A systems approach reveals regulatory circuitry for Arabidopsis trichome initiation by the GL3 and GL1 selectors

Abstract

Position-dependent cell fate determination and pattern formation are unique aspects of the development of plant structures. The establishment of single-celled leaf hairs (trichomes) from pluripotent epidermal (protodermal) cells in Arabidopsis provides a powerful system to determine the gene regulatory networks involved in cell fate determination. To obtain a holistic view of the regulatory events associated with the differentiation of Arabidopsis epidermal cells into trichomes, we combined expression and genome-wide location analyses (ChIP-chip) on the trichome developmental selectors GLABRA3 (GL3) and GLABRA1 (GL1), encoding basic helix-loop-helix (bHLH) and MYB transcription factors, respectively. Meta-analysis was used to integrate genome-wide expression results contrasting wild type and gl3 or gl1 mutants with changes in gene expression over time using inducible versions of GL3 and GL1. This resulted in the identification of a minimal set of genes associated with the differentiation of epidermal cells into trichomes. ChIP-chip experiments, complemented by the targeted examination of factors known to participate in trichome initiation or patterning, identified about 20 novel GL3/GL1 direct targets. In addition to genes involved in the control of gene expression, such as the transcription factors SCL8 and MYC1, we identified SIM (SIAMESE), encoding a cyclin-dependent kinase inhibitor, and RBR1 (RETINOBLASTOMA RELATED1), corresponding to a negative regulator of the cell cycle transcription factor E2F, as GL3/GL1 immediate targets, directly implicating these trichome regulators in the control of the endocycle. The expression of many of the identified GL3/GL1 direct targets was specific to very early stages of trichome initiation, suggesting that they participate in some of the earliest known processes associated with protodermal cell differentiation. By combining this knowledge with the analysis of genes associated with trichome formation, our results reveal the architecture of the top tiers of the hierarchical structure of the regulatory network involved in epidermal cell differentiation and trichome formation.

Figure 1. Gene regulatory network controlling trichome initiation.

(A) BioTapestry wire-graph representation of the regulatory circuitry involved in trichome initiation. GL3 and GL1 bind and activate the transcription of CPC, ETC1, TTG2 and GL2 . CPC (and perhaps ETC1) can move to adjacent cells , and through interactions with GL3, compete for GL1/GL3 complex formation. Under these circumstances, GL3 is predicted to bind and negatively modulate its own expression . Filled lines indicate direct regulation whereas dotted lines indicate indirect regulation or cases for which the regulatory mechanisms are unknown. Colored lines show active network connections in trichome cells and grey lines indicate network connections more likely to become active in epidermal cells adjacent to trichomes. (B) Complementation of the trichome phenotype of gl3 egl3 and gl1 mutants by pGL3::GL3-YFP and pGL1::GL1-YFP-MYC, respectively. Bar = 1 mm.

Figure 2. Genome-wide identification of GL1 and GL3 associated regions.

Signal enrichment location displayed using IGB of GL1 (orange) and GL3 (blue) associated regions as well as signals obtained from the negative ChIP-chip controls using IgG and αGFP on Ler wild type plants (gray). The gene annotation, shown in green, was obtained from TAIR (http://www.arabidopsis.org/index.jsp). Large boxes correspond to exons; small boxes to untranslated regions and lines to introns. Gene orientations are indicated on the left side of the picture. The black boxes indicate the region utilized for ChIP-PCR verification.

Figure 3. Validation of GL1 and GL3 direct target genes.

(A) Semi-quantitative PCR of ChIP experiments performed on gl3 egl3 pGL3::GL3-YFP, gl1 pGL1::GL1-YFP-MYC or gl1 pGL3::GL3-YFP plants of genes selected by ChIP-chip. Serial 4-fold dilutions used for PCR are represented by the black triangles. PCR were performed on regions 500 bp upstream from the TSS, except for Act 2/7, for which PCR was performed on the 5′UTR. (B) Semi-quantitative PCR of ChIP experiments performed on gl3 egl3 pGL3::GL3-YFP, gl1 pGL1::GL1-YFP-MYC or gl1 pGL3::GL3-YFP plants of trichome genes (Table S1) identified as GL3 and GL1 targets (Figure S5). Serial 4-fold dilutions used by PCR are represented by the black triangles. PCR were performed on regions 500 bp upstream from the TSS, except for Act 2/7, for which PCR was performed on the 5′UTR.

Figure 4. Genome-wide expression changes induces by GL3 or GL1.

(A) Venn diagram representation of the genome-wide alterations in mRNA accumulation in gl1 pGL1::GL1-GR plants induced 4 (GL1GR 4 hours) or 24 hrs with DEX (GL1GR 24 hours), gl3 egl3 pGL3::GL3-GR plants induced 4 (GL1GR 4 hours) or 24 hrs with DEX (GL1GR 24 hours), differentially expressed genes between wild type and the gl1 mutant (WT/gl1) or between wild type and the gl3 egl3 mutant (WT/gl3 egl3). Numbers inside the diagrams represent the total umber of genes affected at the indicated P value. Genes estimated to be affected by GL1 and GL3, forming a minimal “Trichome genes” set, were deduced using a meta-analysis statistical approach (see Materials and Methods). (B) Venn diagrams indicate the number of genes that were affected by two or more of the contrasts described in (A). (C) Overlap between gene expression and ChIP-chip analyses. The AGI numbers for the genes in each intersection are indicated.

Figure 5. Expression of a select set of GL3 and GL1 direct targets after DEX induction of GL3-GR or GL1-GR.

qRT-PCR experiments performed on mRNA obtained from gl1 pGL1::GL1-GR or gl3 egl3 pGL3::GL3-GR seedlings DEX- or Mock-treated for various times (15 min, 1 hr, 4 hrs, 24 hrs, 3 days and 7 days). It should be noticed that the y-axis is drawn in linear-scale whereas the x-axis is in log-scale. Black circles correspond to expression in gl1 pGL1::GL1-GR plants and white circles to expression in gl3 egl3 pGL3::GL3-GR plants. For each time-point, the ratio between the DEX and Mock treatment was calculated. For clarity purposes, the pre-induction condition (time = 0) is shown, and corresponds to no induction (ratio = 1). Induction times, corresponding to early stages of trichome development (0–24 hrs), are indicated shaded gray. For each time point, biological triplicates were collected and the error bar indicates the standard error. Thick lines correspond to gene expression profiles for which ANOVA statistical analyses suggested a significant difference (P<0.05). Thin lines indicate cases for which statistical significance could not be supported to biological variation between the triplicates.

Figure 6. Co-expression of the minimal set of 513 trichome genes with GL2, TTG2, SCL8 and MYC1.

(A) The heat map shows a distribution of PCC values after hierarchical clustering. Red and blue colors indicate positive and negative PCC values, respectively. Clusters I–V, discussed in the text, are indicated with red bars on the right side of the heat map. (B–D) Scatter plots comparing the PCC valued of all genes in the trichome set with (B) TTG2 and GL2, (C) TTG2 and SCL8, and (D) MYC1 and SCL8.

Figure 7. Regulatory motifs controlled by GL1 and GL3.

Circle and squares indicate the nodes of the directed graph representing the regulatory network, and correspond to transcription factors, and enzymes or structure proteins, respectively. Filled lines indicate the edges of the regulatory network, and correspond to direct interactions between transcription factors and the corresponding target genes. The color of the lines originating in the GL3/GL1 first tier regulators indicate whether they correspond to direct targets of GL3 and GL1 (purple), GL3 alone (or perhaps with MYB23, black) or GL1 alone (black). Lines with arrowheads indicate activation by GL3/GL1 whereas lines without arrowheads mean gene regulation is uncertain. The dotted lines derived from the second tier regulators SCL8, GL2, TTG2 and MYC1 correspond to genes from the minimal trichome set most likely downstream of each of the regulators. Boxes shown in the “Trichome genes” tier indicate the corresponding functional classifications. Numbers in bold represent the gene functions classified by FunCat (http://mips.gsf.de/proj/funcatDB/search_main_frame.html). All the genes shown here are described in more detail as part of Table S1.