Molecular Framework of a Regulatory Circuit Initiating Two-Dimensional Spatial Patterning of Stomatal Lineage

Abstract

Stomata, valves on the plant epidermis, are critical for plant growth and survival, and the presence of stomata impacts the global water and carbon cycle. Although transcription factors and cell-cell signaling components regulating stomatal development have been identified, it remains unclear as to how their regulatory interactions are translated into two-dimensional patterns of stomatal initial cells. Using molecular genetics, imaging, and mathematical simulation, we report a regulatory circuit that initiates the stomatal cell-lineage. The circuit includes a positive feedback loop constituting self-activation of SCREAMs that requires SPEECHLESS. This transcription factor module directly binds to the promoters and activates a secreted signal, EPIDERMAL PATTERNING FACTOR2, and the receptor modifier TOO MANY MOUTHS, while the receptor ERECTA lies outside of this module. This in turn inhibits SPCH, and hence SCRMs, thus constituting a negative feedback loop. Our mathematical model accurately predicts all known stomatal phenotypes with the inclusion of two additional components to the circuit: an EPF2-independent negative-feedback loop and a signal that lies outside of the SPCH•SCRM module. Our work reveals the intricate molecular framework governing self-organizing two-dimensional patterning in the plant epidermis.

Fig 1. Molecular framework of the SPCH•SCRM positive feedback for stomatal-lineage specification.

(A) SPCH and SCRM are mutually required for initiating the entry asymmetric division of stomatal cell lineages. Shown are false-colored confocal microscopy images of abaxial rosette leaf epidermis from 10–12 day-old seedlings. Wild type (left) epidermis gives rise to stomatal lineage cells: Cyan, early meristemoids; light green, late meristemoids and guard mother cells; green, immature and mature guard cells; white, stomatal-lineage ground cells or pavement cells. spch or scrm scrm2 mutant epidermis is solely composed of pavement cells (white). Scale bars, 20 μm. (B) Expression heat map of SPCH, SCRM, and SCMR2 from a microarray study [13] in wild type and mutants enriched in specific epidermal cells: scrm-D mute (M: meristemoids); scrm-D (GC: stomatal guard cells); spch (PC: pavement cells). (C) Promoter GFP reporter expression patterns of SPCHpro::nucGFP (top), SCRMpro::nucGFP (middle), and SCRM2pro::nucGFP (bottom) in early protoderm of 11-day-old wild-type (left), spch (middle), and scrm scrm2 (right) seedlings. SPCH does not require itself or SCRMs for its own promoter activity. In contrast, SCRMs require SPCH and themselves, indicating that SCRMs form a positive feedback loop essential for pattern formation. Scale bar, 20 μm. (D, E) ChIP assays on SCRM (D) and SCRM2 (E) promoter regions using anti-GFP antibody on control wild type or transgenic seedlings expressing functional SPCH-GFP in scrm-D, GFP-SCRM, GFP-scrm-D, or GFP-SCRM2. Each amplicon is indicated in a red letter. Shown as a graph are mean ± SEM of fold enrichment over wild-type Col from three biological replicates. Line, intergenic region or intron; arrow, transcription start site; filled rectangle, coding region. (F) Transactivation assays in N. benthamiana. Reporter luciferase expression driven by SCRM promoter is strongly induced when both SPCH and SCRM are present. Reporter firefly luciferase activity was normalized against constitutively expressed Renilla luciferase, and the values are normalized against controls without effector proteins. Bars indicate means of three biological replicates; error bars, S.E.M.

Fig 2. Molecular framework of the negative-feedback loop between SPCH•SCRM and EPF2 for stomatal-lineage specification.

(A) Shown are confocal images of abaxial protoderm of rosette leaf primordia of 10-11-day-old seedlings expressing EPF2pro::erGFP in wild type (left), spch (middle), and scrm scrm2 (right). No EPF2 promoter activity is detected in the absence of SPCH or SCRMs. Scale bars, 20 μm. (B) ChIP assays on EPF2 promoter region using anti-GFP antibody on control Col or transgenic seedlings expressing functional SPCH-GFP in scrm-D, GFP-SCRM, GFP-scrm-D, or GFP-SCRM2. Each amplicon is indicated in a red letter. Mean ± SEM of fold enrichment over wild-type Col from three biological replicates are shown. ACT2 serves a control. Line, intergenic region or intron; arrow, transcription start site; filled rectangle, coding region. (C) Transactivation dual luciferase reporter assays in N. benthamiana. Strong EPF2 reporter expression is detected when both SPCH and SCRM are present. Bars indicate means of biological triplicates; error bars, S.E.M. (D) Effects of bioactive recombinant MEPF2 peptide application on promoter activity and protein accumulation of SPCH and SCRMs. MEPF2 application has no effect on SPCH promoter activity (SPCHpro::nucGFP) despite the fact that no-stomatal cell linages are initiated (top left). In contrast, MEPF2 application results in loss of GFP signals in SPCHpro::SPCH-GFP (top right), SCRMpro::nucGFP (middle left), SCRMpro::GFP-SCRM (middle right), and SCRM2pro::GFP-SCRM2 (bottom left). GFP-scrm-D protein is insensitive to MEPF2 application (bottom right). Six-day-old cotyledons are imaged under the same magnification. Scale bar, 20 μm. (E) Abaxial epidermis from 5-6-day-old seedling rosette leaf primordia expressing SPCHpro::SPCH-GFP in wild-type (left) or scrm-D (right) background, showing that more protodermal cells accumulate SPCH-GFP protein (green) in scrm-D. Scale bar, 20 μm.

Fig 3. Differential regulation of receptors by SPCH•SCRM module.

(A) Expression/accumulation patterns of functional ERECTA-YFP (top) and TMM-YFP (bottom) in protoderm from first rosette leaf primordia of 5-8-day-old erecta tmm (left), spch (middle), and scrm scrm2 (right) seedlings. No TMM-YFP signal can be detected in the absence of SPCH or SCRMs. Scale bars, 150 μm. (B) qRT-PCR analysis of EPF2, TMM, ERECTA, and STOMAGEN transcripts levels from five-day-old spch (pavement cells only), scrm scrm2 (pavement cells only), scrm-D mute (meristemoid enriched), and scrm-D (stomata enriched) seedlings compared to wild-type. Both EPF2 and TMM transcripts are highly enriched in meristemod-enriched population (scrm-D mute) while undetectable in spch or scrm scrm2. In contrast, ERECTA and STOMAGEN show no such trends. (C) Higher magnifications of protoderm expressing ERECTA-YFP levels (top left and middle) and TMM-YFP (top right) co-stained with PI (middle) to highlight cell periphery. Presented at the bottom are line scan analyses of each panel corresponding to lines indicated in the confocal images. Cell boundaries between a stomatal-lineage cell and an adjacent epidermal cell (asterisks), between a meristemoid and an SLGC (x), between a GC and adjacent epidermal cells (v), and between two paired GCs (+) are indicated. ERECTA-YFP levels are reduced in stomatal precursors and not detectable in GCs, while TMM-YFP levels are stomatal-lineage-specific (D) ChIP assays on TMM promoter region using anti-GFP antibody on control Col-0 or transgenic seedlings expressing functional SPCH-GFP in scrm-D, GFP-SCRM, GFP-scrm-D, or GFP-SCRM2. Each amplicon is indicated by a letter. Shown are the means ± SEM of fold enrichment over wild type Col from three biological replicates. Line, intergenic region or intron; arrow, transcription start site; filled rectangle, coding region. (E) Transactivation dual luciferase reporter assays using N. benthamiana. TMM expression is upregulated when both SPCH and SCRM are present. Bars indicate means of triplicate; error bars, S.E.M.

Fig 4. Regulatory circuit modeling two-dimensional patterns of stomatal initial cells.

(A) Diagram outlining the regulatory circuit used for modeling. (Top) Example of two adjacent protodermal cells undergoing fate determination process. Arrow designates activation and T-bar designates inhibition. Concentrations of each components are abbreviated as the following: u ₁, SPCH; u ₂, SCRM; u ₃, SPCH•SCRM heterodimer; v ₁, EPF2; w, TMM; v ₂, EPF2-independent hypothetical component, most likely BR pathway; m, strength of MAPK cascade-mediated inhibition. S, a component that competes for receptor pools, most likely Stomagen. The site of bikinin action is also indicated. Initially, all cells possess and operate identical regulatory circuit. Stochastic noise will be amplified in such a way that a cell expressing more activator will self-activate its stomatal-lineage character (light blue), while the neighboring cell will lose stomatal-lineage character (white). The regulatory relationships that are not experimentally verified are in green. It is not known which protodermal cells produce BR, or whether BR acts in neighboring cells. (Bottom) Simplified diagram showing the putative range of inhibitor action. (B) Spatial patterns of each component in wild-type and each mutant background simulated in silico based on the mathematical models. Each square represents a sheet of protoderm with 400 cells (each cell represented by a hexagon). White cells indicate no expression/accumulation of a given component, while dark-blue cells express/accumulate high amounts.

Fig 5. Bikinin treatment represses GFP-SCRM accumulation independent of EPF2-and ERECTA-family.

The bikinin-sensitive, EPF2-independent pathway may constitute the second feedback loop predicted by our modeling. (A-D) wild-type seedlings carrying SCRM::GFP-SCRM mock treated (A, B) or treated with 30 μM bikinin (C, D). (E-H) epf2 seedlings carrying SCRM::GFP-SCRM mock treated (E, F) or treated with 30 μM bikinin (G, H). (I-L) wild-type seedlings carrying SCRM::GFP-scrm-D mock treated (I, J) or treated with 30 μM bikinin (K, L). (M-P) er erl1 erl2 seedling carrying SCRM::GFP-SCRM mock treated (M, N) or treated with 30 μM bikinin (O, P). (Q-T) wild-type seedlings carrying AtML1::nucGFP mock treated (Q, R) or treated with 30 μM bikinin (S, T). Shown are 5-day-old cotyledon epidermis (A, C, E, G, I, K, M, O, Q, S) and protoderm of primary leaf primordial (B, D, F, H, J, L, N, P, R, T) after 2-day exposure to bikinin. Under bikinin treatment, GFP-SCRM signal disappears from stomatal precursors (arrowheads), while GFP-SCRM in stomata (asterisks) is still detected. Reduction of the GFP-SCRM signal was evident ~ 8 hrs after bikinin treatment and the signals became undetectable 2 days after treatment. For cotyledons, cell periphery was highlighted by propidium iodide; scale bars, 50 μm. For primary leaves, scale bars, 100 μm.