Comparative Transcriptomics Reveals Jasmonic Acid-Associated Metabolism Related to Cotton Fiber Initiation

Analysis of mutants and gene expression patterns provides a powerful approach for investigating genes involved in key stages of plant fiber development. In this study, lintless-fuzzless XinWX and linted-fuzzless XinFLM with a single genetic locus difference for lint were used to identify differentially expressed genes. Scanning electron microscopy showed fiber initiation in XinFLM at 0 days post anthesis (DPA). Fiber transcriptional profiling of the lines at three initiation developmental stages (-1, 0, 1 DPA) was performed using an oligonucleotide microarray. Loop comparisons of the differentially expressed genes within and between the lines was carried out, and functional classification and enrichment analysis showed that gene expression patterns during fiber initiation were heavily associated with hormone metabolism, transcription factor regulation, lipid transport, and asparagine biosynthetic processes, as previously reported. Further, four members of the allene-oxide cyclase (AOC) family that function in jasmonate biosynthesis were parallel up-regulation in fiber initiation, especially at -1 DPA, compared to other tissues and organs in linted-fuzzed TM-1. Real time-quantitative PCR (RT-qPCR) analysis in different fiber mutant lines revealed that AOCs were up-regulated higher at -1 DPA in lintless-fuzzless than that in linted-fuzzless and linted-fuzzed materials, and transcription of the AOCs was increased under jasmonic acid (JA) treatment. Expression analysis of JA biosynthesis-associated genes between XinWX and XinFLM showed that they were up-regulated during fiber initiation in the fuzzless-lintless mutant. Taken together, jasmonic acid-associated metabolism was related to cotton fiber initiation. Parallel up-regulation of AOCs expression may be important for normal fiber initiation development, while overproduction of AOCs might disrupt normal fiber development.


Introduction
Gossypium hirsutum (cotton) is an important economic crop and is a major source of both natural textile fiber and cottonseed oil. Cotton fibers are single-celled trichomes from individual epidermal cells on the outer integument of the ovules. Fiber development involves four distinct but overlapping stages: initiation, elongation/primary cell wall (PCW) synthesis, secondary cell wall (SCW) deposition and maturation. Fiber cell initiation usually begins around anthesis and is characterized by the enlargement and protrusion of epidermal cells from the ovular surface. All epidermal cells have the potential to become fibers, but only about 30% will differentiate into commercially viable lint fibers that can be spun [1,2]. During 5-25 days post anthesis (DPA), fiber cells elongate rapidly with peak growth rates of over 2 mm per day [3,4]. During secondary cell wall deposition , cellulose biosynthesis predominates until fibers contain approximately 90% cellulose and the cell wall is adequately thickened. During the final maturation stage (45)(46)(47)(48)(49)(50) fibers quickly dehydrate to become mature [1,2,5].
Fiber initiation progresses through two morphological phases: (1) differentiation of the prefiber from the epidermal cell ovule, and (2) expansion and protuberance of the pre-fiber cell [6]. The second stage involves numerous genes that control the cell cycle, hormone regulation, cytoskeletal features, signal transduction, and formation and deposition of complex carbohydrates and cell wall proteins [7]. In addition, maintaining homeostasis of reactive oxygen species (ROS) is crucial for cotton fiber initiation [8].
Mutant organisms are powerful tools for elucidating the molecular mechanisms of important genes. In cotton, important fiber mutants have been characterized including fuzzless-lintless (fl), such as MD17 [9], SL 1-7-1 [10], Xz-142 WX [11], naked seed (N1 and n2) [12,13], ligon-lintless (Li1 and Li2) [14,15], and immature fiber (im) [16]. A combination of fiber development mutants and high-throughout gene expression analysis has identified genes involved in key stages of fiber development. Padmalatha et al. [17] compared transcriptome analysis of fl with wild-type G. hirsutum cv. MCU5 during fiber initiation (0 DPA) and found that genes involved in calcium and phytohormone-mediated signalling pathways played crucial regulatory roles during fiber differentiation, while mitochondria mediate energy metabolism during fiber elongation. Shi et al. [18] found ethylene biosynthesis to be one of the major biochemical pathways to be upregulated during fiber elongation by combining cDNA library sequencing and microarray analysis. Transcription profiling of the im mutant and its near-isogenic line (NIL) TM-1 during fiber secondary cell wall development led to the identification of an asynchronous fiber development at the transition between secondary cell wall biosynthesis and earlier stages [19].
XinWX is a lintless-fuzzless fiber mutant, showing the similar phenotype with Xz-142 WX. XinFLM is a fuzzless-linted mutant isolated from XinWX. Genetic analysis revealed that XinFLM was a dominant mutant that only differed from XinWX at a single locus for lint [20]. The two mutant lines showed no significant phenotypic differences except for the mature bolls (S1 Fig). Seeds of XinFLM have lint but no fuzz, while those of XinWX are lintless and fuzzless. In this study, we performed a comparative analysis of the transcriptional profile during fiber initiation development of lintless-fuzzless XinWX and linted-fuzzless XinFLM using 28K cDNA microarrays. Differentially expressed genes (DEGs) were identified and their potential functional roles during fiber differentiation and initiation were investigated. Besides largerly agreed with previously reported, this study established a foundation for elucidating the functional role of allene oxide cyclases (AOCs) in the regulation of the development of cotton fiber initation, and suggested that jasmonic acid-associated metabolism was invovled in cotton fiber initiation.
G. hirsutum acc. TM-1 was used for tissue/organ expression analysis. Roots, stems and leaves were collected from two-week-old seedlings, and ovules and fibers were excised from developing flower buds or bolls. G. hirsutum cv. Jinmian 19 was used for JA treatment, and seedlings at the four leaf stage were treated with 100 μM JA (with water as a control). Leaves were harvested at the appropriate times (0, 0.5, 1, 2, 4, 6, 8, 10, 12, and 24 h following treatment). All tissues collected were flash-frozen in liquid nitrogen and stored at -70°C.

RNA isolation
The CTAB-acidic phenolic method [22] was used to extract total RNA that was purified using the NucleoSpin RNA clean-up kit (MACHEREY-NAGEL, Germany). Spectrophotometry and 1.2% agarose formaldehyde denaturing gel electrophoresis were used to monitor RNA quantity and integrity.

RNA amplification and labeling
The cRNA amplification and labeling Kit (CapitalBio, China) was used to amplify and label RNA following the manufacturer's protocol. cDNA samples from XinWX and XinFLM collected during the same fiber developmental stage were labeled with Cy3 and Cy5-dCTP, respectively, and these dyes were interchanged to label cDNA samples in a dye swap experiment.

Microarray hybridization and statistical analysis
Expression patterns were analyzed using the dual-dye 28 k cotton cDNA microarray platform (GPL8569) containing 29,184 probes from Upland cotton (G. hirsutum L. cv. Xuzhou 142) ovule and fiber samples collected between -3 and 25 DPA. Microarray hybridization was carried out as described [19] using -1, 0 and 1 DPA samples from XinWX and XinFLM. The experiment was finished by pairwise hybridizations between the two lines at the same sampling timepoint with two biogical replicates and a dye swap. All microarrays were scanned with a LuxScan 10 K/A scanner (CapitalBio) and array images were analyzed with LuxScan TM 3.0 (CapitalBio). Microarray data have been deposited in the Gene Expression Omnibus of NCBI (GEO series accession number GSE59208; http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi? acc=GSE59208).
The LIMMA program [23] based on the open source Bioconductor project (http://www. bioconductor.org) was used to perform statistical analyses. Feint spots were removed if the intensity was less than 400 units following background subtraction. The LOWESS [24] and Aquantile [25] methods were then used to normalize the data within and across the arrays, respectively. In order to analyze DEGs, normalized data were fitted to linear models in LIMMA and a moderated t-statistic (eBays) calculation was performed [26]. Hybridization signal ratios from each slide were used to determine DEGs between the two accessions at each time point, and single channel data in silico were compared to determine DEGs from the same accession at different time points. Genes were designated as differentially expressed if FDR <0.05 and the fold change 2. Blast2GO (http:// www.Blast2go.com/) was then used for functional categorization of DEGs, GO analysis, and enrichment analysis with Fisher's exact test for each GO term, in which FDR P values (FDR <0.05) were used for the assessment of over-represented GO terms [27].  Table). The His3 gene was used as a control and RT-qPCR was performed using a LightCycler FastStart DNA Master SYBR Green I kit (Roche, Switzerland) in an ABI7500 sequence detection system (Applied Biosystems, USA) according to the manufacturer's protocol. The reaction system was as follows: 1μL cDNA; 0.5 μL forward or reverse primers; 10 μL 2 × SYBR; 8 μL ddH 2 O. The PCR program was as follows: initial denaturation at 95C for 10 min, then 40 cycles of 95C for 15 s, 58 C for 15 s, and 72C for 30 s. The expression of each gene relative to His3 was calculated according to the formula: fold change = 2 -4Ct (4Ct = Ct Target gene -Ct His3 ) [28]. For all real-time PCR reactions, three technical replicates were performed in each of the three biological experiments.

In vitro culture of cotton ovules
Bolls were collected from TM-1 at -1 DPA. The bolls were sterilized in 70% ethanol for 5~10 min and washed three times with sterile distilled water. Ovules were then removed from the bolls under sterile conditons and floated on 30mL liquid BT medium in the flask with 5μM IAA and 0.5μM GA 3 . For JA treatment, JA was added to the liquid BT medium at concentrations of 0.1, 0.5, and 2.5μM, respectively. The ovules were cultured in the dark at 30°C for 2 days. Three replicates were done in each assay.

Fiber initiation differences between XinWX and XinFLM
To identify differences between XinWX and XinFLM in fiber initiation, scanning electron microscopy (SEM) was used to compare the protrusion of epidermal cells during fiber initiation development (Fig 1). The entire ovule and flanking surface lacked any protrusion from -1 DPA to 1 DPA in XinWX (A1-A3, B1-B3). XinFLM also lacked spherical protrusions at the day before anthesis (C1, D1), but the whole ovule and flanking surface was covered in many spherical protrusions at 0 DPA (C2, D2) and 1 DPA (C3, D3), and elongated lint fibers were present at 1 DPA. This indicated that lintless-fuzzless XinWX failed to complete fiber initiation in the normal manner, and the fiber initiation phenotype of linted-fuzzless XinFLM was observed at the day of anthesis. Transcriptional differences between the two lines during fiber initiation (-1, 0, and 1 DPA) were subsequently investigated using a cDNA microarray approach.

Differential gene expression in XinWX and XinFLM during fiber initiation
Differences in gene expression during fiber initiation in the mutant lines were compared at -1, 0 and 1 DPA. Numbers of differentially expressed genes (DEGs) at adjacent time points within and between the lines were determined (Fig 2). The distribution of DEGs exhibiting a false discovery rate (FDR) <0.05 and a fold change 2 suggested the level of transcriptional variation differed significantly. In XinFLM, 1774 genes showed altered expression between 1 and -1 DPA, whereas only 68 DEGs were detected in XinWX, suggesting the gene expression profiles in XinWX had relatively few changes during the fiber initiation stage. Differences in gene expression between XinWX and XinFLM were most pronounced at -1 DPA (834 DEGs) and 1 DPA (1000 DEGs), while only 357 DEGs were detected at 0 DPA. In combination with the SEM results, these findings indicated that genes differentially expressed at 1 DPA may be involved in fiber elongation.
Identified DEGs were further investigated using RT-qPCR. A total of 12 genes involved in various functional categories were selected to validate the microarray data. These included eight DEGs and four genes (28K_001_D11, 28K_118_C05, 28K_141_D12, 28K_206_B07) that did not exhibit differential expression between XinWX and XinFLM in microarray data analysis. For most genes, a similar expression pattern was obtained using both methods. Generally, the results of RT-qPCR were in agreement with the microarray data (correlation coefficient R = 0.87; Fig 3), which validated the microarray data.

Gene ontology (GO) categories of differentially expressed genes within XinWX and XinFLM
Blast2GO was used to obtain GO functional annotation of all identified DEGs, with all of the genes on the chip as a reference group and DEGs as a test group to perform Fisher's exact test (FDR <0.05). To obtain child GO and tip GO terms, the 'show only most specific terms' option was chosen. The main enriched GO terms from adjacent time points in both XinWX and XinFLM mutants are listed in Table 1 and discussed below. Complete information on over-represented GO terms is presented in S2 Table. Between -1 and 0 DPA, XinWX and XinFLM shared some over-represented GO terms in common for up-regulated DEGs, for example 'flavonoid biosynthesis process' (GO:0009813,  two lines, no over-represented GO terms were identified between 0 and 1 DPA or 1 and -1 DPA in XinWX, and between 1 and 0 DPA in XinFLM, for genes up-regulated at 1 DPA. In XinFLM, between -1 and 0 DPA, and -1 and 1 DPA, a large proportion of the DEGs that were up-regulated at 0 and 1 DPA involved in "sucrose metabolic processes" (GO:0005985, FDR = 2.7E-04 and 0.00960), "oxidation-reduction processes" (GO:0055114, FDR = 1.03E-04 and 2.39E-05) and "carbohydrate metabolic processes" (GO:0005975, FDR = 6.74E-07 and 1.70E-04) categories. This suggested that these processes were involved in fiber initiation. These GO terms included genes associated with fiber early development that were reported previously, such as sucrose synthase (Sus), which plays a rate-limiting role in the initiation of single-celled fibers [29]. Additionally, most genes that were differentially expressed during different intervals in XinFLM were associated with hormone metabolism. Between -1 and 0 DPA, many genes up-regulated at 0 DPA belonged to the category "response to growth hormone stimulus" (GO:0060416, FDR = 0.00536), likewise for those up-regulated at -1 DPA involved in "response to hormone stimulus" (GO:0009725, FDR = 0.03591). Between 1 and -1 DPA, many genes up-regulated at 1 DPA belonged to the category "response to peptide hormone stimulus" (GO:0043434, FDR = 0.04961), while many of those up-regulated at -1 DPA occupied the categories "response to hormone stimulus" (GO:0009725, FDR = 0.004169), "hormone-medicated signaling pathway" (GO:0009755, FDR = 0.008339), and "allene-oxide cyclase activity" (GO:0046423, FDR = 0.005497). These results indicated that plant hormones play an important role in fiber initiation development.
Between -1 and 0 DPA, over-represented GO terms for up-regulated DEGs at -1 DPA in XinWX included "jasmonic acid mediated signaling pathway" (GO:0009867, FDR = 0.00328), "salicylic acid mediated signaling pathway" (GO:0009863, FDR = 0.00468), and "abscisic acid mediated signaling pathway" (GO:0.0009738, FDR = 0.00485). These pathways may therefore negatively regulate fiber initiation. Abscisic acid (ABA) is known to have an inhibitory effect on fiber development [30,31], however less information is available regarding the potential roles of jasmonic acid and salicylic acid in fiber development. Annotation of DEGs invovled in "jasmonic acid mediated signaling pathway" showed this GO term contained some genes such as bHLH transcription factor. In previous study, MYC2, a bHLH transcription factor, differentially modulates diverse jasmonate-dependent functions in Arabidopsis [32]. Here, the transcription factors enriched in the GO term of "jasmonic acid mediated signaling pathway" might be related to cotton fiber intiation development.
Over-represented GO categories of differentially expressed genes between XinWX and XinFLM during fiber initiation Functional annotation and enrichment analysis of DEGs between XinWX and XinFLM at -1, 0 and 1 DPA were performed using Blast2GO. The most specific GO terms (Table 2) are discussed below, and complete information on over-represented GO terms is presented in S3 Table. A comparison of DEGs between the two lines showed that the "response to carbohydrate stimulus" category was over-represented in XinFLM both at -1 and 0 DPA (GO:0009743, FDR = 4.97E-04 and 0.02550, respectively), while the "lipid transport" (GO:0006869, FDR = 2.44E-04, 2.40E-07) and "organic substance transport" (GO:0071702, FDR = 0.00710 and 0.00389) categories were simultaneously enriched at 0 and 1 DPA in this mutant line. These GO terms included lipid transfer proteins (LTPs) and transcription factors such as ERFs, WRKYs and Zinc-finger proteins which were indicative of a positive regulatory effect on early fiber development. At -1 DPA, the "xyloglucan: xyloglucosyl transferase activity" (GO:0016762, FDR = 4.07E-08) category was highly represented in XinFLM. Xyloglucan is known to be involved in cell wall loosening and expansion during fiber elongation [33]. At 1 DPA, the GO term "2-hydroxy-1,4-benzoquinone reductase activity" (GO: 0050625, FDR = 0.004528) was over-represented, which included the cotton benzoquinone reductase known to be up-regulated in cotton ovules during fiber initiation [34]. XinWX also generated some particularly over-represented GO categories, including "oxidation-reduction processes" (GO:0055114, FDR = 2.27E-11), "carbohydrate catabolic processes" (GO: 0016052, FDR = 3.13E-10) at -1 DPA, "response to stress" (GO: 0006950, FDR = 1.43E-06 and 2.12E-04) at -1 and 1 DPA, respectively, and "glutamate synthase (NADH) activity" (GO: 0016040, FDR = 0.004267 and 9.84E-05) at -1 and 1 DPA, respectively. During the early stages of fiber development, carbohydrate and energy metabolism categories were over-represented more than once. This was consistent with previous reports suggesting these processes provide the carbon skeletons (polysaccharides and fatty acids) needed for cell wall synthesis and are therefore important for fiber development [35,36]. At 1 DPA, the "allene-oxide cyclase activity" (GO:0046423, FDR = 2.12E-04) was enriched in XinWX, which was also enriched in XinFLM at 1 DPA compared to -1 DPA (GO:0046423, FDR = 0.005497). AOC is a key enzyme in the biosynthesis of JA [37]. Plant hormones such as indole-3-acetic acid (IAA), gibberellic acid (GA3) and brassinosteroid (BR) are known to play important roles in cotton fiber initiation [38,39], and these results suggest that JA may also be involved in the progression of fiber initiation.

Expression of AOCs plays a central role in cotton fiber initiation
The enriched allene-oxide cyclase activity GO category included four cotton ESTs, which corresponded to the four members (Gorai.004G046100, Gorai.004G046300, Gorai.004G046400, Gorai.008G185600) in G. raimondii (http://www.phytozome.net). In order to design the genespecific primers, we downloaded the four AOC family genes from G. raimondii. Blast analysis showed that the G. hirsutum acc. TM-1 orthlog of Gorai.004G046100 (GhAOC) has been cloned previously [40], and was named as GhAOC1 in this study, the other three AOCs in TM-1 were named GhAOC2, GhAOC3, and GhAOC4 (corresponding to Gorai.004G046300, Gorai.004G046400, and Gorai.008G185600, respectively). Using gene-specific primers, it was confirmed that all four AOCs were upregulated during fiber initiation in XinWX compared with XinFLM, especially at -1 DPA (Fig 4), and all four AOCs were preferentially expressed in fiber tissues (Fig 5). GhAOC1 was predominantly expressed during fiber development, especially during fiber elongation and secondary cell wall deposition, while expression in root, stem, and leaf was moderate. The GhAOC2 transcript was much more abundant during fiber initiation (-3 to 1 DPA, with peak expression at -1 DPA), and also in fibers, roots, leaves and anther tissues at 25 DPA. GhAOC3 expression also peaked at -1 DPA, and expression was much lower at 5 and 25 DPA in fibers, while expression was negligible in other tissues and organs compared to the peak at -1 DPA. GhAOC4 was preferentially expressed in roots, stems and fibers during the early stages of development (-3 to 5 DPA), with expression again peaking at -1 DPA. Taken together, all four AOCs were preferentially expressed during the fiber initiation stage, especially at -1 DPA. To further investigate the role of AOCs in fiber initiation, their expression levels were measured between -3 and 3 DPA in different fiber mutant lines (Fig 6). Analysis of RT-qPCR data showed that AOCs exhibited a similar expression profile in all mutant lines tested. GhAOC1, GhAOC2 and GhAOC3 were preferentially expressed in ovules at -1 DPA, while GhAOC4 mRNA was abundant at both -1 and 3 DPA. GhAOC2, GhAOC3, and GhAOC4 were more highly expressed in lintless-fuzzless lines (XinWX, Xz-142 WX) (P<0.01) than in linted-fuzzless lines (XinFLM, n2) and linted-fuzzed lines (TM-1, Xz-142) at -1 DPA. In contrast, GhAOC1 was more highly expressed in XinWX and Xz-142 WX than in NILs XinFLM and Xz-142, and expression was lowest in TM-1 and n2 (S4 Table). In summary, expression of AOCs in lintless-fuzzless mutants was generally higher than in linted-fuzzless and linted-fuzzed lines. These results, in combination with expression data from different tissues and organs in TM-1, suggested that upregulation of four AOCs was required for normal fiber initiation, but excessive overproduction of AOCs may lead to a suppressive effect on fiber initiation.

Expression of AOCs is stimulated by JA
In plants, AOCs are known to play a crucial role in JA biosynthesis. In order to investigate the relationship between expression of AOCs and JA, RT-qPCR was performed on leaf RNA of G. hirsutum cv. Jinmian 19, which was treated with JA for different time periods (with water as a control; Fig 7). All four AOCs exhibited a similar expression profile and all were significantly up-regulated following JA treatment, with a peak at 8 h after exposure.

JA biosynthesis-associated genes were up-regulated during fiber initiation in the fuzzless-lintless mutant
To determine whether JA biosynthesis was co-regulated during fiber initiation, seven genes (AOS, 28K_069_H08 and 28K_156_D09; LOX, 28K_275_B02 and 28K_150_B07; OPR, 28K_299_E12; GH3, 28K_086_B03 and 28K_236_A10), which were differentially expressed between XinWX and XinFLM in the microarrays and their homologs reported previously were involved in JA biosynthesis in other plants [41], were subjected to RT-qPCR analysis (Fig 8).  Other than 28K_086_B03 which showed significant expression differences at 1 DPA, the other six DEGs showed significantly higher expression in XinWX than in XinFLM between -1 and 1 DPA, especially at -1 DPA. Of these, two ESTs (28K_086_B03 and 28K_236_A10), encoding GH3 proteins, are known to be important for jasmonate signaling [35,36]. GH3 genes are induced by auxin and conjugate JA with amino acids. GH3.11 conjugates JA with isoleucine, and the resultant jasmonoyl-L-isoleucine (JA-Ile) is a crucial signalling molecule for the major jasmonate signaling pathway involving COI1 [42]. GH3.3, GH3.5, and GH3.6 also conjugate JA to various amino acids and by doing so regulate JA homeostasis in Arabidopsis adventitious root initiation [43]. These results and the expression profiles of the AOCs suggested that the JA biosynthesis pathway is up-regulated during fiber initiation in the fuzzless-lintless mutant, who may result in overproduction of JA in XinWX compared to XinFLM.

Discussion
Several key genes and pathways are responsible for progression of fiber initiation Cotton fiber initiation is a complex biological process involving genes controlling the cell cycle, hormone regulation, signal transduction, cytoskeletal features, and formation and deposition of complex carbohydrates and cell wall proteins [7]. Transcription factors are particularly important for fiber cell initiation [44]. GhMYB25, GhMYB25-like, and GhHD1 were previously identified as regulators of lint fiber initiation [45], and silencing of GhMYB25 in cotton led to delayed fiber initiation and formation of shorter fibers [46], and overexpression of GhHD1 stimulated fiber initiation, while silencing delayed this process [47]. In addition to transcription factors, phytohormones play key roles in plant development and have been well-studied. At anthesis, ovules initiate fiber development in cultures supplemented with both IAA and GA3, whereas ABA, ethylene, and cytokinins inhibit fiber development [30,31,35]. Numerous genes involved in IAA, GA, and BR biosynthesis and signalling pathways have been cloned and confirmed to be associated with the development of fiber initiation [48,49]. Additionally, genes involved in calcium signalling pathways and calcium and ROS homeostasis play crucial regulatory roles during fiber cell initiation and differentiation, and the Sus gene encoding sucrose synthase is also involved [8,17,29]. In this study, the Fisher's exact test was finished to XinWX and XinFLM at three time points during fiber initiation development (-1, 0 and 1 DPA), and transcription factors such as ERFs, MYBs, WRKYs and Zinc-finger proteins were indicative of a positive regulatory effect on early fiber development, indicating important roles in the fiber initiation. Further, in hormone-associated metabolic pathways, GO terms in JA pathway were particularly enriched, especially AOCs, the key enzymes in JA biosynthesis, which suggests that JA is related to progression of fiber initiation.
Additionally, "Lipid transport" and "asparagine biosynthetic process" was also enriched during fiber initiation development. LTPs transport lipids from the endoplasmic reticulum (ER) to the plasma membrane (PM) and the cell exterior, and are known to be associated with fiber elongation [50,51]. LTPs are usually activated during the early stages of fiber development [5], and expression levels peaked at around 10 DPA [52]. Arfinine is the predominant nitrogen storage unit in cotton and accounts for at least 25% of total stored nitrogen, and the majority of nitrogen transported between cotyledons and the growing axis is via the asparagine pathway [53]. Nitrogen availability strongly influences cotton fiber yield and quality, and changes in nitrogen concentration can also influence sucrose inversion and the activity of invertase, sucrose synthetase and sucrose phosphate synthetase [54,55]. In this study, LTPs and genes associated with asparagine biosynthesis were upregulated in the microarray data of XinFLM compared to that of XinWX at 0 and 1 DPA, indicating important roles in the early stages of fiber development for these proteins. Further functional characterization is required to explore this possibility.

Allene oxide cyclases and Jasmonic acid-associated metabolism regulate JA and affect fiber initiation
In plants, JA is synthesized from α-linolenic acid, which is oxygenated by 13-lipoxygenase. Allene oxide synthase (AOS) and allene oxide cyclase activity subsequently results in cis-(+)-12-oxo-phytodienoic acid (OPDA). AOC is required for formation of the cis-(+)-enantiomer (9S, 13S) of OPDA, which is thought to be the precursor of JA [56]. In Medicago truncatula, suppressing MtAOC1 expression in roots lowered JA levels significantly [37], which highlighted the importance of AOC in JA biosynthesis. Genes encoding enzymes involved in JA biosynthesis generally induced by JA in Arabidopsis and tomato, indicating feed-forward regulation [38,57]. JA, OPDA and their derivatives can induce AOC expression in Arabidopsis [58]. Biotic or abiotic stresses that lead to endogenous increases in octadecanoids and jasmonates are usually accompanied by upregulation of AOC, AOS and OPR3 transcription [59][60][61]. AOC transcription and JA levels therefore appear to be coordinated during plant growth and development.
In this study, AOC, which was up-regulated in XinWX, was also enriched in XinFLM at 1 DPA compared to -1 DPA. The expression pattern of AOCs in different tissues, organs, and fibers at different development stages in TM-1 showed that they had preferential expression in fiber initiation stage (-3~1 DPA), especially at -1 DPA, suggesting a critical regulatory role in fiber initiation. Additionally, expression was significantly higher in lintless-fuzzless mutants than in lintedfuzzless and linted-fuzzed accessions, especially at -1 DPA, indicating that overexpression of AOCs may have a negative effect on progression of fiber initiation in lintless-fuzzless lines. These results, in combination with SEM images that showed no protrusions during fiber cell initiation at the ovular surface at 0 DPA in XinWX, indicated that AOCs are important for progression of fiber initiation but may suppress this process upregulated above a threshold value.
The development of cotton fibers shares many similarities with leaf trichome development in Arabidopsis, in which exogenous application of JA increases the number and density of leaf trichomes and has a synergistic effect on trichome induction with Gibberellin [62][63][64][65]. In cotton, continuous exogenous JA application inhibited fiber initiation and elongation, and this effect was dependent on both dosage and development stage [66]. GhTCP, which directly participates in the regulation of JA biosynthesis, influences fiber elongation via a complex system. Overexpression of GbTCP in cotton fibers increased the JA content but resulted in fewer fibers that were also shorter. However, in GbTCP-silenced cotton lines, the mature fibers were also shorter with lower quality [67], indicating that fiber elongation requires an optimal JA concentration. During fiber initiation, JA content at -1 DPA ovules is also higher, and JA response genes are up-regulated [66]. In this study, "jasmonic acid mediated signaling pathway" was enriched within XinWX suggested JA involved in fiber initiation. JA induced the expression of AOCs in cotton, with significant upregulation between 4 and 12h, and a peak in expression at 8 h. The expression profiles showed four AOCs and other genes encoding the main JA biosynthetic enzymes were up-regulated, indicating a high level of JA during the fiber initiation stage. However, expression levels were significantly higher in XinWX than in XinFLM, therefore JA appears to be related to normal fiber initiation in cotton, but overproduction of JA may suppress fiber development. Furthermore, we also sampled -2 DPA ovules from TM-1 treated with JA of different concentrations. After 2 days of culture of exogenous JA application, we found as low as 0.1μM JA to inhibit severly fiber initiation, and the inhibition was dose dependent (S2 Fig). This confirmed that high concentration of JA suppressed fiber development.
Cotton fiber initiation is a complex process involving many different plant hormones. GH3.3, GH3.5, GH3.6 and GH3.11, all of which are induced by auxin, have been shown to be associated with JA homeostasis during initiation of adventitious roots in Arabidopsis [43].
ARF6 and ARF8, transcription factors that are also transcriptionally regulated by auxin, are positive regulators of JA biosynthesis during flower development [68,69]. MYC2, a bHLH transcription factor, have been reported to play a central role within the JA signaling pathway in regulating diverse JA responses [32]. Furthermore, expression of the plant defense gene ERF1 is rapidly activated by either ethylene or jasmonate, and synergistically by both hormones [70]. In this study, two GH3 proteins were differentially expressed in XinWX and XinFLM, with higher levels of expression in XinWX. This is consistent with the expression patterns observed for JA biosynthesis-associated genes. In summary, there appeared to be crosstalk between different hormones during fiber development in cotton. The exact regulatory mechanisms of the cross-talk between JA and other hormones require further investigation.

Conclusions
Developmental mutants are ideal for exploring the key genes and pathways in plant fiber development. In this study, several important genes and pathways involved in cotton fiber initiation were identified, including hormone-associated metabolic pathways, lipid transport and asparagine biosynthetic processes, and various transcription factors. Additionally, JA was found to be related to normal fiber initiation, but overproduction of JA is unfavorable for fiber development.
Supporting Information S1