Direct roles of SPEECHLESS in the specification of stomatal self-renewing cells

See allHide authors and affiliations

Science  26 Sep 2014:
Vol. 345, Issue 6204, pp. 1605-1609
DOI: 10.1126/science.1256888

A complex network makes simple pores

Stomata, the pores found on the surface of plant leaves, form at intervals from stem cells. Development of stomata is controlled by the SPEECHLESS transcription factor. Lau et al. surveyed the genes that SPEECHLESS itself controls. Targets include genes involved in hormone signaling, control of cell proliferation, and the specification of asymmetric cell fates. Despite the apparent simplicity of a single pore, the genetic network that generates that pore is anything but simple.

Science, this issue p. 1605


Lineage-specific stem cells are critical for the production and maintenance of specific cell types and tissues in multicellular organisms. In Arabidopsis, the initiation and proliferation of stomatal lineage cells is controlled by the basic helix-loop-helix transcription factor SPEECHLESS (SPCH). SPCH-driven asymmetric and self-renewing divisions allow flexibility in stomatal production and overall organ growth. How SPCH directs stomatal lineage cell behaviors, however, is unclear. Here, we improved the chromatin immunoprecipitation (ChIP) assay and profiled the genome-wide targets of Arabidopsis SPCH in vivo. We found that SPCH controls key regulators of cell fate and asymmetric cell divisions and modulates responsiveness to peptide and phytohormone-mediated intercellular communication. Our results delineate the molecular pathways that regulate an essential adult stem cell lineage in plants.

In multicellular organisms, the need to generate and maintain diverse cell types and tissues is fulfilled by lineage-specific stem cells (1). These stem cell lineages, active postembryonically, produce a defined set of cell types. Although the origins of these lineage-specific stem cells during development are largely obscure, master transcription factors are implicated in their specification in both animals and plants (13). However, low expression levels and/or presence in a limited number of cells makes genome-wide study of these transcriptional regulators by standard chromatin immunoprecipitation (ChIP) assays, the most common technique for studying protein-DNA interactions, technically challenging.

Stomata are epidermal valves that mediate gas exchange between the plant and atmosphere. In Arabidopsis, stomatal guard cells are derived from an epidermal cell lineage (Fig. 1A) (4, 5). Two populations of stomatal precursor stem cells, meristemoid mother cells and meristemoids, have limited self-renewing properties and proliferate without the benefit of a stem cell niche (46). These stem cells are created through the postembryonic activity of SPEECHLESS (SPCH) in a subset of protodermal cells (7, 8). SPCH is a control point through which developmental, environmental, and phytohormone signals are integrated (4, 5). However, no targets of SPCH have been reported, and thus the sphere of its regulatory influence is unknown. Here, we develop a ChIP method optimized for rare developmental regulators and profile the genome-wide binding of SPCH in vivo. In combination with multiple transcriptional response data sets, our ChIP-sequencing (ChIP-seq) data indicate that SPCH programs an entire lineage by promoting fate transitions and asymmetric cell divisions (ACDs). SPCH also modulates the sensitivity of stomatal lineage cells to hormone and peptide/receptor–mediated signaling. Our results suggest how this lineage exhibits considerable autonomy while still coordinating with the overall organ development program.

Fig. 1 Chromatin immunoprecipitation (ChIP) optimized for cell-type–specific studies in vivo.

(A) Arabidopsis stomatal development scheme. SPCH controls the initiation and proliferation of the stem cell–like stomatal lineage precursors (pink and red cells). (B) Model for improving target enrichment in ChIPs through increasing experimental scale. (C and D) ChIPs at larger scales improve target enrichments. ChIP-qPCR assays of a SPCH variant on the TMM promoter performed at the indicated conditions (C). SPCH ChIP-seq profiles at TMM (D) generated from ChIPs at 4x and 16x [blue and red box in (C), respectively]. The y axis represents the enrichment values; note scales. Dashed box marks the SPCH-binding region.

Like many developmental regulators, SPCH expression is transient and limited to few cells (Fig. 1A). Standard ChIP assays on SPCH yielded only modest target enrichment (~4-fold, Fig. 1C, blue box), and thus we needed improved ChIP sensitivity for the detection of endogenously weak signals. We hypothesized that if background signals in a ChIP assay could be kept low, increasing the experimental scale would lead to a disproportional increase in signals from targets (true signal) over background (Fig. 1B). Therefore, performing ChIP at a large scale may achieve high target enrichment even for low-abundance proteins. We tested this hypothesis with ChIPs at three different scales on a spch mutant line bearing a complementing, Myc-tagged SPCH variant driven by its native promoter (SPCHpro:SPCH2-4A-MYC) (fig. S1). The scales represented 4, 8, and 16 times (or 6, 12, and 24 g) the input materials used in a typical Arabidopsis ChIP experiment. ChIP-qPCR (quantitative polymerase chain reaction) assays of SPCH on the promoter of TOO MANY MOUTHS (TMM) showed that scale increase improves target enrichment up to 600-fold at 16x (or a >30-fold increase in enrichment with a 4-fold scale increase) (Fig. 1C, three rightmost columns). Thus, weak signals can be enhanced by maximizing input. We termed this method Maximized Objects for Better Enrichment (MOBE)–ChIP.

To profile genome-wide binding events of SPCH, we performed and pooled six MOBE-ChIPs on SPCHpro:SPCH2-4A-MYC and on a wild-type (WT) control for high-throughput sequencing (scale, 16x; total, 144 grams per genotype) (Fig. 1C, red box, and fig. S2B). For comparison, standard ChIP-seq was also included [pooled from nine independent ChIPs on SPCHpro:SPCH2-4A-YFP (yellow fluorescent protein) and nucGFP (green fluorescent protein) at 4x] (Fig. 1C, blue box, and fig. S2A). MOBE-ChIP-seq confirmed the ChIP-qPCR results at the TMM promoter, revealing a single peak with an enrichment score of 178 [–log10(q value), 1.2 × 106]; the corresponding peak from our 4x run had a score of 1.2 [–log10(q-value), 5.7] (Fig. 1D). Low background signal is also a genome-wide trend. Using the peak-calling algorithm ChIP-seq Analysis in R (CSAR) (9), we detected peaks with an enrichment score as low as 1.62 at a false discovery rate (FDR) of 1 × 10−6, in contrast to other studies whose peaks above threshold scores of 1.85 and 79.6 were detected at FDRs of 0.01 and 0.001, respectively (table S1) (10, 11). The ability to identify these low-coverage peaks is indicative of the power of signal enrichment. Thus, through MOBE-ChIP-seq, we generated a comprehensive in vivo genome-wide binding map of SPCH.

Using two complementary peak-calling pipelines, we identified 8327 SPCH-bound regions (tables S2 and S3). Seventy percent of the SPCH binding peaks are associated with gene promoters, mostly within 500 base pairs upstream of the transcriptional start site (Fig. 2A and fig. S3). De novo discovery of enriched motifs in the binding peaks identified CDCGTG as the top-scoring motif; this variant of the E-box (CACGTG), to which basic helix-loop-helix (bHLH) proteins typically bind, is enriched at the summit of the SPCH peaks (Fig. 2B and fig. S4).

Fig. 2 Genome-wide analysis of SPCH-binding targets reveals direct roles in lineage specification and asymmetric cell divisions.

(A) Distribution of SPCH-binding peaks relative to gene structure. (B) Top-scoring motif (E value, 7.5 × 10−365) and the position of its three variants in SPCH-binding peaks. (C and D) Percentage of SPCH targets among differentially expressed genes in RNA-seq analysis of inducible SPCH1-4A (C), and microarray analysis of meristemoid-enriched (scrmD mute) or -depleted (spch) mutants (13) (D). P values are calculated by Fisher’s exact test. Dashed line indicates percentage by chance. (E) Select enriched GO terms of SPCH target genes. (F and G) SPCH binds and activates key stomatal regulators. ChIP-seq profiles of select stomatal genes (F). The y axis represents peak score (CSAR), and arrows indicate gene orientation and transcriptional start sites. Gene expression changes upon induction of SPCH in RNA-seq analysis (G). (H) Importance of SPCH-binding motif (red) on ICE1 expression. Mutation of two motifs (purple, mICE1pro) within the SPCH-binding peak (blue shading) abrogates ICE1 expression (yellow). Confocal images of 4-dpg abaxial cotyledons have ML1pro:mCherry-RCI2A–marked cell outlines (purple). Scale bar, 40 μm.

To focus on loci most likely to respond transcriptionally to SPCH binding, we generated a “high-confidence” subset of peaks that were nonintergenic with enrichment scores ≥10 (table S2). Among the high-confidence targets, Gene Ontology (GO) terms for genes involved in regulation of transcription, signaling, response to stimulus, and regulation of hormone levels were significantly enriched (Fig. 2E, fig. S5, and table S4). This suggests that in the initiation of the stomatal lineage, SPCH could act as a mediator of environmental and hormone inputs that are translated into further downstream transcriptional and signaling networks. The enrichment of the GO term “protein targeting to membrane” is interesting given the membrane-associated polarization of stomatal lineage proteins BASL and POLAR during asymmetric divisions (12, 13).

To correlate SPCH binding with transcriptional responses on a genome-wide scale, we compared the high-confidence SPCH targets to data sets representing genes expressed in response to SPCH induction (fig. S6 and table S5) and those enriched for genes preferentially expressed in the stomatal lineage (13) (fig. S7). Significant enrichment of the SPCH targets was found among genes both up- and down-regulated in response to SPCH induction (27 and 20%, respectively) (Fig. 2C) and in plants with excess or no meristemoids (31 and 12%) (Fig. 2D). By chance, SPCH would be predicted to bind to ~4.5% of genes in the data sets (1517 targets out of 33,602 Arabidopsis genes). Overall, these comparisons indicate that nearly a quarter (23%) of the SPCH targets are differentially expressed (table S6) and SPCH may activate or repress a large number of its targets directly.

Meristemoid-active stomatal regulators are among the direct SPCH targets (Fig. 2F, fig. S8A, fig. S9, and table S7). SPCH binds to its own promoter and to the promoter of genes that encode its heterodimeric bHLH partners, ICE1/SCRM and SCRM2 (14), and induces their expression (Fig. 2, F and G, and fig. S8A). Although initial activation of SPCH may not require SPCH protein (fig. S10), this positive feedback loop may be an essential part of a bistable switch that converts the initially low and stochastic expression of SPCH into an active SPCH-SCRM heterodimer to drive stomatal lineage fates. SPCH also binds and activates expression of genes encoding the secreted ligand EPF2, the receptor TMM, and the ERECTA family of receptor-like kinases (Fig. 2, F and G, and fig. S8A), all of which enforce proper patterning by restricting proliferation in the early stomatal lineage and act upstream of kinases that target SPCH for posttranslational down-regulation (4, 1517). Further, SPCH binds to the promoters and activates expression of BASL and POLAR, which encode polarly localized proteins, suggesting a direct role in regulating the ACD process (Fig. 2, F and G). SPCH binding is not associated with genes that encode a later expressed stomatal lineage EPF (EPF1), an EPF not expressed in the stomatal lineage (CHALLAH) or the broadly expressed mitogen-activated protein kinase kinase kinase (MAPKKK) YODA (fig. S8B). Taken together, our ChIP-seq and RNA-seq data reveal the broad and direct roles of SPCH in sustaining a SPCH transcriptional cascade, establishing meristemoid identity and mediating ACDs.

The CDCGTG motif appears in the SPCH-bound regions of stomatal targets like ICE1, TMM, and ERL2. In ICE1, SPCH binds in two peaks centered on the locations of two CDCGTG motifs (fig. S11). To test the role of SPCH-binding motifs in ICE1 expression, we generated a reporter bearing point mutations in the two peak-associated motifs (Fig. 2H) to compare to the WT reporter. Consistent with previous reports (14), expression of the WT promoter reporter (ICE1pro) was observed in the stomatal lineage; however, the mutant reporter (mICE1pro) was nearly undetectable (Fig. 2H and fig. S12). Similar dependence was seen with the SPCH–up-regulated gene At2g34510, which contains CDCGTG within a strong intronic SPCH binding peak. Deletion of the SPCH binding region abrogated early stomatal lineage-specific expression (fig. S13).

An intriguing meristemoid behavior is the ability to self-renew through ACDs. Beyond requirements for SPCH activity and the polarly localized BASL (Fig. 2F), however, little is known about the ACD process. Among SPCH targets, ARK3/AtKINUa (Fig. 3A) caught our attention as it encodes a plant-specific kinesin in the preprophase band (18). In plants, the preprophase band marks the future division plane (19). Confocal analysis of ARK3pro:ARK3-YFP showed localization to preprophase bands of asymmetrically dividing meristemoids (Fig. 3B). Coexpression with SPCHpro:SPCH-CFP indicated that SPCH precedes ARK3, consistent with SPCH activating ARK3 expression (Fig. 3, C to E). To ascertain its function in the stomatal lineage, we reduced ARK3 expression by driving an artificial microRNA against it with the SPCH promoter (SPCHpro:amiR-ark3). In the cotyledon epidermis of amiR-ark3–expressing plants, we observed clusters of meristemoid-like small cells at 4 days postgermination (dpg) that developed into clusters of stomata at 11 days (Fig. 3, G and I, brackets). These small cell clusters, which displayed diminished physical asymmetry, appear to arise from misplaced but complete division planes. Notably, cell wall stubs or other evidence of incomplete divisions were not observed. The amiR-ark3 phenotypes resembled those associated with basl mutants (12) and are hallmarks of loss of ACD capacity. Thus, ARK3 appears to be a new player essential for ACD, possibly through regulating preprophase band placement, and establishes a direct link between SPCH and the ACD machinery.

Fig. 3 SPCH regulates ACD through a preprophase band-localized kinesin.

(A) SPCH ChIP-seq profile of ARK3/KINUa. (B to E) Expression of ARK3pro:ARK3-YFP (yellow) (B) and its coexpression with SPCHpro:SPCH-CFP (blue) [(C) to (E)] before (C), during (B) and (D), and after (E) a stomatal ACD. Arrowheads indicate the preprophase band. (F to I) ACD defects in SPCHpro:amiR-ark3 [(G) and (I)], compared with Col [(F) and (H)]. Brackets mark clusters of small cells (G) or guard cells (I). Confocal images are of 3- [(B) to (E)], 4- [(F) and (G)] and 11-day [(H) and (I)] abaxial cotyledons with ML1pro:mCherry-RCI2A–marked cell outlines. To display the cells in (E) in the same orientation as panels (C) and (D), we rotated the image in (E) and filled in the nonimaged space in the upper right corner with a black triangle (dashed line). Scale bars, 10 μm [(C) to (E)], 50 μm [(F) to (I)].

SPCH initiates a lineage with autonomous control over cell division and fate determination. Nonetheless, the stomatal lineage is also coordinated with developmental programs operating across tissues and organs. Phytohormones play critical roles in coordinating development, and recent reports indicate that auxin, brassinosteroid (BR), and abscisic acid regulate stomatal development (2023). BR controls stomatal development through phosphorylation of YODA and SPCH by its central glycogen synthase kinase 3–like kinase, BIN2 (Fig. 4F) (21, 22). Among SPCH target categories, BR biosynthetic and response genes show significant enrichment (fig. S14). Notably, SPCH binds to the promoters of BIN2 and CPD, which encodes an essential enzyme for BR biosynthesis (Fig. 4, A and F), and the absence of CPD results in stomatal overproduction (24). We tested the effect of SPCH on the expression of BR genes by reverse transcription qPCR (RT-qPCR) in the meristemoid-enriched line SPCHpro:SPCH2-4A-YFP (Fig. 4B). Consistent with inhibition of BR signaling, we found that BIN2 expression is elevated, whereas CPD is repressed (Fig. 4B). Supporting BIN2’s role in promoting SPCH function, stomatal lineage-specific expression of BIN2-1 led to small cell clusters in cotyledons, similar to those observed upon SPCH overexpression (Fig. 4C). Thus, our results suggest the presence of feedback by SPCH counteracting BR signaling (Fig. 4F). SPCH also binds to genes encoding the BR signaling effectors, the BZR1 family of transcription factors (BZR1, BES1/BZR2, and BEH1 to BEH4), and BIM2, the putative dimeric partner of BES1 (2527). BEH1 to BEH4 and BIM2 were up-regulated in the meristemoid-enriched mutant, and BIM2 exhibited stomatal lineage-specific expression (Fig. 4, B and D). Epidermal expression of bes1-D (26) correlates with an increase in stomatal density, whereas a bes1 RNA interference (RNAi) knockdown line (27) exhibited a trend toward lower stomatal density (Fig. 4E). This role in promoting stomatal development may be explained through the known repression of the BR biosynthetic genes by the BZR1 family (28). Thus, SPCH-mediated induction of BIN2 and repression of CPD (either directly or indirectly through the BZR1 family) leads to higher BIN2 activity and derepression of SPCH, promoting accumulation of SPCH in active meristemoids (Fig. 4F). Overall, this feedback mechanism by SPCH would serve to reinforce differences between SPCH-expressing meristemoids and nonexpressing neighbors, which may be important for local patterning and coordinating the lineage with overall BR-mediated growth controls.

Fig. 4 Feedback regulation of brassinosteroid biosynthesis and signaling by SPCH.

(A) ChIP-seq profiles of select brassinosteroid (BR) pathway genes (labeled as in Fig. 2F). (B) RT-qPCR analysis of BR genes in 4-dpg SPCHpro:SPCH2-4A-YFP and Col seedlings. Values are means ± SEM. (C and D) Confocal images of 3-dpg adaxial cotyledons with propidium iodide-stained cell outlines (purple). Stomatal lineage-specific expression of hyperactive BIN2 (yellow) induces lineage proliferation (bracket) (C). Stomatal lineage expression pattern of BIM2pro:YFP-YFP (yellow) (D). (E) Alteration of stomatal density in gain-of-function BES1pro:bes1-D and bes1-RNAi knockdown. *P < 0.05, ***P < 0.001 (Wilcoxon rank sum test). (F) Model of SPCH-BR pathway interactions. SPCH, a target of BR signaling, feeds back (positively, red arrows, or negatively, red T-bar) upon transcription of multiple pathway members.

Here, we revealed the broad influence of SPCH in stomatal lineage specification through MOBE-ChIP. This technique, which is based on a simple scale increase, could be widely applicable in other tissues or organisms to obtain high-quality binding information about cell-type–specific regulators. The large number of SPCH-binding regions reported here is reminiscent of the behavior of the bHLH transcription factor MyoD, a master regulator of mammalian myogenesis, which associates with more than 30,000 regions in the human genome and is responsible for resetting global transcriptional and epigenetic states during development (29). Additional experiments are needed to establish definitively how often and by what mechanisms SPCH binding alters gene expression. However, our data that hundreds of genes, including those mediating abiotic and hormone responses, are directly regulated by SPCH supports previous functional studies (20, 22) that place SPCH in a critical position to integrate physiological and environmental information into a developmental program that optimizes leaf properties (stomatal density and size) for prevailing environments.

Supplementary Materials

Materials and Methods

Figs. S1 to S15

Tables S1 to S8

References (3062)

References and Notes

  1. Acknowledgments: We thank Z.-Y. Wang (Carnegie) for the antibody to YFP and the bin2-1 allele; Y. Yin (Iowa State University) for the BES1pro:bes1-D-GFP construct; J. Chory (SALK) for the bes1 RNAi line; and members of our laboratory for critical comments. Funding for this work was provided by National Institutes of Health (NIH) 1R01GM086632. O.S.L. was a Croucher Fellow, K.A.D. was supported by Cellular and Molecular Biology Training Program NIH5T32GM007276 and by a National Science Foundation graduate research fellowship, and J.A. was supported by the DAAD. D.C.B. is a Gordon and Betty Moore Foundation Investigator of the Howard Hughes Medical Institute. The ChIP-seq and RNA-seq data in this study can be found in National Center for Biotechnology Information’s Gene Expression Omnibus repository ( as GSE57954. The supplementary materials contain additional data.
View Abstract

Stay Connected to Science

Navigate This Article