A japonica rice (Oryza sativa) cultivar, Zhonghua11 (ZH11), was used as background for all the analyses. Plants were grown either in field under natural conditions or in a growth chamber (GXZ-800D) with settings: 30°C day and 28°C night, 10-h/14-h light/dark cycle. When grown in the chamber, half-strength Murashige and Skoog basal salt mixture (1/2MS) was used as nutrient source. Grain size is measured using an integrated photographing and analysis system (WSeen SC-G).

Total RNA was isolated from the samples collected using Trizol (Invitrogen), and cDNA was prepared using a reverse transcription kit (Toyobo) following the product instructions. Quantitative real-time PCR (qRT-PCR) was performed on a LightCycler 96 system (Roche) using SYBR Green PCR mix (Roche). The gene expression levels were normalized to the transcript level of the reference gene Ubiquitin. Primers used for qRT-PCR are listed in Supplementary Table 1

The serk2 mutants were produced using CRISPR/Cas9 system (Lu et al., 2017) targeting 5′-TGGGACAATACCTAATGAA CTGG-3′ corresponding to 333-355 of the SERK2 coding sequence. For overexpression analysis, the full-length cDNA of SERK2 was amplified and then introduced into an empty vector named p1300-35S-FLAG by in-fusion technology (Clontech). See Supplementary Table 1 for primers used for vector construction. The resulted vector p1300-35S-SERK2-FLAG was introduced into ZH11 callus to produce the overexpression plants by Agrobacterium-mediated method.

Rice protoplasts were prepared according to the previous description (Bart et al., 2006). A vector p2300-35S-SERK2-GFP expressing SERK2-GFP fusion protein was used to transfecting rice protoplast. The empty vector expressing sole GFP was used as control and a vector expression plasma-membrane localized protein was used as reference. See Supplementary Table 1 for primers used for vector construction. After one-night incubation, the protoplast cells were observed under a confocal fluorescence microscope (Zeiss) and those emitting fluorescence were photographed.

Empty vectors for this analysis have been introduced previously (Chen et al., 2008). SERK2 and OsBRI1 were introduced into the empty vectors, resulted in p1300-split-nLUC (nLUC-BRI1) and p1300-split-cLUC (cLUC-SERK2) respectively, which were used for luciferase complementation analysis to test the potential interactions in tobacco (N. benthamiana) leaves. See Supplementary Table 1 for primers used for vector construction. The vectors were transformed into agrobacteria and the desired couples were mixed and then infiltrated into tobacco leaves as detailed previously. Chemiluminiscence was photographed using an imaging system equipped with a cold CCD (NightSHADE LB985).

One-week-old seedlings were treated with 200 mM NaCl for 6-8 days and recovered for 3 days to count survival frequency, or treated for different times for transcription and protein analyses. Sterilized seeds were sown on 1/2MS agar medium with or without ABA (Sigma-Aldrich) supplementation and grown for 10 days prior to root and shoot measurements. Different concentrations of BL were used for BR sensitivity test. Briefly, BL was dissolved in ethanol for leaf inclination assay (0, 10, 100, and 1000 ng) and dissolved in DMSO for cultivation on agar medium (0, 0.1, and 1 μM) to test the shoot and root growth as described previously (Tong and Chu, 2017). For molecular analysis, one-week-old seedlings were treated with 1 μM BL for different times.

Four samples, including ZH11, serk2, salt-treated ZH11, and salt-treated mutant (200 mM NaCl, 8 h), were prepared for transcriptome analyses. The aerial parts of one-week-old seedlings were harvested for total RNA extraction. The purified RNAs were used for library construction using a NEBNext Ultra RNA Library Prep Kit for Illumina (NEB). Three biological replicate libraries were prepared for each sample. Clustering of index-coded samples was performed on a cBot Cluster Generation System using a Nova-seq Cluster Kit (Illumina). The libraries were sequenced on the Illumina NovaSeq S6000 platform and 150 bp paired-end reads were generated. FDR (false discovery rate) value < 0.01 and log2FC (fold change) ≥ 1 were used to identify differentially expressed genes (DEGs). Analysis of overlapping DEGs was performed online using Venny¹. Complete DEG lists used for overlapping analysis were provided in Supplementary Tables 2, 3.

The SERK2-overexpression plants were treated with salt or BL for different times, and the seedling shoots were harvested for immunoblotting analysis. Total proteins were extracted as described previously. Commercialized anti-Flag (1:2,000, Sigma-Aldrich) and anti-HSP (1:5,000, BPI) antibodies were used for immunoblotting analyses on a Trans-Blot Turbo Transfer System (Bio-Rad) according to the manufacturer’s instructions.

The serk2 mutants were generated by targeting a specific sequence within SERK2 (LOC_Os04g38480) coding region (Figure 1A). A number of allelic mutants containing various mutations at the targeting site were obtained. Three representative lines containing frame-shift mutations, designated serk2-1 to serk2-3, were selected for further analysis (Figure 1A). At the vegetative growth stage, a marked phenotype could be observed after 14-d growth. Both the seedling height and leaf angles were decreased compared with the wild type (Figure 1B). At the reproductive stage, the mutants exhibited a much more compact structure, with reduced plant height and erect leaves (Figures 1C–F). The decreases of mutant leaf angles were more obvious because at this stage the lamina joints of the wild-type plants greatly bended downward whereas the mutants had basically no bending (Figures 1C–F). Surprisingly, all the mutants had increased grain width, with slight change of grain length (Figures 1G–I), which are reminiscent of BR signaling mutant such as dlt, suggesting the involvement of SERK2 in BR responses. OsBAK1 is the closest homolog of SERK2 and has been functionally characterized previously (Li et al., 2009; Yuan et al., 2017). We also edited OsBAK1 gene using a same strategy, and found osbak1 had a consistent BR-defective phenotypes as reported, including dwarfism, erect leaves, and decreased grain size (Li et al., 2009; Zuo et al., 2014; Yuan et al., 2017). Thus, it appears that SERK2 plays a relatively specific role in BR responses compared to OsBAK1. Given the beneficial effects in the mutants, SERK2 could be valuable for engineering plant architecture.

In rice, a high level of active BRs is able to induce leaf bending but inhibit shoot and root growth (Tong et al., 2014). To test the BR sensitivities of serk2 mutants, we treated plant seedlings with BL and then compared the growth of the three tissues with those of wild type. In lamina bending assay, while the wild type showed gradually enlarged leaf angles along with the increase of BL amount, the serk2 mutants had basically no response to BL (Figures 2A,B). Similar results were obtained in growth inhibition analysis. The growths of both the shoot and root in the mutants were much less inhibited by BL compared to the cases in the wild type (Figures 2C,D). All these results strongly suggested the critical role of SERK2 in BR responses.

Expression analysis by qRT-PCR showed that SERK2 was constitutively expressed in various tissues, with a preference in leaf blade and culm, but less in other tissues (Figure 3A). This expression pattern was somewhat consistent with the mutant phenotypes with main defections in leaf and culm. In addition, SERK2 had greatly decreased expressions in the mutant (Figure 3B), implying that the dysfunctional SERK2 transcripts were somehow suppressed in the mutants. However, we failed to detect increased expressions of the two BR synthetic genes, D2 and D11, suggesting that the feedback regulation between BR signaling and synthesis somehow was not significant or not strong enough for detection in serk2. We also analyzed the subcellular localization of SERK2 by evaluating the fluorescence of SERK2-GFP fusion proteins in rice protoplast. The results showed that SERK2 was specifically localized on plasma membrane, where can be co-localized with OsBRI1 which serves as a plasma membrane marker (Figure 3C). Furthermore, we tested the potential interaction between SERK2 and OsBRI1 using split-luciferase complementation analysis, and detected the marked signal in tobacco leaves (Figure 3D). Together with the mutant phenotypes, these analyses demonstrated that SERK2 plays a critical role in BR signaling in rice, possibly as a co-receptor of OsBRI1.

Next, we tested whether the mutant had altered resistance to salt stress. One-week-old seedlings were used for the analysis as at the time the mutant had little change of plant height. Strikingly, after high salt treatment (200 mM NaCl, 6 days) followed with recovery, all the three mutants showed greatly decreased survival frequencies compared to the wild type (Figure 4A). On average, the survival frequency of ZH11 was ∼65%, whereas of the mutant were ∼20–40%, strongly suggesting that SERK2 is required for plant resistance to salt stress (Figure 4B). To test whether SERK2 is responsive to salt at molecular level, we treated ZH11 with salt, and then analyzed SERK2 expression at different time points. Strikingly, SERK2 was greatly suppressed by salt treatment after 1-h treatment, and the inhibitory effect could be kept to 24 h (Figure 4C). As control, expression of SalT, a salt induced gene (Zhang et al., 2000), was promoted by salt treatment (Figure 4D). Thus, SERK2 is involved in salt stress response and positively regulates salt resistance.

The hypersensitivity of serk2 mutants to salt prompted us to test whether the plants also have altered sensitivity to ABA, the well-known stress hormone. Without ABA, no significant difference was detected between the mutant and the wild type regarding the seed germination process (Figure 5A). However, when ABA (3 μM or 8μM) was supplemented, all the three allelic mutants showed decreased germination with detectable significance at the certain time points (84 h or 96 h) (Figure 5A), suggesting that the mutant has enhanced ABA sensitivities. Similar results were obtained when evaluating the ABA inhibitory effects on shoot and root growth. The inhibitory effects of ABA (3 μM or 6 μM) on either shoot and root in the mutant were greater than those in the wild type (Figures 5B,C). Given that enhanced ABA response should facilitate plant in cope with stress, these results indicated that the increased salt sensitivity of serk2 mutants should be independent of ABA, and SERK2 as a BR signaling component could somehow affect ABA responses.

To verify the roles of SERK2 in stress responses, we performed transcriptome analysis to identify the salt-regulated genes in the wild type and the mutant respectively. Four samples, including ZH11, serk2-1, ZH11 treated with salt (200 mM NaCl, 8 h), and serk2-1 treated with salt, were prepared and the seedling shoots were collected as materials for the analysis. The DEGs in salt-treated samples, namely salt-regulated genes, were used for the comparison (Supplementary Tables 2, 3). As results, we identified 586 salt-regulated genes in ZH11 background, but identified 975 in serk2 background (Figure 6A). Intriguingly, most of these DEGs were up-regulated ones, and consistently, salt treatment apparently had upregulated more genes in serk2 than in ZH11 (Figure 6A). We identified 329 overlapping DEGs between salt regulated genes in ZH11 and serk2. Notably, all of them were consistently regulated (Figure 6B). Among the salt upregulated DEGs in ZH11, 77.7% (279/359) were also upregulated by salt in serk2 (Figure 6B). Importantly, the change folds of most overlapping DEGs in serk2 were higher than those in ZH11 (Figure 6C). We selected three genes for verification, and found these genes indeed were much more obviously induced in the mutant than in the wild type (Figure 6D). Taken together, these analyses further confirmed that loss-of-function of SERK2 enhances susceptibility of the mutant to salt treatment at molecular level.

Compared to ZH11, we identified 129 DEGs from serk2, including 72 upregulated and 57 downregulated. As revealed by gene ontology (GO) analysis, these DEGs were markedly enriched in biological processes such as “growth” and “signaling” (Figure 6E), consistent with the roles of SERK2 in regulating plant growth and development and hormone signaling. In addition, the enrichment in GO terms such as “immune system process” and “response to stimulus” indicated that SERK2 is also involved in plant immunity, as has been reported previously (Chen et al., 2014), and stress responses, as revealed in this study (Figure 6E). Moreover, 20 of the DEGs were also differentially regulated in the salt treated ZH11 samples (salt_ZH11, Figure 6F). Notably, all these 20 DEGs, had consistent expression tendency, either upregulated or downregulated, in serk2 and salt_ZH11 (Figure 6F). Thus, it appeared that loss-of-function of SERK2 has more or less activated salt responses even without salt treatment. One possibility is that these genes represent a subset of salt responsive genes that are unfavorable for plant adaptability to salt stress.

To gain more insight into SERK2 function, we introduced a construct expressing SERK2-Flag fusion protein into ZH11 and obtained a number of transgenic lines. However, we failed to observe a clear morphological difference compared to the wild type (Figure 7A). In two lines with enhanced SERK2 expression, no difference was detected in term of plant height (Figures 7B,C). Regarding the leaf angle, we can only detect a slight increase of the second leaf (the flag leaf was counted as the first leaf), but not in the first and the third ones (Figures 7D,E). Immunoblotting analysis further confirmed the accumulation of SERK2-FLAG fusion proteins in both the two lines as well as two additional lines (Figure 7F). Notably, all these transgenic lines evaluated showed obviously increased grain length and width (Figures 7G–I). We also tested the stress adaptability of three lines and found that all of them showed consistently increased survival frequencies after high salt treatment (200 mM NaCl, 8 d) (Figures 7J,K). Thus, our results suggested that overexpression of SERK2 is able to simultaneously promote grain size and stress resistance. The little effect on the plant architecture further enhances the feasibility of utilization of the gene for crop improvement.

Since SERK2 positively regulates salt tolerance but is suppressed by salt at transcription level, we are curious how SERK2 is regulated by salt at protein level. When the overexpression lines were treated with high salt for different times, the SERK2-FLAG fusion proteins showed a dynamic expression pattern. Upon treatment for 1 h, the protein abundance was slightly suppressed by salt (Figures 8A,B). After that, the protein was gradually increased and was accumulated to a high level after 4- or 6-h-treatment (Figures 8A,B). However, when treated with BL, the fusion proteins showed unaltered or only slightly altered levels (Figures 8C,D). For both treatments, we obtained consistent results using two independent lines for the analysis. Thus, it appears that SERK2 is slightly inhibited and then greatly induced upon stress treatment. This result further demonstrates the involvement of SERK2 in salt stress responses. The eventually highly accumulation of the protein could favor the plant in cope with the adverse condition.