Interactive CLV3, CLE16 and CLE17 signaling mediates stem cell homeostasis in the Arabidopsis shoot apical meristem

In both animals and plants, pluripotent stem cell populations persist and continuously divide to supply building materials for growth and regeneration. Primary above-ground growth in plants is mediated by the shoot apical meristem (SAM), which is maintained by an intricate intercellular communication network (Aichinger et al., 2012; Heidstra and Sabatini, 2014; Steeves and Sussex, 1989). The most apical region of the SAM harbors the stem cell reservoir. Descendants of these cells that are pushed into the periphery undergo rapid divisions before differentiating to form leaf and flower primordia, the latter of which also harbor their own stem cell reservoir to produce floral organs before terminating in fruit formation. Stem cell identity is maintained by molecular factors originating from directly underlying cells, collectively called the organizing center (OC) (Barton, 2010; Gaillochet and Lohmann, 2015; Soyars et al., 2016). In Arabidopsis thaliana, the master regulator of stem cell identity in the SAM is WUSCHEL (WUS) (Laux et al., 1996). WUS is expressed in the OC, and encodes a homeodomain transcription factor that promotes stem cell identity and suppresses differentiation in the overlying cells (Mayer et al., 1998; Yadav et al., 2011). The WUS protein moves into the stem cell domain to directly activate the transcription of CLAVATA3 (CLV3), which encodes a pre-propeptide that is secreted to the extracellular space and processed into a small 12-13 amino acid peptide hormone (Fletcher et al., 1999; Ito et al., 2006; Ohyama et al., 2009; Ni et al., 2011; Rojo et al., 2002; Xu et al., 2013; Yadav et al., 2011). This mature peptide interacts with receptors in the underlying cell layers, where it limits stem cell accumulation by restricting the WUS expression domain to the OC (Brand et al., 2000). CLV3-WUS thus forms a core negative-feedback loop that maintains stem cell homeostasis in the SAM as well as in the transient floral meristem (Somssich et al., 2016; Dao and Fletcher, 2017). As a result, loss-of-function clv3 mutants have enlarged shoot meristems that produce flowers with extra organs, whereas CLV3 overexpression leads to premature SAM termination and growth arrest (Brand et al., 2000; Clark et al., 1996; Strabala et al., 2006).

Several complexes of leucine-rich repeat receptor-like kinases function to transduce the CLV3 signal. CLAVATA1 (CLV1) forms homodimers that localize to the cells directly below the CLV3 expression domain and physically interact with CLV3 peptide (Clark et al., 1997; Ogawa et al., 2008). CLAVATA2 (CLV2) and its co-receptor CORYNE (CRN) form heterodimers throughout the SAM and transduce the CLV3 signal independently of CLV1, although it is unclear whether CLV2/CRN directly binds CLV3 (Müller et al., 2008; Bleckmann et al., 2010; Guo et al., 2010). Three BARELY ANY MERISTEM (BAM) receptors localize to the periphery of the SAM in a complementary manner to CLV1 (DeYoung et al., 2006, 2008; Nimchuk et al., 2015). BAM1 binds CLV3 in vivo, and BAM1 and BAM3 expression is excluded from the SAM interior by CLV1 activity (Nimchuk, 2017; Shinohara and Matsubayashi, 2015). Whereas bam1/2/3 triple mutants show a reduction in SAM size, clv1 bam1/2/3 plants display a dramatic enhancement of stem cell accumulation beyond that of clv3 plants. Thus, a dual activity model suggests that under normal conditions, the BAM receptors prevent CLV3-CLV1 signaling on the flanks of the SAM to maintain stem cell accumulation. However, when CLV1 is absent, BAM gene expression extends into the SAM center, which allows the receptors to partially compensate for the absence of CLV1 by interacting with CLV3 peptides (Nimchuk, 2017).

CLV3 is a member of a family of polypeptides called the CLAVATA3/EMBRYO SURROUNDING REGION-RELATED (CLE) peptides (Cock and McCormick, 2001). CLE genes are conserved throughout the land plant lineage, and their origination is concurrent with the evolution of three-dimensional growth in plants (Goad et al., 2016; Whitewoods et al., 2018). The Arabidopsis genome contains 32 CLE sequences encoding 27 distinct peptides (Jun et al., 2008). Only a handful of CLE genes have been functionally characterized, all of which are involved in diverse developmental processes (Wang et al., 2016). However, the majority of cle single mutants have a wild-type phenotype, likely due to the high degree of redundancy typical of large gene families (Jun et al., 2010b). Indeed, evidence suggests that multiple CLE peptides may also function in the Arabidopsis SAM, as a previous study reported the promoter activity of several CLE genes in or around the shoot meristem (Jun et al., 2010b). A dodeca-cle mutant in which CLV3 and 11 other CLE genes were knocked out using genome editing showed a more severe meristem phenotype than clv3 plants, demonstrating that at least some of these 11 CLE peptides can partially compensate for CLV3 activity (Rodriguez-Leal et al., 2019). However, exactly which CLE genes are responsible for this compensation activity and the signaling pathway(s) through which they function are still unknown.

Previously, we analyzed lines containing null alleles of CLE16 and CLE17, which, despite their promoter activity in the SAM, displayed wild-type phenotypes under normal growth conditions (Gregory et al., 2018). In this study, we address the biological function of these CLE peptides by analyzing higher order cle16, cle17 and clv3 mutant combinations. Our work reveals that CLE16 and CLE17 partially compensate for CLV3 activity in stem cell maintenance beginning in embryogenesis and continuing throughout development, and that CLV3, CLE16 and CLE17 together limit axillary branch number. Using peptide treatments, we demonstrate that CLE16 and CLE17 peptides are perceived by the BAM receptor kinases but not by CLV1 or CLV2. Finally, we show that, like CLV3, CLE16 and CLE17 signal in the same pathway as WUS to restrict shoot and floral stem cell accumulation. These results show that CLE16 and CLE17 are part of a molecular mechanism that compensates for CLV3 activity to maintain stem cell homeostasis by acting through a subset of CLV3 receptors at the shoot apex.

In a previous comprehensive expression analysis, the CLE16 and CLE17 promoters were both shown to be active in the vegetative shoot apex (Jun et al., 2010b). To independently assess their expression dynamics, we performed mRNA in situ hybridization (ISH) to detect CLE16 and CLE17 transcripts in wild-type Col-0 vegetative meristems (VMs) and inflorescence meristems (IFMs). Neither the CLE16 nor CLE17 transcript was detectable in wild-type VMs (Fig. 1A,C). Sense probes derived from CLE16- and CLE17-coding sequences were used as negative controls and showed no non-specific hybridization (Fig. S1A,C). In wild-type IFMs, CLE16 and CLE17 antisense probes also produced signals indistinguishable from those of the sense probes (Fig. 1I,L, Fig. S1E,H). Our results indicate that, although their promoters are active in shoot apex tissues, CLE16 and CLE17 transcripts are expressed at levels too low to be detected by in situ hybridization in wild-type shoot meristems.

Considering the possibility that CLE peptides act redundantly in the shoot apex, we used in situ hybridization to examine CLE16 and CLE17 expression in clv3 and clv3 cle16 cle17 SAMs. We observed a broad signal for CLE16 transcripts in clv3 VMs, with the strongest expression in the outer cell layers as well as in the organ primordia (Fig. 1B). In contrast, we were unable to detect CLE17 transcripts in clv3 vegetative shoot apices (Fig. 1D). In clv3 IFMs, the CLE16 antisense probe produced a signal in L1 cells that resembled that of the sense control, suggesting it might be an artifact (Fig. 1J, Fig. S1F). In contrast, the CLE17 antisense probe produced a strong and specific signal in L1 cells (Fig. 1M, Fig. S1I). In clv3 cle16 cle17 IFMs, both CLE16 and CLE17 were strongly upregulated in the L1 cell layer compared with the negative controls (Fig. 1K,N, Fig. S1G,J). These results indicate that CLE16 and CLE17 transcription is dynamically repressed by CLV3 signaling during different phases of development. CLE16 but not CLE17 expression is downregulated by CLV3 during the vegetative stage, whereas CLE17 but not CLE16 expression is downregulated during the reproductive stage. Detection of both transcripts in clv3 cle16 cle17 IFMs suggests that CLE16 and CLE17 signaling pathways potentially regulate themselves and/or each other.

Consistent with previous reports, CLV3 and WUS transcripts were detected in the stem cells and organizing center, respectively, of Col-0 VMs and IFMs (Fig. 1E,G,O,R). The CLV3 and WUS expression domains were likewise expanded in clv3 and clv3 cle16 cle17 VMs and IFMs (Fig. 1F,H,P,Q,S,T), indicating that CLE16 and CLE17 regulate stem cell homeostasis in a manner largely analogous to that of CLV3.

We also re-investigated CLE16 and CLE17 promoter activity in shoot apices of pCLE16:GUS-eGFP and pCLE17:GUS-eGFP reporter lines (Jun et al., 2010b) using confocal laser scanning microscopy (CLSM). Although neither reporter could be detected in wild-type or clv3 VMs, we detected both pCLE16:eGFP and pCLE17:eGFP signal in the L1 layer of wild-type IFMs (Fig. S1K,M), and pCLE17:eGFP but not pCLE16:eGFP signal in the L1 layer of clv3 IFMs (Fig. S1L,N,O). However, signals for both reporters were discontinuous across the L1 layer. The disruption of pCLE16:eGFP and pCLE17:eGFP patterns across the L1 layer, as well as their incomplete overlap with the mRNA transcription patterns, suggest that these reporter constructs do not carry the full regulatory region. Nevertheless, both the in situ hybridization and GFP reporter experiments agree on CLE16 and CLE17 being expressed in the L1 layer of the IFM.

Finally, we examined CLE16 and CLE17 expression in a published gene expression map of Arabidopsis shoot apex domains (Tian et al., 2019). CLE16 and CLE17 showed the highest expression levels in the L1 layer and very low expression levels in the other regions of the SAM (Fig. S2A,B), whereas CLV3 transcripts accumulated specifically in the stem cells (Fig. S2C). The absolute value of CLV3 transcript levels as normalized RPKM was higher than that of both CLE16 and CLE17 (CLV3, 30.29; CLE16, 14.74; CLE17, 24.55, Fig. S2A-C). Together, these data indicate that CLE16 and CLE17 are expressed at very low levels in the SAM, and their transcripts are concentrated in the L1 layer of the shoot apex.

CLE16 and CLE17 encode mature CLE peptides that differ from one another by a single amino acid and differ by five amino acids relative to the CLV3 peptide (Fig. S3A). Unlike clv3 mutants, cle16 and cle17 null mutants do not display shoot apical meristem phenotypes in Col-0 under normal growth conditions (Gregory et al., 2018). However, because the CLE16 and CLE17 promoters displayed overlapping expression patterns in the SAM, we hypothesized that the two genes function redundantly with each other and/or with CLV3. We tested this by generating cle16-2 cle17-2, clv3-15 cle16-2, clv3-15 cle17-2 and clv3-15 cle16-2 cle17-2 plants carrying null alleles of CLV3, CLE16 and/or CLE17 (Table 1, Fig. S3B,C). We then compared SAM size across the embryonic, vegetative and reproductive stages in these cle16 cle17, clv3 cle16, clv3 cle17 and clv3 cle16 cle17 mutants, and in wild-type Col-0 plants.

The CLV-WUS signaling pathway limits shoot apical meristem stem cell accumulation as early as the mature embryo stage (Schoof et al., 2000). We visualized the SAMs of mature embryos carrying the above allelic combinations using CLSM and found that the dimensions of cle16 cle17 embryo SAMs were indistinguishable from those of the wild type (Fig. 2A,B,S). In contrast, the diameter and height of clv3 embryo SAMs was significantly greater than wild type (Fig. 2A,C,S). Neither clv3 cle16 nor clv3 cle17 embryos displayed dramatic changes in SAM size relative to clv3 embryos (Fig. 2C-E,S), although clv3 cle16 embryos did show a small but significant increase in SAM diameter (P=0.03). In contrast, clv3 cle16 cle17 embryos formed SAMs that were significantly taller and broader than those of clv3, but not definitively larger than clv3 cle16 or clv3 cle17 embryos (Fig. 2C-F,S). These results show that CLE16 and CLE17 together restrict stem cell accumulation in the absence of CLV3 and that this begins during embryogenesis. In addition, they suggest that either CLE16 and CLE17 act redundantly with one another at the embryonic stage, or their individual contributions are too subtle to measure at this stage under our experimental conditions.

At 7 days after germination (DAG), the vegetative meristems (VMs) of cle16 cle17 plants remained indistinguishable from those of wild-type plants, whereas clv3 plants showed a dramatic VM size enhancement (Fig. 2G-I,T). clv3 cle16 VMs were significantly larger in both height and diameter than clv3 VMs (Fig. 2I,J,T), whereas clv3 cle17 VMs were significantly wider but not taller than clv3 VMs (Fig. 2I,K,T). clv3 cle16 cle17 seedlings produced VMs that were significantly larger in all dimensions than those of clv3 VMs, but were indistinguishable from clv3 cle16 or clv3 cle17 VMs (Fig. 2I-L,T). These results suggest that CLE16 and CLE17 independently restrict stem cell accumulation in clv3 seedlings during vegetative development.

Analysis of IFM height and projected surface area revealed no differences in the dimensions of cle16 cle17 IFMs compared with wild type (Fig. 2M,N,U, Fig. S4A,B,G,H). In contrast, clv3 plants formed fasciated IFMs that were significantly larger in both surface area and height than those of wild-type IFMs (Fig. 2M,O,U, Fig. S4A,C,G,I). Although clv3 cle16 plants did not exhibit increased IFM height or surface area compared with clv3 plants, clv3 cle17 plants showed significantly increased IFM surface area (P<10⁻⁴, Fig. 2O-Q,U, Fig. S4C-E,I-K), and clv3 cle16 cle17 plants formed the largest IFMs on average (Fig. 2O-R,U, Fig. S4C-F,I-L). Both CLE16 and CLE17 therefore appeared to limit IFM activity in the absence of CLV3 activity during reproductive development, with CLE17 contributing more to stem cell restriction at this stage than CLE16.

Taken together, our results indicate that CLE16 and CLE17 each function to control stem cell accumulation in shoot meristems throughout development in the absence of endogenous CLV3 activity. Furthermore, clv3 cle17 plants form larger IFMs than clv3 cle16 plants, indicating that CLE17 plays a greater role than CLE16 in restricting stem cell accumulation during reproductive development.

Because shoot meristematic activity directly influences the production of lateral organs such as leaves, axillary branches and flowers, we also quantified these traits in Col-0, cle16 cle17, clv3, clv3 cle16, clv3 cle17 and clv3 cle16 cle17 plants. Rosette leaf number was quantified at 21 DAG, once the plants had completed vegetative development and entered the reproductive phase. Col-0 and cle16 cle17 plants produced similar numbers of rosette leaves, whereas rosette leaf production was significantly enhanced in clv3 plants (Fig. 3A). clv3 cle16 plants produced a similar number of rosette leaves to clv3 plants, but clv3 cle17 plants generated significantly more leaves and clv3 cle16 cle17 plants had the highest rate of rosette leaf production (Fig. 3A). Thus, CLV3 plays a role in limiting rosette leaf production during vegetative growth, most likely tied to its role in restricting shoot stem cell accumulation. CLE17 performs this function in the absence of CLV3, and CLE16 makes a minor contribution in the absence of the other two genes.

To measure axillary branching, plants were grown until 35 DAG, when axillary stems extending from the rosette were counted (Fig. 3B). Col-0 and cle16 cle17 plants displayed a similar extent of axillary stem outgrowth, with an average of two per plant. In contrast, very little axillary stem outgrowth was observed in clv3, clv3 cle16, clv3 cle17 and clv3 cle16 cle17 plants (Fig. 3B). Thus, the number of visible axillary stems does not correlate with rosette leaf number, but rather their outgrowth appears to be inhibited in plants lacking CLV3 activity.

We next examined cauline leaf production and branch outgrowth from cauline leaf axils during reproductive development. At 35 DAG, Col-0 and cle16 cle17 plants produced similar numbers of cauline leaves (Fig. 3C), whereas both clv3 and clv3 cle16 plants formed more than twice as many cauline leaves as wild-type plants. The cauline leaf production rate was significantly higher in clv3 cle17 plants and the highest in clv3 cle16 cle17 plants (Fig. 3C), mirroring the phenotypes observed for the rosette leaves. With the exception of the triple mutant, the mean number of cauline axillary branches equaled the number of cauline leaves in all genotypes (Fig. 3D). In clv3 cle16 cle17 plants, a few cauline leaves in each plant developed without a visible AM branch. As a result, despite having more cauline leaves, the mean number of visible AM buds in clv3 cle16 cle17 plants was similar to that of clv3 cle17 plants (Fig. 3C,D). Together, our data indicate that CLV3, CLE16 and CLE17 together limit cauline axillary branch number by limiting the rate of cauline leaf production from the inflorescence meristem.

Finally, because the CLV-WUS signaling pathway is also active in the floral meristem (FM), we tested whether CLE16 and/or CLE17 could compensate for CLV3 activity during flower development. We used carpel number as a readout for floral meristem activity, because larger FMs produce more floral organs, and the carpels are the final whorl of organs to form as the FM undergoes termination. Whereas wild-type and cle16 cle17 plants invariably made two carpels per flower, clv3 plants made an average of one additional carpel (Fig. 3E-G,K). Mean carpel number was significantly enhanced in clv3 cle16 and clv3 cle17 flowers compared with clv3 flowers, whereas clv3 cle16 cle17 flowers generated significantly more carpels than either of the double mutants (Fig. 3G-K). We repeated this experiment with an independent set of null alleles (clv3-10, cle16-3 and cle17-3, Fig. S3C, Table 1) and obtained similar results (Fig. S5). These data suggest that CLE16 and CLE17 also compensate for CLV3 signaling to restrict floral organ production.

In addition to our observation that CLV3 signaling represses CLE16 transcript levels in vegetative meristems and CLE17 transcript levels in reproductive meristems, we explored the possibility that CLV3 might also interfere with the perception and/or transduction of CLE16 and CLE17 signals at the post-translational level. To test this scenario, we artificially induced CLE signaling in the SAM by germinating Arabidopsis seedlings on agar plates containing 30 μM of synthetic CLV3, CLE16 or CLE17 peptides (Table 2). We used the unmodified 12 amino acid CLE sequences of the three peptides, as this region of the CLV3 peptide is of comparable biological potency to the mature arabinosylated peptide at the concentration used (Kim et al., 2017). We also used a peptide-free water treatment and a scrambled CLV3 (sCLE) amino acid sequence as negative controls.

We germinated Col-0, cle16 cle17, clv3, clv3 cle16, clv3 cle17 and clv3 cle16 cle17 seeds on the peptide plates and measured vegetative SAM size in 7 DAG seedlings. Seedling SAM size among the genotypes was indistinguishable when grown on mock-treated or on sCLE plates (Fig. 4A-L), indicating that the scrambled peptide had no biological activity and that the experimental conditions had no effect on meristem function. In contrast, treatment with CLV3 peptide strongly reduced the SAM size of all genotypes, suggesting that the synthetic CLE peptides were effectively transported to the shoot apex, and that the unmodified CLV3 dodecapeptide was sufficient to limit WUS and stem cell activity in the SAM. In contrast, treatment with CLE16 or CLE17 peptides did not affect wild-type or cle16 cle17 seedling meristem size (Fig. 4A,B,G,H), indicating that neither CLE16 nor CLE17 synthetic peptides restrict SAM size in these genetic backgrounds. However, application of either CLE16 or CLE17 peptides significantly reduced SAM size in clv3 plants compared with mock or sCLE treatments, albeit not to the same efficacy as CLV3 treatment (Fig. 4C,I). Similarly, CLE16 and CLE17 application significantly restricted stem cell accumulation in clv3 cle16, clv3 cle17 and clv3 cle16 cle17 plants (Fig. 4D-F,J-L). These results suggest that endogenous CLV3 signaling prevents the biological activity of CLE16 and CLE17 peptides in the SAM.

To test the combined effect of CLE16 and CLE17 peptide treatment on meristem activity, we performed a peptide treatment of 7 DAG Col-0, clv3 and clv3 cle16 cle17 plants using 15 μM of CLV3, CLE16 or CLE17 synthetic peptides separately as well as a combination of 15 μM CLE16 and 15 μM CLE17 peptides (Fig. S6). Wild-type SAM height and diameter were restricted only by the application of CLV3 peptide (Fig. S6A), whereas clv3 and clv3 cle16 cle17 SAM size was restricted by all peptide treatments (Fig. S6B,C). In the last two backgrounds, CLV3 application had the strongest effect, while application of CLE16, CLE17 or CLE16+CLE17 peptides produced equivalent results (Fig. S6B,C). Notably, treatment with CLE16 and CLE17 together did not reduce SAM size more than treatment with either peptide alone, despite the total peptide concentration being higher. This observation suggests that the peptides reached an endogenous saturation level, and that CLE16 and CLE17 may act through the same signaling components that have rate-limiting capacity in these assays.

The CLV3 signal at the shoot apex is transduced by a suite of receptor-like kinase complexes, including CLV1, BAM1, BAM2 and BAM3, and potentially CLV2-CRN. Our peptide treatment experiment provides evidence that CLV3 blocks CLE16 and CLE17 signal transduction. Therefore, we tested the hypothesis that CLE16 and CLE17 could act through the same receptors as CLV3 by performing peptide treatment assays using clv1-11, clv2-3 and bam1-4 bam2-4 bam3-2 loss-of-function receptor mutants, as well as clv1 clv3, clv2 clv3 and bam1/2/3 clv3 higher order mutant combinations.

Quantification of VM size in 5 DAG seedlings showed that treatment with synthetic CLV3 peptide significantly reduced both the height and diameter of clv1 SAMs compared with mock treatment (Fig. 5A,I). The observation that clv1 mutants respond to CLV3 signaling is consistent with previous research showing that the BAM receptors compensate for the loss of CLV1 function (Nimchuk, 2017). CLE16 and CLE17 treatment had no effect on clv1 SAMs, most likely due to endogenous CLV3 activity (Fig. 5A,I). On the other hand, SAM size was effectively restricted when clv1 clv3 plants were exposed to CLV3, CLE16 or CLE17 peptides (Fig. 5B,J), indicating that CLV1 is also dispensable for CLE16 and CLE17 signal transduction. clv1 clv3 SAMs grown on CLE16 or CLE17 plates were not significantly different from those grown on CLV3 plates (Fig. 5B,J), suggesting that CLV1 receptor activity is responsible for most of the observed gap in biological efficacy between CLV3 and the CLE16 and CLE17 peptides.

Our assays revealed a complex role for the CLV2 receptor-like protein in CLE peptide perception. clv2 SAM size was unaffected by CLV3, CLE16 or CLE17 peptide treatment (Fig. 5C,K). These data suggest that the CLE16 and CLE17 peptides can signal through CLV2, and additionally that either CLV2 is required to sense the synthetic CLV3 peptide or CLV2 controls stem cell accumulation independently of CLV3 signaling. However, clv2 clv3 SAMs were clearly susceptible to the meristem-restricting effects of the CLV3, CLE16 and CLE17 peptides (Fig. 5D,L). CLV2 therefore appears to be only partially responsible for the full signal transduction of the CLV3, CLE16 and CLE17 peptides in VMs. In addition, whereas clv1 clv3 SAMs were very similar in size to clv3 SAMs (Fig. 5B,H,J), clv2 clv3 SAMs were significantly enlarged compared with clv3 SAMs (Fig. 5D,H,L). This observation again indicates that at least part of the meristem-restricting activity of the CLV2 pathway occurs independently of CLV3 signaling.

All genotypes examined exhibited a similar response to CLE16 and CLE17 peptide treatments, except for clv2 clv3 SAMs, which were more effectively restricted by CLE17 than by CLE16 application (Fig. 5D,L). Indeed, CLE17 was as effective as CLV3 in reducing SAM size in clv2 clv3 plants (Fig. 5D,L). Thus, CLV2 might play a role in monitoring the stem cell response to the CLE16 and CLE17 peptides during vegetative development, either by increasing sensitivity to CLE16 or by dampening sensitivity to CLE17.

CLV3, CLE16 and CLE17 peptide treatments had distinct effects in the bam mutant background. bam1/2/3 SAMs (Fig. 5E) were smaller than wild-type SAMs (Fig. 4A) when grown on mock-treated plates (Fig. 5G,M), consistent with previous research suggesting a role for the BAM receptors in preventing excessive CLV3 signaling in the SAM periphery (Nimchuk et al., 2015). Like clv1, bam1/2/3 SAM size was also restricted by exogenous CLV3 treatment, but was resistant to CLE16 and to CLE17 application (Fig. 5E,M). CLV3 treatment was also able to significantly reduce bam1/2/3 clv3 SAM size (Fig. 5F,N). The BAM receptors are therefore not required for CLV3 signaling, a role that under normal circumstances is already performed by CLV1. In contrast, bam1/2/3 clv3 SAMs were unaffected by the presence of CLE16 and CLE17 peptides (Fig. 5F,N). These results suggest that the BAM receptors are required to transmit the CLE16 and CLE17 signals. Finally, clv1 bam1/2/3 plants were resistant to CLV3, CLE16 and CLE17 peptide application (Fig. 5O,P), indicating that these receptor kinases together play a crucial role in transducing multiple CLE signals at the shoot apex.

Although CLV3, CLE16 and CLE17 have differential interactions with downstream receptors, our current understanding of CLV1 and the BAM receptors suggests that these receptors all function upstream of the WUS transcription factor in the SAM. Therefore, we assessed whether CLE16 and CLE17 act in the same genetic pathway as WUS by crossing cle16 cle17, clv3, clv3 cle16, clv3 cle17 and clv3 cle16 cle17 plants into the wus-1 null mutant introgressed into the Col-0 background. Loss of WUS function leads to a SAM maintenance deficiency following initiation (Laux et al., 1996). As a result, homozygous wus SAMs typically produce a few rosette leaves during vegetative growth and then prematurely arrest. Indeed, at 14 DAG, a wild-type Col-0 population segregating for the wus-1 allele displayed a 3:1 ratio of wild-type to meristem-deficient phenotypes, with precocious termination of the primary shoot (Fig. 6A, Fig. S7A, Table 3). cle16 cle17, clv3, clv3 cle16, clv3 cle17 and clv3 cle16 cle17 populations segregating for the wus-1 allele showed the same proportion of plants with terminated primary shoots (Fig. 6A, Fig. S7B-F, Table 3), indicating that wus is epistatic to the clv3, cle16 and cle17 alleles. As a proxy for shoot meristem activity, we quantified the number of rosette leaves from these segregating populations, and observed that all individuals homozygous for wus-1 produced the same limited number of rosette leaves, regardless of whether they carried the clv3, cle16 and/or cle17 alleles (Fig. 6B). In contrast, plants that were not homozygous for wus-1 produced many additional rosette leaves. Col-0 and cle16 cle17 plants formed comparable numbers of rosette leaves, whereas plants homozygous for the clv3 allele generated significantly more. Average rosette leaf number was similar between clv3 and clv3 cle16 plants, significantly higher in clv3 cle17 plants, and higher still in clv3 cle16 cle17 plants (Fig. 6B). Therefore, WUS acts downstream of CLE16 and CLE17 as well as of CLV3, and the function of CLE16 and CLE17 in primary SAM maintenance is dependent upon WUS activity.

Finally, we explored the possibility that, like CLV3, CLE16 and CLE17 expression is regulated by WUS activity. We queried a published WUS-regulated transcriptome dataset (Yadav et al., 2013), and found that neither CLE16 nor CLE17 was among the lists of transcripts directly or indirectly regulated by WUS. Although it is possible that CLE16 and CLE17 expression is too low to be detected in this experiment, we currently lack evidence that WUS regulates CLE16 and CLE17 transcription in the SAM.

Plant stem cell reservoirs require molecular signals from neighboring cells to maintain homeostasis and continuously supply new cells for growth and organogenesis. In this study, we identified two CLE peptides, CLE16 and CLE17, that play important roles in buffering shoot stem cell homeostasis in the absence of CLV3 activity. CLE16 and CLE17 act in the same genetic pathway as the BAM receptor kinases and the WUS transcription factor to control stem cell accumulation in the SAM.

Our previous study showed the CLE16 and CLE17 promoters are active in overlapping domains in vegetative and reproductive SAMs (Jun et al., 2010b). Because mature CLE peptides are cleaved from both ends of their precursor proteins (Ni et al., 2011; Xu et al., 2013), translational fusions of CLE-coding sequences to fluorescent reporters currently do not exist. We therefore examined the CLE16 and CLE17 expression patterns in greater detail using in situ hybridization. Although we were unable to detect CLE16 or CLE17 mRNA transcripts in Col-0 VMs or IFMs, a published RNAseq-based gene expression database of Arabidopsis shoot apex domains shows that CLE16 and CLE17 transcripts accumulate in the outer L1 layer while being expressed at very low levels in the rest of the wild-type SAM (Tian et al., 2014). Consistently, CLE16 mRNA expression was observed throughout the outer cell layers of clv3 VMs and in the L1 cells of clv3 IFMs (Fig. 1), whereas CLE17 mRNA was not detected in clv3 VMs but was present in the L1 cells of clv3 IFMs. CLE16 was more strongly upregulated in the vegetative stage, and CLE17 in the reproductive stage, in clv3 IFMs (Fig. 1), consistent with CLE17 contributing more than CLE16 to restricting stem cell accumulation during reproductive development (Figs 2 and 3). Finally, both transcripts were detected in the L1 layer of clv3 cle16 cle17 IFMs. We observed an imperfect overlap between the activities of the reporter lines and the transcript accumulation patterns (Fig. S1), suggesting that cis-regulatory elements downstream of the transcription start site, which are absent from the reporter constructs, may be required for the dynamic pattern of CLE16 and CLE17 expression within the shoot apex (Fig. 7A). Such cis-elements exist downstream of the CLV3 stop site and are required for its dose-dependent regulation by WUS (Perales et al., 2016). RT-qPCR assays did not detect upregulation of CLE16 or CLE17 in clv3 plants (Rodriguez-Leal et al., 2019). However, given the dramatic change in tissue size caused by the loss of CLV3 function, and the higher sensitivity of in situ hybridization and reporter line analysis compared with other methods, we provide evidence that CLE16 and CLE17 expression is dynamically regulated by CLV3 activity. Owing to the limitations of both in situ hybridization and RT-qPCR, methods such as FACS-based transcriptional profiling and/or single-cell RNA-seq will need to be employed to confirm the extent to which CLE16 and CLE17 are dynamically regulated within their respective expression domains.

Although CLE16 and CLE17 are dispensable for maintaining SAM function on their own (Gregory et al., 2018), our functional analysis reveals that, in the absence of CLV3, both peptides play an important role in limiting shoot stem cell accumulation throughout the entire life cycle, with CLE17 contributing more to restricting stem cell activity than CLE16 in inflorescence and floral meristems. The major downstream output of the CLV3 signaling pathway is the negative regulation of the WUS expression domain (Mayer et al., 1998). WUS likewise acts downstream of CLE16 and CLE17 in the SAM, because in the absence of WUS even clv3 cle16 cle17 plants are unable to sustain primary SAM growth (Fig. 6, Fig. S7). Such conditional cross-complementation among the CLV3 and CLE16/CLE17 signaling pathways may contribute to plant resiliency by buffering a critically important stem cell signaling network.

CLE16 and CLE17 signaling in the SAM is blocked at the post-translational level by endogenous CLV3 activity, because treatment with synthetic CLE16 or CLE17 peptides did not alter SAM activity in wild-type or cle16 cle17 plants yet effectively restricted stem cell accumulation in clv3 mutants (Fig. 4). Although CLE17 limits stem cell activity more effectively than CLE16 during reproductive development, we see no evidence that the application of CLE17 peptide is more effective than that of CLE16 peptide in clv3 seedlings (Fig. 4). Therefore, the biological response to endogenous CLE16 and CLE17 signaling may depend on a combination of dose, expression patterns and other dynamic downstream components that can act in a stage-specific manner.

CLV3 peptide can block CLE16 and CLE17 activity either by outcompeting them for binding to the same receptor kinases or by preventing the biological activity of their cognate receptors in the SAM. Our peptide assays indicate that the BAM receptors, but not CLV1, transmit CLE16 and CLE17 signals in the SAM, consistent with CLE16 binding to the extracellular domain of the BAM1 receptor in vivo (Crook et al., 2020). These data are consistent with a CLE stem cell signaling model in which CLV1 is a specific receptor for CLV3 and CLV3-CLV1 signaling prevents BAM expression in the central SAM region (Fig. 7B). In the absence of CLV3 signal, the BAM gene expression domains extend into the SAM interior, enabling signaling via both CLV1 and the BAM receptors to affect stem cell activity. This model can explain our observation that synthetic CLV3 peptide is more effective than either CLE16 or CLE17 peptide at restricting stem cell accumulation in clv3 plants, as CLV3 is able to interact with both CLV1 and the BAM proteins, whereas CLE16 and CLE17 signal only through the BAM receptors.

Although peptide binding assays analyzing the specificity of the CLV1 and BAM receptors for the CLE16 and CLE17 peptides will be needed to further evaluate this scenario, it is tempting to speculate that the binding of CLE16 and CLE17 to the BAM receptors may be part of a mechanism that prevents CLV3 signaling on the meristem flanks. cle16 cle17 SAMs are indistinguishable from wild-type SAMs (Figs 2–4), suggesting that CLE16 and CLE17 alone are not sufficient to perform this role. Yet, in addition to CLV3, CLE16 and CLE17, other CLE peptides are perceived by the CLV1 and/or BAM receptors in the SAM, as both clv3 cle16 cle17 plants and dodeca-cle plants display weaker meristem phenotypes than clv1 bam1/2/3 plants that show severe defects in shoot architecture and flower development (Rodriguez-Leal et al., 2019). CLE16 and CLE17 are not among the genes mutated in the dodeca-cle background, suggesting some of these other CLE peptides may function redundantly with CLE16 and CLE17 to signal through the BAM receptors and regulate peripheral shoot stem cell activity.

Our work also sheds new light on the role of CLV2 in SAM stem cell signaling. CLV2 is implicated in various developmental processes in planta, and its function in the SAM requires interaction with other signaling components, including CRN, CLV1 and the BAM receptors (Bleckmann et al., 2010; Zhu et al., 2010). Although CLV2 has not been shown to bind the arabinosylated 13 amino acid CLV3 peptide, it is capable of binding the non-arabinosylated 12 amino acid CLV3 peptide form, as well as other CLE peptides with varying affinities (Guo et al., 2010; Shinohara and Matsubayashi, 2015). Our analysis shows that untreated clv2 clv3 SAMs are larger than untreated clv3 SAMs, and clv2 SAM size is minimally affected, whereas clv2 clv3 SAMs are effectively restricted by application of 12 amino acid CLV3, CLE16 or CLE17 peptides (Fig. 5). We also observe that clv2 clv3 plants respond differently to CLE16 and CLE17 peptide treatments than do other genotypes. Whereas clv1 clv3 and bam1/2/3 clv3 plants respond similarly to CLE16 and CLE17 application, clv2 clv3 plants are more strongly affected by CLE17 than CLE16 peptide. Thus, during vegetative development, CLV2 may dampen the stem cell-restricting effect of CLE17 or promote that of CLE16, perhaps by forming complexes with additional receptors and mediating their binding specificity. Our analysis therefore supports a dual function for CLV2: one that is independent of CLV3/CLE16/CLE17 signaling, and one that involves fine-tuning the meristem response to CLE16 and CLE17 in the absence of CLV3 activity.

Redundancy and compensation among CLE peptides and their cognate receptors appear to be a common feature throughout the land plant lineage. A dual CLE signaling system exists in the maize SAM, where the ZmFCP1 and ZmCLE7 peptide signals are transduced by a pair of receptor kinases (Je et al., 2016, 2017). In the tomato SAM, SlCLV3 is the dominant signal that restricts stem cell accumulation and SlCLE9 buffers the loss of SlCLV3 activity (Rodriguez-Leal et al., 2019). Observations in the moss Physcomitrella patens also suggest that multiple CLE peptides may control apical cell activity in leafy shoots via the partially redundant receptors PpCLV1a/b and PpRPK2 (Whitewoods et al., 2018). Plant stem cell activity underlies multiple yield traits, such as branch number, fruit and seed number, and fruit size, which are crucial for agricultural productivity (Li et al., 2021; Wang et al., 2021). It is therefore important to identify all of the genetic factors that control stem cell function, because manipulating any number of them could lead to agronomic improvements. By identifying two additional Arabidopsis CLE genes that signal to mediate stem cell activity, this work significantly expands our repertoire of candidate targets for crop yield enhancement.

All Arabidopsis thaliana plants were in the Columbia-0 (Col-0) accession. The clv3, cle16, cle17 and bam alleles were generated in Col-0, whereas the clv1, clv2 and wus alleles were generated in Landsberg erecta (L-er) and introgressed three or four times into Col-0 prior to analysis. Plants were grown on a mixture of 1:1:1 perlite:vermiculite:top soil under long day conditions (16 h light/8 h dark, light intensity ∼120 μmol/m²/s) at 21°C. Seeds were stratified at 4°C for 5 days before exposure to light. Seedlings were watered daily with a 1:1500 dilution of Miracle-Gro 20-20-20 fertilizer prior to flowering and once a week with fertilizer thereafter. Homozygous plants were confirmed by PCR-based genotyping and restriction digestion (primers and restriction enzymes are listed in Table S1).

In situ hybridization was carried out as described previously (Jun et al., 2010a) with a few modifications in the fixation process. Samples were fixed in a neutral buffered formalin solution (Sigma) overnight at 4°C, washed in methanol at least three times for 15 min each, washed in ethanol twice for 10 min each, then transferred to a Leica TP1020 tissue processor for automated tissue infiltration into McCormick Paraplast X-tra. Samples were sectioned at 8 µm and transferred to microscope slides. After proteinase K digestion, samples were hybridized overnight at 55°C with sense and antisense RNA probes labeled with digoxigenin (Roche) in a hybridization solution of 50% deionized formamide, 6×SSC, 3% SDS and 0.5 mg/ml tRNA. After washing with 2× SSC and with 0.2× SSC and 0.2% SDS at 55-60°C, probes were detected by incubation with anti-digoxigenin antibody linked to alkaline phosphatase (Roche) for 60 min to overnight. Visualization was carried out by incubation with nitro blue tetrazolium chloride – 5-bromo-4-chloro-3-indolyl-phosphate (NBT-BCIP, Roche) for 4-48 h after washing. Tissue sections were imaged using a Leica DM4000 B light microscope (Leica Microsystems).

For peptide treatment of seedlings, Murashige and Skoog (MS) growing medium and 0.7% agar (pH 5.7) were prepared and autoclaved. Before pouring, either water (mock) or a 3 mM solution of either synthetic scrambled peptide (sCLE), CLV3, CLE16 or CLE17 peptide (Genscript, purity ≥90%, no residue modification) in water was added to the MS medium to a final concentration of 30 µM. Arabidopsis seeds were sterilized in 70% ethanol followed by 10% bleach and 0.2% SDS, and then plated on mock-treated or peptide-treated MS plates. Plants were grown under long-day conditions and the shoot regions of the seedlings harvested 5 days after germination for confocal microscopy.

Propidium iodide (PI) staining of whole seedlings or organs was carried out as described by Truernit et al. (2008). To observe embryonic meristems, Arabidopsis seeds were imbibed in water overnight at 4°C, and embryos were dissected from the seed coat for PI staining. Embryos and vegetative tissues were then cleared in methyl salicylate, whereas inflorescences were partially embedded in 1% agarose in an upright position and submerged in water, before mounting on depression slides. Samples were visualized using a Leica TCS-SP8 spectral confocal laser scanning microscope (Leica Microsystems) with a 40× lens. The excitation/emission wavelengths were 488 nm/600-650 nm for PI and 488 nm/500-530 nm for eGFP.

Shoot meristem measurements were performed as described previously (Monfared and Fletcher, 2014) using Fiji software. Parameters for each meristem were determined by defining three sets of coordinates: one for the apical point; and two for the notches where the meristematic zone ends and organ primordia begins. Meristem height and diameter were calculated as the height and base, respectively, of the resulting triangle. Z-stacks of inflorescence meristems were projected into 2-D images for surface area measurement, and reconstructed in Fiji's Volume Viewer for height measurement. Floral organ counting was performed as previously described (Fiume et al., 2011).

We thank Annis Richardson for technical consultation with CLSM, and Zachary Nimchuk and Sarah Hake for helpful comments on the manuscript. The authors acknowledge the Confocal Microscopy Facility of the Center for Genome Research and Biocomputing at Oregon State University.