Heterotrimeric G-Protein Interactions Are Conserved Despite Regulatory Element Loss in Some Plants1[OPEN]

Heterotrimeric G-proteins are key modulators of multiple signaling and development pathways in plants and regulate many agronomic traits, including architecture and grain yield. Regulator of G-protein signaling (RGS) proteins are an integral part of the G-protein networks; however, these are lost in many monocots. To assess if the loss of RGS in specific plants has resulted in altered G-protein networks and the extent to which RGS function is conserved across contrasting monocots, we explored G-protein-dependent developmental pathways in Brachypodium distachyon and Setaria viridis, representing species without or with a native RGS, respectively. Artificial microRNA-based suppression of Ga in both species resulted in similar phenotypes. Moreover, overexpression of Setaria italica RGS in B. distachyon resulted in phenotypes similar to the suppression of BdGa. This effect of RGS overexpression depended on its ability to deactivate Ga, as overexpression of a biochemically inactive variant protein resulted in plants indistinguishable from the wild type. Comparative transcriptome analysis of B. distachyon plants with suppressed levels of Ga or overexpression of RGS showed significant overlap of differentially regulated genes, corroborating the phenotypic data. These results suggest that despite the loss of RGS in many monocots, the G-protein functional networks are maintained, and Ga proteins have retained their ability to be deactivated by RGS.

The growth and development of living organisms entail proper integration of responses to a multitude of environmental and endogenous cues by several intersecting signaling modules. One such module composed of heterotrimeric G-proteins plays a vital role in all eukaryotic species, from yeast to humans and plants (Gilman, 1987;Cabrera-Vera et al., 2003;Siderovski and Willard, 2005;Stateczny et al., 2016;Pandey, 2019). The core of heterotrimeric G-proteins consists of three dissimilar proteins Ga, Gb, and Gg, which remain in an inactive trimeric conformation when the Ga is GDP bound. Signal-dependent exchange of GTP for GDP on Ga results in the dissociation of the trimeric complex. The resulting GTP-Ga and the freed Gbg dimer both can transduce the signal by interacting with downstream effectors (McCudden et al., 2005;Oldham and Hamm, 2008;Pandey, 2019). The inherent GTPase activity of Ga causes hydrolysis of bound GTP to regenerate Ga-GDP, which associates with the Gbg dimer to reconstitute the inactive trimeric complex, which is available for the next round of activation Pandey, 2017). The activation and deactivation steps of this switch-like mechanism need to be synchronized for effective signal transduction. However, the inherent GTPase activity of Ga is significantly slower than the GDP-to-GTP exchange rates Pandey, 2017). Therefore, the roles of proteins, such as the regulator of G-protein signaling (RGS), which act as GTPase-activity accelerating proteins (GAPs), are central in maintaining the optimal output of signal transduction .
G-proteins modulate a variety of critical growth and development processes in plants, encompassing responses to both environmental and endogenous signals (Urano and Jones, 2014;Pandey, 2019;Wang et al., 2019;Zhong et al., 2019). While the defining biochemical properties and interaction specificities of the three core subunits of G-proteins are conserved across eukaryotes, the broader networks in which they are placed seem to differ, even among different plant lineages. For example, Arabidopsis (Arabidopsis thaliana) and maize (Zea mays), representing model dicot and monocot species, respectively, possess one canonical Ga and one Gb protein each, with considerable sequence similarities (Hackenberg et al., 2017;Gao et al., 2019;Wu et al., 2020). However, the loss of function of these genes has distinct effects in each species. The Arabidopsis Ga knockout mutant (gpa1) exhibits several subtle developmental and stress-responsive phenotypes but maintains its overall architecture, growth rate, and fertility (Perfus- Barbeoch et al., 2004;Chen et al., 2006;Fox et al., 2012;Urano et al., 2016b;Roy Choudhury et al., 2020). In contrast, the maize Ga knockout mutant (ct2) is dwarf, with an overall change in plant architecture and exhibits delayed growth and reduced yield (Bommert et al., 2013;Wu et al., 2018). Similarly, knockout of the sole Gb gene (AGB1) in Arabidopsis results in altered development and stress-related phenotypes, but the plants grow to complete their life cycle and produce viable seeds (Roy Choudhury et al., 2020). In contrast, the complete loss of the Gb gene function in maize or rice (Oryza sativa) results in seedling lethality (Utsunomiya et al., 2011(Utsunomiya et al., , 2012Gao et al., 2019;Wu et al., 2020).
In the context of G-protein signaling, rice and maize differ from eudicots by the absence of an RGS gene in their genome (Hackenberg et al., 2017). In fact, RGS has been lost from many monocot species, without an apparent phylogenetic pattern (Hackenberg et al., 2017). Even within the grass family, species differ in the presence or absence of an RGS gene. However, when present, RGS does play important roles in the regulation of G-protein signaling in plants. Consistent with its role as a deactivator of Ga, reduced expression of RGS results in phenotypes similar to those from overexpression of Ga and vice versa Chen and Jones, 2004;Fan et al., 2008;Pandey, 2015, 2017). Moreover, precise regulation of Ga activity is important for its physiological roles, as has been shown by expressing proteins with altered biochemical characteristics in Arabidopsis, maize, and rice (Oki et al., 2005;Zhang et al., 2009;Wu et al., 2018). In maize, complementation of ct2 mutants with a native versus a constitutively active version of CT2 (CT2 CA ) results in distinct phenotypes (Wu et al., 2018). Similar results were observed in rice (Oki et al., 2005) and Arabidopsis (Ullah et al., 2003;Urano et al., 2016b). These observations confirm that accurate fine-tuning of the activation/deactivation rates of the G-protein cycle is crucial for their roles in affecting plant development and productivity.
This apparent need for RGS in most organisms and its absence in many monocots raises the question of to what extent G-protein function, particularly that of Ga, is affected by the presence or absence of RGS. One possibility is that in plants lacking RGS, the entire G-protein signaling apparatus is remodeled in a way that Ga's function has altered to compensate for this change. An alternative is that the signaling network is robust to loss of RGS such that the downstream network is minimally disrupted.
While rice and maize differ from Arabidopsis in the absence of RGS, the grasses and Brassicaceae are separated by more than 100 million years of evolution and differ in many other respects. To undertake a comparison with better phylogenetic control, we instead focused on the grass Setaria viridis, which has a native RGS gene, and compared it with Brachypodium distachyon, which does not (Hackenberg et al., 2017). We hypothesized that if loss of RGS function leads to compensatory changes or rewiring of the Ga networks, then mutations in Ga should have different phenotypes in these two plant types. In addition, overexpressing RGS in S. viridis should create phenotypes opposite to those of the Ga mutations, but overexpressing RGS in B. distachyon should have either no effect or an effect different from the S. viridis overexpression plants. Accordingly, we suppressed Ga in B. distachyon (amiR-BdGa) and S. viridis (amiR-SvGa) with artificial micro-RNA technology and performed detailed phenotypic and molecular characterization of the transgenic plants to compare the G-protein phenotypes in the context of the presence/absence of a native RGS gene. We also generated gain-of-function transgenic plants expressing the RGS gene from Setaria italica, which is the domesticated form of S. viridis (Hu et al., 2018), in B. distachyon (Bd-SiRGS-OE) and in S. viridis (Sv-SiRGS-OE) to elucidate its effect on Ga-dependent development and molecular phenotype.
Our results are surprising. As shown here, despite the lack of a native RGS in B. distachyon, suppression of Ga results in similar phenotypes in B. distachyon and S. viridis. Moreover, the B. distachyon plants overexpressing the SiRGS transgene exhibited phenotypes similar to amiR-BdGa plants, suggesting that the functional modules involving Ga and RGS are intact in plants, even when the native RGS is lost. RNA sequencing (RNA-seq) transcriptomic analysis showed extensive overlap between transcripts differentially expressed in amiR-BdGa and Bd-SiRGS-OE plants, which include many major development-and hormone-related genes. Overall, our results show that the functional interaction and the intracellular networks of Ga:RGS proteins remain conserved in plant species despite the loss of a native RGS gene.

Suppression of Ga in B. distachyon and S. viridis Results in Similar Developmental Phenotypes
To study whether Ga-affected development in grasses depends on the presence of a native RGS protein, we generated transgenic plants with reduced levels of Ga in B. distachyon and S. viridis as contrasting species models for the absence and presence of an RGS in their genome, respectively. S. viridis has two Ga genes (Sevir.9G524500 and Sevir.3G399300), which show ;96% similarity in their amino acid sequences (Supplemental Fig. S1). The artificial microRNA sequences were selected to target both these genes (Supplemental Fig. S2). Appreciable reduction (,70%) in transcript level was observed in both amiR-BdGa and amiR-SvGa plants compared with their respective controls (Supplemental Fig. S3, A and B).
In both B. distachyon and S. viridis, plants expressing reduced levels of Ga exhibited altered architecture. The amiR-BdGa plants were shorter and exhibited delayed growth throughout their life, based on plant height from the 2-week-old to the 6-week-old stage (Table 1; Supplemental Fig. S4). The plants maintained a dwarf stature at full maturity ( Fig. 1, A and B). The amiR-BdGa plants also had fewer and smaller leaves (Fig. 1C) and tillers compared with the control plants. Tiller development in these plants was delayed, and the first tiller emerged ;6 to 7 d later than in control plants. The overall reduction in amiR-BdGa plant height was primarily due to the reduction in internode length, as shown for the second internode ( Fig. 1D; Table 1). Additionally, almost all tissues showed reduced longitudinal expansion (e.g. the plants also had shorter and broader leaves and seeds). Reduced length and fewer tillers made the amiR-BdGa plants appear less bushy at maturity. Mature seeds of amiR-BdGa plants were not only smaller and wider but also had abundant, large trichomes on the lemma not seen in the control plants ( Fig. 1E; Table 1). Along with the size difference, the amiR-BdGa florets had a whitish palea compared with the brown palea of the seeds from the control plants ( Fig. 1E).
Suppression of Ga in S. viridis (amiR-SvGa) resulted in similar phenotypes, with reduced height (Fig. 1, F and G), reduced longitudinal expansion of most organs, including leaves and seeds, fewer panicles, slower growth, delayed tiller emergence, and dwarf stature at maturity compared with the control plants ( Fig. 1, H  The overall dwarf stature of the amiR-Ga plants throughout development suggested an effect on cell expansion. To examine this at the cellular level, we analyzed cell size in the epidermis of 2-week-old leaves. The epidermal cells of the Ga-suppression lines in both B. distachyon and S. viridis were shorter in length and wider as compared with their respective controls. This asymmetric expansion of cells resulted in a significant difference in the ratio of length to width as compared with the control plants. Along with cell size, the cell number on the leaf surface was also significantly reduced in both amiR-BdGa and amiR-SvGa plants (Fig. 2). The overall shorter stature of amiR-BdGa and amiR-SvGa plants, their slower growth, and altered developmental phenotypes are likely results of the altered cellular expansion as well as changes in cell division, similar to what has been reported for Arabidopsis and Camelina sativa G-protein mutants (Ullah et al., 2001;Oki et al., 2005;Roy Choudhury et al., 2020). These data demonstrate that the Ga-dependent developmental phenotypes in grasses with or without a native RGS are similar and the overall wiring of G-protein-dependent networks that control development is not altered in these two contrasting plant types.
Overexpression of a Nonnative RGS in B. distachyon Phenocopies amiR-BdGa Plants As another approach to assess whether the absence of RGS in specific monocot species has altered the inherent G-protein functional networks, we generated a true gain-of-function RGS plant by overexpressing the S. italica RGS gene in B. distachyon, which has no RGS of its own (Hackenberg et al., 2017). To compare the effect of native versus nonnative RGS overexpression, we also generated S. viridis plants overexpressing its native RGS gene. Furthermore, to ascertain that the in planta effects of RGS are truly due to its effect on the activity of Ga, we generated B. distachyon plants overexpressing a point mutant version of the S. italica RGS protein (RGS E319A ) that exhibits no GAP activity (Oki et al., 2005;Hackenberg et al., 2017). Overexpression of each transgene was confirmed by reverse transcription quantitative PCR (RT-qPCR), where ;80-fold increased expression was observed in both native and mutant RGS as compared with controls (Supplemental Fig. S3C).
The results of suppression of Ga or overexpression of RGS levels were similar at the cellular level as well. As observed with amiR-BdGa plants, leaf epidermal cell length in RGS overexpression plants was significantly reduced and the cell width was increased, altering the overall cell length-to-width ratio. Along with the size, the number of cells was also reduced significantly, similar to the amiR-BdGa plants (Fig. 4).
Collectively, these data confirm that the regulation of the classical G-protein activity influences a range of plant growth and developmental phenotypes and that the signaling networks controlled by Ga/RGS proteins are conserved, even in plants that have lost the native RGS. The Ga protein of these plants is able to functionally interact with and be affected by the nonnative RGS protein.

Suppression of Ga or Overexpression of RGS Results in Overlapping Transcript Level Changes
Phenotypic similarities between the Bd-SiRGS-OE and amiR-BdGa plants led us to explore the extent to which transcriptional networks are shared between these plants, i.e. are there gene expression networks that are similarly affected by overexpression of a nonnative RGS gene and loss of Ga in B. distachyon? As the developmental phenotypes were most obvious at 2 weeks after germination, we chose 14-d-old whole seedlings for this analysis. Suppression of Ga and overexpression of RGS resulted in 8,103 and 7,022 differentially expressed genes (DEGs), respectively, compared with the wild-type plants (Supplemental Table S1). Over 60% of these DEGs were shared between amiR-BdGa and Bd-SiRGS-OE plants ( Fig. 5A; Supplemental Table S1). Within these overlapping DEGs, a large majority (;90%) showed similar changes, with 2,905 and 1,025 transcripts showing higher and lower expression, respectively, in both sets of transgenic plants compared with the control plants ( Fig. 5B; Supplemental Table S1). Gene enrichment analysis of common upregulated genes identified integral component of membranes, oxidation-reduction processes, metabolic pathways, and cell periphery as some of the most enriched categories. The down-regulated genes showed significant enrichment in integral component of membrane followed by several nucleotide-binding categories and kinase activity groups (Supplemental Fig. S5). Overall, these analyses suggest that the overexpression of a nonnative RGS gene in B. distachyon or the suppression of a native Ga gene affects similar transcriptional networks.
To gain further insight into the types of transcripts affected by Ga suppression or RGS overexpression, we queried the publicly available B. distachyon eFP browser (http://bar.utoronto.ca/efp_brachypodium/ cgi-bin/efpWeb.cgi) for tissue-specific expression of overlapping DEGs from our data set. A total of 2,314 upregulated genes showed significantly enriched expression in shoot-specific samples, with few gene clusters identified in internode, leaf, and coleoptiles ( Fig. 5C; Supplemental Table S2). A similar trend was observed in the 701 down-regulated genes. The specific expression of DEGs identified in our data set in multiple developing tissues ( Fig. 5D; Supplemental Table S2) relates to the altered vegetative development of these plants.
Identification of tissue-specific expression clusters led us to further examine these data using Weighted Correlation Network Analysis (WGCNA), which generated 11 clusters for up-regulated genes (Supplemental Fig. S6; Supplemental Table S3) and 10 clusters for downregulated genes (Supplemental Fig. S7; Supplemental Table S3). Three tissue-specific clusters corresponding to internode (133 genes), shoot (1,162 genes), and leaf (177 genes) in the up-regulated DEGs (Fig. 6A) and three tissue-specific clusters in the down-regulated DEGs (i.e. coleoptile [52 genes], shoot [208 genes], and leaf [72 genes; Fig. 6B]) were chosen for gene enrichment analysis using ShinyGO.v4. Each cluster had genes with well-established roles in controlling specific developmental phenotypes in other grasses such as maize or rice. For example, the internode cluster showed enrichment of integral component of membranes followed by transporter activity (Supplemental Fig. S8). This cluster contains Bradi1g301610, which is homologous to maize thick tassel dwarf1 (TD1) and Arabidopsis CLAVATA1. Maize Ga functions via the CLAVATA signaling pathway to control meristem development (Bommert et al., 2005). Similarly, the upregulated shoot-specific cluster showed enrichment of cell wall biogenesis as the most enriched category (Supplemental Fig. S9). This cluster contains homologs of rice Expansin-A13 (Bradi5g19340) that has a role in cell expansion and hormonal responses (Lee et al., 2001). A few PMR5-N-terminal domain genes were also listed in the shoot cluster that play important roles in the maintenance of cell wall composition (Chiniquy et al., 2019). The up-regulated leaf-specific cluster was enriched in chloroplast and plastids (Supplemental Fig.  S10). The down-regulated gene clusters showed significant enrichment of several classes of protein kinases, including receptor-like kinases and wall-associated kinases (Supplemental Figs. S11-S13).
To identify the protein interaction network of BdGa, the genes present in the above-mentioned six tissuespecific clusters were screened using the STRING database (Supplemental Figs. S14 and S15; Supplemental Table S4). From the entire network, 12 proteins were identified as direct interactors of BdGa, along with some of its established interaction partners: BdGb (the Gb protein, which is its cognate signaling partner), BdTD1 (homolog of Arabidopsis CLAVATA1), and RACK1 (reported to modulate gene expression of AtGPA1 and AtAGB1 in Arabidopsis; Fig. 6C; Supplemental Table S4; Guo et al., 2009). Additional proteins identified in this primary network have not been characterized, to date, in any plant system, but given their similar scores as bona fide interactors, these are expected to play an important role in Ga-mediated signaling.
Due to the known role of G-proteins in regulating hormone signaling pathways and the altered developmental phenotypes of the mutants, we explored the comparative transcript levels of corresponding genes in amiBdGa and Bd-SiRGS-OE plants. Expression levels of the core abscisic acid signaling component homologs, such as the pyrabactin re-sistant1 (PYR1)/PYR1-like (PYL) receptors, protein  (Fig. 7). Similarly, transcript levels of the key homologs of gibberellic acid (GA) signaling pathway genes (Niu et al., 2019), such as GID1 (GA-insensitive dwarf1), SLY2 (Sleepy2), DELLA, SCR (Scarecrow), SCL (SCRlike), and SHR (Short root), differed from the control plants similarly in both sets of transgenic plants. Similar patterns of expression were observed for transcripts of known homologs of auxin (Shirley et al., 2019), cytokinin (Tsai et al., 2012), and brassinosteroid (Corvalán and Choe, 2017) signaling pathways (Fig. 7), suggesting that the suppression of Ga causes widespread changes in overall hormone networks in plants.
Receptor-like proteins and receptor-like kinases also showed significant and similar changes in both sets of transgenic plants compared with the control plants (Fig. 8). These include proteins such as the homologs of G-protein coupled receptor1 (GCR1), CLAVATA1, and FERONIA, which are established functional interactors of G-proteins in plants (Pandey and Assmann, 2004;Warpeha et al., 2006Warpeha et al., , 2007Bommert et al., 2013;Ishida et al., 2014;Chakraborty et al., 2015;. Similarly, transcripts coding for the homologs of proteins involved in cell expansion, cell size regulation, and key developmental processes, such as expansins, COBRA, and TCP transcription factors, were also significantly altered in both Bd-SiRGS-OE and amiR-BdGa plants (Fig. 8). We independently confirmed the expression levels of different transcripts by RT-qPCR analysis. In total, 12 genes were selected: BdSLC3 (Scarecrow-like3), BdTCP5/7 (TCP transcription factor5/17), CLAVATA1 precursor, BdEXP4/9/16 (Expansin4/9/16), BdGID1L2 (GA receptor), BdCOBL4 (Cobra-like4), BdBZR1 (Brassinazole-resistant1), BdSHR (Short-root), BdGPCR (G-protein coupled receptor), BdLectin-like RLK7, BdLectinlike RLK, and BdReceptor protein kinase, based on their known or predicted roles in plant growth and development. The selected genes showed similar increased or decreased transcript abundance patterns as observed in RNA-seq analysis (Fig. 9). Strikingly similar patterns of these transcript changes, either due to the suppression of Ga or higher expression of RGS, suggest their involvement in classical G-protein signaling pathways. These data also confirm that the RGS-dependent inactivation mechanisms of Ga proteins are maintained, even at the level of global transcriptional regulation, despite the absence of a native RGS gene in specific grasses.

DISCUSSION
Despite the well-established functional roles of RGS and the existence of its highly networked interaction with other proteins of the G-protein complex, the genomes of many monocot species do not possess a gene encoding a canonical RGS protein. Our extensive DEGs common between amiR-BdGa and Bd-SiRGS-OE plants using the expression values derived from the publicly available eFP browser (http://bar.utoronto.ca/ efp_brachypodium/cgi-bin/efpWeb.cgi). The scale shows the row Z-score pattern, with red and blue representing higher and lower expression, respectively. Numerical values with shoots, internode, leaf, and coleoptile show the range of age of young tissues from 10 to 27 d after germination. evolutionary analysis has confirmed that the gene was indeed lost in the monocot lineage multiple times (Hackenberg et al., 2017). This observation has led to several interesting questions, for example: is the G-protein cycle regulated differently in plants that possess RGS proteins versus those that do not? Do RGS proteins have a role in the regulation of the G-protein cycle in plants that have lost it? And what is, if any, the significance of the regulation of G-protein activity in plants?
The nearly identical developmental differences observed in both amiR-BdGa and amiR-SvGa compared with their respective control plants confirm that the obvious effects of reduced levels of Ga proteins on plant growth, development, and physiology are independent of the presence of a native RGS protein in monocot plants ( Fig. 1; Table 1). In fact, at the cellular level, suppression of Ga in both B. distachyon and S. viridis led to the alteration of longitudinal cell expansion (Fig. 2), which has also been reported for G-protein mutants in dicots (Ullah et al., 2001;Chen et al., 2006;Roy Choudhury et al., 2019). Additionally, the overall similar phenotypes of amiR-BdGa and amiR-SvGa clarify that the relatively severe phenotypes of monocot plants with suppressed Ga levels compared with those of the corresponding dicot mutants are not due to the absence of RGS proteins in specific monocot species.
The overexpression of RGS in both B. distachyon and S. viridis also allowed us to test the extent to Figure 6. Correlation networks derived from WGCNA for tissue-specific clusters of DEGs common between amiR-BdGa and Bd-SiRGS-OE plants. A, Up-regulated DEG clusters specific for internode, shoot, and leaf comprising 133, 1,162, and 177 genes, respectively. The y axis represents mean expression values, and the x axis represents tissue samples. B, Clusters for down-regulated genes specific for coleoptile, shoot, and leaf, with 132, 208, and 72 genes, respectively. The graph represents mean expression values on the y axis and tissue samples on the x axis. DAG, Days after germination. C, Proteinprotein interaction network of genes from six selected clusters of up-and downregulated DEGs. Putative interactions obtained using the STRING database comprise two down-regulated genes denoted by blue color and 10 up-regulated genes denoted by red color, showing direct interaction with BdGa (central node).
which RGS protein is functional in monocot species, i.e. does it have a biological role or is it an evolutionary leftover? We envisioned three possible scenarios. One, the overexpression of RGS has no effect on G-proteindependent phenotypes in either B. distachyon or S. viridis, suggesting that the Ga protein in monocots has evolved to function independently of an RGS protein and the gene present in specific plants has likely lost its function. Alternatively, overexpression of RGS leads to G-protein-dependent phenotypes in S. viridis (native species) but not in B. distachyon. This would suggest that some monocot Gas, including those of B. distachyon, have lost their functional interaction with the RGS proteins in planta. Finally, a more likely possibility, also supported by our extant in vitro data (Hackenberg et al., 2017), is that overexpression of RGS has a similar effect on both B. distachyon and S. viridis. This would confirm that RGS proteins do act as GAPs in monocots, or at least in grasses, and that the functional interaction between Ga:RGS is conserved even in plants that do not possess the native protein. Our data support this final possibility. Overexpression of RGS in both native and nonnative species resulted in similar phenotypes, also seen by the suppression of Ga genes (Figs. 3 and 4). Because the overexpression of a variant protein version with no GAP activity (RGS E319A ) did not show any effect, our data also confirm that the developmental phenotypes seen by the suppression of Ga or overexpression of RGS are indeed linked to classical G-protein activity in plants. This is especially important because under specific conditions, G-proteins have also been reported to regulate certain plant phenotypes independent of their classical activity (Maruta et al., 2019;Roy Choudhury et al., 2019).
The substantial overlap between the transcript changes due to the suppression of Ga or overexpression of RGS corroborates the phenotypic data. In fact, the majority of genes known to be involved in hormone signaling pathways or plant development show similar patterns of differential expression in transgenic plants when compared with controls ( Figs. 6 and 9). Detailed analysis of DEGs also confirms that the G-protein cycle in plants is involved in the regulation of major plant hormone signaling pathways, similar to what has been reported for dicots but remained debated in certain monocots ( Fig. 7; Supplemental Table S5). Inferring the interaction network of the Ga protein identified its cognate interactors as well as many other proteins with yet unknown functions (Supplemental Figs. S14 and S15). Elucidating the roles of these proteins is expected to add significantly to our current knowledge of G-protein signaling networks.
The data presented here confirm that reduction of Ga level, either genetically (by amiR-mediated suppression) or biochemically (by overexpressing RGS), results in several growth-and development-related phenotypes. However, plants without a native RGS protein do not seem to have any obvious disadvantages, thus presenting a perplexing scenario: is the regulation of the GTPase activity of the Ga protein significant in the context of overall plant growth and development? The extra-large Ga proteins of plants, which function with the core trimeric proteins, do not seem to require an RGS protein and seem to display very little, if any, GTPase activity (Urano et al., 2016a). The answer to this question comes from studies in maize, where the ct2 mutants were complemented with a constitutively active version of the Ga protein. Mutants expressing CT2 CA showed phenotypes distinct from those expressing the native CT2 gene, confirming that altering the GTPase activity has major consequences and thus must be regulated (Bommert et al., 2013).
Similar results have been seen in Arabidopsis and rice, where the phenotypes of plants expressing a GTPase activity-deficient version of the protein were different from those expressing the native protein (Oki et al., 2005;Ferrero-Serrano and Assmann, 2016). In maize, the plants expressing CT2 CA show improved yield due to higher seed number per plant, suggesting that the precise regulation of the G-protein cycle is agronomically relevant. In fact, many of the phenotypes of monocot plants lacking the Ga gene, especially dwarfism, were deemed highly desirable for breeding useful traits but remained underutilized due to the associated lower yield and significant fasciation of the ears in the mutants (Bommert et al., 2013). The maize ct2 mutants expressing CT2 CA remain dwarf but show considerably reduced fasciation, erect leaves, and higher yield, making them an ideal breeding target (Wu et al., 2018).
How or why the loss of RGS proteins in specific plant lineages is inconsequential remains an open question at this point. We speculate that possibly there are yet unidentified proteins that can biochemically or functionally complement for a canonical RGS in plants. Alternatively, there may exist parallel networks that are activated in plants without RGS to compensate for its loss. There is already evidence for the role of proteins such as phospholipases in deactivating the Ga in dicots (Roy Choudhury and Pandey, 2016;Roy Choudhury et al., 2019). Identification of additional plant-specific G-protein components or regulatory mechanisms will certainly help answer some of these questions.

Plant Growth Conditions and Phenotypic Analysis
Wild-type and transgenic Brachypodium distachyon and Setaria viridis plants were grown from mature seeds collected under identical growth conditions (for each plant species). Seeds were directly planted into Berger 7% to 35% mix soil and kept at 4°C for 2 d for stratification. B. distachyon plants were grown in growth chambers with 24°C (day)/18°C (night) temperature, 20 h of light, and 50% relative humidity conditions. S. viridis plants were grown in growth chambers with 31°C (day)/22°C (night) temperature, 12 h of light, and 50% relative humidity conditions. The light and temperature conditions were maintained throughout the life cycle of the plants. For early to late growth phases and morphological phenotypic analysis of B. distachyon transgenic plants, the Phenovation CropReporter (www.phenovation.com/cropreporter) system was used.
For early development analysis, 2-week-old seedlings were carefully removed from the soil without damaging the roots. Number of tillers, leaf number, leaf length, and root length were measured manually using a ruler. Additional aboveground phenotype data were compiled from the image data of the Phenovation CropReporter. Length of the second internode was measured from 4-week-old plants. A minimum of 15 plants per genotype were used for whole plant-based phenotypic measurements in each experiment. At least three independent replicates were performed for each experiment (45-50 plants Figure 9. Relative transcript abundances of development-and hormone-related genes show similar expression patterns in amiR-BdGa and Bd-SiRGS-OE. Relative transcript levels are shown for 12 genes selected from the DEGs in RNA-seq analysis from the development and hormone pathways. BdSCL3, Homolog of Arabidopsis SCARECROW LIKE3; BdTCP5/17, homolog for Arabidopsis Transcription factor5 and -17; CLAVATA1 precursor, receptor protein kinase; BdEXP4/9/16, homologs of three Arabidopsis genes, EXPANSIN4, -9, and -16; BdGID1L2, GA receptor; BdCOBL4, COBRA-LIKE4; BdBZR1, homolog of Arabidopsis BRASSINAZOLE-RESISTANT1; BdSHR, Short-root; BdGPCR, G-protein-coupled receptor; BdLec-L RLK7, Lectin-like receptor like kinase7; BdLec-L RLK, Lectin-like receptor-like kinase; BdRec. Pro.K, receptor protein kinase. BdGAPDH and BdUbi16 genes were used as controls. The x axis depicts the wild-type (WT) control and transgenic (amiR-BdGa and Bd-SiRGS-OE) plants, and the y axis depicts the relative mRNA abundances. Three independent biological replicates of the experiment were performed, with three technical replicates per experiment, and data were averaged. Error bars denote SE. Asterisks indicate Student's t test values (**, P , 0.05 and *, P , 0.1). The heat map at right shows the log 2 fold change (FC) values of the selected genes in the RNA-seq analysis. A, amiR-BdGa; R, Bd-SiRGS-OE; W, wild type. The scale bar indicates up-regulated genes in red and down-regulated genes in blue. total), and data were averaged. Statistically significant differences were determined using Student's t test.

Generation of Transgenic Plants
To generate B. distachyon plants with suppressed levels of Ga, the complete coding sequence region of BdGa (Bradi2g60350) was screened for efficient silencing regions using the P-SAMS amiRNA Designer tool (http://p-sams. carringtonlab.org/) as described (Carbonell et al., 2015). The selected regions were cloned into pMDC32B-OsMIR390a-B/c (Addgene plasmid 51776) to generate OsMIR390a-B/c:BdGa (Supplemental Fig. S2A). For the generation of B. distachyon plants overexpressing the native or mutant Setaria italica SiRGS gene, the full-length coding sequence of SiRGS (Seita.2G153100) and the mutant version of the protein (SiRGS E319A ) that lacks GTPase activity (Hackenberg et al., 2017) were cloned into the pMDC32 vector. Mutant SiRGS E319A was generated using site-directed mutagenesis (Supplemental Fig. S2B). Constructs were transformed into Agrobacterium tumefaciens strain GV3101. Plant transformations were performed using the embryogenic calli from B. distachyon 21-3 plants using established protocols (Vogel and Hill, 2008). Confirmed transformants were screened and propagated until the T4 generation and used for phenotypic assays. Seeds from two independently transformed lines of the T4 generation, named amiR-BdGa-L1 and amiR-BdGa-L2 for BdGa suppression; Bd-SiRGS-OE-L1 and Bd-SiRGS-OE-L2 for SiRGS overexpression; and Bd-SiRGS E319A -OE-L1 and Bd-SiRGS E319A -OE-L2 for SiRGS E319A overexpression, together with control seeds grown and collected under identical conditions, were used for all phenotypic analysis. Transgenic S. viridis plants with suppressed levels of SvGa (Sevir.9G524500 and Sevir.3G399300) or overexpression of SiRGS (Seita.2G153100) were generated using the same vector system (Supplemental Fig. S2C). S. viridis transformation was performed at the plant transformation facility at the Boyce Thompson Institute of Plant Research. T0 transgenic plants obtained from the facility were propagated and confirmed by sequencing and RT-qPCR. Homozygous plants of the T3 generation were used for phenotypic characterizations.

Cell Shape and Size Analysis
To quantify cell number, cell size, and overall organization, the second leaf of B. distachyon and the third leaf of S. viridis were covered with dental impression (hydrophilic vinyl polysiloxane, Cinch; Parkell). The resin was removed after it set, and the tissue impression was painted with transparent nail polish. The nail polish was peeled off, placed onto glass microscope slides, and imaged with a microscope (Leica DM 750). FIJI was used to quantify cell number and size. A minimum of six biological replicates per genotype (six leaves from six individual plants) were measured, and the experiment was repeated five times. The data presented are averaged from all experiments. Statistical significance was calculated using Student's t test, and P , 0.05 was considered significant.
Gene Expression, RNA-Seq Analysis, and RT-qPCR Total RNA was extracted from 2-week-old seedlings using the RNeasy Mini Kit (Qiagen). The first-strand cDNA was synthesized from total RNA using a SuperScript III cDNA synthesis kit (Invitrogen) after DNase I (Ambion) treatment as per the manufacturer's instructions. Transcript levels of specific genes in transgenic plants were quantified using RT-qPCR with primers specific for BdGa and SiRGS genes as per previous protocols (Roy Choudhury et al., 2019). To confirm the transcript abundance seen in RNA-seq analysis, RT-qPCR analysis was performed on a selected set of genes. RNA was isolated from 2-week-old seedlings using TRIzol reagent (Thermo Fisher) and treated with DNase I. cDNA was synthesized using a SuperScript III cDNA synthesis kit. Gene-specific primers (Supplemental Table S6) were used to quantify the transcript levels. BdGAPDH and BdUbi16 genes were used as controls, and fold changes were calculated by geometric averaging of the two control genes (Vandesompele et al., 2002). Three independent biological replicates of the experiment were performed, with three technical replicates per experiment, and data were averaged.
For RNA-seq analysis, RNA was isolated from the wild type and two independent transgenic lines each of amiR-BdGa and SiRGS-OE B. distachyon seedlings. RNA isolation and follow-up analysis were performed in triplicate, using three independent sets of plants of each genotype. RNA-seq library preparation, sequencing, and initial quality-control analysis were performed by Novogene. For quality control, the FastQC software was used (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Transcripts per million abundance was estimated by pseudoaligning the reads with that of the k-mer index available from the reference genome of B. distachyon from Phytozome (https://phytozome.jgi.doe.gov/pz/portal. html#!info?alias5Org_Bdistachyon) using Kallisto (https://pachterlab.github.io/ kallisto/). DEGs were sorted using the Sleuth program (https://pachterlab.github.io/ sleuth/) with likelihood ratio cutoff of false discovery rate , 0.05. The log 2 fold change was calculated from the transcripts per million values derived for the wild-type and transgenic plants. DEGs were selected by both the false discovery rate cutoff and log 2 fold change . 2 cutoff. Tissue expression analysis of the DEGs was performed using the publicly available data set on B. distachyon eFP browser (http://bar.utoronto.ca/efp_B. distachyon/cgi-bin/efpWeb.cgi). Expression values of young tissues of roots, internode, shoots, coleoptile, and leaf were selected with development age from 3 to 27 d after germination for further analysis. The expression correlation network was determined using WGCNA in R (Langfelder and Horvath, 2008). Three clusters of both up-and down-regulated DEGs were selected for further gene enrichment analysis using ShinyGO.v4 (http://bioinformatics.sdstate.edu/go41/). The protein-protein interaction network was determined using the STRING database for B. distachyon (https://string-db.org/cgi/input.pl?sessionId58oIrCD4MnERN&input_page_ show_search5on). Additional pathway-specific B. distachyon homologous genes were retrieved from Phytozome (https://phytozome.jgi.doe.gov/pz/portal. html#!info?alias5Org_Bdistachyon).

Accession Numbers
The sequences of genes used in the study are available at Phytozome (https:// phytozome.jgi.doe.gov/pz/portal.html#!info?alias5Org_Bdistachyon) with the following accession numbers: Bradi2g60350 (BdGa), Sevir.9G524500 (SvGa1), Sevir.3G399300 (SvGa2), and Seita.2G153100 (SiRGS). The raw and processed files for RNA-seq analysis used in this study are submitted at the National Center for Biotechnology Information Gene Expression Omnibus repository with accession number GSE153188.

Supplemental Data
The following supplemental materials are available.
Supplemental Figure S1. Coding sequence alignment of full-length Ga genes of Arabidopsis, S. viridis, and B. distachyon.
Supplemental Figure S2. Construct design for the generation of transgenic plants.