Analyses of open-access multi-omics data sets reveal genetic and expression characteristics of maize ZmCCT family genes

Abstract Flowering in maize (Zea mays) is influenced by photoperiod. The CO, CO-like/COL and TOC1 (CCT) domain protein-encoding genes in maize, ZmCCTs, are particularly important for photoperiod sensitivity. However, little is known about CCT protein-encoding gene number across plant species or among maize inbred lines. Therefore, we analysed CCT protein-encoding gene number across plant species, and characterized ZmCCTs in different inbred lines, including structural variations (SVs), copy number variations (CNVs), expression under stresses, dark-dark (DD) and dark-light (DL) cycles, interaction network and associations with maize quantitative trait loci (QTLs) by referring to the latest v4 genome data of B73. Gene number varied greatly across plant species, more in polyploids than in diploids. The numbers of ZmCCTs identified were 58 in B73, 59 in W22, 48 in Mo17, and 57 in Huangzao4 for temperate maize inbred lines, and 68 in tropical maize inbred line SK. Some ZmCCTs underwent duplications and presented chromosome collinearity. Structural variations and CNVs were found but they had no germplasm specificity. Forty-two ZmCCTs responded to stresses. Expression of 37 ZmCCTs in embryonic leaves during seed germination of maize under DD and DL cycles was roughly divided into five patterns of uphill pattern, downhill-pattern, zigzag-pattern, └-pattern and ⅃-pattern, indicating some of them have a potential to perceive dark and/or dark-light transition. Thirty-three ZmCCTs were co-expressed with 218 other maize genes; and 24 ZmCCTs were associated with known QTLs. The data presented in this study will help inform further functions of ZmCCTs.


Introduction
Photoperiod affects flowering in plants including maize (Zea mays) (Jackson 2009;Song et al. 2015). Modern maize gradually evolved two major types of germplasm after domestication of a tropical teosinte from Mexico and Central America: (i) tropically/subtropically adapted and photoperiod-sensitive, and (ii) temperate-adapted and less photoperiod-sensitive (Doebley 2004;Hung et al. 2012). Temperate maize is an autonomous day-neutral plant, and teosinte is an obligate short-day plant that requires uninterrupted long nights to induce flowering (Minow et al. 2018). The differences in photoperiod sensitivity are observed in and between maize populations (Jiang et al. 1999;Huang et al. 2018). Photoperiod sensitivity hinders the improvement of temperate maize through the utilization of subtropical/tropical maize germplasm (Lewis and Goodman 2003;Yamasaki et al. 2007;Wang et al. 2008).
Growing evidence indicate that many key genes in photoperiodic flowering-time regulatory pathways are conserved across diverse plant species, but unique regulatory pathways are also present in some phylogenetic groups (Coles et al. 2010). Maize response to photoperiod is affected only by few of the flowering-time quantitative trait loci (QTLs) (Yang et al. 2013). The photoperiod-sensing pathway consists of the conserved upstream genes of conz1 as the closest homolog of CONSTANS (CO) in Arabidopsis, gigz1A, gigz1B, id1, and the differential downstream FLOWERING LOCUS T (FT)-like genes such as ZCN8, of which the upstream gene components are conserved in maize (Miller et al. 2008).
The proteins containing the CO, CO-like/COL and TOC1 (CCT) domain in the C terminus are transcription factors involved in photoperiod sensitivity of plants . The CCT domain-containing proteins were usually divided into three clades including COL, PSEUDO RESPONSE REGULATOR (PRR)-like and CCT MOTIF FAMILY (CMF)-like proteins on the basis of the domains in the N terminus . Many transcription factors related to photoperiod sensitivity in flowering plants contain a CCT domain in their predicted sequence .
The maize CCT domain-containing protein (ZmCCT) genes, ZmCCTs, appear to be particularly important for the photoperiod sensitivity of maize (Ducrocq et al. 2009;Yang et al. 2013) because their consistent and high expression results in the delay in flowering of maize under long day (Hung et al. 2012). It was reported that the ZmCCT family in maize inbred line B73 contains 53 ZmCCT genes based on the version two (v2) B73 genome, with four clades, including COL, PRR-like, CMF-like and TIF[F/Y]XG (TIFY)-like proteins . ZmCCTs are the upstream genes of ZCN8 in the photoperiod pathway, and they repress ZCN8 expression (Dong et al. 2012;Huang et al. 2018). A few ZmCCTs were functionally identified, of which ZmCCT9 and ZmCCT10, and ZmCOL3 contribute to flowering-time adaptation of maize from tropical to temperate regions (Yang et al 2013;Huang et al. 2018;Jin et al. 2018). In addition, ZmCCTs may also be involved in growth, development, and stress response (Ku et al. 2016;Li et al. 2016Li et al. , 2017Wang et al. 2017;Xu et al. 2017;Zhang et al. 2018). There is a report in which the CCT protein number of numerous genomes was previously informed, but little is yet known about CCT protein-encoding gene number across plant species or the genetic characteristics of ZmCCTs among maize inbred lines. We speculate that the CCT protein gene number varies with plant species, and that the expression patterns of ZmCCTs in maize are different under continuous dark-dark (DD) versus dark-light (DL) cycles. In this study, we analyse the number of CCT protein-encoding genes of plants and disclose characteristics of ZmCCTs including structural variations (SVs); copy number variations (CNVs); expression under stresses, DD and DL cycles; interaction network; and associations with maize QTLs by referring to the latest v4 genome data of B73.
Second, to prevent from losing potential CCT proteins in the above HMM-based search, Arabidopsis CCT proteins from the Arabidopsis Information Resource database (http:// www.arabidopsis.org/), and rice CCT proteins from the Rice Genome Annotation database (http://rice.plantbiology.msu. edu/) were used to search the above proteomes from maize and proteomes from other plants through the basic local alignment search tool for proteins (BLASTp) according to 1E < 10 −5 . After removing redundant sequences, CCT proteins were further confirmed by searching the conserved domain database with a threshold of 0.01 and a maximum hit of 500 (Marchler-Bauer et al. 2017) and the Pfam database under 1E < 0.05 (Finn et al. 2016).
Third, the candidate protein identification (ID) number of maize were used to search and obtain the corresponding gene ID number in the maizegdb database (https://maizegdb.org/) for maize ZmCCTs and in the phytozome database (https:// phytozome.jgi.doe.gov/pz/portal.html) for CCT proteinencoding genes of other plants. When there were multiple candidate proteins produced by different transcripts of the same gene, the protein with the longest amino acid sequence was selected as the representative to search the corresponding gene ID number.

Construction of phylogenetic trees of plant species
With the above information, the v0.66839 Toolkit for Biologists integrating various biological data-handling tools (TBtools) was used to generate the Newick (nwk) files with Latin names of species as input files according to the previous methods under the default parameters . The nwk files were employed to construct phylogenetic trees using the v6 Molecular Evolutionary Genetics Analysis software under the default parameters (Tamura et al. 2013).

Analysis of chromosomal localization, collinearity and duplication time of ZmCCTs
The multiple sequence alignment was performed with the target proteins against the v4 B73 proteome data set (https:// maizegdb.org/) by using the BLASTp to generate an m8-format file under 1E < 10 −5 . The chromosome collinearity of the genes was analysed based on both the m8-format file and the General Feature Format (GFF) file of the v4 B73 genome by using the toolkit for collinearity detection based on an adjusted MCScan . Duplicate gene pairs were extracted from the collinear genes with gene's ID number. The collinearity and chromosome localization were plotted using the Amazing Simple Circos tool .
The non-synonymous substitution (K a ) and synonymous substitution (K s ) rates of the duplicate gene pairs were calculated using the Parallel Alignment and back-Translation tool (Zhang et al. 2012) and K a K s _Calculator 2.0 (Wang et al. 2010). The gene duplication time was estimated according to the formula: K s /2λ, where λ = 6.5 × 10 −9 (Koch et al. 2000).

Analysis of characteristics and domains/motifs of ZmCCT-encoded proteins ZmCCTs, as well as introns and exons of ZmCCTs
The molecular weight and isoelectric point of each protein were analysed according to Artimo et al. (2012), and subcellular localization was predicted as Yu et al. (2006). The amino acid motifs were predicted (Bailey et al. 2015) and annotated through the InterProScan database (Mitchell et al. 2019). The protein domains were predicted according to Letunic and Bork (2017). The introns, exons and the domains/motifs in the genes were mapped based on the GFF files of the genes following the methods in Chen et al. (2018).

Analysis of SVs and CNVs of ZmCCTs
Structural variation analysis of the ZmCCTs was based on the maize inbred line B73 v4 genome SV loci data, which were generated by Yang et al. (2019) against the SV data set of the maize inbred line SK genome. In brief, with the SV loci of ZmCCTs in B73 as reference, the SVs of the ZmCCTs in the genomes of other 521 maize inbred lines (http://maizego.org/Resources. html) were scanned according to the methods described by Yang et al. (2019).
For CNV analysis, the maize genome re-sequencing data sets with a depth of > 30×, including 17 tropical/subtropical maize inbred lines and 24 temperate maize inbred lines (see Supporting Information- Table S1; https://www.ncbi.nlm.nih. gov/sra/), were first subjected to quality control (Bolger et al. 2014). Afterwards, the genome sequences were aligned to the v4 B73 genome by using the V0.7.15 Burrows-Wheeler Alignment software (Li and Durbin 2009), where the parameters used were under threads of 4, a minSeedLen of 32 and mark shorter split hits as secondary (Li and Durbin 2009). The sequences were converted into Binary Alignment/Map (BAM) format files using the Sequence Alignment/Map tools ). The BAMformat files were used to identify CNVs in other maize lines by using the Picard 1.129 software (http://sourceforge.net/projects/ picard/) and the cnvnator 0.3.2 software under 1E < 0.01 (Abyzov et al. 2011).

Expression analysis of ZmCCTs
The raw data sets used were the transcriptomes of embryonic leaves during seed germination of Z. mays cv. White Crystal under DL [dark (11 h)/light (13 h)] cycles (Liu et al. 2013) and DD , and the transcriptomes of abiotic (Makarevitch et al. 2015;Zenda et al. 2019; http://childslab.plantbiology.msu.edu) and biotic stress-treated maize lines (Swart et al. 2017). These raw data had three biological replicates.
The gene expression in the embryonic leaves was presented as Fragments Per Kilobase of transcript per Million-fragments mapped (Trapnell et al. 2010). For analysis of gene expression in the tissues under stresses, the reads of the ZmCCT sequences in the Sequence Read Archive (SRA) file (https://www.ncbi.nlm.nih. gov/sra/) were transformed into FASTQ format by using the SRA toolkit 2.9.2 tool, subjected to quality control by the FastQC 0.11.8 and Trimmomatic 0.38 tools, and then aligned to the bowtie 2-indexed v4 B73 genome (https://maizegdb.org/) by using the TopHat 2.1.1 software (Trapnell et al. 2012) to generate the BAMformat files. The reads of ZmCCTs were counted by using the featureCounts tool (Liao et al. 2014) and then used for differential expression analysis by the OmicShare tools (www.omicshare. com/tools) under the default parameters (Robinson et al. 2010). Differentially expressed genes were defined according to |a log2 fold change of read counts| ≥ 1 at P < 0.05.
The gene expression heat maps were plotted by using the Amazing Simple HeatMap tool in TBtools ).

Analysis of association of ZmCCTs with QTLs
The QTLs associated with ZmCCTs were identified through searching the B73 data set (https://bigd.big.ac.cn/gwas/) in the genome-wide association study (GWAS) Atlas (Tian et al. 2020) by ID number of ZmCCTs.

Analysis of gene co-expression and gene ontology enrichment
The genes co-expressed with ZmCCTs were achieved by searching the maize co-expression data set in the Arabidopsis thaliana trans-factor and cis-element prediction database-II (ATTED-II) (Obayashi et al. 2018; http://atted.jp/) with ID number of ZmCCTs. The co-expression networks of the genes were constructed by using the Cytoscape software (Shannon et al. 2003). Gene ontology (GO) enrichment was performed using the OmicShare tools under default parameters with a false discovery rate (Q) < 0.05 (www.omicshare.com/tools).

CCT protein-encoding genes and CCT proteins in plants
Based on the CCT HMM model, the number of CCT proteinencoding genes was found to differ among 68 plant species, with up to 114 genes in Gossypium hirsutum and only 17 genes in Amborella trichopoda (Fig. 1A). The ID number of CCT proteinencoding genes is listed in Supporting Information- Table S2.
The temperate maize lines of B73, Huangzao 4, W22 and Mo17 had 58, 37, 59 and 48 ZmCCTs [see Supporting Information- Table S3], respectively. The tropical maize line SK ) possessed 68 ZmCCTs. The ID number of the ZmCCTs is shown in Supporting Information- Table S3. The ZmCCTs had high hydrophilicity because of the negative overall average of hydropathicity, and they differed in amino acid sequence length, molecular weight, isoelectric point, and in subcellular locations [see Supporting Information- Table S4].

Chromosome distribution and duplication, collinearity and genomic structure of ZmCCTs in B73
Except for ZmCCT58, which was located on the unassembled genome scaffold, the remaining 57 ZmCCTs were mapped to 10 chromosomes of B73 (Fig. 1B). A total of 26 pairs of the genes with segment duplication (Table 1) showed collinear relationships (Fig. 1B). Duplication happened between 1.1 MYA (million years ago) for the ZmCCTs 28 and 37 pair and 38.947 MYA for both ZmCCT9 and ZmCCT20 (Table 1). ZmCCTs showed easily noticeable differences in genomic DNA structure in B73 genome ( Fig. 2A). The number of introns in ZmCCTs of B73 inbred line varied greatly, with no introns in ZmCCT55 [see Supporting Information- Table S4].

Phylogenetic relationships and conserved domains/ motifs of ZmCCTs
As previously reported   Table S5). In addition to the above domains, ZmCCTs had many conserved amino acid motifs ( Fig. 2B; see Supporting Information- Table S5). The motif 3 existed in 57 ZmCCTs but not in ZmCCT5 (Fig. 2B).

SVs of ZmCCTs in different maize inbred lines
With the SK genome as reference, ZmCCTs 8, 29, 30, 41 and 43 were found to have SVs in the v4 B73 genome, showing deletions and insertions of DNA segments (Table 2). These SVs were divided into three types: AA, for SVs of the same structure as B73 but different structure from SK; TT, for SVs of the same structure as SK but different structure from B73; and NN, for SVs unable to evaluate [see Supporting Information- Table S6].
The genomes of 521 maize inbred lines (http://maizego.org/ Resources.html) were scanned by reference to the SVs of ZmCCTs in the B73 genome according to the methods described by Yang et al. (2019). Consequently, one TT-type SV, SV7, of ZmCCT41 was found in 287 maize lines. Five TT-type SVs, SVs 9-13, of ZmCCT43 were found in 375 to 472 maize lines depending on the SV. One AA-type SV, SV2 of ZmCCT29 was found in 85 maize lines. One AA-type SV, SV4, of ZmCCT30 was found in 152 lines. One AA-type SV, SV8, of ZmCCT43 was found in 46 maize lines [see Supporting Information- Table S6].

Expression of ZmCCTs under biotic and abiotic stresses
Expression levels of 58 ZmCCTs from B73 were analysed in maize lines under abiotic and biotic stresses (Table 3). Forty-two (72.4 %) of 58 ZmCCTs responded to the stresses, and they together made a total of 84 differential expression events when compared to those in corresponding unstressed tissues of the same maize lines. The responses to drought, cold and heat accounted for 73 (86.9 %) of 84 differential expression events (Table 3), of which 53 (63.1 %) resulted from ZmCCTs in the COL clade, 17 (20.2 %) from       Table 3. Continued ZmCCTs in the PRR clade, 12 (14.3 %) from ZmCCTs in the CMF clade and 2 (2.4 %) from a TIFY clade (Table 3).

Expression rhythm of ZmCCTs
Based on the transcriptomes of embryonic leaves during seed germination under DL cycles and DD, there were 37 (63.8 %) of 58 ZmCCTs in inbred line B73 that showed differential expression (Fig. 4). From the beginning to the end of the treatments, ZmCCT expression could be roughly divided into five patterns: uphill pattern (i.e. the expression level tended to increase gradually); downhill-pattern (i.e. the expression level tended to decrease gradually); zigzag-pattern (i.e. the expression level had single or multiple obvious peaks and valleys); └-pattern (i.e. the expression level was high initially, and then it decreased suddenly and remained flat till the end); and ⅃-pattern (the expression level remained low until suddenly increased sharply at the last time point) (Fig. 4).

Association of ZmCCTs with QTLs
Analysis of the relevant data sets in the GWAS Atlas database revealed that 24 (41.4 %) of 58 ZmCCTs were associated with QTLs in the v4 B73 genome (Table 4) in inbred line B73. ZmCCTs 11,34,43,57 and 58 were linked to four QTLs. ZmCCT23 was associated with three QTLs. ZmCCTs 1, 2 and 31 were related to two QTLs. Other 17 ZmCCTs were associated with one QTL (Table 4), respectively.

Co-expression of ZmCCTs with other maize genes
Analysis of the transcriptome data sets of tissue samples of B73 under both control and stresses in the ATTED-II version 9.2 database showed that 33 (56.9 %) of 58 ZmCCTs in inbred line B73 were co-expressed with 218 other maize genes, including 16 transcription factors [see Supporting Information- Table S7]. ZmCCTs 6, 30, 37 and 54 were in isolated co-expression networks containing a few genes each (Fig. 5). In order to visualize the co-expression networks, the jre_8u_windows_64.exe or the jre_8u_windows_32.exe file was first downloaded and installed followed by installing the Cytoscape_3_3_0_windows_32bit or Cytoscape_3_3_0_windows_32bit file depending on the local Windows system. After installing this software, the reader can double-click and then watch the file 'Genes co-expressed with ZmCCTs in maize inbred line B73' through the button 'Zoom In' on the screen. These files are provided for readers The functional categorization by GO analysis indicated that the co-expressed genes were significantly associated with cellular component (Fig. 6A) and biological process (Fig. 6B) rather than molecular function (Fig. 6C).

Discussion
In this study, it was found that the number of CCT domaincontaining genes varies across plant species (Fig. 1A) and among maize inbred lines [see Supporting Information- Table S3]. The number of ZmCCTs identified in maize inbred line B73 based on the v4 B73 genome [see Supporting Information- Table S3] differed from the 53 ZmCCTs that were reported in B73 based on the v2 B73 genome ) likely because of the upgrade of gene annotation.
The CNV is one of the most common and most studied forms of SVs (Swanson-Wagner et al. 2010). They are most likely caused by non-allelic homologous recombination in plants (Gabur et al. 2019). CNVs are closely associated with the chromosome ploidy (Tang and Amon 2013). In this study, plants with higher ploidy had more CCT domain-containing genes compared to diploid species, such as allotetraploid G. hirsutum (Hu et al. 2019), allotetraploid Glycine max (Kyriakidou et al. 2018), allotetraploid Triticum aestivum (Kyriakidou et al. 2018), autotetraploid Solanum tuberosum (Kyriakidou et al. 2018) and tetraploid to octoploid Panicum virgatum (Grant 2017) (Fig. 1A). This phenomenon also occurred in different species of the same genus, for example, G. hirsutum had 1.9 times as many genes as diploid G. raimondii (Hu et al. 2019), P. virgatum had 1.86 times as many genes as diploid P. hallii (Grant 2017) and S. tuberosum had 1.96 times as many genes as diploid S. lycopersicum (Al Shaye et al. 2018) (Fig. 1B).
The ancestor of maize was an ancient tetraploid (White and Doebley 1998). However, over time, its genome has reverted to functional diploid (White and Doebley 1998). With diploidization, separation of chromosome segments will lead to the change of maize inbred line-specific CNVs (Eichten et al. 2011) which have been found in maize populations (Swanson-Wagner et al. 2010). This may partly explain why the genome sizes of B73, Mo17, W22 and SK are marginally different (Springer et al. 2018;Yang et al. 2019), but the ZmCCT number differed among maize inbred lines [see Supporting Information- Table S3].
The gene SVs are important clues to domestication and/ or breeding (Swanson-Wagner et al. 2010) and directly affect trait variations in maize (Gabur et al. 2019). More than 3000 SVs have been found in maize. The average length of an individual SV event is about 20 kb but ranges from 1 kb to over 1 Mb in length (Gabur et al. 2019). Expression of ZmCCTs delays flowering of maize under long day (Hung et al. 2012). Therefore, it is reasonably to believe that SVs of 5 ZmCCTs 8, 29, 30, 41 and 43 in populations of tropical/subtropical and temperate maize lines (Table 2; see Supporting Information-Table S6) were likely specific changes exerted by domestication and/or artificial selection objectives.
Several ZmCCTs are associated with flowering-time QTLs (Hung et al. 2012), drought and heat tolerance (Ku et al. 2016) and stalk rot resistance Wang et al. 2017). Quantitative trait loci associated with ZmCCTs are important nodes linking the photoperiod to stress tolerance responses under long day, and to how photoperiod affects the adaptability of plants to stresses (Ku et al. 2016) and vice versa (Hill and Li 2016). Association of 24 ZmCCTs with QTLs suggests that they can be considered important candidate genes of the QTLs, particularly those associated with flowering-related QTLs (Table 4). Responses to  Table 4. Association between ZmCCTs and QTLs in the v4 B73 genome. The information was extracted from the public data set in the GWAS Atlas (Tian et al. 2020 (Song et al. 2012). The COL proteins show a range of sequence identity with CO proteins (Valverde 2011). The genes with COL, CMF and PRR domains promote flowering under short day or delay flowering under long day in cereal crops . In maize, ZmCOL3 represses circadian clock but enhances ZmCCT expression and therefore delays flowering .
The TIFY family, previously known as ZIM, is characterized by a TIF[F/Y]XG sequence motif (Vanholme et al. 2007), which has been associated with abiotic and biotic stresses in plants such as maize (Zhang et al. 2015). Zf-B/B-box domain-containing proteins from the COL clade are a class of zinc-finger transcription factors with multiple functions (Gangappa and Botto 2014) that act as bridges between light and hormones in plants (Vaishak et al. 2019). MBD domain-containing proteins from the TIFY-like clade can enhance transcriptional repression of CCT domain-containing genes and are involved in DNA demethylation and abiotic stress responses. Also, mutation in AtMBD8 results in a late-flowering phenotype in the C24 ecotype of Arabidopsis (Parida et al. 2018). GATA domain-containing proteins from the TIFY-like clade have been implicated in light-dependent and nitrate-dependent control of transcription (Reyes et al. 2004). The ZmCCTs were divided into four distinct clades ( Fig. 2A) but the degrees of responses of the clades to the stresses differed in terms of the number of stress-responsive genes (Table 3). Taken all together, in addition to the response to photoperiod, the functions of four clades in ZmCCT family are diversified, and even ZmCCTs in the same clade appear to have also different functions. Motif 1, identified as a CCT domain motif of ZmCCTs, was only 29-amino acid residue long [see Supporting Information- Table S5], shorter than the CCT domain PF06203 which is defined as a 43-to 45-amino acid residue domain in the database (http://pfam.xfam.org/). This is because these amino acid residues identified were only the most conserved core motifs in the CCT domain. Unknown motif 3 was 21-amino acid residue long [see Supporting Information- Table S5], existed in 57 ZmCCTs not in ZmCCT, and was closely adjacent to CCT domain (Fig. 2B), suggesting that motif 3 is likely a motif that loses region of the CCT domain.

Conclusions
The CCT protein gene number varies greatly across plant species. ZmCCT number also varies with maize inbred lines. The   ZmCCT, ZmCCT gene. GO, gene ontology; Q, false discovery rate. data analysed suggest that ZmCCTs are involved in photoperiod response and stress adaptability. Based on their expression, it appears that some ZmCCTs are induced during dark-light or light-dark transitions. The data presented in this study are informative to further investigating the functions of the ZmCCTs.

Supporting Information
The following additional information is available in the online version of this article- Table S1. Re-sequencing data sets of maize inbred lines in the Sequence Read Archive (SRA) database (https://www.ncbi. nlm.nih.gov/sra/). Table S2. Accession no. of the CCT protein-encoding genes of other plant species other than maize in the database (https://phytozome.jgi.doe.gov/pz/portal.html). Table S3. Accession no. of the ZmCCTs in maize inbred lines of B73, Huangzao4, W22, Mo17 and SK. Table S4. ZmCCTs and ZmCCTs in maize inbred line B73. Table S5. Conserved domains/motifs of maize ZmCCTs. Table S6. Distribution of structural variations (SVs) of ZmCCTs in genomes of 521 maize inbred lines. Table S7. Genes co-expressed with ZmCCTs in maize inbred line B73.
Network files. In order to visualize the co-expression networks, the jre_8u_windows_64.exe or the jre_8u_ windows_32.exe file should be downloaded and installed followed by installing the Cytoscape_3_3_0_windows_32bit or Cytoscape_3_3_0_windows_32bit file depending on the local Windows system. After installing this software, the reader can double-click and then watch the file 'Genes co-expressed with ZmCCTs in maize inbred line B73' through the button 'Zoom In' on the screen. These files are provided for readers as a compressed supplemental file package.

Sources of Funding
This work was supported by the Guangxi Natural Science Fund Project (2017GXNSFEA198003), and the Science and Technology Major Project of Guangxi (Guike AA17204064), Department of Science and Technology of Guangxi, Guangxi, P. R. China.