Differing Patterns of Stomatal Development in Terrestrial and Amphibious Species of Callitriche.

To identify unique temporal patterns of stomatal development in C. palustris, we also investigated the time course of stomatal development in three other Callitriche species with different lifestyles. In addition to C. palustris, we analyzed a second amphibious species from a different clade within the genus, Callitriche stagnalis. We also examined two terrestrial species, Callitriche japonica and Callitriche terrestris. C. japonica belongs to the ancestral clade, while C. terrestris is a more derived species that is phylogenetically closest to C. palustris among the three newly analyzed species (26, 27) (Fig. 2A). These differences among species allowed us to examine lifestyle-associated traits with minimal phylogenetic constraints.

Fig. 2.

Species in the ecologically diverse genus Callitriche have diverse stomatal development patterns. (A) Phylogenetic tree of the species included in this investigation, based on previous studies (26, 27). The morphologies of single mature leaves of each species are shown on one side. (Scale bar: 1 mm.) (B) Quantification of the stomatal index throughout the developmental period. Each dot corresponds to one leaf primordium. (C–E) Confocal images of Calcofluor-stained developing leaf primordia of C. palustris (C and D, adaxial epidermis) and C. japonica (E, abaxial epidermis). Primordium lengths: 495 μm (C), 1,198 μm (D), and 889 μm (E). (Scale bars: 50 μm.) Yellow arrowheads in E indicate meristemoids undergoing amplifying divisions; the white arrow indicates a stoma surrounded by three cells that probably arose via amplifying divisions (see text). Radially arranged cell groups and rounded cells with strong Calcofluor signals, indicated by the blue arrowheads in C, are hair cells and their primordia, respectively. (F) Quantification of the number of pavement cells surrounding a stoma. For each species, at least 48 mature stomata were examined.

Stomata were distributed at high densities on both surfaces of the leaves of C. terrestris, but the other species had biased adaxial-abaxial distributions of stomata (SI Appendix, Fig. S1A). C. japonica had more stomata on the abaxial side, and the amphibious species C. palustris and C. stagnalis had more stomata on the adaxial side. Thus, we focused on the abaxial epidermis of terrestrial species and the adaxial epidermis of amphibious species in further analyses. We observed the epidermis of developing leaf primordia using confocal microscopy and calculated the stomatal index over the course of leaf development (Fig. 2B). We found that, as reported previously, stomatal development in C. palustris occurred in restricted development phases (∼500 to 1,000 μm leaf primordium length), and also that this was the case in another amphibious species, C. stagnalis. In contrast, the terrestrial species started to develop stomata in much smaller leaf primordia (∼200 μm long). Terrestrial species developed stomata in more developmental stages, and the duration of stomatal development in terrestrial plants appeared to be longer compared with the amphibious species. This difference persisted even after normalizing the leaf primordium length by the maximum leaf length of each species (SI Appendix, Fig. S1B).

The two amphibious species had highly synchronous stomatal development; stomatal lineage cells in close proximity were all at similar developmental stages (Fig. 2 C and D and SI Appendix, Fig. S1C). In addition, we rarely identified polygonal meristemoid-like cells in these species. In both of the terrestrial species, on the other hand, we observed a wide spectrum of developmental stages in the stomatal lineage and found a number of polygonal meristemoid-like cells near the developed stomata (Fig. 2E and SI Appendix, Fig. S1 D and E). These observations imply that stomatal lineage cells in terrestrial species go through amplifying divisions before differentiating into GMCs, whereas those of amphibious species do not, which may explain the gradual increase in stomatal number in terrestrial species and the sudden increase in amphibious species.

Our analyses of epidermal cell arrangement provided another line of evidence supporting this postulate. Helical amplifying divisions successively produce three pavement cells and finally a pair of GCs in the center, forming a monoclonal “anisocytic complex” (16, 28) (Fig. 1A). We found that some mature stomata were surrounded by three pavement cells in terrestrial species, but not in amphibious species (Fig. 2 E and F and SI Appendix, Fig. S1D). The difference in the adaxial-abaxial distribution of stomata between terrestrial and amphibious species should be interpreted with caution (SI Appendix, Fig. S1A), because the pattern of stomatal development may be affected by the genetic system defining leaf polarity (29). Nevertheless, we did observe meristemoid-like cells in the adaxial epidermis of C. terrestris (SI Appendix, Fig. S1E), and the arrangement and developmental time course of stomata in the adaxial epidermis of C. terrestris were similar to those in the abaxial epidermis (SI Appendix, Fig. S1 F and G). Thus, we conclude that it is not the adaxial-abaxial stomatal distribution, but rather the lifestyle of each species, that is associated with the presence of amplifying divisions.

Amphibious Species Lack Amplifying Divisions in the Stomatal Lineage.

To obtain direct evidence of the presence or absence of amplifying divisions in Callitriche species, we conducted time-lapse observations of the leaf epidermis through the construction of serial epidermal replicas using dental impression medium. As expected, we observed clear amplifying divisions before the differentiation of stomata in both terrestrial species (Fig. 3 A and B and SI Appendix, Fig. S2B). We did not find any amplifying divisions in the stomatal lineages in the two amphibious species; that is, all the meristemoids produced by entry division differentiated directly into GMCs without any additional asymmetric divisions (Fig. 3 C and D and SI Appendix, Fig. S2A). We cannot exclude the possibility that some stomata in amphibious species are produced by the direct establishment of GMCs from protodermal cells, as observed in the moss and hornwort species (11, 30), but the morphology of the earliest stomatal lineage cells had asymmetric polygonal shapes, even in the amphibious species (Fig. 3 C and D), which indicates that most of these cells were produced by an entry asymmetric division (Fig. 3C). Whereas both terrestrial and amphibious species had active cell divisions outside the stomatal lineage (Fig. 3E), only terrestrial species had prolonged asymmetric divisions in the stomatal lineage (Fig. 3F). Thus, the meristemoids in terrestrial species clearly underwent a series of amplifying divisions, but the meristemoids in amphibious species skipped this step and proceeded directly to differentiation. This discovery demonstrates a clear correlation between plant lifestyle and meristemoid behavior.

Fig. 3.

Meristemoids in terrestrial species undergo amplifying divisions, whereas those in amphibious species do not. (A–D) A series of epidermal impressions showing the process of stomatal differentiation in C. japonica (A), C. terrestris (B), C. palustris (C), and C. stagnalis (D). Dotted lines indicate a thin cell wall that was assumed to have been newly formed just before the impression was made. The colors in the traced images correspond to individual stomatal differentiation stages (Fig. 1A). (Scale bars: 20 μm.) (E and F) Cell division profile obtained by time-lapse analysis of epidermal impressions. Each dot indicates a result obtained from a single observed region (comprising >45 cells on the first day) in different leaf primordia from different individuals, reflecting a biological replicate. (E) Number of symmetric cell divisions outside the stomatal lineage. (F) Number of asymmetric cell divisions in the stomatal lineage.

Identification of SMF Orthologs in Callitriche.

We next aimed to elucidate the molecular basis for this diversity in the meristemoid division patterns in Callitriche. We focused on the SMF genes, which constitute the most fundamental component of the molecular pathways regulating stomatal development in Arabidopsis (12–3) (Fig. 1A) and, arguably, all land plant lineages (13). To identify putative SMF orthologs in Callitriche, we used the Basic Local Alignment Search Tool (BLAST) to compile amino acid sequences similar to the AtSMFs in our RNA-sequencing (RNA-Seq) datasets for Callitriche (31) and public databases (details in Materials and Methods). Based on the amino acid sequences collected, we constructed a maximum likelihood tree and found three monophyletic groups corresponding to each SMF protein (Fig. 1B). We identified two putative orthologs per SMF in C. palustris, and one putative ortholog in the other three species within each clade. Since previous karyotype analyses have indicated that C. palustris is a tetraploid and that C. terrestris and C. stagnalis are diploids (32, 33), the two orthologs in C. palustris are likely a pair of paralogs after genome duplication, and therefore we named them CpSPCH A and CpSPCH B, for example. Note that each putative paralogous gene in C. palustris was ≥90% identical in its amino acid sequence to the other paralog. Given this information and our RNA-Seq data showing that the expression patterns under various conditions were almost the same between paralogs (31) (SI Appendix, Fig. S3A), we assumed that there was no essential difference in protein function between them. We did not strictly discriminate each pair of paralogs in the following analyses and labeled them “CpSPCH”, for example (details in Materials and Methods). We were unable to find an FAMA ortholog in C. stagnalis, but this may have been due to the low-quality assembly for this species (SI Appendix, Table S1).

We checked the expression patterns of the SMF orthologs identified in Callitriche by performing whole-mount in situ hybridization (WISH). As representatives of each lifestyle, we analyzed C. palustris, the species in which we recently established the experimental system (23), and C. japonica, the species that underwent more rounds of amplifying division (Fig. 3). Expression of both CpSPCH and CjSPCH was observed in the basal regions of developing leaf primordia, but the CpSPCH signal was found more toward the apex compared with the CjSPCH signal (Fig. 1C). Although localization to specific cells was not observed in SPCH orthologs, the CpSPCH signal was clearly in the adaxial epidermis, where stomata are actively produced (SI Appendix, Fig. S4), and probably included stomatal precursor cells in the proliferation zone of this species (23). Both MUTE and FAMA orthologs were expressed in rounded GMC-like epidermal cells (Fig. 1C). Compared to CpMUTE, CpFAMA was expressed more toward the apex. Furthermore, CpFAMA-expressing cells were more widely distributed in larger leaf primordia, with signals also found in young GCs (SI Appendix, Fig. S5). CjMUTE and CjFAMA showed similar expression patterns as the C. palustris orthologs, but their signals were more scattered across the primordia (Fig. 1C). Overall, these findings suggest that the sequential expression patterns of SPCH, MUTE, and FAMA in the stomatal lineage are conserved in Callitriche. We further confirmed that all of the characteristic domains of each SMF protein (34) were identifiable in Callitriche orthologs (SI Appendix, Fig. S6), implying that the protein function of SMF is also conserved in Callitriche.

Different Expression Patterns of SPCH and MUTE Underlie the Diversity in Meristemoid Division.

Among the SMF proteins, SPCH and MUTE, transcription factors that maintain (3) and terminate (2) the amplifying division, respectively, are the most likely candidates underlying the interspecific diversity in meristemoid behavior (12, 16). We hypothesized that either different functions or expression patterns of SPCH and/or MUTE were responsible for the diversity in cell division among Callitriche species. Because we did not find any notable substitution in amino acid sequences that might account for the different protein functions among Callitriche species (SI Appendix, Fig. S6), we focused primarily on the gene expression patterns.

The expression patterns of SPCH and MUTE throughout the course of development were analyzed by real-time PCR. We collected leaf primordia at five different developmental stages (defined by size) from healthy growing shoots of each species. The youngest stage tested corresponded to the primordium length at which stomata started to differentiate (Fig. 2B; primordium length of 0.5 mm for amphibious species and 0.25 mm for terrestrial species), followed by four further stages in larger primordia. We extracted RNA from each sample and quantified the expression levels of SPCH and MUTE orthologs. Interestingly, we found two distinct expression patterns related to the lifestyles of the species. In the terrestrial species C. japonica and C. terrestris, SPCH expression was highest at the earliest developmental stage examined, while MUTE expression peaked at later stages (Fig. 4 A and B). In C. palustris, however, the expression levels of both genes peaked at the same developmental stage (Fig. 4C). A similar trend was observed in the other amphibious species, C. stagnalis (Fig. 4D). This unique pattern in amphibious species can be interpreted thus: SPCH and MUTE expression levels are synchronous; that is, the time lag between SPCH and MUTE expression is shorter in amphibious species than in terrestrial species (see Fig. 7).

Fig. 4.

Expression patterns of SPCH and MUTE orthologs in different stages of leaf development (represented by primordium length) in Callitriche species. The expression of each gene was standardized to SAND1 expression (internal control) and shown in log scale. The data obtained from a single individual are shown by the same symbol for each species and connected with a thin line (n = 3; biological replicates). Mean values from the replicates are shown with bold lines. (A) C. japonica. (B) C. terrestris. (C) C. palustris. (D) C. stagnalis.

Most importantly, the observed expression patterns were sufficient to construct a model explaining the molecular basis of meristemoid division diversity in Callitriche (see Fig. 7). According to our model, SPCH-expressing meristemoids in terrestrial species undergo amplifying divisions until MUTE is expressed, as in Arabidopsis. In amphibious species, however, MUTE is prematurely expressed immediately after SPCH expression, leaving no time for the meristemoids to divide; therefore, they differentiate directly into GMCs.

Differences in the Timing of SPCH and MUTE Expression Confirmed at the Cellular Level.

Although the model that we developed based on the results of the real-time PCR experiments seemed to effectively explain the interspecific variation observed (see Fig. 7), the RNA samples were all extracted from whole leaf primordia, so that the results should be interpreted with caution. Whether individual meristemoids from each species followed the same pattern of gene expression that we observed in the bulk experiment remains to be determined. To analyze the expression patterns at higher resolution, we developed a double-color whole-mount fluorescent in situ hybridization (WM-FISH) system for Callitriche species. Combining the WM-FISH system with cell wall staining allowed us to visualize gene expression at cellular-level resolution with a high signal-to-noise ratio. Furthermore, simultaneous detection using fluorescein isothiocyanate (FITC) and digoxigenin (DIG)-labeled probes allowed us to analyze the spatial relationship of the expression domains of two different genes. Using this system, we analyzed the expression patterns of SPCH and MUTE orthologs in developing leaf primordia of C. palustris and C. japonica (Figs. 5 and 6 and SI Appendix, Figs. S7 and S8). Using antisense probes, we detected positive FISH signals characterized by aggregated granular dots, which are typical of the tyramide signal amplification (TSA) system (35), for all the genes that we tested (Figs. 5 A, C, D, E, and G and 6 and SI Appendix, Fig. S8). We observed no such specific signals by using sense probes (Fig. 5 B and F and SI Appendix, Fig. S7).

Fig. 5.

FISH detection of SPCH (green) and MUTE (magenta) orthologs in C. palustris (A–D) and C. japonica (E–G) leaf primordia; cell walls stained (gray) with Calcofluor. (A and B) The results from antisense (as) and sense (s) probes of CpSPCH. (A') and (B') show close-up views of the white boxed frames in A and B. (C) Results from the antisense probe of CpMUTE. Results from the sense probe are shown in SI Appendix, Fig. S7A. (D) Serial optical sections of a C. palustris leaf primordium after double-detecting CpSPCH and CpMUTE with antisense probes. Cross-sectional images were constructed by Fiji and are shown in D'. The positions of the three sections (x, y, and z) are shown by the white dotted line. In the cross-sections, a transition-state cell (y; Fig. 6C), an SPCH-only cell (y), and MUTE-only cells (x) are identifiable in the epidermal cell layer. In the basal region of the primordia, SPCH is broadly expressed but is still restricted to the epidermal (possibly protodermal) cells (z). (E and F) Results from antisense (as) and sense (s) probes of CjSPCH. E' and F' show close-up views of the white box frames in E and F. (G) Results from the antisense probe of CjMUTE. Results from the sense probe are shown in SI Appendix, Fig. S7B. (Scale bars: 50 μm.)

Fig. 6.

Distribution of transition-state cells, in which SPCH and MUTE are coexpressed, in C. japonica and C. palustris. (A and B) FISH detection of SPCH (green) and MUTE (magenta) orthologs in C. palustris (A; primordium length, 1,137 μm) and C. japonica (B; primordium length, 352 μm); cell walls stained (gray) with Calcofluor. (C) Cells in a transition state showing coexpression of SPCH and MUTE are identified for each species (A', B', and B''; close-up views of the box frames in A and B). (Scale bars: 50 μm.) (D and E) Quantification of the positions of cells in a transition state. (D) Graphic summary of the results. (E) The coordinate position of each cell was calculated along the main axis of the leaf primordium and plotted against the relative primordium length standardized by the length of the mature leaf. Therefore, each dot corresponds to one transition-state cell (n = 43 for C. palustris and n = 73 for C. japonica), and each row of dots corresponds to a different leaf primordium (n = 13 for each species). Linear regression plots are shown with 95% confidence intervals.

In C. palustris, CpSPCH was expressed widely in the basal region of developing leaf primordia (Figs. 5 A and D and 6A and SI Appendix, Fig. S8). The signals were restricted to the adaxial epidermis (Fig. 5D'), consistent with the colorimetric WISH result (SI Appendix, Fig. S4B). In smaller primordia, the expression domain of CpSPCH was located more apically (SI Appendix, Fig. S8A), as also demonstrated by colorimetric WISH (Fig. 1C and SI Appendix, Fig. S4A). This may relate to the slow development of the CpSPCH peak in the qPCR analysis (Fig. 4C). Through FISH analyses, we were able to detect the localization of CpSPCH to specific cells, primarily in apical regions of the expression domain (Fig. 5 A' and D'). Some of the CpSPCH-expressing cells were polygonal in shape, which is typical of early meristemoids. Wider CpSPCH expression in the basal region indicates general expression in protodermal cells (Fig. 5D'). We assume that after these protodermal cells undergo an entry division in the more apical region of the primordium, CpSPCH expression becomes limited to the newly produced meristemoid. This SPCH expression pattern has been observed in Arabidopsis during the very early stage of leaf development (3). A similar pattern was also reported recently in an Arabidopsis line, in which the SPCH promoter of tomato (Solanum lycopersicum) drove green fluorescent protein (GFP) expression (10). These findings support our contention that the expression domain of SPCH is conserved in C. palustris.

We observed clear localization of CpMUTE signals to polygonal or rounded cells in the epidermis (Figs. 5 C and D and 6A and SI Appendix, Fig. S8A). These cells had morphologies typical of late meristemoids and young GMCs, as shown by colorimetric WISH (Fig. 1C). Notably, we found a small number of cells expressing both CpSPCH and CpMUTE (Figs. 5D' and 6A' and SI Appendix, Fig. S8A'). CpMUTE expression was relatively weak in those cells, which displayed typical meristemoid-like morphology and were found in the border region between the expression domains of CpSPCH and CpMUTE. These findings indicate that the cells coexpressing CpSPCH and CpMUTE were meristemoids in a transition state; they had just begun to express MUTE while residual SPCH expression continued to decline (Fig. 6C).

In C. japonica, the expression domain of CjSPCH was broader than that of CpSPCH and extended into the apical region, where the expression was localized to polygonal cells (Figs. 5 E and F and 6B and SI Appendix, Fig. S8B). This pattern differs slightly from that observed in the colorimetric system (Fig. 1C), in which a CjSPCH signal was found only in the very basal region of leaf primordia. This indicates that the greater sensitivity of FISH enabled detection of the apical expression of CjSPCH. Cells expressing CjMUTE were similar in morphology to CpMUTE-expressing cells (Fig. 5G) but had a wider distribution in leaf primordia (Fig. 6B and SI Appendix, Fig. S8B), which is consistent with the colorimetric WISH results (Fig. 1C). As for C. palustris, we found cells in the putative transition state that coexpressed CjSPCH and CjMUTE (Fig. 6 B' and B''). Since the expression domains of CjSPCH and CjMUTE were wide and overlapped in C. japonica, the transitioning cells were found not only in the basal region, but also in more apical regions, where mature stomata were located close by (Fig. 6B'' and SI Appendix, Fig. S8 B and B'). When the relative position of the transitioning cells was plotted against relative primordium length, it was clear that the distribution in C. japonica was wider, extending to more apical regions than in C. palustris (Fig. 6 D and E). Unlike C. palustris, C. japonica has a petiole-like region in the basal-most region of the leaf, where little cell division activity was seen (SI Appendix, Fig. S8C). Even after taking account of the presence of the petiole in C. japonica, the difference in distribution patterns remained (SI Appendix, Fig. S8 D and E).

In C. palustris, cell proliferation occurred in the basal region of the leaf blade (23), as in C. japonica (SI Appendix, Fig. S8C), and SPCH orthologs were expressed in dividing cells in both species. Therefore, the basally biased distribution of transition-state meristemoids in C. palustris demonstrated that the meristemoids began expressing CpMUTE immediately after they had been formed, at a time when they were expressing CpSPCH (Fig. 6 D and E). In C. japonica, however, some of the transitioning cells were located more apically, indicating that the expression of CjMUTE was delayed, as shown in our model (Fig. 7).

Fig. 7.

Graphic summary of the results. Meristemoids in the terrestrial species of Callitriche (Left) undergo a series of amplifying divisions, while those of amphibious species (Right) do not divide further but instead differentiate directly into stomata. Different expression patterns of the orthologs of SPCH, which maintains division, and MUTE, which suppresses SPCH activity, underlie this diversity in the behavior of meristemoids. MUTE expression is delayed in terrestrial species, but MUTE is expressed shortly after the onset of SPCH expression in amphibious species; hence, division is terminated early during the development of amphibious taxa.

Notably, the distributions of transitioning cells in the two species overlapped to some extent in the basal regions (Fig. 6 D and E and SI Appendix, Fig. S8 D and E). Since some meristemoids directly differentiated into stomata without any amplifying divisions even in C. japonica (Fig. 2E and SI Appendix, Fig. S8B), we expected to see early expression of CjMUTE in some proportion of C. japonica meristemoids. The overlapping distribution of transitioning cells supported this expectation. In summary, the expression pattern that we inferred from the bulk qPCR experiment was confirmed at the cellular level (Fig. 7).

Plant Materials and Culture Conditions.

The Callitriche plants used in this study were originally collected from their wild habitats in Japan and maintained in our laboratory. We collected C. stagnalis, C. terrestris, and C. japonica specimens in Ibaraki Prefecture, Hyogo Prefecture, and Kanagawa Prefecture, respectively, and included them in our analysis together with the previously described C. palustris NH-1 strain (23). All plants were maintained in a growth chamber at 23 °C under long-day conditions (16 h light/8 h darkness), with a light intensity of 60 μmol photons m⁻² s⁻¹. The specimens used in the experiments were transplanted aseptically to plant boxes containing autoclaved soil (Aqua Soil Amazonia; Aqua Design Amano) and grown under the conditions described above for at least 2 wk (after which sufficient numbers of new shoots and leaves had been formed). For C. palustris, C. terrestris, and C. japonica, the single plant strain of each species collected from their wild habitats was selfed for at least two generations to produce more stable and homozygous lines. In the subsequent experiments, three siblings from one established strain were analyzed and used as the three biological replicates (Figs. 3 E and F and 4). As C. stagnalis did not produce seeds under our culture conditions, asexually propagated wild strains originating from the same locality were used for the experiments.

Observation of Epidermal Cells.

Shoots and leaves were immersed in formalin-acetic acid-alcohol (FAA) fixative (10% formalin, 5% acetic acid, 50% ethanol [vol/vol]). The fixed samples were then cleared using methods based on either thiodiethanol (TDE) (47) or ClearSee (48). For observation by differential interference contrast microscopy (DM4500; Leica Microsystems), we first transferred the leaf samples into an ethanol series and then treated them with TDE for >1 h before mounting. Smaller samples were treated with 1% Calcofluor White (Sigma-Aldrich) in ClearSee for 2 d to stain cell walls before observation by confocal microscopy (FV10i; Olympus) with UV excitation. Captured images were analyzed using Fiji v1.0 (49). The stomatal index (Fig. 2B and SI Appendix, Fig. S1 A, B, and G) was calculated based on multiple images of the epidermis captured in at least three different fields of view in the apical halves of the leaf primordia.

Time-Lapse Observation of the Epidermis.

A dental impression medium-based method (37, 50) was used for time-lapse observations of the epidermis. Leaf surfaces were covered with freshly prepared dental impression medium (Provil novo Light; Kulzer) to make the molds. This procedure was repeated every 1 to 3 d on each leaf to obtain a series of molds throughout the course of leaf development. Finally, we obtained time-lapse epidermal impressions by applying nail polish (AC Quick-Dry Topcoat; Do-Best, Inc.) to each mold, which was then peeled off after the polish had dried. The epidermal impressions were directly mounted onto glass slides and observed by light microscopy (DM4500; Leica Microsystems).

De Novo Transcriptome Assembly.

To reconstruct the transcriptome, we cultured shoots from single individuals of C. stagnalis and C. japonica under either aseptic aerial or submerged conditions. RNA was extracted from whole plant bodies using the RNeasy Plant Mini Kit (QIAGEN) with on-column applications of RNase-free DNase (QIAGEN). The quality of the RNA was measured using the Agilent 2100 Bioanalyzer system with the RNA6000 nano kit; quantities were measured by Qubit fluorometric quantitation (Thermo Fisher Scientific). The libraries for Illumina sequencing were prepared with a stranded mRNA-Seq kit (KAPA Biosystems) using a protocol optimized for a 300- to 400-bp insert size. We sequenced the libraries using the Illumina HiSeq 1500 sequencing system in rapid-run mode and obtained 150-bp paired-end reads for each sample. The raw reads were deposited in the DNA Data Bank of Japan (DDBJ) Sequence Read Archive (DRA; accession nos. DRX227815 [C. stagnalis] and DRX227816 [C. japonica]). The strategy of de novo transcriptome assembly and decontamination followed that of our previous study (31). We used trimmomatic v0.36 (51), Trinity v2.2.0 (52), rnammer v1.2.1 (53), Corset v1.07 (54), and TransDecoder v3.0.0 software. We also used assemblies of C. palustris and C. terrestris that had been reconstructed previously (31). The statistical parameters of the assemblies are listed in SI Appendix, Table S1. Assembly sequences have been deposited in figshare (http://dx.doi.org/10.6084/m9.figshare.13450931).

Identification of SMF Orthologs in Callitriche Species.

The transcriptome of each Callitriche species was searched for SMF orthologs using BLASTp (protein-protein BLAST). To construct a protein phylogenetic tree, we also conducted BLASTp searches against public protein databases and obtained putative SMF orthologs from other species. The sequences retrieved were aligned using MAFFT v7.453 (55). After trimming nonhomologous regions using trimAL v1.4 (56), we constructed a maximum likelihood tree using RAxML v8.2.12 (57) with 1,000 bootstraps (Fig. 1B).

Quantification of Gene Expression by Real-Time PCR.

Total RNA was extracted from leaf primordium samples of Callitriche species using the FastGene RNA Premium Kit (Nippon Genetics). Each sample comprised >10 leaf primordia of approximately the same length. The extracted RNA was then used for reverse transcription with the PrimeScript RT Reagent Kit (TaKaRa Bio) to synthesize cDNA. Real-time PCR was performed using Thunderbird SYBR qPCR Mix (Toyobo) on a Roche LightCycler 480 II PCR platform. We used the SAND1 gene, which shows stable expression in multiple plant species (58, 59), as an internal control. We also confirmed that CpSAND1 was stably expressed in all tissues and conditions in our RNA-Seq data (31) (SI Appendix, Fig. S3A). We quantified the expression of SPCH and MUTE orthologs in the three biological replicates described above using the comparative CT method (Fig. 4). The results from three technical replicates were averaged and are shown in Fig. 4. The primers used for real-time PCR are listed in SI Appendix, Table S2.

For C. palustris, we designed primers based on the sequence of one of the two putative paralogs of SPCH and MUTE. However, considering the high sequence similarity and highly similar expression patterns of the two paralogs (SI Appendix, Fig. S3), we did not design primers that amplified each paralog specifically. We assumed that our primers amplified both paralog and thus labeled the results as “CpSPCH”, for example.

WISH.

We used colorimetric in situ hybridization to detect stomata-related genes in whole-mount leaf primordia. The whole-length coding regions of the target genes (CpSPCH A, CpMUTE A, CpFAMA A, CjSPCH, CjMUTE, and CjFAMA) were PCR-amplified from the cDNA that we had synthesized previously (as explained above) using the primers listed in SI Appendix, Table S2. The amplified fragments were cloned into the pZErO-2 vector (Thermo Fisher Scientific). After amplification of the template with M13F and M13R primers, we used either SP6 or T7 polymerases (Roche) and the DIG RNA Labeling Kit (Roche) for reverse transcription to synthesize DIG-labeled probes. Using these probes, we performed in situ hybridization as described previously (23), using a modification of the cell wall enzyme treatment (1 h at 37 °C). Samples were visualized by light microscopy (DM4500; Leica Microsystems). For CpSPCH, the samples were embedded in Technovit 7100 (Kulzer) and sliced into 15-μm-thick sections using a HM360 rotary microtome (Thermo Fisher Scientific) to examine the signal localization pattern (SI Appendix, Fig. S4B). Information on the cloned sequences was deposited in the DDBJ (accession nos. LC565133 to LC565137 and LC596432).

For C. palustris, we synthesized probes using one of the putative paralogs for each SMF gene as in our real-time PCR analysis. Because of the high similarity between each pair of paralogs, we anticipated a fair amount of cross-hybridization between these paralogs (e.g., MUTE; SI Appendix, Fig. S3B). Therefore, we did not strictly discriminate one paralog from the other in our in situ hybridization experiments and thus labeled the results as “CpSPCH”, for example.

FISH.

To detect gene expression at higher resolution, we applied FISH to whole-mount leaf primordia of the Callitriche species. The method followed the previously described protocol for the Arabidopsis inflorescence meristem (35), with some modifications. DIG- and FITC-labeled RNA probes of Callitriche MUTE and SPCH were synthesized with the DIG RNA Labeling and Fluorescein RNA Labeling mixes (Roche), as described above, using the same plasmids. To detect the DIG probes, we used the mouse anti-digoxigenin antibody IgG1k (Roche) and goat anti-mouse antibody in a tyramide amplification system (Alexa Fluor 488 Tyramide SuperBoost Kit, goat anti-mouse IgG; Invitrogen). We detected the FITC probes using rabbit anti-FITC polyclonal antibody (Invitrogen) and goat anti-rabbit antibody in a tyramide amplification system (Alexa Fluor 594 Tyramide SuperBoost Kit, goat anti-rabbit IgG; Invitrogen). Probe hybridization, antibody reactions, and tyramide signal amplification (TSA) were performed as described previously (35). We successfully obtained specific granular-like signals, which are typical of the TSA system (35). The samples were then treated with 1% Calcofluor White in ClearSee solution for 2 d to stain the cell walls before visualization by confocal microscopy (Fv10i; Olympus). Using our SPCH and MUTE double-detection procedure, we obtained signals in C. palustris clearly showing identical tendencies in the combination of FITC-labeled CpSPCH and DIG-labeled CpMUTE probes and in the combination of DIG-CpSPCH and FITC-CpMUTE probes. The results were combined for further analyses (Fig. 6E and SI Appendix, Fig. S8E). In C. japonica, however, the FITC-CjSPCH probe did not produce specific signals (for an unknown reason); thus, we used only the results from DIG-CjSPCH and FITC-CjMUTE probes in our analysis.

Detection of the Cell Proliferation Zone in C. japonica.

The cell proliferation zones in the leaf primordia of C. japonica were visualized using a 5-ethynyl-2′-deoxyuridine (EdU)-based procedure (SI Appendix, Fig. S8C). We used the Click-iT EdU Alexa Fluor 488 Imaging Kit (Thermo Fisher Scientific), and followed previously described protocols (23, 60) to incorporate EdU and detect the fluorescence signal. To incorporate EdU, we incubated the samples for 10 h under the normal culture conditions described above. After signal detection, the samples were visualized by fluorescence microscopy (DM4500; Leica Microsystems).