ebook img

Parental DNA methylation states are associated with heterosis in epigenetic hybrids PDF

69 Pages·2017·2.24 MB·English
by  
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Parental DNA methylation states are associated with heterosis in epigenetic hybrids

Plant Physiology Preview. Published on December 1, 2017, as DOI:10.1104/pp.17.01054 1 Short title: Epigenetic divergence associated with heterosis 2 § Corresponding co-last authors: 3 Maike Stam, University of Amsterdam, Swammerdam Institute for Life Sciences, Science 4 Park 904 1098XH Amsterdam, The Netherlands. 5 Frank Johannes, Population epigenetics and epigenomics, Department of Plant Sciences, 6 Technical University of Munich, Liesel-Beckmann-Str. 2, 85354 Freising, Germany 7 8 Parental DNA methylation states are associated with heterosis in 9 epigenetic hybrids 10 11 12 Kathrin Lauss1*, René Wardenaar2*, Rurika Oka1, Marieke H.A. van Hulten3, Victor Guryev4, 13 Joost J.B. Keurentjes3, Maike Stam1§, Frank Johannes5,6§ 14 15 1 University of Amsterdam, Swammerdam Institute for Life Sciences, Science Park 904 16 1098XH Amsterdam, The Netherlands. 17 2 University of Groningen, Groningen Bioinformatics Centre, Faculty of Mathematics and 18 Natural Sciences, Nijenborgh 7, 9747 AG Groningen, The Netherlands. 19 3 Wageningen University & Research, Laboratory of Genetics, Droevendaalsesteeg 1, 20 6708PB Wageningen, The Netherlands. 21 4 Genome structure aging, European Research Institute for the Biology of Ageing, University 22 Medical Centre Groningen and University of Groningen, Antonius Deusinglaan 1, Building 23 3226, 9713 AV Groningen, The Netherlands 24 5 Population epigenetics and epigenomics, Department of Plant Sciences, Technical 25 University of Munich, Liesel-Beckmann-Str. 2, 85354 Freising, Germany 26 6 Institute for Advanced Study, Technical University of Munich, Lichtenbergstr. 2a, 85748 27 Garching, Germany 1 Downloaded from on April 9, 2019 - Published by www.plantphysiol.org Copyright © 2017 American Society of Plant Biologists. All rights reserved. Copyright 2017 by the American Society of Plant Biologists 28 * co-first authors 29 § co-last authors 30 31 One sentence summary: 32 DNA methylation differences between isogenic parental lines can directly or indirectly 33 trigger heterosis in Aradidopsis hybrids. 34 35 Author Contributions 36 K.L., M.S. and F.J. designed the study, interpreted the data and wrote the manuscript with 37 contributions from J.J.B.K. and R.W. K.L. and M.H.A.v.H. planned and performed the 38 phenotypic screen. R.W., K.L., V.G., F.J. and R.O. performed data analysis. 39 40 Funding 41 We acknowledge the support of the Centre for Improving Plant Yield (CIPY; part of the 42 Netherlands Genomics Initiative and the Netherlands Organization for Scientific Research) 43 for Kathrin Lauss. The first phenotypic screen was supported by the Enabling Technologies 44 Hotels (ETH) Programme from CIPY. We gratefully acknowledge the financial support of the 45 European Commission 7th Framework-People-2012-ITN Project EpiTRAITS (Epigenetic 46 Regulation of Economically Important Plant Traits; 316965) to Rurika Oka and Maike Stam. 47 Frank Johannes acknowledges support from the Technical University of Munich-Institute for 48 Advanced Study funded by the German Excellence Initiative and the European Union 49 Seventh Framework Programme under grant agreement #291763, and the 50 SFB924/Sonderforschungsbereich924 of the Deutsche Forschungsgemeinschaft (DFG). 51 52 Corresponding co-last authors: [email protected] (MS), [email protected] (FJ) 53 2 Downloaded from on April 9, 2019 - Published by www.plantphysiol.org Copyright © 2017 American Society of Plant Biologists. All rights reserved. 54 Abstract 55 Despite the importance and wide exploitation of heterosis in commercial crop breeding, the 56 molecular mechanisms behind this phenomenon are not completely understood. Recent 57 studies have implicated changes in DNA methylation and small RNAs in hybrid performance, 58 however, it remains unclear whether epigenetic changes are a cause or consequence of 59 heterosis. Here, we analyze a large panel of over 500 A. thaliana epigenetic hybrid plants 60 (epiHybrids), which we derived from near-isogenic but epigenetically divergent parents. This 61 proof-of-principle experimental system allowed us to quantify the contribution of parental 62 methylation differences to heterosis. We measured traits such as leaf area (LA), growth rate 63 (GR), flowering time (FT), main stem branching (MSB), rosette branching (RB) and final plant 64 height (HT) and observed several strong positive and negative heterotic phenotypes among 65 the epiHybrids. Using an epigenetic quantitative trait locus mapping approach, we were able 66 to identify specific differentially methylated regions (DMRs) in the parental genomes that 67 are associated with hybrid performance. Sequencing of methylomes, transcriptomes and 68 genomes of selected parent-epiHybrid combinations further showed that these parental 69 DMRs most likely mediate remodeling of methylation and transcriptional states at specific 70 loci in the hybrids. Taken together, our data suggest that locus-specific epigenetic 71 divergence between the parental lines can directly or indirectly trigger heterosis in 72 Arabidopsis hybrids independent of genetic changes. These results add to a growing 73 body of evidence that points to epigenetic factors as one of the key determinants of 74 hybrid performance. 75 76 Introduction 77 Heterosis describes an F1 hybrid phenotype that is superior compared to the phenotype of 78 its parents. The phenomenon has been exploited extensively in agricultural breeding for 3 Downloaded from on April 9, 2019 - Published by www.plantphysiol.org Copyright © 2017 American Society of Plant Biologists. All rights reserved. 79 decades and has improved crop performance tremendously (Chen, 2010; Schnable and 80 Springer, 2013). Despite its commercial impact, knowledge of the molecular basis underlying 81 heterosis remains incomplete. Most studies have focused on finding genetic explanations, 82 resulting in the classical dominance (Jones, 1917; Crow, 1998; Schnable and Springer, 2013) 83 and overdominance (Crow, 1948; Crow, 1998) models of heterosis. In line with genetic 84 explanations, interspecies hybrids have been frequently observed to show a higher degree 85 of heterosis than intraspecies hybrids, indicating that genetic distance correlates with the 86 extent of heterosis (East, 1936; Chen, 2010). 87 Genetic explanations do not, however, sufficiently explain nor predict heterosis. There is 88 growing evidence that also epigenetic factors play a role in heterosis (Groszmann et al., 89 2013; Springer, 2013; Dapp et al., 2015a; Zhu et al., 2017). It has, for example, been shown 90 that altered epigenetic profiles at genes regulating circadian rhythm play an important role 91 in heterotic Arabidopsis hybrids (Ni et al., 2009a). Moreover, heterotic hybrids of 92 Arabidopsis, maize and tomato are shown to differ in levels of small regulatory RNAs and/or 93 DNA methylation (5mC) relative to their parental lines (Groszmann et al., 2011; Barber et al., 94 2012; Shen et al., 2012; Shivaprasad et al., 2012; Zhang et al., 2016b). 95 Remodeling of the methylome during hybridization has been proposed to be involved in 96 heterosis in genetic hybrids, and was implicated in the formation of novel epialleles in a 97 met1-derived epiHybrid (Groszmann et al., 2011; Greaves et al., 2012; Shen et al., 2012; 98 Rigal et al., 2016). Processes such as the transfer of 5mC between alleles (trans- 99 chromosomal methylation, TCM), or a loss of 5mC at one of the alleles (trans-chromosomal 100 demethylation, TCdM) have been indicated to contribute to the observed remodeling of the 101 epigenome (Greaves et al., 2012; Shivaprasad et al., 2012; Groszmann et al., 2013). These 102 TC(d)M events occur between homologous sequences and have been shown to require the 103 RNA-directed DNA methylation (RdDM) pathway, which involves small regulatory RNAs 104 (Zhang et al., 2016b). Strikingly, some of these changes in 5mC levels have been shown to be 4 Downloaded from on April 9, 2019 - Published by www.plantphysiol.org Copyright © 2017 American Society of Plant Biologists. All rights reserved. 105 stable over multiple generations (Greaves et al., 2012; Greaves et al., 2014). However, 106 parental lines used for these studies, differed in both their genetic and epigenetic profile, 107 making it challenging to disentangle genetic from epigenetic effects. 108 It has been shown in isogenic epigenetic recombinant inbred lines (epiRILs) that 109 heritable morphological variation in Arabidopsis plants can be exclusively caused by 110 epigenetic factors (Johannes et al., 2009; Roux et al., 2011; Cortijo et al., 2014; Kooke and 111 Keurentjes, 2015). To specifically address the contribution of parental epigenetic variation to 112 F1 heterosis we made use of the same epiRILs (Johannes et al., 2009; Reinders et al., 2009) 113 to generate F1 epigenetic hybrids, hereafter called epiHybrids (Dapp et al., 2015a). EpiRILs 114 are near isogenic, but display mosaic patterns in terms of their epigenomes. The two 115 Arabidopsis epiRIL populations reported have been created by crossing the wild-type 116 Columbia accession (Col-wt) with Col-wt lines carrying a mutation in either 117 METHYLTRANSFERASE 1 (MET1-3) or DECREASE IN DNA METHYLATION 1 (DDM1–2) 118 (Johannes et al., 2009; Reinders et al., 2009). MET1 maintains DNA methylation at cytosines 119 in a CG sequence context (mCG), but also affects non-CG methylation (Finnegan et al., 1996; 120 Mathieu et al., 2007; Stroud et al., 2013). Loss of MET1 causes an almost complete 121 elimination of mCG genome-wide, which is associated with misregulated gene expression 122 and transcriptional activation of TEs (Zhang et al., 2006; Cokus et al., 2008; Lister et al., 123 2008). DDM1 is a nucleosome remodeler and the ddm1-2 mutation leads to a ~70% 124 reduction in DNA methylation (Kakutani et al., 1995), predominantly affecting mCG and 125 mCHG (where H represents A, C or T), and to a lesser extent mCHH (Zemach et al., 2013), in 126 primarily long transposable elements (Zemach et al., 2013). Loss of DDM1 also affects genic 127 loci where it reduces mCG and causes CHG hypermethylation in gene bodies (Zemach et al., 128 2013). 129 Recently, epiHybrids have been generated by crossing a met1-derived epiRIL with 130 Col-wt, and heterosis for biomass was reported for one of the epiHybrids (Dapp et al., 5 Downloaded from on April 9, 2019 - Published by www.plantphysiol.org Copyright © 2017 American Society of Plant Biologists. All rights reserved. 131 2015a). The effect was only observed with the epiRIL as maternal parent, suggesting a strong 132 maternal effect (Dapp et al., 2015a). In this study we created epiHybrids by crossing Col-wt 133 as a maternal parent to nineteen different near-isogenic but epigenetically divergent ddm1- 134 derived epiRILs, allowing the assessment of epigenetic variation in identical maternal 135 backgrounds. Using high-throughput phenotyping, we observed various positive and 136 negative heterotic effects in one or more of the six traits monitored among the nineteen 137 epiHybrids, indicating that epigenetic divergence among parents has a direct or indirect role 138 in triggering heterosis. Furthermore, in contrast to previous studies, our experimental design 139 allowed us to employ an epigenetic quantitative trait locus mapping approach that, in 140 combination with methylome and transcriptome profiling, allowed quantifying and 141 characterizing the contribution of parental methylation differences to heterosis. Using this 142 approach we were able to identify specific differentially methylated regions (DMRs) in the 143 parental genomes that are associated with heterotic phenotypes in the epiHybrids. We 144 provide evidence that these parental DMRs mediate local methylome and transcriptome 145 remodeling at putatively causative genes, thus providing molecular mechanisms underlying 146 the observed heterotic phenotypes in the epiHybrids. 147 148 Results 149 Construction of epigenetic Hybrids 150 Hybrids are usually generated from parental lines that vary at both the genomic and 151 epigenomic level and disentangling those two sources of variation is challenging. To 152 overcome this limitation, we generated epigenetic A. thaliana F1 hybrids (epiHybrids) from 153 near-isogenic but epigenetically divergent parental lines by crossing Col-0 wildtype (Col-wt) 154 as maternal parent to near-isogenic ddm1-2-derived epigenetic recombinant inbred lines 155 (epiRILs) (Johannes et al., 2009) as the paternal parents (Fig. 1a). EpiRILs carry chromosomes 156 that are a mosaic of Col-wt and hypomethylated ddm1-2-derived genomic regions (Johannes 6 Downloaded from on April 9, 2019 - Published by www.plantphysiol.org Copyright © 2017 American Society of Plant Biologists. All rights reserved. 157 et al., 2009; Colomé-Tatché et al., 2012; Cortijo et al., 2014) (Fig. 1A). Nineteen epiRIL 158 parental lines were selected that sample a broad range of 5mC divergence from the Col-wt 159 reference methylome (Fig. 1B, Supplemental Table S1). Besides, lines were chosen that have 160 a wildtype methylation profile at FLOWERING LOCUS WAGENINGEN (FWA) (Supplemental 161 Fig. S1, Table S1), as loss of DNA methylation at the FWA locus is known to affect flowering 162 time and, as such, many related developmental traits (Soppe et al., 2000). Furthermore, we 163 selected epiRILs covering a wide range of phenotypic variation in flowering time and root 164 length, two traits that have previously been monitored (Supplemental Table S1) (Johannes et 165 al., 2009). 166 167 Heterotic phenotypes occur in the epiHybrids 168 The phenotypic performance of the 19 epiHybrid lines and their 20 parental lines was 169 assessed. In total, we monitored about 1090 plants (~28 replicate plants per epiHybrid or 170 parental line) for a range of quantitative traits: leaf area (LA), growth rate (GR), flowering 171 time (FT), main stem branching (MSB), rosette branching (RB), plant height (HT) and seed 172 yield (SY) (Supplemental Tables S2-7). The hybrids and parental lines were grown in parallel 173 in a climate-controlled chamber with automated watering. The plants were randomized 174 throughout the chamber to level out phenotypic effects caused by plant position. Leaf area 175 was measured up to 14 days after sowing (DAS), using an automated camera system (Fig. 176 1C), and growth rate was determined based on these data (Supplemental: Supplemental 177 Methods). Flowering time was scored manually as the day of opening of the first flower. 178 After all plants started flowering, they were transferred to the greenhouse and grown to 179 maturity. Main stem branching, rosette branching and height were scored manually after 180 harvesting of the plants. The phenotypic observations for seed yield were inconsistent in a 181 replication experiment, therefore those datasets were excluded from further analysis. 7 Downloaded from on April 9, 2019 - Published by www.plantphysiol.org Copyright © 2017 American Society of Plant Biologists. All rights reserved. 182 The extent of heterosis was evaluated by comparing the phenotypic performance of 183 the hybrids to that of their parental lines. We distinguished five effects: additivity, positive 184 mid-parent heterosis (positive MPH), negative mid-parent heterosis (negative MPH), high- 185 parent heterosis (HPH) and low-parent heterosis (LPH; Fig. 1D-F). Briefly, an additive effect is 186 defined by a hybrid mean phenotype that is equal or close to the average phenotype of the 187 two parents (the mid-parent value, MPV). Mid-parent heterosis, by contrast, refers to 188 positive or negative deviations of the hybrid mean phenotype from the MPV. High-parent 189 heterosis and low-parent heterosis are important special cases of mid-parent heterosis in 190 which the hybrid mean phenotype either exceeds the mean phenotype of the high parent, 191 or falls below that of the lowest performing parent. In crop breeding, the focus is usually on 192 obtaining high parent heterosis (HPH) and low parent heterosis (LPH) as these present novel 193 phenotypes that are outside the parental range. Depending on the trait and commercial 194 application, either high-parent heterosis or low-parent heterosis can be considered superior. 195 For instance, early flowering may be preferable over late flowering; in such cases maximizing 196 low-parent heterosis may be desirable. For other traits, such as yield or biomass, it is more 197 important to maximize high-parent heterosis. However, in order to obtain a comprehensive 198 view of hybrid performance it is informative to also monitor mid-parent heterosis, as many 199 traits of mature plants are affected by other traits that may not display fully penetrant 200 heterotic effects. 201 We observed a remarkably wide range of heterotic phenotypes among the 202 epiHybrids (Fig. 1G, Supplemental Tables S2-7). The magnitude of these phenotypic effects 203 was substantial (Fig. 1H-J, Supplemental Fig. S2, Tables S8-19) and similar to that typically 204 seen in hybrids of Arabidopsis natural accessions (Groszmann et al., 2014; Wang et al., 205 2015). Many epiHybrids (16/19) exhibited significant mid-parent heterosis in at least one of 206 the six monitored traits (FDR = 0.05, Fig. 1G). Across all epiHybrids and traits we identified 30 207 cases of positive mid-parent heterosis and negative mid-parent heterosis. Among those, four 8 Downloaded from on April 9, 2019 - Published by www.plantphysiol.org Copyright © 2017 American Society of Plant Biologists. All rights reserved. 208 cases show low-parent heterosis and nine cases show high-parent heterosis (Fig. 1G). 209 Interestingly, in 11 out of the 17 remaining cases of mid-parent heterosis, the phenotypic 210 means of the epiHybrids were in the direction of the phenotypic means of the epiRIL parent 211 rather than in the direction of the Col-wt parent (Fig. 1G, Supplemental Tables S2-7, F1 212 trend). Also, all four low-parent heterosis and two of the high-parent heterosis cases were in 213 the direction of the epiRIL parent (Fig. 1G, I-J, Supplemental Fig. S2). This observation 214 illustrates that ddm1-2-derived hypomethylated epialleles are often (partially) dominant 215 over wild-type epialleles. 216 We observed cases of high-parent heterosis for leaf area (LA), height (HT) and main 217 stem branching (MSB), and cases of low-parent heterosis for flowering time (FT) and main 218 stem branching. High-parent heterosis for leaf area (LA) occurred in 3 out of 19 epiHybrids, 219 namely 232H, 195H and 193H (the number in the ID refers to the epiRIL parent and the H 220 stands for F1 epiHybrid line). EpiHybrids 232H, 195H and 193H significantly exceeded their 221 best parent (Col-wt) by 17%, 18% and 15%, respectively (Fig. 1H, Supplemental Table S19). 222 Interestingly, although growth rate is developmentally related to leaf area, hybrid effects in 223 growth rate were only moderately, albeit positively, correlated with leaf area (rho = 0.57, P = 224 0.02), which implies that leaf area heterosis is determined by other traits besides growth 225 rate. 226 For height, we detected five cases of significant high-parent heterosis with up to 6% 227 increases in height (Fig. 1I, Supplemental Table S14). One may expect leaf area high-parent 228 heterosis to strongly correlate with height high-parent heterosis, as the rosette is providing 229 nutrients for the developing shoot (Bennett et al., 2012). However, high-parent heterosis for 230 both leaf area and height occurred only in one epiHybrid (193H; Fig. 1G). For main stem 231 branching, we detected one case of high-parent heterosis (64H; Fig. 1G and Supplemental 232 Fig. S2). 9 Downloaded from on April 9, 2019 - Published by www.plantphysiol.org Copyright © 2017 American Society of Plant Biologists. All rights reserved. 233 Besides positive heterosis, our phenotypic screen revealed strong negative heterotic 234 effects for flowering time (earlier flowering) and main stem branching (less main stem 235 branching). Significant low-parent heterosis occurred in epiHybrids 232H, 208H and 344H 236 (FT) and 438H (MSB) (Fig. 1J, Supplemental Fig. S2, Tables S15 and S17). In the most 237 prominent case for flowering time, the epiHybrid (232H) flowered about 10% earlier than 238 that of the earliest flowering parent. EpiHybrids 208H and 344H flowered 3% and 4% earlier 239 than their earliest parent (epiRIL 208 and epiRIL 344), respectively. EpiHybrid 438H showed 240 14% less main stem branching than its least branched parent (Supplemental Fig. S2). 241 The reproducibility of our findings was tested by performing replicate experiments, 242 using seeds from newly performed crosses under the same growth conditions as before. We 243 focused on epiHybrids that exhibited relatively strong positive or negative heterotic 244 phenotypes in the initial screen (193H, 150H, 232H; Fig. 1G), and measured leaf area, 245 flowering time and height in the epiHybrids and their parents. We monitored about 540 246 plants for leaf area (~60 replicates per line) and 270 plants for flowering time and height 247 (~30 replicates per line). The direction of the heterotic effects in leaf area, flowering time 248 and height was reproducible in all tested cases (Fig. 2A and B). Importantly, the leaf area and 249 height high-parent heterosis observed for 193H, and the strong flowering time low-parent 250 heterosis for 232H were perfectly reproducible, while leaf area high-parent heterosis 251 observed for 232H was reduced to mid-parent heterosis (Fig. 2A). Taken together, these 252 results show that the heterotic effects observed in the epiHybrids are relatively stable for 253 leaf area, height and flowering time even across fresh parental seed batches and 254 independently performed crosses, which is not always the case for Arabidopsis phenotypes 255 (Massonnet et al., 2010). 256 257 Parental DMRs are associated with epiHybrid performance 10 Downloaded from on April 9, 2019 - Published by www.plantphysiol.org Copyright © 2017 American Society of Plant Biologists. All rights reserved.

Description:
Medical Centre Groningen and University of Groningen, Antonius Deusinglaan 1, Building. 22. 3226, 9713 AV DNA methylation differences between isogenic parental lines can directly or indirectly. 32 . epiHybrids, indicating that epigenetic divergence among parents has a direct or indirect role. 137.
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.