In grass primary and secondary cell walls, hydroxycinnamate modifications of polysaccharides occur as acylation of the 5-carbon of arabinofuranosyl (Araf) decoration of the xylan backbone in glucuronoarabinoxylan (GAX; Figure 2 ) (Ishii, 1997; Ralph et al., 1998; Ralph et al., 2004; Buanafina, 2009; Bartley et al., 2013). Xylan is the most abundant polysaccharide other than cellulose in both primary and secondary cell walls of grasses, accounting for ~50% of grass hemicellulose (Scheller and Ulvskov, 2010), which represents, for example, 20-25% of dry switchgrass biomass (David and Ragauskas, 2010). Xylan is much more abundant in primary cell walls (PCWs) of grasses than in dicots (~30% compared to 5% of cell wall, respectively), displacing pectins as the most abundant non-cellulose polysaccharide. The 3-linked Araf decoration of xylan is rare or absent in dicots and the FA and pCA acylation of this Araf in GAX are believed to be completely specific to grass and other recently evolved monocots, known as commelinids (Harris and Trethewey, 2010). Grass GAX also possesses other substitutions on Araf, such as β-(1->2)Xyl-(1->2)Gal (Saulnier et al., 1995), β-(1->2)-Gal and β-(1->2)-Xyl (Wende and Fry, 1997; Chiniquy et al., 2012), and substitutions shared with dicot xylan, such as acetylation and (4-O-methyl-) glucuronosyl at the O2- position (Scheller and Ulvskov, 2010). Other HCAs also occur at lower abundance ester-linked to GAX in grasses. Recent mass spectrometry analysis of products of mild acidolysis of rice cell walls has detected caffeic acid on Araf of GAX (Feijao et al., 2022), and sinapate also occurs ester-linked to arabinoxylan in cereal grain (Bunzel et al., 2003). The presence of FA on GAX in particular confers a mode of cross-linking to grass primary cell walls absent in those of dicots since FA can undergo radical oxygen-mediated coupling to form ether bonds or C-C bonds, making diferulates and triferulates that result in xylan-xylan cross-linking (Takahama and Oniki, 1994; Bunzel et al., 2008) ( Figure 1C ). Across studies in various species and organs (Hatfield et al., 1999; Bartley et al., 2013; de Souza et al., 2018), the 8-5 and 8-O-4 dimers are often the most abundant diferulates ( Figure 1E ).

In PCWs, the functions of hydroxycinnamate modifications on GAX remain to be fully elucidated. The greater abundance of GAX and lower abundance of pectin and xyloglucan in grass PCWs compared to dicot PCW suggests grass GAX may have taken over some of the roles performed by pectin and xyloglucan, which is supported by solid-state NMR analysis of PCWs (Wang et al., 2014b). Potentially, the FA dimer and trimer cross-links on grass GAX partially substitute for the roles played by ionic cross-linking of pectin and for the oxidative cross-linking of extensin proteins in dicot PCWs. Consistent with this, the simplest form of extensins, those lacking a signaling domain, are not found in grasses; (Johnson et al., 2017). Also, FA is especially abundant per mass cell walls in very young tissue (Obel et al., 2002; Lin et al., 2016). Furthermore, abundance of cell wall FA and FA dimers was found to be negatively correlated with cell wall extensibility in wheat coleoptiles, suggesting an important role in control of PCW expansion (Wakabayashi et al., 1997). Additionally, hydroxycinnamates have antimicrobial properties (Akin, 2008) so their presence in grass primary cell walls also may serve to inhibit microbial attack and FA dimers ( Figure 1B, D ) may inhibit digestion (Grabber et al., 1998). Thus, the hydroxycinnamates on GAX in grass PCWs may confer evolutionary advantages by making young grass tissue with many expanding cells less readily digestible.

In lignified secondary cell walls (SCW) of both grasses and dicots, solid-state NMR suggests xylan in a twofold screw conformation (Xn^2f) binds to cellulose microfibrils (Simmons et al., 2016); whereas, distorted twofold or threefold screw xylan (Xn^3f) interacts closely with lignin (Kang et al., 2019; Duan et al., 2021). Thus, xylan bridges the two main components of SCW, although a study on sorghum SCW suggested Xn^2f was much less prevalent there (Gao et al., 2020). A clear difference in grass compared to dicot SCW is that the FA on GAX covalently bonds lignin via oxidative coupling, although the extent of this is difficult to determine (Ralph, 2010). GAX-FA is abundant on both Xn^2f and Xn^3f conformations in Brachypodium stems, and the authors proposed a model of grass SCW where FA on Xn^2f bound to cellulose crosslinks with other xylan FAs, and FA on Xn^3f covalently links to lignin (Duan et al., 2021). This model fits with several lines of evidence that show the abundance of GAX-FA and linkage of FA to lignin are correlated with recalcitrance to digestion of grass biomass (reviewed in (Buanafina, 2009; de Oliveira et al., 2015; Terrett and Dupree, 2019)). Thus, the FA-mediated linking of GAX to lignin inhibits access of hydrolytic enzymes to the cellulose to release glucose (the normal measure of digestibility).

In addition, a key role of FA in initial deposition of lignin is suggested by abundant Araf-FA coupled to coniferyl alcohol, the G-lignin monomer released from grass SCW by mild acidolysis (Lapierre et al., 2019; Feijao et al., 2022). This supports a model that GAX-FA act as the nucleation sites from which the lignin polymers grow that was developed from biomimetic studies of lignification of maize suspension culture cell walls (Grabber et al., 2002). Interestingly, this mode of nucleation with many separate sites (i.e. abundant GAX-FA) may explain the lower molecular weight of grass lignin polymers compared with those of other plants, which could allow some flexibility in developing tissue (Hatfield et al., 2017).

The role of pCA on GAX is less apparent than that of FA because pCA oxidatively couples much less readily than FA and whereas FA-GAX is found in every tissue in grasses, pCA-GAX has low abundance in stems (Fanelli et al., 2021; Möller et al., 2022). One possibility is that, analogous to the putative role of pCA on lignin discussed below, pCA-GAX may participate in radical transfer, thus catalyzing the oxidative coupling of neighboring FA on GAX. This is compatible with the observation that pCA on GAX rapidly increases in response to jasmonic acid application to Brachypodium callus (Hyde et al., 2018) which could be part of a priming of defense, allowing rapid cross-linking to occur in response to additional signals.

Lignin biosynthesis occurs by generation of three main monolignols (p-coumaryl alcohol, coniferyl alcohol, and sinapyl alcohol) in cytosol and subsequent radical coupling of these in the apoplast (Boerjan et al., 2003). Monolignols acylated by phenolic acids (especially pCA; FA; and p-hydroxybenzoate, a simple phenolic with two fewer carbons than hydroxycinnamates) and acetate, are now established as additional monomers of lignification in various species ( Figure 3 ). Attachment of pCA to lignin has been found in a diversity of grass species (Soreng et al., 2015) including maize, bromegrass, bamboo, sugarcane, elephant grass, rice (Withers et al., 2012; Karlen et al., 2016; Takeda et al., 2017), switchgrass (Shen et al., 2009), and Brachypodium (Petrik et al., 2014). Recently, p-coumaryl lignin was also found in other commelinid monocots (Zingiberales, Commelinales, and Arecales) (Karlen et al., 2018) and in the dicot mulberry (Moracacea) (Hellinger et al., 2022). Another phenolic acid, p-hydoxybenzoate, also occurs ester-linked to lignin in the poplar, willows, and oil palms (de Vries et al., 2021; Zhao et al., 2021). Like pCA, p-hydoxybenzoate does not readily oxidatively couple, so terminates lignin chains ( Figure 3 ). The acylation of monolignols by the other major hydroxycinnamate, FA, is a topic of great biotechnological interest because the FA becomes incorporated into lignin polymer via its propensity to oxidatively couple, thereby introducing alkaline-labile ester bonds ( Figure 3 ) making the lignin much easier to break apart (Wilkerson et al., 2014). We discuss this further in biotechnological applications below. It is now clear that FA-lignin occurs at low abundance naturally in all commelinids examined as well as sporadically within eudicots (Karlen et al., 2016).

The role of lignin acylation by hydroxycinnamates is uncertain. One possibility is that pCA (and hydroxybenzoate) moieties on lignin act as “radical catalysts.” Model studies of pCA show that it is readily oxidized. However, the fact that it has not been observed to oxidatively couple in muro has led to a model that oxidized p-coumaryl esters rapidly pass radicals to sinapyl alcohols, thereby facilitating lignin polymerization (Takahama and Oniki, 1994; Ralph, 2010). For the lower abundance acylation by FA, the biological functions are an open question.

Looking for candidate genes for addition of FA to GAX, Mitchell et al. (2007) searched for genes that are highly expressed in grasses while the most similar genes in dicots are much less expressed and differ substantially in protein sequence, since feruloylation is abundant in every grass tissue and absent in dicots. They found a small subclade of BAHD genes that met these criteria in Clade V [Clade Va of Tuominen et al. (2011)] and as acyl transferases these were postulated as involved in feruloylation. Furthermore, some of these BAHD grass genes are co-expressed with other genes responsible for GAX synthesis (Mitchell et al., 2007; Molinari et al., 2013). To facilitate communication about these grass BAHD acyltransferases, Bartley et al. (2013) called the group of 20 rice genes the “Mitchell Clade” and identified subclade i and subclade ii containing, Oryza sativa (Os) acyltransferases (AT), OsAT1-OsAT10, and OsAT11-OsAT20, respectively. Figure 4 shows subclade i for selected model and economically relevant grass species. Alternative names were proposed of the form BAHD01-BAHD20 (Molinari et al., 2013) and are used in some publications; here we show the equivalent names in Figure 4 but will use the AT nomenclature in the text. As discussed in greater detail below, grasses generally possess 8-10 subclade i ATs per haploid genome ( Figures 4 , 5 ). The function of Mitchell subclade ii genes, which have undergone grass species-/tribe-specific expansions/deletions (Karlen et al., 2016) remains unknown. Due to the absence of studies about them and their generally low expression (Bartley et al., 2013) we have excluded the subclade ii genes from this review and use “Mitchell subclade” to refer exclusively to subclade i.

There is now good evidence that several Mitchell subclade ATs are indeed involved in acylation of GAX, as predicted; while others acylate monolignols. Genetic manipulation of the Mitchell subclade was first achieved by Piston et al. (2010) by simultaneous downregulation of OsAT7, OsAT8, OsAT9, and OsAT10 in rice which resulted in decreased amounts of ester-linked FA in a cell-wall enriched fraction from leaves. The authors observed 2- to 3-fold reductions in gene expression of these ATs and an average of 20% reduction in cell wall FA content in the leaves, but due to use of constructs targeting multiple ATs could not determine which of the silenced genes were responsible and did not demonstrate that the FA was attached to GAX. Bartley et al. (2013) provided the first genetic evidence on single ATs involved in addition of hydroxycinnamates to GAX. Using mild acidolysis to break glycosidic bonds, they were able to show effects on a five-carbon sugar-esterified hydroxycinnamates, likely Ara-pCA and Ara-FA, released from rice cell walls (rather than saponification to release ester-linked FA and pCA from cell wall polymers, in general). They designated OsAT10 as a putative p-coumaroyl CoA arabinofuranose transferase (PAT) since overexpression of OsAT10, via an activation tagged line in rice (OsAT10-D1), induced a 5-fold increase in pCA levels in young green tissues. The observed increased saccharification yields, in the transgenic line were possibly due to concomitant 50% decrease in FA linked to GAX (Bartley et al., 2013). Since then, other studies also achieved several-fold increases in pCA-GAX by heterologous expression of OsAT10 in switchgrass (Li et al., 2018) and sorghum (Tian et al., 2021) and of sugarcane AT10 (ScAT10) in maize (Fanelli et al., 2021). As in Bartley et al. (2013) this was sometimes (Li et al., 2018; Fanelli et al., 2021), but not universally, accompanied by a decrease in FA-AX. Assumed to be an indirect effect, the mechanism of the alternating abundance of pCA-AX and FA remains an open question. Since then, a complete knock out mutants of OsAT10 in rice have been generated using CRISPR/Cas9 rice plants, leading to an almost complete lack of pCA-GAX, which was found to be most abundant in rice husks, compared to mature leaf and stems, of wild-type plants (Möller et al., 2022). The gene edited AT10 lines also exhibited an increase in FA, but no differences in cell wall composition or digestibility. Giving further support for the AT10 PAT function across species, a natural allele of the AT10 ortholog in barley (HORVU7Hr1G085100; Figure 4 ), predicted to encode a defective enzyme, has less ester-linked pCA and more ester-linked FA in grain cell walls (Houston et al., 2020).

Genetic analysis has provided various strengths of support for several ATs acting as putative feruloyl arabinofuranose transferases (FATs). RNAi silencing of Brachypodium BdAT1 showed an approximately 25% reduction in FA amounts and BdAT1 overexpression resulted in an approximately 15% increase in FA in leaves and stems (Buanafina et al., 2016). On the other hand, Mota et al. (2021) showed different results, with RNAi suppression of SvAT1, the Setaria viridis BdAT1 ortholog, decreasing pCA not FA on GAX. They therefore suggest that BdAT1 and SvAT1 have differing specificities for pCA-CoA and FA-CoA donors and conduct some protein structural modelling to support this. The clearest evidence of FAT activity thus far, was obtained through RNAi-based silencing of SvAT9 (SvBAHD01) in Setaria resulting in a 60% decrease in FA-GAX with a significant increase pCA-GAX; whereas, downregulation of OsAT9 ortholog in Brachypodium showed only small effects on FA (de Souza et al., 2018). AT7 ortholog downregulation in Brachypodium did not yield any significant changes in FA (Buanafina et al., 2016), though preliminary evidence suggested that a rice T-DNA insertion line for this gene has less leaf sheath FA (Bartley et al., 2013). In general, whilst genetic manipulation of ATs in grasses has achieved many fold increases and abolition in some tissues of pCA-GAX, this has not been reported for FA-GAX where the range of effects is narrower. This may point to a critical functionality of FA-GAX in grass cell walls where abolition would be lethal and large increases are difficult to achieve due to tight regulation.

Independent of the bioinformatics identification of the Mitchell subclade as candidates for GAX feruloylation, other groups established that some members of this subclade add pCA to lignin by acylating monolignols. Withers et al. (2012) showed that OsAT4 functions in vitro as a pCA monolignol acyltransferase (PMT) that transfers pCA from pCA-CoA onto H and S monolignols. The maize ortholog of OsAT3 also shows PMT activity, and RNAi suppression of the maize led to large decreases in pCA ester-linked to lignin (Marita et al., 2014). The strongest line showed a reduction in S lignin. Similarly, a complete knock-out mutant of BdPMT1, the Brachypodium ortholog of OsAT3, had <0.5% pCA on mature lignin; whereas pCA on GAX was unaffected. Conversely, overexpression of BdPMT1 boosted pCA-lignin above wild-type levels (Petrik et al., 2014). Heterologous expression of BdPMT1 and BdPMT2 (ortholog of OsAT8) in Arabidopsis, under the control of the Arabidopsis cinnamate-4-hydroxylase promoter, introduced pCA onto lignin, showing a gain of function since there is no pCA on lignin in wild-type Arabidopsis (Sibout et al., 2016). Though not focused on lignin modification, an early study found an enzyme from the commelinid species, Musa sapientum (i.e., banana alcohol acyltransferase, BanAAT), to have the highest activity on an aromatic acceptor substrate (Beekwilder et al., 2004). In retrospect, based on phylogenic analysis (Bartley et al., 2013), this enzyme is likely a banana PMT.

BAHD enzymes that acylate monolignols with FA have also been discovered. A feruloyl-monolignol transferase (FMT) from Chinese angelica [Angelica sinensis (As), a dicotyledonous medicinal plant, was heterologously expressed in hybrid poplar generating monolignol-FAs that were incorporated into lignin polymers (Wilkerson et al., 2014). The use of this activity to facilitate cell wall deconstruction are discussed further below (see Biotechnological Applications). The AsFMT is in Clade III of the BAHD superfamily, which is distant from the Mitchell subclade within Clade V. Surprisingly, overexpression of OsAT5 in rice increased feruloylated monolignols, suggesting that OsAT5 also encodes an FMT (Karlen et al., 2016). Thus, AsFMT and OsFMT are the result of convergent evolution, one of several examples in the BAHD family (Luo et al., 2007). A recent discovery on substrate specificity was made by Smith et al. (2022) looking at FMT and PMT enzymes from sorghum (Sorghum bicolor) and switchgrass (Panicum virgatum) as synthesized with wheat germ extract followed by in vitro characterization. The FMT enzymes, including OsAT5, produced both monolignol FA and monolignol pCA conjugates; whereas, the PMT enzymes produced exclusively monolignol pCA conjugates. A tolerance of differing acyl-CoA donors is another known feature of many BAHD enzymes (D'Auria, 2006).

The pCA-CoA and FA-CoA molecules that act as donors for the ATs are metabolites within the phenylpropanoid pathway that synthesizes monolignols. In grasses, recent evidence points to the presence of two largely separate pCA-CoA pools derived from the two phenylpropanoid pathway precursors phenylalanine and tyrosine (Barros et al., 2016; Wang et al., 2018; Simpson et al., 2021; Barros et al., 2022). Current thinking is that endoplasmic reticulum (ER)-associated cytochrome P450 enzymes, C4H, C3′H, and F5H, form a metabolon with the soluble enzymes, PAL, 4CL, HCT, facilitating metabolic channeling (Winkel, 2004; Bassard et al., 2012; Gou et al., 2018; Zhang et al., 2022). (See the legend of Figure 6 for the enzyme acronyms used here.) PTAL, a bifunctional enzyme, might also be involved in metabolon formation with ER-associated enzymes. Initial evidence for multiple pCA pools is the observation that PTAL’s tyrosine ammonia-lyase activity provides half the total lignin in Brachypodium stems, and wall-bound pCA with minimal contribution to wall-bound FA (Barros et al., 2016). Further, downregulation of C3′H and F5H in rice decrease unacylated G/S-lignin but do not alter amounts of p-coumaroylated G- or S-lignin; C3′H downregulation also significantly decreases wall-bound FA (Takeda et al., 2017; Takeda et al., 2018; Takeda et al., 2019). Thus, one explanation for the observation that unacylated monolignols and FA on GAX and pCA-monoligols appear to require different enzymes, is that there are separate pCA pools, though other explanations, such as metabolic compensatation (Vanholme et al., 2012), is possible.

Figure 6 summarize a model, which remains to be tested, of separate tyrosine and phenyalanine-derived hydroxycinnmate pools and partially distinct cell wall products. When tyrosine enters as a substrate to the PTAL, the product, pCA, is not utilized by C4H and hence escapes into the cytosol avoiding the initial metabolon channel. The pool of “escaped” pCA is utilized by the cytosolic enzymes C3H, COMT, 4CL, CCR, and CAD to produce a part of pCA-CoA, FA-CoA, and γ-hydroxycinnamoyl acylated monolignols. [C3H is a recently discovered cytosolic enzyme that directly catalyzes the 3-hydroxylation of 4-coumarate to caffeate, bypassing the previously known shikimate shunt involving C3’H and HCT (Barros et al., 2019)]. A part of the pCA-CoA and FA-CoA produced from the “escaped pCA” enters back to the monolignol pathway, which is supported both by the results with the PTAL mutant and the observation that heavy atom labeled tyrosine feeding studies in sorghum do result in labeled p-coumaryl shikimate (Simpson et al., 2021). Thus, these recaptured hydroxycinnamates can contribute to producing minor amounts of FA-CoA, utilized by FATs to substitute FA on GAX. In contrast, the phenylalanine precursor-based PAL/PTAL-ER-associated enzymes contribute to a major part of cell-wall-associated monolignols, FA-CoA, and p-CA-CoA which are utilized by FATs and PATs to decorate GAX. Recently, loss-of-function of two rice 4CL homologs, Os4CL3 and Os4CL4, differentially altered non-acylated and acylated monolignol content (Afifi et al., 2022, indicating divergent roles of 4CL protein isoforms and providing further support for the model. A final step of lignin acylation, it was recently hypothesized that monolignol-FA and monolignol-pCA are synthesized in the cytosol and exported into the cell wall by the same simple diffusion mechanism as monolignols (Vermaas et al., 2019).

Figure 6 also illustrates a plausible model for how Mitchell subclade ATs can be responsible for FA and pCA incorporation into arabinoxylans (AX). Biosynthesis of AX is carried out by glycosytransferases confined to the Golgi lumen where IRX9 and IRX10 proteins participate in a xylan synthase complex (Zeng et al., 2016) and grass XAT proteins mediate Araf decoration (Anders et al., 2012). However, the BAHD ATs are known to be cytosolic, as expected from their sequences which lack transmembrane domains and secretory pathway sequences. In addition, their hydroxycinnamoyl-CoA substrates are cytosolic and not known to occur in the Golgi lumen. Therefore, it seems that F/PATs must acylate a cytosolic precursor to AX synthesis just as P/FMTs acylate cytosolic lignin precursors. This conclusion can also explain the apparently surprising early result that feruloylation activity was found in the cytosolic fraction, not the membrane fraction, of rice cell cultures (Yoshida-Shimokawa et al. (2001); the Araf-Xylp-Xylp acceptor used there is presumably not the natural one but is sufficiently close to be recognized by an endogenous FAT).

The obvious candidate for the natural cytosolic AX precursor is UDP-β-L-arabinofuranose (UDP-Araf) since the UDP-arabinose mutase (UAM) responsible for its generation is localized outside the Golgi lumen, either in the cytsosol or to the Golgi perihpheral region (Konishi et al., 2011; Rautengarten et al., 2011). This is believed to be the last cytosolic step, and UDP-Araf would then be transported by a nucleotide sugar transporter (UAfT) into the Golgi lumen. Therefore, the simplest model is that cytosolic BAHD ATs catalyze the acylation of UDP-Araf to give UDP-Araf-FA/pCA as intermediates ( Figure 6 ). However, these putative products have not been identified despite targeted searches in grass tissues. One possibility is that these metabolites are only stable when bound to proteins, being generated by the action of ATs on UDP-Araf whilst this is still bound to UAM before it is transferred to the UAfT transporter (Hatfield et al., 2017). A protein complex involving both UAM and AT localized to Golgi periphery has been postulated (Hatfield et al., 2017) but proteomics from Brachypodium callus suggest that whereas UAM occurs both in peripheral and cytosolic fractions, ATs occur only in cytosol (JF and RACM, unpublished). One possibility is that UAM with bound UDP-Araf shuttles from the Golgi periphery to the cytosol where acylation of UDP-Araf occurs before returning to the periphery to engage with UAfT. This transporter could be similar to known UDP-Araf transporters (Rautengarten et al., 2017) with variation that permits the FA/pCA modification, which are small in comparison to UDP. Sharing most of the machinery for generating FA and pCA acylated UDP-Araf could also explain the apparent trade off in abundance of FA- and pCA-GAX in many experiments on different grass species when PAT or FAT expression is modified.

This model also necessitates a Golgi-localized GT enzyme to attach FA/pCA-Araf to the growing xylan molecule i.e. a xylan (hydroxcinnamoyl)-Araf transferase (XHAT; Figure 6 ). Addition of non-acylated Araf to xylan is mediated by grass-specific enzymes in GT family 61 (Anders et al., 2012) and it was reported that a closely related GT61 enzyme was responsible for addition of a xylosyl residue to GAX, so this enzyme was named XAX1 (Chiniquy et al., 2012). However more recent LC-MS analysis of sugar products released by mild acid treatment from the rice xax1 mutant suggests that XAX1 functions in the transfer of hydroxycinnamoyl-Araf to xylan, as all FA-Araf and pCA-Araf decorations of GAX were decreased in the mutant compared with the wild type (Feijao et al., 2022). This study therefore provides strong evidence that XAX1 is an XHAT responsible for the incorporation of FA/pCA-Araf onto xylan in the Golgi lumen.

Overall, the models in Figure 6 highlight the similarity of action of PMT/FMT and FAT/PAT in acylating hydroxyl groups on, respectively, monolignols and the Araf sugar in the cytosol, consistent with their similar primary sequences. Structural studies of these enzymes are required to understand the factors determining their specificities for acceptor and donor substrates.

The AT genes that modify GAX represent promising targets to improve the suitability of grass lignocellulosic biomass for biofuel production. Since FA on GAX is believed to be the main means by which polysaccharide is cross-linked to lignin, grass SCW FA amounts are therefore a key to recalcitrance. Suppression of the putative FAT-encoding SvBAHD01/SvAT9 in the model grass Setaria viridis resulted in a ~40% increases in ease of digestion of cell wall polysaccharides into sugars in the modified plants compared with the wild type (de Souza et al., 2018). Similarly, suppression of the ortholog in sugarcane (ScBAHD01/ScAT9) improved the digestibility of sugarcane straw by approximately 20% after Organosolv pretreatment, compared to non-transformed plants (de Souza et al., 2019). These results are exciting because sugarcane (Saccharum spp.) covers vast areas of land (around 25 million ha worldwide), and its processing is already linked into infrastructure for producing bioethanol in many countries, especially in Brazil. Furthermore, sugarcane straw and bagasses are the main industrial residues after sugarcane processing (Menandro et al., 2017). Also, the Organosolv process involves the use of an organic liquid and water to partially hydrolyze lignin bonds and lignin-carbohydrate bonds, resulting in a solid residue consisting of mainly cellulose and some hemicellulose (Zhao et al., 2009). Thus, the biomass of suppressed ScBAHD01/ScAT9 plants combined with Organosolv pretreatment is an interesting approach to be incorporated in the sugarcane industry for bioethanol production (de Souza et al., 2019). In addition, the reduction in FA-AX that often accompanies increases in pCA-AX due to altered expression of PATs in grasses, has been accompanied by a 10 to 40% increase in saccharification depending on the assay conditions. Thus, PAT enzymes like AT10 are also an attractive biotechnological target (Bartley et al., 2013; Li et al., 2018; Mota et al., 2021).

Recently, the world’s first CRISPR-edited sugarcane plants, the so-called Flex I and Flex II sugarcane, were reported (Brazilian Agricultural Research Corporation, 2021). Both Flex I and Flex II plants have CRISPR/Cas9-edited AT genes, and these sugarcane varieties presented higher cell wall digestibility and higher concentration of sucrose in plant tissues, respectively. The precise acyltransferase genes that were edited in these plants were not revealed, but both varieties have decreased levels of ferulate in the cell wall. Moreover, these CRISPR-edited plants were considered non-transgenic by the Brazilian National Technical Commission on Biosafety, representing an important step towards the use of this modified biomass by the bioethanol industry, as edited plants lacking foreign DNA can bypass the costly process of genetically modified-crop regulation.

Feruloyl lignin, i.e., lignin containing feruloyl monolignol conjugates (ML-FAs), facilitates depolymerization of lignin polymers by industrial processes due to the introduction of mild base-labile ester bonds into the lignin polymer. Early work on this technology revealed that incorporation of synthetic coniferyl ferulate into lignin of cell cultures enhanced alkaline delignification and enzymatic hydrolysis (Grabber et al., 2008; Ralph, 2010). Wilkerson et al. (2014) then introduced the Chinese angelica feruloyl-monolignol transferase (AsFMT) to poplar to generate ML-FAs that were incorporated into lignin polymers. The resulting biomass presented improved saccharification after mild base pretreatment (Wilkerson et al., 2014). The generation of the “zip-lignins” can be achieved either through a linear linkage, by extending the polymer chain, or by crosslinking two lignin polymers, as demonstrated elsewhere (Ralph, 2010; Rencoret et al., 2013; Lu et al., 2015; Smith et al., 2015; Kaal et al., 2018). Both biophysical and chemical changes in cell wall accessibility have been observed due to the introduction of ML-FAs in poplar lignin (Shen et al., 2019). As discussed above, FMT activity has also been demonstrated for AT5s in grasses (Karlen et al., 2016). One way to boost the effect of this endogenous enzyme was demonstrated in maize by suppression of the first lignin specific biosynthetic enzyme, cinnamoyl-CoA reductase (CCR) resulting in an increase in the intercellular pool of feruloyl-CoA and in ML-FAs and an overall decrease in lignin content thereby enhancing the digestibility of stem rind tissue (Smith et al., 2017). Similarly, ectopic expression of PMT genes increased saccharification yields under some reaction pretreatment conditions both in Brachypodium (Petrik et al., 2014) and via heterologous expression in Arabidopsis (Sibout et al., 2016). The mechanism could be due to the tendency of pCA-acylated monolignols to end lignin polymerization and not be included within the lignin polymer, consistent with the greater alkali solubility of Arabidopsis lignin esterified with pCA (Sibout et al., 2016). These results show that ML-hydroxycinnamate conjugates are a promising means for engineering bioenergy crops and waste streams of mainstream crops by conferring low-cost lignin breakdown and separation for biorefining applications.

Grains with increased feruloylated arabinoxylans are emerging as a potential multifunctional food and hydroxycinnamates themselves are being used as precursors for material applications. Some have reported that the presence of ferulic acid on the AX can contribute to antioxidant, anticancer and prebiotic properties (Srinivasan et al., 2007; Snelders et al., 2014). In addition, the cross-linking of FA-AX can form covalently linked gels with potential as drug delivery systems with anticancer or antioxidant properties (Mendez-Encinas et al., 2018). Recently, a study demonstrated the optimization of FA-AX isolation from wheat bran at a pilot scale using subcritical water extraction, demonstrating the feasibility of multifunctional FA-AX-based products for food and material applications in industrial scale (Rudjito et al., 2019). Furthermore, hydroxycinnamates in pure or mixed forms are themselves being used in higher value applications. For example, bacteria have been engineered to use HCAs to synthesize muconic acid (Johnson et al., 2016), a precursor of nylon. Techno-economic analyses indicate the value of engineering biomass to predominantly (>80%) produce only a single hydroxycinnamate for use as a high-value precursor (Karlen et al., 2020). Therefore, the manipulation of AT genes in different plant species can improve not only the production of biofuels but may also prove important for food and pharmaceutical applications.