses an epigenetic component that can be directed by artificial selection. In other words, it is feasible to select for a complex trait solely based on the epigenetic component. The main difference with the selection used in classical breeding schemes is that it is not only done at the population, but primary, at the plant level and in a recursive manner. However, the underlying mechanism that Fig. 3. Characterization of the histone modifications and quantification of stabilizes the distinct epigenetic states is not understood. Start- the global methylation of the genomic DNA by the cytosine extension assay. ing from an isogenic population composed of plants with a (A) Histones prepared from leaf 3 of the control line ‘‘Simon’’ and selected variable epigenome, subpopulations with at least a partially fixed sublines. Below the immunoblots, the corresponding SYPRO Ruby-stained gel epigenome can be selected rather efficiently. It becomes even is shown to demonstrate the equal loading. (B) Cytosine extension assay using genomic DNA prepared from leaf 3 of ‘‘Simon’’ (control) and selected sublines. more puzzling when, during plant development, the epigenome Leaves were harvested from the same population as used for studying the of the selected sublines changes in opposite ways versus the histone modifications. (C) Cytosine extension assay with genomic DNA pre- control. For example, in hybrids generated by crossing parental pared from cotyledons and the fourth leaf of a nonselected (control) and a lines that have been selected for higher EUE, the total genomic selected hybrid. The cytosine extension assays were done on 10 individual DNA is hypermethylated in the cotyledons, while in the fourth plants. The error bars indicate the standard error. leaf, the genomic DNA is hypomethylated versus the control hybrid (Fig. 3C). It has to be mentioned that we found in our selections for higher EUE subpopulations that in the first rounds methylation patterns found in the selected lines corresponded of selection returned to their original state. Only by recurrent with methylation patterns of individual plants from the original selection the higher EUE could be fixed. It is not unlikely that population, indicating that recurrent selection for altered respi- the epigenetic stabilization can be tracked down to the expres- ration at the individual plant level allows the selection for genes sion of specific miRNAs. We are testing this possibility by with specific and stable methylation states and distinct pheno- making an inventory of the small RNAs in the selected and types, as a consequence. control lines. The ‘‘epigenetic EUE component’’ is stably inherited, allow- The Subline Specific Respiration Rate Is Transmitted in Self-Fertilized ing the creation of distinct isogenic sublines that can be used in Seeds and Backcrosses. From the approximately eight generations of breeding. Probably as importantly, these isogenic sublines with seed upscalings in the greenhouse and 3 years of elaborate field distinct yield characteristics have a low ‘‘genetic noise’’ and, trials, it became obvious that the epigenetic EUE component of the therefore, are excellent material to study yield and EUE at the lines was stably inherited by self-fertilized siblings (Fig. S3). As molecular level. Initial experiments with cDNA-AFLP show already described for the epigenetic recombination component in clear differences in transcription between the sublines (Fig. S5)

20112 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0908755106 Hauben et al. Downloaded by guest on September 28, 2021 and microarray experiments being carried out to obtain a Physiological and Biochemical Assays. Cellular respiration of hypocotyl ex- broader picture. Making an inventory of the small RNAs of the plants was quantified by measuring the reduction of TTC as described (14). ATP sublines and ChIP experiments with differentially methylated and total NAD(P)H contents were quantified as described (24, 25). EUE was expressed as the ratio of the percent-normalized values versus the control of and/or expressed genes will contribute to a better understanding NAD(P)H content to the amount of reduced TTC. Ascorbate content and of the relation between DNA methylation (as the discrepancy in complex I activity were measured with the reflectometric ascorbic acid test DNA methylation between cotyledons and leaves in the hybrids), (Merck) and the MitoProfile Dipstick Assay kit (MitoSciences), respectively. The histone modifications, transcription, and the resulting pheno- photorespiration intensity was quantified in vitro by floating cotyledons or types. The fact that the concept can be used to further improve leaf pieces on ammonium-free medium containing 3–5 mg/L glufosinate, Ϸ ␮ Ϫ2 Ϫ1 yield in already top-performing hybrids open perspectives to go incubated for 24 h under continuous light ( 70 mol m s ). The ammonium production was evaluated as described (26). Photorespiration was expressed beyond heterosis. as percentage produced ammonium versus the control. The central role of plant mitochondria in energy metabolism and their involvement in biochemical pathways, such as photorespira- Field Trials. For the field trials, five locations with different (from sandy to tion, amino acid biosynthesis, and maintenance of redox status, is loamy) soil qualities were used. Each location was subdivided in six plots per conserved in most plant species. Indeed, ongoing experiments line. Each plot was 10 m2 in size, with different arrangements of the plots for indicate that analogous selections for EUE can be applied in rice each location. Yield was expressed as kg seed/plot. (Oryza sativa) and tomato (Solanum esculentum). The implemen- Statistics. Data were statistically analyzed by one-way ANOVA with Dunnett’s tation of the ‘‘epigenetic’’ selection for EUE has the potential to posttest with the Prism version 5.00 (GraphPad Software). increase yield in many crops and this will further contribute to a better knowledge of the epigenetic mechanism, especially in crops, Molecular Analyses. Methylation-sensitive AFLP (27) was done with the restric- for which the genomics tools are well developed. tion enzyme combinations EcoRI-MspI/HpaII. The methylation state of the total genomic DNA was quantified with the cytosine extension assay as Materials and Methods described (28). In summary, genomic DNA was prepared from cotyledons, leaf 3, or leaf 4 and assayed with the Quant-iT dsDNA High-Sensitivity Assay kit Artificial Selection for Respiration and EUE. Seedlings were grown in vitro for (Invitrogen). Of 10 individual leaves per line, 150 ng were used for the cytosine 2 weeks on agar in half-strength Murashige and Skoog medium supple- extension assay and digested with a 10-fold excess of HpaII and MspI. After mented with 2% sucrose. The shoot tips of the seedlings were placed on the 24 h, again a 10-fold excess of the restriction enzymes was added for another medium for rooting, while five hypocotyl explants per seedling were 24-h incubation. After 48 h, the single-nucleotide extension reaction was done cultured for 5 days on callus-inducing medium (Murashige and Skoog with the digested genomic DNA. Of the reaction mixture, 10 ␮L were spotted medium supplemented with 3% sucrose and 1 mg/L 2,4-dichlorophenoxy- on DE81 Whatman filters and washed three times with phosphate buffer (0.5 ␣ acetic acid, 0.25 mg/L -naphthaleneacetic acid, and 1 mg/L 6-benzylamin- M, pH 7.0). The radioactive incorporation was measured (cpm). opurine). Cellular respiration of the hypocotyl explants was measured. The rooted shoot tips of approximately five seedlings with the highest and Analyses of Histone Modifications. Histones were acid-extracted, separated in lowest respiration were transferred to the greenhouse for seed production NuPAGE 4–12% Bis-Tris gel (Invitrogen), and subjected to a standard proce- by self-fertilization. Both respiration and NAD(P)H content of approxi- dure for immunoblotting with anti-H4-acetyl-Lys-8, anti-H4-acetyl-Lys-12 an- mately 35–40 seedlings of the obtained progenies were measured. Lines tibodies (Millipore), and with anti-H3-dimethyl-Lys-9 and anti-H3-trimethyl- with the lowest and highest respiration and highest and lowest EUE, Lys-27 antibodies (kind gifts of Thomas Jenuwein, Max Planck-Institute of respectively, were retained. The next rounds of selections were done in one Immunobiology, Freiburg, Germany). direction for low or high respiration in lines with the lowest and highest respiration, respectively (Fig. 1). Three to five rounds of selection were ACKNOWLEDGMENTS. We thank T. Jenuwein (Max Planck-Institute of Immu- sufficient to generate lines with distinct respiration and EUE. The same nobiology, Freiburg, Germany) for providing the antibodies against anti-H3- protocol was used for the production of more EUE hybrids. The male-sterile dimethyl-Lys-9 and anti-H3-trimethyl-Lys-27 and Dr. Martine De Cock for help in preparing the manuscript. This work was supported by the Institute for the female, the maintainer, and the male restorer lines were each selected for Promotion of Innovation by Science and Technology in Flanders, Belgium low respiration and high EUE. After three cycles, the male-sterile female (Research and Development project ‘‘Phoenix’’) and an Industrial Agreement and restorer were crossed to produce hybrid seeds (Fig. S4). for Training through Research (CIFRE) fellowship (to H.A.).

1. Molinier J, Ries G, Zipfel C, Hohn B (2006) Transgeneration memory of stress in plants. 14. De Block M, De Brouwer D (2002) A simple and robust in vitro assay to quantify the Nature 442:1046–1049. vigour of oilseed rape lines and hybrids. Plant Physiol Biochem 40:845–852. 2. Enfield FD, Comstock RE, Braskerud O (1966) Selection for pupa weight in Tribolium 15. Nunes-Nesi A, et al. (2005) Enhanced photosynthetic performance and growth as a castaneum. I. Parameters in base populations. Genetics 54:523–533. consequence of decreasing mitochondrial malate dehydrogenase activity in transgenic 3. Kaufman PK, Enfield FD, Comstock RE (1977) Stabilizing selection for pupa weight in tomato plants. Plant Physiol 137:611–622. Tribolium castaneum. Genetics 87:324–341. 16. Juszczuk IM, et al. (2007) Effect of mitochondrial genome rearrangement on respira- 4. Enfield FD, Braskerud O (1989) Mutational variance for pupa weight in Tribolium tory activity, photosynthesis, photorespiration and energy status of MSC16 cucumber castaneum. Theor Appl Genet 77:416–420. (Cucumis sativus) mutant. Physiol Plant 131:527–541. 5. Jacobsen SE, Meyerowitz EM (1997) Hypermethylated SUPERMAN epigenetic alleles in 17. Rich PR, Mischis LA, Purton S, Wiskich JT (2001) The sites of interaction of triphenyltetrazolium Arabidopsis. Science 277:1100–1103. chloride with mitochondrial respiratory chains. FEMS Microbiol Lett 202:181–187. 6. Jacobsen SE, Sakai H, Finnegan EJ, Cao X, Meyerowitz EM (2000) Ectopic hypermeth- 18. Millar AH, et al. (2003) Control of ascorbate synthesis by respiration and its implications ylation of flower-specific genes in Arabidopsis. Curr Biol 10:179–186. for stress responses. Plant Physiol 133:443–447. 7. Stokes TL, Kunkel BN, Richards EJ (2002) Epigenetic variation in Arabidopsis disease 19. Pineau B, Layoune O, Danon A, De Paepe R (2008) l-Galactono-1,4-lactone dehydro- resistance. Genes Dev 16:171–182. genase is required for the accumulation of plant respiratory complex I. J Biol Chem 8. Jablonka E, Raz G (2009) Transgenerational epigenetic inheritance: Prevalence, mecha- 283:32500–32505. nisms, and implications for the study of heredity and evolution. Q Rev Biol 84:131–176. 20. Gutierres S, et al. (1997) Lack of mitochondrial and nuclear-encoded subunits of 9. Johannes F, et al. (2009) Assessing the impact of transgenerational epigenetic variation complex I and alteration of the respiratory chain in Nicotiana sylvestris mitochondrial on complex traits. PLoS Genet 5:e1000530. deletion mutants. Proc Natl Acad Sci USA 94:3436–3441. 10. Wilson D (1982) Response to selection of dark respiration rate of mature leaves in 21. Sabar M, De Paepe R, de Kouchkovsky Y (2000) Complex I impairment, respiratory Lolium perenne and its effects on growth of young plants and simulated swards. Ann compensations, and photosynthetic decrease in nuclear and mitochondrial male sterile Bot 49:303–312. mutants of Nicotiana sylvestris. Plant Physiol 124:1239–1249. 11. Wilson D, Jones JG (1982) Effect of selection for dark respiration rate of mature leaves 22. Priault P, et al. (2006) The lack of mitochondrial complex I in a CMSII mutant of on crop yields of Lolium perenne cv. S23. Ann Bot 49:313–320. Nicotiana sylvestris increases photorespiration through an increased internal resis- 12. Kraus E, Aydemir Y, Duin S, Kollo¨ffel C, Lambers H (1993) Yield advantage of a tance to CO2 diffusion. J Exp Bot 57:3195–3207. ‘slow-’over a ‘fast-’ respiring population of Lolium perenne cv. S23 depends on plant 23. Tsaftaris AS, Polidoros AN, Kapazoglou A, Tani E, Kovacevic NM (2008) in Plant density. New Phytol 123:39–44. Breeding Reviews, eds Janick J (Wiley, New York, NY), vol 30, pp 49–178. 13. Kraus E, Lambers H, Kollo¨ffel C (1993) The effect of handling on the yield of two 24. Rawyler A, Pavelic D, Gianinazzi C, Oberson J, Braendle R (1999) Membrane lipid

populations of Lolium perenne selected for differences in mature leaf respiration rate. integrity relies on a threshold of ATP production rate in potato cell cultures submitted PLANT BIOLOGY Physiol Plant 89:341–346. to anoxia. Plant Physiol 120:293–300.

Hauben et al. PNAS ͉ November 24, 2009 ͉ vol. 106 ͉ no. 47 ͉ 20113 Downloaded by guest on September 28, 2021 25. Nakamura J, et al. (2003) Quantitation of intracellular NAD(P)H can monitor an 27. Matthes M, Singh R, Cheah S-C, Karp A (2001) Variation in oil palm (Elaeis guineensis imbalance of DNA single strand break repair in base excision repair deﬁcient cells in Jacq.) tissue culture-derived regenerants revealed by AFLPs with methylation-sensitive real time. Nucleic Acids Res 31:e104. enzymes. Theor Appl Genet 102:971–979. 26. De Block M, De Sonville A, De Brouwer D (1995) The selection mechanism of 28. Boyko A, et al. (2007) Transgenerational changes in the genome stability and meth- phosphinothricin is inﬂuenced by the metabolic status of the tissue. Planta ylation in pathogen-infected plants (virus-induced plant genome instability). Nucleic 197:619–626. Acids Res 35:1714–1725.

20114 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0908755106 Hauben et al. Downloaded by guest on September 28, 2021