Despite the described role of some TCP factors in plant development [reviewed in Uberti Manassero et al. (2013)] little is known about the function of a subset of class I TCP factors, AtTCP7, AtTCP8, AtTCP22, and AtTCP23. We determined the phylogenetic relationships of class I TCP factors using sequences from Arabidopsis, tomato, and rice (Figure 1 and Supplemental Figure 1). This analysis shows that AtTCP7, AtTCP8, AtTCP14, AtTCP15, AtTCP21, AtTCP22, and AtTCP23 form a distinct clade when compared to the rest of the factors in this subfamily such as AtTCP20, AtTCP9, AtTCP19, AtTCP11, AtTCP6, and AtTCP16. AtTCP7 and AtTCP21 are close homologs and show a 62% of sequence similarity at the protein level. AtTCP14 and AtTCP15 cluster together (Figure 1). Due to high sequence similarity, AtTCP8, AtTCP22, and AtTCP23 do not show clear phylogenetic relationships in this subgroup of class I TCP factors.

We analyzed the expression pattern of class I TCP genes by RT-PCR in several tissues of Arabidopsis wild type plants (Figure 2A). AtTCP7, AtTCP14, AtTCP21, and AtTCP23 are expressed in all the tissues analyzed. AtTCP15 and AtTCP8 are expressed at low levels in siliques and mature leaves, while AtTCP22 is not detected in inflorescences and mature leaves (Figure 2A). Expression data from Genevestigator (Hruz et al., 2008), the Arabidopsis eFP browser at BAR (Winter et al., 2007), AtGenExpress (Schmid et al., 2005), and GeneCAT (Mutwil et al., 2008) supports our RT-PCR analysis for AtTCP8, AtTCP14, AtTCP15, AtTCP21, and AtTCP23, while for AtTCP7 and AtTCP22 there is no expression data yet available in these resources. Is interesting to note that according to this data AtTCP8 shows a high expression level in dry seeds.

To determine the spatial expression of class I TCP factors, we fused the complete 5′ gene promoter regions of AtTCP7, AtTCP8, AtTCP22, and AtTCP23 to the coding sequence of the uidA marker gene and analyzed the GUS expression pattern in 12 days-old plants of T3 lines (Figures 2B–I). We found that the localization of GUS is restricted to young emerging leaves in plants ProAtTCP7:GUS (Figure 2B), ProAtTCP8:GUS (Figure 2C) and ProAtTCP22:GUS (Figure 2D). In plants ProAtTCP23:GUS the GUS signal is detected at the base of growing leaves (Figure 2E). Analysis of cross-sections showed that the GUS signal is located at the surface of the lateral organs and mainly in the adaxial side of the young leaves for plants ProAtTCP7:GUS (Figure 2F), ProAtTCP8:GUS (Figure 2G), and ProAtTCP22:GUS (Figure 2H). In plants ProAtTCP23:GUS the signal is found at the base of the growing leaves (Figure 2I). We did not detect signal at later stages of development in structures such as stem internodes, inflorescences, or flowers (data not shown). Thus, the class I TCP factors we studied have a similar expression pattern hinting at a similar function for these genes.

TCP genes are reported to code for a plant-specific family of transcription factors (Cubas et al., 1999). We analyzed the subcellular location of AtTCP7, AtTCP15, and AtTCP22 through fusions of the coding sequences of the corresponding genes to the marker GFP. We found that the fusion proteins are localized in the nucleus in case of AtTCP7 (Figures 3A–C), AtTCP15 (Figures 3D–F), and AtTCP22 (Figures 3G–I), suggesting a role of these TCP factors as transcription factors.

To determine the function of selected class I TCP genes, we characterized T-DNA insertion lines (Figure 3J). AtTCP7, AtTCP8, AtTCP15, AtTCP22, and AtTCP23 are intron-less genes. For AtTCP8, AtTCP15, and AtTCP23 we used the null alleles tcp8-1, tcp15-1, and tcp23-1, respectively (Figure 3K). For AtTCP22, the allele tcp22-1 showed a reduction in the level of the gene product (Figure 3K). In case of AtTCP7 we used four available insertion lines, however none of them was located in the coding sequence. RT-PCR expression analysis of the alleles tcp7-1 and tcp7-2 indicated that insertions outside the coding region do not lead to a reduction of transcript levels of AtTCP7 compared with the wild type (Figure 3K).

Phenotypically the single mutants are overall similar to the wild type (Figures 4A,B). We determined however that tcp15-1 has reduced number of rosette leaves (Figure 4C). A quantitative analysis of leaf traits revealed that tcp15-1 has shorter petiole length than wild type (Figure 4D). Kieffer et al. (2011) found a slightly shorter fruit pedicel length and shorter inflorescence height in this mutant. We also found that tcp23-1 has increased blade length (Figure 4E), blade width (Figure 4F), blade perimeter (Figure 4G) and blade area (Figure 4H) compared to wild type. However we found that tcp8-1 and tcp21-1 plants are indistinguishable from wild type. For AtTCP7 we made RNAi lines (tcp7-RNAi) but the transgenic plants were similar to the wild type. As the analyzed single mutants only exhibited subtle phenotypic changes compared to the wild type we made multiple mutant combinations in the class I TCP genes reasoning that loss of function at multiple class I TCP genes would reveal more evident changes in plant development. We analyzed the quadruple mutant tcp8-1 tcp15-1 tcp22-1 tcp23-1 and the pentuple mutant tcp8-1 tcp15-1 tcp22-1 tcp23-1 tcp21-1. The overt phenotype of both the quadruple and pentuple mutant was similar to the wild type (Figures 4A,B), although we found that these multiple mutants have fewer rosette leaves (Figure 4C) and bigger leaves than wild type (Figures 4E,G,H).

Mutations in AS1 produce leaves with altered leaf shape, with the leaf lamina highly lobed (Byrne et al., 2000). CIN-like members of class II TCP genes also regulate leaf shape (Nath et al., 2003; Palatnik et al., 2003). Indeed, CIN-like TCP genes regulate the expression of AS1 (Koyama et al., 2007). We wanted to determine if class I TCP mutations interact with the leaf shape phenotype of as1. We used the double mutant tcp8 tcp22 to generate the triple mutant as1 tcp8 tcp22. We found that the number of rosette leaves produced in the triple mutant plants is reduced compared with the number of rosette leaves in as1 while the leaf shape resemble the leaves of as1 (Figure 4I), reinforcing the finding of class I TCP genes as regulators of leaf initiation rate.

In summary, this analysis indicates that single mutations in class I TCP genes have mild phenotypes. The quadruple and pentuple mutants have increased defects relating to number of leaves and leaf size while still maintaining normal morphology, indicating that other members in this subgroup of genes may act redundantly with the five genes analyzed.

It has been reported that some Arabidopsis class I TCP genes such as AtTCP20 (Li et al., 2005), AtTCP15 (Kieffer et al., 2011; Li et al., 2012a), and AtTCP14 (Kieffer et al., 2011) modulate the expression of cell-cycle genes. Furthermore, it has been shown that class I TCP genes can regulate the expression of genes involved in SAM and leaf development, acting in part redundantly with the class II TCP genes controlling the same set of genes (Uberti-Manassero et al., 2012). We wanted to determine if the expression of genes involved in cell-cycle control and SAM maintenance are affected in the pentuple tcp mutant tcp8-1 tcp15-1 tcp22-1 tcp23-1 tcp21-1. We found that genes involved in the control of SAM maintenance and leaf form such as the KNOX1 genes SHOOT-MERISTEMLESS (STM) and BP and to a lesser extent AS1 are overexpressed in the pentuple tcp mutant (Figure 5A). We also determined that genes regulating the cell-cycle such as CYCA1;1 and CYCA2;3 are up-regulated in the pentuple tcp mutant (Figure 5A). It is known that ectopic expression of KNOX1 genes produces leaf lobing (Chuck et al., 1996). Although in the tcp pentuple mutant we found no dramatic leaf lobing alteration, the mutant plants have bigger leaf blades, and the up-regulation of cell-cycle genes can partially explain this increased leaf size.

The up-regulation of KNOX1 genes in the tcp pentuple mutant indicates that class I TCP proteins could act directly or indirectly as regulators of KNOX1 genes. We tested if class I TCP genes directly bind the STM promoter using electrophoretic mobility shift assay (EMSA). We used a probe of 76 bp that belongs to the K-box, a highly conserved region in the STM 5′-flanking sequence involved in the coordination of leaf development (Uchida et al., 2007). We found that AtTCP7 is capable to bind the K-box in vitro (Figure 5B). Mechanistically TCP factors can act as dimers and there is evidence that the DNA binding capacity is increased when they form heterodimers (Kosugi and Ohashi, 2002; Viola et al., 2011). We found that heterodimers AtTCP7-AtTCP23, AtTCP7-AtTCP22, and AtTCP22-AtTCP23 are also able to bind the K-box (Figure 5B).

This interaction between TCP factors can account for the high degree of genetic redundancy found in this family. We wanted to analyze if members of class I TCP genes analyzed in this study are able of protein–protein interaction using a yeast two-hybrid assay (Figure 5C and Table 1). We found that AtTCP7 interacts with AtTCP23, AtTCP21, and to a lesser extent with AtTCP8. We also found interaction between AtTCP8 and AtTCP21 (Figure 5C and Table 1). Class II TCP factors AtTCP2 and AtTCP3 do not show interaction with any of the class I TCP factors tested. Taken together these results indicate that class I TCP factors can have a similar gene function by virtue of their ability to form higher order complexes to regulate similar downstream targets.

The lack of a severe phenotype in the pentuple mutant can be ascribed to genetic redundancy, particularly prominent in the TCP gene family (Schommer et al., 2008; Koyama et al., 2010). One way to circumvent genetic redundancy and to analyze the function of transcription factors is to transform them into dominant negative forms by fusing the coding sequences of the genes to the repressor domain SRDX, producing loss-of-function phenotypes even in the presence of redundant genes (Hiratsu et al., 2003). To further analyze the function of class I TCP genes, we generated transgenic lines with fusions of the coding sequences of AtTCP7 and AtTCP23 to the SRDX repressor domain and compared their phenotypes with the wild type (Figure 6). At the seedling stage lateral organs (cotyledons and first true leaves) of TCP7-SRDX and TCP23-SRDX plants show an upward growth (Figures 6A–C). At the bolting stage compared to the wild type (Figure 6A), TCP7-SRDX and TCP23-SRDX rosette leaves are hyponastically curved and smaller (Figures 6E,F). The basis of the differences in leaf size and form can be due to an altered cell growth pattern. We therefore analyzed the leaf epidermis and found that the leaves of the transgenic lines have groups of small cells on both adaxial (Figures 6G–I) and abaxial (Figures 6J–L) surfaces. Flowers are also affected in the transgenic plants (Figures 6M–R). Compared with the wild type (Figures 6M,N) the flowers of TCP7-SRDX (Figures 6O,P) and TCP23-SRDX (Figures 6Q,R) plants have shorter pedicels and are closed, with irregularly shaped larger sepals, shorter petals and less trichomes. Despite these alterations, the flowers have a normal number of sepals, petals, stamens, and carpels (data not shown). Gene expression analysis through qPCR on SAM gene markers indicated that STM is up-regulated in TCP7-SRDX (Figure 6S) and TCP23-SRDX (Figure 6T) transgenic plants, while the expression of the cell-cycle marker CYCA1;1 is not changed and the expression of CYCA2;3 is down-regulated (Figures 6S,T). In summary, dominant negative forms of AtTCP7 and AtTCP23 produce similar phenotypes with alterations in lateral organ growth, possibly based in differences in cell proliferation. Therefore is likely that AtTCP7 and AtTCP23 both function in regulating cell proliferation.

In conclusion, our results expand our knowledge of a group of members of the TCP gene family which has important roles in plant development. The lack of a severe phenotype in multiple mutants, the phenotype of the leaf epidermis and leaf shape found in dominant negative forms in these genes, their expression in young leaves, the demonstrated protein-protein interaction between selected members, the altered KNOX1 gene expression in the knockout or dominant negative mutations, indicates that they may function in the control of leaf development by regulating similar gene networks.

Alleles tcp8-1 (SAIL_656_F11), tcp7-2 (SAIL_180_B09), and tcp23-1 (SAIL_443_F02) were obtained from the Syngenta Arabidopsis Insertion Library (SAIL) (Sessions et al., 2002). Alleles tcp15-1 (SALK_011491C) and tcp22-1 (SALK_027490) were obtained from the Salk Institute Genomic Analysis Laboratory (SIGnAL) (Alonso et al., 2003) through the Arabidopsis Biological Resource Center (ABRC). The mutant allele for AtTCP21 was che-1 (Pruneda-Paz et al., 2009). The allele tcp7-4 (WISCDSLOX240D10) was obtained from the Wisconsin DsLox collection (Woody et al., 2007). tcp7-1 (GK-511D02) belongs to the GABI-Kat collection (Li et al., 2007). The gene allele tcp7-3 was obtained from the INRA-Versailles T-DNA collection (Beschtold et al., 1993). tcp7-3 is in Wassilewskija (Ws) background, all the rest are in Columbia-0 (Col-0) ecotype background. as1-1 was from Uchida et al. (2007).

Plants were grown under long-day conditions (16 h light/8 h dark) at 22°C in a soil mixture (Sunshine Mix1). For the phenotypic analysis plants were grown in chambers at 23°C at 50% humidity and 60–70 μmoles light and long day conditions.

For in vitro culture, Arabidopsis seeds were surface-sterilized in 70% bleach, before transfer on agar plates containing half-strength salts and vitamins, 1.5% sucrose, and 0.8% agar, with or without kanamycin (50 μg/ml). Plants transferred to soil were grown under the same growth chamber conditions as mentioned above.

A region of 69 aminoacids spanning the highly conserved TCP box was used to build the phylogenetic trees. We used the sequences of the 24 TCP genes in Arabidopsis and the 25 genes in rice respectively. We followed the notation as in Martín-Trillo and Cubas (2010). For tomato we isolated the sequences from the ITAG 2.3 release of the sequenced tomato genome (http://solgenomics.net/organism/Solanum_lycopersicum/genome; Supplemental Table 1). The sequence alignments were made using Muscle (http://www.ebi.ac.uk/Tools/msa/muscle/). The tree was made using MEGA5 software (Tamura et al., 2011; http://www.megasoftware.net/) using the maximum likelihood method with 1000 bootstrap replicates.

To generate RNAi lines for AtTCP7, a fragment of 195 bp of the coding sequence was amplified using primers M2LRNA2f (5′-TATCCGTGGAGCCACCAATT CTACTTCT-3′) and M2LRNA2r (5′-GCTGGTGTTGTCGTAAAGGTCTCATCTT-3′) and cloned into pCR8/GW/TOPO TA entry vector (Invitrogen) to generate pCR8-TCP7RNAi. As destination vector we used pK7GWIWG2(II) (Karimi et al., 2002).

For the GFP fusions with AtTCP7, AtTCP15, and AtTCP22, we cloned the coding sequences of these genes into the entry vector pCR8/GW/TOPO TA and LR cloned into the plasmid pK2GW7 (Karimi et al., 2002) generating TCP7-OEX, TCP15-OEX, and TCP22-OEX, respectively (for primers used see Supplemental Table 4).

For the SRDX constructs, the coding sequence of AtTCP7 was cloned into the entry vector pCR8/GW/TOPO TA using forward primer M2Lf (5′-ATGTCTATTAACAACAACAACAACAACA-3′) and the reverse primer TCPSRDXrev (5′-TTACGCAAAGCCGAGGCGAAGTTCGAGATCAAGATCGAGACGTGGATCTTCCTCTCTTCGATCCGA-3′) without the stop codon of AtTCP7 and with the EAR peptide sequence LDLDLELRLGFA plus the terminator codon TAA (Hiratsu et al., 2003). The construct was LR cloned into the plasmid pK2GW7 generating TCP7-SRDX. Similarly, for AtTCP23 we used primers P23f (5′-ATGGAGTCCCACAACAACAACCAGAGCA-3′) and TCP23SRDXrev (5′-TTACGCAAAGCCGAGGCGAAGTTCGAGATCAAGATCGAGAGGAGAACCATCTATAGTAGGATTTTG-3′). The PCR fragment was cloned into pCR8/GW/TOPO TA and LR cloned in pK2GW7 to generate TCP23-SRDX.

To generate ProAtTCP7:GUS, ProAtTCP8:GUS, ProAtTCP2:GUS, and ProAtTCP23:GUS constructs, respectively, 1990, 1588, 2180, and 2494 bp of the complete 5′ intergenic region preceding the ORFs was amplified from Col-0 genomic DNA using primers indicated in Supplemental Table 4 and inserted into the binary vector pKGWFS7 (Karimi et al., 2002). Constructs were introduced into Agrobacterium strain GV3101 (pMP90) (Koncz and Schell, 1986) by electroporation. Plant transformation was done using the “floral dip” method (Clough and Bent, 1998).

To determine the cellular location of class I TCP factors, the GFP signal of roots in 50% glycerol of transgenic plants TCP7-OEX, TCP15-OEX, and TCP22-OEX was observed under an epifluorescence microscope (Nikon Eclipse E600, Nikon; excitation at 488 nm and emission at 495–575 nm).

Total RNA was extracted with RNeasy Plant Mini Kit (QIAGEN), and treated with on-column DNase (QIAGEN). cDNAs were generated with QuantiTect Reverse Transcription Kit (Qiagen). qPCR was performed on a MyiQ two-color real time PCR detection system with iQ SYBR Green supermix (Bio-Rad). Real-time PCR experiments were repeated using three biological replicates and one technical replicate. For tissue expression analysis of class I TCP genes the roots, young leaves, mature leaves and shoot apices were taken from 2-week old plants, the inflorescences were taken from the tip of the main stem. To analyze expression levels in mutant and transgenic backgrounds, three sets of 5–6 pooled entire 2-week old plants were collected and analyzed. ACTIN8 was used as an endogenous control gene. The number of cycles for the RT-PCR was 28 for AtTCP7, AtTCP8, AtTCP14, AtTCP15, AtTCP21, and AtTCP22, 30 for AtTCP23 and 26 for ACTIN8. Primers used for RT-PCR and qPCR are specified in Supplemental Table 2.

To generate multiple tcp mutants, we made first the double mutants tcp8-1 tcp15-1 and tcp22-1 tcp23-1. Double homozygous recessive plants were crossed and from F3 progenies we selected by genotyping the quadruple mutant tcp8-1 tcp15-1 tcp22-1 tcp23-1. This multiple mutant was crossed with tcp21-1 and through selection by genotyping we isolated the pentuple mutant tcp8-1 tcp15-1 tcp22-1 tcp23-1 tcp21-1.

The primers used for the genotyping of the mutants are indicated in Supplemental Table 3. For tcp8-1 we used the combination tcp8-1Lp and tcp8-1Rp for the wild type copy and tcp8-1Lp and LB1SAIL and tcp8-1Rp and LB1SAIL for the insertions. Similarly, for tcp15-1, we used the combinations tcp15-1Lp and tcp15-1Rp for the wild type copy and tcp15-1Lp and LBa1 for the mutation. For tcp22-1, we used primers tcp22-1Lp and tcp22-1Rp for the wild type copy and tcp22-1Rp and LBa1 for the insertion. For tcp23-1, we used the combinations tcp23-1Lp and tcp23-1Rp for the wild type copy and tcp23-1Lp and LB1SAIL and tcp23-1Rp and LB1SAIL for the insertions. For tcp7-1, we used the primers TCP7p1000 and M2LPROr for the wild type copy and TCP7p1000 and NS343_GK-LB and M2LPROr and NS344_GK-RB for the insertions. For tcp7-2, the primers NS256_S18Lp and NS257_S18Rp for the wild type copy and NS256_S18Lp and LB1SAIL and NS257_S18Rp and LB1SAIL for the insertions. For tcp7-3, the primers NS254_S17Lp and NS255_S17Rp for the wild type copy and NS255_S17Rp and NS290_RB4INRA for the mutant copy. In case of tcp7-4, the combinations NS254_S17Lp and NS255_S17Rp for the wild type copy and NS255_S17Rp and NS314_p745Wisc for the mutant copy. For tcp21-1 or che-1, the primers NS550_CHEgF and NS551_CHEgR for the wild type copy and NS550_CHEgF and LBa1 for the insertion. For the design of the primers we used the web tool at http://signal.salk.edu/tdnaprimers.2.html. All the positions of the insertions were determined by sequencing.

For histochemical GUS activity assays, plants were submerged in 90% acetone for 20 min. on ice, washed twice with 50mM sodium phosphate buffer (pH 7.2) and subsequently incubated in staining solution (0.1 M sodium phosphate buffer (pH 7.2), 0.5 mM Fe(CN)2, 0.5 mM Fe(CN)3, 0.1% Tween-20 and 2 mM 5-bromo-4-chloro-3-indolyl-β-d-glucuronide) at 37°C for 6 h. Plants were cleared and dehydrated through a graded ethanol series. For sections, FAA-fixed and dehydrated material was embedded in Paraplast Plus (McCormick Scientific, St. Louis, MO). The sections were dewaxed by immersion in Histoclear and mounted in Permount (Fisher Scientific).

For the yeast two-hybrid, the full-length coding sequences of AtTCP7, AtTCP8, AtTCP15, AtTCP21, AtTCP23, AtTCP2, and AtTCP3 were amplified by PCR using Pwo polymerase (Roche) and cloned into either pGBKT7 bait vector (fused with Gal4 DNA-binding domain; BD) or pGADT7 prey vector (fused with GAL4 activation domain; AD, Clontech, http://clontech.com). The constructs were verified by sequencing and co-transformed into yeast strain AH109. To test for interactions, the bait and prey constructs co-expressed in yeast were subjected to β–Galactosidase filter assay according to the manufacturer's instructions (Clontech).

We amplified the coding sequences of AtTCP7, AtTCP15, AtTCP22, and AtTCP23, and cloned them into the pCR8/GW/TOPO TA cloning vector (Invitrogen). We generated GST-fusion–protein-expressing constructs through LR reactions between the pCR8 entry clones and the destination vectors pDEST15 (Invitrogen) containing an N-terminal GST tag. All the fusion proteins and empty control plasmids were expressed in Escherichia coli strain BL21 (DE3).

The probe used in EMSA is 76 bp long, from −239 to −315 bp upstream of the starting codon of STM and was prepared by annealing the two biotin-labeled anti-parallel primers Kboxf 5′-AACTATCACTATTTAGAATTTTCAATGTGGAAAAGGAAGCTGATTGTTGAAGCATAAATCCCGGGAGACCACTTTT-3′ and Kboxr 5′-AAAAGTGGTCTCCCGGGATTTATGCTTCAACAATCAGCTTCCTTTTCCACATTGAAAATTCTAAATAGTGATAGTT-3′. The same two unlabeled primers were annealed for competition. For the EMSA, recombinant GST-tagged proteins on pDEST15 were purified on an immobilized glutathione column (Pierce). Binding reactions and detection of biotin-labeled probe was performed according to the manufacturer's instructions (LightShift Chemiluminescent EMSA Kit, Pierce).

For the number of rosette leaves the plants were at bolting stage with 0.5 cm of the main stem elongated. We did not take account of the cotyledons. The petiole length, blade length, blade width, blade perimeter and blade area were determined using the sixth leaf of plants at bolting stage. The data were analyzed using LeafJ (Maloof et al., 2013) with ImageJ software (http://rsb.info.nih.gov/ij/). For leaf epidermis analysis, dental resin imprints and nail polish copies (Kagan et al., 1992) were taken from the medial leaf lamina zone from rosette leaves of Col wt, TCP7-SRDX, and TCP23-SRDX plants and analyzed by light microscopy in a Nikon Eclipse E600 (Nikon).

Sequence data from this article can be found in The Arabidopsis Information Resource (TAIR; http://www.arabidopsis.org/) or GenBank/EMBL databases under the following accession numbers: AtTCP7 (At5g23280), AtTCP8 (At1g58100), AtTCP22 (At1g72010), and AtTCP23 (At1g35560).