Homologs of the MBW complex proteins in rice
In previous experiments, we have identified OsGL1A, OsGL1B and OsGL1C as rice homologs of Arabidopsis GL1, OsGL3A, OsGL3B and OsGL3C as rice homologs of Arabidopsis GL3, and OsTTG1A, OsTTG1B and OsTTG1C as rice homologs of Arabidopsis TTG1[40].
To examine if these MBW homologs in rice can form MBW complexes, we first analyzed their interaction relationship on STRING (https://string-db.org/). We found that OsGL1D and OsGL1E were predicted as potentially interaction proteins with OsGL3B. As shown in Fig. 1A, OsGL1 proteins showed a 29.4%~36.3% identity, and a 43.9%~54.6% similarity with GL1 at amino acid level (Fig. 1A). Phylogenetic analysis showed that OsGL1A is closely related to OsGL1B, whereas OsGL1D is closed related to OsGL1E. Together with OsGL1C, these five OsGL1s formed a clade (Fig. 1B). On the other hand GL1 is closed related to MY23, and they formed another clade together with MYB82 (Fig. 1B). Sequence alignment showed that the most conserved region of the OsGL1s is the R2R3 MYB domain (Fig. S1). The [D/E]L×2[R/K]×3L×6L×3R amino acid signature required for the interaction of MYB transcription factors with R/B-like bHLH transcription factors [41], and the S residue has been shown to be required for the activation of GL2 [42], are fully conserved in all the five OsGL1s (Fig. S1).
As for the OsGL3s, both OsGL3A and OsGL3B showed a more than 33% identify and a more than 51% similarity with GL3 and EGL3, whereas that for OsGL3C are only about 25% and 40%, respectively (Fig. 1C). Phylogenetic analysis showed that OsGL3A is closely related to OsGL3B, and they formed a clade with GL3 and EGL3 pair (Fig. 1D). Sequence alignment showed that the most conserved regions of the OsGL3s are the N-terminal and C-terminal domains (Fig. S2). At both the first 97 amino acids of GL3 that are required to interact with GL1 [9], and the HLH domain region, OsGL3A and OsGL3B, but not OsGL3C showed high similar to GL3 and EGL3 (Fig. S2).
Among the MBW complex protein homologs in rice, OsTTG1s are the most conserved ones when compared with their Arabidopsis homologs. OsTTG1A and OsTTG1B showed a 60.5% and 49.3% identity, and a 75.4% and 63.7% similarity, respectively to TTG1 (Fig. 1E). Phylogenetic analysis showed that OsTTG1A is closely related to TTG1 (Fig. 1F). Sequence alignment showed OsTTG1s and TTG1 are highly conserved at full-length amino acid sequence level (Fig. S3), including the 25 amino acid sequence that is required for interaction of TTG1 with GL3 [9].
In order to get a better pictures on the relations that exists between the Arabidopsis MBW complex component proteins and their rice homologs, we identified MBW complex component protein homologs, i.e., proteins with highest amino acid similarity with GL1, GL3 and TTG1, respectively, in the Brassicaceae family plants Brassica rapa, Capsella grandiflora and Capsella rubella, the Fabidae family plant Glycine max, the Malpighiales family plant Populus trichocarpa, and the Panicoideae family plants Zea mays, Setaria italica and Panicum hallii, and expended the phylogenetic analysis. We found that for OsGL1s and the Arabidopsis GL1, MYB23 and MYB82 are still in two different clades (Fig. S4). The Arabidopsis GL1, MYB23 and MYB82 are closely related to homologs from the three Brassicaceae plants and the Malpighiales family plant P. trichocarpa, whereas OsGL1s are closely related to homologs from the three Panicoideae family plants and the Fabidae family plant G. max (Fig. S4). On the other hand, OsGL3C and TT8 formed a clade, whereas GL3, EGL3, OsGL3A and OsGL3B formed another clade with homologs from all the eight plants mentioned above, in which OsGL3A and OsGL3B formed a sub-clade with homologs from the three Panicoideae family plants, and GL3 and EGL3 formed another sub-clade with homologs from the three Brassicaceae plants, P. trichocarpa and G. max (Fig. S5). For the WD40 proteins, OsTTG1B alone formed a clade, whereas OsTTG1A, TTG1 and homologs from all the eight plants formed another clade, in which the OsTTG1A and homologs from the three Panicoideae family plants formed a sub-clade, and TTG1 and homologs from the three Brassicaceae plants, P. trichocarpa and G. max formed another sub-clade (Fig. S6).
Subcellular localization of the MBW complex homolog proteins
Previous reports have shown that GL3, GL1 and TTG1 are all localized in the nucleus [43]. Based on the above bioinformatics analysis, OsGL1A, OsGL1B, OsGL1C, OsGL1D, OsGL1E, OsGL3B and OsTTG1A were chosen for subcellular localization assays. OsGL1A and OsGL1B were chosen because they showed relative high amino acid identify and similarity to GL1, whereas OsGL1D and OsGL1E are potential interactors of OsGL3B according to STRING assays. OsGL3B was chosen because both OsGL3A and OsGL3B showed relative high amino acid identify and similarity to GL3, OsGL3B was predicted to interact with OsGL1D and OsGL1E on STRING, whereas OsGL1C is not paired with other OsGL1 proteins. OsTTG1A was chosen because it showed relative high amino acid identify and similarity to TTG1.
We examined their subcellular localization in Arabidopsis protoplasts. GFP fused constructs of the MBW complex homolog genes were transfected into Arabidopsis protoplasts, and GFP fluorescence was observed under a confocal microscope. We found the OsGL1D, OsGL1E, OsGL3B and OsTTG1A were predominantly localized in nucleus, whereas OsGL1A and OsGL1B may be localized in nucleus and likely some other organelles such as cell membranes and chloroplasts (Fig. 2).
OsGL3B is a transcriptional activator and it interacts with GL1 and TTG1 in Arabidopsis protoplasts
Preciously we have shown that GL3 functions as a transcription activator in transfected Arabidopsis protoplasts [44]. To examine if the MBW complex homologs in c rice can indeed form MBW complexes, we then examined if OsGL3B may also functions as a transcription activator. Plasmids of effector gene GD, GD-OsGL3B or GD-GL3, together with the reporter gene Gal4-GUS were co-transfected into Arabidopsis protoplasts, and GUS activities were examined by using a microplate reader. We found that, similar to GD-GL3, cotransfection of GD-OsGL3B activated the reporter gene expression (Fig. 3A).
Having shown that OsGL3B functions as a transcriptional activator, we examined if OsGL3B may form a MBW complex with GL1 and TTG1 by examining their interactions in yeast cells and Arabidopsis protoplasts. As shown in Fig. 4, OsGL3B interacts with GL1 and TTG1 in yeast cells. Cotransfection of OsGL3B with GD-GL1 and GD-TTG1, respectively activated reported gene expression in protoplasts (Fig. 3B), indicating that OsGL3B may able to interact with GL1 and TTG1 in plant cells. Yet this result may not indicate a direct interaction.
Interactions of OsTTG1A and OsGL1s with OsGL3B and GL3
The above results suggest that OsGL3B is able to form a MBW complex with GL1 and TTG1. We then further examined if it may form MBW complex with OsGL1s and OsTTG1A. To do that, we examined interaction of OsGL3B with OsGL1s and OsTTG1A in yeast cells and Arabidopsis protoplasts. As shown in Fig. 4, OsGL3B is able to interact with OsGL1D, OsGL1E and OsTTG1. Similar, cotransfection of OsGL3B with GD-OsTTG1 activated reported gene expression in protoplasts, whereas cotransfection of OsGL3B with GD-OsGL1A or GD-OsGL1B failed to do so (Fig. 5A). However, cotransfection of OsGL3B with GD-OsGL1E activated reporter gene expression (Fig. 5B). These results suggest that OsGL1E, OsGL3B and OsTTG1A may able to form a MBW complex.
Our protoplast transfection assays also suggest that both OsTTG1A and OsGL1E may able to interact with GL3 (Fig. 5). MBW complex proteins in Arabidopsis and rice are interchangeable in forming MBW complex.
Surprisingly, we found that transfection of GD-OsGL1D activated reporter gene expression (Fig. 5B), suggesting that unlike GL1 and other OsGL1A examined, OsGL1D is able to function as a transcription activator. We also found that GUS activities were increased when GL3, but not OsGL3B was cotransfected with confection of GD-OsGL1D (Fig. 5B), indicating that OsGL1D is able to interact with GL3, but not OsGL3B.
Ectopic expression of OsTTG1A rescued ttg1 phenotypes
After showing that OsGL1E, OsGL3B and OsTTG1A are able to form a MBW complex, we wanted to further examine if they may have similar functions as their Arabidopsis homologs. Considering that the ttg1 mutant has a variety of obvious phenotypes relate to trichome and root hair cell fate determination and secondary metabolism including seed color, mucilage production and anthocyanin biosynthesis [12, 45, 46], and OsTTG1A showed high amino acid identity and similarity to TTG1, we decided to examine if OsTTG1A is a functional analogue of TTG1 by examine if ectopic expression of OsTTG1A could rescue the ttg1 mutant phenotypes.
Transgenic plants were generated in the ttg1 mutant plants by expressing OsTTG1A under the control of the 35S promoter (35S:OsTTG1A/ttg1). Two independent homozygous lines were used for phenotypic analysis. As shown in Fig. 6A, transcript of TTG1 was only detectable in the Ler wild type plants, whereas transcript of OsTTG1A was only detectable in the 35S:OsTTG1A/ttg1 transgenic plants, and relative high transcript level of OsTTG1A was observed in seedlings of the 35S:OsTTG1A/ttg1 #1 line. We observed that plants of both 35S:OsTTG1A/ttg1 transgenic lines produced trichomes on rosette leaves and stems (Fig. 6B). Quantitative analysis showed that plants of the 35S:OsTTG1A/ttg1 #1 line produced more trichomes on rosette leaves (Fig. 6C), consistent with the relative high transcript level in seedlings of this line. On the other hand, reduced root hairs formation was observed in both of the 35S:OsTTG1A/ttg1 transgenic lines when compared with the ttg1 mutants (Fig. 7A), and quantitative analysis showed that root hair density in the 35S:OsTTG1A/ttg1 transgenic seedlings is similar to the Ler wild type (Fig. 7B).
We also found that seed color phenotype of the ttg1 mutant was recovered in the 35S:OsTTG1A/ttg1 transgenic plants, but also to different degree in the two different lines (Fig. 8A). Whereas mucilage production in the 35S:OsTTG1A/ttg1 #1 line was recovered nearly to the Ler wild type, but that in #2 line was largely similar to the ttg1 mutants (Fig. 8B), anthocyanin biosynthesis was also largely recovered in the 35S:OsTTG1A/ttg1 #1 line, but not #1 line (Fig. 8C). These results indicate that OsTTG1A is likely the functional analogue of TTG1.