Evolutionary and molecular analysis of Dof transcription factors identified a conserved motif for intercellular protein trafficking


Author for correspondence:

Jae-Yean Kim

Tel: +82 (0)55 772 1361

Email: kimjy@gnu.ac.kr


  • Cell-to-cell trafficking of transcription factors (TFs) has been shown to play an important role in the regulation of plant developmental events, but the evolutionary relationship between cell-autonomous and noncell-autonomous (NCA) TFs remains elusive.
  • AtDof4.1, named INTERCELLULAR TRAFFICKING DOF 1 (ITD1), was chosen as a representative NCA member to explore this evolutionary relationship. Using domain structure–function analyses and swapping studies, we examined the cell-to-cell trafficking of plant-specific Dof TF family members across Arabidopsis and other species.
  • We identified a conserved intercellular trafficking motif (ITM) that is necessary and sufficient for selective cell-to-cell trafficking and can impart gain-of-function cell-to-cell movement capacity to an otherwise cell-autonomous TF. The functionality of related motifs from Dof members across the plant kingdom extended, surprisingly, to a unicellular alga that lacked plasmodesmata. By contrast, the algal homeodomain related to the NCA KNOX homeodomain was either inefficient or unable to impart such cell-to-cell movement function.
  • The Dof ITM appears to predate the evolution of selective plasmodesmal trafficking in the plant kingdom, which may well have acted as a molecular template for the evolution of Dof proteins as NCA TFs. However, the ability to efficiently traffic for KNOX homeodomain (HD) proteins may have been acquired during the evolution of early nonvascular plants.


In the plant kingdom, the intercellular trafficking of regulatory proteins and mRNAs is facilitated by symplasmic communication channels, established by plasmodesmata (PD), to allow for the noncell-autonomous (NCA) control of plant development (Zambryski & Crawford, 2000; Cilia & Jackson, 2004; Lucas & Lee, 2004; Oparka, 2004; Gallagher & Benfey, 2005; Maule, 2008; Lucas et al., 2009; Guseman et al., 2010). Transcription factors (TFs), which act as key regulators of major developmental programs, are of special interest in this context. The first reported endogenous NCA TF was the maize KNOTTED1 (KN1; Vollbrecht et al., 1991; Jackson et al., 1994; Lucas et al., 1995). KN1 was shown to be expressed in the L2 layer of the shoot apical meristem, whereas the protein was detected in the L1 layer, where it controls cell fate determination (Vollbrecht et al., 1991; Sinha et al., 1993).

The Antirrhinum floral homeotic proteins, DEFICIENS and GLOBOSA (Perbal et al., 1996), together with TFs involved in root development, such as the Arabidopsis GRAS family member, SHORT ROOT (SHR; Nakajima et al., 2001), and CAPRICE, a Myb-like DNA-binding domain protein (Wada et al., 2002; Kurata et al., 2005), represent examples of other well-characterized NCA TFs. In the case of SHR, it is expressed in the stele and its site of action includes the cortex/endodermal initials, where it controls asymmetric cell division and endodermal cell fate. Interestingly, the conserved GRAS domain in SHR and nuclear localization appear to be required for movement (Gallagher & Benfey, 2009).

Intercellular movement of such NCA TFs can involve either a selective (Lucas et al., 1995; Wada et al., 2002; Lee et al., 2003; Kurata et al., 2005; Gallagher & Benfey 2009) or nonselective (Oparka et al., 1999; Crawford & Zambryski, 2000, 2001; Wu et al., 2003) pathway. Some proteins can move nonselectively by simple diffusion through PD. LEAFY (LFY) and cytoplasmically localized green fluorescent protein (GFP) have been suggested to follow this pathway (Crawford & Zambryski, 2001; Wu et al., 2003; Kim et al., 2005a). Such proteins appear to require no specific motifs to interact with PD components and move cell to cell by diffusion through PD. By contrast, selective trafficking of NCA TFs involves specific peptides, or motifs, which mediate in their targeting to and trafficking through PD (Gallagher et al., 2004; Kim et al., 2005b; Taoka et al., 2007).

In this study, we report the NCA capacity of the Arabidopsis Dof family of TFs. A detailed analysis was performed on one member, INTERCELLULAR TRAFFICKING DOF 1 (ITD1; At4g00940), to elucidate its mode of cell-to-cell trafficking. Domain structure–function analysis identified a motif within ITD1, termed the intercellular trafficking motif (ITM), that is necessary and sufficient for intercellular trafficking. Related motifs from both cell-autonomous and NCA Dof family members were examined to gain insights into the evolution of this capacity for plant Dof TFs to function as NCA regulatory agents.

Materials and Methods

Plant materials and growth conditions

Enhancer trap line J0571 was obtained from the Arabidopsis Biological Research Center, and a gl1-1 mutant (Oppenheimer et al., 1991) was used in our trichome rescue assays. Leaves from Nicotiana benthamiana, Oryza sativa cv japonica, Glycine max and Chlamydomonas reinhardtii strain CC-503 cw92 mt+ and Physcomitrella patens cultures were used for the isolation of RNA and genomic DNA. Seeds were germinated on either Murashige and Skoog (MS) plates or directly in soil. The floral dip method was conducted using Col-0, J0571 or gl1 for the production of transgenic plants. Transgenic plants were selected on MS plates containing 50 mg l−1 kanamycin in phytagar (Duchefa Biochemie, Haarlem, the Netherlands) or 30 mg l−1 hygromycin in phytagel (Sigma, St Louis, MO, USA). T1 and T2 plants were employed for these assays. Two-week-old transgenic plants were transferred to soil. Plants were grown at 22°C and 60% relative humidity under a 16 h : 8 h light : dark regime with a light intensity of 150 μmol m−2 s−1.

Plasmid construction

The genomic DNA and total RNA were isolated using a DNeasy Plant Mini kit and RNeasy Plant Mini kit (Qiagen GmbH, Hilden, Germany), respectively, according to the manufacturer's instructions. After RNA extraction, RNase-free DNase (Promega, Madison, WI, USA) was used to eliminate potential contamination of genomic DNA. cDNA synthesis was performed using the SuperScript III First-Strand Synthesis System (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. The open reading frames of full length and protein fragments were amplified with PCR using Pfu DNA polymerase (Solgent, Dae-jeon, South Korea) according to the manufacturer's instructions. The PCR-amplified fragments in all constructs were analyzed by DNA sequence analysis to ensure that no mutations had occurred. Sequence information of the PCR primers used in this study is listed in Supporting Information Table S1.

The vector pZY375 (a gift from David Jackson, Cold Spring Harbor Laboratory, NY, USA) was restricted with SalI/SacI to remove the mGFPer sequence flanked by SalI/SacI sites and, subsequently, the mCherry sequence flanked by SalI/SacI sites was ligated into these sites to yield pCCL276 (UAS::mCherry). The histone-2B (H2B) fragment was amplified with the primer set H2B-AvrII_d1 and H2B-HpaK_r1 and inserted into the AvrII/KpnI sites in pCCL276 to yield pCCL284 (UAS::mCherry-H2B). The GW fragment was obtained from pEarlyGate103 as a XhoI fragment and inserted into the SalI site in pCCL276 to yield pCCL292 (UAS::GW-mCherry). pCCL292 was restricted with AvrII/KpnI and the H2B fragment was inserted to yield pCCL306 (UAS::GW-mCherry-H2B). For the cloning of the pCCL702 (pRbcS::GL1-GW) gateway destination vector, the GW fragment was amplified from pDest22 (Invitrogen), with the primer set attR-Bsgl_d1/attR-BKpn_r1, and inserted into the BglII/KpnI sites in pK1401 (a gift from David Jackson). The PCR products flanked by the attB sites were cloned into pDonr207 using Gateway BP Clonase II Enzyme Mix (Invitrogen), according to the manufacturer's instructions. The target fragments cloned in entry vectors, as described above, were transferred to the desired destination vectors using Gateway LR Clonase II Enzyme Mix (Invitrogen), according to the manufacturer's instructions, to obtain the required expression vectors.

Transcriptional GFPer and translational mCherry reporter fusion constructs for ITD1 were generated as follows. The 1779-bp 5′ upstream sequence from the ITD1 first codon was cloned into the modified gateway destination vector, pCCL310, to yield pCCL492 (pITD1::GAL4/UAS::GFPer). The 885-bp open reading frame (ORF) of ITD1, without the stop codon, was cloned by recombination into pCCL292 to yield pCCL493 (UAS::ITD1-mCherry). The UAS::ITD1-mCherry fragment and nopaline synthase (NOS) terminator were amplified by PCR and then inserted into the ApaI site, in pCCL492, to yield the final expression vector pCCL494 (pITD1::GAL4/UAS::GFPer/UAS::ITD1-mCherry).

To make various mCherry and mCherry-H2B fusion constructs, the cDNA sequences of full-length Dofs, Dof_ITM, truncated ITD1_∆ITM and other genes from different species were amplified by PCR, using specific primers, and then inserted into pENTR/D-TOPO (Invitrogen). Subsequently, these entry vectors were cloned into pCCL292 or pCCL306 gateway vector, as described above, by recombination.

To generate mesophyll-expressed GLABROUS1 (GL1) fusion constructs, the coding sequences of ITD1 deletions and other genes were amplified by PCR and then cloned into pCCL702 gateway vector. Alternatively, cMyc-tagged vectors used in this assay were generated by first synthesizing 2xcMyc flanked by BamHI/BglII and inserted into the single BglII site of pK1573 (Kim et al., 2005b), generating the pRbcS2b::GL1-2xcMyc-KN1_HD (pK1573-2xcMyc) construct. To generate pRbcS2b::GL1-2xcMyc-ITM/HD, the sequences of ITD1_ITM and KN1_HD homologous peptides from the various species were amplified by PCR using gene-specific primers, flanked by BglII/KpnI or BamHI/KpnI, and cloned into BglII/KpnI in pK1573-2xcMyc by replacing KN1_HD. To generate pRbcS2b::GL1-2xcMyc (a negative control), the KN1_HD sequence in the pK1573 construct was substituted with the PCR-amplified 2xcMyc peptide sequence flanked by BglII/KpnI.

For the cloning of various epidermal-expressed GL1-ITM/HD fusion vectors, the fragments of GL1-Dof_ITM/HD were PCR amplified, using the common forward GL1-AscI primer and specific reverse primers, flanked by the BamHI or KpnI site. These PCR products were then cloned into the AscI and BamHI sites in the pML1::GFP-KN1 (pK1204) construct (Kim et al., 2002), replacing GFP-KN1. Alternatively, the PCR products were generated using the aforementioned pRbcS2b::GL1-2xcMyc-ITM/HD as template and then cloned into the same AscI/KpnI sites in the pML1::GL1-ITM construct, as described above.

Scoring in the trichome rescue assay

Trichome rescue assays were conducted as described previously (Kim et al., 2005b); 2-wk-old seedlings were employed and the first four leaves were scored for the number of trichomes using a stereomicroscope (Carl Zeiss AG, Oberkochen, Germany).

Confocal laser scanning and electron microscopy

One-week-old Arabidopsis seedlings were used to examine the cellular distribution of GFPer and mCherry in roots and leaves. Fluorescence imaging was carried out using a model FV1000 confocal laser scanning microscope (Olympus, Tokyo, Japan) with excitation at 488 and 543 nm, and emission at 510–540 and 587–625 nm, for GFPer and mCherry, respectively. For scanning electron microscopy (SEM), the second pair of leaves from 1-wk-old seedlings was prepared, as described previously (Rim et al., 2009), and observed on a JSM-6380LV SEM (JEOL, Tokyo, Japan).

Immunoblot analysis

Tissue from 3-wk-old plants was collected and ground in liquid nitrogen. Protein was extracted in grinding buffer containing 50 mM Tris-HCl, pH 8.0, 50 mM EDTA, 5 mM EGTA-NaOH, 25 mM NaF, 100 mM NaCl, 10 mM MgCl, 0.1% Triton X-100, 0.2% Nonidet P-40, 10% glycerol, 1 mM phenylmethylsulfonylfluoride (PMSF), 1 mM dithiothreitol (DTT), Protease inhibitors cocktail (Roche Diagnostics, Mannheim, Germany) and 20 μM MG-132 (Calbiochem, San Diego, CA, USA). The various GL1-2xcMyc fusion proteins, in sodium dodecylsulfate (SDS) buffer and 100 mM DTT, were detected by western blotting using monoclonal anti-cMyc antibody (GenScript, Piscataway, NJ, USA) and horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG antibody (Santa Cruz, CA, USA), according to the manufacturer's instructions. Signal was detected by an ECL Advance Western Blotting Detection Kit (GE Healthcare, Little Chalfont, Buckinghamshire, UK). Equal loading controls were verified by staining the membrane with Ponceau S.

Software and web resources

cNLS Mapper software (Kosugi et al., 2009) was employed to locate the predicted nuclear localization sequences (NLSs). The classification of TF families was performed using PlnTFDB (http://plntfdb.bio.uni-potsdam.de/v3.0/) and PlantTFDB (http://planttfdb.cbi.edu.cn/). The amino acid sequences of the conserved DNA-binding domain in CrCO_CCT343–387 (XM_001698040) and CrYABBY36–103 (XM_001695861) were tested for the capacity of intercellular trafficking. The DNA-binding domains of ITD1_ITM61–110 and KN1_HD256–326 were used to search for potential homologous genes in different species (Tables S2, S3) through BLASTP, TBLASTN at the National Center for Biotechnology Information (NCBI; http://www.ncbi.nlm.nih.gov/Blast.cgi), Plant GDB (http://www.plantgdb.org/), Phytozome (http://www.phytozome.net/) and Chlamy Center (http://www.chlamy.org/) websites. Phylogenetic trees were constructed with MEGA5 software (Tamura et al., 2011). Sequence alignments were carried out using ClustalW2 software (Larkin et al., 2007). Evolutionary reconstruction was inferred using the neighbor-joining (NJ) method (Saitou & Nei, 1987). Construction of vector maps and in silico cloning were performed in the Vector NTI program, version 11 (Invitrogen).


ITD1 can traffic cell to cell in Arabidopsis root and leaf tissues

To explore the evolutionary relationship between cell-autonomous and NCA TFs, a genome-wide screen of Arabidopsis TFs was performed (Rim et al., 2011). During this screen, we isolated a putative NCA TF from the Dof family, AtDof4.1 (hereafter designated as ITD1). In an earlier study, Lee et al. (2006) tested the cell-to-cell movement capacity of several Dof proteins by comparing the expression patterns of transcriptional and translational GFP reporters. The translational reporters of ITD1 and Dof3.7 (At3g61850) showed an expanded pattern relative to their transcriptional reporters, whereas both reporters for Dof1.1 (At1g07640), Dof2.2 (At2g28810) and Dof2.4 (At2g37590) displayed equivalent patterns. Although these findings were consistent with ITD1 and Dof3.7 being NCA TFs, a comparison of the promoter-driven reporters from different transgenic loci may not always reflect the endogenous situation.

To confirm that ITD1 acts as an intercellular trafficking protein, ITD1 promoter-driven expression of cell-autonomous GFPer and ITD1-mCherry was examined from a single transgenic locus. In the root, GFPer expression was confined to pericycle cells (Fig. 1a,c). By contrast, ITD1-mCherry was detected in both the endodermal and cortical cell layers, suggesting that endogenously expressed ITD1 can move from its site of synthesis into neighboring cells (Fig. 1b,c). The expression of ITD-mCherry did not appear to alter the pattern of root development.

Figure 1.

The INTERCELLULAR TRAFFICKING DOF 1 (ITD1) transcription factor can traffic cell to cell within Arabidopsis roots. (a–c) Imaging of pITD1::GAL4/UAS::GFPer/UAS::ITD1-mCherry plant lines. The green fluorescent protein (GFP) signal defined the endogenous expression domain of ITD1 in the mature region of the root. ITD1-mCherry (arrowheads) moved out from its site of synthesis within the pericycle into endodermal and cortical cells. (d–o) Imaging of J0571 GAL4/UAS expressing GFPer and various forms of mCherry. (d–f) mCherry showed extensive movement from its site of synthesis in the cortex (Co) and endodermis (En) into cells throughout the root tip, including the lateral root cap (L), epidermis (Ep) and stele (St). (g–i) Myb88-mCherry failed to move outside of its expression zone. (j–l) ITD1-mCherry showed one cell layer outward and multicell layer inward movement (arrowheads). (m–o) Dof5.4-mCherry failed to move, although a strong signal was detected (boxed area) in cortical and endodermal cells in the meristematic region. Insets: magnified views of boxed areas. Bars, 20 μm.

The movement capacity of ITD1 was further investigated using various ITD1-mCherry fusions, expressed under the control of the Upstream Activation Sequence (UAS) in a background of the J0571 enhancer trap line (Birnbaum et al., 2003; Rim et al., 2011). This J0571 line expresses a GFPer reporter within the cortex and endodermis of the root (Figs 1d,g, S1). As a control for these experiments, the cellular pattern of fluorescence was examined for the UAS::mCherry (27 kDa) reporter. As illustrated in Fig. 1(e), the free form of mCherry could move from its site of synthesis (shown in Fig. 1d) into most cell types throughout the root tip (Fig. 1e,f). This indicated that the PD interconnecting the majority of root cells are open sufficiently to permit nonselective cell-to-cell movement of the 27-kDa mCherry via diffusion. By contrast, a larger Myb88-mCherry fusion (81 kDa) could not move beyond its site of synthesis (Fig. 1g–i). Red fluorescence from Myb88-mCherry in roots was restricted to the nuclei of cortical and endodermal cells. This indicated that either the increased size and/or the nuclear localization of Myb88-mCherry prevented its diffusion through PD.

Analysis of the UAS::ITD1-mCherry (61.7 kDa) transgenic J0571 enhancer trap line revealed the presence of a fluorescent signal, albeit weak, on both sides of the GFPer marked boundary (Fig. 1j–l). Inward movement from the endodermis to the stelar tissues appeared to be more effective than outward movement, as a stronger fluorescent signal was detected in the stele; no movement was observed to the lateral and distal root cap cells. The significant difference in the movement patterns between mCherry and ITD1-mCherry suggested either a differential pattern in the PD size exclusion limit or selectivity in ITD1 trafficking. In Arabidopsis, AtDof5.4 is a distant relative of ITD1 (Yanagisawa, 2002) and its size (33.6 kDa) is equivalent to that of ITD1 (33.4 kDa). To test whether AtDof5.4 is a cell-autonomous TF, we next examined the intercellular movement of AtDof5.4-mCherry in root tissues. The AtDof5.4-mCherry protein (61.8 kDa) was restricted to the expression zone marked by GFPer (Fig. 1m–o). Here, it should be noted that similar signal strengths were observed for both AtDof5.4-mCherry (Fig. 1k, small box) and ITD1-mCherry (Fig. 1n, boxed area). Thus, these findings offer support for the hypothesis that ITD1, but not AtDof5.4, has the capacity for selective intercellular trafficking.

In the J0571 background, the UAS::GFPer reporter is also expressed in the mesophyll and mature guard cells within leaves (Figs S1, S2a). As mature guard cells have been reported to be symplasmically isolated by the truncation of PD during the final stages of maturation (Wille & Lucas, 1984; Ding et al., 1997), we next tested whether ITD1 could move between the mesophyll and epidermal layers. The ITD1-mCherry signal was detected in leaf epidermal cells, consistent with its movement from the subepidermal layer (Fig. S2b,c). Similarly, free mCherry could be detected in epidermal cells (Fig. S2d–f), whereas a cell-autonomous fusion of mCherry-H2B was restricted to the expression zone (Fig. S2g–i).

ITD1 traffics by a selective NCA pathway

The trichome rescue assay developed previously (Kim et al., 2005b) was employed to further test for the selective trafficking capability of ITD1. Here, we asked whether this NCA TF, like KN1, could confer gain-of-trafficking function on the otherwise cell-autonomous protein GL1. To this end, a GL1-ITD1 fusion protein was expressed specifically in the mesophyll of gl1 mutant plants. Analysis of transgenic lines (T1) indicated that GL1-ITD1 could, indeed, rescue trichome development in the gl1 genetic background (Fig. 2a). Controls for these experiments were the gl1 mutant (Fig. 2b) and wild-type Columbia-0 (Fig. 2c) plants. SEM analysis of leaves from these same plant lines confirmed the presence of trichomes on both the GL1-ITD1 (Fig. 2d) and wild-type (Fig. 2f) plants, and the glabrous nature of the gl1 mutant line (Fig. 2e).

Figure 2.

An intercellular trafficking motif (ITM) imparts noncell-autonomous (NCA) property to INTERCELLULAR TRAFFICKING DOF 1 (ITD1). (a–c) Trichome rescue assay. (a) gl1 transgenic line expressing GL1-ITD1 exhibits trichome rescue. (b) gl1 plants develop glabrous leaves. (c) Wild-type leaves displaying normal trichome pattern. Insets: magnified views of boxed areas. Bars, 1 mm. (d–f) Scanning electron micrographs of leaves from the plant lines shown above (a–c, respectively). Bars, 200 μm. (g) Predicted domain structure of ITD1 and deletion mutants tested in the trichome rescue assay. Numbers correspond to amino acid residues at the specific motif borders and deletion sites. ZFM, zinc finger motif; NLS, nuclear localization sequence; ITM, intercellular trafficking motif. Lower bars: 11 deletion mutants; (−), no trichome rescue; (+), trichome rescue. (h) Trichome rescue assays conducted with the indicated GLABROUS1 (GL1) fusions. Trichome rescue statistics are provided in Supporting Information Table S4. KNOTTED1 (KN1) trichome rescue data from Kim et al. (2005b). (i) Functional property of 2xcMyc-tagged chimeric protein in trichome rescue and Western blotting assays. Blue bars, pRbcS2b::GL1-2xcMyc-; gray bars, pML1::GL1-2xcMyc-. All epidermal-expressed GL1 fusion proteins (pML1::GL1-2xcMyc-ITD1 fragment) rescued trichome development in the gl1 background, whereas only the mesophyll-expressed GL1-ITM fusion protein could rescue trichomes in gl1. The number above each bar is the average number of trichomes per leaf pair in 3-wk-old plants. Trichome rescue statistics are provided in Tables S4 and S5. Western blots showing similar protein levels for each mesophyll-expressed GL1 fusion transgene in 3-wk-old seedlings. Ponceau S staining of the Rubisco 55-kDa large unit was used as loading control. (j–m) Cellular patterns observed in Arabidopsis root for mCherry-H2B alone or fused to the indicated Dof motifs using the J0571 line. (j) Control using mCherry-H2B. (k) The Dof5.444–93-mCherry-H2B motif remained confined to cortical and endodermal cells. (l) ITD1_ITM-mCherry-H2B showed multicell layer movement into the stele. (m) The ITD1_∆ITM-mCherry-H2B signal was confined to nuclei in cortex/endodermal cells. Co, cortex; En, endodermis; St, stele. Bar, 20 μm.

The statistics for trichome rescue experiments using the NCA TF, KN1 (full-length clone), was on the order of 27% (Kim et al., 2005b); an equivalent value was obtained in our GL1 ITD1 study. Here, we observed trichome rescue in 75 of 271 transgenic plants examined, that is, a 28% rescue (Fig. 2h, Table S4). As a control, we employed the cell-autonomous TF, AtDof5.4; only five of 207 independent transgenic plants had trichomes, mostly only a single trichome, on their leaves (Fig. 2h). To confirm whether the GL1-AtDof5.4 construct could rescue trichome development, when expressed directly in the epidermal cells, an epidermal-specific ML1 promoter was used to express GL1-AtDof5.4. Here, 78% of T1 plants showed trichome rescue (Table S5). These results confirm the functionality and stability of the fusion. This result is consistent with the inability of AtDof5.4-mCherry to move within the root (Fig. 1m–o).

ITD1_ITM imparts the capacity for intercellular trafficking

Domain structure analysis was next performed on ITD1 to identify potential motifs that might have evolved to impart NCA function to this TF. As with other members of the Dof TF family, ITD1 has a conserved CX2CX21CX2C zinc finger motif (ZFM) within its N-terminal region (Fig. 2g; amino acid residues 70–98). A putative NLS comprising the bipartite basic region (amino acid residues 99–122) was also predicted based on a previous study (Kreb et al., 2010) and a search using cNLS Mapper software (Kosugi et al., 2009); this bipartite NLS is located immediately adjacent to the ZFM (Fig. 2g).

The trichome rescue assay was again employed to test the role of these identified motifs in mediating ITD1 intercellular trafficking. Various deletions were engineered to give a series of GL1-ITD1Δ constructs which were then transformed into the gl1 mutant line. Initially, the ITD1 ORF was divided into four fragments (Fig. 2g). Interestingly, the ITD11–207, ITD161–207 and ITD161–294 fragments, which all included the ZFM and NLS regions, mediated significantly higher trichome rescue (44%, 30% and 24%, respectively) compared with ITD11–60 (2%; Fig. 2h, Table S4). We next dissected the ITD161–207 region to further explore the functional motif required for intercellular trafficking (Fig. 2g,h). These studies established that ITD161–110 and ITD161–158 (i.e. residues spanning the ZFM) gave 40% and 25% rescue rates, respectively. By contrast, ITD1111–158, ITD1111–207, ITD1159–207, ITD1208–294 and ITD1Δ61–110 all failed to mediate in trichome rescue (4%, 3%, 2%, 1% and 1%, respectively). Therefore, these findings indicated that the ITD161–110 region containing a canonical ZFM is required for cell-to-cell trafficking; this region is hereafter referred to as the ITM.

To further investigate the protein stability and expression properties for these test fusion proteins, we also conducted studies on five cMyc-tagged fusion proteins, of a similar size range, by expressing each under the mesophyll-specific RbcS2b promoter or the epidermis-specific ML1 promoter (Fig. 2i). In the gl1 background, only mesophyll expression of ITD1_ITM was found to rescue trichome development significantly. None of the other four chimeric proteins rescued trichome development significantly, whereas, when expressed under the ML1 promoter, all were effective for trichome rescue, thereby indicating the functionality of these fusion proteins.

Western blotting analysis conducted on protein extracted from constructs driven by the mesophyll-specific RbcS2b promoter indicated similar expression levels for all tested chimeric proteins. Thus, the absence of trichomes must reflect the inability of the tested construct to confer NCA function on the GL1 chimeric protein, rather than being caused by protein instability. Taken together, these experiments established that the region located within ITD1_ITM, including the DNA-binding ZFM, is necessary and sufficient for intercellular trafficking of ITD1.

In order to further test the role of the ITM in selective NCA TF trafficking, we next developed a root-based selective trafficking system using H2B (16 kDa). This chromatin-binding protein has a strong NLS that can mediate the sequestration of a fusion protein within the nucleus (Fig. S2g–i; Boisnard-Lorig et al., 2001). In contrast with the expression of mCherry (Fig. 1e,f), an mCherry-H2B fusion protein (44.2 kDa) was confined to the GFPer marked expression zone in the root tip (Fig. 2j). Thus, the fusion of H2B prevented mCherry from diffusing out through PD into the surrounding cells.

This H2B system was next used to analyze the distribution pattern of ITD1_ITM, ITD1_∆ITM and an equivalent region of ITD1 ITM from AtDof5.4 (AtDof5.444–93) fused to mCherry-H2B. In marked contrast with AtDof5.444–93 (Fig. 2k) and ITD1_∆ITM (Fig. 2m) fusion proteins, the ITD1_ITM fusion protein moved from the endodermis into the stele, where it accumulated in nuclei (Fig. 2l). These studies provided further support for the role of ITD1_ITM in imparting cell-to-cell gain-of-function movement capacity to the otherwise cell-autonomous mCherry-H2B protein.

Trafficking motif conservation among Arabidopsis Dof family members

Sequence alignment of the region surrounding the ZFM for the Arabidopsis Dof family revealed a high degree of conservation (Fig. S3). Hence, we next tested whether equivalent regions to ITD1_ITM from other family members could also impart cell-to-cell trafficking capacity to GL1. Five additional Dof members from the various clades (Fig. S4) were selected for testing in the trichome rescue assay (Yanagisawa, 2002). Importantly, all five Dof members are expressed in both leaf and root organs (Genevestigator). As predicted from the phylogenetic analysis and level of sequence homology of these putative ITMs (Fig. 3a), AtDof3.7 and AtDof4.5, belonging to the same subclade as ITD1, conferred significant trichome rescue (Fig. 3c, Table S6).

Figure 3.

Evolutionary relationship between ITD1_ITM and corresponding regions of other Dof family members. (a) Multiple sequence alignment of the region spanning the ITD1_ITM and the zinc finger motifs (ZFMs) of selected AtDof members; note the high degree of sequence conservation between these regions. Phylogenetic tree (left) shows the relationship between these protein motifs. Bar, 0.1% sequence divergence. Numbers refer to the ITD1_ITM residues. N-var, N-terminal variable region; ZFM, CX2CX21CX2C; B1, region containing the first motif (shown as red lines) of the bipartite nuclear localization sequence (NLS). (b) Multiple sequence alignment of the region spanning ZFMs from selected members of the Dof family (species identities given in Table S2). Phylogenetic tree is shown on the left. Bar, 0.1% sequence divergence. The first two letters of the names indicate the initials of the species (Table S2). (c) Trichome rescue assays conducted with the indicated ITD1_ITM and related motifs from the indicated genes fused to GLABROUS1 (GL1). Blue bars, pRbcS2b::GL1-2xcMyc-; gray bars, pML1::GL1-2xcMyc-. Most mesophyll-expressed GL1 fusion proteins could rescue trichomes in the gl1 background. Note that mesophyll expression of GL1-AtDof5.4 and GL1-AtDof2.2 constructs failed to rescue trichome development; however, their expression in the epidermis resulted in efficient trichome rescue. Trichome rescue statistics are provided in Tables S5 and S6. (d) Domain swapping assay for ITD1_ITM and the related AtDof5.4 motif. Both the ITD1 N-var region and ZFM are required for conversion of AtDof5.4 to a noncell-autonomous transcription factor (NCA TF). Constructs employed in trichome rescue assays incorporated the indicated exchange of AtDof5.4 (yellow) and ITD1 (blue) N-var and ZFMs. The numbers of plants showing trichome rescue/total number of plants examined (as %) are shown on the right of each construct. (e, f) Cellular patterns observed for different mCherry-H2B fused to putative moss and algal Dof_ITMs. Physcomitrella patens PpDof19 (e) and Chlamydomonas reinhardtii CrDof (f) showed multicell layer movement (arrowheads) into the stele. Co, cortex; En, endodermis; St, stele. Bar, 20 μm.

The putative ITMs from the more distantly related AtDof2.2 and AtDof5.4 family members were ineffective in rescuing trichome development (Figs 3a,c, S4), but epidermal expression of these chimeric proteins resulted in effective trichome rescue (Fig. 3c, Table S5). However, the putative ITM from the most distantly related AtDof5.6 was found to facilitate trichome rescue (Fig. 3a,c). Previously, it has been reported that ITD1 and AtDof3.7, but not AtDof 1.1, AtDof2.2 and AtDof2.4, appear to be NCA TFs (Lee et al., 2006), which is consistent with our findings from the trichome rescue assays. Based on these findings, and particularly that AtDof5.6 ITM has the capacity to impart NCA function to GL1, it would seem that the capacity for selective intercellular trafficking of AtDof TFs could well have been an early derived trait. In this regard, mutations in the ancestral Dof ITM could have generated dysfunctional ITMs, thereby converting these family members into cell-autonomous TFs.

Functional analysis of the N-variable and canonical ZFM regions of ITD1 and AtDof5.4

The ITM comprises the N-terminal variable region (N-var), the canonical ZFM and the first motif of the bipartite NLS. To further probe the role of the ITM, in terms of imparting cell-to-cell trafficking capacity in the Arabidopsis Dof TF family, we next developed a series of constructs in which the N-var and ZFM of ITD1 and cell-autonomous AtDof5.4 (Fig. 3a,b) were swapped (Fig. 3d). Here, trichome rescue was not observed when the ITD1 N-var region was swapped into AtDof5.4; the reciprocal swap of the AtDof5.4 N-var region into ITD1 also resulted in a loss of its movement capacity (Fig. 3d). These experiments established that the ITD1 N-var region is necessary, but not sufficient, to impart NCA function to AtDof5.4. Equivalent experiments, in which both the N-var region and ZFM (i.e. the full ITM) were swapped between ITD1 and AtDof5.4, indicated that the modified AtDof5.4, but not ITD1, could rescue trichome development (Fig. 3d). This result indicated that both the ITD1 N-var region and the ZFM are necessary, or that the entire ITM, including the first motif of the bipartite NLS, is necessary and sufficient to mediate selective intercellular trafficking.

Evolution of intercellular protein trafficking capacity within the Dof TF family

The Dof family of TFs is plant specific, with members being present in unicellular green algae, lower and higher plants (Shigyo et al., 2007). A single Dof gene has been reported in Chlamydomonas (Shigyo et al., 2007), and this TF might represent the ancient prototype of the Dof family. As a Dof TF pre-existed in unicellular organisms that had yet to develop PD, we hypothesize that specific Dof TFs acquired NCA trafficking capacity through gene duplication, followed by subfunctionalization.

A phylogenetic analysis based on the related ITM regions from seven disparate species indicated a high degree of conservation across these representatives (Fig. 3b, Table S2). Equivalent regions to the ITD1_ITM from these species were employed in trichome rescue assays to test whether the algal and moss (a nonvascular plant) sequences would be ineffective in mediating the cell-to-cell trafficking of GL1 in Arabidopsis. Surprisingly, and of considerable interest, the Dof_ITMs from the moss Physcomitrella and the unicellular alga Chlamydomonas were equally effective in rescuing trichome development, as were the ITMs from Nicotiana tabacum (tobacco) and Glycine max (soybean; Fig. 3c, Table S6). In agreement with these results, the ITM regions from P. patens (PpDof19_ITM) and Creinhardtii (CrDof_ITM) clearly supported the movement of their ITM-mCherry-H2B inward towards the stele (Fig. 3e and f, respectively). Taken together with our results obtained from the different AtDof_ITMs tested, these results suggested that this motif, which imparts NCA function to members of the Dof TF family, must have predated the evolution of the PD-selective trafficking pathway.

To address whether or not the evolution of Dof NCA function represents a general case, we expanded our study to explore the KNOX homeodomain (HD), one of the best characterized domains of NCA TFs. Based on analysis of the HD phylogenetic tree (Fig. S5 and Table S3) and sequence alignment of HDs from candidate genes (Fig. 4a), three Physcomitrella (Champagne & Ashton, 2001) and two Chlamydomonas (Lee et al., 2008) HDs were tested for their ability to restore trichome development. Interestingly, as with the maize KN1_HD, all three moss KNOX HDs (MKN2, MKN4 and MKN1-3) from P. patens were effective in rescuing trichome development (Fig. 4b, Table S7). By contrast, the two C. reinhardtii KNOX HDs (GSM1 and HDG1) proved to be ineffective in the rescue of trichome development. In addition, each fusion protein expressed under the ML1 promoter was sufficient to rescue trichome development (Fig. 4b, Table S5), indicating that these fusion proteins are biologically active in epidermal cells.

Figure 4.

Evolutionary relationship between different KNOX homeodomain (HD) regions. (a) Sequence alignments of the HDs from the indicated genes from maize, moss and the alga Chlamydomonas. Phylogenetic tree is shown on the left. Bar, 0.1% sequence divergence. The nuclear localization sequence (NLS) and three helices are marked above the sequence. Numbers indicate KN1 HD residues. Yellow, conserved residues; green, identical residues; blue, similar residues. (b) Trichome rescue assays conducted with the indicated KNOX HDs from Zea mays (KN1), Physcomitrella patens (MKN1–3, MKN2 and MKN4) and Chlamydomonas reinhardtii (GSM1 and HDG1) fused to GLABROUS1 (GL1) driven by either the RbcS2b (mesophyll) or ML1 (epidermal) promoter. Chlamydomonas reinhardtii CO and YABBY sequences were used as controls. Red bars, pRbcS2b::GL1-2xcMyc-HD; gray bars, pML1::GL1-2xcMyc-HD. Species identities are given in Table S3. Trichome rescue statistics are provided in Tables S5 and S7. Abbreviations: KN1, KNOTTED1; MKN2 and MKN4, moss Class 1 KNOX proteins; MKN1–3, moss Class 2 KNOX protein; GSM1, Gamete-specific minus1; HSG1, HOMEODOMAIN GLABROUS1; CO, 45-amino-acid sequence containing the CCT domain of CONSTANS; YABBY, construct includes a 68-amino-acid C2C2 zinc finger-like domain and a helix–loop–helix motif.


We identified ITD1, a member of the Dof TF family, as having the capacity to mediate its selective intercellular trafficking. This property was mapped to the highly conserved region spanning the ZFM that functioned as an ITM. However, not all AtDof TFs tested exhibited this ability for cell-to-cell movement. Analysis of the Dof ITM from a broader range of plant species revealed that the NCA property of ITD1 probably represents an early derived evolutionary trait.

One possible evolutionary scenario is that, during the transition from algal to early land plants, PD evolved as a pathway for the efficient exchange of both nutrients and information molecules, including proteins and RNA (Lucas & Wolf, 1993; Lucas et al., 1993; Franceschi et al., 1994; Qiu, 2008). Subsequent changes in PD substructure are thought to have imposed limitations on the size of proteins that could diffuse through PD and, subsequently, led to the evolution of PD-selective trafficking pathways (Lucas et al., 2009). It is possible that the algal Dof ITM could have served as a molecular template for the evolution of one such selective NCA TF pathway. Following gene duplication, random mutations within this ancestral Dof ITM could then have given rise to family members that were no longer compatible with this NCA trafficking machinery; these duplicated genes would be represented by current cell-autonomous Dof TFs.

In the case of the KNOX HD proteins, the ability to efficiently traffic cell to cell appears to have been acquired during the evolution of early nonvascular land plants. In this regard, the weak trichome rescue activity of the algal GSM1 (5.6%, with an average of only one trichome per leaf pair, compared with 72.4% and an average of 20 trichomes per leaf pair for KN1 HD) might reflect a limited level of homology to the P. patens HD that is also retained in higher plant NCA KNOX HDs.

An alternative scenario is that the Dof ITM confers movement via an unknown non-PD-mediated pathway that exists in both plants and algae. In the case of the KNOX HD, this domain can move between cells in plants and animals (Kim et al., 2002, 2003; Ruiz-Medrano et al., 2004; Tassetto et al., 2005; Lee et al., 2008). Thus, we cannot discount the possibility that PD- and non-PD-mediated trafficking of TFs may establish parallel trafficking pathways in the plant. Further studies are required to test for the operation of such non-PD pathways in plants that are capable of trafficking TFs.

Inspection of the residues within each N-var region and ZFM for the 13 Dof proteins analyzed (Fig. 3a,b) indicated that no obvious correlation could be found regarding specific residue changes and cell-autonomous vs NCA TF function. Similarly, a comparison of the ITM with previously characterized motifs for other NCA proteins (Fig. S6) failed to establish any clear sequence or structural features common to this class of protein. This suggests that this ITM, required for cell-to-cell trafficking of Dof NCA TFs, involves a three-dimensional structural element that can be composed of a class of residues, rather than a simple sequence motif. Recently, a component of a type II chaperonin complex, CCT8, has been suggested to regulate the transport of the KN1 HD and viral movement proteins (Xu et al., 2011; Fichtenbauer et al., 2012). It will be interesting to test whether this protein is involved in the ITM trafficking pathway.

Our current study provides important insights into the evolution of selective intercellular trafficking by plant TFs. As plants have acquired a wide range of proteins having the capacity to traffic through PD, the approaches developed in the current study should provide a means to probe the evolutionary processes underlying the emergence of the individual pathways utilized by these proteins.


We thank Sunseon Kim and Mohammad Nazim Uddin for technical assistance. This work was supported by the World Class University (WCU) program (R33-10002) through the National Research Foundation of Korea, funded by the Ministry of Education, Science and Technology and a grant from the Next-Generation BioGreen 21 Program (SSAC no. PJ009495), Rural Development Administration, Korea. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.