The genome sequence of turnips was downloaded from www.bioinformatics.nl/brassica/turnip. To find all TCP genes in turnips, NCBI BLASTn searches against a local database built using nucleic acid sequences were performed using sequences from all 24 known TCPs from Arabidopsis. Subsequently, the Pfam database was used to determine if each candidate TCP sequence was a member of the TCP gene family. To exclude overlapping genes, all candidate TCP genes were aligned using DNAMAN 4.0 (Lynnon Biosoft) and checked manually. All nonoverlapping TCP genes were used for further analysis.

Conserved motifs of BrrTCP proteins were analyzed using MEME (http://meme-suite.org/tools/meme) with the following parameters: (1) the optimum motif width was set from 6 to 200, and (2) the maximum number of motifs was set to identify 20 motifs.

BrrTCP genes were mapped on chromosomes by confirming their detailed chromosomal positions supplied by the Turnip Genome Database. To illustrate the structure of introns and exons of BrrTCP genes, full-length genome and coding sequences of BrrTCP genes were subjected to online GSDS analysis (http://gsds.cbi.pku.edu.cn/). To determine their physical location, the starting positions of all BrrTCP genes on each chromosome were confirmed based on a local database of the complete sequence of the turnip genome through BLASTn searching. The segmental and tandem duplication regions were obtained from MCscanX. For synteny analysis, synteny block of the turnip gene was visualized using Circos (http://circos.ca/). Synonymous (Ks) and nonsynonymous (Ka) substitution rates were estimated by the codeml program of PAML4 (Yang, 2007). The divergence time (T) of turnip gene pairs was calculated using the formula T = Ks/2λ, where λ represents the divergence rate of 1.5 × 10⁻⁸ for Arabidopsis (Gaut et al., 1996).

Brassica rapa ssp. rapa “KTRG-B48a” from Xianggelila City, Yunnan Province, China, was used. For root collection, seedlings were grown on a Whatman filter paper and watered with 1/2 MS medium for 2 weeks. For other tissues, plants were grown in the greenhouse (22°C) under long-day conditions (16 h light/day).

Total RNA was extracted using Eastep™ Universal RNA Extraction Kit (Promega, Shanghai, China) from roots of 1-week-old plants, leaves and stems of 8-week-old plants, and floral buds of 10-week-old plants. RNA quality and concentration were assessed using electrophoresis and ND-1000 Spectrophotometer (NanoDrop Technologies, Delaware, USA). Two micrograms of total RNA were reverse transcribed using GoScript™ Reverse Transcription System (Promega). Quantitative RT-PCR (qRT-PCR) was performed with ABI7500 Real-Time PCR System using TransStart® Top Green qPCR SuperMix (TransGen, Beijing, China). BrrACT2 was used as reference gene. The primers are listed in Table S1. Triplet biological replicates were analyzed.

All BrrTCP ORFs were amplified from seedling cDNA of KTRG-B48a into pENTR vector used primers (Table S2). The entry vectors were subcloned into the pGADT7 prey vector (pDEST-GSDT7) and pGBKT7 bait vector (pDEST-GBKT7) using gateway method according to Bai et al. (2016). The prey vector was transformed into yeast strain Y187, and all bait vectors were transformed into yeast strain Y2H gold and selected on SD plates lacking Leu and Trp. After selection, the yeasts harboring prey and bait plasmids were spotted onto control medium (SD plates lacking Leu and Trp) and test medium (SD plates lacking Leu, Trp, and His). Yeast growth was observed daily in a growth chamber at 28°C for 2–5 days.

The entry vectors of BrrTCP were subcloned into pRI101-GFP using gateway method. 35S:GFP-TCPs were introduced into Agrobacterium GV3101. Positive transformants were selected on LB agar medium (50 mg/L kanamycin, 50 mg/L gentamicin, and 50 mg/L rifampicin). After 3 days, a single colony was inoculated into 2 mL LB liquid medium. Target inserts were confirmed through PCR. Nicotiana benthamiana plants were grown at 22°C under 16 h light/8 h dark conditions. One-month-old plants were used for infiltration. Two days before infiltration, 2 mL cultures of the Agrobacterium strains were inoculated from single colonies on plates and grown for 24 h at 28°C. The working cultures were inoculated from the starter culture at a 1:100 ratio. Cells were harvested through centrifugation at 3,000 g, 22°C, for 5 min. Cell pellets were resuspended in infiltration medium (10 mM MgCl₂; 10 mM MES, pH 5.7; and 150 μM acetosyringone) with OD600 adjusted to 1. Resuspended cell cultures were kept at room temperature for 3 h before infiltration. Leaf infiltration was conducted by depressing a 1 mL disposable syringe to the surface of fully expanded leaves and slowly depressing the plunger. Leaf discs were collected 72 h after infiltration for measurements using green fluorescent protein (GFP) fluorescence assay. Fluorescence images were obtained under an Olympus FV1000 laser confocal microscope (Olympus, Tokyo, Japan), with excitation at 488 nm and emission at 505–530 nm.

The 35S:GFP-BrrTCP2 construct was introduced into Agrobacterium GV3101, which was used to transform wild-type Arabidopsis via floral dip (Clough and Bent, 1998). T0 seeds were screened on MS medium containing 30 mg L⁻¹ kanamycin. All transgenic lines overexpressing 35S:GFP-BrrTCP2 were verified by Western blot using GFP antibody. OXBrrTCP2 x tcp2/4/10 plants were obtained by crossing OXBrrTCP2 with tcp2/4/10.

The sixth rosette leaves were selected for scanning electron microscopic analysis as previously described by Sun et al. (2013). The samples were observed and photographed under a scanning electron microscope (Gemini Sigma 300/VP SEM, Carl Zeiss, Germany) at an accelerating voltage of 10 kV.

The release of the complete turnip genome sequences allowed the genome-wide identification of turnip genes (Lin et al., 2014). In the present study, BLAST was carried out to search BrrTCP in the turnip genome. The obtained sequences were further verified through hidden Markov model. Finally, a total of 39 nonredundant BrrTCPs were identified from turnip genome. The BrrTCP genes were named following the nomenclature of Arabidopsis thaliana depending on protein sequence similarities (Figure 1). Sequence analysis revealed that AtTCP11 and 16 had no orthologs in turnip. AtTCP1, 4, 7, 9, 13, 15, 17, 18, 20, 21, and 24 had more than one ortholog in the turnip genome. Given the lack of standard annotations assigned to the 39 TCPs in turnips, the orthologs were designated as shown in Table 1 based on protein sequence similarities to their orthologs in Arabidopsis.

Information of the BrrTCP gene family members is shown in Table 1. The ORF length of BrrTCP genes varied from 603 to 1,401 bp, encoding polypeptides of 200–466 amino acids, with a predicted molecular mass of 22.05–50.25 kDa. The theoretical pI ranged from 5.48 to 10.20. Genomic localization of each BrrTCP in turnips is shown in Figure 2. These genes were distributed in all 10 turnip chromosomes. The maximum number of BrrTCP genes per chromosome was found for chromosome A02 with 9 TCP genes. Eight genes were located at chromosome 3, and five and four genes were located at chromosomes 7 and 5, respectively. Chromosomes 4 and 10 had the minimum BrrTCP genes, only one each. Three genes each were located on chromosomes 1, 6, and 9. Chromosome 8 contained two BrrTCP genes. A total of seven pairs of putative BrrTCP paralog proteins were produced by segmental duplication, which were distributed in different chromosomes. These results indicated the large-scale segmental duplication events involved in the expansion of BrrTCP gene family in turnips.

The divergence time (T) of seven pairs of BrrTCP paralog proteins was estimated by measuring the synonymous (Ks) and nonsynonymous (Ka) mutation rates (Table 2). The estimated divergence time (T) for the BrrTCP paralogs was from 10.3067 to 27.8600 million years ago (MYA), with an average duplication time of approximately 18.7862 MYA. Schranz et al. (2006) estimated the time of very early radiation of Brassicaceae at 34 MYA. Comparative physical mapping studies have confirmed genome triplication in a common ancestor of B. oleracea (O'Neill and Bancroft, 2000) and B. rapa (Park et al., 2005) since its divergence from the A. thaliana lineage at least 13–17 MYA (Yang et al., 1999; Town et al., 2006; Beilstein et al., 2010). The divergence time of three pairs of BrrTCP paralogs (BrrTCP9/9a, BrrTCP22/23, and BrrTCP24/24a) occurred precedent to the period of the origin of the B. rapa. The ω values for the BrrTCP paralogs were less than one. However, two pairs of BrrTCPs (BrrTCP1a/1b, ω = 0.3794; BrrTCP24/24a, ω = 0.3646) achieved relatively large ω values, which suggested that they may have evolved rapidly from those of the last common ancestor.

To better understand the diversification of the TCP genes in turnips, the exon/intron organization and conserved motifs of BrrTCPs were analyzed. A new neighbor-joining phylogenetic tree was constructed using the protein sequences of BrrTCPs (Figure 3A). The genome sequences and corresponding coding sequences of TCP genes in turnip analysis revealed that most BrrTCP genes have similar gene structures in the same group (Figure 3B). A total of 32 members of BrrTCP gene family have one exon (82%), 4 genes have two exons (10%), and 3 have four exons (8%). In Arabidopsis, the number of exons ranged from one to four, and 82% of the genes contained only one exon. The TCP genes in turnips exhibited similar gene structure. The MEME online tool was used to predict the conversed motif composition of BrrTCPs (Figure 3C). The number of motifs varied from 2 to 11. The motifs were evaluated using InterProScan for annotation. The results revealed that motifs 1, 2, and 3 were identified as TCP domain. In addition, some motifs only presented at specific subclades, such as motifs 6, 7, 10, and 19 in BrrTCP1, 1a, and 1b; motif 9 in BrrTCP15, 15a, and 15b; and motif 16 in BrrTCP4, 4a, and 4b, suggesting that they may have subclade-specific functions.

In Arabidopsis, miR319 controls jasmonate biosynthesis and senescence by cleaving TCP transcription factors (Schommer et al., 2008) and petal development (Nag et al., 2009). The AtmiR319-targeted TCP genes, namely, AtTCP2, AtTCP3, AtTCP4, AtTCP10, and AtTCP24, all belong to the CIN clade. Similarly, four TCP genes in turnips contain miR319 binding sites (Figure 4), and all of them were CIN family members.

The known members of the TCP gene family function as transcription factors. The GFP gene was fused with BrrTCPs as a reporter. The GFP signals of BrrTCP-GFPs were detected in the nucleus using laser confocal microscopy (Figure 5), which suggested that BrrTCPs functioned as transcription factors.

The expression pattern of a gene is always relative to its function (Xu et al., 2015). To probe possible functions of BrrTCP genes in turnips, qRT-PCR was performed to examine the expressions of BrrTCP genes in different organs of turnips (Figure 6). Interestingly, expression levels of most CIN-type BrrTCP genes were high in leaves and weak in the roots and stems. BrrTCP18 and BrrTCP18a, the members of CYC/TB1 subclade, are highly expressed in flower buds and weakly expressed in roots. In contrast, class I clade showed more widespread but less tissue-specific expression patterns; for example, BrrTCP7 and BrrTCP9 are highly expressed in roots, stems, leaves, and flower buds; BrrTCP8, 14, 18a, 20, 21b, 22, and 23 are highly expressed in the stems, leaves, and flower buds; and BrrTCP7b, 13a, 15, and 21 are highly expressed in leaves and flower buds. These results indicated that every clade possessed a characteristic expression profile.

TCP proteins tend to form homodimers or heterodimers with other TCP proteins, and dimerization may be required for their DNA-binding activity and hence for their biological activity. Yeast two-hybrid screening was carried out to investigate the interactions among the BrrTCP proteins, as shown in Figure 7, where the proteins are arranged according to their subclades except their autoactivation activity. A total of 8 out of 39 AD-fusion proteins tested had autoactivation activity in yeast. Meanwhile, out of the 39 BD-fusion proteins tested, 13 had autoactivation activity in yeast. Among them, five BrrTCP proteins had autoactivation activity in AD- and BD-fusion proteins. Although we selected 219 interactions, this number may not be accurate due to autoactivation. A total of 90 (45 pairs) proteins interacted in AD- and BD-orientations, including 9 homodimer formations. The class I BrrTCP transcription factors formed 91 homo- or heterodimers within class I members. Meanwhile, class II BrrTCP transcription factors formed 42 homo- or heterodimers within class II. The members of class I proteins more preferentially formed dimerization properties in the same class than class II.

BrrTCP2 is a member of the CIN subclade of TCP in turnips. Given the unavailability of tcp-related mutant in turnips, we constructed transgenic Arabidopsis plants overexpressing BrrTCP2. As shown in Figure 8, the sixth rosette leaf of tcp2/4/10 triple mutants showed the most enhanced leaf size, with some signs of crinkling (Figures 8B,F). Meanwhile, the OXBrrTCP2 plants showed the most diminished leaf size (Figures 8C,G). OXBrrTCP2 line was crossed with tcp2/4/10, and the phenotype of the homozygote F2 plants restored the leaf phenotype of tcp2/4/10 to the phenotype of wild type (Figures 8D,H). Western blot analysis showed that OXBrrTCP2 and F2 plants had high expression levels, whereas no signal was detected in WT and tcp2/4/10 plants (Figure 8I). Arabidopsis leaf size is normally regulated by the cell size and number. To assess the cell size and number, we selected a site midway along the length of the lamina and between the margin and the midvein of the expanded sixth rosette leaf for analysis using scanning electron microscopy. The adaxial epidermal cell size of tcp2/4/10 was larger than WT (Figures 9A,B), and fewer cells were noted per unit area (Figure 9E). Meanwhile, the adaxial epidermal cell size of OXBrrTCP2 plants was smaller than WT (Figures 9A,C), and more cells were noted per unit area (Figure 9E). The F2 plant had similar cell size and cell number per unit area with WT. The abaxial epidermal cell size and cell number were similar to the adaxial epidermis. The tcp2/4/10 had larger cell size and fewer cell number per unit area than WT (Figures 9F,G,J), whereas OXBrrTCP2 plants had smaller cell size and more cell number per unit area than WT (Figures 9F,H,J). Overexpression of BrrTCP2 in tcp2/4/10 also restored the abaxial epidermal phenotype of tcp2/4/10 to WT (Figures 9D,I). BrrTCP2 may have a function in cell division and/or differentiation.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.