Origin and function of stomata in the moss Physcomitrella patens

Abstract

Stomata are microscopic valves on plant surfaces that originated over 400 million years (Myr) ago and facilitated the greening of Earth's continents by permitting efficient shoot-atmosphere gas exchange and plant hydration¹. However, the core genetic machinery regulating stomatal development in non-vascular land plants is poorly understood^2-4 and their function has remained a matter of debate for a century⁵. Here, we show that genes encoding the two basic helix-loop-helix proteins PpSMF1 (SPEECH, MUTE and FAMA-like) and PpSCREAM1 (SCRM1) in the moss Physcomitrella patens are orthologous to transcriptional regulators of stomatal development in the flowering plant Arabidopsis thaliana and essential for stomata formation in moss. Targeted P. patens knockout mutants lacking either PpSMF1 or PpSCRM1 develop gametophytes indistinguishable from wild-type plants but mutant sporophytes lack stomata. Protein-protein interaction assays reveal heterodimerization between PpSMF1 and PpSCRM1, which, together with moss-angiosperm gene complementations⁶, suggests deep functional conservation of the heterodimeric SMF1 and SCRM1 unit is required to activate transcription for moss stomatal development, as in A. thaliana⁷. Moreover, stomata-less sporophytes of ΔPpSMF1 and ΔPpSCRM1 mutants exhibited delayed dehiscence, implying stomata might have promoted dehiscence in the first complex land-plant sporophytes.

Figure 1. The moss Physcomitrella patens genome encodes orthologues of the basic helix loop helix (bHLH) transcription factors regulating stomatal development in flowering plants

(a) Developing P. patens sporophyte, arrow indicating region of stomatal placement, and (b) excised sporophyte with stomata (orange/brown pores) forming a ring around the base. (c) Close-up of the sporophyte epidermis with single celled guard cells and central pores. (d and e) Bootstrapped Maximum Likelihood phylogenies of the SMF gene family comprising the FAMA, SPCH and MUTE subfamilies and the SCRM/ICE gene family in sequenced land plants. Internal node names in bold red indicate inferred subfamily ancestry. Internal nodes are coloured to indicate either duplication (red), speciation (green) or haplotype (blue) origin of the descendant nodes. Edge values represent bootstrap values. External node names comprise species abbreviations, original accession numbers of the protein sequences and accepted gene names of experimentally studied representatives in bold red. Species abbreviations in five-letter-code: Arabidopsis thaliana, Populus trichocarpa, Oryza sativa, Sorghum bicolor, Selaginella moellendorffii and Physcomitrella patens. (f, g and h) Relative expression of PpSMF1, PpSMF2 and PpSCRM1 in the developing sporophyte (grey bars) and protonema tissue (black bars) analysed by qRT-PCR. Error bars indicate standard error of the mean. Three replicates per tissue type were used. The scale bar in a = 100μm, in b = 100μm, in c = 25μm.

Figure 2. PpSMF1 and PpSCRM1 are required for stomatal development in the moss Physcomitrella patens

(a) Stacked UV fluorescence images (upper panel), scanning electron microscope images (middle panel) and bright field images (bottom panel) showing the spore capsule base and epidermal close-ups from P. patens wild-type, ΔPpSMF1, ΔPpSMF2 and ΔPpSCRM1 knock-out mutants, respectively. The top panel wild-type representative is from Villersexel K3 ecotype of P. patens, the middle panel wild-type representative is from the Gransden D12 ecotype and the bottom panel wild-type relates to the Gransden 2004 ecotype. There were no discernible differences between the sporophytes of the different background lines. For both of the ΔPpSCRM1 lines generated we observed one such instance of aborted stomata (see bottom right panel) in the 7 capsules of each line surveyed. (b) Number of stomata formed per sporophyte in two independent lines of each genotype versus wild-type controls. Error bars indicate one standard error of the mean. For ΔPpSMF1 and ΔPpSCRM1 and the corresponding wild-types, n = 7 capsules of each line were analysed. For ΔPpSMF2 and wild-type background, 5 capsules were surveyed. A One-way ANOVA was performed to test for differences between the wild-type and ΔPpSMF2 lines and no significant differences (denoted ns) were found. (c) RT-PCR to confirm loss of the respective transcript in each of the P. patens knock-out lines (top panel). A Rubisco (RBCS) control was run to verify the integrity of the produced cDNA (Bottom panel). For labelling purposes the wild-types Villersexel K3, Gransden D12 and Gransden 2004 are denoted Vx, GrD12 and Gr04. For PpSMF2 two bands were amplified in the control for which the smaller 239bp product represents the size expected for PpSMF2. Scale bars in a = 50 μm in the top and middle panels, in the bottom panel = 15 μm.

Figure 3. Bimolecular fluorescence complementation and Yeast 2-Hybrid assays demonstrating PpSMF1 and PpSCRM1 protein-protein interactions

(a) Representative bright-field, fluorescence and overlay/merged images of BiFC analysis showing pairwise combinations of bHLH constructs, each fused with a complementary, half-YFP molecule (nYFPn fusion and YFPc fusions, respectively). In the intact Allium cepa epidermis using bimolecular fluorescent complementation (BiFC), PpSMF1 and PpSCRM1 showed strong heterodimerization in the nuclei. Controls are described in Supp. Info. Fig 9. Scale = 100 μm. (b–d) Yeast two-hybrid analysis: (b) Growth on minimal medium. (c) Growth on stringent selection medium. Blue indicates reporter activation. (d) Key to patch plate assays is shown in (b) and (c).

Figure 4. Loss of PpSMF1 and the PpSCRM1 gene functions results in delayed dehiscence of spore capsules

Box-whisker plots of the percentages of ruptured sporophyte capsules in the wild-type, ΔPpSMF1 and ΔPpSCRM1 lines over a developmental time series experiment ranging from second and seventh week after induction of fertilization. Vertical lines within boxes mark the median. The boxes indicate the upper (75 %) and lower (25 %) quartiles. Whiskers indicate the ranges of the minimal and maximal values. Inset photograph depicts an open/ruptured spore capsule in the Gransden wild-type strain. Significance of differences between mutants and the wild type was tested using a binomial model with a nested error term correcting for repeated measurements in the combined data set, and for each genetic background independently, with consistent results. Significant (P < 0.05) deviations from the wild type are indicated by asterisks.