The single Marchantia polymorpha FERONIA homolog reveals an ancestral role in regulating cellular expansion and integrity

The functioning of plant cells is contingent upon the cell wall acting as a barrier between the cell and its environment. This complex matrix of polysaccharides, glycoproteins and other organic compounds defines the growth and shape of a cell and provides protection against biotic and abiotic stresses (Zhu et al., 2021; Showalter, 1993; Franck et al., 2018). Moreover, the cell wall has to resist turgor pressure while also allowing cell expansion during growth. Thus, sensing and controlling cell wall integrity (CWI) is crucial for plant cells, and receptor kinases (RKs) play an important role in sensing extracellular cues that activate intracellular pathways. Different RK subfamilies are defined by their extracellular domain (ECD), which is more variable than their transmembrane and kinase domains (Shiu and Bleecker, 2001). Over the last decade, the Catharanthus roseus RLK1-like (CrRLK1L) subfamily emerged as important CWI sensors (Steinwand and Kieber, 2010; Wolf et al., 2012).

Functions have been assigned to 15 of the 17 CrRLK1L members encoded in the Arabidopsis thaliana genome: they are required for sensing environmental cues and cell-cell communication in diverse contexts, including reproduction, hormone signaling, cell expansion, innate immunity and various stress responses (Escobar-Restrepo et al., 2007; Hématy et al., 2007; Boisson-Dernier et al., 2009; Guo et al., 2009a,b; Kessler et al., 2010; Bai et al., 2014; Schoenaers et al., 2018; Miyazaki et al., 2009; Keinath et al., 2010; Duan et al., 2010; Galindo-Trigo et al., 2020; Zhu et al., 2018; Li et al., 2015; Richter et al., 2017, 2018; Stegmann et al., 2017). For instance, FERONIA (AtFER), THESEUS1 (AtTHE1), HERCULES1 (AtHERK1) and AtHERK2 are required for cell elongation as Atfer single, Atthe1 Atherk1 double, and Atthe1 Atherk1 Atherk2 triple mutants exhibit stunted growth (Hématy et al., 2007; Guo et al., 2009a,b; Kessler et al., 2010). Root hairs burst in mutants in which AtFER and [Ca²⁺]cyt-ASSOCIATED PROTEIN KINASE1 (AtCAP1) [also known as ERULUS (AtERU)] are affected (Bai et al., 2014; Duan et al., 2010). Moreover, AtFER plays a role in powdery mildew resistance (Kessler et al., 2010), innate immunity (Keinath et al., 2010; Stegmann et al., 2017), calcium signaling (Ngo et al., 2014), phytohormone signaling (Guo et al., 2009b; Yu et al., 2012; Mao et al., 2015; Deslauriers and Larsen, 2010), and mechano- and heavy metal sensing (Richter et al., 2017; Shih et al., 2014). In most contexts, the CrRLK1Ls tend to promote cellular growth and cell elongation. In contrast, AtTHE1 actively inhibits cell elongation in hypocotyls when cell wall perturbations occur, e.g. as a result of mutations in the cell wall biosynthesis machinery or upon treatment with isoxaben, a cellulose synthesis inhibitor (Hématy et al., 2007; Denness et al., 2011).

In addition to regulating cellular growth during the vegetative phase of the life cycle, several CrRLK1Ls play a role during fertilization by controlling the growth and reception of the pollen tube. In the female gametophyte, AtFER is highly expressed in the two synergid cells that flank the egg cell and are important for double fertilization (Escobar-Restrepo et al., 2007; Okuda et al., 2009). AtFER is crucial for pollen tube reception and the release of the two sperm cells (Escobar-Restrepo et al., 2007; Guo et al., 2009b; Ngo et al., 2014; Huck et al., 2003; Rotman et al., 2003), a process also involving two other, redundantly acting synergid-expressed CrRLK1Ls, AtHERK1 and ANJEA (AtANJ) (Galindo-Trigo et al., 2020). Moreover, AtFER and AtANJ also play a role in pollen-stigma recognition during pollen germination (Zhu et al., 2021). In contrast to these synergid-expressed CrRLK1Ls, four pollen-expressed CrRLK1Ls, named ANXUR1 (AtANX1), AtANX2, BUDDHAS PAPER SEAL1 (AtBUPS1) and AtBUPS2 (Boisson-Dernier et al., 2009; Miyazaki et al., 2009; Zhu et al., 2018; Ge et al., 2017), are redundantly required for pollen tube integrity and tip growth.

To transduce extracellular cues, CrRLK1Ls bind secreted peptides of the RAPID ALKALINIZATION FACTOR (RALF) family (Stegmann et al., 2017; Ge et al., 2017; Haruta et al., 2014; Dünser et al., 2019; Xiao et al., 2019), which induce complex formation with members of LORELEI (LRE) family of GPI-anchored proteins to signal to the interior of the cell (Ge et al., 2017; Xiao et al., 2019; Mecchia et al., 2017; Capron et al., 2008). Hence, it was suggested that at least some CrRLK1Ls act as CWI sensors that coordinate cell elongation in response to changes in the cell wall (Boisson-Dernier et al., 2011; Hématy and Höfte, 2008; Cheung and Wu, 2011; Nissen et al., 2016; Feng et al., 2018; Lindner et al., 2012). A role as CWI sensors is also consistent with recent findings that AtFER interacts with pectin (Feng et al., 2018; Tang et al., 2021; Lin et al., 2021).

The 17 A. thaliana CrRLK1Ls act partially redundantly and sometimes even have opposite effects on cellular growth (reviewed by Li et al., 2015; Nissen et al., 2016; Lindner et al., 2012; Galindo-Trigo et al., 2016). To investigate the original function of CrRLK1Ls, we characterized them in a system with reduced genetic complexity: the liverwort Marchantia polymorpha, the genome of which encodes only one CrRLK1L (Bowman et al., 2017; Rövekamp et al., 2014). Liverworts represent an early diverging land plant lineage and have been hypothesized to retain, at least in part, characteristics of the earliest land plants (Bowman et al., 2017, 2007; Nishiyama et al., 2003; Qiu et al., 2006; Wickett et al., 2014; Mishler and Churchill, 1984; Kenrick and Crane, 1997). The presence of similar types of gene families in all plant lineages suggests that the differences in development evolved by co-opting and modifying existing developmental programs and genetic networks, rather than through the evolution of novel genes (Bowman et al., 2007; Nishiyama et al., 2003; Floyd and Bowman, 2007). M. polymorpha is thus an ideal system in which to study the ancestral role of genes and how it was modulated and diversified during land plant evolution (Bowman et al., 2017, 2016; Ishizaki, 2017).

We refer to the single CrRLK1L encoded in the M. polymorpha genome, Mp4g15890 (Mapoly0869s0001), as MpFERONIA (MpFER) (Galindo-Trigo et al., 2016; Bowman et al., 2017; Rövekamp et al., 2014) [also known as MpTHESEUS (MpTHE)] (Honkanen et al., 2016). An Mpfer mutant was identified in a large T-DNA screen for defective rhizoid formation, developing short and irregularly shaped rhizoids with brown tips, indicating rhizoid rupture (Honkanen et al., 2016). In the same screen, a mutant affecting the ortholog of the A. thaliana receptor-like cytoplasmic kinase MARIS (AtMRI), known to act downstream of AtFER, was identified with a similar phenotype [named Mpmri or Mppti; Mp7g17560 (Mapoly0051s0094)] (Honkanen et al., 2016; Westermann et al., 2019). As in A. thaliana, a dominant active version of MpMRI can partially rescue the Mpfer rhizoid phenotype (Westermann et al., 2019). This suggests that at least some of the machinery associated with cell elongation and CWI sensing is conserved between A. thaliana and M. polymorpha.

Here, we report that MpFER function is required for a variety of aspects of M. polymorpha gametophyte development. Phylogenetic analyses suggest that CrRLK1L genes first appeared in the common ancestor of land plants. The characterization of lines with reduced MpFER levels indicates that, in addition to its function in rhizoid formation, MpFER controls cell size and organ growth and is involved in male gametogenesis and fertility. Analysis of the MpFER expression pattern points to an involvement in female sexual organogenesis and sporophyte development. Our data suggest that the broad involvement of CrRLK1L function in the development and physiology of land plants is an ancestral and conserved characteristic, although the functions of CrRLK1Ls were extended and adapted to control additional developmental processes in the sporophyte during the course of land plant evolution.

The M. polymorpha genome encodes a single CrRLK1L homolog [Mp4g15890 (Mapoly0869s0001)] (Galindo-Trigo et al., 2016; Bowman et al., 2017; Rövekamp et al., 2014). Like other CrRLK1L members, this gene encodes an intracellular serine/threonine kinase domain, a transmembrane domain and an extracellular malectin-like domain with two malectin domains (Fig. 1A). Given that the ECD of CrRLK1Ls in A. thaliana is crucial for their function and, in contrast to the kinase domain, not interchangeable (Kessler et al., 2014), we performed a phylogenetic analysis with the amino acid sequences of the predicted malectin-like domain of Mp4g15890, the CrRLK1Ls of the mosses Physcomitrium patens (five members), Sphagnum fallax (seven members) and Ceratodon purpureus (three members), the hornworts Anthoceros agrestis Bonn (one member), Anthoceros agrestis Oxford (one member) and Anthoceros punctatus (one member), the lycophyte Selaginella moellendorffii (two members), the basal angiosperm Amborella trichopoda (nine members) and the 17 CrRLK1Ls of A. thaliana (Fig. 1B). The M. polymorpha CrRLK1L member grouped together with the CrRLK1Ls of basal land plants and was in the same clade as, for example, AtFER, AtHERK2 and AtANX1/2, whereas AtTHE1 was in a different one (Fig. 1B). Based on this phylogenetic position and the fact that the malectin-like domain of the M. polymorpha CrRLK1L shared the highest amino acid identity with AtFER (Fig. S1A), we named it MpFER (Bowman et al., 2017; Rövekamp et al., 2014; Westermann et al., 2019).

Interestingly, no significant similarity to the MpFER malectin-like domain was found among available genomic data of six Chlorophyte algae (Chlamydomonas reinhardtii, Dunaliella salina, Volvox carteri, Coccomyxa subellipsoidea, Micromonas pusilla, Ostreococcus lucimarinus) and expression data of six Charophycean algae (Klebsormidium nitens, Nitella mirabilis, Coleochaete orbicularis, Spirogyra pratensis, Mesostigma viride, Closterium peracerosum–strigosum–littorale complex). In the Closterium complex, an RK named CpRLK1 expressed during sexual reproduction was suggested to be a CrRLK1L homolog (Hirano et al., 2015). However, comparing the full sequence or the ECD of CpRLK1 with the predicted M. polymorpha proteome, higher identity to another RK was found [Mp1g17720 (Mapoly0001s0111)] (Fig. S1B,C). Moreover, no RALF orthologs have been reported in algae, suggesting the origin of the CrRLK1L pathway in land plants (Bowman et al., 2017). A similar conclusion was recently reached by analyzing the RALF family in P. patens (Ginanjar et al., 2021). Moreover, the best candidate obtained by BLAST analysis using the MpFER malectin-like domain against the genome of the Charophycean alga Chara braunii clustered outside the CrRLK1L phylogenetic tree (Fig. 1B, Fig. S1B).

To support MpFER as a member of the CrRLK1L subfamily, the 3D structure of its ECD was modeled (Schwede et al., 2003) using the structure of the AtFER ECD (PDB 6a5b.1A; Xiao et al., 2019) as a template (Fig. 1C). Overall, the predicted structure of the MpFER ECD is highly similar to that of AtFER, keeping an almost identical number of secondary structures [seven α-helices and 33 β-sheets in the MpFER ECD (Xiao et al., 2019) and eight α-helices and 34 β-sheets in the AtFER ECD, respectively], and a similar spatial disposition for 32 of the β-sheet structures and six of the α-helices (Fig. 1C). When comparing the structural superposition of the AtFER and MpFER ECDs (Fig. 1D, Fig. S1D), we observed that most conserved parts of the malectin-like domain are in the core of the protein, and the more variable parts are located peripherally. Moreover, the 3D structure of the CpRLK1 ECD could not be properly modeled using the AtFER ECD (Fig. S1D,E), and neither could that of C. braunii (Fig. S1D,F). Taken together, these results suggest that the malectin-like domain of MpFER could interact with similar proteins as AtFER does in A. thaliana and may thus be involved in similar pathways. Sequence comparison of the MpFER kinase domain showed conservation of the activation loop and a Lys residue important for catalytic activity (Fig. 1E). Furthermore, several determined phosphorylation sites found in AtFER, AtERU and CrRLK1 were conserved in MpFER (Fig. 1E) (Schoenaers et al., 2018; Haruta et al., 2014; Liu et al., 2016; Schulze-Muth et al., 1996). Given that the genomes of M. polymorpha and Anthoceros spp. encode a single CrRLK1L and this RK family is not present in Charophycean algae, we propose MpFER to be basal and orthologous to all other land plant CrRLK1Ls, and that the CrRLK1L family arose as plants conquered land.

In A. thaliana, the expression patterns of the 17 CrRLK1L members span vegetative and reproductive stages and both generations of the life cycle (reviewed by Lindner et al., 2012), pointing to the importance of CrRLK1Ls for basic cellular functions. To assess whether this pattern is reflected in the land plant with the most ancestral characteristics, a promoter fragment of 3.2 kb (_proMpFER) was used to drive expression of a triple yellow fluorescent protein (trpVNS) with a nuclear localization signal (_proMpFER:trpVNS) and transformed into M. polymorpha sporelings. All transformants expressing the trpVNS fluorescent protein (45 of 48 independent lines) exhibited an indistinguishable expression pattern (Fig. 2). During vegetative stages of gametophyte development, expression was observed in most cells of the thallus (Fig. 2A,B). In concordance with the importance of MpFER for rhizoids (Honkanen et al., 2016), high trpVNS expression was observed in these tip-growing cells (Fig. 2A′,B′). Although trpVNS was expressed in gemma cups, no expression was observed in mature gemmae (clonal, asexual propagules produced in the gemma cups), possibly owing to their dormancy (Fig. 2C,D).

In mature female sexual organs, strong expression was observed in the entire archegoniophore (Fig. 2E,F). The archegonia expressed trpVNS in most cells, except for the egg cell (Fig. 2G). However, after fertilization, _proMpFER:trpVNS became active in the zygote (Fig. 2H). In male sexual organs, trpVNS was broadly detected in the antheridial splash platform (Fig. 2I,J) and, specifically, in the spermatogenous tissue of the antheridia and the non-reproductive jacket cells surrounding them at different developmental stages (Fig. 2K). Inside the sporogonium, the sporogenous cells, and subsequently the developing spores, but not the elater cells, strongly expressed trpVNS (Fig. 2L).

Taken together, MpFER is expressed in most tissues during the vegetative stages of gametophyte development as well as in the antheridia and archegonia during sexual reproduction. Interestingly, the _proMpFER promoter was not active in mature gemmae and the unfertilized, quiescent egg cell, although expression was initiated after dormancy and in the zygote. In concordance with the collectively ubiquitous expression of CrRLK1Ls in A. thaliana, the MpFER expression pattern reinforces the importance of this family for basic cellular functions in land plants.

To analyze the function of MpFER, three independent artificial microRNA constructs targeting MpFER were designed (amiR-MpFER), based on the endogenous microRNA MpmiR160 (Fig. S2; Flores-Sandoval et al., 2016) and driven by the ubiquitous _proMpEF1 promoter (Althoff et al., 2014). Whereas amiR-MpFER1 and amiR-MpFER2 target sites in the ECD coding sequence, amiR-MpFER3 targets the sequence encoding the kinase domain (Fig. 3A,B). For each amiR-MpFER construct, several independent transgenic lines were obtained that showed a reduction in thallus size (Fig. 3C,D, Fig. S3A). MpFER levels were quantified by quantitative reverse transcription PCR (qRT-PCR) in two independent amiR-MpFER2 and amiR-MpFER3 lines each, which showed a similar phenotype (Fig. 3E). MpFER expression ranged from 10% to 20% of the wild-type level in all amiR-MpFER lines analyzed (Fig. 3E).

Previously, the T-DNA insertion mutant Mpfer-1 was identified in a screen for mutants with defective rhizoids (Honkanen et al., 2016; referred to as Mpthe). As the T-DNA inserted into the 3′ UTR of the MpFER gene (Fig. 3A; Honkanen et al., 2016), MpFER expression was analyzed by qRT-PCR, but it showed no reduction compared with the wild type (Fig. S3B), suggesting that the T-DNA insertion does not affect MpFER transcription. Sequencing of the MpFER coding sequence in the Mpfer-1 mutant did not show any mutation, and 3′ RACE-PCR showed that transcript terminated about 1.3 kb inside the T-DNA and was not polyadenylated (Fig. S3C). Because the thallus size of Mpfer-1 mutants is similar to that of amiR-MpFER lines (Fig. 3C,D, Fig. S3A) and efficient translation depends on the presence of a poly(A)-tail (Sachs and Varani, 2000), we propose that MpFER protein levels are reduced in the Mpfer-1 mutant despite normal transcript levels.

Rhizoids are tip-growing cells analogous to the root hairs of flowering plants (angiosperms). Reduced MpFER activity strongly impaired rhizoid formation and the rhizoids collapsed (Fig. S3D). This had also been reported for the Mpfer-1 mutant (Honkanen et al., 2016) and is in concordance with the function of AtFER and AtANX1/2 in tip-growing root hairs (Bai et al., 2014; Duan et al., 2010; Haruta et al., 2014; Gonneau et al., 2018) and pollen tubes (Boisson-Dernier et al., 2009; Miyazaki et al., 2009), respectively. As CrRLK1L members are important for cell expansion in angiosperms (Hématy et al., 2007; Guo et al., 2009a; Kessler et al., 2010; Li et al., 2015; Dünser et al., 2019; Gachomo et al., 2014; Merz et al., 2017), the size of epidermal thallus cells in fully differentiated regions with minimal growth was measured in Mpfer-1 and amiR-MpFER lines, showing a reduction in cell area compared with the wild type (Fig. 3F, Fig. S3E,F). The average epidermal cell area was similarly reduced, although the distribution of cell areas differed between amiR-MpFER lines and Mpfer-1 (Fig. 3F, Fig. S3E,F).

In conclusion, MpFER has a fundamental role in rhizoid formation and cellular growth during vegetative development. These results suggest an ancestral and conserved function of the CrRLK1Ls in polar tip growth and cell expansion.

Given that Mpfer-1 and amiR-MpFER lines retain some MpFER activity, we generated Mpfer knockout mutants with the goal of unveiling potentially hidden CrRLK1L functions by generating a plant without any CrRLK1L activity. Using CRISPR/Cas9, two sites in the ECD-coding sequence of MpFER were targeted (Fig. 4A,B). At least ten independent lines were selected and analyzed for each target site. All plants with severely affected thallus development contained a frame-shift mutation at the respective target site (Fig. 4B,C) but MpFER was not affected in transgenic plants with normal development.

The thallus area of the newly generated, amorphic knockout mutants was more strongly reduced compared with Mpfer-1 (Honkanen et al., 2016), confirming that, like the amiR-MpFER lines, Mpfer-1 is a hypomorphic mutant (Fig. 4C,D). In contrast, rhizoid integrity was similarly affected in hypo- and amorphic Mpfer alleles, suggesting that rhizoid formation is more sensitive to reduced MpFER activity than is thallus growth (Fig. 4E,F). Despite the strong impact on thallus development, both hypo- and amorphic Mpfer mutants form gemma cups.

The strong disruption of thallus growth in amorphic Mpfer mutants prompted us to investigate whether cells of the thallus died in the absence of MpFER activity. Indeed, using Tryphan Blue staining, we detected dead cells in these mutants, indicating a loss of morphological integrity (Fig. 4G). To investigate the mechanical properties of thallus cells, we measured their apparent stiffness using cellular force microscopy on 3-day-old gemmalings (plantlets derived from a gemma) and found that amorphic Mpfer-2 mutants had a lower apparent stiffness than the wild type (Fig. 4H). The apparent stiffness of the thalli depends on both the stiffness of the cell wall and the turgor pressure of the cells. Because the apparent stiffness is lower in Mpfer-2 mutants, it is likely that a reduction in the stiffness of their cell wall causes the cells to burst and die.

In A. thaliana, at least seven CrRLK1L family members play a role in reproduction, with AtANX1/2 and AtBUPS1/2 being important for pollen tube growth (Boisson-Dernier et al., 2009; Miyazaki et al., 2009; Zhu et al., 2018; Ge et al., 2017) and AtFER, AtHERK1 and AtANJ for pollen tube reception by the synergid cells (Escobar-Restrepo et al., 2007; Galindo-Trigo et al., 2020). To characterize the function of MpFER in reproductive development, amiR-MpFER3 lines were transferred to sexual organ-inducing conditions (Chiyoda et al., 2008). The number of antheridiophores produced was significantly lower in amiR-MpFER lines compared with the wild type (Fig. 5A, Fig. S3G) and the antheridiophore splash platforms were also smaller (Fig. 5B). The reduction in MpFER expression in antheridiophores of amiR-MpFER lines was confirmed by qRT-PCR and results were comparable to those using thallus tissue (Fig. 5C). A pronounced reduction in spermatogenous tissue was observed in many antheridia isolated from amiR-MpFER3-3 lines (Fig. 5D,E). Whether this phenotype is caused by reduced cell expansion, as was observed in thalli, or by a lower cell division rate is currently unclear. Moreover, although the plants were grown under optimal conditions in axenic culture boxes, the effects on reproductive structures could be partly indirect and due to the reduction in plant size and/or rhizoid function.

To determine male fertility of these amiR-MpFER lines with reduced spermatogenous tissue, we performed crosses with wild-type female plants. The spermatocyte concentration harvested from wild-type antheridia was almost threefold higher than from amiR-MpFER lines and was adjusted using a hemocytometer. In just one of the crosses (n=20), a single sporophyte was formed, resulting in 0.05±0.05 (mean±s.e.m.) sporophytes per archegoniophore, whereas the same females crossed with wild-type males yielded 9.95±2.33 (mean±s.e.m.) sporophytes per archegoniophore. In summary, MpFER plays a role in antheridiophore development, and spermatocytes of plants with reduced MpFER activity are poorly fertile. Thus, CrRLK1Ls have a conserved role in reproduction in addition to their role in cell expansion and integrity during vegetative development.

Some A. thaliana CrRLK1Ls are involved in surveying CWI and in inhibiting growth when CWI is impaired. Cell wall defects caused either by mutations affecting cellulose biosynthesis or by treatment with isoxaben, a cellulose synthesis inhibitor, can be suppressed by a mutation in AtTHE1 (Hématy et al., 2007; Denness et al., 2011). To investigate whether MpFER also acts as CWI sensor, we treated wild-type, Mpfer-1 and Mpfer-2 gemmae with isoxaben (Fig. 5F,G). As in the wild type, isoxaben affected the development of gemmae with reduced MpFER activity, all growing progressively less with increasing isoxaben concentration (Fig. 5F,G). Moreover, isoxaben treatment was lethal for Mpfer-2 knockout mutants (Fig. 5F,G). Similar observations were reported for Atfer-4, where isoxaben treatment increased the number of dead cells in hypocotyls (Malivert et al., 2021). These results suggest that the function of AtTHE1 in repressing cellular growth when CWI is impaired appeared later in the course of land plant evolution or was lost in the Marchantia lineage.

As downregulation of MpFER or AtFER produce similar phenotypes concerning polarized growth and cell expansion in M. polymorpha and A. thaliana, respectively, interspecific complementation was attempted. First, the coding sequence of MpFER fused to the Green Fluorescent Protein (GFP) gene under the control of the viral 35S promoter (_pro35S:MpFER-GFP) was transformed into heterozygous Atfer-1 mutants. Then, GFP expression at the cell periphery and complementation of the bursting root hair and reduced rosette size phenotypes were assessed in transgenic plants (Fig. 6A). Atfer-1 homozygotes expressing the MpFER-GFP protein look similar to Atfer-1 mutants, indicating that MpFER does not complement these Atfer-1 loss-of-function phenotypes (Fig. 6A). Although we cannot exclude the possibility that this is because we used a heterologous promoter, we consider this unlikely as expression the 35S promoter is routinely used to complement root hair phenotypes (Baumberger et al., 2001; Kim and Dolan, 2016). In a complementary experiment, the coding sequence of AtFER fused to the Citrine gene under the control of the MpEF1 promoter (_proMpEF1:AtFER-Cit) was transformed into amiR-MpFER3-2 plants. Ten independent lines with Citrine expression at the cell periphery were phenotypically characterized (Fig. 6B,C). All had bursting rhizoids and a reduced thallus size, similar to the parental amiR-MpFER3-2 line, suggesting that AtFER does not rescue the vegetative phenotypes produced by downregulation of MpFER. A sequence alignment of AtFER with amiR-MpFER3 showed no complementarity (Fig. 3B), confirming that the failure to complement is not caused by a potential downregulation of AtFER by amiR-MpFER3. Taken together, these results show that MpFER and AtFER cannot substitute for each other with respect to their function in cell expansion and tip growth.

In A. thaliana, it was shown that AtFER forms a complex with AtRALF peptides and the LRE/LLG co-receptors (Xiao et al., 2019). As no interspecific complementation was observed, the conservation of binding surfaces between CrRLK1L, RALF and LRE homologs were analyzed. The M. polymorpha genome encodes three RALF peptides [MpRALF: Mp1g27120 (Mapoly0002s0166); MpRALF2: Mp2g21670 (Mapoly0040s0047); MpRALF3: Mp7g07270 (Mapoly0076s0067)] (Galindo-Trigo et al., 2016; Bowman et al., 2017). We performed phylogenetic analysis using the amino acid sequences of the predicted mature peptides of A. thaliana and M. polymorpha RALFs (Fig. S4A). All MpRALF peptides clustered together with AtRALFs known to interact with CrRLK1L receptors, including AtRALF1 and AtRALF23, which serve as AtFER ligands (Stegmann et al., 2017; Haruta et al., 2014) (Fig. S4A). The MpRALFs shared the four conserved Cys residues as well as the YXXY and YY motifs with the AtRALF1 subgroup (Fig. S4B). However, only MpRALF1 and MpRALF3 had an RRXL motif, which is important for cleavage by the S1P protease (Fig. S4B,C) (Srivastava et al., 2009), consistent with the presence of one S1P ortholog in basal plants [Mp8g07990 (Mapoly0155s0018); Fig. S4D]. There are two MpLRE/LLG proteins encoded in the M. polymorpha genome (Fig. S5A,B). Both the MpLRE1 [Mp5g09600 (Mapoly0048s0110)] and MpLRE2 [Mp4g22100 (Mapoly0090s0020)] proteins showed conservation of the eight Cys residues and the ND motif characteristic of the family (Fig. S5B); however, MpLRE1 did not contain a GPI-anchoring site (Fig. S5B). Structure prediction using AtLLG1 as a template (Xiao et al., 2019) showed a conserved general structure (Fig. S5C,D). Some of the MpRALF and MpLRE family members showed a similar expression pattern as MpFER (Fig. S5F), suggesting that the corresponding proteins could form a complex similar to that described in A. thaliana.

To gain insights into the formation of a putative CrRLK1L signaling complex, the MpFER/MpRALF/MpLRE complex was modeled. The general structure appeared to be conserved (Fig. S5E); however, analysis of conserved sites between M. polymorpha and A. thaliana showed a low amino acid conservation in the interacting surfaces of the complex subunits (Fig. 6D). This suggests that the lack of interspecific complementation may be due to differences in amino acids that are important for complex formation, and that the proteins forming the complex have diverged, following distinct routes of co-evolution in the two lineages.

Because interspecific complementation of the amiR-MpFER phenotypes was unsuccessful, we investigated whether suppression using a downstream component of CrRLK1L signaling identified in A. thaliana was possible. AtMRI, a receptor-like cytoplasmic kinase, acts downstream of AtFER or AtANX1/2 in the regulation of polar tip growth (Boisson-Dernier et al., 2015; Liao et al., 2016). The dominant AtMRI^R240C mutation can suppress pollen tube rupture in the Atanx1 Atanx2 double mutant and the root hair defects of Atfer-4 (Boisson-Dernier et al., 2015; Liao et al., 2016). A mutation in MpMRI (also known as MpPTI), with a bursting rhizoid phenotype similar to that observed in Mpfer-1, was identified in the same genetic screen for M. polymorpha mutants with defective rhizoids (Honkanen et al., 2016). To test whether MpMRI is a conserved downstream component of the MpFER signal transduction pathway, a dominant-active form of the protein equivalent to AtMRI^R240C was transformed into the amiR-MpFER3-2 line, driven by 2 kb of the endogenous promoter (_proMpMRI:MpMRI^R240C). Several independent lines showed a partial restoration of rhizoid growth; however, we also observed defects in thallus development of lines with higher levels of MpMRI expression, mainly abnormalities in the epidermis (Fig. 6E, Fig. S6A-E). This suggests that MpMRI acts downstream of MpFER in the signal transduction pathway controlling polarized growth since the origin of land plants, in agreement with a recent report on the functional characterization of MpMRI (Westermann et al., 2019).

Because overexpressing AtFER in M. polymorpha did not lead to any obvious phenotypes or complementation of amiR-MpFER lines (Fig. 6B,C), we also overexpressed AtFER and MpFER in the wild type. As in the amiR-MpFER background, overexpression of AtFER (_proMpEF1:AtFER-Cit) in a wild-type background had no effect, consistent with the idea that AtFER is unable to form a complex with the corresponding M. polymorpha proteins. However, when expressing an MpFER-Citrine fusion protein (_proMpEF1:MpFER-Cit) in wild-type plants, thallus development was strongly affected (Fig. 7A). As expected, MpFER-Cit localized to the membrane (Fig. 7A, inset), and higher protein levels correlated with a more severe phenotype (Fig. S6F,G). Scanning electron microscopy of the epidermis of plants expressing high levels of MpFER showed defects in the formation of air chambers and air pores (Fig. 7B), similar to what we observed when expressing the dominant-active MpMRI^R240C (Fig. S6A-C). This phenotype is reminiscent of the quasimodo1 mutant in A. thaliana, in which cell walls have a reduced pectin content and cell adhesion is compromised (Bouton et al., 2002). Moreover, overexpression of MpFER affects the number of lobes in the antheridial receptacle, producing only four instead of eight lobes (Fig. S6H), supporting the importance of MpFER for sexual development. Considering the importance of MpFER for normal rhizoid formation, we also analyzed rhizoids in the _proMpEF1:MpFER-Cit lines: although rhizoid morphology looked normal, a decrease in rhizoid number was observed (Fig. 7C,D). To analyze cell wall properties of MpFER overexpressor lines, chlorophyll efflux rates were measured as an indicator of epidermal permeability (Fig. 7E). Overexpression of MpFER clearly reduced the chlorophyll efflux compared with the wild type, whereas the opposite was observed in Mpfer-2 mutants (Fig. 7E). Taken together, the results obtained by overexpressing MpFER support a role of MpFER in cell wall formation and the maintenance of tissue integrity during plant morphogenesis.

During land plant evolution, many developmental aspects have changed in order to adapt to new environments, producing the enormous diversity observed in the plant kingdom. However, the control and maintenance of CWI remained a key aspect for the biology of a plant cell. CWI sensing is not only important for vegetative growth by cell expansion, but it is also essential for sexual reproduction and defense responses.

The CrRLK1L family was first characterized for its importance during angiosperm fertilization, a process comprising aspects of polar cell elongation, CWI control, and cell-cell communication (Escobar-Restrepo et al., 2007; Huck et al., 2003; Rotman et al., 2003). Lately, different members of this family were found to carry out diametrically opposite roles in diverse aspects of plant development, which has made it difficult to define the core or ancestral function of this gene family comprising 17-members in A. thaliana (Escobar-Restrepo et al., 2007; Hématy et al., 2007; Boisson-Dernier et al., 2009; Guo et al., 2009a,b; Kessler et al., 2010; Bai et al., 2014; Miyazaki et al., 2009; Keinath et al., 2010; Duan et al., 2010; Galindo-Trigo et al., 2020; Zhu et al., 2018; Li et al., 2015; Richter et al., 2017; Ge et al., 2017).

Here, we report the characterization of the single CrRLK1L gene encoded in the genome of M. polymorpha. Structurally, the MpFER ECD and kinase domain share similarities with previously characterized CrRLK1L members, including the malectin-like domain and conserved phosphorylation sites. Conservation of RALF and LRE members in M. polymorpha suggests that MpFER also forms a complex at the plasma membrane (Xiao et al., 2019). Although modeling predicted that the general structure of the complex was conserved, the interaction surfaces seem to have diverged and co-evolved independently in the respective lineages.

We observed a broad involvement of MpFER at both vegetative and reproductive stages of development. During vegetative growth of the gametophyte, MpFER is required for rhizoid formation and cell expansion, but it also plays a role in male gametogenesis and is expressed in female reproductive organs. These findings suggest that the CrRLK1Ls have, also in ancestral land plants, held roles in various aspects of both vegetative and reproductive development. Thus, the role of CrRLK1Ls in angiosperm reproduction does not represent a derived feature from a purely vegetative function in cell elongation in bryophytes but constitutes a conserved feature of this gene family.

Moreover, the almost ubiquitous expression of MpFER in M. polymorpha is consistent with the collective expression of different CrRLK1L members in essentially all tissues and organs of A. thaliana (Lindner et al., 2012). Thus, gene duplication allowed the expansion of the CrRLK1L family in angiosperms, diversifying their expression patterns and functions but also leading to genetic redundancy (Boisson-Dernier et al., 2009; Galindo-Trigo et al., 2020; Ge et al., 2017; Lindner et al., 2012).

CrRLK1L members are important sensors of CWI during polarized growth, both in pollen tubes (Boisson-Dernier et al., 2009; Ge et al., 2017) and root hairs (Duan et al., 2010). In early divergent land plants, rhizoids are tip-growing cells with a function analogous to that of root hairs, important for taking up nutrients and water (Honkanen et al., 2016; Bowen, 1935). Based on the conservation of many genes controlling the development and growth of tip-growing cells forming rooting systems, it was suggested that these pathways were active in the earliest land plants that existed about 470 million years ago (Honkanen et al., 2016). That MpFER also controls rhizoid integrity points to the importance of the CrRLK1L pathway for polarized growth since the origin of land plants. Moreover, the fact that amorphic Mpfer mutants contain many dead cells demonstrates the importance of the CrRLK1L pathway for cellular integrity during vegetative growth. As the occurrence of dead cells was recently also reported in Atfer-4 mutant seedlings (Malivert et al., 2021), the function of this pathway in maintaining cellular integrity may also be ancestral. Whether MpFER also plays a central role in plant innate immunity, as it does in angiosperms (Keinath et al., 2010; Stegmann et al., 2017), remains to be determined.

The signaling mechanisms downstream of MpFER in M. polymorpha development are still unclear. Mutations affecting MpMRI, the M. polymorpha ortholog of AtMRI that acts downstream of AtFER and AtANX1/2 in A. thaliana during polarized growth (Boisson-Dernier et al., 2015), lead to a rhizoid phenotype similar to that of Mpfer (Honkanen et al., 2016; Westermann et al., 2019). The suppression of the rhizoid phenotype in amiR-MpFER transformants with the constitutively active form MpMRI^R240C corroborates a conserved CrRLK1L signaling cascade during polarized growth in M. polymorpha. However, whether this pathway also relies on Ca²⁺ and reactive oxygen species as second messengers, as it does in A. thaliana, remains to be determined.

In A. thaliana, loss-of-function mutations in the motor protein AtKINESIN-13a, a microtubule-based motor, cause cell elongation in petals, leaves and hypocotyls concomitant with changes in cell wall composition in leaves (Fujikura et al., 2014). This effect depends on functional AtTHE1 and does not occur in Atkinesin-13a Atthe1 double mutants. Interestingly, in this case, AtTHE1 stimulates cell elongation in response to defects in cell wall deposition (Fujikura et al., 2014). Based on our results, we have no indication that MpFER acts as a CWI sensor and regulates growth similar to AtTHE1 in response to cell wall disturbances. We have not observed a context in M. polymorpha development, in which MpFER inhibits cell expansion. Therefore, the ability to restrict growth in response to disturbances in CWI could represent an evolutionarily derived feature, as gene duplication allowed for the diversification and specialization of CrRLK1Ls to regulate cellular growth in a more complex and context-dependent way. The capacity to regulate growth flexibly appears to be fundamentally context dependent, as a similar discrepancy in promoting or inhibiting growth is also known for some well-studied growth-promoting agents, such as the phytohormones auxin and brassinosteroid, which can have growth-inhibitory effects dependent on tissue and concentration. The reverse was noted for the phytohormones ethylene and abscisic acid, which are usually considered growth inhibitors but can also promote growth, depending on the context (Pierik et al., 2006; Wolf and Höfte, 2014; Humplík et al., 2017).

Taken together, our results suggest an ancestral and conserved function of the CrRLK1Ls in polar cell growth and cell expansion, but also during sexual reproduction. In addition, we probed the relevance of CrRLK1L members for cell integrity and morphogenesis. In angiosperms, CrRLK1Ls occupy an important position at the interface of developmental and environmental inputs, which are integrated by a downstream signaling machinery that controls cell shape and polar growth to ensure normal development, e.g. by preventing cell rupture upon CWI defects. Whether the same downstream machinery is utilized in a similar fashion in M. polymorpha remains to be shown, even though the similarity in function points to a related mechanism. Even less is known about upstream aspects of CrRLK1L signaling. Future studies on the transcriptional regulation and the upstream components, such as the putative RALF ligand(s), in the highly simplified M. polymorpha system could provide fundamental insights into the molecular mechanism of the CrRLK1L pathway, which is so central to land plant physiology, growth and development.

M. polymorpha subsp. ruderalis (Bischl. & Boissel.-Dub.) plants of the Tak-1 or Tak-2 accession were grown on sterile culture on half-strength Gamborg's B5 basal medium (PhytoTech Labs), supplemented with 1% phytoagar in Petri dishes sealed with air-permeable tape. The plants were kept under fluorescent light and long-day conditions (16 h light at 22°C, 8 h dark at 20°C) in Percival growth cabinets (models AR-41L3 and AR-41/L2). Alternatively, to induce sexual reproduction, plants were cultivated in Jiffy pots filled with a 1:1 mixture of soil (‘Einheitserde D73+Bims’, Universalerde) and sand or in culture boxes (Duchefa Eco2 Box, E1654) in sterile culture on half-strength Gamborg's B5 basal medium, under fluorescent light supplemented with far-red light (GreenPower LED module HF far-red, 929000464503 and 929000632103 Philips).

For Arabidopsis thaliana L. (Heyn), wild-type plants in all experiments were of the Ler-0 accession. The Atfer-2 mutant was obtained from the Signal Collection at the Salk Institute. Plants were grown in soil (ED73; Universalerde), covered with a thin quartz sand layer, under long photoperiods (16 h light/8 h dark) at 23°C and 60% relative humidity.

Protein sequences were identified in plant genomes via BLASTp searches in https://phytozome.jgi.doe.gov/pz/portal.html. To elucidate the evolutionary relationship across land plant evolutionary history, we focused on A. thaliana, Amborella trichopoda, Selaginella moellendorffii, Physcomitrium patens, Spagnum fallax, Ceratodon purpureus, Antoceros agrestis, A. punctatus and M. polymorpha as representative species for angiosperms, lycophytes, mosses, hornworts and liverworts, respectively. Complete or partial coding protein sequences were aligned using ClustalW parameters and were conducted in MEGA7. Phylogenetic trees were constructed with the ‘neighbor-joining’ method, with a bootstrap test of 1000 replicates. The evolutionary distances were computed using the Poisson correction method.

A BJ36 plasmid (Flores-Sandoval et al., 2015) containing a tripleVENUS-NLS (trpVNS) fragment was modified by adding a ligation-independent cloning (LiC) adapter site (De Rybel et al., 2011). A promoter fragment of MpFER of 3.2 kb was amplified by PCR using the primers listed in Table S2 and cloned into the BJ36 vector via LiC cloning. The _proMpFER promoter fragment fused to trpVNS was then shuttled via NotI restriction sites into the HART01 (Flores-Sandoval et al., 2015) expression vector. The expression vector was also directly modified to contain the LiC sites, so promoter fragments could be directly cloned in front of trpVNS in the HART01 vector via LiC cloning. This vector was called VHL (trpVNS-containing HART01 vector with LiC sites).

For functional analyses, three independent artificial microRNA (amiRNA) constructs were made (Fig. S2). The endogenous microRNA precursor MpmiR160 was used as both backbone and template to design the constructs, which generated single species small RNA molecules of 21 nt length, complementary to the target gene transcript (Flores-Sandoval et al., 2016; Rövekamp et al., 2016). Three miR160 backbones containing independent amiR-MpFER/amiR-MpFER* duplexes were synthesized by GenScript and fused to the _proMpEF1 promoter in the BJ36 shuttle vector and shuttled to the HART01 expression vector as described previously (Rövekamp et al., 2016). The folding structure and design procedures have been explained in detail previously (Flores-Sandoval et al., 2016; Rövekamp et al., 2016). The sequences for the amiRNAs were designed to retain the exact physical properties of the endogenous MpmiR160 template. The folding patterns were analyzed in Mfold software (Zuker, 2003) and the sequences synthesized (Fig. S2).

Selection of target sites was performed using CasFinder (Aach et al., 2014 preprint). Complementary oligos were designed and annealed for ligation into gRNA, in the pMpGE_En03 vector, previously digested with the BsaI restriction enzyme. Resulting gRNAs were incorporated into binary vector pMpGE011 through Gateway recombination.

The full-length MpFER sequence was amplified from genomic M. polymorpha DNA with attB sites and Gateway-cloned via pDONR207. For A. thaliana expression, MpFER was introduced into the expression vector pMDC201, which contains a 2X35S promoter fragment and an in-frame C-terminal mGFP6 (Curtis and Grossniklaus, 2003). For M. polymorpha expression, MpFER was introduced intro the binary vector pMpGWB308, which contains the _proMPEF1 promoter and an in-frame C-terminal Citrine (Ishizaki et al., 2015). Similarly, for expression of AtFER in M. polymorpha, the coding sequence of AtFER was introduced into the pMpGWB308 binary vector for the generation of the _proMpEF1:AtFER-Cit construct.

RNA extraction from thallus tissue (∼100 mg including apical notch) was carried out using the RNeasy Plant Mini Kit (74904, QIAGEN). To remove contaminating DNA, the TURBO DNA-free Kit (AM1907, Ambion) was used according to the manufacturer's recommendations. To extract RNA from the gametophores (two antheridiophores of >5 mm diameter were pooled, respectively, for each replicate), the Direct(-zol) RNA MiniPrep (R2050, ZymoResearch) was used with the TRIzol Reagent (15596026, Ambion) according to the manufacturer's protocol, including on-column DNase digestion and subsequent TURBO DNA-free Kit treatment. RNA samples were quantified using the Qubit RNA HS Assay Kit (Q32852, Life Technologies), a Nanodrop ND-1000 Spectrometer or an Agilent 2100 BioAnalyzer.

cDNA was synthesized from 0.5 μg or 1.0 μg of total RNA. Reverse transcription was performed in 25 μl with 20 μg/ml Oligo(dT) 12-18 primers and 200 units of SuperScript II Reverse Transcriptase (18064-014, Invitrogen) according to the manufacturer's protocol. The resulting cDNA was diluted 1:9 by adding nuclease-free water. To control for genomic DNA contamination, control replicates of all samples were incubated without SuperScript II reverse transcriptase.

Primer efficiency and concentration tests were carried out for MpFER and suitable reference genes as described previously (Rövekamp et al., 2016; Althoff et al., 2014). Amplification experiments were carried out using a 7500 Applied Biosystem Fast Real-Time PCR System. Reactions were performed in 20 μl volumes containing 10 μl 2× SYBR Green Master Mix (SsoAdvanced™ Universal SYBR Green Supermix), 200 nM (MpEF1 and MpFER) or 250 nM (MpACTIN) forward and reverse primers, and 1 μl diluted cDNA. Where possible, three technical and biological replicates were performed for each reaction. The primers used for qRT-PCR are summarized in Table S1.

Individual PCR reactions were performed in a total volume of 25 μl, using 1× ddPCR EvaGreen Supermix, with droplets generated according to the manufacturer's recommendations. Reading of the PCR-amplified droplets was carried out using a QX200 Droplet Reader (Bio-Rad) and analyzed using QuantaSoft™ Software (v1.4, Bio-Rad). The counts from ddPCR were normalized through a log2 transformation. Afterwards, relative expression was calculated against the geometric mean of the counts for all three reference genes (MpACT1, MpACT7 and MpAPT3) (Vandesompele et al., 2002): [log2(gene tested+1)−log2(geom. mean of all references+1)]. The primers used for the ddPCR are summarized in Table S1.

RNA from wild-type and Mpfer-1 mutant plants were extracted as previously described. A poly(G)-tail was added at the 3′ end of the RNA according to a Poly(A) Polymerase protocol (Thermo Scientific), using GTP instead of ATP. Reverse transcription was performed using anchored Oligo(dC) instead of Oligo(dT). Nested PCR on the cDNA was performed using two forward primers that matched the 3′ UTR sequence of MpFER (3′RACE-Fwd1 and 3′ RACE-Fwd2) and Oligo(dC) that are listed in Table S1. PCR fragments were purified and sequenced.

Agrobacterium-mediated transformation using regenerating thalli of M. polymorpha was performed by co-cultivation of 15-day-old plant fragments with transformed Agrobacterium tumefaciens (GV3101) cells (Kubota et al., 2013; Ishizaki et al., 2008). After 3 days of incubation, positive transformants were selected on Gamborg's B5 plates supplemented with 10 μg/ml hygromicin B (Invitrogen), 0.5 μM chlorosulfuron or kanamycin, and 100 μg/ml of cefotaxime sodium. Isogenic lines were obtained using plants derived from gemmae of the T1 generation (G1 generation). G1 or subsequent gemmae generations were used for all experiments, as they are derived from single cells (Barnes and Land, 1907; Hughes, 1971). Transformation of A. thaliana via A. tumefaciens (GV3101) was performed by floral dipping according to the method of Bent (2006). Primers used for amplification of promoter fragments of MpFER are listed in Table S2.

Plants were observed in a Lumar.V12 (ZEISS) or a Leica MZFLIII dissection microscope and imaged with an Axiocam HRc (ZEISS) or Leica DFC 420C camera. Clearings were observed using a Leica DMR microscope and photographed with an Axiocam 105 color camera. Fluorescence reporter expression was analyzed using a Leica SP5 confocal microscope. Fiji (Schindelin et al., 2012) and GIMP (version 2.8.10) software were used for adjustments of brightness, contrast, channel selection and image size.

M. polymorpha tissues were fixed in Carnoy's solution and incubated at 4°C overnight, followed by rehydration in an ethanol (EtOH) series of 85%, 70%, 50% and 30%. Samples were incubated for 1 h at 4°C for each step and then incubated in chloral hydrate solution at 4°C overnight. Samples were mounted in chloral hydrate.

Pieces of fresh thallus tissue located in fully differentiated zones (Solly et al., 2017) of plants grown on plates were dissected out and put into water. The pieces were transferred to PBS pH 6.1 containing 100 μg/ml Calcofluor White and vacuum-infiltrated for 1 h before being mounted on slides. The area of the epidermal cells was measured using Imaris 8.3.1. software (Bitplane AG). Cells of air pores and the two adjacent cell layers were excluded from the analysis, as well as the spiraling cells surrounding newly developing air pores.

Seven-day-old gemmalings were stained with lactophenol-Trypan Blue (10 ml of lactic acid, 10 ml of glycerol, 10 g of phenol, 10 mg of Trypan Blue, dissolved in 10 ml of distilled water) by boiling them for approximately 1 min in the stain solution and then decolorized in chloral hydrate (2.5 g of chloral hydrate dissolved in 1 ml of distilled water) for at least 30 min. They were mounted in chloral hydrate and imaged using a dissection microscope.

Twenty-day-old plants of wild type and _proMpEF1:MpFER-Cit lines #5 and #9 were harvested and fixed in glutaraldehyde (0.25% in 20 mM HEPES) overnight at 4°C. After fixation, dehydration with an acetone series (from 50% to 100% acetone ‘ultradry’) was carried out. The samples were then critical-point dried, without the following sputter-coating step. Microscopic imaging was performed on a JSM-6010 scanning electron microscope (JEOL) at the Center for Microscopy and Image Analysis at the University of Zurich.

To evaluate the stiffness of the gemmaling tissue, we used cellular force microscopy (Felekis et al., 2011; Vogler et al., 2013). A microelectromechanical systems (MEMS)-based FT-S100 force sensor (FemtoTools AG) with a glued-on tungsten probe (T-4-22; Picoprobe by GGB Industries) was used to indent the gemmaling tissue. The sensor has a force range of ±100 μN and a resolution of 15 nN measured at 500 Hz. The sensor was attached with a custom-made acrylic arm to an xyz positioner (SLC-2460-M, SmarAct), which was fixed on an xyz piezo stage (P-563.3CD PIMars; Physik Instrumente). The xyz positioner was used for the rough positioning of the sensor on a part of a gemma that was as much as possible in contact with the supporting glass slide and away from the meristematic zones. Nine evenly spaced measurements were then taken in a 3×3 grid, again using the xyz positioner to place the sensor, while the indentation at each position was driven by the piezo stage. This procedure was repeated a second time on each gemmaling, whenever possible on the opposite lobe.

The stiffness analysis was achieved with MATLAB (MathWorks) based on the slope fitted to the force-displacement curves, which were acquired by cellular force microscopy. In a first round of data curation, each force-indentation curve was visually inspected, and curves that did not have a clear contact point or showed a clear displacement of the gemmaling, indicated by an irregular slope, were discarded. The remaining data exhibited a bimodal distribution with a high density at very soft values around 1 µN, indicating the presence of an artifact due to gemmalings that were not in contact with the slide and, therefore, were displaced rather than indented. To identify such failed measurements, we pooled all data and fitted a Gaussian mixture model with two components. We then removed all measurements that were closer than three standard deviations to the mean of the distribution with the smallest mean (mean=1.16 µN and s.d.=0.56 µN). Hence, all measurements with a value below 2.9 µN were removed.

Boxplots and statistical analysis were performed in MATLAB. The statistical tests used to assess the significance of differences between groups are indicated in the figure legends.

M. polymorpha explants were ground in liquid nitrogen and resuspended in 300 μl of Laemmli buffer (Laemmli, 1970). After centrifugation at 6000 rpm (3800 g) for 5 min at 4°C, the supernatant was recovered. Protein accumulation was confirmed by western blotting using an anti-GFP antibody (1:3000; TP401, Torrey Pines Biolabs). As secondary antibody, a goat anti-rabbit antibody (1:20,000; S09602, Agrisera) was used followed by ECL detection using the FUSION FX - Western Blot & Chemi Imaging system (Vilber Lourmat).

Gemmae of wild-type, Mpfer-1 and independent transformation lines of amiR-MpFER lines were grown on plates and the thallus area was measured over time. Ten or 30 gemmae of four independent transformation lines were grown on half-strength Gamborg's B5 and scanned on an Epson 2450 Photo scanner at 600 dpi. Plant area was measured using color threshold settings and particle analysis in Fiji software (Schindelin et al., 2012).

Gene expression of CrRLK1L signaling pathway components was calculated from publicly available RNA-seq data of 39 samples of M. polymorpha Tak-1 and Tak-2. Samples were grouped by tissue and analyzed under the same pipeline: Nextera paired-end adapters were trimmed from sequencing reads using the bbduk tool embedded in the BBMap software package (Bushnell, 2014). Read ends with quality below 20 were also trimmed and the minimum read length was set to 25. The rest of parameters of bbduk were set at default values. Reads were mapped to M. polymorpha reference genome (v3.1) (Bowman et al., 2017) using the TopHat software (v2.1.1) (Trapnell et al., 2009), designed for RNA-sequencing read mapping. TopHat contains Bowtie (v2.3.2.0) (Langdon, 2015) as the aligner software. All parameters were set to default values except for the RAM memory and number of used Cores. Mapped reads were sorted using samtools (v1.3.1) (Li et al., 2009) and counted with the software HTSeq-count (v.0.9.1) (Anders et al., 2015). Gene expression values were calculated using the package DESeq2 (Release 3.6) (Love et al., 2014) for R software (v3.4.4) (https://www.R-project.org/). Another R package, ggplot2 (release 2.2.1) (Wickham, 2016), was used for producing the figures. A summary of the samples that were used in this study is provided in Table S3.

To measure the chlorophyll efflux rate (Lolle et al., 1997), 10-day-old gemmalings were used. Sixteen replicates of MpFER-overexpressor, Mpfer-2 mutant, and wild-type lines were immersed in an 80% EtOH solution in 5-ml tubes. The tubes were agitated gently on a shaker platform. Aliquots of 2 µl were removed every minute, during the first 10 min. The amount of chlorophyll extracted by the EtOH solution was quantified with a spectrophotometer (NanoDrop 8000, Thermo Fisher Scientific) and calculated from UV light absorption at 647 and 664 nm. The micromolar concentration of total chlorophyll per gram of fresh weight of tissue per ml of EtOH was calculated using the following equation: total micromoles chlorophyll=7.93(A₆₆₄)+19.53(A₆₄₇).

Comparative modeling of protein 3D structures was performed using the SWISS-MODEL online tool (Schwede et al., 2003). The software was fed with the three protein sequences of MpFER, MpRALF1 and MpLRE1 as input to model. The templates chosen for modeling the three sequences were selected from the best homology result provided by SWISS-MODEL (Table S4). Quality control of the model was assessed using the same tool and the results are summarized in Table S4. Spatial comparisons of the M. polymorpha predicted structures with the crystal structures of A. thaliana were performed in PyMOL (DeLano, 2002).

Analyses were performed in InfoStat (http://www.infostat.com.ar), the R base package and MATLAB. Statistical tests are indicated in the corresponding figure legend.

We thank Liam Dolan for kindly providing the Mpfer-1/Mpthe mutant, Valeria Gagliardini for help with qRT-PCR and ddPCR, Anja Grossniklaus for help in measuring epidermal cell area, Célia Baroux and Ethel Mendocillo-Sato for instructions on confocal microscopy and the use of the Imaris software, Tiago Meier for helping with sample preparation for scanning electron microscopy, Isabel Monte and Cyril Zipfel for comments on the manuscript, and Christof Eichenberger, Frédérique Pasquer, Arturo Bolaños, Daniela Guthörl, Daniel Prata, and Peter Kopf for general lab support.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.200580.reviewer-comments.pdf.