Introduction

Trichomes are highly differentiated epidermal cells of leaves representing physical and chemical barriers against environmental conditions. Cistus creticus is a dicotyledonous xerophytic shrub that grows in arid Mediterranean regions (Papaefthimiou et al. 2014). The multicellular glandular trichomes of C. creticus secrete “ladano”, a resin enriched in labdane-type diterpenes, renowned for antimicrobial and cytotoxic properties (Falara et al. 2008, 2010).

In the model-species Arabidopsis thaliana, numerous genes and molecular regulatory networks have been identified to modulate epidermal cell differentiation (Hulskamp et al. 1999; Balkunde et al. 2010; Tominaga-Wada et al. 2011). The WD40 repeat protein TRANSPARENT TESTA GLABRA1 (TTG1) is the common denominator of overlapping regulatory complexes that control cell fate and differentiation (Galway et al. 1994; Walker et al. 1999). In the root, TTG1 participates in the assembly of two distinct MYB-bHLH-WD40 (MBW) complexes that determine epidermal cell patterning (Schiefelbein et al. 2009, 2014). In non-hair cell files, TTG1 together with the MYB-domain protein WEREWOLF (WER) (Lee and Schiefelbein 1999), and two redundantly acting bHLHs, GLABRA3 (GL3) and ENHANCER OF GLABRA3 (EGL3) (Bernhardt et al. 2003, 2005; Payne et al. 2000; Simon et al. 2013), form a stable ternary WER-GL3/EGL3-TTG1 complex. This assembly positively regulates the expression of homeodomain-leucine-zipper (HD-Zip) transcription factor GLABRA2 (GL2), which represses transcription of the downstream hair-cell promoting genes (Di Cristina et al. 1996; Masucci et al. 1996; Costa and Shaw 2006; Bruex et al. 2012). In the root-hair cells a set of small, one-repeat MYB proteins that lack transcriptional activation domains, including CAPRICE (CPC1), TRIPTYCHON (TRY), and ENHANCER OF TRY AND CPC1 (Kirik et al. 2004; Schellmann et al. 2002; Simon et al. 2007; Wada et al. 1997) competitively inhibit WER binding to the MBW complex (Tominaga et al. 2007; Ishida et al. 2008). This site-specific complex is not competent to stimulate GL2 expression and thereby promotes root-hair cell fate.

In trichomes GL2 is a positive regulator of epidermal differentiation (Masucci et al. 1996; Szymanski et al. 1998; Zhao et al. 2008). The MBW complex is still composed of TTG1, GL3 or EGL3 (Payne et al. 2000; Zhang et al. 2003) and the R2R3-MYB transcription factor GLABRA1 (GL1) (Kirik et al. 2005). Trichome development and regular spacing are controlled by TRY (Schellmann et al. 2002). TRY protein migrates to neighboring cells and competes GL1 binding leading to the formation of GL2 inactive TRY-GL3/EGL3-TTG1 complex thereby inhibiting trichome fate (Schnittger et al. 1999; Esch et al. 2004).

The MBW transcriptional activation complex additionally modulates the late flavonoid biosynthetic genes. TTG1 associates with an MYB protein of the PRODUCTION OF ANTHOCYANIN PIGMENTS1 (PAP1)/PAP2/MYB113/MYB114 group, and a bHLH factor of the TRANSPARENT TESTA8 (TT8)/GL3/EGL3 group (Gonzalez et al. 2008; Zhang et al. 2003).

As the trichomes show diverse structural appearance and secondary metabolism, dissecting the interplay of the regulatory components involved in trichome patterning is stimulating. Here, we report that two CcSPLA/B proteins and their highly similar Arabidopsis homologues AtSPL4 and AtSPL5, physically interact with CcTTG1 and AtTTG1, respectively. Ectopic expression of miR156-resistant forms of AtSPL4/5 genes in transgenic Arabidopsis lines diminished the function of the TTG1/GL2 developmental pathway. Our results support the role of SPLs in trichome patterning control by potential modification of the MYB-bHLH-WD40 complex that is composed of heterogeneous protein partners.

Materials and methods

Plant materials and growth conditions

The Arabidopsis ttg1 mutation in the Landsberg erecta (Ler-0) genetic background has a single base transition that introduces a premature termination codon resulting in C-terminal truncation (Walker et al. 1999). The Arabidopsis transgenic lines of Columbia (Col-0) ecotype background express the miR156-resistant forms of SPLr4/5 genes under the control of the CaMV35S promoter (Wu and Poethig 2006). Cistus creticus plants were grown in pots under conditions as previously described (Falara et al. 2008). Arabidopsis thaliana and Nicotiana benthamiana plants were grown at 22 °C in a Fitotron (Weiss Gallenkamp; http://weiss-uk.com/) growth chamber with 100 μmol m−2 s−1 light intensity under 16 h light/8 h dark photoperiod. Morphometric analysis of epidermal cells performed as described (Rigas et al. 2009).

Isolation of CcTTG1

A partial cDNA clone flanking the 5′ sequence of a WD40 repeat protein was identified from sequencing analysis of 2022 ESTs from a C. creticus cDNA library prepared from mRNA isolated from trichomes (Falara et al. 2008). The full-length cDNA, designated CcTTG1, was isolated by PCR amplification with a gene specific primer annealing at the 5′ of the EST clone using the Phusion HF polymerase (Finnzymes-Life technologies; https://www.lifetechnologies.com/) and gene specific primers (Supplementary Table 2). The 1023 bp product was cloned and fully sequenced.

Generation of transgenic Arabidopsis thaliana plant lines

CcTTG1 was cloned into the BamHI site of the pBI121 binary vector. The Agrobacterium tumefaciens strain C58C1 RifR carrying the non-oncogenic Ti plasmid pGV3101 was transformed by electroporation with the pBI121 construct (GenePulser II, Bio-Rad; http://www.bio-rad.com/). The construct was introduced into ttg1 mutant background by vacuum infiltration as previously described (Rigas et al. 2009). T2 and T3 generation of transgenic plant lines selected on 40 mg l−1 kanamycin were assessed for capacity to complement ttg1 pleiotropic defects.

Gene expression analysis

Total RNA was isolated using the phenol-sodium dodecyl sulfate (SDS) extraction method (Rigas et al. 2009). RNA concentrations were determined spectrophotometrically and verified by ethidium bromide staining on agarose gels. DNA was eliminated with RQ1 RNase-free DNase (Promega; http://worldwide.promega.com/). Reverse transcription (RT) was performed on 3 μg of total DNA-free RNA using Superscript II Reverse Transcriptase (Invitrogen-Life technologies; https://www.lifetechnologies.com/). RT-PCR gene expression analysis was performed using specific primers (Supplementary Table 3). The linear phase of semi-quantitative expression analysis was validated by preliminary experiments comparing the relative amounts of PCR products under low and high number of PCR amplification cycles. Quantitative gene expression analysis was performed by PikoReal 96 Real-Time PCR System (Thermo scientific; http://www.thermoscientific.com/) using the SYBR Select Master Mix (Applied Biosystems-Life technologies; https://www.lifetechnologies.com/) and calculated by the ∆∆Ct method (Daras et al. 2009). Normalization of gene expression data was performed by using the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (AtGAPDH) as a reference. Total RNA isolation from C. creticus was performed as previously described (Pateraki and Kanellis 2004). RNA analysis by electrophoresis on denaturing formaldehyde agarose gels, Northern blotting onto nylon membranes (Schleicher and Schuell; http://www.schleicher-schuell.de/), hybridization and detection were performed as described (Sambrook et al. 1989).

Extraction and quantitative analysis of anthocyanins

Seedlings were grown under anthocyanin inductive conditions composed of high intensity light 300 μmol m−2 s−1 and increased supplemental carbon source 2 % sucrose as previously described (Poustka et al. 2007). Anthocyanin content was determined as reported (Mehrtens et al. 2005).

Yeast-two-hybrid interactions

The CcTTG1 cDNA was cloned into the BamHI/XhoI sites of pEG202 vector, containing the LexA DNA binding domain (BD) under the control of the ADH promoter. The C. creticus cDNA library of trichomes was constructed with 2 μg of total RNA using the SMART cDNA Library Construction Kit (Clontech; https://www.clontech.com/). The fragments of double stranded cDNA digested with SfiI restriction enzyme, larger than 500 bp, were cloned into the SfiI site of pJG4-5 plasmid, containing the B42 transcriptional activation domain (AD) under the control of the GAL1 promoter. Saccharomyces cerevisiae EGY48 yeast cells were co-transformed with BD and AD fusion constructs using the lithium acetate method (Gietz et al. 1992). Transformed yeast cells were tested on solid drop-out medium lacking histidine, tryptophan and leucine supplemented with 2 % galactose/1 % raffinose. To eliminate false positives yeast cells were tested on 2 % glucose.

BiFC assays

The PCR products of CcSPLA and CcSPLB cDNAs amplified using the primers listed in Supplementary Table 2 were cloned into the SmaI site of pSPYCE vector to generate in-frame fusions with YFPC (Walter et al. 2004). Likewise, CcTTG1 was cloned into the SmaI site of pSPYNE vector generating in-frame fusion with YFPN. For the isolation of AtTTG1, AtSPL4 and AtSPL5, 5 µg of total DNA-free RNA isolated from Arabidopsis leaves were reverse transcribed and the full length cDNAs were PCR amplified using the gene specific primers shown in Supplementary Table 3. The PCR products were initially cloned into the SmaI site of pUC19 and then isolated with SalI/XmaI restriction enzymes. AtTTG1 was inserted into the SalI/XmaI sites of pSPYNE vector, whereas AtSPL4 and AtSPL5 were introduced into the SalI/XmaI sites of pSPYCE vector, generating in-frame fusions with YFPN and YFPC, respectively. The basic leucine zipper (bZIP) transcription factor bZIP63 cloned into pSPYCE and pSPYNE vectors was used as positive control due to homodimers formation in the nucleus (Walter et al. 2004). The constructs were inserted into Agrobacterium tumefaciens cells. Transient transfection of the split-YFP fusion constructs in Nicotiana benthamiana epidermal cells and live cell imaging were performed as previously described (Daras et al. 2014).

Bioinformatics

The software packages and online tools used in this study are shown in Supplementary Table 4.

Accession numbers

Sequence data from this article can be found in the EMBL/GenBank databases under accession numbers: CcTTG1: KT892927; CcSPLA: KU145276; CcSPLB: KU041720; CcEF1a: EF062868.

Results

Structural differences occur between CcTTG1 and AtTTG1

The leaves of A. thaliana are covered by unicellular non-glandular trichomes, whereas C. creticus (Supplementary Fig. 1) has multicellular non-glandular and glandular trichomes, with the latter being the primary accumulation sites of “ladano” (Fig. 1a, b). To decipher the molecular mechanism that controls C. creticus epidermal cell differentiation to trichomes, the homolog of TTG1 gene (CcTTG1) was identified. Comparison of CcTTG1 deduced amino acid sequence with the Arabidopsis TTG1 (AtTTG1) homologue shows 77 % identity and 88 % similarity retaining a number of amino acid residues important for the three-dimensional conformation of the entire protein. CcTTG1 predicted to contain four WD40 repeat motifs ranging in length from 41 to 47 amino acids residues that perfectly match with the AtTTG1 motifs (Fig. 1c). The third motif (III), which is the longest, bears no proline residues contrary to the second (II) and fourth (IV), each having three prolines. However, the first (I) WD40 repeat motif of CcTTG1 consists of five prolines compared to the first motif of AtTTG1 that has eight prolines (Fig. 1c).

Fig. 1
figure 1

Structural properties and expression profile of CcTTG1. a The leaves of C. creticus are covered by glandular trichomes (GT) and stellate trichomes (ST). Scale bar 100 µm. b In Arabidopsis, the leaf trichomes are unicellular epidermal hairs with a branched morphology. Scale bar, 100 µm. c Sequence comparison between the deduced amino acid sequences of C. creticus TTG1 (CcTTG1) and A. thaliana TTG1 (AtTTG1). Dark shading depicts identical amino acids, whereas light grey shading represents amino acid residues with similar physicochemical properties. Boxed areas (I-IV) indicate the conserved WD-repeats. The proline residues are marked with dots. d Comparative molecular modeling of CcTTG1 and AtTTG1 protein structure. The ribbon model distinguishes four highly conserved WD-repeats (red) from the most variable N- and C-terminal domains depicted in yellow and green color, respectively. e Northern blot analysis of TTG1 expression in C. creticus tissues. The elongation factor 1a is used as loading control

Full size image

Phylogenetic analysis showed that CcTTG1 clusters together with TTG1-like WD40 proteins from dicotyledonous plants (Supplementary Fig. 2a; Supplementary Table 1). The Brassicaceae TTG1 homologs, including AtTTG1, group together consistent with the evolutionary relationships of these species (Supplementary Fig. 2a, b). These TTG1 homologs slightly deviated from CcTTG1, which is highly similar to the TTG1-like protein from cotton (Gossypium hirsutum), given that both species belong in the Malvales order.

Molecular models that project the structure of AtTTG1 and CcTTG1 predict distinctive structural differences as the primary structure of the N- and C-terminal domains is variable between the two polypeptides (Fig. 1d). Moreover, the proline residues located within the WD40 repeat motifs possibly determine the periodicity and regularity of blade folding that drives the formation of the circular propeller structure. Hence, CcTTG1 compared to AtTTG1 has subtle structural differences, albeit the typical TTG1-like WD40 protein conformation, that potentially could reflect the functional interplay with different regulatory components.

Expression profile analysis revealed that the CcTTG1 transcript is present in all tested organs (Fig. 1e), in agreement with previous reports in Arabidopsis (Walker et al. 1999). Nevertheless, in C. creticus, TTG1 appears to have a prominent expression in organs bearing high trichome and root-hair content including leaves, stems and roots.

CcTTG1 structural differences determine its functional capacity to control Arabidopsis trichome patterning

To test whether these subtle structural differences are important for the capacity of CcTTG1 to complement the functional impairments of ttg1, the Arabidopsis mutants were transformed with the CcTTG1 homologue (Supplementary Fig. 3). Phenotypic analysis of transgenic Arabidopsis ttg1 plants, referred as CcTTG1 transgenic plants, revealed a partial functional competence of CcTTG1 to restore trichome patterning (Fig. 2a). The number of trichomes on the cauline leaves of transgenic plants were significantly reduced by approximately one-third (Fig. 2b). However, the ectopic root-hair patterning of ttg1 mutant was fully restored (Fig. 2c). The epidermal cells differentiated to root-hairs were indistinguishable from the wild-type seedlings (Fig. 2d). Hence, the inability of CcTTG1 to fully complement the glabrous phenotype of ttg1 plants could be attributed to the subtle, nevertheless critical differences of CcTTG1 conformation and to the heterogeneity of the protein interaction partners within the Arabidopsis leaf epidermal cells that likely result in the formation of a rather unstable ternary transcriptional active complex. However, the results show that CcTTG1 is able to trigger the developmental cascade of Arabidopsis genes towards epidermal cell differentiation.

Fig. 2
figure 2

CcTTG1 reflects the heterogeneity of the transcription factors controlling Arabidopsis trichome formation. a CcTTG1 partially complements the glabrous phenotype restoring the formation of trichomes in cauline leaves. Scale bar represents 1 mm. b Trichome density calculated by the analysis of 15 distinct cauline leaves from each transgenic line. Asterisks indicate significant difference between the wild-type plants (Ler-0) and transgenic lines (P < 0.01; Student’s t test). c Primary root architecture of vertically grown 5-days-old seedlings. Scale bar represents 1 mm. d Root-hair density of 5-days-old seedlings calculated by the analysis of 20 primary roots from each transgenic line. No significant difference was found between the wild-type plants (Ler-0) and transgenic lines (P < 0.01; Student’s t test). e At the junction between hypocotyl and cotyledons of 9-days old seedlings the accumulation of anthocyanins is restored by CcTTG1. Scale bar represents 5 mm. f CcTTG1 complements the transparent testa phenotype restoring the accumulation of proanthocyanidins in the seed coat mucilage. Upper panel, Scale bar represents 1 mm. Lower panel, Scale bar represents 1 mm. g Extracts of anthocyanins from 9-days old seedlings. h Quantitative analysis of anthocyanin content. Values are means ± standard error (n = 4) normalized per g of fresh weight. No significant difference was found between the wild-type plants (Ler-0) and transgenic lines (P < 0.01; Student’s t test)

Full size image

Additionally, TTG1 forms a regulatory complex with MYB-bHLH transcription factors activating the expression of the late genes involved in the biosynthetic pathway of flavonoids. These genes result in the accumulation of anthocyanins and proanthocyanidins (PAs) in vegetative tissues and the seed coat, respectively (Li 2014; Xu et al. 2015). To assess the ability of CcTTG1 to complement the biosynthesis of flavonoids, the transgenic plant lines were grown under anthocyanin inductive conditions (Poustka et al. 2007). The biosynthesis of purple anthocyanins was restored at the junction between hypocotyl and cotyledons of the trangenic plants, whereas the yellow color of ttg1 seeds turned brown due to the accumulation of oxidized proanthocyanidins (Fig. 2e, f). To evaluate the reconstitution of the metabolic pathway, measurements of anthocyanins accumulation were made. The data pinpoint that CcTTG1 completely repaired the defects in anthocyanins biosynthesis, restoring their accumulation to wild-type levels (Fig. 2g, h). On the basis of these observations, the TTG1 homolog of C. creticus forms a stable ternary complex compensating the metabolic disorders of Arabidopsis ttg1 plants in different tissues.

CcTTG1 activates the GL2 regulator implicated in epidermal cell differentiation of Arabidopsis seedlings

The GL2 transcription factor is, among other genes, a downstream molecular target of the MBW complex (Masucci et al. 1996; Szymanski et al. 1998; Zhao et al. 2008). Hence, the functional competence of CcTTG1 to activate GL2 expression in ttg1 background was validated. As ttg1 is a nonsense mutation resulting in a premature stop codon, the expression level of AtTTG1 in the transgenic lines was similar to that in wild-type seedlings (Fig. 3a). Contrary to ttg1 seedlings, the ectopic expression of CcTTG1 highly increased the abundance of AtGL2 (Fig. 3a). Quantification analysis confirmed that the elevated expression level of AtGL2 was identical to wild-type seedlings and significantly different from ttg1 seedlings (Fig. 3b). Hence, in the seedlings that have cotyledons without trichomes, CcTTG1 fully restored the AtGL2 expression, a negative regulator of root-hair morphogenesis, in consistency with the reduction of the root-hair density observed in the transgenic lines (Fig. 2c, d). By using RNA from the first pair of cauline leaf primordia, the expression of the trichome positive regulator AtGL2 is also activated in CcTTG1 transgenic lines (Fig. 3c). However, the expression level of AtGL2 was, on average, equal to half of that observed in wild-type cauline leaves (Fig. 3d). This expression pattern coincides with the reduced number of trichomes observed on the cauline leaves of CcTTG1 transgenic plants (Fig. 2a, b). These results provide evidence for reconstitution of the TTG1/GL2 developmental pathway upon heterologous expression of CcTTG1 in Arabidopsis ttg1 seedlings and suggest that partial restoration of GL2 expression on cauline leaves trichomes is likely attributed to structural differences of CcTTG1 and to the heterogeneity of the protein interaction partners.

Fig. 3
figure 3

CcTTG1 restores the expression of AtGL2 key regulator of epidermal cell patterning in Arabidopsis ttg1 mutant background. Expression levels of AtTTG1, AtGL2 and AtGAPDH genes by semi-quantitative (a, c) and Real-time (b, d) RT-PCR analysis. No RT reactions indicate that the RNA samples are free from genomic DNA contamination. Values are means ± standard error for fold changes relative to wild-type ecotype Ler-0 detected by quantitative RT-PCR of three biological replicates, each analyzed in triplicate. In seedlings (b), no significant difference was found between the wild-type plants (Ler-0) and transgenic lines (P < 0.01; Student’s t test). In cauline leaves (d), the asterisks indicate significant difference between the wild-type plants (Ler-0) and transgenic lines (P < 0.01; Student’s t test)

Full size image

CcSPL transcriptional activators interact with CcTTG1 in the nucleus

Given the central role of TTG1 in multiple aspects of cell differentiation and metabolism, a screen was performed to identify protein interaction partners by the yeast two-hybrid assay. Using CcTTG1 as bait a cDNA library prepared from C. creticus trichomes was screened. Two SPL transcriptional activators, hereafter called CcSPLA/B, interacted with CcTTG1 (Fig. 4a). Consistent with the stability of CcSPLA/B-CcTTG1 interactions, the yeast cells failed to grow in the presence of glucose that repressed the transcription of the galactose regulatory system. To confirm these interactions in planta the bimolecular fluorescence complementation (BiFC) assay was applied in transiently transfected Nicotiana benthamiana epidermal cells. Reconstituted fluorescence was detected upon interaction of CcTTG1 with either CcSPLA or CcSPLB and localized exclusively in the nucleus of the transfected cells (Fig. 4b), confirming that CcSPLA/B directly bind to CcTTG1 in vivo.

Fig. 4
figure 4

Identification of CcTTG1 interaction partners. a CcSPLA/B bind to CcTTG1 in yeast. Yeast cells were co-transformed with combinations of binding domain (BD) and activation domain (AD) fusion constructs as indicated. Transformed yeast cells were grown on synthetic drop out (SD) selection plates lacking histidine, tryptophan and leucine (−HWL) in the presence of galactose (Gal) supplemented with raffinose or glucose (Glc). The pJG4-5 (AD) with pEG202 (BD), and the AD-CcSPLA/B fusions expressed with the BD alone were used as negative controls. b BiFC analysis of CcTTG1 interactions in the nucleus of transiently transfected Nicotiana benthamiana epidermal cells. bZIP63 that forms homodimers in the nucleus was used as positive control. Green fluorescence indicates the direct protein–protein interactions (arrowheads) between CcTTG1 and CcSPLA/B. Red fluorescence corresponds to chloroplasts due to chlorophyll auto-fluorescence. DAPI staining (5 µg mL−1) was used for visualization of the nuclei (middle panel). Scale bar represents 50 µm

Full size image

Phylogenetic analysis revealed that CcSPLA/B belong in the group of the miR156-targeted SPL transcription factors containing 11 members out of totally 17 SPL proteins encoded by the Arabidopsis genome (Xing et al. 2010). CcSPLA/B share a high level of protein similarity with particularly three miR156-targeted AtSPL3/4/5 factors (Supplementary Fig. 4a). Phylogenetic analysis revealed that CcSPLA/B and their Arabidopsis homologs AtSPL3/4/5 cluster together and they are clearly separated from the members of the AtSPL9 subfamily (Supplementary Fig. 4b). Furthermore, the identity of CcSPLA with AtSPL4 and AtSPL5 is 46 and 45 %, respectively, whereas this identity with AtSPL9 decreases to 18 %. Likewise, the identity between of CcSPLB with AtSPL4 and AtSPL5 is 35 and 37 %, respectively, whereas this identity with AtSPL9 decreases to 15 %. As the gaps upon the protein alignment between CcSPLA/B and AtSPL9 increase dramatically, the homologs of CcSPLA/B, SPL4 and SPL5, were selected as candidates to interact with TTG1. To determine whether this interaction is conserved among plant species, BiFC assays were performed using the Arabidopsis AtSPL4/5 homolog proteins. The fluorescent signal was also evident upon the interaction in planta of AtTTG1 with the AtSPL4 and AtSPL5 counterparts (Supplementary Fig. 4c). The nuclear localization pattern supports the conclusion that the in vivo binding between TTG1 and SPLs is conserved among evolutionary distant plant species (Supplementary Fig. 2b) revealing its importance in biological processes.

Although AtTTG1 has no obvious nuclear localization signal (NLS; Walker et al. 1999), the protein is present in both the nucleus and the cytoplasm (Zhao et al. 2008; Balkunde et al. 2011; Cheng et al. 2014). The most plausible hypothesis is that TTG1 is not transported to the nucleus but the protein interacting factors promote its nuclear localization (Zhao et al. 2008; Balkunde et al. 2011). Unlike AtTTG1, the AtSPL4/5 proteins contain a bipartite NLS that overlaps with the highly conserved SBP-domain of 76 amino acids (Birkenbihl et al. 2005), which is shown in Supplementary Fig. 4a. The SBP-domain has been reported to bind specifically to GTAC core motifs of Arabidopsis promoters (Birkenbihl et al. 2005; Wang and Wang 2015) and therefore may be involved in the nuclear localization of the AtTTG1-AtSPL4/5 complex (Supplementary Fig. 4c). In silico analysis of CcTTG1 sequence failed to predict any putative NLS, whereas CcSPLA/B have a putative NLS overlapping with the highly conserved SBP-domain (Supplementary Fig. 4a) most likely involved in the CcSPLA/B-CcTTG1 complex nuclear translocation. Like their Arabidopsis homologs, the protein domains except the SBP-domain of CcSPLs are highly variable. Based on this insight, CcSPLA/B in terms of physical interaction with CcTTG1 likely followed a conserved evolutionary path to modulate specific TTG1-dependent biological processes.

SPL4/5 interfere in vivo with the regulation of MBW complex transcriptional activity

Our results support the notion that TTG1 is recruited by the SPL factors and thereby rearranges the MBW complex. We tend, therefore, to favor the hypothesis that elevated levels of SPLs may retract TTG1 and diminish the transcriptional activity of the MBW complex, causing defects reminiscent to ttg1-like phenotypes. To validate this hypothesis, we used A. thaliana transgenic plants overexpressing the SPLr4/5 miR156-resistant (Wu and Poethig 2006) gene homologs of CcSPLA/B (Supplementary Fig. 4b). The stability of SPL4/5 mRNAs against miR156 accelerates vegetative-to-reproductive phase transition promoting flowering by direct activation of LEAFY, FRUITFULL and APETALA1 (Wu and Poethig 2006; Wu et al. 2009; Yamaguchi et al. 2009; Wang and Wang 2015). Our results confirmed that the high abundance of SPLr4/5 is tightly coupled with rapid entrance into the reproductive stage (Supplementary Fig. 5). The overexpressing SPLr4/5 lines affected a broad range of developmental processes including leaf morphology and lamina area (Fig. 5a, b). Interestingly, while the trichome pattern was evident, the number of trichomes on the adaxial surface of cauline leaves was reduced to approximately half upon elevation of SPL4/5 levels (Fig. 5c). Given that TTG1 interacts with SLP4/5 in both species, this phenotypic defect provides strong evidence that elevated SPL4/5 levels recruit TTG1, and therefore interfere with the stability of the MBW complex mimicking the ttg1-like trichome disorder.

Fig. 5
figure 5

Elevated SPL4/5 levels accelerate vegetative-to-reproductive phase transition and reduce trichome density through GL2 down-regulation. a and b Rapid phase transition modifies leaf morphology and leaf lamina geometry (n = 8). Scale bar represents 2 mm. c Trichome density per leaf lamina area on the adaxial surface of cauline leaves is reduced (n = 8). d and e Expression of the trichome inducer GL2 or trichome repressors TCL1 and TRY in the first pair of cauline leaf primordia. Relative gene expression levels by semi-quantitative (d) and real-time (e) RT-PCR analysis. No RT reactions indicate that the RNA samples are free from genomic DNA contamination. Values are means ± standard error for fold changes relative to wild-type ecotype Col-0 detected by quantitative RT–PCR of three biological replicates, each analyzed in triplicate. Student’s t test was used to indicate significant difference between the wild-type plants (Col-0) and the SPLr4/5-overexpressor lines at P < 0.01 (single asterisks) or P < 0.05 (double asterisks)

Full size image

To confirm that SPL4/5 attenuate the transcriptional activity of MBW complex, the expression of GL2 was monitored. RNA was isolated from the first pair of cauline leaf primordia when the plants were bolting, immediately after entering into the reproductive stage (Supplementary Fig. 5). Upon ectopic expression of SPLr4 and SPLr5, the levels of GL2 expression were decreased, whereas the expression of TTG1 remained constant (Fig. 5d). Quantification analysis revealed that GL2 expression was significantly reduced to approximately half and one-third in SPLr4 and SPLr5 plants, respectively (Fig. 5e), corroborating the phenotypic analysis (Fig. 5a). Moreover, in the first pair of cauline leaf primordia of SPLr4 and SPLr5 plants, the expression of trichome repressors TRICHOMELESS1 (TCL1) and TRY was moderately increased (Fig. 5e). Unlike the trichome patterning phenotype, ectopic expression of miR156-resistant SPL4/5 genes failed to reveal any evident defects associated with other TTG1-dependent biological processes. The root-hair patterning was not affected and AtGL2 expression pattern was similar to Col-0 seedlings (Supplementary Fig. 6). Likewise, the ectopic expression of SPL4/5 had no effect on flavonoid biosynthesis (Supplementary Fig. 7). Taken together, these results provide conclusive evidence that the trichome patterning observed during overexpression of SPL4/5 genes coincides with the decreased levels of GL2 expression.

Discussion

Trichomes are differentiated epidermal cells that vary considerably between plants. Trichomes protect against insects, herbivores and UV radiation, control the temperature of leaf surface and prevent excessive transpiration (Johnson 1975; Balkunde et al. 2010). TTG1, a WD-repeat protein, forms a ternary complex together with R2R3-MYB and bHLH transcription factors, necessary for the expression of genes promoting trichome formation including GL2 (Payne et al. 2000; Zhang et al. 2003; Kirik et al. 2005). Despite trichome formation, the interacting transcription factors form with TTG1 distinct MBW complexes that act as regulatory hubs controlling root-hair morphogenesis or flavonoids biosynthesis. In ttg1 mutants, the formation of these MBW complexes is hindered, which prevents trichome formation causing the glabrous phenotype, promotes ectopic root-hair morphogenesis, and disrupts the biosynthesis of anthocyanins in vegetative tissues and proanthocyanidins in the seed coat mucilage resulting in the transparent testa phenotype (Galway et al. 1994; Walker et al. 1999).

Herein we demonstrate that similar to its Arabidopsis counterpart, the CcTTG1 polypeptide bears four WD40 repeat motifs, albeit the first repeat has five proline residues instead of eight. These distinctive differences together with the slightly variable N- and C-terminal domains most likely reflect the evolutionary distance between C. creticus and A. thaliana. It is therefore reasonable to assume that the structural differences modify the circular propeller core structure of CcTTG1 compared to AtTTG1 that partially affects the docking of heterogeneous Arabidopsis MYB and bHLH protein interaction partners in the trichome forming cells. Consequently, CcTTG1 fully complemented the root-hair and flavonoid phenotype of Arabidopsis ttg1 mutant, whereas resulted in hemi-complementation of the glabrous phenotypic defect. Hence, MBW complex integrity efficiently activates the expression of downstream regulatory genes that control root-hair formation and flavonoid biosynthesis contrary to the MBW complex formed in trichomes that is less active but still efficient. Identification of CcTTG1 and its functional role in Arabidopsis gives an interesting twist to comprehend the molecular switch that controls trichome morphogenesis coupled with the accumulation of secondary metabolites in medicinal plants.

While in the absence of TTG1 the ectopic expression of the heterogeneous MYB and bHLH factors activate the downstream regulatory pathways, the presence of TTG1 is crucial stabilizing the ternary transcriptional complex (Baudry et al. 2004; Li 2014; Xu et al. 2015). Our results show that miR156-targeted SPL transcription factors bind to TTG1 attenuating the transcriptional activity of MBW complex, thus causing trichome reduction. A subset of 11 SPLs is targeted by miR156 (Xing et al. 2010). The levels of miR156 gradually decrease with age, while in turn the expression of its target SPL genes increases. The miR156-regulated SPL genes control a suite of temporal changes during development including transition from the juvenile to adult phase (Wu and Poethig 2006; Wang et al. 2009; Wu et al. 2009; Yamaguchi et al. 2009; Wang and Wang 2015). The switch from vegetative growth to flowering is biphasic depending on gibberellin (Yu et al. 2012; Yamaguchi et al. 2014). Although gibberellin promotes termination of vegetative development, it inhibits flower formation. First, gibberellin levels increase and stimulate production of key flowering factors including LEAFY that stimulates the gibberellin degradation pathway. As gibberellin levels fall, the gibberellin-sensitive DELLA proteins accumulate and directly bind to SPL transcription factors promoting flowering through synergy between LEAFY and APETALA1. In addition, elevated levels of SPLs negatively regulate the TTG1-dependent biological processes of acropetal anthocyanin accumulation and trichome distribution on the inflorescence stems or floral organs, possibly by interfering with the integrity of a MBW complex (Yu et al. 2010; Gou et al. 2011).

Ectopic expression in A. thaliana of the miR156-resistant forms SPLr4/5, homologs of C. creticus SPLA/B, resulted in the reduction of trichomes on the adaxial surface of cauline leaves. Moreover, the trichome phenotype was correlated with downregulation of GL2 trichome inducer expression. As GL2 is the downstream molecular target of MBW, the SPLs most likely modify the complex by recruiting TTG1 and disrupt the TTG1/GL2 developmental signaling pathway that controls trichome formation. It will be of great interest to explore in future work whether the SPLs negatively regulate trichome formation through destabilization of a MBW transcriptional activation complex. Moreover, two conclusions can be drawn given that SPLr4 and SPLr5 interference exclusively affects trichome formation without disturbing root-hair patterning or flavonoids biosynthesis. Firstly, SPL4/5 are not attached on the ternary MBW complex to form a quaternary one to prevent complex binding to GL2 promoter as GL2, which is a negative regulator of root-hair formation, is normally expressed in seedlings. Secondly, upon root-hair morphogenesis and flavonoids biosynthesis MBW complex integrity is retained due to the heterogeneity of MYB and bHLH interacting partners. These heterogeneous factors protect the complex by SPL penetration.

The negative effect of SPLs on trichome formation has been recently enlightened (Xue et al. 2014). Our results show that SPL4/5 overexpression has a moderately positive effect on the expression of trichome TCL1 and TRY repressor genes, in agreement with previous results on SPL9 (Yu et al. 2010). However, constitutive expression of LOST MERISTEMS (LOMs) transcription factors belonging to the GRAS-domain containing family promoted trichome formation. Physical interaction between LOMs and SPLs underpinned the repression of SPL activity. Moreover, microarray data support the opposite expression pattern between especially SPL4 and GL2 (Supplementary Fig. 8). While TTG1 expression is constant during vegetative-to-reproductive phase transition or leaf expansion, the expression of GL2 progressively decreases contrary to elevated expression of SPLs. The data support the notion that the MBW complex is likely modified by SPLs having a major effect on trichome formation.

Nevertheless, the fate of the SPL-TTG1 interaction in C. creticus still remains vague. A tantalizing hypothesis is that the SPL-TTG1 hub first terminates trichome development and then initiates terpenes biosynthesis. A remarkable experimental evidence in this context is the progressive regulation of sesquiterpene biosynthesis by miR156-SPLs (Yu et al. 2015). In Pogostemon cablin, the age-regulated SPLs are the molecular link between developmental timing and sesquiterpene production. Although very little is known about C. creticus, it will be tempting to shed light on SPL-TTG1 function as a check point switch between developmental and biochemical processes.