Tuber Oil Contents Differed Markedly Between Yellow and Purple Nutsedge
Under our experimental conditions, new shoots from seed tubers of the two nutsedges appeared above soil at 5-9 days after sowing on April 16, 2016. New tubers began to appear from 6 to 8 weeks after shoot emergence. Both species lasted around five months for their growth and development.
Analyses of the proximate constituents of mature tubers (Fig. 1A) indicated that there were marked differences between two species in the contents of major storage reserves including starch, oil, sugar, and protein on a tuber dry weight basis (Fig. 1B), where starch was the component at highest level in tubers of two species. The most notable difference was that yellow nutsedge stored more than 25% oil of dry weight in mature tubers, whereas purple nutsedge contained less than 3% oil, indicating that there is around 10-fold difference in oil content.
Analysis of fatty acid composition of oil from mature tubers showed that yellow nutsedge predominated with oleic acid (C18:1) that accounted for more than 60% of total fatty acids, while purple nutsedge was represented with palmitic acid (16:0), C18:1, and linoleic acid (18:2) as major fatty acids, with concentrations ranging from 25% to 35% (Fig. 1C). These results indicated that significant difference also occurred in fatty acid composition of tuber oil between these two species, where purple nutsedge contained less oleic acid and more saturated fatty acids than yellow nutsedge.
To check whether there was also a difference in oil accumulation occurred in developing tubers, the changes in oil contents during tuber development were determined for two species. The oil accumulation patterns in the development period spanning around 110 days after tuber formation (DAF) were shown in Fig. 1D. The results indicated that the oil accumulation in the two types of tubers continued to increase throughout the tuber development. In all developmental stages, however, yellow nutsedge contained a significantly higher percentage of oil in tubers than purple nutsedge.
Overall, the striking differences in oil content and the fatty acid composition were present between these two types of tubers, suggesting that there existed distinct transcriptional control of oil production the two species.
Overall Level of Transcripts for Oil Production Is Higher in Yellow Nutsedge Than in Purple Nutsedge
To uncover the difference in the transcriptional control of oil production between the two species, we systematically conducted comparative transcriptome analyses of oil-related genes in developing tubers. Tuber samples at three different developmental stages (i.e., the early stage 20 DAF, the middle stage 50 DAF and the late stage 90 DAF) were used for transcript analysis.
Oil production involves the conversion of sucrose up to TAG assembly or storage, primarily including cytosolic and plastidial carbon metabolism toward pyruvate generation in plastid, fatty acid (FA) synthesis in plastid, TAG synthesis in ER and TAG storage in oil body or lipid droplet. These four major metabolic processes are related to expression of more than 400 genes (Additional file 1: Table S2 and S3). In this study, transcript levels were represented by FPKM (fragments per kilobase of exon model per million mapped reads) per protein, where multiple transcripts for genes encoding for isoforms or subunits of the same protein family were summed. Among these metabolic pathways, more than 70% of the transcripts in yellow nutsedge were associated with genes involved in TAG storage and only 8% and 3% were related to FA synthesis and TAG synthesis, respectively(Fig. 2A). In contrast, in purple nutsedge the transcripts related to carbon metabolism were most abundant, while those of TAG storage were the least.
For every metabolic pathway, transcript levels were on average higher in yellow nutsedge than in purple nutsedge (Fig. 2A). The largest difference was noted for TAG storage, for which the transcript level was more than 20-fold higher in yellow nutsedge compared to purple nutsedge. Clear difference was also present for FA synthesis, where there was over two times higher in yellow nutsedge than in purple nutsedge. Unexpectedly, there was no substantial differences in carbon metabolism or TAG synthesis between two species. A similar contrast between the two plants also occurred across the three developmental stages of tubers (Fig. 2B). Notably, transcript patterns for other lipid-related metabolic pathways were comparable in two species (Fig. 2C).
Overall, the results mentioned above suggested that transcriptional control of genes involved in TAG storage along with FA synthesis rather than TAG synthesis might be the major factors required for high oil accumulation of yellow nutsedge in relative to purple nutsedge.
Transcripts for Carbon Metabolism Toward Fatty Acid Synthesis Were Slightly Higher in Yellow Nutsedge Than Purple Nutsedge
The generation of plastid pyruvate for FA synthesis from sucrose primarily involved sucrose degradation in cytosol, glycolysis and pentose phosphate pathway (PPP) occurring in both cytosol and plastid (Fig. 3A, B).
It was found that there were only 1.4-fold higher transcripts for sucrose degradation pathway in yellow nutsedge compared to purple nutsedge (Fig. 3B), which was catalyzed either by sucrose synthase (SUS) into uridine diphosphate glucose (UDP-Glu) and fructose (Fru), or by extracellular cell wall invertase (CWINV) or intracellular neutral invertase (CINV) into glucose (Glu) and Fru (Fig. 3A). SUS genes in both nutsedge tubers were highly expressed at levels of over 2800 FPKM/protein, which were at least 15-fold higher than CINV or CWINV (Fig. 3A, Additional file 1: Table S2), implying that SUS might play an important role as the preferred enzyme in initial sucrose metabolism in nutsedge tubers. This result could support the evidence that SUS activities in plant seeds or potato tuber were significantly higher than INV activities [13, 14, 15, 16].
Similar patterns of gene expresses involved in glycolysis were also present in the two tuber tissues (Fig. 3B). Transcripts for nearly all of glycolytic enzymes in cytosolic or plastidial compartments were at slightly higher or similar levels in yellow nutsedge compared with purple nutsedge (Fig. 3A). Only expression levels for cytosol isoforms of ATP-dependent phosphofructokinase (PFK) and fructose-bisphosphate aldolase (FBA), and plastid fructose kinase (FK), which were over 2.5-fold higher in yellow nutsedge than in purple nutsedge.
Comparing the transcripts for both plastid and cytosol glycolysis revealed some conserved features between the two tuber tissues. The transcript levels were much higher for almost all genes involved in cytosolic glycolysis than their counterparts for plastidial glycolysis (Fig. 3C), except for genes encoding for phosphoglucomutase (PGM) and glucose-6-phosphate isomerase (GPI), where transcripts were distributed in somewhat balanced manner between the two glycolytic pathways. For a complete glycolytic pathway in the two tubers, the glycolytic genes involved in latter steps associated with seven downstream glycolytic pathways (from fructose-bisphosphate aldolase (FBA) to pyruvate kinase (PK)) were overall more abundantly expressed in relative to those in early steps. Notably among these genes, those encoding for cytoplasmic glyceraldehyde-3-phosphate dehydrogenase (GAPDH), FBA and enolase (ENO) were significantly highly expressed (>1,800 FPKM). As a result, these data suggested that the cytosolic glycolysis metabolic pathway, particularly the latter steps, was highly active and might produce more carbon precursors toward FA synthesis in tubers. The gene expression patterns of these two species resembled those observed in heterotrophic non-green oil seeds such as castor, safflower and sunflower [17], but differ from those in photoheterotrophic green seeds such as Arabidopsis, rapeseed and soybean [18] and oil-rich mesocarps of oil palm as well avocado that displayed more balanced distribution of transcripts between cytosol and plastid glycolysis [19, 20, 21].
Pentose phosphate pathway, a glycolysis bypass process, has been shown to provide carbon sources for pyruvate generation, which also occurred in both cytosol and plastid [22]. However, the overall transcripts for this pathway in cytosol or plastid was similar between the two species (Fig. 3B).
Significant Difference of Transcripts for Plastid Rubisco Bypass Is Present Between Two Species
Glycolysis is also bypassed by the production of 3-phosphoglycerate (3-PGA) catalyzed by ribulose-1,5-bisphosphate carboxylase/oxygenase (RubisCO), a key enzyme for fixing carbon dioxide. This process without full Calvin cycle is called Rubisco bypass or shunt [23]. Intriguingly, the expression of genes encoding for RubisCO orthologs also appeared in two nutsedge tubers. In this study, six unigenes encoding for RubisCO small subunit (RbcS) and one for RubisCO large subunit (RbcL, ATCG00490) were detectable to express during tuber development, though RbcL was merely slightly transcribed (Fig. 4A; Additional file 1: Table S2). The presence of RubisCO genes, particularly RbcS in nutsedge tubers is surprising, since tubers are non-green and non-photosynthetic underground tissues that require little or no light for their development and growth. It is unclear why these RbcS proteins are maintained in the non-photosynthetic tubers, but one can expect that this type of RbcS may be evolved to accomplish different functions other than Calvin cycle occurred in photosynthetic tissues. Orthologous of tuber RbcS were also found in other non-photosynthetic tissues [24, 25]. This type of RbcS was first identified in secretory cells of glandular trichomes of Nicotiana tabacum [26] and therefore named as T-type RbcS. So far, the T-type RbcS was found to almost entirely absent in photosynthetic tissues, but essentially expressed in non-photosynthetic tissues of diverse plant species, such as Oraza sativa (leaf sheath, culm, anther, and root central cylinder), Setaria italica (seed), Solanum lycopersicum (stamen, pistil, and green fruit), Lotus japonicus (root, nodule, seed, and various floral organs), Vitis vinifera (mature leaf and green berry), and Selaginella moellendorffii (rhizome and root) [25]. A phylogenetic analysis (Fig. 4B) based on the deduced amino acid sequences revealed that T-type RbcS homologues underwent an evolutionary separation from the photosynthetic relatives, implying that their function may shift when photosynthesis is faint or lost. It was suggested that this type of RbcS might adapt Rubisco to particular environment, for example high CO2 concentration generated by intense metabolic pathways in non-photosynthetic tissues that are less permeable to gas exchange [27].
In this study, the RbcS genes were transcribed at low levels in purple nutsedge. In contrast, they were abundantly expressed in yellow nutsedge and displayed up-regulation patterns during tuber development (Fig. 4A), with an average transcript level more than 12 times higher than that of purple nutsedge (Fig. 3A). A recent study also showed that the ortholog of RbcS in potato starchy tuber displayed much more abundant transcripts in oil-accumulated transgenic lines expressing Arabidopsis WRI1 gene than in the control [28]. These results might reinforce the enzyme assay of the silique of rapeseed, which revealed that high oil content was associated closely with enhanced RbcS expression levels and its activities [29]. Evidence has shown that changes in the RbcS transcript abundance were directly correlated with the changes in Rubisco level [30, 31]. High expression of RbcS genes possibly reflected the high CO2 environment [32] and was probably associated with the ability of capturing CO2 resulting from the conversion of malate to pyruvate or pyruvate to acetyl-CoA occurring in the plastid catalyzed by NADP-dependent malic enzyme (ME) or pyruvate dehydrogenase complex (PDHC), respectively [15]. Previous studies demonstrated that RubisCO bypass skipping Calvin cycle for CO2 recapture in green oil-rich seeds brought about less loss of carbon as CO2 and produced more 3-PGA, thus improving carbon conversion efficiency toward for fatty acid synthesis [23, 32, 33, 34, 35].
Transcripts for Plastid Malate and Pyruvate Metabolism Are Strikingly Distinct Between Two Species
Pyruvate is the important carbon precursor required for plastid fatty acid synthesis. It can be generated either from phosphoenolpyruvate (PEP) catalyzed by PK or pyruvate phosphate dikinase (PPDK), or through the malate metabolism involved in malate dehydrogenase (MDH) and ME that catalyze the sequential production of malate and pyruvate from oxaloacetate (OAA) (Fig. 3A). In this study, the abundant expression of genes encoding for MDH, ME, and PK occurred in both cytosol and plastid, and the transcript level of MDH plus ME was comparable to that of PK (Fig. 4B), implying that the malate metabolism may be as important as PK in pyruvate generation for fatty acid synthesis in nutsedge tubers. Previous studies indicated that malate was a major substrate for fatty acid synthesis in non-green seeds of safflower, castor bean, and sesame [36, 37, 38].
Higher expression of MDH, ME, and PK was noted in the cytosol than in the plastid (Fig. 4B), suggesting that the cytosolic pyruvate generation might be more prominent in two species, and malate and pyruvate produced in the cytosol could be transported to the plastid for subsequent pyruvate metabolism for fatty acid synthesis.
One interesting aspect from this comparison was that transcripts for cytosolic MDH, ME and PK were similar in both nutsedge species (Fig. 4C). By contrast, transcript levels for plastid counterparts, along with PPDK, were over two times higher in yellow nutsedge than in purple nutsedge. This suggested that plastid malate and pyruvate metabolism was more active and might produce more carbon source required for fatty acid synthesis in yellow nutsedge relative to purple nutsedge. Therefore, our results implied that up-regulation of genes involved in plastid malate and pyruvate metabolism might play an important role in providing pyruvate for high oil synthesis.
Transcripts for Fatty Acid Synthesis Enzymes in Plastid Were More Abundant in Yellow Nutsedge than in Purple Nutsedge
At least fourteen proteins required for de novo fatty acid synthesis from pyruvate in the plastid were all detectable to transcribe in two species (Fig. 5A). Among these proteins, PDHC, acetyl-CoA carboxylase (ACCase), and acyl-carrier protein (ACP) were more abundantly expressed than any other enzymes (Fig. 5B), implicating the important roles of them played in fatty acid synthesis. The overall transcripts for the three proteins accounted for over 50% of the total fatty acid synthesis gene expression at each stage of tuber development. Similar phenomena were also observed in developing oil-rich seeds and mesocarps [17, 19, 21].
Almost all proteins were transcribed at comparatively higher levels in yellow nutsedge, coinciding with high oil accumulation in its tubers. Overall, transcript levels for these plastidial proteins were on average 2.5-fold higher in yellow nutsedge than in purple nutsedge (Fig. 2C). Significant individual differences were represented by ACCase, hydroxyacyl-ACP dehydratase (HAD), stearoyl-ACP desaturases (SAD), and acyl-ACP thioesterase A (FATA), for which their transcripts were more than 3-fold higher in yellow nutsedge as compared with purple nutsedge (Fig. 5A), suggesting that the metabolic pathways catalyzed by the four enzymes perhaps were much more active in yellow nutsedge.
SAD and FATA are two important enzymes controlling levels of unsaturated fatty acids. Typically, in oil seeds and fruits that rich in unsaturated fatty acids, transcript levels of SAD along with FATA were higher than that of FATB [17, 19, 21] (Fig. 6A). In yellow nutsedge, SAD or FATA were much abundantly transcribed in developing tubers as compared to purple nutsedge (Fig. 6B). In particular, FATA genes were up-regulated during tuber development and expressed at more than 10-fold higher levels in yellow nutsedge than in purple nutsedge at tuber maturation, correlating with their fatty acid profiles of oil (Fig. 1C). It is noteworthy that FATA expression was more than 3-fold higher relative to FATB in yellow nutsedge, whereas it was lower or comparable to that of FATB in purple nutsedge. Intriguingly, high ratio of FATA to FATB transcript was also observed in other plant oil storage tissues rich in oleic acid or its derivatives, such as oil seeds of rapeseed and castor bean [17] as well as hickory [39], and oil fruits of avocado [21] and olive [40, 41]. By contrast, the transcript ratio of FATA/FATB is only 0.1, 0.2, and 1.0, respectively, for seeds of Arabidopsis [17] and soybean [42], and mesocarp of oil palm [19] that contain low levels of oleic acid. Altogether, the expression patterns of SAD and FATA genes in two tubers might reflect the fatty acid composition, and the ratio of FATA to FATB expression was likely correlated with the content of C18:1.
Transcripts for Most TAG Synthesis Genes in Endoplasmic Reticulum Were Similar or Less in Yellow Nutsedge Than in Purple Nutsedge
In contrast to transcript patterns for plastidial fatty acid genes, most TAG synthesis genes in endoplasmic reticulum were expressed at similar or lower levels in yellow nutsedge as compared to purple nutsedge (Fig. 7A). For example, phosphatidic acid phosphohydrolase (PAP), phospholipid:diacylglycerol acyltransferase (PDAT), lysophosphatidylcholine acyltransferase (LPCAT) and D12-oleate desaturase (FAD2), an enzyme responsible for synthesis of C18:2 fatty acid, displayed similar expression patterns between yellow and purple nutsedge, while lysophosphatidyl acyltransferase (LPAAT) and two enzymes that catalyze fatty acyl exchange between phosphocholine (PC) and diacylglycerol (DAG) through "acyl exchange/editing" processes [43], cytidine-5-diphosphocholine:diacylglycerol cholinephosphotransferase (CPT) and phosphatidylcholine:diacylglycerol cholinephosphotransferase (PDCT), had 2- to 5-fold lower transcripts in yellow nutsedge than in purple nutsedge. It was demonstrated that acyl editing under the reversible action of PDCT and/or CPT was the major mechanism directing flux of PC-derived polyunsaturated fatty acids such as C18:2 and C18:3 into TAG synthesis [43, 44]. Mutation of PDCT gene was shown to bring about a significant decrease in contents of C18:2 and C18:3 in seed oil, with a concomitant increase in C18:1 content [43]. Heterologous expression of flax PDCT in Arabidopsis was indicated to increase the levels of C18:2 and C18:3 of seed TAG, while at the same time decrease the proportion of C18:1 [45]. Intriguingly, coinciding with the down-regulation expression of PDCT and CPT genes in yellow nutsedge, the contents of C18:2 and C18:3 of tuber oil were lower as compared to purple nutsedge (Fig. 1C).
Two exceptions were noted for glycerol-3-phosphate acyltransferase (GPAT9) and diacylglycerol acyltransferase (DGAT) that possessed 2.5- and 3.4-fold higher transcripts in yellow nutsedge than in purple nutsedge, respectively (Fig. 7B). DGAT is the key enzyme that catalyzes the final step of acyl-CoA dependent TAG synthesis [46]. In the two species, transcripts were much higher for DGAT2 than for DGAT1, suggesting a more prominent role of DGAT2 than DGAT1, and DGAT2 may be a key mediator in tuber oil production. Our recent study of DGAT1 and DGAT2 functional analysis has provided an evidence to support this hypothesis [6].
It was noted that PDAT, an enzyme responsible for the transfer of fatty acyl moiety from PC to DAG destined to TAG synthesis, and an ortholog of Arabidopsis thaliana DGAT3 (AT1G48300), the soluble cytosolic enzyme that might catalyze TAG synthesis using cytosolic acyl-CoA pool [47], were also detectable to have more abundant transcripts than DGAT1 during tuber maturation (Fig. 7C), but displayed similar expression patterns between the two species. Furthermore, transcripts for DGAT2, DGAT3 and PDAT enzymes in yellow nutsedge were all up-regulated during tuber maturation, coinciding with tuber oil accumulation. Taken together, our data implicated the important roles of DGAT2, DGAT3, and PDAT rather than DGAT1 played in transcriptional regulation of TAG synthesis in the nutsedge tubers.
Great Transcriptional Divergence of TAG Storage Genes Between Two Species
Similar to oil seeds of plants [17], oil-tuber of yellow nutsedge were represented by the abundant expression of a large number of genes encoding for seed-like oil body proteins such as oleosin (OBO), caleosin (CALO), steroleosin (STERO), oil body associated protein (OBAP), and seed lipid droplet protein (SLDP), with OBO transcript being the most abundant (Fig. 8A, B; Additional file 1: Table S3). In addition, their expression levels increased constantly during tuber maturation, consistent with tuber oil accumulation. However, large difference was noted for these proteins between two species, for which their transcripts were over 6- to 160-fold higher in yellow nutsedge. Similar contrast was also detectable across all the developmental stages. Therefore, these results might imply the importance of these oil body proteins particularly oleosin in stabilizing TAG and producing high oil content in yellow nutsedge. Previous studies have shown that oleosin was accumulated in a coincident manner with TAG accumulation and the abundant expression of oleosin was associated with relatively high oil content in seeds [48, 49, 50, 51, 52,53]. For example, overexpression of OBO genes in transgenic plant seeds increased oil content by up to 46% as compared to the non-transgenic controls [54, 55].
The high transcripts of oil body proteins in tubers of yellow nutsedge were in sharp contrast to the case in non-seed oily mesocarp tissues of olive, oil palm and avocado, where these structural proteins were poorly transcribed and considered less contribution to TAG storage or assembly in oil mesocarp tissues [19, 21, 56, 57]. Indeed, other lipid droplet protein such as lipid droplet-associated protein (LDAP) were found to display plentiful transcripts in these oil mesocarps [19, 21].
Intriguingly, a number of genes encoding for LDAP (At3g05500), LDAP-interacting protein (LDIP, At5g16550) and lipodystrophy protein (SEIPIN1, AT5G16460; SEIPIN2, AT1G29760) were also detectable to moderately express in developing tubers of two species (Fig. 8A, B; Additional file 1: Table S3). The transcript levels for LDAP or LDIP were even higher than that of STERO. In addition, the temporal transcript patterns of the three proteins were similar to those of seed-like oil body proteins and in accordance with oil accumulation during tuber development. However, transcripts for the three proteins showed no substantial differences between yellow and purple nutsedge.
It is noteworthy that among these proteins, OBO was most highly expressed in yellow nutsedge, followed by CALO and LDAP in this descending order, similar to the case in oil seeds where OBO proteins are most abundant on oil body [51]. In purple nutsedge, however, LDAP exhibited highest transcripts, which were five- to sixty-fold abundant than those of other oil body proteins.
Taken together, our results described above indicated that expression patterns of oil body proteins involved in TAG storage were positively associated with oil accumulation, but transcriptional control of these proteins was significantly different between the two species.
WRI1 and ABI3 Show Much Higher Transcripts in Yellow Nutsedge Than in Purple Nutsedge
WRINKLED1 (WRI1) is an important transcriptional regulator in controlling oil accumulation in aboveground oil-rich seeds and fruits of plants, and has been well functionally characterized particularly in the model plant Arabidopsis thaliana [58, 59]. In the two nutsedge species, WRI1 and its multiple orthologs, WRI2 (AT2G41710), WRI3 (AT1G16060), and WRI4 (AT1G79700), belonging to the APETALA2-ethylene responsive element binding protein (AP2/EREBP) family, were all detectable to express in tubers, but they displayed low transcript levels (<25 FPKM) (Fig. 9A). Particularly, WRI3 and WRI4 were only barely expressed, suggesting that they are unlikely to participate in the control of oil production in tubers. This may support the fact that in Arabidopsis, only WRI1 can activate fatty acid synthesis in oil seeds for oil production, and WRI3 and WRI4 are required for cutin synthesis in floral and stem tissues [60]. Although WRI2 transcript levels were relatively higher compared to WRI3 and WRI4, they were comparable between yellow and purple nutsedge. It has been shown that WRI2 was unlikely to be associated with fatty acid synthesis [60].
A significant difference between two species was noted for WRI1 ortholog, which showed on average 14.3-fold higher transcripts in yellow nutsedge over purple nutsedge (Fig. 9A; Additional file 1: Table S3). A similar great contrast was also observed between oil palm and date palm, in which oil palm mesocarp showed 57-fold higher expression of WRI1 orthologue [19]. In yellow nutsedge, the expression level of WRI1 was increased with tuber development (Fig. 9B), which was in accordance with oil accumulation pattern in tubers. Furthermore, the temporal expression pattern of WRI1 matched that of its potential targets such as PDH-β (AT2G34590), BCCP1 (AT5G16390), FAB2 (AT2G43710), FATA (AT3G25110) and FATB (AT1G08510).
Unlike oil seed tissues, the maturation master regulators that directly or indirectly control the expression of WRI1, such as LEAFY COTYLEDON (LEC1 (AT1G21970) and LEC2 (AT1G28300) and FUSCA3 (FUS3, AT3G26790) [22], were not detectable in two nutsedge tubers, as in oil mesocarps of oil palm and avocado [19,21], supporting the previous report that FUS3-type proteins are present only in seed tissues while LEC2-like proteins only appear in dicot plants [61]. Lack of these seed-like regulators suggests that the regulation of WRI1-related transcriptional network in non-seed tissues is different from that of oil seed tissues.
It is noteworthy that in yellow nutsedge tuber, an ortholog of ABSCISIC ACID INSENSITIVE 3 (ABI3, AT3G24650), the member of B3 domain superfamily, was expressed at comparable levels to WRI1, and both of them showed similar temporal expression patterns (Fig. 9A, B). Similar to WRI1, ABI3 ortholog was significantly up-regulated in yellow nutsedge as compared to purple nutsedge. Genes that share similar expression patterns are likely to interact with each other and have regulatory relationships of functionally importance [62]. In this respective, however, it is unclear whether ABI3 is the upstream regulator of WRI1 and controls WRI1 expression in the nutsedge tubers as in plant oilseeds. Recent study indicated that ABI3 played an important role in regulating plant oil accumulation, which might be independent from WRI1 and LEC2 regulation networks occurred in oil-rich seed tissues [63].
To determine the relevant relations between WRI1 and ABI3 in regulating oil accumulation in oil tuber, a gene co-expression network was constructed and analyzed using the method of weighted gene co-expression network analysis (WGCNA) [64, 65]. The produced network indicated that these two transcriptional factors were classified as hub genes within a same cluster or module in the co-expression network (data not shown). An interesting finding from the co-expression analysis is that ABI3 and WRI1 are co-expressed in concert with the oil genes involved in carbon metabolism, fatty acid synthesis, TAG synthesis, and TAG storage pathways (Fig. 10A; Additional file 1: Table S4). Our analysis result of WRI1 in concerted regulation with the genes related to carbon metabolism, fatty acid synthesis, and TAG synthesis in yellow nutsedge tuber is similar to recent reports [66, 67]. These oil-related target genes are co-regulated by both of ABI3 and WRI1, indicating an overlapping set of targets for the two transcription factors. It was also shown that transcriptional regulation of ABI3 and WRI1 is closely interconnected and WRI1 is one of direct targets for ABI3, suggesting that ABI3 is the regulator of WRI1 in yellow nutsedge, in contrast to oil-rich mesocarps of oil palm and avocado where WRI1 is not under the control of ABI3 [19,21]. In oil palm, WRI1 transcriptional activation can be activated by three ABA-responsive transcription factors, NF-YA3, NF-YC2 and ABI5 [68].
It is noted that a large number of genes encoding for late embryogenesis abundant (LEA) proteins, which were charactered well as under control of ABI3 [69,70,71], were also identified as potential targets of ABI3 and WRI1, as shown in the co-expression network (Additional file 1: Table S4). Much higher transcripts for these LEA proteins in yellow nutsedge than in purple nutsedge reflect the enhanced desiccation tolerance of yellow nutsedge (Fig. 10B), which allows tubers can be stored for long time [8].
Collectively, these results reinforce the fact that WRI1 and ABI3, as the master regulators of oil accumulation, not only regulate gene expressions involved in fatty acid synthesis, but also control the expression of genes encoding for seed-specific oil-body storage proteins [53, 63, 72, 73]. Therefore, WRI1-related regulation network of oil production in yellow nutsedge is most likely to differ either from oil seeds or oil fruits, possibly tuber-specific.