The functional specificity of ERECTA-family receptors in Arabidopsis stomatal development is ensured by molecular chaperones in the endoplasmic reticulum

Stomata are epidermal pores that play important roles in gas exchange between plants and the atmosphere. In Arabidopsis, a subpopulation of protodermal cells adopts the stomatal lineage identity to become meristemoid mother cells (MMCs). An MMC undergoes asymmetric cell division to create two daughter cells with distinct identity: a smaller meristemoid and a larger stomatal lineage ground cell (SLGC). Both meristemoids and SLGCs have the potential to further divide asymmetrically, but each has a different cell-fate trajectory. Meristemoids differentiate into guard mother cells (GMCs) and eventually produce a pair of kidney-shaped guard cells (GCs) after GMC symmetric division. SLGCs ultimately become puzzle-shaped pavement cells in the epidermis (Nadeau and Sack, 2002; Dong and Bergmann, 2010; Torii, 2012).

Stomatal formation and pattern comply with the ‘one-cell-spacing’ rule, by which each stoma is spaced out by at least one pavement cell. It was believed that short-distance signaling mediated by peptide ligand-receptor signal transduction specifies cell-to-cell communication during stomatal pattern formation. The identified key components include secretory peptides from the epidermal patterning factor (EPF)/EPF-like (EPFL) protein family (Hunt and Gray, 2009; Sugano et al., 2010), the leucine-rich repeat (LRR) receptor protein TOO MANY MOUTHS (TMM) (Nadeau and Sack, 2002) and the ERECTA family (ERECTAf) and somatic embryogenesis receptor-like kinase (SERK) family receptors (Meng et al., 2015; Shpak et al., 2005). Three ERECTAf receptors, including ERECTA, ERECTA-LIKE1 (ERL1) and ERL2, redundantly govern the initial decision of stomatal lineage in the epidermis. The fine-scale genetic analyses revealed that individual members may contribute differently to stomatal division and differentiation. Loss of function of all three ERECTAf genes was found to cause large stomatal clustering (Shpak et al., 2005). The single mutation of erecta resulted in excessive stomatal precursor-like cells (PrCs) that failed to differentiate into mature stomata. With an additional erl2 mutation in erecta erl2 double mutants, the formation of excessive PrCs was enhanced. By contrast, additional mutation of erl1 promoted the differentiation of arrested PrCs into stomata in erecta and erecta erl2 double mutants (Shpak et al., 2005). Thus, although the collective function of the ERECTA family is to suppress stomatal initiation, individual members contribute differently to different aspects of stomatal division and differentiation (Shpak et al., 2005).

In eukaryotic cells, secretory proteins, including cell-surface receptors, are synthesized by the endoplasmic reticulum (ER)-bound ribosomes. Nascent proteins are then translocated into the ER lumen where carbohydrates are added primarily to newly synthesized, unfolded proteins. As the above processes are error prone, eukaryotic cells have an ER quality control (ERQC) system that monitors protein conformation and eliminates terminally misfolded polypeptides through ER-associated degradation (ERAD) or autophagic degradation (Strasser, 2018).

The maintenance of protein homeostasis of cell-surface receptors involves ERQC events in plants. For example, misfolded growth receptor BRASSINOSTEROID INSENSTITIVE 1 (BRI1) was shown to be retained in the ER and degraded by a glycan-dependent ERAD process (Hong et al., 2012). Limited evidence has already shown that VAP-RELATED SUPPRESSOR OF TMM (VST, PapD-like superfamily protein) proteins facilitate ERECTAf signal transduction, possibly through an interaction between VST1 and the ER-localized proteins SOYBEAN GENE REGULATED BY COLD-2 (SRC2) and SYNAPTOTAGMIN A (SYT1) (Ho et al., 2016). Also, it has recently been shown that EPFL9/STOMAGEN triggers the retention of ERL1 in the ER (Qi et al., 2020).

The ER-resident SDF2-ERdj3-BiP chaperone complex plays an important role in protein folding, maturation and degradation in ERQC (Buchberger et al., 2010; Strasser, 2018). In Arabidopsis, luminal binding proteins (BiPs) have three homologs in the genome (BIP1, BIP2 and BIP3). No phenotype was found in single mutants of these genes under normal conditions, but higher-order mutants displayed a pollen development defect (Yamamoto et al., 2008; Maruyama et al., 2014a,b). There are two ERDJ3 homologs in Arabidopsis, including ERDJ3A/TMS1 and ERDJ3B. ERDJ3A is mainly expressed in pollen, and its mutation in a Ler background led to a pollen defect at high temperatures (Yang et al., 2009). By contrast, ERDJ3B is widely expressed in all organs, but no obvious phenotype was found in the erdj3b single mutant. Recent results showed that ERDJ3B was involved in anther and ovule development at high temperatures (Yamamoto et al., 2020; Leng et al., 2022). However, whether the SDF2-ERdj3B-BiP chaperone complex is required for stomatal development by regulating ERECTAf receptors has not been experimentally examined.

Here, we employed pharmacological treatments combined with genetic, cell biological and biochemical strategies to demonstrate that, among the stomatal receptors, individual ERECTAf receptors are distinctly modulated by the SDF2-ERdj3B-BiP chaperone complex. Despite the ERECTAf receptors having high sequence similarity, the SDF2-ERdj3B-BiP chaperone complex appeared to specifically maintain the protein abundance of ERECTA, while promoting the ER exit and translocation of ERL1/ERL2 to the plasma membrane (PM).

Protein N-glycosylation is closely linked with protein folding and plays an important role in ERQC (Helenius and Aebi, 2004). Several conserved N-glycosylation motifs have been predicted in the LRR ectodomain of all three ERECTAf receptors (Chakraborty et al., 2019). To assay possible involvement of ERQC in stomatal development, wild-type Arabidopsis seedlings were treated with tunicamycin (TM), an inhibitor of N-glycosylation (Olden et al., 1982). At a low concentration (30 ng/ml) of TM, no obvious stomatal phenotype was found (Fig. 1A,B), but treatment with a high concentration (60 ng/ml) induced the formation of stomatal clusters (Fig. 1C), consistent with previous observations (Ho et al., 2016). When higher concentrations of TM (120 ng/ml or 240 ng/ml) were applied, the stomatal clustering phenotype became more severe and unpaired GCs within stomatal clusters appeared (Fig. 1D-F). Thus, TM treatment affects stomatal development in a dosage-dependent manner.

ERECTAf receptors redundantly regulate the initial decision and distribution of the stomatal lineage in the epidermis (Shpak et al., 2005). Likely owing to the redundant function of ERECTA in stomatal patterning control, the stomatal phenotype in erl1-5 erl2-3 double mutant was normal (Fig. 1G). However, the TM-induced stomatal clustering phenotype in erl1-5 erl2-3 was more severe than that found in wild-type cotyledons after 60 ng/ml TM treatment (Fig. 1H versus Fig. 1C). For instance, about 37% stomatal units in erl1-5 erl2-3 consisted of more than four GCs, compared with 10% in the wild type, indicating a redundant function of ERECTA in resistance to TM-induced stomatal clustering (Fig. 1I).

We analyzed the expression level and subcellular localization of fluorescent protein-tagged ERECTAf receptors. ERECTA:ERECTA-YFP complemented the stomatal phenotype of the erecta mutant er-105 and predominantly localized at the PM in all epidermal cells (Fig. 1J). After TM treatment for 24 h, no obvious differences in the expression level and subcellular distribution pattern of ERECTA-YFP were found (Fig. 1K). However, a prolonged TM treatment of 48 h resulted in an increase of intracellular ERECTA-YFP fluorescence signals (Fig. 1L,M).

Immunoblotting results confirmed the increase in ERECTA protein levels in TM-treated ERECTA-FLAG transgenic plants (Fig. 1N). However, a large portion of ERECTA-FLAG proteins from the TM-treated plants displayed a faster mobility, which might be due to the increase of newly synthesized proteins with defective protein glycosylation. As ERECTA-FLAG expression was driven by the 35S promoter, we examined the transcription of the ERECTA gene in non-transgenic wild-type plants. However, no significant change in ERECTA transcript levels was found after TM treatment (Fig. S1A). Thus, the enhanced abundance of the ERECTA-FLAG protein in TM-treated plants might be caused by increased protein synthesis or reduced protein degradation. Next, ERECTA-FLAG seedlings were co-treated with TM and cycloheximide (CHX), a protein synthesis inhibitor, or MG-132, a widely used protein degradation inhibitor that can repress the activity of the 26S proteasome. As detected by immunoblotting, the abundance of the ERECTA-FLAG protein was significantly reduced by CHX treatment (Fig. S1I). TM-induced protein accumulation was also abolished by CHX treatment. Inhibition of protein degradation by MG-132 caused an increase in ERECTA-FLAG protein levels. However, the combination of TM and MG-132 treatment conferred a stronger accumulation of ERECTA-FLAG proteins (Fig. S1J). These results indicate that the TM-induced increase of ERECTA protein abundance was mainly caused by the increased protein synthesis instead of reduced degradation. Similarly, the enhanced accumulation of ERL1-FLAG and ERL2-FLAG proteins in TM-treated ERL1-FLAG and ERL2-FLAG plants was due to increased protein synthesis (Fig. 1T; Fig. S1K-N).

To further analyze the glycosylation status of the ERECTA protein, the enzyme endoglycosidase H (Endo-H), which primarily cleaves N-linked glycans of ER-localized proteins but not the complex glycans of proteins that are transported into the Golgi apparatus and further in the secretory pathway, was used to digest the proteins extracted from ERECTA-FLAG transgenic plants (Jin et al., 2007; Nekrasov et al., 2009; Qi et al., 2020). In control plants (DMSO mock treatment), two bands were detected by using anti-FLAG antibodies for immunoblotting after Endo-H digestion, a lower band (N-glycans removed) and an upper band (non-cleaved protein), indicating that the ERECTA protein was partially retained in the ER. However, in TM-treated ERECTA-FLAG plants, the upper band could barely be detected after Endo-H digestion, suggesting that TM treatment greatly enhanced the retention of the ERECTA protein in the ER (Fig. 1O; Fig. S1D).

In ERL1:ERL1-YFP plants, ERL1-YFP mainly localized at the PM and partially appeared in the endosomes of stomatal precursor cells (Fig. 1P,R) (Qi et al., 2020). Interestingly, unlike ERECTA-YFP, ERL1-YFP showed significant accumulation in the ring-like structure after TM treatment (Fig. 1Q,S). ERL2:ERL2-YFP plants exhibited punctate and PM localization of ERL2 in stomatal lineage cells (Ho et al., 2016). Like ERL1-YFP, ERL2-YFP signals also accumulated in a ring-like structure after TM treatment (Fig. S1F,G). Similar to ERECTA-FLAG, both ERL1-FLAG and ERL2-FLAG from TM-treated plants were hypersensitive to Endo-H digestion (Fig. 1U; Fig. S1E,H).

In addition, we examined the effect of TM on YFP-tagged SERK3 and GFP-tagged BRI1, which are PM-localized receptors and have functions in stomatal development (Meng et al., 2015; Kim et al., 2012). However, after 48 h of TM treatment, no obvious changes in the expression level and subcellular localization of SERK3-YFP and BRI1-GFP were found (Fig. S1O-R). Taken together, our results suggest that protein N-glycosylation affects protein biogenesis, subcellular localization and function of ERECTAf receptors in stomatal development.

ERdj3B is the core component of the SDF2-ERdj3B-BiP complex, but none of the three ERECTAf genes showed significant changes in transcript levels in the erdj3b-1 mutant (Fig. S2). Then, we analyzed the total protein abundance of the ERECTAf receptors in the erdj3b-1 mutant by using rabbit polyclonal antibodies that recognize all three ERECTAf proteins. The total ERECTAf protein levels were 60% lower in the erdj3b-1 mutant than those in wild-type plants (Fig. 2A).

To further narrow down whether individual ERECTAf proteins were differentially affected in the erdj3b-1 mutant, by using the same antibodies, we evaluated the total ERECTAf protein amount in double mutants for different combinations of ERECTAf genes; each mutant would contain the third family member intact in the genome. We found a dramatic reduction of protein abundance in erdj3b-1 erl1-5 erl2-3, compared with erl1-5 erl2-3 (Fig. 2B). By contrast, the total protein amounts in erdj3b-1 er-105 erl1-5 and erdj3b-1 er-105 erl2-3 were slightly changed compared with those in er-105 erl1-5 and er-105 erl2-3, respectively (Fig. 2C,D). These results suggested that the lowered total protein levels of ERECTAf receptor proteins in erdj3b-1 were mainly caused by a decrease in the levels of ERECTA, rather than those of ERL1 or ERL2.

Then, we investigated whether the erdj3b mutation affected the subcellular localization of the ERECTAf receptors. ERECTA-YFP (er-105 complementary) was colocalized with the styryl dye FM4-64 at the PM (Fig. 2E). Consisting with immunoblotting results, the overall fluorescence intensity of ERECTA-YFP in erdj3b-1 was dramatically reduced, especially at the PM (Fig. 2F). In addition, the reduced expression level of ERECTA-YFP was also observed in the other chaperone mutant sdf2-2 (Fig. S3), suggesting that the SDF2-ERdj3-BiP complex is involved in maintaining the protein abundance of ERECTA.

Kifunensine (Kif) is an inhibitor of ER α-mannosidase, which prevents protein degradation via the ERAD pathway (Hüttner et al., 2014; Chen et al., 2020). The expression of ERECTA-YFP (er-105 complementary) was significantly enhanced by Kif or MG-132 treatment (Fig. 2G-I). However, these effects were completely abolished in the erdj3b-1 mutant (Fig. 2J-L), as supported by the quantitative analysis of ERECTA-YFP fluorescence intensities in stomatal lineage cells (Fig. 2M). Collectively, our results indicated that the SDF2-ERdj3B-BiP complex was involved in the modulation of ERECTA protein abundance.

To test the possible influence of the erdj3b-1 mutation on intracellular trafficking of ERECTA-YFP, we used brefeldin A (BFA), an inhibitor of ADP-ribosylation factor guanine-nucleotide exchange factors (Geldner et al., 2003; Langhans et al., 2011; Naramoto et al., 2014; Qi et al., 2020). Application of 50 μM BFA for 1 h resulted in aggregation of ERECTA-YFP in ‘BFA-body’ structures. After a 1 h water washout, the BFA bodies disappeared and ERECTA-YFP relocated to the PM (Fig. 2N-P; Fig. S4). Although the expression level of ERECTA-YFP in erdj3b-1 was lower than that in the wild-type, the response of ERECTA-YFP to BFA treatment in erdj3b-1 was similar to that in the wild-type (er-105 complementary), indicating that the erdj3b mutation has no discernible consequence on the intracellular trafficking of ERECTA (Fig. 2Q-S).

ERL1:ERL1-YFP (erl1-5 complementary) plants displayed a predominant localization of ERL1-YFP at the PM (Fig. 3A). In the erdj3b-1 mutant background, ERL1-YFP exhibited a ring-like structure surrounding the nucleus (Fig. 3B). The results of quantitative image analysis demonstrated that in the erl1-5 stomatal precursor cells, ERL1-YFP displayed partial colocalization with an ER marker mCherry-HDEL (Fig. 3C,G). However, in erdj3b-1 precursor cells, ERL1-YFP highly colocalized with mCherry-HDEL in close proximity to the nucleus (Fig. 3D,G). Meanwhile, the fluorescence level of ERL1-YFP was decreased at the PM, which was labeled by FM4-64 (Fig. 3E-G). The ER retention of ERL1-YFP in erdj3b-1 was recapitulated in the sdf2-2 mutant (Fig. S5A,B), suggesting that the functional SDF2-ERdj3B-BiP chaperone complex is required for the transport of the ERL1 protein from the ER to the PM.

To further check the effects of erdj3b-1 mutation on ERL1 retention in the ER, we performed the Endo-H assay using the ERL1-FLAG proteins extracted from erdj3b-1 plants. In contrast to the two bands with similar intensity that were observed in wild-type ERL1-FLAG plants, the erdj3b-1 mutant produced a stronger lower band of ERL1-FLAG (cleaved protein) after Endo-H digestion, demonstrating that the erdj3b-1 mutation induced an increase in the amount of ERL1-FLAG that remained in the ER (Fig. 3H). By contrast, the ERECTA-FLAG proteins extracted from wild-type and erdj3b-1 plants showed a similar sensitivity to Endo-H digestion (Fig. S5C), indicating the distinct effects of the erdj3b-1 mutation on ERL1 and ERECTA.

We also monitored the BFA responses of ERL1-YFP in the erdj3b-1 mutant. In the epidermis of erl1-5 precursor cells, ERL1-YFP was sensitive to 50 µM BFA treatment by forming BFA bodies (Fig. 3I,J). However, ERL1-YFP in erdj3b-1 mutant failed to colocalized with FM4-64 in the BFA bodies (Fig. 3K,L). Treatment with wortmannin (Wm), which can cause a fusion of multivesicular bodies/late endosomes by inhibiting phosphatidylinositol-3 and phosphatidylinositol-4 kinases (Foissner et al., 2016; Qi et al., 2020), was found to induce the dilation of late endosomal ERL1-YFP signals (Fig. 3M,N). However, the formation of these dilated ERL1-YFP-positive ‘Wm bodies’ was suppressed in erdj3b-1 (Fig. 3O,P). These results suggested that the defective transport of ERL1 to PM via the secretion pathway in erdj3b-1 might lead to a reduction of PM localization and consequent internalization of ERL1.

ERL2-YFP in the erdj3b-1 mutant, like ERL1-YFP, lost its predominant localization at the PM and accumulated mostly in the ER (Fig. S5D,E). Consistently, ERL2-FLAG proteins extracted from erdj3b-1 plants displayed hypersensitivity to Endo-H treatment (Fig. S5F). Therefore, the ER-resident chaperone complex SDF2-ERdj3B-BiP modulates ERECTAf receptors in distinct manners, i.e. by affecting the protein abundance for ERECTA versus the subcellular dynamics for ERL1 and ERL2.

BRI1, TMM, and the auxin transporters AUX1 and PIN3 are also involved in stomatal development (Nadeau and Sack, 2002; Swarup et al., 2004; Kim et al., 2012; Le et al., 2014). However, none of them exhibited visible changes in localization in the erdj3b-1 stomatal lineage cells (Fig. S6). Thus, the impact of SDF2-ERdj3B-BiP on ERECTAf receptors does not seem to be a universal phenomenon.

Compared with wild type, no obvious stomatal phenotypes were observed in the erdj3b-1 mutant leaf epidermis (Fig. 4A,B). As reported previously (Shpak et al., 2005), er-105 produced an excessive number of arrested stomatal lineage cells that were unable to differentiate into mature stomata (Fig. 4C,K, colored in pink). However, the additional erdj3b mutation in er-105 plants reduced the percentage of small PrCs (over the total number of PrCs and mature stomata) and induced stomatal clusters, leading to an increased stomatal density (Fig. 4D,K). Similar phenotypes were also found in the sdf2-2 er-105 mutant, confirming the requirement of a functional SDF2-ERdj3B-BiP complex in stomatal development (Fig. S7A-F). Introduction of ERDJ3B-GFP restored the appearance of small PrCs in the erdj3b-1 er-105 epidermis (Fig. S7G,H). ERDJ3B-GFP highly colocalized with the ER marker mCherry-HDEL, supporting that ERdj3B, as predicted, is an ER-resident molecular chaperone (Fig. S7I,J).

Mutation of erl1 was able to resume the stomatal precursor cell differentiation in er-105 (Fig. 4E) (Shpak et al., 2005). However, comparing with the er-105 erl1-5 double mutant, the erdj3b-1 er-105 erl1-5 triple mutant exhibited a more severe stomatal clustering phenotype and increased the number of mature stomata (Fig. 4F,K), indicating that the erdj3b mutation caused the dysfunction of other regulator(s), like ERL2, apart from ERECTA and ERL1.

In contrast to the function of the erl1 mutation, an additional erl2 mutation enhanced the formation of excessive PrCs in the er-105 mutant (Fig. 4G,L). The erdj3b-1 er-105 erl2-3 triple mutant displayed a reduced percentage of PrCs compared with the er-105 erl2-3 double mutant (Fig. 4H,L). We suspected that the action of ERdj3B is possibly through ERL2, which is antagonistic to ERL1 in repressing precursor cell differentiation. However, the erl1-5 erl2-3 mutant displayed normal stomatal development (Fig. 4I). No observable change of stomatal phenotype was found in the erdj3b-1 erl1-5 erl2-3 mutant (Fig. 4J), suggesting the redundant function of ERECTA in stomatal development control, although the abundance of ERECTA receptors was reduced owing to the erdj3b-1 mutation. Furthermore, we checked the impact of the erdj3b-1 mutation on stomatal development in the hypocotyl, where TMM acts to dampen ERECTAf-mediated signaling (Shpak et al., 2005). In agreement with the statement that ERdj3B functions in stomatal differentiation, mutation of erdj3b-1 conferred stomatal production in tmm-1 hypocotyls, mimicking the stomatal phenotype in tmm-1 erl1-5 mutant hypocotyls (Fig. 4M-Q).

The YODA (YDA)-mitogen-activated protein kinase (MAPK) cascade acts downstream of the TMM-ERECTAf receptor complexes in stomatal development (Lampard et al., 2008, 2014). Notably, the erdj3b-1 mutation neither affected the excessive stomatal phenotype of the loss-of-function yoda mutant, nor altered the stomata-less phenotypes of the constitutively active CA-YDA plants (Fig. 5A-D,M), indicating that ERDJ3B genetically functions upstream of the YODA-MAPK cascade. Loss of BASL function in the basl-2 allele has been shown to induce the formation of clustered PrCs in the epidermis (Dong et al., 2009; Zhang et al., 2015) (Fig. 5E). However, the erdj3b-1 mutation failed to convert these small PrCs into mature stomata, indicating that ERdj3B is unlikely to act in the same pathway with BASL (Fig. 5F,M,N). On the contrary, the stomatal phenotypes induced by overexpression of EPF1, EPF2 or EPFL9 were abolished or reduced in the erdj3b-1 mutant background (Fig. 5G-N), further confirming a role for ERdj3B in ERECTAf receptor-mediated signaling during stomatal development.

The functions of secretory EPF/EPFL peptides and TMM-ERECTAf receptors in Arabidopsis stomatal development have been intensively investigated (Yang and Sack, 1995; Shpak et al., 2005; Hara et al., 2007; Hunt and Gray, 2009; Dong and Bergmann, 2010; Pillitteri and Torii, 2012; Qu et al., 2017). Three ERECTAf receptors redundantly govern the initial decision of stomatal lineage and distribution pattern in the epidermis, but individual members contribute differently to stomatal lineage cell differentiation. A question that arises is what determines the distinct roles of ERECTAf receptors that have high homology.

In eukaryotic cells, the majority of membrane receptor proteins are processed in the ER and controlled by the ERQC system. Only correctly folded receptors can exit from the ER and be transported to their final PM destination via the secretory pathway. Improperly or incompletely folded proteins are retained in the ER for further folding or ultimately degraded by ERAD (Fig. 6). Several conserved N-glycosylation motifs like N(X)S/T have been predicted in the ectodomain (LRR) of the ERECTAf receptors (Chakraborty et al., 2019). Here, we experimentally confirmed that inhibition of protein N-glycosylation by TM significantly affects the protein abundance and subcellular distribution of all three ERECTAf receptors. Whether the N-glycosylation differences cause the functional diversity of ERECTAf receptors needs to be further investigated. However, we found that the dysfunction of the ER-resident chaperone complex SDF2-ERdj3B-BiP differentially affects the biogenesis and subcellular dynamics of ERECTAf receptors.

ERECTA is broadly expressed in the epidermis and plays a predominant role overlapping with ERL1 and ERL2 in stomatal patterning control. In contrast to the normal stomatal phenotypes of erl1, erl2 and erl1 erl2, the single mutation of erecta confers an increase of stomatal lineage cells. Although the abundance of ERECTA is reduced in the erdj3b mutant, the intracellular trafficking of the remaining ERECTA protein is normal. The additional mutation of erdj3b has no impact on the erl1 erl2 stomatal phenotype due to the presence of functional ERECTA. This functional insensitivity of ERECTA, unlike ERL1 and ERL2, to dysfunction of the chaperone complex maintains a minimum signaling that enforces the stomatal ‘one-cell-spacing’ rule.

We found that the transcriptional level of ERECTA was not altered by the erdj3b mutation. The effects of Kif and MG-132 treatment on the abundance of ERECTA in the erdj3b-1 mutant were limited, suggesting that the reduction of ERECTA proteins was mainly caused by the downregulation of protein synthesis, but not of protein degradation (Fig. 6). Similarly, the protein levels of EFR, another LRR receptor kinase, in total extracts were found to be strongly reduced in the sdf2-2 background, although the transcription of EFR was not changed (Nekrasov et al., 2009). Accumulation of misfolded proteins in the ER triggers the unfolded protein response. In yeast and mammalian cells, the unfolded protein response leads to a halt of protein synthesis through the action of protein kinase RNA-like ER kinases (PERKs) (Hetz, 2012). The release of BiP from PERK results in PERK autophosphorylation and subsequently inhibits protein translation via the phosphorylation of eIF2α, the key translation initiation factor. Arabidopsis has eIF2α but no PERK homolog has been found (Kanodia et al., 2020). In response to dithiothreitol (DTT)-induced ER stress, GENERAL CONTROL NON-DEREPRESSIBLE 1 (GCN1) and GCN2 promote the phosphorylation of eIF2α (Izquierdo et al., 2018). In maize, ER-stress induces the decline of protein translation efficiency because of the sequestration of mRNAs in stress granules (Kanodia et al., 2020). Therefore, in response to ER stress, like yeast and mammalian cells, plants have ways to relieve the protein overload in the ER by downregulating protein synthesis.

Both ERL1 and ERL2 are specifically expressed in stomatal lineage cells and exhibit hypersensitivity to ER stress. It was demonstrated that EPF1-induced ERL1 internalization and trafficking, for which TMM is essential, are required for the function of ERL1 in restricting stomatal lineage cell differentiation. However, EPFL9 competes with EPF1 by trigging the retention of ERL1 in the ER (Qi et al., 2020). In agreement with this, the arrested stomatal differentiation in erecta leaves and tmm hypocotyls is rescued in an erdj3b-1 mutant background. In addition, the effects of EPF1 and EPFL9 overexpression on stomatal phenotypes are abolished in the erdj3b-1 mutant, further supporting the finding that ER-resident molecular chaperones are required for the subcellular dynamics and function of ERL1.

ERL2, like TMM, can heteromerize with ERECTA and ERL1 to modulate signaling (Ho et al., 2016). Here, we provide evidence that ERL2 localization and function are modulated by SDF2-ERdj3B-BiP. An additional erdj3b mutation in erecta erl1 plants produced excessive stomata in clusters, resembling the phenotypes of the erecta erl1 erl2 triple mutant. Mutation of erdj3b caused the majority of the ERL1 protein to be retained in the ER and failed to transport to the PM, consequently alleviated the repressive effect of ERL1 on stomata lineage cell differentiation in erecta erl2 double mutants. The study of the molecular chaperone complex SDF2-ERdj3B-BiP as a modulator of ERECTAf receptors in the ER will be helpful to further reveal the finely tuned mechanism of stomatal patterning and development.

Arabidopsis accession Columbia (Col-0) was used as wild-type. The erdj3b-1 (SALK_113364), erl1-5 (SALK_019567), erl2-3 (GK_486E03) and sdf2-2 (SALK_141321) were ordered from the Arabidopsis Biological Resource Center (https://abrc.osu.edu). The following higher-order mutants and marker lines were generated by crossing: erdj3b-1 er-105, er-105 erl2-3, erdj3b-1 er-105 erl2-3, er-105 erl1-5, erdj3b-1 er-105 erl1-5, erl1-5 erl2-3, erdj3b-1 erl1-5 erl2-3, erdj3b-1 tmm-1, sdf2-2 er-105, erdj3b-1 yoda, erdj3b-1 CA-YDA, erdj3b-1 basl-2, ERECTA:ERECTA-YFP erdj3b-1, ERECTA:ERECTA-YFP sdf2-2, ERL1:ERL1-YFP erdj3b-1, ERL1:ERL1-YFP 35S:mCherry-HDEL, ERL1:ERL1-YFP 35S:mCherry-HDEL erdj3b-1, ERL1:ERL1-YFP sdf2-2, ERL2:ERL2-YFP erdj3b-1, ERDJ3B:ERDJ3B-GFP 35S:mCherry-HDEL, PIN3:PIN3-GFP erdj3b-1, AUX1:AUX1-YFP erdj3b-1, BRI1:BRI1-GFP erdj3b-1, SUPER:ERECTA-FLAG erdj3b-1, SUPER:ERL1-FLAG erdj3b-1, SUPER:ERL2-FLAG erdj3b-1 and ERDJ3B:ERDJ3B-GFP 35S:mCherry-HDEL. Sterilized seeds were grown on half strength of Murashige and Skoog (1/2 MS) medium supplemented with 1% sucrose and 0.8% agar at 22°C in a growth room with a 16 h light/8 h dark light regime.

To obtain ERDJ3B:ERDJ3B-GFP, the full-length genomic DNA of ERDJ3B without T-Nos driven by its native promoter was cloned into the pCAMBIA1300-GFP vector (Cambia). For SUPER:ERECTA-FLAG, SUPER:ERL1-FLAG and SUPER:ERL2-FLAG, the full-length genomic DNA of ERECTA, ERL1 and ERL2 without T-Nos was cloned into the pCAMBIA1300-FLAG vector. For the ERECTA:ERECTA-YFP, ERL1:ERL1-YFP, ERL2:ERL2-YFP and SERK3:SERK3-YFP constructs, the genomic coding regions of ERECTA, ERL1, ERL2 and SERK3 (from ATG to immediately ahead of the stop codon) were amplified and cloned into pENTR/D-TOPO (Invitrogen). Next, the promoter regions were amplified and inserted into the NotI site of pENTR/D-TOPO vectors carrying the respective genomic regions. Then, the entry clone was recombined by the Clonase II enzyme mix (Invitrogen) into the destination vector pHGY (Kubo et al., 2005) for expression in plants. To obtain the 35S:EPF1, 35S:EPF2 and 35S:EPFL9 constructs, the cDNA of EPF1, EPF2 and EPFL9 was obtained by PCR amplification. Then the fragments were cloned into a pCAMBIA1300 vector under control of the 35S promoter. Primers used for plasmid constructs are listed in Table S1. All constructs were transformed into the Agrobacterium tumefaciens strain EHA105, and then transformed into wild-type or erdj3b-1 mutant seedlings by floral dipping. Transgenic plants were selected on 1/2 MS medium containing 25 µg/ml hygromycin.

Total RNA from 5-day-old wild-type and erdj3b-1 seedlings was extracted using HiPure Plant RNA Mini Kit (Magen Biotechnology). Reverse transcription was performed using PrimeScript RT Reagent Kit (TaKaRa). Amplified ACTIN2 gene was used as an internal control. Real-time quantitative PCR experiments were repeated independently for three times. The data were averaged, and cDNA was amplified using Hieff qPCR SYBR Green Master Mix (Yeasen Biotechnology) with Roche Applied Science Light Cycler 96. Primers specific to target genes used for real-time quantitative PCR are listed in Table S1.

For endogenous ERECTAf protein analysis, the peptide containing amino acids 25-332 of the ERECTA protein sequence was used to produce an ERECTA family-specific rabbit polyclonal antibody (Abmart, Shanghai, China). Total proteins were extracted from etiolated 4-day-old seedlings of the Col-0, erdj3b-1, er-105 erl1-5, erdj3b-1 er-105 erl1-5, er-105 erl2-3, erdj3b-1 er-105 erl2-3, erl1-5 erl2-3 and erdj3b-1 erl1-5 erl2-3 using protein extraction buffer [100 mM Tris-HCl, pH 7.5, 150 mM NaCl, 2 mM EDTA, 2% Triton X-100, 20% glycerol, 20 mM NaF, 1 mM phenylmethanesulfonyl fluoride, 1× protease inhibitor cocktail (Roche)]. The homogenate was centrifuged at 5000 g for 30 min to remove debris. Supernatants were then used for immunoblotting analysis using the anti-ERECTAf antibody (1:5000). Actin was used as the loading control and detected by an anti-actin antibody (1:5000, CWBIO, CW0264M).

For TM, CHX or MG-132 treatment, SUPER:ERECTA-FLAG, SUPER:ERL1-FLAG and SUPER:ERL2-FLAG seedlings were first grown on 1/2 MS medium plates for 2 days, and then treated with 60 ng/mL TM or in combination with 100 μM CHX or 40 μM MG132 for 48 h. The total proteins were extracted as described above. Supernatants were used for immunoblotting analysis. Anti-FLAG (1:10,000, Sigma-Aldrich, F1804-50UG) and anti-actin (1:5000) antibodies were used for immunoblotting analysis.

Total proteins were extracted from 4-day-old seedlings of SUPER:ERECTA-FLAG, SUPER:ERECTA-FLAG erdj3b-1, SUPER:ERL1-FLAG, SUPER:ERL1-FLAG erdj3b-1, SUPER:ERL2-FLAG and SUPER:ERL2-FLAG erdj3b-1. For TM treatment, SUPER:ERECTA-FLAG, SUPER:ERL1-FLAG and SUPER:ERL2-FLAG seedlings were first grown on 1/2 MS medium plates for 2 days, and then treated with 0.01% DMSO or 60 ng/ml TM for 48 h. Extracted proteins were digested with Endo-H (New England Biolabs) for 5-60 min at 37°C before immunoblotting analysis. Anti-FLAG (1:10,000) and anti-actin (1:5000) antibodies were used for immunoblotting analysis.

To obtain differential interference contrast (DIC) images, cotyledons or hypocotyls were cleared with a destaining solution (containing 75% ethanol and 25% acetic acid) overnight at room temperature. After treatment with a basic solution (7% NaOH in 60% ethanol) for 15 min at room temperature, the samples were rehydrated via an ethanol series (40%, 20% and 10%) for 20 min at each step. Images were captured using an Olympus BX51 microscope.

For FM4-64 (Invitrogen) staining, cotyledons of 4-day-old seedlings of the ERECTA:ERECTA-YFP, ERECTA:ERECTA-YFP erdj3b-1, ERL1:ERL1-YFP and ERL1:ERL1-YFP erdj3b-1 lines were immersed into 2 µM FM4-64 for 30 min. For propidium iodide staining, cotyledons of 4-day-old seedlings were stained with 0.1% propidium iodide for 1 min. Confocal images were captured using an Olympus FV1000-MPE laser scanning confocal microscope. The excitation/emission spectra are 488 nm/501-528 nm for GFP/YFP and 543 nm/600-620 nm for mCherry, FM4-64 and propidium iodide. The line scan across the stomatal precursor cells were analyzed with the Plot Profile function of ImageJ (https://imagej.nih.gov/ij/).

To assess the impact of TM on stomatal development, 10-day-old seedlings grown in 1/2 MS medium supplemented with 30, 60, 120 and 240 ng/ml TM (Macklin) were used. Two-day-old ERECTA:ERECTA-YFP er-105, ERL1:ERL1-YFP erl1-5, ERL2:ERL2-YFP, SERK3:SERK3-YFP and BRI1:BRI1-GFP seedlings were treated with 60 ng/ml TM for 24 or 48 h before imaging.

For kifunensine or MG-132 treatment, 3-day-old seedlings of ERECTA:ERECTA-YFP and ERECTA:ERECTA-YFP erdj3b-1 were treated with 20 µM kifunensine (MedChemExpress) or 40 µM MG-132 (Merck) for 12 h. For BFA treatment, 4-day-old seedlings were first immersed into 2 µM FM4-64 for 30 min before treatment with 50 µM BFA solution (Macklin) for 60 min. For BFA washout, seedlings were rinsed in water for 60 min. For wortmannin treatment, ERL1:ERL1-YFP erl1-5 and ERL1:ERL1-YFP erdj3b-1 seedlings were treated with 25 µM wortmannin (Selleck) for 10 min before imaging.

For quantitative analysis of relative protein abundance of ERECTAf receptors, the mean gray value of bands recognized by anti-ERECTAf antibodies in immunoblots were measured using ImageJ. The relative ratios of mean gray value of bands were used in the pairwise comparison of protein abundance. Data shown represent one of three independent experiments.

For quantitative analysis of the effects of kifunensine or MG-132 on ERECTA-YFP expression, the mean intensity values of YFP fluorescence from stomatal precursor cells were measured using ImageJ. Data shown are from one experiment representative of three independent experiments.

For quantitative analysis of Arabidopsis stomatal phenotypes, the numbers of mature stomata and PrCs in a region of 0.2 mm² from the same quadrant of 12 adaxial cotyledons were scored. Small epidermal cells with an area equal to or smaller than a normal stomatal GMC were classified as PrCs (Le et al., 2014). PrC percentage is indicated as the number of PrCs over the total number of PrCs and mature stomata.

All statistical analyses were conducted with SPSS 25.0 software. To compare two normally distributed groups, F-tests were performed first, and then unpaired two-tailed t-tests were used. For multiple comparisons between normally distributed groups, one-way ANOVA followed by Tukey's post hoc test was used.

We thank Prof. Zu-Hua He (CAS Center for Excellence in Molecular Plant Sciences, Chinese Academy of Sciences, Shanghai, China) for the gifts of er-105 mutant seeds and Prof. Wei-Cai Yang (Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing, China) for the gift of mCherry-HDEL plasmid. We thank Wei Wang (Swedish University of Agricultural Sciences, Sweden) for critical reading of the manuscript and English editing. We thank Jing-Quan Li (Plant Science Facility of the Institute of Botany, Chinese Academy of Sciences) for her excellent technical assistance of confocal microscopy analysis.