The process of flower abscission in yellow lupine was investigated already in late 1950s. Those studies demonstrated a significant influence of the flower's location in the inflorescence (Van Steveninck, 1959), the application of various substances, i.e., IAA and TIBA (2,3,5-triiodobenzoic acid, IAA transport inhibitor), and defoliation (Van Steveninck, 1958, 1959) on the degree of flower abortion. Lately, this issue has been revived since it became possible to study flower abscission in L. luteus at the molecular level, and several papers have been published (Frankowski et al., 2015; Wilmowicz et al., 2016).
This study represents the first deep-sequencing analysis performed in yellow lupine and focused on flower and fruit abscission. Our analyses allow us to investigate not only genes involved in the molecular mechanism of abscission (in pedicels), but also to compare organs that do and do not drop off the plant, which provides an opportunity to describe the probable causes of abscission. According study conducted on other plant species (Ascough et al., 2005; Corbacho et al., 2013; Estornell et al., 2013; Kim et al., 2015; Sundaresan et al., 2016) we analyzed obtained DEGs with special attention paid to the genes related to hormone and cell wall functioning, and we performed metabolic pathway analyses.
Transcriptomic changes related to flower and pod abscission in yellow lupine
Cell wall related DEGs are associated with organ abscission but also with their development
A number of genes regulate the functioning of the plant cell wall. Changes in their expression are associated with aging (Vetten and Huber, 1990; Han et al., 2016), organ abscission (Kim et al., 2015; Roongsattham et al., 2016), development and maturation of fruits (Brummell et al., 2004; Han et al., 2016; Song et al., 2016; Giné-Bordonaba et al., 2017), and organ growth and development (Gunawardena et al., 2007).
Our RNA-Seq analyses of L. luteus generative organs showed bidirectional changes in the expression of the cell wall-related genes, probably associated not only with the ongoing process of abscission but also with progressive development of the organs that are not dropped (Table 2). In a flower transcriptome comparison, we identified 27 DEGs involved in cell wall function. Most of them were up-regulated in FNAB, and only 5 were down-regulated. This is probably related to the inhibition and continuation of growth in flowers being dropped and those remaining on the plant, respectively. This supposition is supported by two facts: (1) non-abscision lupine flowers at this stage of development showed the first signs of ovary growth (data not shown) and (2) pollination in tomato developing flower caused increase in the transcript levels of cell expansion-related genes like SlEXPA5, SlPEC, and SlXTH1 (Vriezen et al., 2007; de Jong et al., 2011). Our results confirm our hypothesis that the abscising flowers are not fertilized, probably because they stopped developing properly even before pollination. In contrast, FNAB at this stage of development showed the first signs of ovary growth that required up-regulation of some cell wall modification-related genes. We strongly believe, that the lack of fertilization is caused by cues triggered from mature flowers on the first whorl and axillary flowers, that elicit modification of expression of specific genes during development of immature flowers. Verification of this hypothesis and identification of these cues need further study.
In our study, the samples were collected at the step of abscission activation (Kim, 2014) (visible lignin accumulation in AZ, Figures 2G,K), where biochemical changes in the cell walls within the AZ are crucial. Many studies (e.g., del Campillo et al., 1988; Lashbrook and Cai, 2008; Roongsattham et al., 2012; Estornell et al., 2013; Tsuchiya et al., 2015) indicate the importance of PG, CEL, XTH, and EXP genes in organ abscission. Our RNA-Seq analyses showed that in flower pedicels many genes that encoded enzymes responsible for cell wall and middle lamella degradation and remodeling factors, which are the main targets at the late abscission stages, were differently expressed. These included EXP, XTH, EDG/CEL, PG, BGal, PME, PAE, PEC genes (Table 2 and Table S4), which were previously demonstrated to be highly expressed in the AZs of a large number of abscising organs (Lashbrook et al., 1994; del Campillo and Bennett, 1996; Cho and Cosgrove, 2000; Taylor and Whitelaw, 2001; Roberts et al., 2002; Ogawa et al., 2009; Meir et al., 2010). However, Tucker et al. (2007) discovered that in the soybean leaf AZ most of the cell wall-related genes displayed increased expression, while some of them showed no change or were down-regulated (e g., EXP1, PG6, PG7, PG16, and CEL11). The authors suggested that different gene products may play very specialized roles in cell wall modifications and down-regulation of some of them may be necessary for the optimal functioning of the others (Tucker et al., 2007). In our case, the identified DEGs showed diverse expression patterns (they were either up- or down-regulated), which may be also related to the fact that we constructed the libraries by taking whole flower pedicels containing not only AZ cells, but also a large number of remaining stalk cells. Additionally, we compared pedicels of growing flowers. The identified DEGs may also be associated with continuation of flower development, which involves the strengthening of pedicels, a further development of vascular bundles, etc. However, active AZ-specific genes (Estornell et al., 2013), namely CEL1 (c33456_g1), QRT2 (c56936_g4), XTH2 (c48301_g1), and BGAL1 (c59734_g1), were highly expressed in FPAB (Table S4) which indicate that in L. luteus organ abscission process are involved similar cell wall modification enzymes as in other plant species.
Pods are shed at the early stages of development (at a length of approximately 1 cm) (Figure 2) (Van Steveninck, 1958), and their location in the inflorescence does not inevitably determine their fate. Pods fall off from higher whorls, but some also from the lowest ones, where most of the pods remain on the plant. They probably compete in some way—the first pod to set has the best chance of achieving full development, while the last one to do so is dropped. Presumably, the signal from older pods causes developmental arrest of younger ones and induce organ abscission. A similar regulation mechanism of shedding fruits is suggested in apple (Botton et al., 2011). In pods of lupine studied here, 76 DEGs were attributed to cell wall metabolism and most of them were over-expressed in the abscising pods that indicate extensive degradation of cell wall structure in these organs (Table 2 and Table S4).
Hormonal metabolism and signaling—related genes involved in lupine organ abscission
Plant hormones are involved in regulating all developmental processes in plants (Arteca, 1996) and a large body of evidence indicates that hormones are the critical regulators of organ abscission (for a review see: Kim, 2014; Sawicki et al., 2015). Our analyses of DEGs showed that in yellow lupine genes involved in the auxin and gibberellins metabolism and action presented greatest expression changes, followed by changes in the expression of genes related to the functioning of ABA and ethylene. There were also differences in JA, CK, BR, SA, and STK metabolism.
Organs abscission in yellow lupine is associated with intensive changing of auxin catabolism and signaling
Our RNA-Seq analyses indicate that the process of generative organ abscission in yellow lupine is associated with significant changes of auxin balance at the level of catabolism, perception, transport, and response to the phytohormone. The expression of genes encoding auxin biosynthetic enzymes does not differ between the generative organs of L. luteus that are shed and those that are not (except for one unigene in flowers). Closer look at the DEGs designed as involved in regulation of auxin availability (metabolism and conjugation) and transport suggest, that during organ abscission there is a huge rebuilding of auxin patterning. In falling flowers this is achieved by inhibition of synthesis, but also break-down and transport of this hormone. In pedicels of falling flowers the pattern probably changes by decreased auxin catabolism and changed transport. In falling pods auxin is probably less available because it is conjugated with proteins. Or it is rather released from conjugates in non-abscission pods.
In the flowers, we identified 11 DEGs involved in auxin signaling (Table 3 and Table S5). Flowers that remain on the plant over-express gene encoding key enzyme in IAA biosynthesisYUCCA8 and DAO, essential for auxin catabolism and maintenance of auxin homeostasis in reproductive organs (Zhao et al., 2013; Panoli et al., 2015). Analyses of auxin biosynthesis mutants (Cheng et al., 2006) indicate that localized IAA synthesis is critical for proper gynoecium morphogenesis in A. thaliana (for a review see: Hawkins and Liu, 2014). Up-regulation of YUCCA8 in FNAP suggests that IAA biosynthesis is also required for proper flower development in yellow lupine. Pollen grains of the Arabidopsis dao mutant did not germinate neither in vitro nor in vivo on the wild-type stigma, and no seeds were produced, which indicates that DAO plays a role in regulating anther dehiscence and pollen grain development (Zhao et al., 2013). This suggests that in falling lupine flowers, where DAO in less expressed, pollen grains germinate less effectively and in consequence fertilization does not occur, which is the cause of flower abscission. The gene encoding basipetal auxin transporter TORNADO 2 (TRN2) was up-regulated in FNAB. In Arabidopsis the auxin pattern resulting from TRN2 activity enables growth and organ organization by cell differentiation (Cnops et al., 2000; Cnops, 2006). Our finding indicate that flower abscission is correlated with inhibition of IAA biosynthesis in one area and simultaneous inhibition of IAA break-down in the other area, and suppression of its transport. This may be caused by cues sent by older flowers, but further study is needed to verify this hypothesis and find possible countermeasures, to increase the number of maintained flowers on the plant. Among the many DEGs analyzed in the flower transcriptomes comparison, two DEGs interpreted as related to IAA signal transduction are worth mentioning: ARF4 and ARF7, up and down-regulated in FNAP, respectively (Table S5). An analysis of tomato SlARF7 indicate that transcript level of this gene decreased after pollination and transgenic plants with decreased SlARF7 expression formed parthenocarpic fruits (de Jong et al., 2009). In our RNA-Seq analysis, the ARF7 mRNA level was higher in the abscising flowers, thus supporting our hypothesis that abscising flowers of L. luteus are unfertilizated. DEGs presenting similarity to ARF4 are present in all comparisons. In the flowers, ARF4 is over-expressed in FNAB. There is no clear evidence that ARF4 is involved in organ abscission. As genetic analyses indicate, ARF4 have functions in leaf and floral organ patterning and specify abaxial cell identity (Pekker et al., 2005). In tomato, SlARF4 expression is high in flowers and young fruits, and decreases during fruit maturation and ripening (Sagar et al., 2013a,b). We suggest that ARF4 in yellow lupine could be associated with auxin regulation of development by way of symmetric cell division and organ polarity control.
Ethylene and auxin (IAA) are important regulators of abscission and the balance between these phytohormones determines where and when separation takes place (Ascough et al., 2005; Estornell et al., 2013). Most of the unigenes associated with auxin signaling found in this study were identified in the pedicels of flowers (44 out of 70) (Table 4 and Table S6). Two of these DEGs are responsible for auxin break-down: DAO, essential for IAA oxidation, and IAMT1, involved in formation of MeIAA. MeIAA demonstrates a different activity, as it more easily forms conjugates with proteins and carbohydrates and as a nonpolar molecule probably diffuses through membranes (Qin et al., 2005). Both DEGs were up-regulated in FPNAB. Three of the four DEGs encoding auxin transport proteins were up-regulated in FPNAB, which presumably resulted in a change in auxin distribution. LAX4 and LAX5, encoding the auxin influx carrier (Bainbridge et al., 2008; Petrášek and Friml, 2009; Vanneste and Friml, 2009) proteins were over-expressed in lupine FPNAB. Auxin efflux transporter BIG that controls elongation of pedicels and stem internodes through auxin action in Arabidopsis (Gil et al., 2001; Yamaguchi et al., 2007) showed a contrary tendency. Also, the expression of the gene encoding PINOID, a kinase that phosphorylates PINs thus leading to their internalization into cells (Petrášek and Friml, 2009), was reduced. We conclude that in FPAB: (i) IAA break-down was reduced, (ii) the amount of its form that could diffuse out of cells was lower, (iii) auxin influx was enhanced and efflux was inhibited, (iv) and polar auxin transport was reduced. This is consistent with literature data that in poplar leaves prior to cell separation during abscission, an auxin maximum in AZ is formed (Jin et al., 2015) and that reduction of polar auxin transport increases sensitivity of the tissue to ET, which stimulates production of hydrolytic enzymes resulting in organ detachment (Estornell et al., 2013). This indicates, that in yellow lupine the mechanism of organ abscission is similar to that of other plants.
The list of DEGs identified in the pedicels was rich in genes encoding other elements of auxin signal transduction: two encoded down- and up-regulated receptors (F-box protein SKP2A and auxin-binding protein ABP19a, respectively), three encoded down-regulated auxin response factors (ARF4, ARF7 and ARF19), and as many as 15 encoded up- or down-regulated AUX/IAA proteins (Table 4 and Table S6). In Arabidopsis, SKP2A is an F-box protein that binds auxin and connects its signaling with cell division (Jurado et al., 2010). Over-expression of SKP2A in pedicels of abscising flowers in yellow lupine is probably associated with the auxin-dependent regulation of cell division in the AZ. ABP receptors are known to regulate many processes. Upon auxin binding, they restrict internalization of PINs by inhibiting clathrin-mediated endocytosis (Robert et al., 2010; Grones et al., 2015), thus they are involved in regulating polar auxin transport (Effendi et al., 2011). They also regulate the expression of genes encoding cell wall remodeling proteins, such as EXP, XTH and pectin (Paque et al., 2014). This suggests that during the growth of the pedicel of a flower remaining on the plant, ABP may be engaged in the regulation of cell growth and PIN-mediated auxin patterning. Among auxin-related DEGs in flower pedicels we also found ARF4 (one DEG up-, another down-regulated in FPNAB), ARF7 and ARF19 (both down-regulated in FPNAB). As mentioned before, ARF4 is probably involved in regulating polar cell division and cell wall modification. Genetic analysis of Arabidopsis mutants suggests that ARF7 and ARF19 are involved in abscission of floral organs (Estornell et al., 2013). In NGS data analysis of the abscission zone in tomato flowers and leaves ARF19 was also over-expressed in the AZ, while in soybean leaf abscission only ARF8 expression changes were observed (Kim et al., 2016). Our results indicate that ARF7 and ARF19 can be associated with flower abscission in lupine, but the mechanism of how these genes regulate abscission needs to be clarified.
In the falling pods we observed reduced amounts of ILR1-like 2 (IAA-amino acid hydrolase) transcripts responsible for IAA conjugation to amino acids (LeClere et al., 2002) and increased DEGs encoding auxin receptors such as TIR1, ABP19, and SKP2A (Table 5 and Table S7). This indicates an increase in free auxin concentrations and a growth of the sensitivity to IAA in these organs. The amount of transcripts encoding ARF2 and ARF4 in PNAB is higher than in PAB. It is suggested that ARF2 is a general repressor of auxin-regulated cell division (Ellis et al., 2005; Okushima et al., 2005) and plays an important role in growth of the ovule before fertilization that determines the final size of the seed (Schruff et al., 2005). Therefore, ARF4 and ARF2 may participate in the regulation of cell division in the developing pods of yellow lupine.
Ethylene
In the flower, we found only three DEGs involved in ethylene (ET) signal transduction, which suggests that in this organ ethylene did not play any primary role in abscission regulation (Table 3 and Table S5).
This was not the case, however, in the flower pedicels, as here 17 of the identified DEGs were involved in ethylene metabolism (Table 4 and Table S6). Three unigenes encoding the ET biosynthesis enzyme ACO1 were down-regulated in FPNAB. This confirms the outcome of study suggesting that in L. luteus, flowering abortion evoked by ABA resulted from its stimulatory effect on the expression of two main ET biosynthesis genes LlACS and LlACO, the ACC content and, consequently, the increase in that phytohormone production (Wilmowicz et al., 2016). As mentioned above, ethylene accumulation causes auxin transport inhibition and organ abscission. The same takes place in yellow lupine pedicels, as an increase in the expression of ET biosynthesis genes correlates with decrease in the expression of genes associated with auxin transport in the pedicels of abscising flowers. Additionally, fourteen genes encoding elements of ET signal transduction were up- or down-regulated in FPAB (Table 4 and Table S6). In RNA-Seq analysis of another plant species many genes related to different steps of the ethylene signaling transduction pathway were expressed in the tomato AZs both after flower removal (FAZ) and leaf deblading (LAZ) (Sundaresan et al., 2016). Also, EIN3 and the genes from ERF family were over-expressed in the tomato flower and leaf AZs (Sundaresan et al., 2016) and in the FPAB of yellow lupine too.
In the pods, all the DEGs attributed to ethylene biosynthesis (ACS1 and ACS7) and signal transduction (ERF3 and ERF1A) were less expressed in PNAB, which indicated that in the abscising pods ET was accumulated in larger amounts (Table 5 and Table S7). Ethylene plays a major role in regulating the expression of cell wall-associated genes encoding PGs (Sitrit and Bennett, 1998; Hiwasa et al., 2004), EXPs (Rose et al., 1997; Hiwasa et al., 2003) and BGals (Karakurt and Huber, 2004; Nishiyama et al., 2007). Therefore, we suggest that an extensive up-regulation of cell wall-related DEGs in the abscising pods could, to some extent, be the result of an increased ethylene biosynthesis in these organs.
Gibberellin biosynthesis is induced in organs maintained on plant
Gibberellins are hormones involved in cell expansion, fruit set and growth (Katsumi and Ishida, 1991; Serrani et al., 2007). In contrast to auxin, GA biosynthesis was mainly altered in the tested samples. Changes also occurred at the level of signal transduction and catabolism, but to a minor extent.
The large number of identified genes related to gibberellin metabolism and signaling indicates the importance of GA in lupine flower fate. By comparing the gene expression between the flowers falling off and control, we found 7 DEGs involved in GA functioning. One of them, GA2OX1 engaged in GA catabolism, was down-regulated, while the others—responsible for its biosynthesis (KAO2, GA20OX2, GA20OX1) and signal transduction (GAI, GASA12, and GASA14)—were up-regulated in FNAB (Table 3 and Table S5). The mRNA levels of GA biosynthesis genes rise after pollination in tomato (Rebers et al., 1999; Serrani et al., 2007). In addition, the transcript level of SlGA2ox2 was found to be lower, that results in higher levels of GA in active form (Serrani et al., 2008). In A. thaliana and P. sativum, auxin probably acts as a signal from successfully fertilized ovules that, in turn, stimulates GA biosynthesis and triggers fruit development (Ozga and Reinecke, 2003; Dorcey et al., 2009). Our RNA-Seq analysis shows that probably a similar scenario occurs in yellow lupine.
Similarly to the flowers, in the pedicels of lupine flowers GA biosynthesis was induced in FPNAB (Table 4 and Table S6). This suggests that in this part of non-abscising flowers the GA signaling pathway was also induced.
Our RNA-Seq analysis shows that GA plays an essential role in retaining the pods on the plant (Table 5 and Table S7). The transcription level of GA biosynthesis genes was already higher in FNAB, and the difference even rose in PNAB. All the 10 DEGs attributed to GA biosynthesis were strongly expressed in PNAB, including GA20OX2 which was the most up-regulated DEG in the developing pods. At the same time, this up-regulation in PNAB means that in the abscising pods lower GA biosynthesis was ceasing their development. Interestingly, DEGs involved in GA signal transduction (mainly GAMYB) were significantly over-expressed in the pods being dropped. GAMYB is known as a regulator of flower induction (Gocal et al., 1999) and α-amylase activation in the aleurone layer of the seeds (Gómez-Cadenas et al., 2001; Kaneko et al., 2002). There is no evidence of their function in fruit abscission or development. Only one study demonstrates that in the seeds of A. thaliana, GAMYB-like genes are able to promote—but are not essential for—the progression of PCD in the aleurone layer (Alonso-Peral et al., 2010). GA also regulates exine formation and PCD of tapetal cells and the direct activation of CYP703A3 by GAMYB is crucial for exine formation (Aya et al., 2009). Probably, GAMYB is involved in PCD induction during lupine pod abscission.
Involvement of other hormones in lupine organs abscission
There is still little understanding of the participation of other plant hormones in cutting off generative organs. Our study aims to fill this gap.
Our RNA-Seq analysis revealed that ABA receptor PYL2 was over-expressed in the abscising flowers and flower pedicels (Tables S5, S6). In the flower pedicels, other DEGs recognized as involved in abscisic acid signaling, such as SAPK2, SAPK8, and PP2CA, were also down-regulated in FPNAB. Simultaneously, CYP707A4 engaged in ABA break-down was up-regulated. These data suggested that in the abscising flowers and flower pedicels ABA signaling pathways were being activated. In the pods of L. luteus, ABA catabolism gene CYP707A1 was up-regulated in PNAB, which means that in the abscising pods this gene was less expressed, and ABA accumulation occurred. In apple, before fruits abscission a chronological increase in the level of ABA and ACC also takes place (Gómez-Cadenas et al., 2000). Additionally, ABI3 involved in ABA signal transduction was more expressed in PNAB. ABI3 has been demonstrated to directly induce the expression of storage protein genes (Ezcurra et al., 1999; Reidt et al., 2000; Kroj et al., 2003; Braybrook et al., 2006; Wang and Perry, 2013).
Among the DEGs in the flowers, we identified unigenes that showed homology to jasmonic acid (JA) and ester methyl jasmonate (JaMe) biosynthesis genes (Table S5). A similar situation took place in the pods. DEGs connected with JA biosynthesis were all up-regulated in PNAB (Table 5 and Table S7), which suggests stimulation of metabolism of these hormones in the pods that were continuing to develop. In contrast, in the pedicels of non-abscising flowers, JA biosynthesis was decreased (Table 4 and Table S6). Mutations in JA biosynthesis genes in A. thaliana causes male sterility due to delayed anther development and shortened filaments (reviews in Wasternack, 2007; Browse, 2009). Additionally, it was shown that JA is also required for proper development of tomato embryo (Goetz et al., 2012). This again confirms our hypothesis that flowers undergoing abscission are not fertilized and in falling pods embryo development is stopped.
Cytokinins are involved in the regulation of cell division and expansion (Skoog et al., 1965) and CK effect on abscission is thought to be mediated by ethylene (Sipes and Einset, 1983; Grossmann, 1991; Cin et al., 2007). In the pedicels of non-abscising flowers (Table 4 and Table S6), more expressed were those DEGs that showed similarity to: CK receptor AHK4, two genes encoding cytokinin biosynthesis enzymes LOG1 and LOG3, and two genes CKX5 and CKX3 encoding enzymes that catalyze CK oxidation to biologically inactive forms (Galuszka et al., 2007). In the pedicels of flowers undergoing abscission two other forms, namely CKX1 and CKX9, were up-regulated. In Arabidopsis roots, CKX proteins show diverse subcellular and tissue-specific localization that suggests specific developmental and physiological functions of each gene (Werner et al., 2003). Opposite changes in the expression of CKX genes in the flower pedicels of L. luteus may be associated with the fact, that different processes regulated by CK occur in various cell compartments and tissues but more study is need to clarify the role of CK in organ abscission in lupine.
Brassinosteroids are endogenous plant hormones essential for the proper regulation of multiple physiological processes required for proper plant growth and development (Clouse and Sasse, 1998; Krishna, 2003; Sasse, 2003; Clouse, 2011). In FPNAB, 8 DEGs were recognized as involved in BR signaling, which suggests that BRs play an important role in the pedicels of the developing flowers (Table 4 and Table S6). Three DEGs involved in signal transduction (Cyclin-D3) were up-regulated in FPNAB, which indicates more intensive cell division in the growing pedicels. DEGs putatively encoding BR biosynthesis enzymes were up- and others down-regulated. BR biosynthesis can be conducted via many alternative pathways which interweave at many nodes and form a sort of metabolic grid (see KEGG map00905; Shimada et al., 2001). In the flower pedicels, some alternative routes are up-regulated, which probably leads to changes not only in BR concentrations, but also in BR composition. BR promote cell expansion through regulation of expression of the genes involved in cell wall modifications, cellulose biosynthesis, ion and water transport, and cytoskeleton rearrangements (Clouse and Sasse, 1998; Schumacher et al., 1999; Morillon et al., 2001; Vert et al., 2005; Kim and Wang, 2010). These findings correlate with our RNA-Seq analysis results; apart from many up-regulated cell wall-related DEGs, 12 DEGs showing similarity to aquaporins were also up-regulated in FPNAB (Table S6).
In the flower pedicels and pods, we identified five and six DEGs (indicating similarity to TGA, SABP2, and PR1) interpreted as SA-related (Tables 4, 5, Tables S6, S7), respectively. This situation is common during organ abscission. For example, in tomato, transcription factor TGA was expressed in FAZ and LAZ (Sundaresan et al., 2016). The transcription of several genes coding for pathogenesis-related (PR) proteins increased concomitantly with the onset of flower abscission of Sambucus nigra (Coupe et al., 1997).
The expression of the DAD2 gene (another name: DR14/DWARF14) encoding the receptor of SL is differential in the flowers and pods (Tables 3, 5, Tables S5, S7). DR14 protein could work as an intercellular signaling molecule to fine-tune SL function (Kameoka et al., 2016). Recently, a hypothesis in which SL may indirectly influence seed size has been created. SL mutants show enhanced shoot branching and delayed leaf senescence, so they may have reduced nutrient remobilization resulting in a reduction in seed production. These findings indicate that SLs may affect the grain crop yield through leaf senescence and shoot branching regulation (Yamada and Umehara, 2015). We suggest that the increased transcription of the DAD2 gene in the non-abscising flowers and pods can be linked to the nutrients mobilization.
Other factors probably involved in generative organ abscission of yellow lupine
Transcription factors involved in generative organ abscission
Some transcription factors (TFs) characteristic for soybean leaf and flower abscission processes, such as NAC, WRKY, and AIL/PLETHORA (PLT) (Kim et al., 2016), were also up-regulated in FPAB of lupine (Table S9). Among the DEGs, there were AIL/PLT (3 and 2 unigenes respectively), NAC (29) and WRKY (12). For example, DEG c60377_g2 with log2FC = −8.08, a putative ortholog of NAC transcription factor 29, also called NAP (NAC-LIKE, ACTIVATED BY AP3/PI), is closely associated with the senescence process of Arabidopsis rosette leaves and, possibly, in other plant species (Guo and Gan, 2006; Zhang and Gan, 2012). The expression of ANAC019, ANAC055, and ANAC072 was induced by drought, high salinity, and abscisic acid (Tran et al., 2004). NAC055 and NAC072 were also over-expressed in the FPAB of lupine (Table S9). These results are consistent with previous studies showing up-regulation of this type of TF genes in the flower AZ of tomato (Sundaresan et al., 2016), and the fruit AZ during abscission of mature melon and olive fruit (Corbacho et al., 2013; Gil-Amado and Gomez-Jimenez, 2013). Another transcription factor, namely MYB108, was up-regulated in FAB, as well. This protein contributes to the regulation of stamen maturation and male fertility in response to jasmonate signaling in A. thaliana (Mandaokar and Browse, 2008).
Some TFs were more expressed in organs continuing development. For example, the bHLH TF, MYC4 is up-regulated in PNAB (Table S9).The maximum expression of MYC2, MYC3, and MYC4 in A. thaliana coincided with the developmental stages during which seed storage reserves accumulated, suggesting these genes may affect the accumulation of seed storage reserves (Gao et al., 2016). Over-expression of MYC4 in the developing pods of yellow lupine suggests that storage reserves accumulated in seeds.
PCD and ROS are associated with abscission process
Organs that completed the abscission, undergo changes that resemble the canonical plant PCD (van Doorn et al., 2011), among others, the stimulation of nucleases and reactive oxygen species (ROS) (Farage-Barhom et al., 2008; Sakamoto et al., 2008a; Meir et al., 2010; Bar-Dror et al., 2011). In tomato, the overexpression of antiapoptotic proteins and inhibition of a PCD-associated ribonuclease trigger abscission (Lers et al., 2006; Bar-Dror et al., 2011), whereas in pepper, ROS inhibitors prevent abscission suppressing H2O2 production (Sakamoto et al., 2008b).
Our results confirm these findings. Among the DEGs from all the compared libraries we identified numerous genes encoding proteins directly involved in protection against reactive oxygen species, such as peroxidase, as well as enzymes belonging to the carotenoid biosynthesis pathway, such as CCD4, ZDS, and beta-carotene hydroxylase (Tables S8–S10). Also, 16 DEGs displaying similarity to various nucleases were more expressed in FPAB (Table S9). Additionally, one of the most intensively expressed DEGs in the flower pedicels with an active AZ showed similarity to the vacuolar processing enzyme VPE, an executor of plant PCD (Hatsugai et al., 2004). VPE is a cysteine protease that cleaves a peptide bond at the C-terminal side of asparagine and aspartic acid (Wang et al., 2009) and plays an essential role in the regulation of the lytic system of plants during the processes of defense and development (Hara-Nishimura et al., 2005).
Water transport plays an important role especially in the developing organs
Our RNA-Seq analysis revealed that water transport played an important role especially in the developing organs. Aquaporin TIP2;1 was one of the common DEGs for all our study libraries. Additionally, transcripts of 12 DEGs encoding aquaporins belonging to the tonoplast intrinsic protein (TIP) and plasma membrane intrinsic protein (PIP) gene families were more accumulated in the pedicels of non-abscising flowers (Table S9).There were also 6 (5 up-regulated) and 5 (4 down-regulated) DEGs presenting similarity to genes encoding aquaporins in the flowers and pods, respectively (Tables S8, S10). TIP1 and TIP2 are preferentially associated with the large lytic vacuoles and vacuoles accumulating vegetative storage proteins, respectively (Gattolin et al., 2009). Moreover, intensive expression of TIP and PIP genes during cell expansion has been observed in numerous plant species (e.g., O'Brien et al., 2002; Liu et al., 2008; Saito et al., 2015). Up-regulation of aquaporin genes in L. luteus FNAB and FPNAB is probably associated with cell expansion. On the other hand, in the pods, beside TIP2;1, other DEGs belong to the NIP aquaporin family (that are involved in transport of water, but also other molecules; Dean et al., 1999; Liu et al., 2003; Wallace and Roberts, 2005; Takano et al., 2006; Choi and Roberts, 2007; Mitani-Ueno et al., 2011) were over-expressed in PAB (Table S10). This was probably associated with the processes of nutrient reutilization.
Calcium signaling pathway plays an important role in yellow lupine pod abscission
Calcium (Ca2+) is crucial for numerous biological functions. In addition to its key roles in ensuring cell wall and membrane system structural integrity, it has been shown to act as an intracellular regulator in many aspects of plant growth and development, including stress responses (White and Broadley, 2003; Ranty et al., 2006), cell division and elongation, and fruit growth (Hepler, 2005). Recent RNA-Seq analyses of apple fruit abscission revealed that Ca2+ deficiency due to the down-regulation of genes encoding transporters of this cation could be a signal for the degeneration of the lateral apple fruits (Ferrero et al., 2015). In contrast, our study the abscising pods demonstrated over-expression of 26 unigenes associated with calcium transport, homeostasis and response (Table S10). Among them there were: a putative calcium-transporting ATPase, a calcium-dependent protein kinase, a probable calcium-binding protein, a cation/calcium exchanger and a calmodulin-like protein. This suggests that the Ca2+ signaling pathway plays an important role in yellow lupine pod abscission, although there is a need for further research in order to clarify which physiological processes these changes are associated with.
Fertilization probably did not occur in abscising flowers
In addition to the cell-wall related, MYB108 and other mentioned genes, there were more unigenes associated with pollination that were differentially expressed between the developing flowers and the abscising ones. In the abscising flowers, we detected more mRNA of MS5, a gene essential for male fertility, especially for microspora and pollen grain production (Glover et al., 1998), and involved in regulating cell division after male meiosis I and II to facilitate meiotic exit and transition to G1 (Ross et al., 1997; Bulankova et al., 2010) (Table S8). Another gene, NFD4, encoding a protein required for polar nuclear fusion during female gametophyte development and karyogamy during fertilization (Portereiko et al., 2006), was also less expressed in FAB. This confirms the hypothesis that shed flowers are unfertilized.
Analysis of DEGs common for all comparisons confirms known data but also reveals MIR169 is involved in generative organ abscission
Our analyses of DEGs common for all comparisons allowed determination of which changes were universally associated with the processes of generative organ abscission. By that means we distinguished 28 unigenes with different expression in all of the libraries compared. 25 of them were down-regulated, 2 were up-regulated in all comparisons, and only one—mainly TIP2;1—was up-regulated in the FPNAB, unlike the other samples (Table 6, Table S11).
Among the common DEGs, there were representatives of factors involved in the already discussed issues such as cell wall reorganization (EGL11) and water transport (aquaporin TIP2;1). The EG45-like domain containing protein (PNP-A) also has a systemic role in H2O and solute homeostasis (Ludidi et al., 2002). Numerous common DEGs are involved in stress response, among them ZAT12, HSF24, RCI2B, 5INPP5A, SPX3, MYB4, or SIB2. ZAT12 play a central role in reactive oxygen and abiotic stress signaling in Arabidopsis (Davletova et al., 2005). SIB2, Arabidopsis sigma factor binding proteins, is an activator of the WRKY33 transcription factor in plant defense (Lai et al., 2011). RCI2B and RCI2A are two developmentally- and stress-regulated cold-inducible genes of Arabidopsis encoding highly conserved hydrophobic proteins expressed during the first stages of seed development and germination, in vascular bundles, pollen, and guard cells (Medina et al., 2001).
The presence of the DEG encoding the putative calcium-transporting ATPase 13 (ACA13), the enzyme catalyzing the hydrolysis of ATP coupled with the translocation of calcium from the cytosol out of the cell or into organelles (Iwano et al., 2014), confirmed that calcium signaling plays an important role in organ abscission. Interestingly, among common DEGs we identified one showing similarity to QKY encoding a predicted membrane-bound protein. By analogy to animal proteins with related domain topology, QKY is speculated to be involved in Ca2+-dependent signaling and membrane trafficking. QKY interacts with receptor-like kinase STRUBBELIG (SUB) at plasmodesm (PD) to promote tissue morphogenesis, for example integument initiation and outgrowth during ovule development (Fulton et al., 2009; Vaddepalli et al., 2014). Our RNA-Seq findings indicate that cell-cell communication via QKY was an important factor determining organ fate.
Surprisingly, none of the common DEGs were associated with hormone metabolism or signaling. This was probably due to the fact that although the significant changes in hormone-related genes occurred in all of the tested samples, they were specific to a particular organ. However, the DEGs from the same cluster but with others gene number (c62886_g2, c62886_g7, c62886_g4) indicating similarity to ARF4 proteins were presented in all comparisons (Tables S8–S10).
Our analysis revealed similarity of the common DEG c125095_g1 to the miR169e precursor of Glycine max (gma-MIR169e) (Figure 5). MicroRNAs (miRNA) are small noncoding RNAs that control development, stress responses and hormone signaling or metabolism, in both animals and plants, by post-transcriptional regulation of gene expression (Bartel, 2009; Voinnet, 2009). In plants, 21 nt miRNAs are processed from long stem-loop precursor RNAs. The mature miRNA is incorporated into the RNA induced silencing complex (RISC) (Schott et al., 2012) to guide post-transcriptional gene silencing (PTGS) of complementary mRNA by cleavage and/or inhibition of translation (Brodersen and Voinnet, 2009). A sequence analysis revealed that the transcript of unigene c125095_g1 sequence forms the hairpin characteristic for pre-miRNAs, wherein the stem was the sequence of the mature molecule of miR169e (Figure 5), thus indicating that it may be a source of functional mature miRNA. MiR169 has been identified in many plant species (Sunkar and Jagadeeswaran, 2008) and their isoforms are involved in regulation various processes during development (Gonzalez-Ibeas et al., 2011) and in response to biotic (Singh et al., 2012) or abiotic stresses (Zhou et al., 2010; Licausi et al., 2011; Zhao et al., 2011). In plants, the main targets of miR169 are genes that encode the subunit A of Nuclear Factor Y (Rhoades et al., 2002). In plants, NF-Y TFs have been linked to development (Lotan et al., 1998; Combier et al., 2006; Wenkel et al., 2006), signaling (Warpeha et al., 2007), stress responses (Nelson et al., 2007; Li et al., 2008; Liu and Howell, 2010; Zhao et al., 2011), carbohydrate metabolism and cell wall modification (Leyva-González et al., 2012). Among yellow lupine DEGs no NF-YA transcripts were found, although this may be related to the fact that miR169 may regulate the expression of its target genes also by inhibiting translation, which cannot be detected using RNA-Seq. Further study is needed to identify the processes regulated by miR169 in yellow lupine.
Several DEGs that we identified, such as ARF2, ARF4 or GAMYBA were also regulated by the regulatory sRNAs - miR390, ta-siARF (Marin et al., 2010) and miR159 (Alonso-Peral et al., 2010), respectively. These findings shed new light on possible regulation of organ abscission by non-coding RNAs.
RT-qPCR validation of RNA-seq results
In order to validate the results obtained via RNA-Seq analyses, we assessed the trends in expression of 11 chosen genes which displayed significantly different transcription level in all of the compared transcriptomes, using RT-qPCR. The results of the expression analysis of these genes supported the validity of our RNA-Seq (Figure S1).
Metabolic changes associated with the processes of generative organ abscission in lupine
KEGG is a highly integrated database providing information on the biological systems and their relationships at the molecular, cellular and organism levels, particularly via the KEGG pathway maps (Kanehisa et al., 2007). An analysis of KEGG metabolic pathways, which was performed using the identified DEGs, allows for a global analysis of the metabolic changes that occur during abscission, both in pedicels with an AZ and the abscising organs. The identified DEGs were grouped into 4 major pathways from the KEGG pathway database (Figures S2A, S3A, S4A): “metabolism,” “genetic information processing,” “environmental information processing” and “organismal systems.” As “metabolism” appeared to be one of the most significant and highly represented categories in our study we ran an in-depth analysis of it, which is presented in Figures S2B, S3B, S4B.
Our analysis of enriched KEGG pathways from all the comparisons indicates that some metabolic pathways were modulated in all the cases, while some seemed to be characteristic for specific plant organs. “Carbohydrate metabolism” and “amino acid metabolism” were common enriched pathway categories for the DEGs from all the comparisons made. This suggests that these metabolism categories were a major target of changes during the abscission of generative organs in yellow lupine. It is worth mentioning that to the “Pentose and glucuronate interconversions” pathway from the “carbohydrate metabolism” category belong i.e., cell-wall related PME and PAE. This explains the path's strong enrichment in all library comparisons, and especially in the pods (Figures 6–8).
The KEGG pathways which were enriched particularly in FNAB were “Terpenoid backbone biosynthesis” from the “Metabolism of terpenoids and polyketides” category (Figure 6). Terpenoids, also known as isoprenoids, are a large class of natural products consisting of isoprene (C5) units. In higher plants, isoprenoids participate in a wide variety of biological functions. Specific examples include photosynthetic pigments (chlorophylls and carotenoids) or hormones (ABA, GA, CK, and BR). The enrichment of these KEGG pathways in the flower could be associated with hormone biosynthesis. Other groups of up-regulated DEGs in FNAB belonged to the “Genetic Information Processing” category, and particularly to the “Replication and repair” and “Translation” pathways (38 DEGs) (Figure 6 and Figure S2A). The fact that, apart from these pathways, amino acid biosynthesis and metabolism and nucleotide metabolism pathways were enriched, too, indicates that cell division and intensive de novo protein synthesis were induced in FNAB or from the other side these processes are inhibited in FAB.
Other pathways from the KEGG metabolism category in which DEGs identified in flower pedicels were indexed fell under the “Environmental Information Processing” heading, and included the “Plant hormone signal transduction pathway,” which confirmed the importance of changes in hormonal balance in the studied pedicels (Figure 7). This could be associated with the “Inositol phosphate metabolism” pathway that was over-represented in FPAB. We found that DEGs showing similarity to, inter alia, non-specific phospholipase C (NPC), belonged to this pathway. NPCs play important roles in many processes as phospholipid-to-galactosyl DAG exchange, growth and development associated with hormone signaling and stress responses (reviewed in: Pokotylo et al., 2013). The same phospholipases were included in the “Ether lipid metabolism” and “Glycerophospholipid metabolism” pathways that were strongly up-regulated in the abscising flower pedicels. In contrast, the “Biosynthesis of unsaturated fatty acid” and “Fatty acid metabolism” pathways that are probably associated with cell membrane functioning were enriched in the flowers that continued to develop.
In the pods, the “Plant-pathogen interaction” pathway from the “organismal systems” major category was strongly enriched due to high numbers of up-regulated DEGs related to PR1, calcium-related genes and WRKY in PAB (Figure 8). Another pathway enriched in the abscising pods was “Phosphatidylinositol signaling system,” which include DEGs encoding calmodulines and phospholipase A and C. In PAB, within the “amino acid metabolism” category we found DEGs showing similarity to enzymes involved in protein degradation, such as Cysteine proteinase (e.g., c134689_g1, c73956_g1, Table S10) (Martínez et al., 2012) or Proline dehydrogenase (e.g., c60845_g3, c53844_g2, Table S10) (Funck et al., 2010; Monteoliva et al., 2014), which confirmed our DEGs analysis indicating a strong induction of degradation processes in PAB. Only one pathway, “diterpenoid biosynthesis” related to GA biosynthesis, was clearly enriched in PNAB, which fits well to our analysis of hormone-related DEGs in the pods.