Identification and positional distribution analysis of transcription factor binding sites for genes from the wheat fl-cDNA sequences

Abstract

The binding sites of transcription factors (TFs) in upstream DNA regions are called transcription factor binding sites (TFBSs). TFBSs are important elements for regulating gene expression. To date, there have been few studies on the profiles of TFBSs in plants. In total, 4,873 sequences with 5ʹ upstream regions from 8530 wheat fl-cDNA sequences were used to predict TFBSs. We found 4572 TFBSs for the MADS TF family, which was twice as many as for bHLH (1951), B3 (1951), HB superfamily (1914), ERF (1820), and AP2/ERF (1725) TFs, and was approximately four times higher than the remaining TFBS types. The percentage of TFBSs and TF members showed a distinct distribution in different tissues. Overall, the distribution of TFBSs in the upstream regions of wheat fl-cDNA sequences had significant difference. Meanwhile, high frequencies of some types of TFBSs were found in specific regions in the upstream sequences. Both TFs and fl-cDNA with TFBSs predicted in the same tissues exhibited specific distribution preferences for regulating gene expression. The tissue-specific analysis of TFs and fl-cDNA with TFBSs provides useful information for functional research, and can be used to identify relationships between tissue-specific TFs and fl-cDNA with TFBSs. Moreover, the positional distribution of TFBSs indicates that some types of wheat TFBS have different positional distribution preferences in the upstream regions of genes.

Graphical abstract

Identification and positional distribution analysis of transcription factor binding sites for genes from the wheat fl-cDNA sequences

Key words:

• transcription factor binding site
• Triticum aestivum
• positional distribution preferences
• tissue

Common wheat (Triticum aestivum L.) is one of the most important crops, accounting for ~20% of the calories consumed by humans worldwide.¹⁾ To meet the demands of an increasing global population, many methods have been explored to improve wheat output and grain quality, such as molecular breeding and genomic selection.^1–3) Wheat is widely cultivated in various climates and soil environments. During their lifetimes, wheat plants might suffer one or more types of abiotic and biotic stresses, such as fungal infection, drought, and temperature stress. Therefore, studies on gene expression and regulation are usually conducted by comparing the expression of a set of genes under normal conditions to reference genes under a stress condition.^4–6)

Transcription factors (TFs) activate and regulate the expression of specific downstream genes.⁷⁾ Each TF has one or several specific sequences called the DNA-binding domain (DBD), which can bind to target DNA sequences of specific downstream genes. TFs can be classified into different families according to their DBDs. Studies of TFs have included two phases; the first phase involved functional research on one or several TFs by experimentation. During this stage, the regulatory mechanisms of various members of different TF families were well studied, such as MYB, bZIP, and WRKY.^8–10) The second phase has been genome-wide prediction of the TFs of different families based on HMM profiles. There are currently three wheat TF databases, wDBTF, PlantTFDB, and WheatTFDB, classified by family.^11–13) In particular, the WheatTFDB provides developmental stage information and tissue sources for each TF, and includes 1313 developmental stage-specific TF sequences and 1145 tissue-specific TF sequences.

The binding sites (approximately 5 to 12-bp short sequences) of TFs in upstream DNA regions are called transcription factor binding sites (TFBSs). When TFs are anchored at their TFBSs, they activate and regulate various biological processes, such as organ development, tissue differentiation, and abiotic and biotic stress responses.¹⁴⁾ Moreover, the numbers of TFBSs can affect the expression of downstream genes. Guo et al. found that the numbers of TaGAMyb binding sites were correlated with the abundance of high-molecular-weight glutenin subunit (HMW-GS) proteins, through binding of the promoters of wheat HMW-GS genes.¹⁵⁾ The regulation mechanisms and relationships between TFs and their TFBSs have been experimentally confirmed in many plants.^16–19) However, studies on the functions and molecular mechanisms of TFBSs can be time-consuming and labor-intensive if there is no prior information.²⁰⁾ Recently, some TFBS profiles related to plant stress responses have been published in bioinformatics methods articles.^14,21) However, a systemic study of wheat TFBSs in silico has not been reported. The recently published survey sequences of the wheat genome enable us to investigate the profiles of different TFBS types. Ordinarily, TFBS prediction is based on position weight matrices (PWMs), which quantitatively score known TFBSs.²²⁾ The study of the TFBS positional distribution in upstream DNA regions is one aspect of gene expression and regulation research. In eukaryotes, the distal region of the start codon plays a central role in binding TFs that regulate gene expression.²³⁾ Some recent studies have indicated that the proximal region dominates transcriptional regulation in general, or in particular biological contexts, even in mammals.²³⁾ To date, there have been only a few systematic studies on the positional distribution of TFBSs. In the present study, we attempted to analyze the positional distribution of TFBSs in wheat based on TFBS prediction. The main objectives of this study were: (i) to analyze the TFBS profiles of wheat full-length cDNAs (fl-cDNA) and the associated tissue information and (ii) to analyze the positional distribution of TFBSs in wheat DNA.

Materials and methods

Transcription factor binding site analysis

The fl-cDNA sequences of wheat were used to analyze their TFBSs. Collectively, 8530 wheat fl-cDNA sequences were obtained from the Triticeae full-length CDS database (TriFLDB, http://trifldb.psc.riken.jp/v3/download.pl).²⁴⁾ The TFBSs from the fl-cDNA sequence analysis pipelines are shown in Fig. 1 The TFBSs of genes are always located upstream of the gene. To determine the upstream regions of the 8530 fl-cDNAs, their positions in the genome were first identified. The BLAST V2.2.18 program was used to search for the genome locations of the 8530 fl-cDNAs against the wheat assembled genome (IWGSC1.0 + popseq.27, downloaded from Ensemble plants; ftp://ftp.ensemblgenomes.org/pub/plants/release-27/fasta/triticum_aestivum/).²⁾ The 100-bp sequences following the transcriptional start site of the 8530 fl-cDNAs were used as input data for BLAST searches (e: 1e-10; m: 8; F: T). The input sequence IDs for BLAST and their transcriptional start site locations are listed in S1 Table. The BLAST output is shown in S1 File. For accurate selection, the subject sequence locations found in the wheat genome were considered fl-cDNA locations if they met the following criteria: (i) the alignment length of the subject sequence was 100 bp, (ii) the identity value was greater than 0.95 (S1 File). Among the 8530 fl-cDNAs, 4873 fl-cDNAs were found in the genome at 7065 locations (S2 Table). We retrieved the upstream regions for these 7065 locations, and their lengths were no longer than 2000 bp. Next, TESS V1.0²⁵⁾ was employed to predict the TFBSs of the 7065 upstream DNA sequences. TESS searched for TFBSs in the DNA sequences using PWMs. The PWMs used in this research were downloaded from the JASPAR CORE database (http://jaspar.genereg.net/html/DOWNLOAD/). Sixty-four non-redundant plant PWMs were downloaded, but only 53 matrices were found in wheat according to the matrix annotations in the JASPAR CORE database. The details of the 53 collected matrices, which were distributed in 14 TF families/superfamilies, are listed in S3 Table. We established a new Dof PWM according to the published wheat Dof TFBS sequence (5′-TGTAAAG-3′),²⁶⁾ based on the method of Wasserman and Sandelin.²⁷⁾ For the 54 PWMs to work with TESS correctly, the matrices needed to be transposed. The TESS analysis (fmt: t; mlo: 12.0) results for the 7065 upstream sequences are deposited in S2 File. According to the TFBS matrices (S3 Table), we classified the TFBS information of the 4873 fl-cDNAs in 7065 upstream sequences from S2 File into 14 groups. The classified TFBS information results are listed in S4 Table. The TFBS information for the 4873 fl-cDNAs organized into 14 TF families was collected and arranged as Table 1. To evaluate the differences among the average numbers of TFBSs of the 14 types in each fl-cDNA, a one-sample t-test was performed using the IBM SPSS statistics software V21.0 (SPSS Inc., Chicago, IL, USA).

Fig. 1. Schematic description of wheat transcription factors and their binding sites analysis workflow.

Table 1. TFBS features of 4873 fl-cDNA sequences.

TFBS type	MADS	bHLH	B3	HB	ERF	AP2/ERF	SBP	MYB	bZIP	RAV	TCP	BES1	WRKY	Dof
Number of PWMs	12	9	3	4	1	3	3	7	2	2	1	1	1	5
Number of TFBS	4572	1951	1951	1914	1820	1725	1365	1114	976	897	474	459	300	924
Number of fl-cDNA	3229	1537	1627	1220	1607	1240	992	763	661	708	270	330	263	553
Average	1.416	1.269	1.199	1.569	1.133	1.391	1.376	1.46	1.477	1.267	1.756	1.391	1.141	1.671

Notes: TFBS: transcription factor binding site; TF: transcription factor; fl-cDNA: full length cDNA; PWMs: positional weight matrix models.

Average number of TFBS in each fl-cDNA represents its number of TFBS divided by the number of fl-cDNA of this TF family.

RNA-seq data for 15 wheat tissue sequences were used to analyze the expression features of the 4873 fl-cDNAs. To investigate the TFBSs and TF profiles related to grain development, we used seven RNA-seq data-sets from grain cells at crucial developmental stages. The tissue and developmental stage information for the 15 RNA-seq data-sets is listed in S5 Table, such as germinated seeds, roots, and starchy endosperm at 30 days post anthesis (DPA). First, Tophat V2.0.14 was employed to align the 15 RNA-seq reads to the 4873 fl-cDNA sequences.²⁸⁾ Next, the Cufflinks program was used to match these alignments against the 4873 fl-cDNA sequences to assemble the reads into transcripts.²⁸⁾ We used the default parameters of Tophat and Cufflinks to assemble the transcript files. There were 15 assembled transcript files. The tissue-related features of the 4873 fl-cDNAs in the 15 wheat tissues are shown in S6 File. Furthermore, we integrated the TFBS information of the 4873 fl-cDNA sequences in S4 Table with the tissue-related features in S6 Table. The integrated tissue and TFBS results of the 4873 fl-cDNA sequences are shown in S7 Table. Table 2 shows the number of fl-cDNAs with detected TFBSs in different tissues. To investigate the distribution of the TFBSs of the 4873 fl-cDNAs in different tissues, we calculated their percentages in each tissue. We used the fl-cDNA numbers from Table 2 and divided by the number of fl-cDNAs found in Table 1. For example, for fl-cDNA with AP2 TFBSs identified in roots we used 863 from Table 2 and divided by 1240 in Table 1. The resulting value was the percentage of fl-cDNA with AP2 TFBSs identified in root tissue. Fig. 2 shows the percentages for fl-cDNA with different TFBSs expressed in the 15 tissues.

Table 2. The number of fl-cDNA sequences with different TFBS expressed indifferent tissues.

TFBS type/Tissue	Root	Flag leaf	Shoot	Pistil	Stamen	Germination seed	Inner pericarp	Outer pericarp	REF 20DPA	SE 20DPA	AL 20DPA	TC 20DPA	SE 30DPA	AL & SE 30DPA	10DPA
AP2/ERF	863	685	812	869	905	875	698	718	653	524	693	664	559	657	609
B3	985	772	912	956	1001	968	784	808	702	581	746	727	626	721	661
BES1	205	147	191	194	209	198	154	159	145	123	152	146	129	149	138
bHLH	1029	785	945	990	1045	1014	818	843	765	633	797	779	677	776	717
bZIP	398	299	374	378	401	389	302	316	270	218	289	281	239	282	253
ERF	1086	846	1015	1072	1115	1082	883	912	812	664	853	825	702	810	749
HB superfamily	722	583	686	686	722	698	583	594	539	443	567	542	481	543	514
MADS	1998	1656	1937	1998	2080	1976	1669	1692	1579	1338	1657	1585	1417	1594	1521
MYB superfamily	546	421	514	529	544	527	441	457	404	333	425	416	361	407	376
RAV	452	358	431	452	460	458	384	407	364	315	384	367	333	370	360
SBP	597	485	575	601	617	606	501	501	475	394	495	477	413	478	470
TCP	192	155	183	182	189	189	147	152	140	118	143	138	131	142	138
WRKY	160	125	150	150	156	160	123	128	121	96	133	113	100	123	107
Dof	317	258	314	316	326	319	266	270	250	215	271	254	221	261	246

Notes: TFBS: Transcription factor binding site; TF: transcription factor; fl-cDNA: full length cDNA; REF: whole endosperm; SE: starchy endosperm; AL: aleuron layer; TC: transfer cells; DPA: days post anthesis; 10DPA: whole endosperm at 10 days post anthesis.

Fig. 2. Percentage of fl-cDNA with different TFBS types expressed in 15 wheat tissues.

Note: The value of each bar is the number of fl-cDNA with TFBS in this tissue dividing by its total number of TFBS.

Tissue information in wheat TF sequences

The tissue information of the TF sequences for the 14 families mentioned above was also analyzed. The original EST sequences of the 14 TF families were downloaded from WheatTFDB (http://xms.sicau.edu.cn/wheatTFDB/). In all, there were 1226 EST sequences distributed among the other 14 TF families. Initially, the transcript files of the 1226 EST sequences in the 15 RNA-seq data-sets were obtained by the same method as the 4873 fl-cDNA sequences. Tissue information for the 1226 sequences was retrieved from the 15 transcript files, and is shown in S8 Table. However, the 1226 TF sequences from WheatTFDB showed expression information for six tissues and six different developmental stages, including leaves, germinated seeds, and roots. Using the expression information for the 1226 TF sequences in WheatTFDB, tissue information for TF members in germinated seeds and roots was added to S8 Table (the added information is marked with a green background). According to the tissue information, the expression information for the 14 TF families in 15 tissues was collected as Table 3. To explore the TF distribution preferences in different tissues, the percentages of TF members in each tissue were calculated. We used the TF numbers from different tissues in Table 3 divided by the total TF number. Fig. 3 shows the percentages of different TFs distributed in the 15 tissues.

Table 3. The number of TFs from different tissues.

TF family	Root	Flag leaf	Shoot	Pistil	Stamen	Germination seed	Inner pericarp	Outer pericarp	REF 20DPA	SE 20DPA	AL 20DPA	TC 20DPA	SE 30DPA	AL & SE 30DPA	10DPA	Total TF number
AP2	8	4	6	5	8	7	3	4	4	3	5	3	2	4	4	9
B3	67	34	40	90	86	65	33	41	34	21	49	45	24	33	48	116
BES1	4	1	4	4	4	4	1	1	1	1	4	4	1	1	1	4
bHLH	83	33	54	63	71	69	28	44	29	23	29	27	22	26	31	140
bZIP	90	57	67	79	88	81	63	57	61	58	63	64	58	61	62	110
ERF	128	32	73	94	100	77	40	49	40	29	58	43	29	42	37	217
HB superfamily	56	27	38	46	50	53	19	30	21	16	25	27	18	25	25	91
MADS	29	20	20	50	41	30	25	34	9	6	9	15	10	10	18	91
MYB superfamily	173	78	110	133	147	140	74	83	87	67	99	88	62	92	88	255
RAV	4	3	3	1	2	3	1	1	1	1	1	1	1	1	1	4
SBP	8	4	7	12	9	7	3	3	6	5	5	5	4	5	6	12
TCP	8	1	8	9	7	9	1	0	1	0	2	2	1	1	1	10
WRKY	87	55	66	74	80	56	23	48	28	22	32	32	20	25	29	135
Dof	21	11	16	21	20	23	7	10	6	6	9	6	4	6	6	32
Average	55	26	37	49	51	45	23	29	23	18	28	26	18	24	26

Notes: TFBS: Transcription factor binding site; TF: transcription factor; fl-cDNA: full length cDNA; REF: whole endosperm; SE: starchy endosperm; AL: aleuron layer; TC: transfer cells; DPA: days post anthesis; 10DPA: whole endosperm at 10 days post anthesis.

Fig. 3. Percentage of transcription factor (TF) sequences in 15 tissues.

Note: The value of each bar represents the percentage which the number of TFs divides by its total number of TFs.

Positional distribution of TFBS

To investigate the positional distribution preferences of the TFBSs, the upstream regions of the fl-cDNAs, from −1 to −2000 bp, were divided into seven distinct regions. The first region was from −1 to −200 bp, and the remaining regions from −201 to −2000 bp were divided into six equal parts. The number of TFBSs distributed in each region was counted, and the total number of different TFBS types was also counted (S2 File). We calculated the ratios of TFBSs in each region by dividing by the total number of distinct TFBSs. Finally, a one-sample t-test was applied to test the difference of TFBS positional distribution in upstream regions of the fl-cDNAs using the IBM SPSS statistics software V21.0 (SPSS Inc., Chicago, IL, USA).

Results

In total, 8530 sequences of wheat fl-cDNAs obtained from TriFLDB were used to characterize the TFBSs for TFs. The BLAST program was employed to identify the locations of the 8530 fl-cDNAs in the wheat genome. There were 4873 fl-cDNAs at 7065 locations in the genomic sequence of the cultivar Chinese Spring (S2 Table). The upstream sequences of the fl-cDNAs at the 7065 locations were used to search for TFBSs using the TESS program. The information for the identified TFBSs of the fl-cDNAs, which were separated into 14 TFBS types, is shown in Table 1. There are four rows in Table 1: “Number of PWMs” (positional weight matrix models), “Number of TFBSs,” “Number of fl-cDNAs,” and “Average number of TFBSs in each fl-cDNA.” “Number of PWMs” is the number of PWMs of each TFBS type used in the TESS program (S3 Table). “Number of TFBSs” is the number of different TFBSs predicted in each of the 14 types (S2 Note). “Number of fl-cDNAs” shows the number of fl-cDNA sequences (S4 Table) in which TFBSs were found in the upstream region. The contents of “Number of TFBSs” are sorted in descending order. The TFBS number for the MADS family was 4572, which was the largest among the 14 types of TFBSs. There were twice as many TFBSs for MADS TFs as for bHLH, B3, HB superfamily, ERF, and AP2/ERF TFs, and four times as many as for the remaining eight TF types. Only one PWM was used to predict ERF binding sites, but 1820 TFBSs were found for ERFs. The average number of TFBSs in each fl-cDNA ranged from 1.133 to 1.756 sites. The TCP family showed the largest number of TFBSs in each fl-cDNA. A one-sample t-test was applied to evaluate the differences among the 14 averages; the P-value of the two-tailed test was 1.000 (>0.05).

The numbers and types of TFBSs identified in the upstream sequences of the 4873 fl-cDNAs are shown in Table 2. The numbers of fl-cDNAs showed large differences among tissues and TFBS types. The four largest TFBS types were MADS, ERF, bHLH, and B3. There were much fewer TFBSs in the fl-cDNA upstream sequences for WRKY, TCP, and BES1 TFs than for the other TFBS types in the 15 wheat tissues. More than 60% of the TFBSs of each type were distributed in roots, shoots, pistils, stamens, and germinated seeds, which was greater than in grain cell types (Fig. 2).

The 1226 TF sequences were separated into 14 TF groups and 15 tissue types (Table 3). There were no more than 10 TFs of the AP2, BES1, RAV, SBP, and TCP families in any tissue. In particular, there were no TCP TF sequences predicted in the outer pericarp and SE 20 DPA tissues. The MYB superfamily, ERF, bHLH, and WRKY were the four largest families, accounting for 60.93% of the 1226 TFs (Table 3). The percentages of TFs from different wheat tissues in the 14 TF families are shown in Fig. 3 The bZIP family was clearly the most common type; more than 60% of bZIP TFs were detected in all tissues (Fig. 3). More than 50% of the TFs in the HB and MYB superfamilies were detected in roots, pistils, stamens, and germinated seeds. Moreover, more than 50% of the members of each of the 14 TF families were found in roots, which had the largest number of TFs among all tissues.

The distribution and frequency of each TFBS type in seven regions (−1 to −2000 bp) of the fl-cDNA upstream sequences was determined (Fig. 4). All of the P-values for TFBS positional distribution of the fl-cDNAs upstream region were less than 0.05, which used to test the distribution difference of TFBS by one-sample t-test. About 1/3 of the SBP type TFBSs were detected from −201 to −500 bp, and 30% of the BES1 type TFBSs were found within −501 to −800 bp of the upstream region. About half of the HB type TFBSs were detected in the upstream region from −801 to −1400 bp. Additionally, the AP2, WRKY, and TCP types of TFBSs were most common between −1401 and −1700 bp of the fl-cDNA upstream sequences. High frequencies of TFBSs for AP2, ERF, Dof, and MYB TFs were found between −1701 and −2000 bp.

Fig. 4. Positional distribution of wheat transcription factor binding sites (TFBS).

Note: X-axis is the upstream region from −1 to −2000 bp of fl-cDNA. Y-axis is the percentage of different TFBS types, which represents the number of TFBS in this region dividing by total number of different TFBSs.

Discussion

Studying the regulation of gene expression by TFs requires elucidation of their TFBS profiles.²⁹⁾ TFBSs are cis-elements that play important roles in the regulation of gene expression in different tissues and at specific developmental stages. The interactions of one or more TFBS types and TFs have been reported for some plants.^30–32) In wheat, the number of wheat MYB TFBSs is correlated with the abundance of glutenin in different wheat cultivars.¹⁵⁾ In the present study, we predicted the TFBSs in the upstream regions of wheat fl-cDNA sequences using PWMs; the profiles of 14 TFBS types were analyzed with 54 PWMs. The P-value of the two-tailed test was 1.000, indicating that there were no significant differences among the average numbers of TFBSs per fl-cDNA for the 14 TFBS types (Table 1). Knowledge of the positional distribution of TFBSs is important to investigate gene expression regulation by TFs. However, the positional distribution of TFBSs has not been widely explored. In wheat, some types of TFBSs exhibit different positional distribution preferences in these upstream regions (Fig. 4). Overall, the distribution of TFBSs in the upstream regions of wheat fl-cDNA sequences had significant difference, according to the results of one-sample t-test. Meanwhile, high frequencies of some types of TFBSs were found in specific regions within the upstream sequences.

In this study, it was found that the numbers of TFs and fl-cDNA with TFBSs were different expressed in different tissues (Tables 2, 3). The results were consistent with the results in a published wheat TF database.^11,12) However, some TFs were not detected in specific tissues. Most TFs were detected in all wheat tissues, except the TCP family in the inner pericarp and SE 20DPA. This may have been because few members of the TCP family are present in WheatTFDB.¹³⁾ This is in agreement with the published wheat transcription factor databases wDBTF and PlantTFDB; the TF numbers of TCP genes in WheatTFDB, wDBTF, and PlantTFDB are 10, 14, and 14, respectively.^11–13)

To elucidate the interaction mechanisms between TFs and their binding sites, it is important to analyze the structure of TFs and TFBSs. The MADS TF family has two subfamilies, the M-type and MIKC families. The key domain of MADS proteins is the serum response factor (SRF).¹²⁾ Some binding sites for the MADS family have been reported; for example, the CArG box with the sequence CC(AT)₆AG and the SRF-binding site with CC[A/T]₆GG and CTA[A/T]₄TAG.^33,34) These binding sites were detected in the 4572 wheat TFBSs for MADS TFs (Table 1, S2 File). MADS proteins, through interaction with their binding sites, play key roles in regulating various biological processes, especially in floral organ development.^33,34) Accordingly, the number of MADS TF genes expressed in pistils and stamens was obviously higher than in other tissues (Table 3).

The basic region/leucine zipper motif (bZIP) regulates biotic and abiotic responses by binding to various TFBSs.³⁵⁾ As shown in Table 3, bZIP proteins were extensively distributed in all tissues and had no distribution preferences. Plant bZIP proteins can recognize the ACGT core in DNA sequences, preferentially the A-box (TACGTA), C-box (GACGTC), and G-box (CACGTG),³⁵⁾ but can also bind to other sites, for example, ((C/T)AAC(T/G)G),³⁶⁾ GTGTC, and GTGTA.³⁷⁾ In this study, we identified 976 TFBSs of bZIP TFs (Table 1) in 15 wheat tissues; the core cis-element sequences of bZIP TFs mentioned above are listed in S2 File. Our results implied that the binding sites of wheat bZIP proteins are quite diverse.³⁷⁾

The AP2 (APETALA2)/ERF family contains three subfamilies, AP2, ethylene responsive factor (ERF), and RAV, which are classified based on the number of AP2/ERF domains.³⁸⁾ In wheat, we detected 1820 binding sites for ERF, 1725 sites for AP2/ERF, and 897 sites for RAV TFs (Table 1). The ERF family is one of the largest TF families in wheat (Table 3), and regulates a set of genes responsive to biotic and abiotic stresses such as drought, salt, and cold.³⁹⁾ The AP2 subfamily is involved in regulating organ-specific growth and development, such as flower development, ovule development, and the formation of seed size, while RAV subfamily genes induce expression in response to ethylene, brassinosteroids, and biotic and abiotic stresses.⁴⁰⁾ Wheat ERF subfamily members bind the conserved nucleotide sequences AGCCGCC⁴⁰⁾ and TACCGACAT³⁹⁾ of the GCC-box in the upstream regions of genes. The RAV binding sites CAACA and CACCTG motif can also be found in S2 File.⁴¹⁾

All WRKY proteins have a WRKY domain, which contains a C-terminal zinc finger and the conserved amino acid sequence WRKYGQK at its N-terminus.⁴²⁾ There were 135 WRKY members in WheatTFDB (Table 3). We found 300 TFBSs for WRKY TFs in 4873 fl-cDNA sequences (Table 1). The core DNA sequence motif recognized by WRKY TFs includes various W-box elements containing a TGAC core sequence.⁴²⁾ The cis-element sequences TTGAC(C/T/A/G) were confirmed in this study (S2 File). WRKY proteins and their TFBSs are associated with responses to different stresses and numerous processes required for diverse plant functions, such as development, embryogenesis, dormancy, and drought tolerance.⁴²⁾

bHLH TFs contain two different regions, the basic region and the HLH region. The basic region contains 15 amino acids with a high number of basic residues, and is located at the N-terminal end of the domain. The HLH region is located at the C-terminal end, and contains hydrophobic residues that form two amphipathic α-helices separated by a loop region of variable sequence and length.³⁰⁾ bHLH TFs have been reported to function in light-induced hormone signaling, wound and drought stress responses, shoot branching, fruit and flower development, root development, and other pathways.⁴³⁾ Table 3 shows that the numbers of bHLH TFs identified in roots, shoots, pistils, stamens and germinated seeds were greater than in leaves and grains. The distribution of bHLH TFs was in accordance with their published functions.⁴³⁾ The core cis-element bound by bHLH TFs is a consensus hexanucleotide sequence called the E-box (5ʹ-CANNTG-3ʹ).³⁰⁾ Here, 1951 TFBSs were predicted as bHLH type sites in wheat (Table 1). The cis-elements with the core sequences (C/T)GCGTG, (C/T)ACGTG, and CACG(A/G)C predicted in the upstream regions of wheat fl-cDNA sequences have been detected in various plants, such as Arabidopsis, rice, and poplar.^21,30,43)

TCP proteins can be distinguished by the specific TCP domain. Their binding sites vary; for example, GGNCCCAC, GTGGNCCC, and TGGGCC.^19,44) These binding sites are listed in S2 File. We identified 474 binding sites of TCP TFs in wheat (Table 1). TCP genes have been reported to affect the development of axillary structures, petals, and stamens.⁴⁵⁾ The TCP family had few members in wheat, and no TF sequences were detected in the outer pericarp or SE 20DPA (Table 3). These results suggest that the main functions of TCP proteins may not affect the development of outer pericarp or SE at 20 DPA.

Plant SBP proteins have been reported to be involved in a variety of important biological processes, including flower and fruit development, architecture formation, responses to copper and fungal toxins, and control of GA levels.⁴⁶⁾ The SBP domain can combine with the cis-element TNCGTACAA, for which GTAC is the essential core sequence.⁴⁷⁾ In wheat, 1365 TFBSs were detected in the non-coding regions of fl-cDNAs (Table 1), and the cis-element containing TNCGTACAA was confirmed in wheat (S2 File).

MYB proteins comprise one to four imperfect repeats, each of which forms a helix–turn–helix structure of ~50 residues.⁴⁸⁾ Here, 1114 cis-elements of the MYB superfamily were predicted in wheat (Table 1). Some published TFBS sequences were detected with the core sequences TAACTG, CAACTG, AACGG, and C/TAACNA/G.^8,49) Wheat MYB TFs have been shown to regulate developmental and physiological processes, especially responses to various biotic and abiotic stresses.⁴⁸⁾ Moreover, Guo et al. found that MYB TFs play an important role in the regulation of the glutenin gene expression during wheat endosperm development by directly binding to the promoter of TaGLU-1 and by recruiting GCN5 to modulate histone acetylation.¹⁵⁾

The plant B3 superfamily has two subfamilies, ARF and B3, and the B3 domain is the defining domain.¹²⁾ The B3 family is involved in flowering time control and in organ growth and polarity.⁴⁹⁾ In this study, we predicted 1951 TFBSs for B3 TFs (Table 1). Cis-elements of the B3 superfamily, with core sequences of CATGC, CATGCA, and TGTCTC, were verified in wheat (S2 File). The numbers of B3 members distributed in pistils, stamens, germinated seeds, and roots was greater than in the other tissues (Table 3). This seems to be consistent with the reported functions of B3 TFs.^49,50)

The BES1 family is a transcription factor family controlled by brassinosteroids (BR) that regulates multiple biological processes, including cell elongation, vasculature differentiation, photomorphogenesis, senescence, and stress responses.⁵¹⁾ BES1 transcription factors bind to sites with various core sequences to modulate gene expression; for example, CACGTG and CGTG(T/C)G.^51,52) We identified 459 TFBSs in 330 fl-cDNA sequences (Table 1). The core sequences of BES1 TFBSs mentioned above were detected in wheat (S2 File).

TFs of the HB superfamily are involved in hormonal pathways, providing a link between gene regulatory networks and signaling in the shoot apical meristem.⁵³⁾ The plant HB superfamily is divided into four subfamilies, HB-other, TALE, HB-PHD, and HD-ZIP, in which the homeobox domain is the defining domain.¹²⁾ However, no HB-PHD and HD-ZIP sequences were detected in wheat.¹³⁾ In this study, 1914 TFBSs for HB TFs were predicted in 1220 different fl-cDNA sequences (Table 1). Some reported cis-elements with the core sequences CAAT(A/T)ATTG and CAAT(G/C)ATTG were found in wheat (S2 File).⁵⁴⁾

Wheat Dof protein previously was isolated from endosperm and suggested to function as an activator of prolamin gene expression during seed development.²⁶⁾ The Dof domain of grass contains Cys2-Cys2 zinc-finger DNA binding proteins.^55,56) There were 32 Dof members in WheatTFDB (Table 3). The Dof TFs can bind with the cis-element 5′-TGTAAAG-3′ in upstream of gene,²⁶⁾ for which AAAG is the essential core sequence of TFBS.⁵⁷⁾ In wheat, 924 TFBSs were identified in the upstream regions of fl-cDNAs (Table 1), and the cis-element containing 5′-TGTAAAG-3′ was confirmed in wheat (S2 File).

Author contribution

Zhen-Yong Chen participated in the design of this work, carried out the experiment, and wrote the manuscript. Xiao-Jiang Guo, Zhong-Xu Chen, and Wei-Ying Chen participated in the design of this work and contributed to writing the manuscript. Ji-Rui Wang planned the study and designed this work, and wrote the manuscript.

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

This work was supported by the National Basic Research Program of China [grant number 2014CB147200]; the National Natural Science Foundation of China [grant number 31571654], [grant number 31230053], and [grant number 31171555]; The national key research and development plan of China [grant number 2016YFD0100100]; Research Fund for the Doctoral Program of Higher Education of China [grant number 20135103110008]; and Sichuan Science and Technology Support Project [grant number 2016NZ0103].

Supplemental materials

The supplemental material for this paper is available at http://dx.doi.org/10.1080/09168451.2014.946390.