Nội dung

Pearce et al. BMC Plant Biology (2014) 14:368 DOI 10.1186/s12870-014-0368-2 RESEARCH ARTICLE Open Access Regulation of Zn and Fe transporters by the GPC1 gene during early wheat monocarpic senescence Stephen Pearce1, Facundo Tabbita2, Dario Cantu3, Vince Buffalo1, Raz Avni4, Hans Vazquez-Gross1, Rongrong Zhao5, Christopher J Conley6, Assaf Distelfeld7 and Jorge Dubcovksy1,8* Abstract Background: During wheat senescence, leaf components are degraded in a coordinated manner, releasing amino acids and micronutrients which are subsequently transported to the developing grain. We have previously shown that the simultaneous downregulation of Grain Protein Content (GPC) transcription factors, GPC1 and GPC2, greatly delays senescence and disrupts nutrient remobilization, and therefore provide a valuable entry point to identify genes involved in micronutrient transport to the wheat grain. Results: We generated loss-of-function mutations for GPC1 and GPC2 in tetraploid wheat and showed in field trials that gpc1 mutants exhibit significant delays in senescence and reductions in grain Zn and Fe content, but that mutations in GPC2 had no significant effect on these traits. An RNA-seq study of these mutants at different time points showed a larger proportion of senescence-regulated genes among the GPC1 (64%) than among the GPC2 (37%) regulated genes. Combined, the two GPC genes regulate a subset (21.2%) of the senescence-regulated genes, 76.1% of which are upregulated at 12 days after anthesis, before the appearance of any visible signs of senescence. Taken together, these results demonstrate that GPC1 is a key regulator of nutrient remobilization which acts predominantly during the early stages of senescence. Genes upregulated at this stage include transporters from the ZIP and YSL gene families, which facilitate Zn and Fe export from the cytoplasm to the phloem, and genes involved in the biosynthesis of chelators that facilitate the phloem-based transport of these nutrients to the grains. Conclusions: This study provides an overview of the transport mechanisms activated in the wheat flag leaf during monocarpic senescence. It also identifies promising targets to improve nutrient remobilization to the wheat grain, which can help mitigate Zn and Fe deficiencies that afflict many regions of the developing world. Keywords: Wheat, Senescence, GPC, Zinc transport, Iron transport, ZIP Background In annual grasses, monocarpic senescence is the final stage of a plant’s development during which vegetative tissues are degraded and their cellular nutrients and amino acids are transported to the developing grain. The regulation of this process is crucial for the plant’s reproductive success and determines to a large extent the nutritional quality of the harvested grain. Among wild diploid relatives of wheat, there exists large variation in Zn and Fe grain content, whereas modern wheat germplasm collections exhibit comparatively lower and less * Correspondence: jdubcovsky@ucdavis.edu 1 Department of Plant Sciences, University of California, Davis, CA 95616, USA 8 Howard Hughes Medical Institute and Gordon & Betty Moore Foundation Investigator, Davis, CA 95616, USA Full list of author information is available at the end of the article variable Zn and Fe concentrations [1,2], demonstrating that improvements in these traits are possible. Zn and Fe deficiency afflict many parts of the developing world where wheat constitutes a major part of the diet, making the development of nutritionally-enhanced wheat varieties an important target for breeders tackling this problem [3]. The main source of protein and micronutrients in the wheat grain is the flag leaf and, to a lesser extent, the lower leaves [4,5]. When applied to the leaf tip, radioactively-labelled Zn is efficiently translocated to the developing wheat grain [6]. The close correlation between Zn and Fe content in the grain suggests some level of redundancy in the regulatory mechanisms used by the plant to transport these micronutrients [1]. However, the regulation of gene expression associated with nutrient © 2014 Pearce et al.; licensee BioMed Central. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Pearce et al. BMC Plant Biology (2014) 14:368 transport from leaves to grain during wheat monocarpic senescence is poorly understood. A detailed understanding of these mechanisms will be required in order to engineer wheat varieties with improved nutritional quality through biofortification [7]. Several studies in other species, including barley, rice and Arabidopsis have revealed distinct mechanisms regulating micronutrient transport in vegetative tissues, which are described below according to their sub-cellular location. Transport between chloroplast and cytoplasm Because of its importance to photosynthesis, Fe is particularly abundant within the chloroplasts, which harbor ~90% of all Fe in the leaf during vegetative development [8]. Therefore, the remobilization of Fe from the chloroplast is an important process during monocarpic senescence. In Arabidopsis a member of the ferric chelate reductase (FRO) gene family is highly expressed in photosynthetic tissues and localizes to the chloroplast membrane, suggestive of a role in the reduction-based import of Fe into the chloroplasts [9]. In rice, certain FRO genes are preferentially expressed in the leaf vasculature rather than the roots, suggesting that this may be a conserved transport mechanism [10]. Certain members of the Heavy Metal ATPase (HMA) family of transporters have been implicated in the reverse process; nutrient export from the chloroplast to the cytoplasm. In Arabidopsis, AtHMA1 localizes to the chloroplast membrane and facilitates Zn export from the chloroplast [11] and in barley, HvHMA1 facilitates both Zn and Fe export from the chloroplast [12]. Transport between vacuole and cytoplasm Additional mechanisms within the leaf exist to facilitate Fe and Zn transport between the vacuole and cytoplasm as part of a sequestration strategy, since high concentrations of either nutrient can be toxic for the plant cell. In rice, two VACUOLAR IRON TRANSPORTER genes, OsVIT1 and OsVIT2, encode proteins which are localized to the vacuolar membrane (tonoplast) and facilitate Zn2+ and Fe2+ import to the vacuole [13]. Likewise, the ZINC-INDUCED FACILITATOR-LIKE (ZIFL) genes encode Zn-transporters which are implicated in vacuole transport. In Arabidopsis, ZIF1 localizes to the tonoplast and zif1 mutants accumulate Zn in the cytosol, suggesting that these transporters promote vacuolar sequestration of Zn by facilitating its import into the vacuole [14]. However, several of the thirteen ZIFL genes recently described in rice are induced in the flag leaves during senescence [15]. This suggests that in monocots, certain ZIFL genes may also play a role in promoting nutrient remobilization during senescence. The NRAMP family of transporters appears to regulate nutrient export from the vacuole. In Arabidopsis, NRAMP3 and NRAMP4 are Page 2 of 23 induced in Fe-deficient conditions and plants combining mutations in both these genes fail to mobilize vacuolar reserves of Fe [16]. Transport from cytoplasm to phloem For their transport to the grain, micronutrients must be transported from the cytoplasm across the plasma membrane to be loaded into the phloem. This process is facilitated by members of the Yellow stripe like (YSL) and ZRT, IRT like protein (ZIP) families of membrane-bound transporters, which transport metal-chelate complexes across the plasma membrane in the leaves of several plant species [17-19]. In Arabidopsis, two Fe-transporting members of the YSL gene family were shown to be essential for normal seed development [20] and in barley, HvZIP7 knockout mutant plants exhibit significantly reduced Zn levels in the grain, suggesting that this family may also be important for nutrient loading into the phloem [21]. Because Zn and Fe ions exhibit limited solubility in the alkaline environment of the phloem, they are transported in association with a chelator [19]. Nicotianamine (NA) is one such important chelator and is a member of the mugineic acid family phytosiderophores [22]. NA biosynthesis is regulated by the enzyme nicotianamine synthase (NAS) by combining three molecules of S-Adenosyl Methionine [23], and can be further catalyzed to 2’-deoxymugineic acid (DMA) by the sequential activity of nicotianamine aminotransferase (NAAT) [24,25], which generates a 3”-keto intermediate and DMA synthase (DMAS, Figure 1) [26]. Although Zn has been shown to associate with DMA in the rice phloem [27], a recent study suggests that it is more commonly associated with NA [28]. In contrast, the principal chelator of Fe in the rice phloem is DMA [29]. It has been hypothesized that phloem transport represents the major limiting factor determining Zn and Fe content of cereal grains [30] and this is supported by several studies which demonstrate that altering NAS expression can have significant impacts on Zn and Fe grain and seed content. In Arabidopsis, plants carrying non-functional mutations in all NAS genes exhibit low Fe levels in sink tissues, while maintaining high levels in ageing leaves [31]. Conversely, NAS overexpression results in the accumulation of higher concentrations of Zn and Fe in Arabidopsis seed [32], rice grains [33,34] and barley grains [35]. Regulation of senescence and nutrient translocation Monocarpic senescence and nutrient translocation to the grain occur simultaneously, requiring a precise coordination of these two processes. This is reflected in the large-scale transcriptional changes in the plant’s vegetative tissues during the onset of senescence, as documented in recent expression studies in Arabidopsis [36,37], barley [38] and wheat [39,40]. These studies consistently identify Pearce et al. BMC Plant Biology (2014) 14:368 S-Adenosyl Methionine (SAM) NAS Nicotianamine (NA) NAAT 3”-keto intermediate Page 3 of 23 stages of monocarpic senescence in tetraploid wheat. We also identified genes that were differentially expressed within each of these stages between tetraploid WT and gpc mutants, which exhibited reduced Zn and Fe grain concentrations. We identified members of different transporter families, which were differentially regulated both during the early stages of senescence and between genotypes with different GPC alleles. Results from this study define more precisely the role of individual GPC genes in the regulation of transporter gene families in senescing leaves and identify new differentially regulated targets for Fe and Zn biofortification strategies in wheat. Results DMAS 2’-Deoxymugineic acid (DMA) Figure 1 Biosynthesis of mugienic acid phytosiderophores. The combination of three molecules of SAM to form one molecule of NA is catalyzed by NAS. NA is converted to DMA through the action of NAAT to form a 3”-keto intermediate and then by DMAS to form DMA. Adapted from Bashir et al. [26]. increased expression levels of a number of transcription factors of different classes. Particularly important roles have been identified for members of the NAC family [38,41-44]. In wheat, one such NAC-domain transcription factor, Grain Protein Content 1 (GPC1, also known as NAM1), has been shown to play a critical role in the regulation of both the rate of senescence and the levels of protein, Zn and Fe in the mature grain [44]. Originally identified as a QTL which enhances grain protein content in wild emmer (Triticum turgidum spp. dicoccoides) [45], the genomic region of chromosome arm 6BS including GPC1 was later shown to also accelerate senescence in tetraploid and hexaploid wheat [44,46,47]. A paralogous gene, GPC2 (also known as NAM2), was identified on chromosome arm 2BS, which shares 91% similarity with GPC1 at the DNA level [44]. Transcripts of GPC1 and GPC2 are first detected in flag leaves shortly before anthesis and increase rapidly during the early stages of senescence. In hexaploid wheat, plants transformed with a GPC-RNAi construct targeting all homologous GPC genes and plants carrying loss-of-function mutations in all GPC1 homoeologs, both exhibit a three-week delay in the onset of senescence as well as significant reductions in the transport of amino acids (N), Zn and Fe to the grain [5,44,46]. Therefore, GPC mutants represent an excellent tool to dissect the mechanisms underlying Zn and Fe transport from leaves to grains during monocarpic senescence. In the current study, we used RNA-seq to identify genes differentially regulated in the flag leaves during three early GPC1 and GPC2 mutations and their effect on senescence and nutrient translocation Field experiments comparing wild type (WT), single (gpc-A1 and gpc-B2), and double (gpc-A1/gpc-B2) mutants showed consistent results across the four tested environments (UCD-2012, TAU-2012, NY-2012 and NY-2013, Figure 2, Additional file 1: Figure S1 and S2). None of the gpc mutants showed significant differences in heading time relative to the WT, which is consistent with the known upregulation of the GPC genes after anthesis [44]. Both the gpc-A1 and gpc-A1/gpc-B2 mutants were associated with a significant delay in senescence relative to the WT and the gpc-B2 mutant. In the Davis field experiment (UCD-2012), these two mutants showed a 27-day delay in the onset of senescence in comparison to WT plants (Figure 2a), and consistent results were observed in field experiments carried out in Tel Aviv and Newe Ya’ar (Additional file 1: Figure S1). The differences in senescence observed between WT and gpc-B2 or between gpc-A1 and gpc-A1/gpc-B2 mutants were comparatively much smaller (Figure 2a). To test the effects of the GPC mutations on yield components in a tetraploid background, we measured thousand kernel weight (TKW) in three field environments and dry spike weight in the Davis field experiment. We detected a marginally significant reduction in TKW associated with the gpc-A1 and gpc-A1/gpc-B2 mutant genotypes (P =0.02, Additional file 1: Figure S2a). These mutant genotypes were also associated with significant reductions in dry spike weight in the Davis field experiment which was lower in both gpc-A1 and gpc-A1/ gpc-B2 mutants at 35 DAA (P <0.001) and in the gpc-A1/ gpc-B2 mutant at 42 and 49 DAA (P <0.001, Additional file 1: Figure S2b). The delays in the onset of senescence in the gpc-A1 and gpc-A1/gpc-B2 mutants relative to WT plants were associated with reductions in protein, Zn and Fe levels in the mature grain (Figure 2, b-d). Similarly, the marginal differences in senescence between WT and gpc-B2 or between gpc-A1 and gpc-A1/gpc-B2 mutants (Figure 2a) Pearce et al. BMC Plant Biology (2014) 14:368 Page 4 of 23 (a) (b) 200 WT gpc-B2 gpc-A1 gpc-A1/ gpc-B2 60 50 40 30 20 160 *** 140 *** *** *** * 120 10 0 WT gpc-B2 gpc-A1 gpc-A1/ gpc-B2 180 GPC (g Kg-1) Chlorophyll (Relative units) 70 ** H 7 22 36 42 48 54 60 66 100 72 UCD Days after anthesis (c) TAU NY (d) 80 *** *** * ** 70 60 Zn (ppm) Fe (ppm) WT gpc-B2 gpc-A1 gpc-A1/ gpc-B2 WT gpc-B2 gpc-A1 gpc-A1/ gpc-B2 50 40 *** *** 30 20 10 0 UCD TAU Figure 2 GPC mutations in tetraploid wheat result in significant delays in senescence and reductions in protein, Zn and Fe content in the grain. (a) Relative chlorophyll content of flag leaves taken from the UCD-2012 field experiment (b) GPC content of mature grains harvested from three experiments, (UCD n = 10, TAU and NY n = 4) (c) Fe and (d) Zn content of mature grains harvested from UCD-2012 and TAU-2012 experiments (n = 5). * = P < 0.5, ** = P < 0.01, *** = P < 0.001, difference when compared to WT control sample from Dunnett’s test. UCD = UC Davis 2012 experiment, TAU = Tel Aviv University 2012 experiment, NY = Newe Ya’ar research center 2012 experiment. were paralleled by the absence of significant differences in protein, Zn and Fe levels in the grain in the different field experiments (Figure 2, b-d). Similar reductions in GPC were observed across the different field experiments (Figure 2b), which ranged between 19.5% (WT vs. gpc-A1) and 13.4% (WT vs. gpc-A1/gpc-B2). Micronutrient concentrations in the mature grain for each genotype in UCD-2012 and TAU-2012 experiments are presented in Additional file 1: Table S1. Fe concentrations in the grain were significantly lower in both the gpc-A1 (20.9% mean reduction) and gpc-A1/gpc-B2 mutants (20.8% mean reduction) when compared to WT samples in both locations (Figure 2c). Zn grain concentrations were also lower for the same mutant genotypes in both locations, but the differences were significant only in the UCD-2012 experiment (Figure 2d). Interestingly, gpc-A1 and gpc-A1/gpc-B2 mutants also exhibited significantly higher grain K concentrations than in WT plants, with increases ranging between 18 and 33% (Additional file 1: Table S1). All GPC and micronutrient values are reported as the concentration within the grain, so are unaffected by the variation in TKW detected between genotypes. Taken together, these results demonstrate that a knockout mutation of the GPC1 gene alone is sufficient to delay the onset of senescence and to perturb the translocation of protein, Zn and Fe to the developing grain in tetraploid durum wheat under field conditions. The gpc-B2 mutation had no significant effect on any of these traits, even in a genetic background with no functional GPC1 genes. Evaluation of the mapping reference used for RNA-seq and overall characterization of loci expressed in each sample To identify GPC-mediated transcriptional changes associated with the onset of senescence, we carried out an RNA-seq study focusing on three genotypes; WT and Pearce et al. BMC Plant Biology (2014) 14:368 the two mutants that showed the largest differences in senescence in the previous field experiments, gpc-A1 and gpc-A1/gpc-B2. None of the plants sampled at heading date (HD), 12 days after anthesis (DAA) or 22 DAA, showed signs of chlorophyll degradation in the flag leaves or yellowing of the peduncles (Additional file 1: Figure S3, a-c), confirming that the selected time points represent relatively early stages of the senescence process. Clear differences between genotypes were apparent five weeks later (60 DAA), when the WT plants showed more advanced symptoms of senescence than either of the two gpc mutants (Additional file 1: Figure S3, d-f). This result indicates that in this greenhouse experiment, the effects of the GPC genes were consistent with those observed in the field experiments described above (Figure 1a). On average, 35 million trimmed RNA-seq reads were generated for each of the four replicates of each of the nine genotype/time point combinations included in this study (Additional file 1: Table S2, total 1.3 billion reads). Most of the reads (average 99.0%) were mapped to the reference genomic contigs generated by the International Wheat Genome Sequencing Consortium (IWGSC) using flow-sorted chromosomes arms of T. aestivum cv. Chinese Spring [48]. Since we were mapping transcripts of a tetraploid wheat cultivar, only the sequences from the A and B genome chromosome arms were used as a reference. A large proportion of the trimmed reads (average 93.4%, Additional file 1: Table S2) mapped within the 139,828 previously defined transcribed genomic loci within this reference (see Methods), suggesting that these loci provide a good representation of the transcribed portion of the wheat genome. However, only 58.5% of these reads mapped to unique locations (Additional file 1: Table S2), most likely due to a combination of the high level of similarity shared by the coding regions of A and B homoeologs (average identity = 97.3%, standard deviation = 1.2%, [49]), and the short length of the reads used in this study (50 bp). Ambiguously mapped reads were excluded from the statistical analyses described below, resulting in an average of 20.4 M uniquely mapped reads per sample. After excluding ambiguously mapped reads, only 80,168 of the genomic loci showed transcript coverage above the selected threshold for the statistical analyses (>3 reads for at least two biological replicates, within at least one genotype/time point pair, see Methods). The complete list of statistical analyses performed for these 80,168 loci is summarized in Additional file 2. Probability values for all four statistical tests are presented in this table so researchers can reanalyze the data using different statistical analyses and levels of stringency for specific sets of genes. Where available, this table also describes the highconfidence protein coding gene corresponding to each Page 5 of 23 genomic locus, derived from the recent annotation of these wheat genomic contigs [48]. Principal component analysis (PCA) of the uniquely mapped reads at each time point showed limited clustering of the samples according to their genotype at HD (Additional file 1: Figure S4a), very clear groupings at 12 DAA (Additional file 1: Figure S4b), and intermediate clustering at 22 DAA (Additional file 1: Figure S4c). The reciprocal analysis, to distinguish samples according to time point within each genotype, showed that in all three genotypes, the HD samples were more clearly separated than the two later time points (Additional file 1: Figure S4, d-f ). The clearer separation of both gpc mutants from the WT, and of gpc-A1 from gpc-A1/gpc-B2 at 12 DAA than at either HD or 22 DAA, suggests that both GPC1 and GPC2 genes have a major regulatory role at this early stage of senescence (12 DAA). Following mapping, we confirmed the genotype of each sample by analyzing pileups of reads which mapped to the genomic loci corresponding to the GPC-A1 and GPC-B2 genes. The expected TILLING mutations (G561A = W114* for gpc-A1 and G516A = W109* for gpc-B2) were confirmed in the expected mutant genotypes and were absent in all WT samples. All GPC genes showed a low number of mapped reads at HD, with significant increases at 12 DAA and 22 DAA (Additional file 1: figure S5). Approximately 3-4-fold more reads mapped to GPC1 homoeologous genes than to the GPC2 genes, a pattern which was consistent across all genotypes (Additional file 1: Figure S5). We detected no significant differences in the expression profiles of GPC-A1 and GPC-B2 between WT and gpc mutant genotypes suggesting that the mutations in these genes did not affect the stability of the transcribed mRNAs, and that neither GPC-A1 nor GPC-B2 functional proteins exhibit a feedback regulatory mechanism on their own transcription (Additional file 1: Figure S5). However, at 22 DAA, GPC-A2 expression was significantly lower in WT plants than in either gpc-A1 (P = 0.024) or gpc-A1/gpc-B2 (P = 0.004) mutants, suggesting that there may exist some GPC-mediated feedback mechanism on the regulation of GPC-A2 transcript levels (Additional file 1: Figure S5). Identification of loci differentially expressed during monocarpic senescence in WT plants Applying stringent selection criteria (significant according to four different statistical tests, see Methods), we identified 3,888 contigs which were differentially expressed (DE) in at least one pairwise comparison among sampling times in the WT genotype (Figure 3a). As expected, the comparison between HD and 22 DAA showed the largest number of DE loci (2,471), followed by the comparison between HD and 12 DAA (1703). The comparison between Pearce et al. BMC Plant Biology (2014) 14:368 (a) Page 6 of 23 Effect of timepoint in WT (b) Effect of genotype HD vs. 12 DAA (1703) WT vs. gpc-A1 (520) 852 168 61 718 12 321 72 1173 1349 504 508 HD vs. 22 DAA (2471) (c) 19 12 DAA vs. 22 DAA (1145) Effect of gpc-A1 and senescence gpc-A1 vs. gpc-A1/gpc-B2 (292) WT vs. gpc-A1/gpc-B2 (1913) (d) 37 224 Effect of gpc-B2 and senescence WT vs. gpc-A1 (520) gpc-A1 vs. gpc-A1/gpc-B2 (292) 114 43 121 147 66 219 1038 WT vs. gpc-A1/gpc-B2 (1913) 535 6 96 3068 WT senescence (3888) 1012 WT vs. gpc-A1/gpc-B2 (1913) 658 3128 WT senescence (3888) Figure 3 Overlap of DE genes (a) Between time points in WT samples, (b) Between different GPC genotype comparisons, (c) Between GPC-A1-regulated loci and senescence regulated loci and (d) Between GPC-B2-regulated loci and senescence regulated loci. 12 DAA and 22 DAA showed the lowest number of DE loci (1,145, Figure 3a). Of the loci which were significantly DE in the WT plants between HD and 12 DAA, a larger proportion were upregulated (76.2%) than were downregulated (23.8%). The reverse was true for loci DE between 12 DAA and 22 DAA, when 30.2% of loci were upregulated and 69.8% were downregulated. This suggests that during the first 12 DAA different mechanisms required to actively prepare the plant for the upcoming senescence are upregulated, which is followed by the shutdown of many biological processes and the downregulation of a large number of genes. We next determined whether any previously characterized senescence associated genes were also differentially expressed in our dataset. In a wheat microarray study, 165 annotated genes were identified which were differentially expressed during eight stages of senescence, ranging from anthesis to yellowing leaves [40]. We identified the corresponding genes within our dataset using BLAST (P ≤ 1e−5) and found that 26 (15.8%) were also significantly differentially expressed during senescence in the current study (Additional file 1: Table S3). This relatively low percent is not unexpected since our study covers only the early stages of senescence whereas the previous study covered a more extended period. A second microarray experiment in barley identified a set of genes differentially expressed between NILs divergent for a high-GPC genomic segment at 14 DAA and at 21 DAA [38]. In the leaves, 2,276 genes were upregulated in at least one of these time-points and 1,193 were downregulated. Among the upregulated genes, we identified 100 which were also significantly up-regulated during senescence, and of the down-regulated genes, 96 were also significantly down-regulated within our dataset, which used different statistical stringency criteria. The use of different technologies (microarray vs RNA-seq) and different species may also contribute to the different Pearce et al. BMC Plant Biology (2014) 14:368 sets of differentially expressed genes detected in these studies. The genes regulated by senescence in both experiments are listed in Additional file 1: Table S4. This study in tetraploid wheat supersedes our previous RNA-seq analysis in hexaploid wheat comparing the transcriptomes of WT and transgenic GPC-RNAi lines with reduced transcript levels of GPC1 and GPC2 at 12 DAA [39]. In the current study, we generated a greater number of reads, studied additional time-points, used targeted knockouts of individual GPC genes and had access to a more comprehensive wheat genome mapping reference. Among the differentially expressed genes common to both studies were three genes of biological interest selected for validation in the previous study [39]. Identification of loci differentially expressed among GPC genotypes We next identified loci which were DE between genotypes. The largest number of DE loci was detected between the WT and the double gpc-A1/gpc-B2 mutants (1,913 loci), an expected result given that this comparison includes genes regulated by both GPC-A1 and GPCB2 (Figure 3b). The comparison between the WT and the single gpc-A1 mutant, expected to detect mainly GPC-A1-regulated genes, showed a much lower number of DE genes (520 loci) than the previous comparison. A total of 321 of these loci (62%, Figure 3b) were DE in both these comparisons and are designated hereafter as high-confidence GPC-A1-regulated genes. The third comparison, between the gpc-A1 and gpc-A1/gpc-B2 mutant genotypes, expected to detect mainly genes regulated by GPC-B2, yielded a lower number of DE loci (292). Most of these loci (224 = 77%, Figure 3b) were also DE in the comparison between the WT and the gpcA1/gpc-B2 double mutant and are designated hereafter as high-confidence GPC-B2-regulated genes. There were 19 loci which were DE in all three comparisons between genotypes, and these likely represent genes redundantly regulated by both GPC-A1 and GPC-B2 genes (Figure 3b). Similarly, the 1,349 loci DE only between the WT and double gpc-A1/gpc-B2 mutants but not in the other two classes (Figure 3b), likely include loci that are redundantly regulated by both genes, but that show significant differences in expression only when mutations in both GPC paralogs are combined. To determine how these differences between genotypes were distributed in time, we made pairwise comparisons between genotypes within each of the three time points. Since both GPC1 and GPC2 expression is relatively low at HD (Additional file 1 : Figure S5), we expected to find a small number of DE loci among GPC genotypes at this time point. Indeed, only ten genes were DE between WT and the gpc-A1 single mutant, only six between WT and the gpc-A1/gpc-B2 double mutant and Page 7 of 23 19 between the gpc-A1 and gpc-A1/gpc-B2 mutants at HD. Two loci were shared between the WT vs. gpc-A1/ gpc-B2 and gpc-A1 vs. gpc-A1/gpc-B2 comparisons, suggesting they may potentially be regulated by GPC-B2 and one gene was common to the WT vs. gpc-A1 and WT vs. gpc-A1/gpc-B2 comparisons, suggesting it may be regulated by GPC-A1. These results confirm that GPC genes have only a marginal effect on the wheat transcriptome at this developmental stage. By contrast, the number of DE loci between genotypes was much greater at 12 DAA. Of the 520 loci DE between WT and the gpc-A1 single mutant, 504 (96.9%) were DE at 12 DAA and only six (1.1%) at 22 DAA. Similarly, of the 1,913 loci DE between WT and the gpcA1/gpc-B2 double mutant 1,525 (79.7%) were DE at 12 DAA, whereas only 385 (20.1%) were DE at 22 DAA. Of the 292 DE genes in the comparison between the gpc-A1 single mutant and the gpc-A1/gpc-B2 double mutant, 239 were DE at 12 DAA, whereas only 38 genes were DE at 22 DAA. These results suggest that even though GPC1 and GPC2 expression continues to rise between 12 DAA and 22 DAA (Additional file 1 Figure S5), the major effect of both these genes on the regulation of downstream genes occurs at 12 DAA. We next compared the two sets of high-confidence GPC-regulated loci with the senescence-regulated loci. A broad overlap was detected between GPC-A1-regulated and senescence-regulated loci, with 206 of the 321 (64.2%) high-confidence GPC-A1-regulated loci also DE during senescence (Figure 3c). By contrast, of the 224 high-confidence GPC-B2-regulated loci only 83 (37.1%) were also DE during senescence (Figure 3d). Surprisingly, 81% of the genes upregulated during the first 12 DAA in WT plants (1,054 genes) were no longer significant in the gpc-A1 mutant. This observation highlights the critical role of GPC1 in the activation of a large number of genes during the early stages of monocarpic senescence, possibly to prepare the plant for the upcoming senescence. Distribution of expression profiles among different genotypic classes To further analyze the loci DE during senescence, we classified them into eight classes based on their upregulation (Up), downregulation (Down) or absence of significant differences (Flat) between HD and 12 DAA, and between 12 DAA to 22 DAA (Figure 4a). Loci which were not significantly DE in either of these comparisons, but were significantly up or downregulated between HD and 22 DAA were included in the ‘Up-Up’ and ‘Down-Down’ classes, respectively. When all 3,888 loci DE during senescence in WT plants were considered (Figure 4, a-b) all eight classes were well represented with slightly higher proportions in the three classes that Pearce et al. BMC Plant Biology (2014) 14:368 Page 8 of 23 (a) (b) (c) WT senescence (3888) (d) GPC-A1 and senescence (219) GPC-B2 and senescence (96) % Up-Up % Up-Flat % Flat-Up % Up-Down % Down-Up % Flat-Down % Down-Flat % Down-Down Figure 4 Expression profiles during senescence. (a) Boxplot of log2 normalized counts for WT samples over three time points during senescence, separated according to their expression profiles in 8 classes. Classes were defined based on the existence of significant (‘Up’ and ‘Down’) or non-significant differences (‘Flat’) between time point comparisons. (b-d) Proportion of expression classes among loci DE during senescence in (b) WT (3888 loci), (c) high-confidence GPC-A1-regulated loci (219 loci) and (d) high-confidence GPC-B2-regulated loci (96 loci). Loci included in C and D are based on the intersections of the three classes shown in Figure 3, c and d. H = Heading Date, 12 = 12 days after anthesis, 22 = 22 days after anthesis. include loci upregulated between HD and 12 DAA (‘Up-Down’: 21.2%, ‘Up-Up’: 20.3% and ‘Up-Flat’: 16.9%). A different picture emerged when, among the loci DE during senescence, we considered only the high-confidence GPC-A1 (219) and GPC-B2 (96) regulated genes. In both cases the ‘Up-Down’ class was dominant, representing 63.5% and 62.5% of the DE loci, respectively (Figure 4, c and d). However, a difference between these two groups was evident in the second most abundant class; ‘Up-Flat’ in the high-confidence GPC-A1-regulated genes (24.7%), and ‘Down-Up’ in the high-confidence GPC-B2-regulated genes (26.0%, Figure 4, c and d). In both groups, the remaining six classes represented less than 12% of the DE loci. These data indicate that while both genes have their greatest effect at 12 DAA, a partial differentiation exists of the loci and processes regulated by the GPC-A1 and GPC-B2 genes. Gene ontology analysis We next used BLAST2GO to generate ‘Biological Process’ Gene Ontology (GO) terms for each locus to compare the proportions of different functional categories between loci up- and downregulated during senescence in WT and between high-confidence GPC-A1- and GPC-B2-regulated loci (Table 1). To simplify the description of these functional analyses, we first combined the eight functional categories from Figure 4a into four: upregulated loci (combining ‘Up-Up’, ‘Up-Flat’ and ‘Flat-Up’ categories), downregulated loci (combining ‘Down-Down’, ‘Down-Flat’ and ‘Flat-Down’ categories), ‘Up-Down’, and ‘Down-Up’. Among loci upregulated during senescence, we observed enrichment in transport functions and catabolism of photosynthetic proteins. Four of the top five most significantly enriched GO terms included those related to transmembrane transporter function (Table 1). By contrast, loci downregulated during senescence were enriched in functions related to biosynthetic processes, especially photosynthesis (Table 1). These results, together with the previous observation that upregulated loci were more abundant between WT and 12 DAA (76.2%) and downregulated loci were more abundant between 12 and 22 DAA (69.8%), are indicative of the early activation of catabolic enzymes and transport systems followed by the downregulation of growth promoting processes in the leaves during these two early stages of senescence. Pearce et al. BMC Plant Biology (2014) 14:368 Page 9 of 23 Table 1 Top significantly enriched ‘Biological Process’ GO terms among upregulated and downregulated genes during monocarpic senescence in wheat and in the 316 high-confidence GPC-A1- and 224 GPC-B2-regulated genes Upregulated Downregulated GPC1-regulated GPC2-regulated Accession Ontology Annotated Significant Expected P GO:0055114 Oxidation-reduction process 3168 171 87.5 2.60E-18 GO:0055085 Transmembrane transport 1622 90 44.8 2.20E-10 GO:0071577 Zinc ion transmembrane transport 20 8 0.55 3.10E-08 GO:0034220 Ion transmembrane transport 369 31 10.19 4.90E-08 GO:0006829 Zinc ion transport 24 8 0.66 1.60E-07 GO:0043562 Cellular response to nitrogen levels 14 6 0.39 1.10E-06 GO:0009064 Glutamine family amino acid metabolic process 65 11 1.8 1.50E-06 GO:0006787 Porphyrin-containing compound catabolic process 76 11 2.1 7.40E-06 GO:0033015 Tetrapyrrole catabolic process 76 11 2.1 7.40E-06 GO:0051187 Cofactor catabolic process 76 11 2.1 7.40E-06 GO:0015979 Photosynthesis 502 123 11.06 <1e-30 GO:0009765 Photosynthesis, light harvesting 76 47 1.67 <1e-30 GO:0019684 Photosynthesis, light reaction 340 69 7.49 <1e-30 GO:0006091 Generation of precursor metabolites and energy 774 78 17.06 1.70E-29 GO:0033014 Tetrapyrrole biosynthetic process 183 30 4.03 9.60E-18 GO:0015977 Carbon fixation 45 17 0.99 3.40E-17 GO:0006779 Porphyrin-containing compound biosynthetic process 161 27 3.55 2.30E-16 GO:0015995 Chlorophyll biosynthetic process 122 23 2.69 2.80E-15 GO:0033013 Tetrapyrrole metabolic process 262 31 5.77 3.20E-14 GO:0055114 Oxidation-reduction process 3168 134 69.81 5.50E-14 GO:0005385 Zinc ion transmembrane transporter activity 29 13 0.14 2.00E-23 GO:0046915 Transition metal ion transmembrane transmembrane activity 68 13 0.32 7.80E-18 GO:0072509 Divalent inorganic cation transmembrane activity 95 13 0.44 7.80E-16 GO:0046873 Metal ion transmembrane transporter activity 317 15 1.48 3.00E-11 GO:0022890 Inorganic cation transmembrane transport activity 472 15 2.21 7.20E-09 GO:0022891 Substrate-specific transmembrane transport 1016 20 4.76 6.60E-08 GO:0015075 Ion transmembrane transporter activity 911 18 4.26 3.00E-07 GO:0022892 Substrate-specific transporter activity 1132 20 5.3 3.70E-07 GO:0008324 Cation transmembrane transporter activity 647 15 3.03 4.30E-07 GO:0005215 Transporter activity 1989 26 9.31 1.90E-06 GO:0009834 Secondary cell wall biogenesis 17 2 0.03 0.00041 GO:0009832 Plant-type cell wall biogenesis 52 2 0.09 0.00382 GO:0007017 Microtubule-based process 379 4 0.67 0.00446 GO:0006812 Cation transport 948 6 1.67 0.00623 GO:0071669 Plant-type cell wall organization or biogenesis 69 2 0.12 0.00664 GO:0006811 Ion transport 1285 7 2.27 0.00701 GO:0007029 Endoplasmic reticulum organization 5 1 0.01 0.00879 GO:0015801 Aromatic amino acid transport 5 1 0.01 0.00879 GO term analysis among the 321 high-confidence GPC-A1-regulated genes showed a significant enrichment of categories similar to the patterns observed for loci upregulated during senescence, with the ten most significantly enriched terms all relating to transporter activity (Table 1). Although transporter functions were also enriched among the 224 high-confidence GPC-B2regulated genes, several unrelated terms were also enriched Pearce et al. BMC Plant Biology (2014) 14:368 in this class but not in the GPC-A1-regulated class, including genes with putative roles in cell wall biogenesis and microtubule organization. The closer similarity in GO term enrichment between senescence-regulated loci and GPC-A1-regulated genes than with GPC-B2-regulated genes is consistent with the greater overlap between senescence-regulated and GPCregulated loci (64.2% overlap for GPC-A1 vs. 37.1% overlap for GPC-B2, Figure 3, c and d) and with the relatively stronger effect of the gpc-A1 mutation on senescence and nutrient transport relative to the gpc-B2 mutation (Figure 1, a-d). Taken together, these results suggest that GPC-A1 plays a more important role than GPC-B2 in the regulation of genes controlling the early stages of monocarpic senescence in wheat. Identification and expression analysis of wheat transporter genes To categorize the wheat transporters upregulated during senescence and to determine the role of GPC1 in their regulation, we identified specific wheat homologues of Fe and Zn transporters previously characterized in other plant species and determined their expression profiles both among different time points during senescence and between GPC genotypes. Chloroplastic transporters Among genes previously known to be involved in the reduction-based import of Fe into the chloroplasts, we identified two FRO genes in Triticum aestivum (Ta), one of which, TaFRO1, was highly expressed at HD and significantly downregulated during senescence in WT plants (Table 2). By comparison, TaFRO2 expression was lower, and although its expression also fell during senescence, differences between time points were not significant. Neither gene was significantly DE among genotypes. Among genes previously known to promote the export of nutrients from the chloroplast to the cytoplasm, we identified five T. aestivum members of the Zn/Co/Cd/ Pb-transporting class of HMA genes (see phylogeny in Additional file 1: Figure S6). Two of these genes, TaHMA2 and TaHMA2-like, which showed the highest similarity to OsHMA2 (Additional file 1: Figure S6), were significantly upregulated during senescence, both showing >6-fold increases in expression between HD and 22 DAA (Table 2). Furthermore, TaHMA2-like expression was significantly reduced in both gpc mutants, implicating a role for GPC in its regulation. Two other genes, TaHMA1 and TaHMA-like1 which are both similar to OsHMA1 (Additional file 1: Figure S6), were not DE during senescence and a third, TaHMA3, was not detected at any time point in this study. Page 10 of 23 Vacuolar transporters Two VIT transporters, which promote Fe and Zn import in to the vacuole, were previously characterized in rice [13]. Both of the corresponding wheat homologues of these genes were downregulated ~4-fold during senescence, but these differences were not significant according to our stringent differential expression criteria (Table 2). Furthermore, neither gene was DE in either of the gpc mutant genotypes (Table 2). Eight wheat ZIFL genes, thought to promote vacuolar sequestration of Zn [14], were identified and annotated in this study (see phylogeny in Additional file 1: Figure S7). Two TaZIFL genes (TaZIFL2 and TaZIFL9) were expressed at negligible levels in all time points included in this study and were excluded from further analyses (Table 2). Among the six TaZIFL genes which showed higher levels of expression during senescence, TaZIFL2-like1 and TaZIFL3 were significantly upregulated during senescence while TaZIFL1 was significantly downregulated. Interestingly, although it was not upregulated during senescence, TaZIFL7 expression was significantly higher in WT plants than in both gpc mutants (Table 2). Among the genes known to promote Fe export from the vacuole to the cytoplasm, eight NRAMP genes were recently described in wheat [7]. Five of these genes showed very low levels of expression in flag leaves during the time points included in our study, suggesting that they may play more important roles during other developmental stages or in other tissues. Of the three NRAMP genes with higher expression levels during senescence, TaNRAMP3 and TaNRAMP7 both exhibited stable expression, but TaNRAMP2 was significantly upregulated, showing a ~5-fold increase in expression between HD and 22 DAA (Table 2). No significant differences among genotypes were detected for any of the NRAMP genes. Plasma-membrane transporters After being transported into the cytoplasm, Zn and Fe must be loaded into the phloem for their transport to different sink tissues, including the grain. In rice and barley, the YSL and ZIP gene families appear to play a prominent role in this process. We identified a total of 14 YSL genes within available wheat databases (see phylogeny in Fig S8), but one of these genes is likely a pseudogene (Table 2). Among the functional YSL genes, TaYSL6 and TaYSL9 were significantly upregulated during senescence and TaYSL18 was significantly downregulated (Table 2). Although not DE during senescence, TaYSL12 expression was significantly reduced in the gpc-A1/gpc-B2 mutant compared to the WT. The largest transporter gene family described in this study is the ZIP family, with a total of 19 wheat genes