Analysis of gene expression in cotton fiber initials
© Taliercio and Boykin; licensee BioMed Central Ltd. 2007
Received: 25 September 2006
Accepted: 16 May 2007
Published: 16 May 2007
Cotton (Gossypium hirsutum L.) fibers are trichomes that initiate from the ovule epidermis. Little is known about the developmental pathway causing fiber to differentiate from ovular epidermal cells even though limits on the number of cells that differentiate into fiber will limit yield.
A method was developed to isolate RNA from fiber initials 1 day post anthesis (dpa). Complementary DNA libraries representing 1 dpa fibers and other cotton tissues were sequenced and analyzed. Assembly of G. hirsutum Expressed Sequenced Tags (ESTs) identified over 11,000 sequences not previously represented in GenBank. New genes identified among these ESTs were represented on microarrays. The microarrays were used to identify genes enriched in fiber initials (1 dpa fibers) and elongating fibers. Analyses of Gene Ontologies (GO) of differentially expressed genes determined that terms associated with the "membranes" were statistically over represented among genes increased in expression in fiber initials and 10 dpa fibers. Staining ovules with a fluorescent dye confirmed an increase in Endoplasmic Reticulum (ER) occurred in fiber initials on the day of anthesis, persisted through 3 dpa and was absent in a fiberless mutant. Two genes similar to the CAPRICE/TRIPTYCHON (CPC) gene that inhibits differentiation of leaf trichomes in Arabidopsis were also characterized. Genes associated with novel regulation of brassinosterols, GTP mediated signal transduction and cell cycle control and components of a Ca+2 mediated signaling pathway were identified. Staining of cellular Ca+2 indicated that fiber initials had more Ca+2 than other ovule cells supporting a role for Ca+2 in fiber development.
Analysis of genes expressed in fiber initials identified a unique stage in fiber development characterized by an increase in ER and Ca+2 levels that occurred between 0 and 1 dpa. The gene similar to CPC has a MYB domain but appears to lack a transcription activating domain similar to the Arabisopsis gene. The method used to stain the ER also can be used to count fiber initials and showed fiber cells develop from adjacent cells unlike leaf trichomes.
Trichomes initiated from cotton ovule epidermal cells develop into spinnable fiber. Little is known about fiber initiation even though the number of ovule epidermal cells that differentiate into fiber impacts yield of this important crop. Cotton fibers are single cells that differentiate nearly synchronously through 4 over-lapping stages of development . From about -2 dpa (days post anthesis) to 2 dpa ovular epidermal cells differentiate into fiber initials, from 2 dpa to 21 dpa fibers rapidly elongate up to 5 cm in length, beginning about 16 dpa massive amounts of cellulose are deposited in the secondary cell wall and finally the fiber matures and dries.
Fiber initials represent a minority of epidermal ovule cells and identification of epidermal cells that develop into fibers is difficult before 0 dpa. Ovules cultured in-vitro become competent to produce fiber in response to auxin and giberellic acid 2 days before anthesis . Fiber initiation also requires brassinosterol production . Fiber differentiation is evident in-vivo by -1 dpa when microtubules reorient in epidermal cells destined to differentiate into fibers . On the day of anthesis the amount of golgi bodies and ER increase [3, 4]. By 1 dpa, fiber initials bulge from the surface of the ovule. Protein biosynthesis and nucleoli size increase in very young fibers . In-vitro cultured ovules indicated that mRNA synthesis is required for fiber initiation up to 2 dpa and the ovules remained competent to initiate fibers up to 5 dpa [6–9]. Conservatively, the period of fiber initiation ends at 2 dpa and may extend to 5 dpa. Fiber initiation requires transcription and therefore transcription factors are likely to play an important role in fiber initiation. The Myb109 and MYB2 transcription factors are expressed in fiber initials . The Myb2 transcription factor is able to complement Arabidopsis thaliana trichome mutants and activate expression of R22-like (RDL) gene expressed in fiber initials [11, 12]. Additionally, the RDL gene along with genes involved in cell structure, long chain fatty acid biosynthesis and sterol biosynthesis have been identified that are absent or reduced in a fiberless mutant of cotton . Most of these genes are expressed in 1 dpa ovules. Evaluation of fiberless cotton mutants has identified genes differentially expressed in very young fiber, including transcription factors shown to play roles in fiber development [14, 15]. A second rounds of fiber initiation occurs that produces the short linters or fuzz fibers.
Fiber elongation occurs by a diffuse growth mechanism . Many genes expressed during the elongation stage of fiber differentiation relate to cell expansion, cell wall loosening, and osmoregulation [17–20]. Ovule culture studies confirmed a role for brassinosterols during fiber elongation in addition to fiber initiation . Genomic analyses by Shi et al. indicated that ethylene plays an important role during fiber elongation . The role for ethylene in fiber elongation was confirmed when longer fibers were obtained with the addition of ethylene to ovule culture. An increase in cellulose and expression of genes encoding cellulose synthase marks the end of the rapid elongation stage of the fiber development.
In this investigation a method to isolate RNA from 1 dpa fiber initials is presented. Genes expressed during fiber initiation and elongation were identified using a custom DNA microarrays representing over 11,000 genes, many of which were originally identified for this study. Genes with known patterns of expression were used to validate the microarray data. Additionally, the differential expression of selected genes was also confirmed by RNA blot analysis and semiquantitative PCR. Analyses of gene ontologies (GO) indicated that endomembranes and a GTP signaling pathway increased in developing fibers. Other genes not falling into the GO categories that are differentially regulated during fiber development also provided insight into fiber initiation and elongation. Genes associated with Ca+2 signaling pathways are differentially regulated during fiber initiation and elongation. Differentially regulated genes similar to GLABRA2 (GL2) and Caprice (CPC) which play a role in Arabidopsis trichome and root hair development were also identified [23–25]. These results were supported by histological methods and more detailed analysis of expression of selected genes to broaden our understanding of cotton fiber development.
Isolation of RNA from 1 dpa fiber
List of primers.
forward primer (5'-3')
reverse primer (5'-3')
Summary of cDNA libraries.
fiber 5 dpa (FFT)
lower stem (between the root and first leaf node 7 weeks after planting)
lower stem (between the root and first leaf node 3 weeks after planting)
fiber initials 1 dpa (from membrane bound polyribosomes)
0 dpa ovules from the fiberless mutant (SL1-7-1)
0 dpa ovules from DES119
root 3 week after planting (from free polyribosomes)
root 3 week after planting (from membrance bound polyribosomes)
Roots 10–12 weeks after planting
stems 7 weeks after planting (membrane bound polyribosomes)
stems 3 week after planting
stems 3 week after planting (from membrane bound polyribosomes)
stems 3 weeks after planting (unsuccesful normalization)
Young fiber (1–5 dpa)
random primed normalized
oligo dT primed normalized
Analysis of ESTs
Assemblies of G. hirsutum ESTs identified 4,303 contigs and 8,601 singletons not previously represented among G. hirsutum EST in GenBank (dbEST). About 43% of the contigs and singletons in this assembly were derived solely from ESTs sequenced for this study. Details of the number of new contigs and singletons identified among these ESTs are shown in Table 2. The high rate of recovery of new genes indicated the importance of representing multiple tissues with normalized libraries. Contigs composed of ESTs solely from these new libraries should be enriched in genes represented in 1 dpa fiber and other tissues represented in the EST libraries. Approximately 11,000 sequences, most unique to this assembly, were represented on a microarray in triplicate or quadruplicate.
Validation of microarray data
Expression profiling was performed using the cotton microarray to assess changes in gene expression in fiber initials compared to whole ovule and elongating fiber. Fluorescently labeled cDNA prepared from 1 dpa fiber, representing the period of fiber initiation, was hybridized against 6 microarrays. Three of these arrays were also hybridized with the complementary labeled cDNA derived from 1 dpa ovule RNA and the other three with the complementary labeled cDNA derived from 10 dpa fiber RNA. Benchmark genes, many of which are known to be differentially regulated during fiber initiation and elongation, were represented on the microarray to validate expression (additional file 1) . An mRNA encoding an Acetyltransferase (GhACY) and a FIDDLEHEAD homolog (GhFDH) were more abundant in 1 dpa fiber compared to 1 dpa ovules consistent with reported expression of these genes. RNA encoding a serine carboxypeptidase (GhSCP) and a Beta-tubulin (Ghtub) were increased in 10 dpa fiber compared to 1 dpa fiber. Expression of a GhSCP and Ghtub were consistent with genes increased in expression in 5 dpa fiber as previously reported. Similarly, mRNAs encoding a cellulose synthase, sterol-C-methyltransferase, flavanone 3-beta-hydroxylase, heat shock protein 70 and another serine protease-like protein were not increased in 1 dpa fiber. Nor were these mRNAs increased in 10 dpa fiber with the exception of the mRNA encoding the serine protease-like protein. These data are in good agreement with the previously reports that showed no differential expression of these genes in 5 dpa fiber.
Expression of genes encoding the transcription factors MYB109, MYB(2–6) comported well with published data (additional file 1) [10, 11]. Messenger RNA encoding MYB109 and MYB2 increased in 1 dpa fiber compared to 1 dpa ovules and persisted in 10 dpa fibers. RNA encoding a RD22-like protein (GhRDL) fell slightly below the threshold for increase at 1 dpa but was substantially increased in 10 dpa fibers. The GhRDL gene had been shown to be activated by the MYB2 homolog in Gossypium arboreum. These genes were included on the microarray because their expression has been investigated in young fiber. The agreement of the known expression of these genes with expression of these genes on microarrays validated the microarray results.
Global analysis of gene expression
Genes that vary greater than two fold in expression between 1 dpa fiber and ovules or 1 dpa fiber and 10 dpa fiber are included in additional file 1. Also included in additional file 1 is the significance of the variation. Comparison of expression of genes between 1 dpa fiber and 1 dpa ovules identified 248 transcripts that were down regulated in 1 dpa fiber and 376 transcripts that were up regulated in 1 dpa fiber. Comparison of expression of genes between 1 dpa fiber and 10 dpa fiber identified 390 transcripts that were down regulated in 1 dpa fiber and 165 transcripts that were up regulated in 1 dpa fiber. There were 59 transcripts that were upregulated in 1 dpa fiber compared to both 1 dpa ovules and 10 dpa fiber.
GO analysis of fiber initials.
structural constituents of ribosome
intracellular nonmembrane-bound organelle
establishment protein localization
cell organization and biosynthesis
intrinsic to membrane
cytoplasmic organization and biogenesis
establishment of cellular localization
GO analysis of 10 dpa fiber.
Membrane bound organelles
hydrolyzing-O glycosyl bonds
cell wall modification
GO analysis of genes upregulated in 10 dpa fiber indicated that transcripts associated with organelles (excluding cell membrane and nucleus) decreased (Table 4). An increase was observed for genes associated with cell wall modifications as would be expected for these rapidly growing cells. Even with a decrease in organelles, there was an increase in membranes, consistent with the need for rapidly expanding cell membrane.
Another excellent source of GO annotated genes was UniProt [30, 31]. UniProt cognates of the genes represented on the microarray were also used to analyze the ontologies of genes differentially expressed in fiber initials and elongating fibers. The UniProt cognates were analyzed using GOstat . The GO analyses of the UniProt cognates were similar to the analyses with the Arabidopsis cognates with one exception. Six genes (Contig10324, Contig85, Contig16430, Contig2338, Contig6782, Contig1658) involved in a small GTPase mediated signal transduction pathway (P = 0.0169) were up-regulated in 1 dpa fiber and persisted in 10 dpa fiber [32, 33]. These genes may also play roles in vesicle trafficking.
Validation of GO analysis
The ER of 0 dpa ovules from fiberless mutant was also stained. No coherent staining was observed in the fiberless mutant (Fig. 3A). The collective evidence supported a rapid increase in ER between -1 and 0 dpa in nascent fibers, which was absent in the fiberless mutant.
The ER of expanding leaves was also stained with DiOC (Fig. 3A). Unstained controls showed minimal autofluorescence that was easily distinguishable from the fluorescence due to DiOC staining (data not shown). Leaf trichomes were not stained more intensely than the surrounding cells, indicating they were not enriched in ER relative to nearby cells. These images required very long exposures (~26 s) compared to the DiOC stained ovules (50 ms).
There is currently no easy way to directly measure fiber density. An image of a 0 dpa DiOC stained ovule was digitalized and the stained cells identified and counted (Fig. 3B). DiOC staining provided sufficient contrast to identify and count fiber initials. We are currently developing high through-put methods to count fiber initials in defined areas.
Transcription factor and other genes differentially expressed in fiber initials
Many genes were not included in the GO analysis though they were clearly differentially expressed. Pathways associated with these may have not been identified because of poor representation of the pathway on the microarray of lack of annotated genes associated with these pathways, among many possible reasons. Fifty nine genes were identified that were up-regulated in the fiber initials, and were less abundant in whole ovule and down-regulated in 10 dpa fiber. Four genes that gave insight into various aspects of fiber initiation were identified by eliminating genes with poor annotations and genes with functions already shown to be important in fiber initiation (Contig3145, Contig1481, Contig3407, and Contig7833). A prohibitin that potentially regulates the cell cycle was identified . A MATE efflux protein that may play a role in lateral root initiation  was identified. Semiquantitative rt-PCR confirmed that a transducin and a ribosomal protein were expressed at the highest levels in 1 dpa fiber compared to whole ovules and 10 dpa fiber (Fig. 2B). Transducins are a component of a GTP-mediated signaling pathway not identified in the GO analysis [32, 33]. A putative steroid sulfotransferase mRNA was also upregulated in fiber initials .
Genes other than transcription factors can have profound affects on expression of other genes. Expression of some other types of regulatory genes increased in 1 dpa fibers and persisted in 10 dpa fiber (additional file 1). Examples include receptor kinases, calmodulin, calmodulin binding proteins and lumen receptors (Contig15340, Contig16628, Contig2019, Contig17143 and Contig10804). RNA blot analysis confirmed differential expression of the calmodulin gene and a receptor kinase. (Fig. 2A). The calmodulin encoding mRNA expressed in the fiber samples was slightly larger than the most abundant calmodulin expressed in the whole ovule, indicating a unique calmodulin characterized by a different size was expressed in fibers. The receptor kinase differentially regulated as the fibers mature was detected in 10 dpa fiber. Apparently the level of the putative receptor kinase was not high enough in 1 dpa fiber to be detected. Semiquantitiative PCR confirmed differential expression of a lumen receptor and another receptor kinase (Fig. 2). Note that the semiquantitative PCR successfully detected the increase in expression between ovules and 1 dpa fiber but failed to detect the further increase in expression in 10 dpa fiber for either gene. It is likely that the method was not sensitive enough to detect a further increase.
Transcription factor and other genes differentially expressed during fiber elongation
Genes potentially important in fiber elongation should be differentially expressed in 10 dpa fibers. Nine putative transcription factors were down-regulated in 10 dpa fibers compared to 1 dpa fibers (Contig13, Contig3089, Contig1984, Contig14677, Contig13751, Contig14961, Contig11028, WTOV_01-01-18R_G01, R10M_10R_E12_invR). This pattern of expression was confirmed by rt-PCR for a homeobox protein gene (Fig. 2). Nine putative transcription factors were up-regulated in 10 dpa fibers compared to 1 dpa fibers (Contig17592, Contig7886, Contig549, Contig4608, Contig18656, Contig17085, Contig11492, Contig15981, Contig1963). An mRNA encoding a potential calmodulin binding protein was also up-regulated in 10 dpa fibers consistent with the previously discussed increase in expression of a fiber specific calmodulin gene (Contig9400). An mRNA encoding Rho GDP dissociation inhibitor (GDPDI) was substantially increased between 1 dpa and 10 dpa fiber (TMIRS_147_F04.F). The GDPDI functions in a GTP mediated signaling pathway identified by GO analysis .
Ca+2in fiber initials
The method developed to isolate RNA from fiber initials of Gossypium hirsutum allowed the direct isolation and analysis of genes expressed during fiber initiation. Profiling of gene expression on microarrays identified genes differentially regulated during fiber initiation and elongation. Statistical methods, inclusion of benchmark genes, RNA blot analysis and semiquantitative rt-PCR validated expression profiling data. Statistical analyses of GO also validated the expression profiling data by identifying an increase in protein biosynthesis in fiber initials and cell wall remodeling in elongating fibers consistent with previously reported aspects of fiber development [17, 39].
Six of the genes used to validate the microarray experiments are MYB-type transcription factors. Transcription factors play a global role in control of gene expression. Additionally, their role in differentiation of leaf trichomes, a structure analogous to fiber, is particularly well characterized . Expression of both MYB transcription factors (MYB2 and MYB109) important in fiber development were abundant in 1 dpa fibers and persisted into the elongation stage of fiber development. We have identified 5 transcription factors with a similar pattern of expression that could play a role in fiber development. Contig15274 was very similar to GL2 in Arabidopsis which acts down stream of GL1 to control trichome development [25, 40]. These partial DNA sequences were about 50% identical to other GL2 type genes reported in cotton (AF530913 and AF530914) and 50% identical to an Arabidopsis GL2 gene (NM_106633). In Arabidopsis gl2 mutants result in expanded trichomes and proliferation of root hairs at position where root hairs would not normally develop . An increase in expression of 2 genes similar to the TTG1 genes isolated from cotton which are able to restore trichome formation in the ttg1 Arabidopsis mutant was observed . Two genes similar to CPC were observed in 1 dpa fiber. CPC acts as a negative regulator of trichome development in Arabidopsis . One of the putative CPC genes was down regulated in 1 dpa fiber compared to ovules. The inhibitors described for Arabidopsis are not down regulated in trichomes; therefore it is not possible to draw a conclusion based in gene expression about which putative CPC gene in cotton was more likely involved in fiber development. If CPC genes in cotton act as inhibitors of fiber initiation, reducing expression of these genes with interfering RNAs would be expected to increase the number of fibers. Therefore a transgenic cotton line with reduced CPC expression could be agronomically valuable. In Arabidopsis GL1, GL2, TTG1, and CTC along with other regulatory genes control trichome development via lateral inhibition. The ability of cotton genes to complement the trichome mutant in Arabidopsis thaliana to restore trichomes supported by the presence of putative GL2 and CPC homologs in fiber clearly shows that fiber initiation and development of leaf trichomes use similar genetic mechanisms [12, 38]. However analysis of the distributions of fiber initials stained with DiOC and previously published observations suggest that fibers often develop adjacent to each other (fig. 3A, 40× magnification) . In the "lateral inhibition" model of trichome development trichomes do not normally develop in close proximity. Therefore other factors that interact with the "lateral inhibitory" pathway may be active in controlling fiber initiation.
GO analyses of genes differentially regulated during fiber development identified an increase in membranes specific to ovular trichomes. The hypothesis that membranes increased in fiber initials was confirmed by staining ovules with DiOC. DiOC may also stain mitochondria but the staining pattern was consistent with ER staining, though an increase in mitochondria in these metabolically active cells is also likely. The increase in ER was not evident until 0 dpa, was limited to fiber initials, was absent in a fiberless mutant, and was not observed in leaf trichomes. While the DiOC stain was not quantitative, the longer exposures required for leaves suggested ER levels were not high in leaf trichomes. We cannot eliminate the possibility that a transient ER increase occurred during leaf trichome initiation that was missed in these experiments. However, it is likely that the marked and long lasting increase in ER in fiber initials was unique to the ovular trichomes, indicating an early departure between the developmental programs that give rise to ovular and leaf trichomes. This increase in ER was consistent with the increase in golgi bodies reported in fiber initials. Abundant ER may play a role in biosynthesis and transport for components of the rapidly expanding cell membrane, cell wall and cuticle. Indeed, analysis of genes differentially regulated during fiber initiation and elongation identify numerous genes associated with these developmental pathways (see additional file 1). We propose that the increase in ER represented the first stages of fiber elongation since increase demands for cell membrane, primary cell wall, and cuticle production will persist through the elongation phase of fiber development.
DiOC stains fiber initials with sufficient contrast to allow direct counts of fiber initials. Digitalization of the DiOC stained image and counting DiOC stained ovule cells will allow us to develop protocols to identify cotton germplasms and mutations with increased density of fibers and follow the trait as it is crossed into elite germplasms.
A correlation between fiber initials, ER increase and Ca+2 localization was also observed. ER membranes are stabilized by Ca+2 therefore calcium is probably redistributed to fiber initials. Increased expression of a calmodulin gene unique to fibers, and differential expression of calmodulin binding proteins were also observed. It seems likely that a calmodulin mediated signaling pathway exists that either causes or responds to the redistribution of calcium into ER. Interestingly, deesterified pectins increase in fiber initials . Deesterified pectins bind calcium; therefore it is likely that the cell walls may also compete for Ca+2. Manipulating expression of the calmodulin or manipulating calcium levels in-vitro should determine whether a calcium mediated pathway exists that causes or responds to the increase in ER and what role a calmodulin mediated response to Ca+2 plays in fiber development.
GO analyses of genes up-regulated in fiber initials using UniProt cognates identified genes associated with a small GTPase mediated signal transduction pathway. This pathway has been implicated in transduction of signals in a variety of plant processes including response to light, pathogen responses and regulation of brassinosteroid biosynthesis [32, 33]. This pathway may also play an important role with vesicle trafficking that is consistent with the increased level of Golgi bodies in fiber .
Genes that peak in expression during fiber initiation then decrease in expression during elongation would be expected to play a specific role in fiber initiation. Four well annotated genes with a fiber initiation-specific pattern of expression give potentially new insight into fiber initiation. A putative sterol sulfotransferase may alter brassinisteroids via sulfonation . Brassinosterols play an important role in fiber initiation [2, 21]. A prohibitin-like gene expressed in 1 dpa fiber may play a role in maintaining the fiber as a single cell . The decrease in prohibitin mRNA correlated with an increase in the ploidy level of fibers cells . A MATE efflux protein plays a role in root development and may play a similar role in fiber initiation . Transducins play potential roles in signal transduction, have WD40 repeat motifs and may bind guanosine nucleotides [33, 44, 45]. This transducin was a fiber initiation specific component of GTP mediated signal transduction pathway different from the pathway identified by GO analyses. There are many other genes that are potentially differentially regulated available in the GEO submission (series accession number GSE6855). For example, 16 more transcripts were defined as fiber initial specific at 1.9 fold change of expression and 13 more transcripts were defined as fiber initial specific if a significance of 0.1 was used. Genes evaluated here may also become relevant as annotation of cotton genes improve. This information is available in GEO. Synchronously differentiating fibers represent a valuable developmental model to determine how developmental signals are integrated to control differentiation and elongation of fiber and how these signaling pathways differ between ovular and leaf trichomes.
We present a new method of isolating RNA from very young fibers that allows the direct examination of genes expressed during fiber initiation. Sequencing cDNAs representing genes in a variety of cotton tissues, including fiber initials, has identified numerous genes not previously represented in GenBank. The expression profile of over 11,000 cotton genes, many unique to this investigation, was evaluated using microarrays. GO analysis identified an increase in genes associated with "membranes". Microscopic methods confirmed a marked increase in ER in fiber initials between 0 dpa and 1 dpa. Changes in expression of genes associated with Ca+2 regulation were also observed and Ca+2 concentrations were observed to be higher in fiber initials than surrounding cells. Genes potentially relevant to transcription regulation, brassinosterol regulation, cell cycle regulation and GTP mediated signal transduction were differentially regulated during fiber initiation. Genes associated with the "lateral inhibition" control of thrichome development in Arabidopsis were also present and many were differentially regulated during fiber development. A gene similar to CPC that acts as an inhibitor of trichome development in Arabidopsis was identified in fiber initials and appeared to possess the MYB domain but lack the transacting domain similar to its Arabidopsis counterpart.
RNA isolation, RNA blots and semiquantitative PCR
Stems, roots and the shoot (meristematic region) harvested from field grown DES119 cotton plants (2004) were frozen in liquid nitrogen and ground in liquid nitrogen in a Waring blender (Torrington, CT). Flowers from fields grown (2005) cotton plants (DES119 and ST4793R) were tagged with the date of anthesis and harvested 0 dpa, 1 dpa or 10 dpa. Fiber from 10 dpa ovules was dissected from the ovule, quickly frozen in liquid nitrogen and stored at -80°C. Polyribosomal RNA was isolated from 10 dpa fiber and 1 dpa ovules as described elsewhere [46–48]. Polyribosomal RNA was isolated from 1 dpa fiber by freezing freshly harvested 1 dpa ovules from 50 bolls in an excess of liquid nitrogen, adding about 0.1 g glass beads (Sigma, Atlanta, GA) and vortexing for 5 min. After the liquid nitrogen evaporated but before the sample warmed, 20 ml of the first buffer for polyribosomal RNA isolation was added and the intact ovules removed by filtering through cheese cloth. Free-polyribosomal RNA, membrane bound-polyribosomal RNA and total polyribosomal RNA was isolated as usual. Between 25 μg and 65 μg of total polyribosomal RNA was typically recovered. RNA quality was confirmed on a BioAnalyzer (Agilent, Palo Alto, CA).
RNA was separated on a 1.2% agarose gel (Phosphate buffer, pH6.5) and transferred to positively charged Nytran membrane (Roche, Alameda, CA) as described elsewhere . The probe was amplified from the 3' end of the selected transcripts using the PCR DIG synthesis Kit (Roche). The blot was hybridized, rinsed and visualized following the instructions in the DIG Wash and Block Buffer Set (Roche). RT-PCR and Semiquantitative PCR were described in Taliercio and Kloth . Primer sequences are presented in Table 1.
EST assembly and analysis
ESTs libraries representing unnormalized sequences were prepared as described in Taliercio et al.  and a comprehensive list of tissues represented is shown in Table 2. Vertis Biotechnologie (Freising-Weihenstephan, Germany) made the normalized libraries. Normalization brings the frequencies of most mRNAs within a narrow range . Normalized cDNA libraries representing oligodT primed or randomly primed RNA were made by pooling RNA from meristematic regions, 1–3 dpa fibers, 7 week old roots, 3–10 week old stems. Complementary DNA representing the pooled RNA was PCR amplified 16 (oligodT) or 21(random) cycles using sequences included in the primers and normalized 1 time over an hydroxylapatite column. The single stranded fraction was amplified 11 cycles and cDNAs >~0.6 kb were cloned into the BamHI-EcoRI site of pBSIIsk+. Aliquots of the cloned cDNA were electroporated into TOP10 cells (Invitrogen) and sequenced from the 3'end at the MidSouth USDA/ARS genomics facility. Selected clones were also sequenced from the 5'end. Sequences were carefully trimmed to remove vector sequences and deposited in GenBank Table 2.
These ESTs were assembled into contigs with Gossypium hirsutum ESTs from GenBank using CAP3 . ESTs that do not assemble with others are called singletons. Orientation of the assemblies were determined by the known orientation of any of the component ESTs, presence of a polyA tail at the end of the contig, and/or a strong match (<E-24) with a protein sequence. An ACCESS database archiving the assembly was used to analyze the distribution of ESTs. We reasoned that contigs consisting of ESTs from the new libraries would be good sources of genes expressed in fiber initials (also stems, shoots and roots) since fiber initials were not previously well represented in GenBank. Probes spotted on the microarray were 40 nt-60 nt oligonucleotides identified using "picky (version 2.00)" software that represented the contigs and singletons unique to the new libraries but not contigs and singletons assembled from ESTs already in GenBank . A microarray using these probes and including probes representing seventeen genes with well characterized expression in young fiber already in GenBank was made by Agilent (Palo Alto, CA). Information about the microarray has been deposited in the Gene Expression Omnibus (GEO) at NCBI under the accession number GPL4739. The complete sequences of the genes represented on the microarray were included on the platform submission to GEO.
The Contigs and singleton sequences used on the microarray were used to query GenBank's nonredundant database and the TAIR database with BLASTx. The multiple matches from GenBank were saved, but only the best match that was better than E-10 were saved from the TAIR database. Approximately 74% of the assembled sequences had TAIR cognates that met the query criterion. A similar method was used to identify UniProt cognates. The complete gene sequences for all of the genes represented on the microarray, the sequences of the probes printed on the microarray, expression profiles of all of the genes on the microarray and more extensive BLAST annotation are available as GEO series accession number GSE6855.
Hybridization and analysis of microarrays
RNA isolated from 3 biological replications of 1 dpa intact ovules, 1 dpa fiber, or 10 dpa fiber was labeled with CY3 or CY5 using the MicroMax kit (Perkin Elmer, Wellesley MA). Biological replications were harvested on different days. Three microarrays were hybridized with labeled cDNA presenting 1 dpa fiber and 1 dpa ovules including one dye swap. Three microarrays were hybridized with cDNA representing 1 dpa fiber and 10 dpa fibers including one dye swap. All hybridizations, array scanning and basic analyses were done by MoGene (St Louis, MO). To test for treatment differences, analysis of variance (ANOVA) was performed using SAS (Cary, NC). The F-test was used to test for significance of expression levels. Since hybridization with 1 dpa fiber was the common treatment, an ANOVA using all treatment and assuming common variance among the 3 treatments resulted in a better estimate of error and an increase the precision of the study. The experimental design including all 6 microarrays was an Incomplete Block design with 3 treatments taken 2 at a time. Some of the genes had more than 1 probe on a plate and some genes had more than 1 spot for a single probe resulting in several levels of subsamples. Therefore, the potential sources of variability (random effects) for each gene represented on the array are: 1) overall block effect of chip, 2) chipXtreatment, 3) oligo within chip and treatment and, 4) residual (spot to spot variability for a given probe). A detailed analysis of the statistics will be published in the Proceeding of Applied Statistics in Agriculture. Unless otherwise noted, 2 fold differences in expression supported by a probability of 0.05 or less are considered significant and probabilities of 0.01 or less are considered highly significant. We note that the statistical analysis of three of the benchmark genes fell slightly outside of this range. Data from this microarray series has been deposited in GEO under accession number GSE6855.
Staining ovules and leaves with DiOC and RhodFF
Cotton plants (ST4793R) were grown in the greenhouse (winter 2005–2006 and spring 2006)) supplemented with 6 h of halogen light. Small unexpanded leaves were harvested from ~3 week old plants, washed in water and incubated for 30 min in 25 μM 3, 3'-Dihexyloxacarbocyanin iodide (DiOC) (Sigma), or in water. Ovules harvested from bolls -1 dpa, 0 dpa, 1 dpa or 3 dpa were washed briefly in phosphate buffered saline (PBS) and incubated for 30 min in 25 μM DiOC or in water. Ovules were rinsed in PBS and also stained in 10 uM RhodFF (Invitrogen), washed with 1 ml PBS and incubated at room temperature for 30 min before visualization. Midribs were cut from the leaves and the leaves were places on a slide and covered with 50% glycerol and a cover slip. The ovules were cut in quarters, with one cut along the chalazal ridge to help orient the section and covered in 50% glycerol and a cover slip. Stained ovules and leaves were visualized using fluorescence on the Axioskop 2 using the FITC/EGFP filter (41012) for DiOC or the red filter (41035) for RhodFF (Chroma, Rockingham, VT). Images are of the central portion of the ovule unless otherwise noted. Images were taken with the Axiovision (Zeiss, Thornwood, NY) camera using the Axiovision 4.4 software (Zeiss). Exposures were in the range of 50 ms for stained ovules and much longer for unstained ovules and stained or unstained leaves. DiOC stained images were digitalized and stained cells counted using the nonproprietary software Image J.
We Thank Drs. Jeff Ray, Rickie Turley, Brian Scheffler, Candace Haigler and Josh Udall for critical review of this manuscript. We thank Pameka Johnson for excellent technical assistance. This work was supported by the USDA-ARS CRIS No. 6402-21000-029-00. Mention of trade names or commercial products in this article is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the USDA.
- Graves AD, Stewart MJ: Chronology of the differentation of cotton (Gossypium hirsutum L.) fiber cells. Planta. 1988, 175: 254-258. 10.1007/BF00392435.PubMedView ArticleGoogle Scholar
- Sun Y, Fokar M, Asami T, Yoshida S, Allen RD: Characterization of the Brassinosteriod insensitive 1 genes of cotton. Plant Molecular Biology. 2004, 54: 221-232. 10.1023/B:PLAN.0000028788.96381.47.PubMedView ArticleGoogle Scholar
- Ryser U: Cotton Fiber Initiation and Histodifferentiation. Cotton Fibers. Edited by: Basra SA. Binghamton, NY, Hawthorn Press, 1999:1-46.Google Scholar
- Berlin JD: The Outer Epidermis of the Cottonseed. Cotton Physiology. Edited by: Mauney JR and Stewart JMD. Memphis, The Cotton Foundation, 1986:375-414.Google Scholar
- Van't Hof J: Production of micronucleoli and onset of cotton fiber growth. Planta. 1998, 205: 561-566. 10.1007/s004250050356.View ArticleGoogle Scholar
- Beasley CA, Ting IP: Effects of Plant Growth Substances on In Vitro Fiber Development from Unfertilized Cotton Ovules. Am J Bot. 1974, 61: 188-194. 10.2307/2441189.View ArticleGoogle Scholar
- Beasley CA: Culture of Cotton Ovules. Cell Culture and Somatic Cell Genetics of Plants,vol 1. Edited by: Vasil IK. 1984, Academic Press, 232-240.Google Scholar
- Beasley JO: The Production of Polyploids in Gossypium. The Journal of Heredity. 1940, 31: 39-48.Google Scholar
- Triplett BA: Stage-Specific Inhibition of Cotton Fiber Development by Adding a-Amanitin to Ovule Cultures. In Vitro cellular and Developmental Biology. 1998, 34: 27-33.View ArticleGoogle Scholar
- Suo J, Liang X, Pu L, Zhang Y, Xue Y: Identification of GhMYB109 encoding a R2R3 MYB transcription factor that expressed specifically in fiber initials and elonging fibers of cotton (Gossypium hirsutum L.). Biochimica et Biophysica acta. 2003, 25-34.Google Scholar
- Loguerico LL, Zhang JQ, Wilkins TA: Differential regulation of six novel MYB-domain genes defines two distinct expression patterns in allotetraploid cotton (Gossypium hirsutum L.). Molecular and General Genetics. 1999, 261: 660-671. 10.1007/s004380050009.PubMedView ArticleGoogle Scholar
- Wang S, Wang JW, Yu N, Li CH, Luo B, Gou JY, Wang LJ, Chen XY: Control of Plant Trichome Development by a Cotton Fiber MYB Gene. The Plant Cell. 2004, 16: 2323-2334. 10.1105/tpc.104.024844.PubMedPubMed CentralView ArticleGoogle Scholar
- Li CH, Zhu YQ, Meng YL, Wang JW, Xu KX, Zhang TZ, Chen XY: Isolation of genes preferentially expressed in cotton fibers by cDNA filter arrays and RT-PCR. Plant Science. 2002, 163: 1113-1120. 10.1016/S0168-9452(02)00320-5.View ArticleGoogle Scholar
- Wu Y, Machado AC, White RG, Llewellyn DJ, Dennis ES: Expression Profiling Identifies Genes Expressed Early During Lint Fibre Initiation in Cotton. Plant Cell Physiol. 2006, 47: 107-127. 10.1093/pcp/pci228.PubMedView ArticleGoogle Scholar
- Lee JJ, Hassan OSS, Gao W, Wei NE, Kohel RJ, Cen XY, Payton P, Sze SH, Stelly DM, Chen ZJ: Developmental and gene expression analysis of a cotton naked seed mutant. Planta. 2006, 223: 418-432. 10.1007/s00425-005-0098-7.PubMedView ArticleGoogle Scholar
- Tiwari SC, Wilkins TA: Cotton (Gossypium hirsutum) seed trichomes expand via diffuse growing mechanism. Canadian Journal of Botany. 1995, 73: 746-757.View ArticleGoogle Scholar
- Arpat AB, Waugh M, Sullivan JP, Gonzales M, Frisch D, Main D, Wood T, Leslie A, Wilkins TA: Functional genomics of cell elongation in developing cotton fibers. Plant Molecular Biology. 2004, 54: 911-929. 10.1007/s11103-004-0392-y.PubMedView ArticleGoogle Scholar
- Hasenfratz MP, Tsou CL, Wilkins T: Expression of Two Related Vacuolar H+ -ATPase 16-Kilodalton Proteolipid Genes isdifferentially Regulated in a Tissue-Specific Manner. Plant Physiol. 1995, 108: 1395-1404. 10.1104/pp.108.4.1395.PubMedPubMed CentralView ArticleGoogle Scholar
- Wan CY, Wilkins TA: Isolation of multiple cDNAs encoding the vacuolar H(+)-ATPase subunit B from developing cotton (Gossypium hirsutum L.) ovules. Plant Physiol. 1994, 106: 393-394. 10.1104/pp.106.1.393.PubMedPubMed CentralView ArticleGoogle Scholar
- Ji SJ, Lu YC, Feng JX, Wei G, Li J, Shi YH, Fu Q, Liu D, Luo JC, Zhu YX: Isolation and analyses of genes preferentially expressed during early cotton fiber development by subtractive PCR and cDNA array. Nucl Acids Res. 2003, 31: 2534-2543. 10.1093/nar/gkg358.PubMedPubMed CentralView ArticleGoogle Scholar
- Sun Y, Veerabomma S, bdel-Mageed HA, Fokar M, Asami T, Yoshida S, Allen RD: Brassinosteroid Regulates Fiber Development on Cultured Cotton Ovules. Plant Cell Physiol. 2005, 46: 1384-1391. 10.1093/pcp/pci150.PubMedView ArticleGoogle Scholar
- Shi YH, Zhu SW, Mao XZ, Feng JX, Qin YM, Zhang L, Cheng J, Wei LP, Wang ZY, Zhu YX: Transcriptome Profiling, Molecular Biological, and Physiological Studies Reveal a Major Role for Ethylene in Cotton Fiber Cell Elongation. The Plant Cell. 2006, 18: 651-664. 10.1105/tpc.105.040303.PubMedPubMed CentralView ArticleGoogle Scholar
- Wada T, Kurata T, Tominaga R, Koshino-Kimura Y, Tachibana T, Goto K, Marks MD, Shimura Y, Okada K: Role of a positive regulator of root hair development, CAPRICE, in Arabidopsis root epidermal cell differentiation. Development. 2002, 129: 5409-5419. 10.1242/dev.00111.PubMedView ArticleGoogle Scholar
- Schellmann S, Schnittger A, Kirik V, Wada T, Okada K, Beermann A, Jürgens G, Hülskamp M: TRIPTYCHON and CAPRICE mediate lateral inhibition during trichome and root hair patterning in Arabidopsis. The EMBO Journal. 2002, 21: 5036-5046. 10.1093/emboj/cdf524.PubMedPubMed CentralView ArticleGoogle Scholar
- Di Cristina M, Sessa G, Dolan L, Linstead P, Baima S, Ruberti I, Morelli G: The Arabidopsis Ath-10 (GLABRA2) is an HD-Zip protein required for regulation of root hair development. The Plant Journal. 1996, 10: 393-402. 10.1046/j.1365-313X.1996.10030393.x.PubMedView ArticleGoogle Scholar
- Yang S, Cheung F, Lee JJ, Ha M, Wei NE, Sze SH, Stelly DM, Thaxton P, Triplett B, Town CD, Jeffrey Chen Z: Accumulation of genome-specific transcripts, transcription factors and phytohormonal regulators during early stages of fiber cell development in allotetraploid cotton. The Plant Journal. 2006, 47: 761-775. 10.1111/j.1365-313X.2006.02829.x.PubMed CentralView ArticleGoogle Scholar
- Covitz PA, Smith LS, Long SR: Expressed Sequence Tags from a Root-Hair-Enriched Medicago truncatula cDNA Library. Plant Physiol. 1998, 117: 1325-1332. 10.1104/pp.117.4.1325.PubMedPubMed CentralView ArticleGoogle Scholar
- Consortium GO: The Gene Ontology (GO) database and informatics resource. Nucl Acids Res. 2004, 32: D258-D261. 10.1093/nar/gkh036.View ArticleGoogle Scholar
- Beissbarth T, Speed T: GOstat: Find statistically overrepresented Gene Ontologies within a group of genes. 2004, 1: 1-2. [http://gostat.wehi.edu.au/]Google Scholar
- Camon E, Magrane M, Barrel D, Lee V, Dimmer E, Masien J, Binns D, Harte N, Lopez R, Apweiler R: The Gene Ontology Annotation (GOA) Database: sharing knowledge in Uniprot with Gene Ontology. Nucl Acids Res. 2004, 32: D262-D266. 10.1093/nar/gkh021.PubMedPubMed CentralView ArticleGoogle Scholar
- Schneider M, Bairoch A, Wu CH, Apweiler R: Plant Protein Annotation in the UniProt Knowledgebase. Plant Physiol. 2005, 138: 59-66. 10.1104/pp.104.058933.PubMedPubMed CentralView ArticleGoogle Scholar
- Yang Z: Small GTPases:Versatile Signaling Switches in Plants. The Plant Cell. 2002, S375-S388.Google Scholar
- Assmann SM: Heterotrimeric and Unconventional GTP Binding Proteins in Plant Cell Signaling. The Plant Cell. 2002, 14: S355-S373.PubMedPubMed CentralGoogle Scholar
- Ruzn SE: Histochemistry and Cytochemistry. Plant Microtechniques and Microscopy. 1999, New York, Oxford University Press, 145-176.Google Scholar
- Pih KT, Yi MJ, Liang YS, Shin BJ, Cho MJ, Hwang I, Son D: Molecular Cloning and Targeting of a Fibrillarin Homolog from Arabidopsis. Plant Physiol. 2000, 123: 51-58. 10.1104/pp.123.1.51.PubMedPubMed CentralView ArticleGoogle Scholar
- Diener AC, Gaxiola RA, Fink GR: Arabidopsis ALF5, a Multidrug Efflux Transporter Gene Family Member, Confers Resistance to Toxins. The Plant Cell. 2001, 13: 1625-1638. 10.1105/tpc.13.7.1625.PubMedPubMed CentralView ArticleGoogle Scholar
- Rouleau M, Marsolais F, Richard M, Nicolle L, Voigt B, Adam G, Varin L: Inactivation of Brassinosteroid Biological Activity by a Salicylate-inducible Steroid Sulfotransferase from Brassica napus. J Biol Chem. 1999, 274: 20925-20930. 10.1074/jbc.274.30.20925.PubMedView ArticleGoogle Scholar
- Humphries JA, Walker AR, Timmis JN, Orford SJ: Two WD-repeat genes from cotton are functional homologues of the Arabidopsis thaliana TRANSPARENT TESTA GLABRA1 (TTG) gene. Plant Molecular Biology. 2005, 57: 67-81. 10.1007/s11103-004-6768-1.PubMedView ArticleGoogle Scholar
- Ryser U, Holloway PJ: Ultrastructure and chemistry of soluble and polymeric lipids in cell walls from seed coats and fibres of Gossypium species. Planta. 1985, 163: 151-163. 10.1007/BF00393501.PubMedView ArticleGoogle Scholar
- Rerie WG, Feldmann KA, Marks MD: The glabra 2 gene encodes a homeo domain protein required for normal trichome development in Arabidopsis. Genes and Development. 1994, 8: 1388-1399. 10.1101/gad.8.12.1388.PubMedView ArticleGoogle Scholar
- Turley RB, Vaughn KC: The primary walls of cotton fibers contain an ensheating pectin layer. Protoplasma. 1999, 209: 237.Google Scholar
- Nadimpalli R, Yalpani N, Johal GS, Simmons CR: Prohibitins, Stomatins, and Plant Disease Response Genes Compose a Protein Superfamily That Controls Cell Proliferation, Ion Channel Regulation, and Death. J Biol Chem. 2000, 275: 29579-29586. 10.1074/jbc.M002339200.PubMedView ArticleGoogle Scholar
- Taliercio E, Hendrix B, Stewart JMD: DNA content and expression of genes related to cell cycling in developing Gossypium hirsutum (Malvaceae) fibers. Am J Bot. 2005, 92: 1942-1947.PubMedView ArticleGoogle Scholar
- Neer EJ, Schmidt CJ, Nambudripad R, Smith TF: The ancient regulatory-protein family of WD-repeat proteins. Nature. 1994, 371: 297-300. 10.1038/371297a0.PubMedView ArticleGoogle Scholar
- Nielsen NC, Beilinson V, Bassuner R, Reverdatto S: A Gb-like protein from soybean. Physiologia Plantarum. 2001, 111: 75-82. 10.1034/j.1399-3054.2001.1110110.x.View ArticleGoogle Scholar
- Larkins BA, Davis E: Polyribosomes from Peas. Plant Physiol. 1973, 52: 655-659.PubMedPubMed CentralView ArticleGoogle Scholar
- Mignery GA, Pikaard CS, Hannapel DJ, Park WD: Isolation and sequence analysis of cDNAs for the major tuber protein patatin. Nucl Acids Res. 1984, 12: 7987-8000. 10.1093/nar/12.21.7987.PubMedPubMed CentralView ArticleGoogle Scholar
- Taliercio E, Ray JD: Identification of transcripts translated on free or membrane-bound polyribosomes by differential display. Plant Molecular Biology Reporter. 2001, 19: 269a-269f.View ArticleGoogle Scholar
- Taliercio E, Kim YJ, Mahe A, Shanker S, Choi J, Cheng HW, Prioul LJ, Chourey PS: Isolation,Characterization and Expression Analyses of Two Cell Wall Invertase Genes in Maize. Journal of Plant Physiology. 1999, 155: 197-204.View ArticleGoogle Scholar
- Taliercio E, Kloth R: Expression and Characterization of a UDP-Glucose Pyrophosphorylase Gene in Cotton. Journal of Cotton Science. 2004, 8: 91-98.Google Scholar
- Taliercio E, Allen RD, Essenberg M, Klueva N, Nguyen H, Patil MA, Payton P, Millena ACM, Phillips AL, Pierce ML, Scheffler B, Turley R, Wang J, Zhang D, Scheffler J: Analysis of ESTs from multiple Gossypium hirsutum tissues and identification of SSRs. Genome. 2006, 49: 306-319. 10.1139/G05-115.PubMedView ArticleGoogle Scholar
- Soares MB, Bonaldo MDF, Jelene P, Su L, Lawton L, Efstatiadis A: Construction and chatacterization of a normalized cDNA library. Proceedings of the National Academy of Science USA. 1994, 91: 9228-9232. 10.1073/pnas.91.20.9228.View ArticleGoogle Scholar
- Huang X, Madan A: CAP3: A DNA Sequence Assembly Program. Genome Res. 1999, 9: 868-877. 10.1101/gr.9.9.868.PubMedPubMed CentralView ArticleGoogle Scholar
- Chou HH, Hsia AP, Mooney DL, Schnable PS: Picky: oligo microarray design for large genomes. Bioinformatics. 2004, 20: 2893-2902. 10.1093/bioinformatics/bth347.PubMedView ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.