Bioinformatics prediction of miRNAs that mainly relies on comparative genome-based EST analysis using known miRNAs in certain species is a well-established approach to discover conserved miRNAs in target species lacking genomic resources (Zhang et al. 2005). This method has been widely used for miRNA discovery in many plant species such as Arabidopsis (Wang et al. 2004; Adai et al. 2005), rice (Bonnet et al. 2004; Jones-Rhoades and Bartel 2004), cotton (Zhang et al. 2007), soybean (Zhang et al. 2008), tomato (Luan et al. 2010), brachypodium (Unver and Budak 2009), apple (Gleave et al. 2008), and other species.

Different known miRNA sequences as a query are used to search against NCBI’s switchgrass EST database. Matts et al. (2010) used miRNA sequences obtained from Arabidopsis (miRBase) as a query for general identification, and miRNA sequences from rice for identification of monocot-specific miRNAs (Matts et al. 2010). Xie et al. (2010) used 1699 known miRNAs from 29 plant species for switchgrass miRNA indetification (Xie et al. 2010). While Matts et al. (2010) used NCBI BLASTN as the search tool to find homologous miRNAs in switchgrass with the criteria of at least 18 nt and left 3 nt match (Matts et al. 2010), Xie et al. argued that BLASTN is not an ideal tool for miRNA discovery and might miss a lot of potential miRNA predictions (Xie et al. 2010). Instead, they adopted WATER to search against the EST database with the criteria of no >2 nt substitution (Xie et al. 2010).

The search by either BLASTN or WATER led to numerous hits among the ESTs, which were then subjected to a more strict screening by using different criteria. Matts et al. extracted the flanking region of the mature miRNA sequences and used a fold-back structure prediction software mFOLD to predict its secondary structures (Matts et al. 2010). These predicted secondary structures were then compared with those deposited in miRBase for verification (Matts et al. 2010; Xie et al. 2010). Xie et al. first removed repeated and protein-coding sequence hits, and then screened the rest of the hits by using 6 standards based on sequence complementarity between EST hits and query miRNA sequences, minimum length of pre — miRNA, secondary structure of predicted pre-miRNA, and sequence complementarity and structure of miRNA: miRNA* (Xie et al. 2010). Application of these criteria reduced some false positives and generated potential candidates for conserved microRNAs in switchgrass.

Switchgrass has a reputation for being difficult to establish, and several factors contribute to its status as a challenging crop. For example, small seed size, high seed dormancy, slow germination and poor seedling vigor can cause slow, and often poor stand establishment (Hsu and Nelson 1986b; Aiken and Springer 1995; Hintz et al. 1998; Evers and Parsons 2003; Parrish and Fike 2005). Economically successful production systems will require producers to achieve high yields during the early years after establishment (Schmer et al. 2006; Perrin et al. 2008). Adequate and timely preparation can go a long way to dispel the issues surrounding switchgrass’ reputation of poor establishment. In the following sections we discuss the many important management factors that must be addressed to enhance switchgrass establishment.

Site Selection

The choice of appropriate sites will be an important consideration for achieving economically viable switchgrass yields. A key presumption of many non-agronomically-oriented researchers and policy makers seems to be that marginal lands will produce suitable yields in switchgrass-for — energy cropping systems. Such an idea certainly is appealing in terms of minimizing competition for existing agricultural lands and to tamp down the food vs. fuel debate. It should be telling however, that the USDA NRCS technical field note on switchgrass establishment counsels producers to choose fields "typically used for row crop agriculture" (Douglas et al. 2009, pg. 1) in order to avoid steep slopes, irregular terrain, and wet sites. Thus, it seems that while switchgrass has broad suitability to sites, not all sites may be well-suited to switchgrass for bioenergy production given the issues associated with logistics, production, and sustainability.

In addition to beneficial bacterial endophytes, beneficial fungi also exist with the potential to enhance switchgrass performance. These beneficial fungi represent both mycorrhizae and endophytes. As with the bacterial endophyte interaction, root exudates, as well as CO2 release, play a role in stimulating development of the initial interaction, enhancing fungal spore germination, hyphal growth towards the root and hyphal branching (Giovannetti et al. 1993). The key exudate molecules are the strigolactones (Akiyama et al. 2005; Besserer et al. 2006), which form a concentration gradient helping the fungus to assess closeness of the host root.

As the fungal hyphae grow and then contact the root epidermis, each contacting hypha produces an appressorium (hypophodium), which flattens out and adheres to the epidermal cell surface. Through the production of localized cell wall-degrading enzymes, and the turgor pressure exerted by the contacting hyphal tip, the fungus is able to penetrate the epidermal cell wall (Bonfante and Perotto 1995). Once across the cell wall, the root host cell membranes invaginate to accommodate the fungus, resulting in the development of an apoplastic space between the fungus and plant cell, providing the interface for exchanges between both organisms (Vierheilig

2004) . In order to extend into the inner cortical tissues, a novel structure, the prepenetration apparatus (PPA) develops within the root, which helps direct the course of hyphal movement through the root (Genre et al. 2008). The PPA formation represents a tunnel or bridge through the cortical cells, with microtubules, microfilaments and rich in ER-cisternae. In addition, a reorientation of the plant cell nucleus occurs, with its movement to the site of fungal attachment, and it then leads and serves as a guide for the elongating PPA (Genre et al. 2008), which provides a tunnel for the growing hyphae as they colonize and grow towards the inner root cortical cells (Parniske 2008). The hyphae may grow along and between cells and eventually colonize the internal cortical cells, including longitudinally into adjacent cells, still under the guidance of the PPA (Genre et al. 2008). It is evident that key changes in growth and behavior of both plant and fungal cells take place to allow this process to occur, and a variety of genes/traits from both partners are involved in the success of this process (Gadkar et al. 2001; Genre et al. 2008; Parniske 2008).

Studies with various endophytic fungi have suggested that fungal entry can occur in the leaf through hyphae in wound sites, stomata, or penetration via appressoria (Ernst et al. 2003). Fungal growth tends to be primarily intercellular, having little effect on the surrounding host cells (Ernst et al. 2003; Gao and Mendgen 2006). As some non-clavicipitaceous fungi can be transferred vertically (Rodriguez et al. 2009), fungal growth may extend into inflorescence primordia, and eventually into the ovules with colonization of the scutellum and embryo axis of the seed (Rodriguez et al. 2009). In the case of root-colonizing fungal endophytes, root surface colonization was followed by direct hyphal penetration or through appressorium formation, and subsequent growth through epidermal and cortical cell walls (Gao and Mendgen 2006).

Lignin is a key polymer in vascular plant secondary cell walls that renders the cell wall impenetrable to solutes and enzymes, blocking biochemical conversion of biomass to biofuel. Lignin removal from plant biomass during pretreatment represents a key inefficiency of biochemical conversion. Hence, the enzymes involved in lignin synthesis are clear targets for modification in plants. Several reviews provide comprehensive information on the study of lignin synthesis, polymerization, acylation, and related topics (Hatfield et al. 2009; Penning et al. 2009; Ralph 2010; Vanholme et al. 2010; Harrington et al. 2012; Vanholme et al. 2012). Here, we will highlight results related to understanding and modifying lignin synthesis in switchgrass.

As with cell wall polysaccharide synthesis, our understanding of the pathway for lignin biosynthesis was determined primarily through genetic and biochemical studies of Arabidopsis (Vanholme et al. 2010; Vanholme et al. 2012). This work has been reinforced by the cloning of some of the so-called brown midrib mutants of the grasses maize and sorghum, the altered coloration of which is caused by aberrant accumulation of phenolics in lignified tissues (Harrington et al. 2012). Monolignols, also known as hydroxycinnamyl alcohols, are typically considered to consist of coniferyl alcohol, sinapyl alcohol, and p-coumaryl alcohol (Fig. 3). Their synthesis starts with the general phenylpropanoid pathway and then proceeds to the monoligonol-specific pathway, with deamination of phenylalanine by phenylalanine ammonium lyase (PAL) representing the first committed step of that pathway (Boerjan et al. 2003). Monolignol synthesis then proceeds via a series of phenyl ring hydroxylation and methylation modifications by the following enzymes: cinnamate 4-hydroxylase (C4H), p-coumaroyl shikimate 3′-hydroxylase (C3’H), caffeoyl-CoA methyltransferase (CCoAMT), ferulate 5-hydroxylase (F5H), and caffeic acid methyltransferase (COMT). These reactions are interspersed with a series of modifications leading to the reduction of the carboxylic acid at the end of the 3-carbon monolignol "tail", as catalyzed by the following enzymes: 4-coumaroyl ligase (4CL), hydroxycinnamoyl-CoA:shikimate transferase (HCT), cinnamoyl-CoA reductase (CCR), and cinnamyl alcohol dehydrogenase (CAD) (Boerjan et al. 2003).

Mutant studies have shown that decreasing plant total lignin content by manipulating key enzymes in the lignin biosynthesis pathway is a good way to reduce the recalcitrance of biomass. Forward or reverse genetics, especially in the dicots, Arabidopsis and poplar, showed that the down regulation of the genes that synthesize PAL (Baucher et al. 2003; Chen et al.

2006) , C4H, 4CL, HCT (Besseau et al. 2007), C3H (Abdulrazzak et al. 2006), CCoAOMT, CCR (Leple et al. 2007; Mir Derikvand et al. 2008), and CAD (Sibout et al. 2005) have an obvious effect on total lignin content. However, reduced lignin content is often associated with abnormal plant growth and development (Shadle et al. 2007; Mir Derikvand et al. 2008; Vanholme et al. 2008; Bonawitz et al. 2013). Other work has noted that increasing the S:G ratio by altering expression of lignin biosynthesis enzymes, such as by increasing the expression of F5H, improves processing efficiencies for pulp and biofuel (Stewart et al. 2009; Li et al. 2010).

Progress has also been made recently in unveiling the group of enzymes involved in the acylation of monolignols by p-coumarate (Ralph 2010). A BAHD acyltransferase in rice from the clade identified by Mitchell et al. (2007) as being differentially expressed in grasses compared to dicots was found in vitro to catalyze the acylation of monolignols with p-coumaroyl — CoA (Withers et al. 2012). This study provides a lead for the idea of engineering phenolic pathways to produce modified lignin precursors that contain ester or amide bonds and that are more efficiently processed to biofuels (Weng et al. 2008; Ralph 2010).

Recent publications have extended the study and manipulation of lignin biosynthesis enzymes to switchgrass. This work has been made more complicated by the fact that most of the lignin biosynthesis enzymes exist as large families of closely related proteins in the grasses, while in Arabidopsis there are fewer members (Tobias et al. 2008; Penning et al. 2009; Escamilla-Trevino et al. 2010; Saathoff et al. 2011; Saathoff et al. 2012). Down-regulation of Pv4CL1, one of the genes that encodes a homolog of 4CL in switchgrass, lowered lignin content and G subunits and enhanced saccharification efficiency by as much as 57% (Xu et al. 2011). Furthermore, silencing of a COMT gene decreased lignin content and S:G ratio, and enhanced bioconversion efficiency of lignocelluloses into ethanol by as much as 38% (Fu et al. 2011). Similarly, two groups published the results of silencing switchgrass genes that encodes CAD proteins (Fu et al. 2011; Saathoff et al. 2011). Again, these manipulations reduced the lignin content and improved digestibility by as much as 40% (Fu et al. 2011; Saathoff et al. 2011).

It was shown many bioenergy-related traits, such as biomass yield, seed yield, starch content, digestibility, etc., are quantitatively expressed in plants (Holland 2007; Rae et al. 2009). Heritability is important to plant breeders because it will help to evaluate the role of genetic factors and increase the efficiency of selection. Many studies about heritability were carried out for quantitative traits in switchgrass. Eberhart and Newell (1959) found the mean broad sense heritability for plant yield in two years was 0.78 for all tested strains from an upland population in Nebraska. Later the same research group estimated narrow sense heritability of plant yield ranged from 0.02 to 0.5, depending on different population types (Newell and Eberhart 1961). Talbert et al. (1983) reported the narrow sense heritabilities for plant dry weight were 0.25 and 0.59 based on individual and family means, respectively. But for plant height, average individual and family narrow-sense heritabilities were 0.80 and 0.83, respectively. Godshalk et al. (1986) studied narrow sense heritability for dry mass was 0.20 and 0.52, within and among half-sib families, respectively. Hopkins et al. (1993) reported narrow-sense heritability for forage yield was 0.22 for polycross families from an upland population in seeded rows (76-cm row spacing) in Nebraska. Missaoui et al. (2005a) found heritability of biomass in ‘Alamo’ was 0.6 for individual plants, 0.69 for family means, 0.76 for parent offspring regression in same environment and 0.45 in different environments. Casler (2005) analyzed variance for 49 switchgrass populations in two years at two locations, and showed broad-sense heritability for biomass yield was 0.63, for plant height was 0.90, and for maturity was 0.95. Rose et al. (2007) reported the narrow sense heritabilities for biomass yield grown in different condition were different: it was 0.73 for high yielding environment and 0.65 for low yielding environments when based on phenotypic family mean performance from half-sib populations. Boe and Lee (2007) indicated narrow-sense heritability estimates for biomass production in ‘Summer’ and ‘Sunburst’ switchgrass were 0.6. Bhandari et al. (2010) used 37 half-sib families and reported narrow-sense heritability for biomass yield ranged from 0.13 to 0.29, and stem thickness had low (<0.27) and plant height and tillering ability had low to moderate (0.26-0.48) heritability. Heritabilities were moderate (0.47-0.70) for heading, flowering, and plant spread and relatively high (> 0.82) for spring regrowth (Bhandari et al. 2010). Later Bhandari et al. (2011) used 46 full-sib families and indicated addictive genes were predominant to controlling biomass yield, tillering ability and spring regrowth. Heritability for plant height and stem thickness was smaller when estimated using full-sib families than half-sib families, while the plant spread was reverse. They suggested tillering ability, plant height and stem thickness could be useful indirect selection traits to improve biomass yield in switchgrass. These above-mentioned results suggest QTL analysis will be feasible to locate genomic regions responsible for quantitative traits and to explore their effects and interactions.

Although QTL analysis has been widely used for quantitative trait mapping, gene cloning, and marker assisted selection in major crop species (Kearsey and Farquhar 1998), little information is available in switchgrass due to the fact that it is a recently emerging bioenergy crop. However, significant progress has been made towards QTL mapping of important traits in switchgrass as three linkage mapping investigations have been reported independently: initial one was in a segregating population derived from a cross of one lowland ‘Alamo’ genotype, AP13 and one upland ‘Summer’ genotype, VS16 (Missaoui et al. 2005b), second one was from a population derived from a cross between selected genotypes of ‘Kanlow’ (lowland ecotype) as the female parent and ‘Alamo’ (another lowland ecotype) as the male parent (Okada et al. 2010); and the third population was derived from selfing a lowland genotype, ‘NL94 LYE 16Х13’ (Liu et al.

2012) . Respective QTL analyses with biomass and related traits, including heading date, spring growth vigor, tillering ability, plant height, base size, girth, stem thickness, plant spread, biomass yield, drought resistance, etc., through multiple-year and at multiple-locations are in progress (Y. Q. Wu, Unpublished).

Breeding of switchgrass as a tailor-made lignocellulosic feedstock has four major objectives: 1) increasing biomass yield under various field and geographic conditions, 2) decreasing input of switchgrass field production, 3) improving bioenergy feedstock quality, and 4) developing value-added switchgrass biomass feedstock. Transgenic approaches can substantially contribute to these targets. Switchgrass improvement via genetic transformation has just started. So far, the research has mainly focused on improving its quality as a biofuel feedstock by reducing lignin content and/ or altering lignin composition, biomass yield, and value addition.

Using appropriate herbicides reduces the time required to establishment and maximum biomass yields (Vogel et al. 2011). This approach is similar to the preferred method of establishing cool-season perennial grasses such as tall fescue when winter annual grass weeds are present. Butler et al. (2008) reported that sequential applications of glyphosate, one application in the spring to prevent annual grass weed production followed by second application in the autumn after rainfall and first flush of weed emergence, was very effective in increasing stand establishment. However, this technique is not well documented on switchgrass, therefore future research is needed.

Iron, one of the most abundant minerals on the planet, is not readily available to bacteria because its most commonly found form, ferric iron (Fe+3), is only slightly soluble and tightly bound to many particles in the soil. To gather iron needed for growth, bacteria and fungi secrete low molecular weight compounds called siderophores. Bacterial siderophores generally act to inhibit pathogenic fungi as a result of having higher affinity to iron than fungal siderophores (Ordentlich et al. 1988). Like many mechanisms of action in bacteria and fungi, environmental factors such as pH, nutrient levels including iron may affect synthesis of siderophores. Siderophore secretion has been confirmed in a number of bacterial taxa including Bacillus, Pseudomonas, Rhodococcus, Serratia, Obesumbacterium and Lysinibacillus (Czajkowski et al. 2012) as well as the fungal endophyte actinomycetes (Nimnoi et al. 2010). Genes encoding siderophores may be more difficult to introduce to other plant growth promoting endophytes since studies have shown that they are located in multiple loci (Osullivan et al. 1990) and have complex control mechanisms (Ovadis et al. 2004).

In addition to those described above, another intriguing approach to achieve process consolidation and cost reduction is in planta expression of lignocellulolytic enzymes. This approach fuses the discovery and development of advanced biocatalysts with engineering higher quality crop biomass. Indeed, expression of GHs and other hydrolytic enzymes in planta offer several advantages, including the following: (1) Growing transgenic plants in the field requires less energy than microbial production of the same enzymes; (2) Proteins can benefit from eukaryotic post-translational processing that can increase stability and activity; (3) Proteins can be targeted to subcellular compartments to prevent damage and allow high accumulation of the protein in the cell; and (4) Expression of proteins in planta juxtaposes the enzyme and the substrate, reducing enzyme demand needs during deconstruction due to inefficient mass transfer (Taylor et al. 2008; Sainz 2009). Reviews by Sticklen et al. (2006), Taylor et al. (2008), and Sainz et al. (2009) summarize numerous studies over expressing hydrolases in dicots and of greater relevance to switchgrass, in other grasses, including tall fescue, barley, maize, and rice. Here, we highlight key goals of, and new results related to, this approach.

Through the resources described above, switchgrass is becoming well — positioned for functional genomic studies. A major tool for functional genomic studies is genetic engineering of targeted genes facilitating a one to one tool relating DNA sequence with function. There are a large number of genome sequences currently available with many in progress allowing researchers a rich source of genes for potential manipulation for tolerance to abiotic stresses (Mittler and Blumwald 2010). A potential novel source of transgenes can be found by looking to sequences from organisms inhabiting extreme environments from desert adapted plants, to freeze tolerant fish, to even the diverse metagenomic projects being assessed for traits of functional interest (Mittler and Blumwald 2010). However, switchgrass is still considered recalcitrant for genetic modification, but significant progress has been made in optimizing transformation conditions and efficiency (see more detail in Chapter 9). There have been reports for Agrobacterium — mediated transformation with a range of efficiencies depending on the genetic background or genotype of the target (Somleva et al. 2008; Fu et al. 2011; Li and Qu 2011; Ramamoorthy and Kumar 2012). Particle bombardment of calli for switchgrass transformation has also been reported (Richards et al. 2001; Mann et al. 2011), but Agrobacterium infection of plants appears to be the method of choice for switchgrass transformation. A recent study suggests that a high-throughput and reproducible transformation system for the cultivar Alamo has been developed (Casler et al. 2011; Li and Qu 2011) reaching up to 90% efficiency. Other reports note that genetic manipulation of single genes can have a large genomic effect on biomass and conversion properties (Fu et al. 2011; Saathoff et al. 2011a; Xu et al.

2011) . The convergence of DNA sequence analysis and functional genomics demonstrate a trend toward understanding and manipulating gene function for desired traits.

Another approach to unraveling and targeting important pathways in switchgrass is to look directly at the protein and small molecule (metabolite) profiles through global proteomics and metabolomics. For example, the process of lignification is critical in biomass quality and the identification and characterization of the key enzymes involved is important in the identification of targets for manipulation. Through a proteomic approach, key enzymes such as cinnamyl alchol dehydrogenases have been identified (Saathoff et al. 2011b). Global studies into the switchgrass proteome have yet to be deployed, but unraveling the entire compliment of proteins and their modifications will lend key insight into the cellular physiology surrounding key traits. Metabolomics efforts are designed to study and characterize the unique chemical fingerprints that remain from specific cellular processes (Daviss 2005) and these approaches and cellular signatures are excellent tools in a functional genomics perspective for determining phenotype caused by genetic manipulation. Metabolomic analysis of switchgrass is still very few, but prospects to applying these techniques to identify elite lines for biofuel production are on the horizon.