Introduction 

Auxin is a key signaling phytohormone regulating almost all aspects of plant growth and development from embryogenesis to senescence [1]. Indole-3-acetic acid (IAA), the main naturally occurring auxin in the plant kingdom, has long been proposed to be synthesized from both tryptophan through trp-dependent pathways (Fig. 1) and directly from tryptophan precursors through a trp-independent pathway [2, 3].

Fig. 1
figure 1

Potential IAA biosynthesis pathways in Arabidopsis. CHA, chorismate; ANT, anthranilate; IGP, indole-3-glycerol phosphate; TSA, tryptophan synthase A; TSB, tryptophan synthase B; Trp, tryptophan; INS, indole synthase; IAOx, indole-3 acetaldoxime; IAN, indole-3-acetonitrile; IAM, indole-3-acetamide; TAA, tryptophan aminotransferase of Arabidopsis; IPA, indole-3-pyruvic acid; YUCCA, a Flavin monooxygenase; IAA, indole-3-acetic acid. Enzymes are indicated in blue color, whereas intermediates are in black. Arrows in red colors refer to undefined or unconfirmed enzymes, whereas black colors refer to identified enzymes

Full size image

Biochemical and genetic experiments have recently revealed the indole pyruvic acid (IPA) pathway as the first complete route of IAA synthesis via the trp-dependent pathway. In this simple two-step pathway, tryptophan aminotransferase (TAA) converts tryptophan into IPA, whereas YUCCA, a flavin-containing monooxygenase, catalyzes the rate-limiting reaction by converting IPA into IAA [4, 5]. Additionally, indole-3-acetaldoxime (IAOx) has been demonstrated as a second trp-dependent pathway [6, 7]. Even though the cytochrome P450 monooxygenase CYP79B2/CYP79B3 has been suggested to produce IAOx from tryptophan, the remaining intermediates and enzymes are still largely unidentified. Furthermore, aldehyde oxidase (AAO), nitrilase (NIT), and amidase convert indole-3-acetaldehyde (IAAld), indole-3-acetonitrile (IAN), and indole-3-acetamide (IAM) to IAA. They all were recognized as naturally occurring enzymes. However, pathways including these enzymes are yet to be characterized [8,9,10,11].

Three decades ago, a combination of labeling, genetic, and metabolic approaches led to the proposal that IAA is biosynthesized from tryptophan precursors via the trp-independent pathway [12, 13]. This pathway was suggested to be ubiquitous as it was proposed to synthesize IAA in broad taxonomic groups of plants, including Arabidopsis, tomato, maize, carrot, Lemnagibba, and bean [14,15,16,17,18]. Recently, indole severe sensitive 1 mutant (iss1) was described to display trp-independent IAA synthesis as well as related physiological activities including longer petioles, adventitious roots, and leaf narrowing [19]. Six years ago, feeding and genetic approaches were uncovered INS (At4g02610) as the first enzyme in the trp-independent pathway and an essential role for this pathway in early embryogenesis was suggested [20]. The cytosolic INS was demonstrated to regulate hypocotyl length at high temperature in concert with the IPA pathway in Arabidopsis based on the observation that ins-1 mutants are not 100% penetrant and have an additive effect with wei8 (weakly ethylene insensitive8) mutants. Wang et al. [20] hypothesis has been questioned by Nonhebel bioinformatics work. The possibility for the INS or its product, indole, to interact with TSB and therefore with the trp-dependent pathway was not excluded [21]. The additive effect of wei8 and ins was further used to support this hypothesis. Therefore, the involvement of INS in either trp-dependent or independent pathways was considered in this article. In this pathway, indole-3-glycerol phosphate is converted to indole by the indole synthase in the cytosol [20]. However, enzymes catalyzing the remaining steps have not been identified so far. Therefore, this piece of work was aimed to find out enzymes potentially act with INS in the INS pathway of IAA synthesis. All previously reported genes to have roles in IAA synthesis were investigated in the list of genes most strongly positively co-expressed with INS. Additionally, a list of genes negatively co-expressed with INS was also studied to shorten the list of genes most probably act with INS in the trp-independent pathway.

Coexpression analysis is a powerful tool that has been used for identifying genes related to a specific metabolic pathway [22]. For instance, this approach has been adopted to identify genes regulating cellulose synthesis and transcription factors involved in the glucosinolate synthesis pathway [23, 24]. In this research, ATTED-II, a web-based tool, was used as a secondary database to analyze publicly available microarray data for the dicot model plant Arabidopsis thaliana [25]. Unlike Hirai et al. [24], who used Pearson’s correlation coefficient for combinations of Arabidopsis genes available in ATTED-II, mutual rank (MR) as a powerful measure was used in the latest release of the database. Co-expressed gene relationships are primarily based on the vast amount of microarray data stored in TAIR as well as NASCArrays [26].

Methods

The co-expression analysis tool (CoExSearch: co-expressed gene list from multiple query genes) available on the ATTED-II (http://atted.jp) database version 9.2 was used to find genes positively and negatively co-expressed with the Arabidopsis INS and TSB genes. Genes previously reported to have a possible role in IAA synthesis are listed in Table 1. The available top 2000-gene lists positively and 300-gene lists negatively co-expressed with seed genes were studied and analyzed. The four TSB genes in Arabidopsis (At5g38530, At5g28237, At4g27070, and At5g54810) were uploaded into the CoExSearch tool of ATTED-II. Affy data were only available to TSB (At5g38530).

Table 1 Summary of IAA-synthesis genes previously reported having a role in IAA synthesis. INS, indole synthase; TSA, tryptophan synthase A; TSB, tryptophan synthase B; YUCCA, a flavin monooxygenase; TAA, tryptophan aminotransferase of Arabidopsis; TAR, tryptophan aminotransferase related; AAO, aldehyde oxidase; NIT, nitrilase
Full size table

Results and Discussion

Tryptophan Synthase A but Not Indole Synthase Interacts with Tryptophan Synthase B to Produce Tryptophan

A significant role for the involvement of indole synthase in the trp-independent pathway of IAA synthesis has been quite recently reported. The cytosolic INS was suggested to produce indole from the indole-3-glycerol phosphate [20, 27]. However, this suggestion was challenged by a proposal that INS may interact with TSB to produce tryptophan and therefore with the trp-dependent pathway [21]. In this work and through the implementing co-expression analysis of Arabidopsis genome obtained from the ATTED-II secondary database [25], we excluded the possibility of INS's interaction with TSB. Among the 21,000 genes analyzed, the available 2000 gene list correspond to 90% cut-off value positively significantly co-expressed with TSB in general conditions (condition-independent manner) were obtained and studied. Furthermore, co-expressed gene lists obtained from specific samples under certain conditions, including hormone, tissue, biotic, and abiotic stress, were also obtained and investigated. Table 2 shows a short list of all genes related to IAA synthesis significantly positively co-expressed with TSB under general and specific conditions.

Table 2 Summary of IAA-synthesis genes positively co-expressed with TSB in the top 2000 genes lists in general and specific conditions. Specific conditions include hormone, tissue, biotic, abiotic, and light. Coexpression index is represented by mutual rank
Full size table

Indole synthase gene was found neither in the general nor in the specific co-expression data lists, suggesting that this enzyme may not interact with TSB to produce IAA via the trp-dependent pathway. In contrast to INS, TSA which its product forms a complex with TSB [28], was strongly co-expressed with the query gene with mutual ranks of 38, 98, and 180 in the hormone, general, and abiotic conditions respectively. Interestingly, INS was not also found in the top 300 gene list negatively co-expressed with TSB, suggesting that trp-dependent and independent pathways may interact to produce IAA in Arabidopsis. Another evidence for the involvement of INS in the trp-independent pathway came from the analysis of correlation patterns. Figure 2 shows the correlation of expression patterns for INS and TSB (A and B) as well as TSA and TSB (C and D) for samples obtained from both developmental and vegetative tissues, respectively. The correlation of expression for INS and TSB is −0.046, whereas for TSA and TSB is 0.291 is further supports the involvement of INS in the trp-independent pathway of IAA synthesis. This suggestion is in agreement with the independent production of indole by TSA and INS suggested in maize [27].

Fig. 2
figure 2

Correlation of the expression obtained from developmental and vegetative samples for INS and TSB (A and B) and for TSA and TSB (C and D) 

Full size image

Amidase May Interact with INS in the trp-Independent Pathway

Even though Wang et al. [20] reported a significant role for INS in IAA production through the trp-independent pathway, the remaining reaction(s) in this pathway needs to be identified. Functional identification of gene(s) expected to work in the same pathway of IAA synthesis with INS was also attempted in this study using the publically available co-expression database ATTED-II [25]. Indole synthase was used as a seed gene to search for other possible genes in the tryptophan-independent pathway of IAA synthesis. Among all genes previously reported to be involved in IAA synthesis, amidase1 (At1g08980) was strongly co-expressed with INS in general and light conditions with mutual ranks of 116 and 141 respectively (Table 3). Additionally, Amidase1 was found to positively correlate with INS (0.297) (Fig. 3). This correlation value was similar to that found between TSA and TSB; their products constitute an enzyme complex involved in the production of trp from indole-3-glycerol phosphate. Taken together and in addition to the observation that this gene was not found to positively co-expressed with TSB, suggests that amidase1 may be exclusively involved in the trp-independent pathway of IAA synthesis. Finding amidase1 in the top 300 gene list negatively co-expressed with TSB (MR 20,628) further supports this hypothesis. It concludes that amidase may catalyze the rate-limiting step in this pathway. These findings are in agreement with the observation that Amidase1 is specifically convert IAM to IAA (Km 972 µM) [29]. Furthermore, the importance of amidase1 in IAA synthesis has been quite recently suggested by Gao et al. [30].

Table 3 Summary of genes co-expressed with INS (The top 1st 2000 genes). TAA1, tryptophan aminotransferase of Arabidopsis. Coexpression index is represented by mutual rank
Full size table
Fig. 3
figure 3

Correlation of the expression for INS and amidase1 obtained from developmental samples (A) and vegetative samples (B)

Full size image

Amidase5 (At4g34880), an Arabidopsis amidase member that functionally not characterized, was co-expressed with INS in the general and light conditions with a MR of 1000 and 926 respectively. However, statistical supportability of the co-expression relation for this gene was found to be weak. The same thing was observed for the expression of this gene with TSβ in the tissue condition (MR 2471) but again with low statistical supportability suggesting that the gene product may be involved in the production of substance other than IAA. Amidase2 (At5g07360) which was proposed to convert substrate other than indole-3-acetamide (IAM) [31] was co-expressed with TSβ in the general, tissue, biotic, and abiotic specific conditions whereas co-expressed with INS in the abiotic conditions only. Amidase6 (At3g25660), another uncharacterized Arabidopsis amidase member, was co-expressed only with TSβ in the light condition (MR 1419).

Role of Aldehyde Oxidase and Nitrilase in IAA Synthesis

Condition-dependent and independent co-expression lists were also searched for previously genes reported to be involved in IAA biosynthesis. A group of IAA synthesis genes including NIT, AAO, YUC, and CYP79B2/B3 is positively co-expressed with TSB. The nitrilase isoform (NIT3), which was reported exclusively in Brassicaceae [32, 33], is co-expressed with TSB in all but biotic conditions (Table 2). Through the analysis of NIT2RNAi lines, it has been shown that the level of IAA is considerably reduced due to non-detected NIT1-3 [34] suggesting that NIT3 may participate in IAA synthesis through the trp-dependent pathway. Additionally, NIT4 found in monocot and dicot plants is strongly co-expressed with TSB in hormone conditions in addition to general and biotic conditions (Table 2). NIT4 is primarily involved in cyanide detoxification [35]. These controversial data need to be clarified for a possible role of NIT4 in IAA synthesis, particularly in the trp-dependent pathway. None of the NIT isoforms was found to co-expressed with INS suggesting that this enzyme may not have any possible role in IAA production through the INS pathway.

Aldehyde oxidase α and γ are homodimers of AAO1 and AAO2 respectively whereas aldehyde oxidase β is a heterodimer of AAO1 and AAO2. They have shown to have limited preferences for abscisic aldehyde [36]. In this work, co-expression of INS was noted only with AAO2 in the hormone-specific condition, suggesting a possible role for AOγ in IAA biosynthesis through the trp-independent pathway. However, aldehyde oxidase1, 2, and 3 were found to co-express with TSB. Aldehyde oxidase1 is co-expressed with TSB in all analyzed general and specific conditions, whereas AAO2 co-expressed with TSB in the tissue, abiotic and light conditions with mutual ranks of 328, 1456, and 302, respectively. Therefore, a possible role for these enzymes in IAA synthesis through the trp-dependent pathway is not excluded. Additionally, AAO3 product which was reported to oxidize abscisic aldehyde efficiently is co-expressed with TSB in general, hormone, and light conditions. This finding may explain the crosstalk between IAA and ABA hormones.

Conclusion

The trp-independent pathway was earlier proposed as a major route of IAA biosynthesis in Arabidopsis based on metabolites quantification and feeding experiments in trp biosynthesis mutants. More recent evidence supporting the involvement of the trp-independent pathway in auxin biosynthesis through INS was further illustrated. However, the interaction of INS with TSB to produce trp and, as a result, a possible role for this enzyme in the trp-dependent pathway was also suggested based mainly on bioinformatics studies. Therefore, the main aim of this work was to collect evidence through the implementation of coexpression analysis of the Arabidopsis genome for the specific role of INS in auxin biosynthesis. Among the 21,000 genes of the Arabidopsis genome available on the ATTED-II (http://atted.jp) database, TSA was found to be strongly positively coexpressed with TSB in the top 2000 gene lists of general, hormone, and abiotic conditions. Mutual ranks under these conditions were found to be 89, 38, and 180, respectively. Furthermore, TSA was found to positively correlate with TSB (0.291) in a large group of vegetative and developmental Arabidopsis samples. The correlation was clearer in developmental samples comparing to vegetative ones. However, INS was co-expressed with TSB neither in general nor in specific conditions and was found to negatively correlate with TSB (− 0.046), suggesting that INS is operated in the trp-independent pathway. Absence of co-expression and correlation between INS and TSB may suggest that trp-dependent and independent pathways may separately and independently operate to synthesize IAA.

Production of IAA through INS in the trp-independent pathway is unclear in terms of other enzymes involved in this pathway. Therefore, all previously reported enzymes to have a possible role in IAA synthesis were analyzed in this research. Among all these enzymes, amidase was proposed to interact with INS in the trp-independent pathway. Amidase is the only gene found to be strongly positively coexpressed with INS. Mutual ranks for coexpression between INS and amidase (in particular amidase1) were 116, and 141 in general and light conditions respectively. Mutual ranks for these two genes were similar to that found between TSA and TSB, in which their products form an active complex. INS and amidase were positively correlated with each other. The value of correlation was similar to that found between TSA and TSB. Taken together and in addition to the observation that amidase is negatively correlated with TSB, we suggested that amidase1 is solely involved in the trp-independent pathway of IAA synthesis.