Introduction

Plants have the ability to develop most of their organs post-embryonically and generate new organs throughout their lives, unlike animals in which organogenesis completes during embryogenesis1. Flowering plants maintain a source of undifferentiated stem cells through the aboveground-shoot and underground-root systems. These systems are mainly derived from the shoot apical meristem (SAM) and the root apical meristem. SAM generates indeterminate vegetative growth of the stem and produces determinate lateral organs: leaves and their modified forms, namely floral organs.

In typical dicotyledonous plants, embryogenesis is classified into a morphogenesis phase with high cell division and differentiation activity, and a maturation phase in which proliferation is ceased2. The morphogenesis phase begins immediately after fertilisation and continues through the globular, early-heart to late-heart stages. Embryogenesis ends after bent cotyledon stages in the maturation phase. SAM becomes histologically visible during embryogenesis with a typical three-layered structure between cotyledons3. This typical three-layered structure, evidence of SAM formation, is established during transition from the late globular to the torpedo stage3.

The formation of SAM between predetermined cotyledon regions and establishing boundaries between organs are controlled by NAM/ATAF/CUC (NAC) family transcription factors4,5,6. In Arabidopsis thaliana, these developmental processes are performed by CUP-SHAPED COTYLEDON1 (CUC1), CUC2 and CUC3, which encode NAC domain proteins4,6,7,8,9. In cuc1 cuc2 double mutant seedlings lack embryonic SAM and fuse cotyledons4. In SAM initiation and boundary establishment, SHOOTMERISTEMLESS (STM), which encodes a class I KNOTTED1-LIKE HOMEOBOX (KNOX1) transcription factor, is required in addition to CUC1 and CUC28,10. At the globular embryo stage, the expression domains of CUCs and STM genes largely overlap in the predetermined cotyledon boundary region6,8. CUCs and STM can bind to their respective promoter regions and mutually promote their expression8,11,12,13. During embryogenesis, STM promotes expression of CUCs, and also represses CUC1/2 by indirectly activating microRNA164 that targets CUC1/2 mRNA11,12,14,15,16. As a result, a boundary region is formed by suppressing the expression of CUC in the region where STM is expressed. Thus, expression of CUCs is excluded from the centre of the STM-expressing region, and STM expression is restricted to the centre of this region at the heart embryo stage. The STM-expressed region develops into domed-shape SAM and the growth of the CUC-expressed region is suppressed, resulting in a furrow. Therefore, separation of CUC and STM expression patterns is crucial to the formation and maintenance of SAM.

In Gesneriaceae, all members of the genus Monophyllaea and some members of the genus Streptocarpus are known to be one-leaf plants17. Monophyllaea and Streptocarpus are not sister genera, although their morphological structures are similar17,18,19,20. One-leaf plants exhibit a unique developmental manner unlike that of typical seed plants such as the model plant A. thaliana17,21. One-leaf plants develop two identical cotyledons after germination (called the isocotyledonous stage). Next, one cotyledon (the microcotyledon) stops growing, whereas the other (the macrocotyledon) continues growing (called the anisocotyledonous stage) as a result of competition between them17,22. A unique phenotype of one-leaf plants is the lack of new vegetative organ production.

One-leaf plants have a unique shoot-like structure called a phyllomorph that is composed of an indeterminately growing lamina and one petiolode (a stem- and petiole-like organ)21. The phyllomorph has three meristems: the groove meristem (GM), the basal meristem (BM) and the petiolode meristem (PM). PM is located at the base of the midrib and is thought to contribute to growth of the midrib and petiolode21,23,24. GM and BM are responsible for the unique characteristics of one-leaf plants: indeterminately expanding leaf lamina without producing additional vegetative organs21,22,24,25,26. In histological studies, GM has been suggested to be a modified SAM since it is located at the base of cotyledons, though it lacks a typical dome-like structure in the vegetative stage. However, GM contributes to the differentiation of inflorescence meristems (IMs) after floral induction, and is seemingly silenced in the vegetative phase21,24,25. In the reproductive phase, the inflorescence that develops from IMs has the typical shoot structure of angiosperms23,24,25. Meanwhile, two BMs located at the base of macrocotyledons contribute to indeterminate leaf lamina growth, suggesting they are modified leaf meristem21,22,24,26.

In Monophyllaea, we previously reported that in GM, homologs of not only SAM-specific genes such as STM, but also non-SAM-expressed genes such as ANGUSTIFOLIA3 (AN3)/Arabidopsis thaliana GRF-INTERACTING FACTOR1 (AtGIF1) are expressed27. STM expression is detected in the GM region by tissue-specific laser dissection and in situ hybridisation during the vegetative phase, suggesting that the GM has SAM properties26,27. AN3 is specifically expressed in leaf meristem that is established in the basal part of leaf primordia and regulates cell division in A. thaliana and rice28,29,30,31,32. Because AN3 is expressed not only in the BM region but also in the GM region, GM also has lateral organ properties, unlike SAM of usual angiosperms. The mixed expression of the leaf meristem gene and the SAM gene suggests that the boundary region between GM and BM may be ambiguous.

In this study, we identified organ boundary gene CUC orthologs of Monophyllaea glabra to evaluate the properties of one-leaf plant-specific meristems, especially GM. We compared the expression patterns of CUCs with those STMs using whole-mount in situ hybridisation (WISH) in different developmental stages. Our findings reveal novel properties related to the maintenance and initiation of meristems in one-leaf plants of the genus Monophyllaea.

Results

Expression pattern of Mg-CUCs in M. glabra during the vegetative phase

To examine the characteristics of specific meristems, we first identified organ boundary gene CUC orthologs in M. glabra. The phylogenetic relationship in NAC family-related genes from several angiosperms identified two groups, CUC1/2 and CUC3, consistent with previous studies (Fig. 1a)33. We identified two CUC1/2 orthologs, Mg-CUC1/2-A and Mg-CUC1/2-B, and one CUC3 ortholog, Mg-CUC3. The phylogenetic relationship suggested that Mg-CUC1/2-A and Mg-CUC1/2-B are paralogs that split in the lineage to Monophyllaea. CUCs of M. glabra possess an NAC domain that includes highly conserved amino acid motifs (Supplementary Fig. 1)34,35. We then isolated their cDNAs to measure gene expression using WISH.

Figure 1
figure 1

Mg-CUCs expression patterns in Monophyllaea glabra during the vegetative phase. (a) Phylogenetic tree of NAC proteins. Numbers at nodes indicate confidence values (%) calculated from 1000 bootstraps. (b–g) Whole-mount in situ hybridisation of Mg-CUCs at the anisocotyledonous stage: 12 days after sowing (DAS). (b,c) Mg-CUC1/2-A. (d,e) Mg-CUC1/2-B. (f,g) Mg-CUC3. (b′,d′,f′) Magnified images in (b,d,f). (b,d,f) Antisense probes. (c,e,g) Sense probes. Black bars represent 100 μm. Ma macrocotyledon, mi microcotyledon, GM groove meristem, BM basal meristem.

Full size image

We performed WISH at the early anisocotyledonous stage in which GM and BM were identifiable in the macrocotyledon (Fig. 1b–g). We detected CUC expression in the basal region of macrocotyledons, and although there were differences in the degree of staining, these CUC expression patterns were similar for each probe of CUC1/2 and CUC3 (Fig. 1b–g). Expression of CUCs was observed near the BM region, with a spot-shaped character rather than a border pattern (Fig. 1b,d,f). Given previous data on the expression region of AN3 in BM27, we assumed that some CUC-expressed cells are included in BM. Interestingly, expression of CUCs was also observed in the GM region regarded as modified SAM, indicating that the expression pattern of CUCs differed from that of vegetative SAM of A. thaliana (Fig. 1b,d,f).

Expression profiles of Mg-CUC and Mg-STM around the basal area of the macrocotyledon

We next assessed cross-section WISH samples to obtain detailed information on CUC ortholog expression patterns (Fig. 2). These cross-section samples were prepared from specimens used for gene expression by WISH (Fig. 1b). Because three CUCs showed similar gene expression patterns (Fig. 1b–g), we used the probe for Mg-CUC1/2-A as a representative in subsequent analysis of CUC expression (Fig. 2a–c). In addition, to identify the location of the GM, we detected the expression of STM using a probe for the Mg-STM-B ortholog expressed in the GM (Fig. 2d–f)27. By making serial cross-sections from the tip of the macrocotyledon to the microcotyledon, we sequentially observed the BM region, the region between the BM and GM, and the GM region (see Fig. 1d). Expression of CUC was confirmed near both BM regions of the macrocotyledon, where STM expression was hardly observed (Fig. 2a,d,g), and was also spot-like rather than border-like in cross-sections, indicating that CUC expression near the BM was restricted to a small region (Figs. 1d and 2a). CUC expression was also detected in the GM region around the basal area of the macrocotyledon, covering a slightly wider and shallower area than that of the STM expression domain (Fig. 2b,c,e,f). It should be noted that CUC and STM were expressed in the same cells of the subepidermal tissue of GM (Fig. 2b,e). These results demonstrate that the CUC1/2 expression pattern observed in GM was different from that observed in SAM of typical seed plants such as A. thaliana.

Figure 2
figure 2

Expression patterns of CUC and STM orthologs around the basal area of the macrocotyledon. (a–i) Cross-sections of whole-mount in situ hybridisation samples of Mg-CUC1/2-A and Mg-STM-B at the anisocotyledonous stage (12 DAS). Examples of the GM region were cross-sectioned from (a,d,g) to the base (c,f,i) of macrocotyledon. (a–c) Antisense (AS) probes of Mg-CUC1/2-A. (d–f) AS probes of Mg-CUC1/2-B. (g–i) Sense probes (SE) of Mg-CUC1/2-A. Black bars represent 100 μm. Black arrowheads indicate gene expression. Ma macrocotyledon, mi microcotyledon, GM groove meristem.

Full size image

Simultaneous detection of Mg-CUC and Mg-STM expression by double-fluorescent signals

To analyse the localisation of CUC and STM expression in the same tissue at higher resolution, we developed a double-colour whole-mount fluorescent in situ hybridisation (WM-FISH) system for M. glabra (Fig. 3). In the WM-FISH system, simultaneous detection using fluorescein isothiocyanate (FITC) and digoxigenin (DIG)-labelled probes allowed us to obtain the spatial cellular localisation information for two different genes. In addition, we stained the cell wall with calcofluor white to visualise the localisation of gene expression at cellular-level resolution. Using this system, we analysed the expression patterns of CUC and STM orthologs during the vegetative stage of M. glabra (Supplementary Fig. 2). To confirm that we were not detecting artificial signals with the probe sets, we prepared two types of probe sets with different combinations of target genes. As shown in Fig. 3a,b, we detected positive FISH signals characterised by aggregated granular dots, which are typical of the tyramide signal amplification system using antisense probes36.

Figure 3
figure 3

Detection of double gene expression for CUC and STM orthologs by whole-mount fluorescent in situ hybridisation. (a–c) Whole-mount fluorescent in situ hybridisation of Mg-CUC1/2-A and Mg-STM-B at the anisocotyledonous stage (12 DAS). (a) Antisense probes of Mg-CUC1/2-A (green) and Mg-STM-B (magenta). Serial optical sections of a M. glabra seedling after double-detection with antisense probes. Cross-sectional images were constructed by Fiji and are shown in (a′) and (a″). The positions of the two sections (a′ and a″) are shown by the white dotted line. (b) Antisense probes of Mg-STM-B (green) and Mg-CUC1/2-A (magenta). Cross-sectional images are shown in (b′) and (b″). (c) Sense probes of Mg-CUC1/2-A (green) and Mg-STM-B (magenta). Cross-sectional images are shown in (c′) and (c″). Cell walls stained (grey) with Calcofluor White. Ma macrocotyledon, mi microcotyledon, GM groove meristem, BM basal meristem.

Full size image

CUC expression, seen as green signals, was detected near the BM region and in the GM region (Fig. 3a and Supplementary Fig. 2a). Since these expression patterns were consistent with the colorimetric WISH results (Fig. 2), the signals detected with this system likely reflect expression of the target genes. Expression of CUC observed near the BM region was restricted to one or two cells, and granular aggregations were detected over the whole cell (Fig. 3a and Supplementary Fig. 2a). On the other hand, for CUC expression observed in GM, aggregation of granular signals was detected in part of the cell rather than the whole cell (Fig. 3a and Supplementary Fig. 2a). Staining reflecting CUC expression was observed throughout the entire cell by colorimetric WISH (Fig. 2), which may be due to limitations of the experimental system, such as probe penetration into the cell. Alternatively, it may reflect actual subcellular localisation differences or differences in expression levels between tissues. Fluorescence signals for STM, seen as magenta signals, were also observed in the same samples where signals for CUC were detected (Fig. 3a and Supplementary Fig. 2a). Expression of STM was narrower and deeper than that of CUC (Fig. 3a and Supplementary Fig. 2a). Expression of these genes overlapped in the second and third layers beneath the epidermis, and CUC expression was also confirmed in the surrounding areas (Fig. 3a and Supplementary Fig. 2a). These expression patterns were observed even when we replaced the combination of FITC and DIG probes with target genes (Fig. 3b and Supplementary Fig. 2b). Since the expression patterns of CUC and STM do not overlap in the post-embryogenesis of A. thaliana8,9,10,37, the overlapping expression patterns of these genes may indicate a GM-specific mechanism of morphogenesis in one-leaf plants. CUC expression in BM and GM partially or totally overlapped in each meristem rather than in the boundary region between BM and GM, suggesting that CUCs of M. glabra may not yet exhibit the function as the organ boundary gene, at least during the vegetative phase.

Observation of transition from GM to floral meristem

The Monophyllaea body is composed of a phyllomorph, a shoot structure unique to one-leaf plants in the vegetative phase, but in the reproductive phase the shoot system of the inflorescence follows the typical developmental pattern observed in A. thaliana, making lateral meristems and lateral organs from IM and floral meristem (FM). Accordingly, we attempted to examine the gene expression patterns in IM and FM, which are morphologically similar to their counterparts in typical plants.

Before examining the expression of CUC and STM in IM and FM, we first observed the process of meristem formation from GM. GM of M. glabra exhibits low cell division activity during the vegetative phase, but accelerated cell division activity after shifting to the reproductive phase, resulting in the formation of GM-derived IM25,27. Because M. glabra is a short-day plant, and flowering is induced earlier under short-day conditions than under continuous light conditions17, we compared the process of IM development under short-day conditions with GM under continuous light conditions (Fig. 4). Under short-day conditions, a dome-shaped structure was formed near GM, and this was IM exhibiting a tunica-corpus structure, which is typical of shoot meristems (Fig. 4b,c)25. As shown in Fig. 4d, FM formed thereafter. On the other hand, we observed no such dome-like structure after the same number of days under continuous light conditions (Fig. 4e–g).

Figure 4
figure 4

Formation of flower meristem from groove meristem. Development of the groove meristem under short-day conditions (a–d) and under continuous light conditions (e–g). (a,e) 28DAS. (b,f) 35 DAS. (c,g) 42 DAS. (d) 49 DAS. Black bars represent 50 μm. Ma macrocotyledon, mi microcotyledon, GM groove meristem.

Full size image

Expression patterns of Mg-CUC and Mg-STM during the reproductive phase

Next, we performed a timecourse analysis of the expression of CUC and STM in GM and its derivative meristems (see Fig. 4) after transition to the reproductive phase (Fig. 5). CUC initially displayed a band-shaped accumulation in front of the pre-dome structure on the macrocotyledon side, and this expression pattern shifted within the dome structure (IM) as the dome structure formed (Fig. 5a–c). By contrast, STM expression was maintained at the centre of the dome structure (Fig. 5d–f). Therefore, expression of CUC and STM became separated during IM formation. In some samples, during the formation of the dome structure (Fig. 5e), STM expression was observed in the basal region (in the petiolode direction) away from the centre of the dome structure. Therefore, STM expression might partially retain GM properties at the base, even after IM formation.

Figure 5
figure 5

Expression patterns of CUC and STM orthologs during the reproductive phase. (a–i) Longitudinal sections of whole-mount in situ hybridisation samples of Mg-CUC1/2-A and Mg-STM-B during the reproductive phase. (a–c,g) Antisense probes of Mg-CUC1/2-A. (d–f,h) Antisense probes of Mg-CUC1/2-A. (i) Sense probes of Mg-CUC1/2-A. (a,b,d,e) 36 DAS. (c,f) 42 DAS. (g–i) 49 DAS. Black bars represent 100 μm (a–f) and 50 μm (g–i). Black arrowheads indicate gene expression. Ma macrocotyledon, mi microcotyledon.

Full size image

Next, we analysed the gene expression patterns of CUC and STM in FMs (Fig. 5g–i). CUC was expressed in the meristem boundary region, and STM was expressed in the meristem centre (Fig. 5g,h). This expression pattern was similar to that in SAM and FM of A. thaliana, suggesting that CUC and STM have similar functions in IM and FM of M. glabra as in A. thaliana. Therefore, CUC of M. glabra likely functions as an organ boundary gene in the reproductive phase.

Expression patterns of Mg-CUC and Mg-STM during embryogenesis

SAM is formed by separating the expression locations of CUC and STM in A. thaliana embryos8. Since these expression patterns in M. glabra overlapped during the vegetative phase and separated during the reproductive phase, we next assessed the expression of these genes during embryogenesis of M. glabra. CUC and STM expression was observed in cells between the cotyledon primordia in the early heart and late heart stages (Fig. 6). The expression patterns of CUC and STM observed during embryogenesis were similar to those in GM during the vegetative phase (Figs. 2, 3 and 6). Therefore, it is probable that the expression patterns of CUC and STM in GM during the vegetative phase are already established during embryogenesis, and they continue during the vegetative phase.

Figure 6
figure 6

Expression patterns of CUC and STM orthologs during embryogenesis. (a–h) Whole-mount in situ hybridisation of Mg-CUC1/2-A (a–d) and Mg-STM-B (e–h) during embryogenesis. (a,c,e,g) Antisense probes. (b,d,f,h) Sense probes. (a,b,e,f) Early heart stage embryos. (c,d,g,h) Late heart embryos. Black bars represent 50 μm.

Full size image

Discussion

The purpose of the present study was to clarify the characteristics of meristems (in particular, GM) specific to one-leaf plants from the expression pattern of the organ boundary gene CUC. We isolated CUC orthologs of M. glabra and assessed gene expression by WISH (Fig. 1). We found unique expression patterns for CUC in GM and BM.

The expression pattern of CUC in GM and SAM regions was analysed at various stages of the life cycle of M. glabra, including the heart embryo stage, vegetative phase, and reproductive phase (Figs. 1, 2, 5 and 6). CUC was expressed between IM and macrocotyledon, and FM and floral organ in the reproductive phase, as seen in authentic SAM of typical angiosperms (Fig. 5). Therefore, the CUC ortholog of M. glabra appears to function as the organ boundary gene during the reproductive phase. High (~ 90%) amino acid sequence conservation shared with NAC domains of M. glabra CUCs further indicates that CUCs perform an organ boundary function also in M. glabra.

Many genes expressed in a leaf primordia-specific manner in standard angiosperms including A. thaliana. We previously showed that homologs of PHANTASTICA, ROUGH SHEATH2, ASYMMETRIC LEAVES1 (PHAN/RS2/AS1) and AN3 are expressed in the BM region of M. glabra26,27, indicating a leaf meristem-like identity of BM. Because BM is much more active in cell proliferation than GM in the vegetative stage, and BM region is separated from GM by several enlarged, vacuolated cells27, hence something must determine the border between GM and BM. However, in the present study we showed that CUCs are not expressed between the GM and BM in a border-like pattern. Therefore, region separation between GM and BM must involve some other border gene(s) in M. glabra.

We also previously showed that AN3 is expressed in both GM and BM27. Because GM and BM are thought to be modified SAM and leaf meristem, respectively, expansion of AN3 expression not only in BM but also in GM suggests incomplete identity separation between BM and GM. This could also be attributed to the absence of CUC-dependent border establishment between GM and BM in reproductive phase.

To further test the above possibilities, the cell fates and cell identities must be traced in the GM and BM regions. Because a suitable genetic transformation system is not yet established in M. glabra, direct cell file tracing or genetic manipulation are not plausible. However, in the present study, we established WM-FISH to reveal spatial information on expression of specific genes at the cellular level. Using WM-FISH and other approaches such as single-cell transcriptome analysis in combination could trace the detailed cell identity and cell fate changes in meristems of M. glabra in the future.

In A. thaliana, CUC and STM are essential genes mediating the formation and maintenance of SAM4,10. They are co-expressed in future SAM regions of embryos by mutual activation, then become separately expressed in mature SAM due to indirect inhibition of CUC by STM via microRNA1648,11,16. However, in M. glabra, we showed that expression of CUC and STM overlaps from embryogenesis until the beginning of the reproductive phase, suggesting a prolonged stage for pre-mature SAM. Indeed, timing of the disappearance of overlapping expression coincides with activation of SAM, resulting in the production of new organs. In A. thaliana, foliage-leaf formation was inhibited in transgenic seedlings in which a microRNA164-resistant mutant form of CUC was expressed under the control of STM regulatory sequences16. In other words, if CUC and STM gene expression regions do not separate, active meristems are not formed. Therefore, we propose that co-expression of CUC and STM leads to the maintenance of an inactive meristem in the vegetative phase of M. glabra. This overlap could be caused by weak suppression of CUCs by microRNA164-resistant mutation in M. glabra, but we do not support this idea at present because abnormal overlap between CUCs and STM is only temporally observed in the vegetative phase. Rather, we hypothesise that phase-dependent expression changes in microRNA164 might be a possible mechanism for this unique CUC-STM overlap in M. glabra meristems. Future analysis of microRNA164 in M. glabra will test the above idea and could elucidate the GM maintenance mechanism.

From an evolutionary standpoint, we believe that Monophyllaea may be the result of neoteny in plants because M. glabra GM appears to retain juvenile characteristics of normal SAM. If so, the unique shoot system of one-leaf plants (phyllomorph) could have evolved via prolongation of an immature state, namely neoteny of SAM. Neoteny is a key evolutionary process that triggers neomorph formation and novel body plans in plants and animals38,39. Jong and Burt (1975) proposed that the unique one-leaf shoot system in Streptocarpus could be due to neoteny, based on the occurrence of inflorescences on cotyledons21. Our present interpretation differs from the above idea, focusing on the maturation of SAM formation. We propose that neoteny of meristem, from where new organs are formed, can be possible mechanism underpinning the evolution of novel traits in plants.

Materials and methods

Plant materials and growth conditions

Seeds of Monophyllaea glabra were originally collected at Strakaew Cave, Thailand under a permission from Mr. T. Wongprasert of the Forest Herbarium, Thailand26,27. Voucher specimens are housed in the Forest Herbarium, Department of National Parks, Wildlife and Plant Conservation, Bangkok (BKF), and the University of Tokyo Herbarium (TI). The strain was maintained by cultivation in growth chambers. Plants were grown in plates containing one third Murashige and Skoog salts and 0.8% agar at 22 °C under continuous light or short-day (SD; 8 h light and 16 h dark) conditions. The light intensity was ca. 45 µmol/m2/s.

Isolation and characterisation of CUC orthologs from M. glabra

Total RNA was extracted from M. glabra using an RNeasy plant mini kit (Qiagen, Hilden, Germany) following the manufacture’s protocol. cDNA was reverse-transcribed using a SuperScript III First-Strand Synthesis Kit (Invitrogen, Waltham, MA, USA). Primers for isolating CUC homologs were designed based on de novo assembled sequences obtained from mRNA-sequencing (mRNA-seq) data for M. glabra27. Primers for cloning Mg-CUC1/2-A, Mg-CUC1/2-B and Mg-CUC3 were Mg-CUC1/2-A_clon-F1 (5′-ACCTGCACACGCACGCATACTG-3′) and Mg-CUC1/2-A_clon-R1 (5′-ACGGACAGAAACCAGAAATCGGA-3′), Mg-CUC1/2-B_clon-F1 (5′-TGTGTTTCTTGACCTCGCCGGA-3′) and Mg-CUC1/2-B_clon-R1 (5′-GCACTACAGCGATCACGACAAGGT-3′), and Mg-CUC3_clon-F1 (5′-ACGTAGAGTTAAGGCGGGGGA-3′) and Mg-CUC3_clon-R1 (5′-CGTGCTTACTCCATCATCGGGC-3′), respectively. Nucleotide sequences have been deposited in the DNA Data Bank of Japan (DDBJ) (http://www.ddbj.nig.ac.jp/) under accession number LC782938 for Mg-CUC1/2-A, LC782939 for Mg-CUC1/2-B and LC782940 for Mg-CUC3. Although most genes in M.glabra would be present in two copies (homeologs)26,27, we used one type of isolated gene in this study.

Molecular phylogenetic analyses

Amino acid sequences from species other than M. glabra were obtained from the following databases: Phytozome (https://phytozome-next.jgi.doe.gov/) for Amborella trichopoda (version 1.0), Arabidopsis thaliana (Araport11), and Solanum lycopersicum (ITAG 4.0); Snapdragon Genome Database (http://bioinfo.sibs.ac.cn/) for Antirrhinum majus (version 3.0); Plant GARDEN for Streptocarpus rexii (version 1.0). Amino acid sequences were aligned by MAFFT40, and poorly aligned sequences were trimmed with TrimAL41. IQ-TREE242 was used to analyse phylogenetic relationships using the maximum likelihood method. Bootstrap analysis with 1000 replicates was performed using UFBoot243 and phylogenetic trees were visualised using TreeViewer (https://treeviewer.org/).

Whole-mount in situ hybridisation

cDNA sequences were amplified using primers for target genes. Amplified fragments were cloned into the pZErO-2 vector. Templates for probe synthesis were amplified by PCR using M13 forward (5′-GTAAAACGACGGCCAGT-3′) and M13 reverse (5′-CAGGAAACAGCTATGAC-3′) primers. Using the templates, we prepared DIG-labelled antisense and sense probes with SP6 or T7 polymerases (Roche, Basel, Switzerland) using DIG RNA Labeling Mix (Roche). Whole-mount in situ hybridisation (WISH) was performed as described previously27,36. For the WISH, we used embryos at the heart stage and plants at the anisocotyledonous stage in which GM and BM were evident and reproductive stage in which an inflorescence was visible. These were immersed in the fixative solution (4% (w/v) paraformaldehyde with 15% (v/v) dimethyl sulfoxide in phosphate-buffered saline with 0.1% (v/v) Tween-20) for 1 h at room temperature. DIG was detected using a DIG Detection Kit (Roche) with anti-digoxigenin antibody (Roche). Samples were visualised using a MZ16F stereomicroscope (Leica Microsystems, Wetzlar, Germany) or an OM-D E-M10 digital camera (Olympus, Tokyo, Japan) with a SZ61 dissecting microscope (Olympus).

Histology and microscopy

To observe the formation of reproductive meristem derived from GM, plants grown under continuous light or SD conditions were fixed in formalin-acetic acid-alcohol (FAA) (10% formalin, 5% acetic acid, 50% ethanol (v/v)). The fixed samples were then embedded in Technovit 7100 (Heraeus Kulzer, Wehrheim, Germany) and sliced into 10–15 μm thick sections using a HM360 rotary microtome (Thermo Fisher Scientific, Waltham, Massachusetts, USA). Sections for observation of the transition from GM to floral meristem were then stained with 0.02% (w/v) Toluidine Blue in phosphate-buffered saline containing 0.5% (v/v) Triton X-100 (Nacalai Tesque, Kyoto, Japan). Samples were visualised using a DM4500 light microscope (Leica Microsystems).

Whole-mount fluorescent in situ hybridisation

FITC-labelled antisense and sense probes were prepared using the same procedure described for the DIG-labelled probe. Whole-mount fluorescent in situ hybridisation was performed using the tyramide amplification system as described previously44 with minor modifications. To detect DIG and FITC probes we used mouse anti-digoxigenin antibody IgG1k (Roche) and rabbit anti-FITC polyclonal antibody (Invitrogen) as primary antibodies, respectively. We then used goat anti-mouse antibody with the tyramide amplification system (Alexa Fluor 488 Tyramide SuperBoost Kit, goat anti-mouse IgG; Invitrogen) and goat anti-rabbit antibody with the tyramide amplification system (Alexa Fluor 594 Tyramide SuperBoost Kit, goat anti-rabbit IgG; Invitrogen). We successfully obtained specific granular signals that were typical of the tyramide amplification system in M. glabra. The FISH signal in this system reflected the expression of the target gene because we observed no such specific signals when using sense probes (Fig. 3c). Samples were treated with 1% (v/v) Calcofluor White in ClearSee solution45 for 1 day to stain the cell walls before visualisation using a FV3000 confocal microscope (Olympus).