Functional Analysis of the teosinte branched 1 Gene in the Tetraploid Switchgrass (Panicum virgatum L.) by CRISPR/Cas9-Directed Mutagenesis

Tillering is an important biomass yield component trait in switchgrass (Panicum virgatum L.). Teosinte branched 1 (tb1)/Branched 1 (BRC1) gene is a known regulator for tillering/branching in several plant species; however, its role on tillering in switchgrass remains unknown. Here, we report physiological and molecular characterization of mutants created by CRISPR/Cas9. We successfully obtained nonchimeric Pvtb1a and Pvtb1b mutants from chimeric T0 mutants using nodal culture. The biallelic Pvtb1a-Pvtb1b mutant plants produced significantly more tillers and higher fresh weight biomass than the wild-type plants. The increased tiller number in the mutant plants resulted primarily from hastened outgrowth of lower axillary buds. Increased tillers were also observed in transgene-free BC1 monoallelic mutants for either Pvtb1a-Pvtb1b or Pvtb1b gene alone, suggesting Pvtb1 genes act in a dosage-dependent manner. Transcriptome analysis showed 831 genes were differentially expressed in the Pvtb1a-Pvtb1b double knockdown mutant. Gene Ontology analysis revealed downregulation of Pvtb1 genes affected multiple biological processes, including transcription, flower development, cell differentiation, and stress/defense responses in edited plants. This study demonstrates that Pvtb1 genes play a pivotal role in tiller production as a negative regulator in switchgrass and provides opportunities for further research aiming to elucidate the molecular pathway regulating tillering in switchgrass.


INTRODUCTION
Switchgrass (Panicum virgatum), a C4 perennial grass with demonstrated high biomass yield is native to North America and is well adapted to marginal land not suitable for food crops (Mitchell et al., 2008;Narasimhamoorthy et al., 2008). The low production cost and high lignocellulose-based biofuel potential make it well-suited for bioenergy crop (McLaughlin and Kszos, 2005). It was named the model species for herbaceous bioenergy crop by the U.S. Department of Energy in 1991 following more than a decade of research (Wright and Turhollow, 2010). Switchgrass is an out-crossing species with varied ploidy levels and is selfincompatible with individual plants being highly heterozygous (Narasimhamoorthy et al., 2008). Upland and lowland ecotypes are the two major representative taxa of switchgrass (Zhang et al., 2011). Most of the low-tillering lowland ecotypes are tetraploid (2n=4x=36), while the high-tillering upland ecotypes contain both tetraploids and octoploids (2n=8x=72) with hexaploids (2n=6x=54) being reported rarely (Zhang et al., 2011). Unlike the lowland ecotype, which has a caespitose growth habit and poor freezing tolerance, upland ecotypes are highly rhizomatous and cold hardy (Mitchell and Schmer, 2012).
The teosinte branched 1 (tb1) gene is an important tillering/ branching-related transcription factor gene that regulates tillering by integrating environmental and developmental cues (Doebley et al., 1997;Whipple et al., 2011;Seale et al., 2017). In cultivated maize (Zea mays L.), an insertion in the regulatory region of the tb1 gene elevates its expression at leaf axils where axillary buds are located and greatly suppresses tillering, while its ancestor teosinte lacking the insertional sequence is highly branched (Studer et al., 2011). tb1 belongs to the TCP (TEOSINTE BRANCHED1, CYCLOIDEA, PCF) gene family that all encode proteins with a 59-amino acid, noncanonical basic helix-loop-helix (bHLH) motif that allows DNA binding and protein-protein interactions (Martin-Trillo and Cubas, 2010). Closely related tb1 genes with similar functions have been identified in both monocots and dicots (Kebrom et al., 2006;Aguilar-Martinez et al., 2007;Martin-Trillo et al., 2011;Choi et al., 2012;Nicolas et al., 2015). For example, the OsTB1 gene functions as a negative regulator for lateral branching in rice (Oryza sativa L.) (Takeda et al., 2003;Choi et al., 2012). The ortholog of tb1 in bread wheat (Triticum aestivum L.) regulates inflorescence architecture and outgrowth of axillary buds (Dixon et al., 2018). In Arabidopsis, two orthologs of tb1, BRANCHED 1 (BRC1) and BRANCHED 2 (BRC2), regulate branching with BRC1 having the major effect (Aguilar-Martinez et al., 2007). Many studies demonstrate tb1 genes integrate multiple signaling pathways to regulate bud outgrowth and tillering (Kebrom et al., 2013;Rameau et al., 2015). Low ratio of red light to far-red light (R/FR), as a result of shade, inhibits the outgrowth of axillary buds in sorghum (Sorghum bicolor (L.) Moench) by upregulating the expression of SbTB1, an ortholog of tb1 (Kebrom et al., 2006;Kebrom et al., 2010). In Arabidopsis (Arabidopsis thaliana (L.) heynh.), BRC1 suppresses lateral bud growth in response to shade by promoting abscisic acid accumulation inside axillary buds (Gonzalez-Grandio et al., 2017). Cytokinins activate axillary bud outgrowth by inhibiting localized PsBRC1 expression in pea (Pisum sativum L.) (Braun et al., 2012) or OsTB1 in rice (Minakuchi et al., 2010). These studies all demonstrate that BRC1/TB1 is a common target for hormonal and environmental signals that regulate tillering/branching across species. Hence, understanding the function of switchgrass tb1 genes can provide valuable information to the understanding of the tillering mechanisms in switchgrass.
The lowland switchgrass cv. Alamo from which the reference genome was produced is an allotetraploid (2n = 4x = 36, NNKK). It is therefore hypothesized that the majority of genes are present as homoeologs. There are two Pvtb1 genes (Pvtb1a and Pvtb1b) in 'Alamo' with 89% DNA sequence identity between them (Panicum virgatum v4.1, DOE-JGI, http://phytozome.jgi.doe. gov/). Because of its self-incompatibility, it is difficult to obtain homozygous mutants in switchgrass by inbreeding hemizygous mutants as is done for self-pollinating transgenic plants. The clustered regularly interspaced short palindromic repeat (CRISPR)/CRISPR-associated protein 9 nuclease (Cas9) based genome editing tools have become a powerful genetic tool for gene function analysis or crop improvement (Zhao et al., 2016;Wang et al., 2018;Yang et al., 2018). A multiplex CRISPR/Cas9 platform can be used to simultaneously edit multiple genes, which is particularly advantageous for a polyploid species such as switchgrass in which homeologs of a same gene exist, or for members of a gene family that are present in tandem (Zhao et al., 2016).
We previously created switchgrass tb1 mutant plants by using CRISPR/Cas9 (Liu et al., 2018). However, the allelic composition of Pvtb1 genes in these T0 mutant plants was not fully characterized. In addition, it remained possible that these primary mutants are chimeric, which prevented us from accurately assessing the function of tb1 genes. In plant species that can be vegetatively propagated, nonchimeric, solid mutants can be successfully isolated from chimeric mutants (Maluszynski et al., 1995). For example, separating mutated sectors from chimeric mutants by in vitro multiplication has been achieved in banana, cassava and other vegetatively propagated crops (Harten et al., 1981;Novaḱ et al., 1990;Maluszynski et al., 1995;Das et al., 2000). Switchgrass can be readily propagated using node culture (Alexandrova et al., 1996). Hence, generation of primary mutants with CRISPR/Cas9, followed by nodal culture would allow us to generate nonchimeric mutants without the need of producing progeny.
The relative ease with which transgene-free mutant plants can be obtained is a unique advantage for CRISPR/Cas9-based genome editing. Transgene-free mutants have been obtained in multiple crops using CRISPR/Cas9-based genome editing tools and stable inheritance of CRISPR/Cas9 induced mutations has been demonstrated in other plant species, (Pyott et al., 2016;Zhang et al., 2016;He et al., 2018) but not yet in switchgrass. Germplasm generated through CRISPR/Cas9-based genome editing method without foreign DNAs may bypass or simplify the regulatory process required for crops created with traditional transgenic approaches (Waltz, 2018).
Here, we report the successful use of micropropagation to generate nonchimeric mutants from chimeric primary mutants induced by CRISPR/Cas9, eliminating the need of obtaining progeny mutants for gene function analysis. Moreover, the transmission of CRISPR/Cas9-induced mutations in switchgrass is demonstrated in this study. Transgene-free progeny mutants, preserving the phenotypic effect of Pvtb1 mutations similar to that exhibited in T0 mutants, were generated in this study. We showed that double biallelic mutant for Pvtb1a and 1b genes enhanced tiller production and increased biomass yield, indicating that Pvtb1 genes negatively regulate tillering in switchgrass. Transcriptome analysis of a Pvtb1 knockdown mutant 52-1 and wild-type (WT) plant WT-1 suggest that Pvtb1 genes are involved in multiple pathways to regulate tillering in switchgrass.

Micropropagation and Seed Propagation of CRISPR/Cas9-Induced Pvtb1 Mutants
Primary mutant plants were generated previously (Liu et al., 2018). Briefly, the CRISPR/Cas9 construct containing two gRNAs targeting the conserved regions of Pvtb1a and Pvtb1b genes was delivered into the genome of switchgrass through Agrobacterium-mediated transformation of caryopsis-derived embryogenic callus from which primary mutant plants were obtained. To produce nonchimeric mutants from chimeric primary mutants, we micropropagated Pvtb1 mutants by culturing nodes containing axillary buds according to (Harten et al., 1981;Alexandrova et al., 1996). Each of the two basal nodes from a stem containing an axillary bud with a portion of (~1.5cm long) internode above and below the node was selected for in vitro culture on the MS-0 medium with 3 gL -1 Phytagel (Sigma Chemical Co., St. Louis) and no plant growth regulator. These nodal segments were surface sterilized in commercial bleach (5% aqueous solution of sodium hypochlorite) for 30 min, and then rinsed three times with sterile distilled deionized water. Three experiments were conducted for the 52-1 and 35-2 mutants and their corresponding WT (WT-1) plants with 20 nodes for each experiment. After 8 weeks of culture in a growth chamber with a light intensity of 140 µmol m -2 s -1 at a photoperiod of 16/8 h light/dark and a temperature of 25°C, regeneration efficiency (regenerated plantlets/total nodes number) were calculated. After roots emerged, plantlets were grown in 6-inch diameter (1 L) pots using commercial soil mix (Sunshine soil mix #1, Sun Gro) of peat moss and perlite and moved into a mist room for 7-10 days before they were placed in a greenhouse at 26°C with a 16/8 h (day/night) photoperiod and a light intensity of approximately 400 µmol m -2 s -1 .
To determine the inheritance of CRISPR/Cas9 induced mutations in switchgrass, BC1 progeny were obtained by crossing mutants with genetically compatible WT plants. Seeds were harvested from the mutant parents and were sowed in a 1,020 tray (54.5 cm L × 27.8 cm W × 6.2 cm H) with the same soil as described previously that was maintained at a mist room until germination. Seedlings were transferred to the same greenhouse as described earlier. The presence of the transgene was analyzed using PCR with gRNA/Cas9-specific primers ( Table S6).
Genotyping of CRISPR/Cas9-Induced Pvtb1 Mutants by Next Generation Sequencing NGS was used to genotype the Pvtb1 mutant plants generated from micropropagation of primary mutants and BC1 Pvtb1 mutant progeny. To amplify the target fragments of each Pvtb1 gene, gene-specific primers were designed for Pvtb1a and Pvtb1b respectively. The amplicon size for Pvtb1a is 250 bp, while that of Pvtb1b is 288 bp, both are sufficiently long to cover the two target sequences. The Illumina overhang adapter sequences were added to the gene-specific primers (Table S6). A second round of PCR was used to add the dual indices with the Nextra XT Index Kit (Illumina, San Diego, CA, USA). The quality of these sequencing libraries was determined by the Qubit ® 2.0 Fluorometer. The 150-cycle HiSeq sequencing was done for all the amplicon libraries at the DNA Facility at Iowa State University (ISU). For each library, at least 5,000 reads were generated to determine the sequence of the respective amplicons. The results were analyzed by CRISPR-DAV pipeline .

Phenotypic Characterization of Pvtb1 Mutant Plants With Hydroponic Culture
To investigate the effects of Pvtb1 genes on phenotype of switchgrass, we established a hydroponic system to observe plant growth. Nodal segments were cut from the wild-type plant (WT, genotype AABB) and mutant 52-1-3 (genotype aabb), 52-1-1 (genotype Aabb) plants, and cultured in vitro as described previously. Plants of similar height that were grown in 6-inch pots for 10 days were transferred into a hydroponic device in a greenhouse where temperature was maintained at 26°C with a 16h/8h (day/night) photoperiod with a light intensity of approximately 400 mmol m -2 s -1 . For hydroponic culture, the stem base of each plant was wrapped around by sponge and placed into a small basket which was then inserted into a hole on the lid of a plastic container (51 cm L × 43 cm W × 15 cm H) that was filled with a nutrient solution (16 -4 -17 Oasis Hydroponic, JR Peters Inc., PA, USA). Electrical conductivity (EC) and pH of the nutrient solution were measured by a portable pH/EC/ Temperature meter (HI9813-6, Hanna Instruments, Inc., RI, USA) daily and maintained at 1.35 mS cm -1 and 6.0, respectively. The pH was lowered or raised by phosphoric acid or sodium hydroxide, respectively, while the EC was maintained by adding concentrated fertilizer stock solution or clear water. The nutrient solution was aerated by an aquarium air pump (ActiveAQUA, Hydrofarm, CA, USA) to ensure adequate oxygen supply.

Statistical Analysis of Morphological Traits
Primary, secondary, tertiary or quaternary tillers were visually determined according to Davidson and Chevalier (1987). Primary tillers were defined as those which arise from leaf axils in nodes of the main stem whereas secondary tillers arise from leaf axils in nodes of the primary tillers. By the same rule, tertiary and quaternary tillers were defined as those arise from leaf axils of nodes of secondary and tertiary tillers, respectively. Adventitious roots of newly formed tillers tend to have creamy white color whereas roots of older tillers appear in darker color. This color variation was used to assist classification of tillers. Plant height was determined by measuring the tallest tiller for each plant. Root length was determined by measuring the length of the longest root for each plant. Stem diameter was determined by the average diameter of the middle internode of five largest tillers for each plant. The number of tillers and roots (longer than 2 cm) were also counted. To determine biomass yield, pot-grown plants were manually cut at the ground level using a pruning shears and weighted for fresh weight. Samples were then oven-dried at 70°C for 4 days when they reached a constant weight to obtain the dry weight. Three to six biological replicates were used in each measurement. Student's t-test was used to determine the significance of difference between the mutant plants and the WT plants.

Transcriptome Sequencing and Analysis of the mRNA-Seq Data
It was shown in the expression database that Pvtb1 genes in switchgrass expressed highly in axillary buds (Panicum virgatum v1.0, DOE-JGI, http://phytozome.jgi.doe.gov/). Thus, we extracted RNAs from axillary buds of tillers. The mutant 52-1 with highly enhanced tiller production and the WT plant (WT-1) generated from the same callus line from which the mutant 52-1 was obtained were chosen for mRNA-seq. Three biological replications were sampled for each plant. All visible axillary buds collected from two to three tillers are treated as one biological replicate. Total RNA was isolated from tissue samples using the Qiagen RNeasy Plant Isolation kit according to the manufacturer's protocol (Qiagen Inc., Valencia, CA, USA) and the RNA quality was checked using BioAnalyzer 2100 (Agilent Technologies, Santa Clara, CA, USA). cDNA libraries were constructed by the ISU DNA Facility according to the instructions in the Illumina sequencing manual. The Illumina HiSeq 3000 150 paired-end platform was used for sequencing by the DNA Facility at Iowa State University (http://www.dna.iastate.edu/). Library adapter and low quality nucleotides were trimmed off using Trimmomatic (Bolger et al., 2014). Then, STAR was applied to align the trimmed reads to the reference genome V4.1 of switchgrass (Dobin et al., 2013). The number of reads mapped to each gene was calculated using HTseq-count (Anders et al., 2015). Normalization was done using the Upper Quartile method and gene expression differences were analyzed with package edgeR (Robinson et al., 2010) in the statistical software 'R' (Version 3.4.3) (The R FAQ, https://CRAN.Rproject.org/doc/FAQ/). Genes are classified as differentially expressed when an absolute log2-fold change value is ≥1 and a false discovery rate is ≤ 0.05 (Benjamini and Hochberg, 1995). Gene Ontology (GO) analysis for DEGs were performed using the Database for Annotation, Visualization and Integrated Discovery (DAVID) (Huang et al., 2009a;Huang et al., 2009b). ReviGO (Supek et al., 2011) and Cytoscape (Shannon et al., 2003) were applied to visualize enriched GOs. Six genes were selected to validate the expression patterns determined by RNA-seq. Primers were listed in Table S6.
To determine the allelic compositions of Pvtb1a and Pvtb1b genes in 10 randomly selected micropropagated plants including 9 plants generated from primary mutants and 1 plant from the wild-type, sequencing libraries for the tb1a amplicons with an insert size of 250 bp and tb1b amplicons with an insert size of 288 bp were constructed for each individual plant. For each library, at least 5,000 reads were generated from Illumina HiSeq3000 to determine the sequence of the respective amplicons. Three distinct genotypes were found among the five micropropagated plants (52-1-1, -2, -3, -4, -5) derived from 52-1, while four distinct genotypes were observed among the four micropropagated plants (35-2-1, -2, -3, -4) derived from 35-2, clearly indicating that the two primary Pvtb1 mutants from which micropropagated plants were obtained are chimeric ( Table 1).
Four nonchimeric, solid mutants, 52-1-1, 52-1-3, 52-1-4, and 52-1-5 were obtained from the T0 mutant 52-1 with 52-1-1, -4, and -5 all having the same genotype, i.e. 50% of WT Pvtb1a allele and 50% Pvtb1a allele with a G deletion at the first target site and a T insertion at the second target site ( Table 1), suggesting these plants were likely originated from a single progenitor mutant cell carrying a monoallelic Pvtb1a mutation. In the other solid mutant 52-1-3, all alleles of the Pvtb1a were mutated with half of the Pvtb1a alleles contains a G deletion at the first target site and a T insertion at the second target site, while the remaining half of Pvtb1a alleles contains only one change, an insertion of G at the first target site ( Table 1), suggesting 52-1-3 was likely derived from a single progenitor mutant cell carrying biallelic mutations of Pvtb1a gene. These four plants all contain the same biallelic mutations for Pvtb1b with no WT alleles ( Table 1). Fifty percent of the amplicons from each of the four solid mutants contained one C deletion at the first target site, while the remaining half amplicons contained the same mixed mutation (deletions, insertions and substitution) at the first target site and one A insertion at the second target site ( Table 1), suggesting these four plants were developed from a single progenitor cell or mutant sector carrying biallelic mutations of the Pvtb1b gene.

Knockout of Pvtb1 Genes Leads to Increased Tillering in Switchgrass
Based on the most recent annotated switchgrass genome in Phytozome (Goodstein et al., 2012) (https://phytozome.jgi.doe.  gov/pz/portal.html), two tb1-like genes, Pvtb1a (Pavir.Ia00838/ Pavir.9NG142700) and Pvtb1b (Pavir.Ib04362/Pavir.9KG031700), were isolated from switchgrass using gene-specific primers. These two Pvtb1 genes reside on homeologous chromosomes 9N and 9K, thus are likely homeologous to each other. Amino acid sequence alignment of Pvtb1 gene products and other tb1 orthologs of related species showed that both PvTB1 proteins have the conserved TCP domain and R domain (Figure 1). To determine the function of Pvtb1 genes, tiller numbers of the Pvtb1a-Pvtb1b double biallelic mutants (aabb), Pvtb1a monoallelic and Pvtb1b biallelic mutant plants (Aabb) and WT (AABB) plants were compared. Because switchgrass tillers develop from nodes near or below soil surface, we grew plants in a hydroponic system, which allows close monitoring of tiller and root growth without the destructive excavation of the underground parts of these plants. As shown in Figure 2, the first primary tiller of WT plants did not develop until the end of the fourth week following establishment in the hydroponic system, while most of the mutant plants developed their first primary tiller by the end of the first week. After 4 weeks, the average tiller number of aabb mutants is 5.7 which is about two times the tiller number of WT plants, while the average tiller number of Aabb mutants is 3.7 which is about 1.5 times the tiller number of WT plants.
To further investigate where the increased tiller originated from mutants, development of tillers of different orders in the aabb mutant plants and WT plants were closely monitored for 8 weeks in a separate hydroponic experiment. As shown in Figure  3A, the pvtb1a-pvtb1b (aabb) mutant plants had significantly more tillers than WT plants. The change of tiller numbers over the course of 8 weeks show the tiller numbers of mutant plants are significantly higher than WT plants at all weekly sampling times ( Figure 3B). The WT plants did not generate new tillers until the end of the fourth week after establishment in the hydroponic system, while the mutant plants started to produce new tillers at the end of the first week, which is similar to the results from the shorter-term study described above. However, the pvtb1a-pvtb1b mutant plants had no significant effect on the number of primary tillers (1°) compared with WT plants ( Figures 3C, D), which suggested that Pvtb1 genes do not regulate the formation of primary axillary buds in the main stem, and the significant increase in secondary (2°), tertiary (3°),  or quaternary (4°) tillers in the mutant plants is the result of hastened outgrowth of primary axillary buds. To further confirm the effect of Pvtb1 on the outgrowth of axillary buds and eliminate possible effect of physiological conditions of the starting materials, primary tillers developed from the pvtb1a-pvtb1b double biallelic mutant plants and the WT plants were grown as starting material in a separate hydroponic study. The total number of new tillers developed from the primary tillers of the mutant plants was three-fold of that produced by the primary tillers of the WT plants ( Figures 4A, B). Similarly, plants established from the primary tillers of the pvtb1a-pvtb1b mutant had similar number of the first order tillers (1°) to that of the WT, but had greatly increased number of secondary (2°) and tertiary tillers (3°) than those from the primary tillers of the WT. These results strongly suggest that, regardless of the origin of the starting materials, the pvtb1a-pvtb1b mutant plants consistently promote axillary bud outgrowth, increasing tiller number. Aerial tillers, however, were not observed in either the mutant plants or the WT plants, suggesting that Pvtb1 genes did not affect the outgrowth of the upper axillary buds. Interestingly, while the first axillary bud (the lowest in the stem base) in the WT was generally dormant, it was frequently elongated to eventually become a tiller in the pvtb1a-pvtb1b mutant ( Figure   S2). These results further demonstrate that Pvtb1 genes affect the outgrowth of the axillary bud rather than the formation of new axillary buds.

Pvtb1 Knock-out Mutant Plants Exhibit Increased Root Production and Biomass Yield
The average root number in mutant plants was similar to that of the WT plants at the beginning of the hydroponic culture ( Figure 5). Starting from the second week, the difference between the average root number in mutant plants and WT plants started to increase as the experiment progressed ( Figure  5). At the end of the experiment, the average root number of the mutant plants was 28, twice the average root number of the WT plants ( Figure 5). Hence, disrupting Pvtb1 gene function in switchgrass enhances root production.
To eliminate the potential unintended effect of the hydroponic growth environment on growth and development, we also evaluated tiller number, stem diameter, plant height, fresh weight and dry weight of pvtb1a-pvtb1b mutant plants and WT plants grown in the soil. Similar to the hydroponic experiments, pvtb1a-pvtb1b mutant plants produced significantly more tillers (2.53-fold) compared with the WT plants ( Table 2). Plant height of the mutant plants was similar to that of the WT, while stem diameter of the mutant plants was 13.6% smaller than that of the WT ( Table 2). The fresh and dry biomass of the mutant plants increased 29.6% and 15.5% over these of the WT plants, respectively, suggesting the enhanced numbers of tillers of mutants were more than sufficient to compensate for the loss in biomass from decreased stem diameter ( Table 2).

CRISPR/Cas9-Induced Mutations Are Transmitted to Progeny
Only transgene-free BC1 plants, derived from crossing the T0 mutants as maternal parents with a compatible WT plant, were examined to avoid complications that could have resulted from the continuing action of the CRISPR/Cas9 transgene. Among the 30 progeny of the primary mutant 52-1, 12 were transgene-free with ten of them carrying mutations. Meanwhile three of six progeny of the cross between mutant 35-1 and a WT plant carried no transgenes and of these, only one carried mutations (Table 3). Additionally, 18 of 20 progeny from the mutant 35-2 were transgene-free ( Table  3), but only two carried mutation. The Pvtb1 mutations observed in all three primary mutants were also observed in their progeny, indicating a stable transmission of the original mutations to the progeny (Table  S1). Seven BC1 plants of the 52-1 had both Pvtb1 genes mutated (Table S1), while the other five plants, in BC1 generation, only contained mutations of Pvtb1b (Table S1). All mutations in these BC1 plants were present in the primary mutant 52-1 except for the 52-1-BC1-24, which has a mutated Pvtb1a allele carrying a 128 bp deletion which was not observed in the primary 52-1 mutant ( Table S1). Two monoallelic Pvtb1a mutants (35-2-BC1-1 and -6) were obtained from the progeny of the 35-2 ( Table 3). Both carried identical mutations observed in the primary mutant 35-2 (Table S1). Furthermore, one doubly monoallelic Pvtb1a-tb1b mutant with big deletions for both tb1 genes was found in the progeny of 35-1 (Table S1).
To test if mutant progeny still retains the phenotypic effects of Pvtb1 mutations, tiller numbers of Pvtb1b monoallelic mutants (genotype AABb with A represents the WT and a represents the mutant allele of Pvtb1a and B represents the WT and b represents the mutant allele of Pvtb1b), and Pvtb1a-Pvtb1b doubly monoallelic mutant plants (genotype AaBb) in the BC1 generation were investigated. Tiller numbers of doubly Pvtb1a-Pvtb1b monoallelic mutants and single Pvtb1b monoallelic mutants were significantly higher than the siblings carrying no mutations (null segregants), suggesting that Pvtb1 genes act in dosage-dependent manner. The average percentage increase in tiller number was 45% for doubly Pvtb1a-Pvtb1b monoallelic mutant plants and 30% for the Pvtb1b monoallelic mutant plants ( Table 4).

mRNA-Seq of Pvtb1 Mutants Suggests Pvtb1 Genes Are Involved in Multiple Pathways That Regulate Tillering in Switchgrass
Although the primary mutant 52-1 was later found to be chimeric, its enhanced tiller production (Liu et al., 2018) with 30%-50% of the tb1a alleles and 61%-94% of the tb1b alleles mutated are sufficient to be treated as a knockdown mutant of Pvtb1 gene ( Figure S3). Using a generalized linear model with the R package edgeR (Robinson et al., 2010), 831 genes showed significantly differential expression between the WT and 52-1 when the false discovery rate (FDR) was controlled at 0.05 (Benjamini and Hochberg, 1995) (Table S2). Among them, 364 genes were significantly upregulated while 467 genes showed significantly down regulation in the mutant 52-1 ( Figure S4). To validate the RNA-seq data, qRT-PCR was performed on 4 differentially expressed genes and 2 genes with   All data are shown as mean ± standard deviation. A and a represent the wild-type and mutant allele of Pvtb1a gene, respectively whereas B and b represent the wild-type and the mutant allele of the Pvtb1b gene, respectively. The percentage increase of tiller number for each genotype was calculated by comparing the mutant plants with the wild-type plants at the same developmental stage using the one-tailed Student's t-test. Different letters at the same column indicate significant differences at P < 0.05 level. similar expression level between the WT and 52-1. As shown in Figure S5, the expression trends of these 6 genes observed in qRT-PCR and RNA-seq results were similar in WT_vs_52-1 and a high expression correlation was observed between fold changes of 6 genes in RNA-seq and qPCR experiments using linear regression analysis (R 2 = 0.871, p=0.0065), indicating the results of RNA-seq are reliable. Orthologous Arabidopsis genes for these DEGs were subjected to Gene Ontology (GO) analysis using the Database for Annotation, Visualization and Integrated Discovery (DAVID) (Huang et al., 2009a;Huang et al., 2009b). Regarding the biological process, 27 terms (P<0.05, Table S4) were enriched for downregulated genes in the mutant, and 12 terms (P<0.05, Table S3) for the upregulated genes. A major GO term for these DEGs was DNA-templated transcription (GO: 0006351) (Figures 6A, B). Among the genes enriched in this GO term, transcript levels for 34 genes increased significantly in the Pvtb1 gene knockdown mutant 52-1 while the remaining 43 genes were downregulated ( Table S5). The majority of the genes in this GO term encode transcription factors (TFs). For instance, genes for 3 TCP family TFs (TCP2, 4, and 5) and 4 MADS-box family TFs (AP1) were upregulated in the mutant 52-1, while genes for 8 WRKY family TFs (WRKY30,33,40,41,46,and 53) and 3 NAC family TFs (NAC02, 036, and 102) were downregulated. GO terms associated with cell differentiation and positive regulation of development were also enriched among the upregulated genes in the mutant. Other GOs enriched in both up-and downregulated genes were involved in response to various phytohormones suggesting Pvtb1 genes are, as in other species involved in various hormonal signaling pathways in regulating tillering in switchgrass (Tables S3, S4). For example, BRH1, a brassinosteroid-responsive RING-H2 gene, was downregulated in the Pvtb1 gene knockdown mutant (Figure 7). In contrast, genes for auxin-responsive factor and ABA-induced proteins were both upregulated in the mutant (Figure 7). In addition, GO terms involved in negative regulation of flower development and heterochromatin maintenance were overrepresented among the downregulated genes ( Figure 6B). For example, Arabidopsis DRD1 and EDM2 both regulate DNA methylation patterns (Cho et al., 2016;Duan et al., 2017) and the orthologs of these two were downregulated in the present study. Hence, we speculate that epigenetic modification plays a role in Pvtb1-mediated regulation in tillering in switchgrass. Intriguingly, GOs associated with stress/defense responses were significantly enriched in downregulated genes, indicating Pvtb1 genes might function in maintaining homeostasis in the regulation of growth and defense ( Figures 6B, 7). These data suggest the gene regulatory network of tillering in switchgrass is complex and Pvtb1 genes are important hubs in this network.

DISCUSSION
CRISPR/Cas9-based genome editing tool affords new possibilities for reverse genetics-based studies in switchgrass (Liu et al., 2018). In the present study, through the node culture of chimeric primary mutants induced by the CRISPR/Cas9 system, we successfully obtained a nonchimeric biallelic Pvtb1a-Pvtb1b mutant, and two nonchimeric mutants carrying biallelic Pvtb1b mutations and one Pvtb1a mutated allele. These results demonstrate that micropropagation is an effective way of isolating mutant sectors from chimeric mutants in switchgrass and obtaining nonchimeric mutants. Solid nonchimeric mutants were isolated from the chimeric primary mutant, 52-1. Among five plants regenerated from nodes culture of 52-1, four plants are nonchimeric mutants, suggesting the primary mutant plant 52-1 is composed primarily of mutant cells. However, for the primary mutant 35-2, none of the regenerated plants are nonchimeric mutants, indicating this plant, or regions where axillary buds arise are composed primarily of nonmutant cells. Overall, we demonstrated that nonchimeric mutants can be successfully isolated from chimeric mutants using micropropagation, but the chimeric pattern of chimeric mutants is a determining factor on the efficiency of isolating nonchimeric mutants, and varies greatly between mutants. Strong genetic-incompatibility of switchgrass is an obstacle to generate homozygous mutants. To set seeds carrying homozygous mutations, individual mutants used for crossing must have different alleles of S and Z genes (Martinez-Reyna and Vogel, 2002), which is difficult to determine because the molecular basis of the self-incompatibility in switchgrass has not been studied. Also, it takes over 6 months for switchgrass seedlings to reach reproductive stage (Hopkins et al., 1995;VanEsbroeck et al., 1997) for crossing. Given all these obstacles, the successful purification of chimeric mutants using micropropagation clearly has advantages. In addition, nonchimeric homozygous mutants can be obtained from T0 chimeric mutants within one month of using the micropropagation method, which is time-saving compared to traditional breeding method.
Switchgrass cv. Alamo is an allotetraploid with two homeologous subgenomes, but detailed information about chromosome pairing, whole or partial genome duplications, and allelic diversity of specific genes is lacking (Missaoui et al., 2005;Okada et al., 2010). With limited information gained by sequencing clones of PCR amplicons spanning the target regions, the nature of mutants was not resolved unequivocally in our previous study (Liu et al., 2018). Using the Next Generation Sequencing (NGS) technology, genotypes of micropropagated mutants were fully characterized. The two primary mutants, 52-1 and 35-2 were both revealed by NGS to be chimeric mutants, which were likely the results of continuous action of CRISPR/ Cas9 in somatic cells. Indeed, chimeric mutations induced by CRISPR/Cas9 have been reported in different species (Feng et al., 2014;Pan et al., 2017). For instance, Pan et al. (2017) showed that 63.9% T0 transgenic plants carried chimeric mutations in tomato (Solanum lycopersicum L.). In rice, the segregation ratio of CRISPR/Cas9-induced mutations in BC1 generation did not follow the expected segregation ratio, indicating the chimeric mutations in T0 plants (Xu et al., 2015).
Transgene-free mutants were successfully obtained by crossing primary mutant plants with WT plants. The majority of the mutations observed in progeny were identical to the mutations in the primary mutants. For example, BC1 progeny of the 52-1 carried the Pvtb1b mutations that were present in all five tillers of the primary mutant 52-1. These results demonstrate mutations induced by the CRISPR/Cas9 in T0 mutants were stably transmitted to the BC1 generation without alteration. However, a mutation (128bp deletion) of Pvtb1a, not detected in the primary mutant 52-1, was observed in the BC1 transgenefree mutant plant 52-1-BC1-24. This deletion likely escaped detection in T0 plants due to the chimeric nature. The similar phenomenon has also been observed in maize (Lee et al., 2019).
Pvtb1 genes are closely related to the tb1 genes in other monocots. The Pvtb1a-Pvtb1b double biallelic mutants produced significantly higher tiller number than the WT plants under both hydroponic and soil culture conditions. Results from the hydroponic experiments in which starting plant materials were primary tillers validated the observed differences between the mutants and the WT plants were the result of genetic alteration instead of the existing physiological difference between the WT and the mutant. Our results strongly indicate Pvtb1 genes play an important role in regulating tillering in switchgrass. There was no significant difference in the number of primary tillers between the WT and 52-1-3, suggesting the function of Pvtb1 genes in switchgrass is only to regulate the rate of outgrowth of axillary buds that are destined to become tillers with one possible exception, i.e., the release of the outgrowth of the bud at the lowest node ( Figure S2). This is similar to orthologs of tb1 in other species, reflecting the functional conservation of tb1 genes across species (Takeda et al., 2003;Kebrom et al., 2006;Aguilar-Martinez et al., 2007;Braun et al., 2012).
The Pvtb1a-Pvtb1b double biallelic mutant plants also produced more roots compared with WT plants. This result is consistent with Gaudin et al. (2014) who reports a decrease in tb1 function in maize resulted in a larger root system. The increased root growth and development is likely the indirect result of increased tiller production, as each tiller normally develops its own adventitious roots. Increased tiller number has the potential to increase biomass yield in switchgrass (Chuck et al., 2011;Fu et al., 2012). The Pvtb1a-Pvtb1b mutants produced 29.6% and 15.5% more fresh and dry biomass, respectively. Although the difference on dry biomass between the wild-type and mutant plants is not statistically significant, this may change if mutant plants with more tillers and increased root mass are grown in the field where more resources are available.
In maize, studies have shown that tb1 regulates branching in a dosage-dependent manner (Doebley et al., 1995;Hubbard et al., 2002). Maize heterozygous tb1 mutants had slightly more tillers than WT plants, while homozygous tb1 mutants produced more tillers than heterozygous tb1 mutants (Doebley et al., 1995). In this study, we noticed the tiller number in monoallelic heterozygous Pvtb1b mutants (AABb) was significantly higher than WT plants, suggesting that Pvtb1b functions in a dosage-dependent manner. In addition, due to the high amino acid sequence identities between PvTB1A and PvTB1B, it is reasonable to expect they regulate tillering of switchgrass redundantly or additively. However, comparing monoallelic Pvtb1b mutant plants (AABb) with the doubly monoallelic Pvtb1a-Pvtb1b mutant plants (AaBb) did not show significant differences on tiller numbers, suggesting there is no significant additive effect between Pvtb1a and Pvtb1b. Therefore, Pvtb1a might have a minor effect on tillering in switchgrass. This is similar to Arabidopsis in which only BRC1 regulates branching, despite both BRC1 and BRC2 have the conserved TCP and R domains (Aguilar-Martinez et al., 2007;Gonzalez-Grandio et al., 2013;Seale et al., 2017).
To have a better understanding of the functions of Pvtb1 genes, we examined global transcriptional changes caused by the downregulation of Pvtb1 genes. Increased expression level of genes for TCP TFs that are associated with cell differentiation and positive regulation of development in the mutant suggested that Pvtb1 genes inhibit the tiller production through deactivating cell differentiation. In addition, HAIRY MERISTEM 3 (HAM3), the gene for GRAS family TF that interacts with WUSCHEL (WUS) TF to promote shoot meristem development (Zhou et al., 2015), was upregulated in the mutant, suggesting increased shoot stem cell proliferation in the Pvtb1 gene knockdown mutant.
Our transcriptomic analysis results suggest PvTB1a and PvTB1b regulate tillering in switchgrass by interacting with complex hormonal signaling pathways. Six cytochrome P450 genes were up regulated in the mutant ( Table S5). The members of cytochrome P450 family catalyze the biosynthesis of several phytohormones including auxin, brassinosteroids, and strigolactones which regulate branching across various plants species (Zhao, 2008;Kebrom et al., 2013). In addition, increased expression of ABA-responsive genes, PROTEIN PHOSPHATASE 2C (PP2C) genes, was observed in the mutant. These results suggest that Pvtb1 genes regulate bud development by modulating phytohormone biosynthesis and signaling. In Arabidopsis and maize, it has been shown TB1/ BRC1 promotes ABA accumulation and the expression of ABA response factors to inhibit bud outgrowth (Gonzalez-Grandio et al., 2013;Yao and Finlayson, 2015;Gonzalez-Grandio et al., 2017;Holalu and Finlayson, 2017). Although it is well-known that TB1/ BRC1 are involved in hormonal signaling pathways in different plant species, these regulation pathways are not conserved across various species (Kebrom et al., 2013). For instance, cytokinins repress the expression of TB1 in rice, while they act in a pathway independent of BRC1 in Arabidopsis (Aguilar-Martinez et al., 2007;Minakuchi et al., 2010). The timing of hormonal signaling in tillering has not been decided in switchgrass. Hence, more studies are needed to understand how the hormonal signals are involved in the Pvtb1-mediated regulation of branching.
Altered expression of genes in response to red or far-red light in the mutant suggest that Pvtb1 genes integrate the light signal to regulate tillering in switchgrass (Figure 7). FAR-RED-IMPAIRED RESPONSE1 (FAR1)-related sequence (FRS) family of transcription factors regulate plant growth and development in response to farred light in Arabidopsis (Wang and Wang, 2015). The homolog of FAR1, FAR-RED ELONGATED HYPOCOTYLS3 (FHY3) promotes shoot branching in Arabidopsis (Stirnberg et al., 2012). Further, FAR1/FHY3 promotes FHY1/FHL gene expression to facilitate phyA nuclear accumulation under far-red light condition. The downregulation of FRS TF genes in the switchgrass mutant suggested that Pvtb1 genes may be involved in promotion of phyA nuclear accumulation to inhibit the axillary bud initiation or outgrowth. Additionally, knockdown of Pvtb1 genes increased the expression level of the Phytochrome interacting factor 4 (PIF4) gene that has been shown to regulate expression of genes involved in cell expansion (Huq and Quail, 2002). Because PIF4 is a TF regulated by phyB-mediated signaling, its activity is regulated by the red light signal (Xu, 2018). These results suggest Pvtb1 genes regulate tillering through light signaling pathways. It has been reported that tb1 genes inhibit bud outgrowth in the process of shade-avoidance-syndrome (SAS) (Kebrom et al., 2013). In Arabidopsis and sorghum, the expression levels of BRC1 and SbTB1 were both upregulated under shade (Kebrom et al., 2006;Kebrom et al., 2010;Gonzalez-Grandio et al., 2013). Therefore, Pvtb1 genes might also sense the low R:FR ratio to inhibit bud outgrowth in switchgrass.
Several genes associated with stress/defense responses were significantly downregulated in the mutant. For example, 13 genes for Heat-shock proteins (Hsps)/chaperones which assist in protein refolding under stress conditions (Wang et al., 2004) were downregulated in the mutant. It has been shown overexpression of alfalfa (Medicago sativa L.) MsHSP70 gene could enhance Arabidopsis drought and heat stress tolerance (Li et al., 2017). Soybean (Glycine max (L.) Merril) Hsp 90 family members respond differentially to abiotic stresses and reduce the damage caused by abiotic stresses in Arabidopsis (Xu et al., 2013). Hence, the downregulation of Hsp genes in the switchgrass mutant suggest hastened outgrowth of axillary buds of the mutant might trigger increased expression of stress/defense responsive genes. Further, the expression levels of several WRKY TF genes responsive to chitin elicitation functioning in plant defense to fungal pathogens (Libault et al., 2007) also decreased in the mutant. In Arabidopsis, WRKY TFs are necessary for resistance to pathogen infection (Zheng et al., 2006) or resistance to abiotic stresses (Chen et al., 2010). It is well-known tolerance-growth trade-offs occur in plants under the low-resource conditions (Koziol et al., 2012;Bristiel et al., 2018). It has been shown that TB1 may influences sucrose levels and energy balance within dormancy buds in maize (Dong et al., 2019). Exploration of the mechanism of PvTB1s controlling energy balance in switchgrass would provide valuable information for the improvement of switchgrass for biomass production and development of enhanced stress-tolerant cultivars.

CONCLUSIONS
We successfully isolated mutated segments from chimeric mutants using micropropagation. This method overcomes the difficulties of obtaining nonchimeric mutants in selfincompatible species. Further, transgene-free mutants were obtained in this research, which provided valuable germplasm for switchgrass genetic research and breeding. More importantly, we proved the stable transmission of mutations induced by the CRISPR/Cas9 system in switchgrass. We propose that Pvtb1 genes negatively regulates tillering in switchgrass. RNA-seq analysis revealed a complex regulatory network potentially regulating tillering in switchgrass and provided some clues to the pathways of Pvtb1 genes.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories: The repository name is Sequence Read Archive (SRA). The project name is' Transcriptome sequencing of axillary buds of switchgrass Pvtb1 mutant and WT plant' The accession number is PRJNA660286. To access the data, please use the link: https://www.ncbi.nlm.nih.gov/Traces/study/? acc=PRJNA660286.

AUTHOR CONTRIBUTIONS
S-zF and YL conceived the original research plans. YL conducted the experiments on gene editing, micropropagation, and sequencing and analyzed relevant data. YB provided technical assistance to YL. WW conducted the hydroponic experiment. CC provided technical assistance to WW. YL and S-zF wrote the article with contributions from all authors. S-zF agrees to serve as the author responsible for contact and ensures communication.