Biochemical and Structural Analysis of Substrate Specificity of a Phenylalanine Ammonia-Lyase

The major phenylalanine ammonia-lyases from Sorghum bicolor were characterized through crystal structures, molecular docking, site-directed mutagenesis, and kinetic and thermodynamic analyses.

Abstract

Phenylalanine ammonia-lyase (PAL) is the first enzyme of the general phenylpropanoid pathway catalyzing the nonoxidative elimination of ammonia from l-phenylalanine to give trans-cinnamate. In monocots, PAL also displays tyrosine ammonia lyase (TAL) activity, leading to the formation of p-coumaric acid. The catalytic mechanism and substrate specificity of a major PAL from sorghum (Sorghum bicolor; SbPAL1), a strategic plant for bioenergy production, were deduced from crystal structures, molecular docking, site-directed mutagenesis, and kinetic and thermodynamic analyses. This first crystal structure of a monocotyledonous PAL displayed a unique conformation in its flexible inner loop of the 4-methylidene-imidazole-5-one (MIO) domain compared with that of dicotyledonous plants. The side chain of histidine-123 in the MIO domain dictated the distance between the catalytic MIO prosthetic group created from ¹⁸⁹Ala-Ser-Gly¹⁹¹ residues and the bound l-phenylalanine and l-tyrosine, conferring the deamination reaction through either the Friedel-Crafts or E₂ reaction mechanism. Several recombinant mutant SbPAL1 enzymes were generated via structure-guided mutagenesis, one of which, H123F-SbPAL1, has 6.2 times greater PAL activity without significant TAL activity. Additional PAL isozymes of sorghum were characterized and categorized into three groups. Taken together, this approach identified critical residues and explained substrate preferences among PAL isozymes in sorghum and other monocots, which can serve as the basis for the engineering of plants with enhanced biomass conversion properties, disease resistance, or nutritional quality.

Phenolic metabolism in plants plays important roles in providing aromatic amino acids, defense-related compounds, chemical attractants or repellents, structural support, UV protection, and color (Vermerris and Nicholson, 2006; Lattanzio et al., 2008). The biosynthesis of phenolic compounds in plants occurs via the concerted action of the shikimate and phenylpropanoid pathways, whereby substrates generated from the catabolism of Glc-6-P are converted to chorismate and p-coumaroyl-CoA, respectively (Aharoni and Galili, 2011). Phenylalanine ammonia-lyase (PAL; EC 4.3.1.5) is the first enzyme in the general phenylpropanoid pathway, and a significant amount of the total fixed carbon is directed through this enzyme (Maeda and Dudareva, 2012; Zhang and Liu, 2015). PAL catalyzes the nonoxidative elimination of ammonia from l-Phe to give trans-cinnamic acid. The 4-methylidene-imidazole-5-one (MIO) group is the coenzyme moiety established autocatalytically by the cyclization and dehydration of an Ala-Ser-Gly tripeptide within the sequence of PAL and a closely related enzyme, histidine ammonia lyase (HAL; Langer et al., 1997; Schwede et al., 1999)

Following PAL activity, the hydroxylation catalyzed by cinnamate 4-hydroxylase (C4H) gives rise to p-coumaric acid (Russell, 1971). Both trans-cinnamic acid and p-coumaric acid are precursors of myriad organic compounds with agricultural, nutritional, and industrial relevance, such as stilbenes, chalcones, flavonoids, cinnamoyl anthranilates, monolignols, lignans, and lignin (Treutter, 2006; Laskar et al., 2010; Vanholme et al., 2010; Schreiner et al., 2012; Cheynier et al., 2013; Shahidi and Ambigaipalan, 2015). Thus, control of the enzymatic activity of PAL and C4H can influence the pool of precursors and their fate, making these two enzymes attractive targets for the engineering of plants with enhanced biomass conversion, disease resistance, and/or nutritional quality.

The reduction of PAL and C4H activities in tobacco (Nicotiana tabacum) through antisense suppression resulted in both reduced content and altered subunit composition of lignin (Sewalt et al., 1997). In addition, PAL and C4H in tobacco were shown to colocalize to the endoplasmic reticulum membrane (Achnine et al., 2004). Thus, this physical association of PAL with C4H might establish a metabolic channeling complex on the endoplasmic reticulum surface, through which intermediates can be processed without unnecessary diffusion into the cytosol. This implies that modification of the interaction between PAL and C4H could have a much greater impact on metabolic flux than based purely on their catalytic mechanism. If the same is true among strategic grasses for bioenergy production, such as sorghum (Sorghum bicolor) and switchgrass (Panicum virgatum), in-depth knowledge about interaction between PAL and C4H together with a detailed understanding of the substrate-binding pockets of PAL and C4H will form the basis for engineering altered carbon flux.

While it is known that PAL enzymes in dicots utilize only l-Phe as substrate, PAL from some monocots, including maize (Zea mays; Rösler et al., 1997) and brachypodium (Brachypodium distachyon; Cass et al., 2015; Barros et al., 2016), also can deaminate l-Tyr, indicating their additional activity of tyrosine ammonia-lyase (TAL). While a His residue at position 123 appears to be critical for TAL activity (Röther et al., 2002; Watts et al., 2006; Hsieh et al., 2010), the biochemical basis and evolutionary benefit of TAL activity, which can bypass C4H in the production of trans-p-coumaric acid, are not well understood. Rösler et al. (1997) speculated that TAL activity might provide metabolic flexibility when Phe concentrations are temporarily low due to reduced biosynthesis, incorporation in proteins, or sequestration in the vacuole. Maeda (2016) suggested that PAL activity remained under selective pressure even though generating p-coumarate via TAL activity is energetically more efficient, because of the importance of cinnamic acid as a precursor for defense-related compounds. The importance of TAL activity in the biosynthesis of lignin in brachypodium was demonstrated by transgenic down-regulation of both BdPAL1 and BdPAL2 (Cass et al., 2015) or just BdPAL1 (Barros et al., 2016) via RNA interference. In both studies, lignin in the transgenic plants displayed a greater syringyl-to-guaiacyl ratio, consistent with the phenotypes observed in transgenic tobacco (Sewalt et al., 1997) and Arabidopsis (Arabidopsis thaliana) pal1/pal2 double mutants (Rohde et al., 2004). The change in syringyl-to-guaiacyl ratio observed in these different species, combined with gene expression analyses (Rohde et al., 2004; Cass et al., 2015; Barros et al., 2016), suggest the existence of complex regulatory mechanisms that alter flux through the different branches of the general phenylpropanoid pathways. Tracer studies in brachypodium with radiolabeled substrates indicated that close to half of the lignin susceptible to thioacidolysis originated from precursors generated from TAL activity (Barros et al., 2016).

The improved saccharification efficiency of biomass from transgenic brachypodium plants in which BdPAL1 and BdPAL2 were down-regulated (Cass et al., 2015) provides evidence that modifying PAL activity can enhance biomass conversion. Given that down-regulation of orthologs encoding monolignol biosynthetic genes in different grass species generates similar phenotypes (e.g. maize and sorghum mutants with reduced caffeic acid O-methyltransferase activity; Vermerris et al., 2007), it is plausible that modification of PAL activity will lead to improved biomass conversion in other grasses. Sorghum has been proposed as a strategic high-yielding biomass crop in the United States, because it is a C₄ species with substantial heat and drought tolerance. Its sequenced genome facilitates genetic improvement (Sarath et al., 2008; Paterson et al., 2009). Analysis of several sorghum brown midrib (bmr) mutants, which contain brown vascular tissue in the leaves and stems as a result of perturbations in the monolignol biosynthetic pathway, has shown that their biomass is more easily converted than that of their wild-type counterparts (Oliver et al., 2005; Vermerris et al., 2007; Saballos et al., 2008; Dien et al., 2009; Sattler et al., 2010, 2012; Jung et al., 2012) without major negative impacts on agronomic performance (Oliver et al., 2005), indicating that it is possible to balance changes in cell wall composition with plant productivity. The use of sorghum bmr mutants also can reduce the severity of thermochemical pretreatment, reducing both the cost of processing and the degradation of monomeric sugars (Dien et al., 2009; Godin et al., 2016). There are no known sorghum bmr mutants with a defective PAL gene. This is not entirely surprising, since the sorghum genome contains eight PAL genes (Xu et al., 2009), similar to the observed numbers in maize (10), rice (nine; Penning et al., 2009), and brachypodium (eight; Cass et al., 2015). Of the eight sorghum PAL genes, SbPAL1 is the most highly expressed (Shakoor et al., 2014). PAL expression appears to be sensitive to perturbations in the biosynthesis of monolignols, based on semiquantitative reverse transcription-PCR analysis of 13 sorghum bmr mutants (Yan et al., 2012).

Here, we report a comprehensive characterization of SbPAL1 including differential activity for l-Tyr versus l-Phe. The key residues that are responsible for PAL/TAL activity were delineated through the high-resolution crystal structure of the apo-form of SbPAL1, which, to our knowledge, is the first PAL structure from a monocot, followed by validation via site-directed mutagenesis.

RESULTS

Oligomeric State and Global Structure of SbPAL1

Recombinant SbPAL1 (75.6 kD) was purified and crystallized in a tetragonal space group I4₁22, and its structure was determined at 2.5 Å resolution (Table I). The lattice packing of SbPAL1 was established through a homodimer as an asymmetric unit, which, in turn, formed a tight interaction with a neighboring homodimer in a 2-fold symmetry manner, indicating its homotetrameric nature with a pseudo D₂ symmetry (Fig. 1A) as similar to other related ammonia-lyases and ammonia-transferases (Calabrese et al., 2004; Ritter and Schulz, 2004; Wang et al., 2005, 2008; Louie et al., 2006; Moffitt et al., 2007; Heberling et al., 2015). In order to investigate the plausible oligomeric state of SbPAL1 in solution, the PISA software package (Krissinel and Henrick, 2007) was applied to evaluate interactions between neighboring molecules in crystal lattices for predicting biologically relevant oligomeric states. The results clearly showed that SbPAL1 will form a stable homotetramer, as indicated in its crystal lattice, where the solvation free energy gain upon formation of the interface (ΔG_int) was estimated as −142.3 kcal mol⁻¹. Upon tetramerization, a solvent-exposed surface area of this tetrameric SbPAL1 was predicted to be 84,010 Ų, and the area buried due to tetramerization was 36,110 Ų.

Table I. X-ray diffraction data and refinement statistics for SbPAL1 (PDB identifier 6AT7).

Parameter	SbPAL1
Data collectiona
Space group	I4122
Cell dimensions
a, b, c (Å)	126.304, 126.304, 337.477
α, β, γ (°)	90.00, 90.00, 90.00
Resolution (Å)	44.66–2.49 (2.54–2.49)
Wavelength (Å)	1.00
Asymmetric unit	2
Total reflections	639,487
Completeness (%)	99.75 (98.8)
I/σ	22.57 (3.24)
CC1/2b	0.995 (0.854)
Redundancy	13.0
Rmeasc	0.109 (0.929)
Rpimd	0.030 (0.255)
Refinement
Resolution (Å)	44.66–2.49 (2.56–2.49)
Unique reflections	47,930
Rwork/Rfreee	0.160/0.202 (0.181/0.276)
B-factors (Å2)
All atoms	33.6
Solvent	38.2
Root mean square deviations
Bonds (Å)	0.003
Angles (°)	0.600
Ramachandran (%)
Favored	96.95
Outliers	0.45
No. of atoms
Protein and ligand	10,185
Water	446

^aNumbers in parentheses refer to the highest resolution shell.  ^bCC_1/2 is the correlation between two data sets each based on half of the data, as defined by Karplus and Diederichs (2012).  ^cR_meas is the multiplicity-weighted merging R factor.  ^dR_pim is the precision-indicating merging R factor.  ^eR_free was calculated as for R_cryst using 5% of the data that was excluded from refinement.

Figure 1.

Ribbon diagram representing the crystal structure of SbPAL1. A, Side and top views of tetrameric SbPAL1. SbPAL1 forms a homodimer in the crystallographic asymmetric unit. In solution, SbPAL1 forms a homotetramer with a nearby homodimer in a 2-fold symmetry. Subunits A and B are shown in yellow and green, respectively, and subunits A′ and B′ are shown in red and blue, respectively. B, Monomeric SbPAL1. The MIO domain is shown in cyan, the core domain in beige, and the shielding domain in magenta. MIO is represented as orange balls and sticks. Molecular graphics images were produced using the UCSF Chimera package.

Among 704 residues of SbPAL1, the electron density for the first nine residues of subunit A and three N-terminal residues, as well as residues 231 to 240 of subunit B in the dimeric asymmetric unit, and the last two C-terminal residues of both subunits were not resolved, probably due to their disordered nature. The Cα positions of the individual SbPAL1 subunits were superimposable with a root mean square deviation value of 0.65 Å. The B-factor values of the C_α atom indicate that three regions of SbPAL1, residues 90 to 110, 310 to 330, and 520 to 650, display high mobility. The root mean square deviation value between two subunits was reduced to 0.12 Å, without including those three high B-factor regions.

Each SbPAL1 subunit contained 20 α-helices and eight short β-strands. Overall, those secondary structural elements established three distinct domains: the MIO, shielding, and core domains, which were named in the 3D structure of PAL from parsley (Petroselinum crispum; Ritter and Schulz, 2004). As shown in Figure 1B, the residues spanning from Ser-10 to Thr-249 establish a MIO domain (cyan) that contains the MIO prosthetic group and a highly flexible inner lid-loop. This attached ammonium group was surrounded by the side chains from Leu-193, Asn-247, Tyr-338, and Phe-387. The shielding domain of SbPAL1, which is depicted in purple (Fig. 1B), spans from Leu-512 to Arg-636 and contains four α-helices. According to our search, this shielding domain seems unique to PAL and is absent in both TAL and HAL that have been deposited in the Protein Data Bank (PDB) so far. Generally, the closely related Phe aminomutases (PAM) have a similar sized shielding domain that shares ∼27% amino acid sequence identity with that of PAL. In the tetrameric configuration of SbPAL1, this shielding domain established an arch-like structure over the active site of the MIO domain. Lastly, the core domain, which is depicted in beige (Fig. 1B), connects those catalytic and shielding domains. The longest α-helix (α19: Gly-481 to Gln-529), which is located at the center of Figure 1B with two-thirds in beige and one-third in purple, runs through the core domain starting right under the stem of the MIO group to the end of the shielding domain.

Active Site of SbPAL1

In the crystal structure of each SbPAL1 subunit, the ¹⁸⁹Ala-Ser-Gly¹⁹¹ tripeptide displayed a well-defined electron density of the MIO motif. Each active site containing this MIO cofactor was established with residues from three neighboring subunits of a homotetramer. For example, the active site of the A subunit was constituted not only by the residues of the A subunit but also by the residues from the B and B′ subunits, which are related by noncrystallographic and crystallographic 2-fold symmetry, respectively. The individual imidazole-5-one ring of the MIO cofactor retained aromaticity where its N₃ atom showed a planar sp² conformation. On the other hand, the N₂ and N₃ atoms on the MIO ring established a hydrogen bond with the side chains of Tyr-338 and Asn-247, respectively.

The above-mentioned two flexible loops near the active site have been referred to previously as the inner lid-loop and the outer lid-loop (Louie et al., 2006). The inner lid-loop of SbPAL1 (residues 90–110) caps the active site, and the outer lid-loop of SbPAL1 (residues 310–330) flanks the inner lid-loop of the dyad-related subunit. It has been known that the conformations of these two loops vary significantly among aromatic ammonia-lyases. For example, the corresponding outer lid-loop of RsTAL from Rhodobacter sphaeroides (Louie et al., 2006) is longer than that of SbPAL1 or parsley PcPAL (Ritter and Schulz, 2004). The longer outer lid-loop of RsTAL seems to be pressing down the capping inner lid-loop and, thus, results in a tighter closure for the active site pocket. Due to the extra-tight closure, the Tyr-60 side chain of RsTAL is positioned closer to the substrate, providing the second proton acceptor for the E₂-type reaction for l-Tyr deamination catalysis (Louie et al., 2006). As PAL catalyzes a Friedel-Crafts-type reaction for the deamination of l-Phe (Hermes et al., 1985; Schuster and Rétey, 1995; Alunni et al., 2003; Louie et al., 2006; Watts et al., 2006; Pilbák et al., 2012), PAL does not require the second proton acceptor. Thus, the outer lid-loop is generally shorter in its structure. The shielding domain of SbPAL1, established by residues 520 to 650, is not present among TALs and HALs that have been characterized so far. Although the exact role of this domain unique to PAL is unknown, it could play a similar role as the outer lid-loop.

The Substrate Docking and Identification of Key Residues for Binding and Catalysis

Despite our numerous attempts, complex crystals of SbPAL1 with either substrate analogs or product were not obtained. Therefore, in order to understand the potential mode of substrate/product binding, an approach with molecular docking software was adopted through AutoDock Vina (Trott and Olson, 2010). Although a complex crystal structure for PAL has not been obtained so far, our docked position and conformation are very similar to those bound ligands in the structures of RsTAL (Louie et al., 2006) and 2,3-aminomutase from Streptomyces globisporus (Cooke and Bruner, 2010). The docking result clearly indicated that the substrate-binding pocket of SbPAL1 was formed with mainly hydrophobic residues. Ionic interaction was noticed between the guanidinium side chain of Arg-341 from subunit B (depicted with green in Fig. 2) and the carboxyl group of the l-Phe and l-Tyr docked in subunit A. In addition, for proper positioning of the aromatic portion of both bound substrates, the side chain of Lys-443 in subunit B′ and the aromatic rings of both substrates docked in subunit A were within an appropriate distance, establishing a potential cation-π interaction. The phenolic side chain of Tyr-338 from subunit B was located on top of the C_β of docked Phe, thus being located at the potential position for a general base during catalysis. In addition, the hydroxyl group of the Tyr-96 side chain from the inner lid-loop was positioned near the C_γ of the docked l-Phe. On the contrary, in the binding conformation of the l-Tyr molecule, the hydroxyl group of the same Tyr-96 was closer to its C_β instead. Thus, it is likely that Tyr-96 serves as one of the two required bases along with Tyr-338 for the E₂-type deamination of l-Tyr. Another noticeable difference between the docked conformation of l-Phe and l-Tyr was observed. The hydroxyl group of the l-Tyr substrate interacts with the imidazole side chain of His-123, which consequently shifts the MIO group closer to C_α of the bound l-Tyr. Absence of this interaction between the imidazole side chain of His-123 and the bound l-Phe positioned the ortho-carbon (C2) of its phenyl ring closer to the MIO group and the amine group closer to the side chain of Asn-474. In order to confirm the position and resulting interactions of bound substrate, the same molecular docking was performed with the model coordinates for the H123F-SbPAL1 mutant. The results confirmed that the loss of polarity in the H123F mutant oriented both l-Tyr and l-Phe in similar positions, where the ortho-carbon is closer to the methylidene carbon of MIO (Fig. 3).

Figure 2.

Active site of SbPAL1 with docked substrates. Two substrates, l-Phe (blue sticks) and l-Tyr (pink sticks), were docked to the active site of SbPAL1. The active site was completed as a tetramer in which three subunits of the tetramer participate. Subunits A, B, and B′ are shown in yellow, green, and blue, respectively. Molecular graphics images were produced using the UCSF Chimera package.

Figure 3.

Substrates docked into the active site of wild-type SbPAL1, H123F-SbPAL1, and H123Y-SbPAL1. A, l-Phe bound to SbPAL1. B, l-Tyr bound to SbPAL1. C, l-Phe bound to H123F-SbPAL1. D, l-Tyr bound to H123F-SbPAL1. E, l-Phe bound to H123Y-SbPAL1. F, l-Tyr bound to H123Y-SbPAL1. Three enzyme subunits that make up the active site are shown in gray, gold, and coral. Interactions between the bound substrate and nearby residues are marked with dotted lines. Molecular graphics images were produced using the UCSF Chimera package.

Steady-State Kinetics of Wild-Type and Mutant SbPAL1 and Isozymes

To confirm our hypothesis based on the crystal structure and molecular docking results, enzyme kinetic assays with the wild-type SbPAL1 and its site-directed mutants were performed. It is known that PAL loses activity fast without any reducing agents and that DTT also forms a covalent adduct to the MIO-N atom of PAL (Ritter and Schulz, 2004). Thus, β-mercaptoethanol (βME) was tested for its inhibitory effect. In the absence of βME, the k_cat and K_m for Phe deamination were 1.76 s⁻¹ and 0.34 mm, respectively. However, SbPAL1 activity was inhibited by βME competitively, where k_cat was unaffected but K_m was increased as the concentration of βME was increased (Fig. 4). The activity of SbPAL1 also was inhibited by tris-2-carboxyethylphosphine (Supplemental Fig. S1). Due to these inhibitory effects of reducing agents, all of our enzyme purification and kinetics measurements were performed within 24 h after harvesting the cells and in the absence of any reducing agents.

Figure 4.

SbPAL1 kinetics. A, Lineweaver-Burk plot of SbPAL1 in the presence of different concentrations of βME. Catalysis of l-Phe deamination was inhibited over increasing concentrations of βME. Enzymatic reaction was carried out in the presence of 10 mm (white circles), 5 mm (black triangles), 2 mm (black squares), and 0 mm (black circles) βME. B, Kinetic parameters of wild-type SbPAL1, its mutants, and two isozymes. The kinetic parameters k_cat, K_m, and k_cat/K_m are compared for wild-type SbPAL1, five SbPAL1 mutants (H123F, H123Y, F102Y, K443E, and Y96F), and two SbPAL isozymes (Sb06g022740 and Sb04g026520). NA, no measurable activity.

The deamination activity of SbPAL1 against four potential substrates, l-Phe, l-Tyr, l-His, and l-3,4-dihydroxy-Phe (l-DOPA), was assayed (Table II). Wild-type SbPAL1 showed catalytic activity against both l-Phe and l-Tyr, with k_cat/K_m values of 5.18 and 2.52 s⁻¹ mm⁻¹, respectively. SbPAL1 displayed activity against l-DOPA with a catalytic efficiency (k_cat/K_m) of 0.76 s⁻¹ mm⁻¹, which is only 14.6% of the activity displayed against l-Phe. SbPAL1 showed no detectable catalytic activity against l-His.

Table II. Kinetic activity of wild-type and mutant SbPAL1 and two isozymes.

The wild type and mutants showed no catalytic activity with His. NA (no activity) indicates that there was no measurable activity

Sample	Amino Acid	kcat	Km	kcat/Km
		s−1	mm	s−1 mm−1
Wild type	Phe	1.76 ± 0.037	0.34 ± 0.030	5.18
	Tyr	0.31 ± 0.004	0.12 ± 0.009	2.52
	l-DOPA	0.30 ± 0.006	0.40 ± 0.034	0.76
H123F	Phe	1.97 ± 0.032	0.06 ± 0.006	32.29
	Tyr	NA	NA	NA
H123Y	Phe	0.17 ± 0.010	0.23 ± 0.063	0.72
	Tyr	NA	NA	NA
F102Y	Phe	2.01 ± 0.092	2.57 ± 0.305	0.78
	Tyr	1.11 ± 0.051	2.89 ± 0.334	0.38
K443E	Phe	NA	NA	NA
	Tyr	NA	NA	NA
Y96F	Phe	NA	NA	NA
	Tyr	NA	NA	NA
Sb06g022740	Phe	2.34 ± 0.096	4.02 ± 0.376	0.58
	Tyr	0.37 ± 0.008	0.42 ± 0.037	0.89
Sb04g026520	Phe	2.20 ± 0.050	1.52 ± 0.102	1.45
	Tyr	NA	NA	NA

To confirm the effect of the above-mentioned His-123, the enzyme kinetics of H123F- and H123Y-SbPAL1 mutants were tested. The H123F-SbPAL1 mutant showed 6.2-fold elevation in catalytic efficiency (k_cat/K_m) for l-Phe, mainly due to a 5.7-fold decrease in K_m. However, H123F-SbPAL1 lost activity for l-Tyr and, thus, became a dedicated PAL. For the H123Y-SbPAL1 mutant, the catalytic efficiency for l-Phe was reduced to 0.72 s⁻¹ mm⁻¹, mainly due to a 10-fold decrease in the turnover rate (k_cat). The H123Y-SbPAL1 mutant also lost activity against Tyr. Neither the H123Y- nor the H123Y-SbPAL1 mutant showed any activity with l-His (Table II; Fig. 4B).

To further confirm a hypothetical role of Phe/His-123 in substrate preference, two additional PAL isozymes, encoded by the sorghum genes Sb06g022740 and Sb04g026520, were tested. At the corresponding position of His-123 of SbPAL1, Sb06g022740 has a His and Sb04g026520 has a Phe. Consistent with the kinetics data of SbPAL1 and the H123Y-SbPAL1 mutant, Sb06g022740 catalyzed the deamination of both l-Phe and l-Tyr with k_cat values of 2.34 and 0.37 s⁻¹, respectively. In contrast, Sb04g026520 only catalyzed the deamination of l-Phe effectively, with a k_cat of 2.2 s⁻¹ (Table II; Fig. 4B).

Close examination of the crystal structure of SbPAL1 also suggested that the spatial position of residue Phe-102 is close to and almost parallel with the side chain of Tyr-96 (Fig. 2) and that an F102Y mutation may provide an alternative catalytic base when the flexible inner lid-loop is in motion. Because the side chain of Tyr-96 serves as the second catalytic base during the deamination reaction of l-Tyr, an F102Y mutation might increase the catalytic activity of SbPAL1 for l-Tyr. To test this hypothesis, the enzyme kinetics of an F102Y-SbPAL1 mutant were examined. As predicted, the turnover rate for l-Tyr was improved 3.6-fold, whereas the turnover rate for l-Phe was unaffected. However, K_m values for both l-Tyr and l-Phe were increased by 7.6- and 24-fold, respectively, which caused reduced enzyme efficiency for both substrates. The F102Y mutant did not show any activity with l-His (Table II; Fig. 4B).

To examine the hypothetical role of residue Lys-443, the enzyme kinetics of a K443E-SbPAL1 mutant were tested. This mutant showed no measurable activity against any of the three substrates, l-Phe, l-Tyr, and l-His (Table II; Fig. 4B), consistent with our prediction about its role for proper positioning of the aromatic portion of the bound substrates through a potential cation-π interaction.

According to the proposed mechanism, residue Tyr-96 is one of the two bases involved in an E₂-type deamination of l-Tyr, but it is not required for a Friedel-Crafts-type deamination of l-Phe (see Fig. 8 below). To examine whether SbPAL1 uses two different mechanisms, the activity of the Y96F-SbPAL1 mutant was tested for both substrates. However, the Y96F-SbPAL1 mutant displayed only trace amounts of activity for both l-Tyr and l-Phe (Table II; Fig. 4B). A subsequent isothermal titration calorimetry (ITC) experiment also showed no measurable affinity between the Y96F-SbPAL1 mutant and the reaction products, trans-cinnamic acid and p-coumaric acid, which reflects significant perturbation of the local conformation. Consistent with that, the equivalent mutation (Y110F) in PcPAL resulted in an almost inactivated enzyme (Röther et al., 2002).

Figure 8.

Mechanism of SbPAL1. A, l-Phe deamination. The methylidene electrophile of MIO is attacked by the ortho-carbon of the substrate, l-Phe, initiating a Friedel-Crafts-type deamination. B, l-Tyr deamination. Due to the different substrate-binding mode, the methylidene electrophile of MIO is attacked by the amide nitrogen of the substrate, initiating an E₂ reaction.

ITC for Wild-Type SbPAL1

ITC was performed to determine the thermodynamic parameters for the products and the substrate analogs. As shown in Figure 5A, a large amount of heat was released (ΔH = −2.3 kcal mol⁻¹) when trans-cinnamic acid, the deamination product of l-Phe, was used as the titrant into the SbPAL1 solution. This enthalpic change also was accompanied by a substantial entropic contribution upon ligand binding (ΔS = 12.6 cal mol⁻¹ K⁻¹), resulting in a K_d of 34.1 μm (Fig. 5A). On the other hand, p-coumaric acid, a product formed from the deamination of l-Tyr, bound to SbPAL1 with a K_d value of 2.14 μm and with ΔH of −9.1 kcal mol⁻¹ and ΔS of 6.29 cal mol⁻¹ K⁻¹. A higher affinity of SbPAL1 for p-coumaric acid than for trans-cinnamic acid is likely due to the hydrogen bond formation between the p-hydroxyl group of p-coumaric acid and the nearby imidazole side chain of His-123 (Fig. 3B). In addition, the K_d value for caffeic acid, a product formed from the deamination of l-DOPA, was 6.21 μm, with ΔH of −2.1 kcal mol⁻¹ and ΔS of 16.7 cal mol⁻¹ K⁻¹. According to the ITC results, SbPAL1 did not show any cooperative binding to either of the products, nor did it display any significant affinity for either d-Phe or d-Tyr.

Figure 5.

ITC assay. A, Binding affinity between the wild type, Y96F-SbPAL1, and reaction products. The trend of heat released by serial injections of cinnamate (black squares), caffeate (black circles), and p-coumarate (black triangles) indicates that p-coumarate (K_d = 2.14 µm) binds tighter to SbPAL1 than cinnamate (34.1 µm) and caffeate (6.21 µm). Y96F-SbPAL1 displayed no affinity to either cinnamate (white circles) or p-coumarate (white squares). B, Binding affinity between the wild type, H123F-SbPAL1, and AIP. The PAL-specific inhibitor AIP showed higher affinity to the H123F-SbPAL1 mutant (black squares; K_d = 0.26 µm) than to the wild type (white circles). K_d for the wild type was above the ITC detection limit; hence, the value is not stated. Solid lines represent the least-square fits to the data. C, Lineweaver-Burk plot of SbPAL1 in the presence of different concentrations of AIP. Enzymatic activity was inhibited over increasing concentrations of AIP. Enzymatic reaction was carried out in the presence of 20 µm (black triangles), 5 µm (black squares), and 0 µm (black circles) AIP. K_i calculated from the kinetics experiment was 0.22 mm.

Our inhibition assays with 2-aminoindan-2-phosphonic acid (AIP) indicated its minor competitive inhibition of SbPAL1 activity. The K_i value for AIP, estimated from enzyme kinetics, was 0.22 mm, which was near the detection limit of our calorimeter. Being consistent with this result, the K_d value between SbPAL1 and AIP from ITC measurement was insignificant. The low affinity of SbPAL1 to AIP was somewhat expected, as AIP is a PAL-specific inhibitor. SbPAL1 has both PAL and TAL activities (Table II), and the conformation of its active site has characteristics of both PAL and TAL (Fig. 3). In supporting these, H123F-SbPAL1 showed higher affinity to AIP with a K_d value of 0.26 μm, with ΔH of −3.2 kcal mol⁻¹ and ΔS of 19.3 cal mol⁻¹ K⁻¹ (Fig. 5, B and C).

Structural Homologs of SbPAL1

To identify homologs of SbPAL1 and correlate their sequences with known substrate specificities, the amino acid sequence of SbPAL1 was used to perform a similarity search with the deposited crystal structures in the PDB using BLAST (Altschul et al., 1997). In addition, a chain fold search using DALI (Holm and Sander, 1993) was performed using the atomic coordinates of SbPAL1 to identify closely related structural homologs in the same PDB. A BLAST search to identify proteins with similar amino acid sequences in the PDB revealed that PAL from parsley (PDB identifier 1W27) showed the highest identity (69%) to SbPAL1, followed by PAM (PDB identifier 4C6G) from Taxus chinensis (Chinese yew) with 45% identity and PAL from the yeast Rhodosporidium toruloides (PDB identifier 1Y2M) and the cyanobacterium Nostoc punctiforme (PDB identifier 2NYF) with substantially lower values of 36% and 35% identity, respectively. Overall, DALI search results were similar to those of BLAST searches. PAL from parsley was again the most similar 3D structure, with a Z-score of 53.9, followed by PAM from T. chinensis (Z = 49.5), TAM from S. globisporus (PDB identifier 2RJR; Z = 40.9), PAL from Anabaena variabilis (2NYN; Z = 40.6), PALs from R. toruloides (1Y2M; Z = 38.7) and Nostoc punctiforme (2NYF; Z = 40.4), PAM from the bacterium Pantoea agglomerans (PDB identifier 3UNV; Z = 39.6), TAM from S. globisporus (2OHY; Z = 39.6), and HAL from Pseudomonas putida (1B8F; Z = 39.5). As shown in both DALI and BLAST searches, SbPAL1 showed significant similarity with a group of plant PAMs, displaying 45% to 46% amino acid sequence identity. Both sequence alignment (Fig. 6) and secondary structure matching of the above structures displayed a clear correlation, where all eukaryotic PALs and PAMs contain the shielding domain with a shorter outer lid-loop. On the other hand, TAL, HAL, and prokaryotic PAL/PAM lack the shielding domain and contain a longer lid-loop instead. Among those, the inner lid-loop of PcPAL was significantly different from the others and was referred to as an open active-site form (Louie et al., 2006; Heberling et al., 2015). However, due to its poor fitting into electron density, the conformation of the inner lid-loop in the deposited PcPAL structure is uncertain.

Figure 6.

Sequence alignment of structural homologs. Sequences of SbPAL1, parsley PAL (PDB identifier 1W27), T. chinensis PAM (4C6G), R. toruloides PAL (1Y2M), N. punctiforme PAL (2NYF), S. globisporus TAM (2RJR), A. variabilis PAL (2NYN), P. agglomerans PAM (3UNV), S. globisporus TAM (2OHY), and P. putida HAL (1B8F) are aligned. The outer lid-loop and shielding domains are indicated with red and black boxes, respectively.

Sequence Comparison and Activity Assay for the Putative SbPAL Isozymes

In addition to SbPAL1, there are seven other genes encoding putative PAL isozymes in the genome of sorghum. According to the alignments of the deduced amino acid sequences (Fig. 7A) and phylogenetic analyses (Fig. 7B), those eight isozymes can be categorized into three groups. The first group contains Sb04g026510 (SbPAL1) and Sb06g022740. The second group contains Sb04g026520, Sb04g026560, Sb04g026550, Sb04g026530, and Sb04g026540. The remaining Sb01g014020 is distinct from those in the other two groups. Significantly, the first group has His at the earlier mentioned residue 123, the second group has Phe, and Sb01g014020 has Tyr. In addition, all isozymes have Lys at residue 443 except Sb01g014020, which has Asn. Expression data in Phytozome (https://phytozome.jgi.doe.gov/pz/portal.html#!info?alias=Org_Sbicolor) indicate that gene Sb01g014020 is expressed only at very low levels in roots. Based on its expression profile, the protein encoded by this gene is not expected to play a significant role in phenylpropanoid metabolism. However, to understand the effect of Tyr in Sb01g014020 at the corresponding position of His-123 in SbPAL1, the enzyme kinetics of an H123Y-SbPAL1 mutant were examined. H123Y-SbPAL1 was inactive under the test conditions for all three tested substrates: l-Phe, l-Tyr, and l-His (Table II). Our kinetics results show that the first group, SbPAL1 and Sb06g022740, which have His at residue 123, is bifunctional. In contrast, Sb04g026520, which has Phe at residue 123, displayed only PAL activity, similar to what was observed for the H123F-SbPAL1 mutant (Table II).

Figure 7.

PAL isozymes of sorghum. A, Sequence alignment. Eight PAL isomers are aligned; Sb04g026510 is SbPAL1. The regions of α-helix and β-strand are represented with coil and arrow bars, respectively. Corresponding positions of the SbPAL1 His-123 residue are highlighted with green for all eight isomers. B, Phylogenetic tree. The optimal tree with the sum of branch length of 0.573 is shown. Branch length represents the evolutionary distances that were computed using the Poisson correction method and is in units of amino acid substitutions per site. The evolutionary history was inferred using the minimum evolution method in MEGA7 (Kumar et al., 2016) that uses the close-neighbor-interchange algorithm at a search level of 1.

DISCUSSION

The overall structure of SbPAL1 was predominantly α-helical, where 52% of its residues are in 23 α-helices. PAL from parsley (PcPAL), which is the most similar in terms of both amino acid sequence and 3D structure, according to our BLAST and DALI searches, shares amino acid sequence with 70% identity with SbPAL1. PcPAL, so far, is the only plant PAL for which a crystal structure is available in PDB. PAL from dicots such as PcPAL, however, are only able to use l-Phe as a substrate and lack the ability to catalyze the deamination of l-Tyr. In contrast, several tested monocot PAL enzymes, including SbPAL1, are able to catalyze the deamination of both l-Phe and l-Tyr. Thus, the TAL activity of SbPAL1 and similar enzymes in other grasses provides an alternative route for generating p-coumaric acid without a need for C4H. The catalytic efficiency (k_cat/K_m) for TAL activity of SbPAL1, however, is 2-fold lower than that of its PAL activity (Table II). On the other hand, the catalytic efficiency for TAL of Sb06g022740, another bifunctional PAL in lignified tissues, is 1.5-fold higher than that of its PAL activity, although its TAL and PAL efficiencies are 35% and 11% SbPAL1, respectively (Table II). Barros et al. (2016) determined that BdPAL1 had a 2-fold greater activity against l-Tyr than l-Phe and that close to half of the lignin susceptible to degradation via thioacidolysis originated from monolignols that could be traced back to TAL activity. Enzymatic properties of ZmPAL1 determined by Rösler et al. (1997) suggest that maize has similar activity toward both substrates and, thus, is intermediate between sorghum and brachypodium. The combined results from these three species, taken together with the observation that different grass species have different numbers of PAL homologs encoding enzymes with both PAL and TAL activities (Barros et al., 2016), imply an inherent variation among individual grass species in the regulation of metabolic flux through the general phenylpropanoid pathway and metabolic pathways leading to specific classes of phenylpropanoids. These differences may reflect adaptation of the individual grass species to the specific environments where they evolved, which differ in terms of climate (temperature, water, photoperiod, and light intensity), pests, and pathogens. This variation in selective pressure may explain the observed variation in photosynthesis (C₃ versus different forms of C₄; Williams et al., 2013) and the ability to produce certain metabolites, including phenylpropanoids. For example, despite a close evolutionary history, sorghum is able to produce the antifungal 3-deoxyanthocyanidins, whereas maize is not (Snyder and Nicholson, 1990).

Active Site and Reaction Mechanism

The diffusing-in substrate docks the active site through replacing water molecules bound in the active site, as indicated in our product ITC data (ΔS = 12.6 and 6.29 cal mol⁻¹ K⁻¹). In the active site, both l-Phe and l-Tyr establish a salt bridge between their respective α-carboxyl group and the guanidinium side chain of Arg-341 and the π-cation interaction between their aromatic ring and the amine side chain of Lys-443. However, l-Tyr seems to be able to form a tighter bond than l-Phe through an additional hydrogen bond with the imidazole side chain of His-123. On the other hand, l-Phe interacts with the side chain of Asn-474. Obviously, the flexible inner lid-loop must be displaced for substrate to enter, of which length and flexibility will affect the K_m and k_cat, as suggested before (Ritter and Schulz, 2004; Pilbák et al., 2006; Wang et al., 2008). In addition, phosphorylation of the hydroxyl side chain of Thr-536 in SbPAL1 could affect this binding and subsequent catalytic events, as proposed for PAL enzymes from Phaseolus vulgaris (Allwood et al., 1999) and parsley (Ritter and Schulz, 2004). As indicated previously in the structure of R. toruloides RtPAL (Calabrese et al., 2004), the prosthetic MIO of SbPAL1 also is under direct influence of the positive poles of three helices, which form a triple coiled-coil motif, possibly increasing the electrophilicity of MIO (Fig. 1). Due to the existence of a hydroxyl group in l-Tyr and its consequent binding mode, SbPAL1 appears to catalyze the deamination of l-Tyr and l-Phe differently. l-Phe in SbPAL1 seems to undergo a Friedel-Crafts-type deamination, as suggested before (Hermes et al., 1985; Schuster and Rétey, 1995; Alunni et al., 2003; Louie et al., 2006; Watts et al., 2006; Pilbák et al., 2012), where the ortho-carbon of the aromatic ring of l-Phe forms a covalent bond with the electrophilic methylidene carbon. Once that bond is formed, the hydroxyl group of Tyr-338, which is within 2.96 Å from the methylidene carbon and 2.84 Å from the N₂ atom of the MIO, deprotonates C_β of the substrate. While the phenolic side chain of Tyr-338 acts as a general base, its protonation can be stabilized by the hydrogen bond with the N₂ atom of the MIO and the amide side chain of Gln-475. Then, as the bond between the MIO and the intermediate dissociates, the amine-C_α bond of the intermediate breaks, producing cinnamate and an ammonium ion (Fig. 8A). On the other hand, the phenolic side chain of Tyr-338 deprotonates to the amine group of the bound l-Tyr before it establishes the N-MIO intermediate. Then, an E₂ elimination step is catalyzed by deprotonation of the intermediate C_β by the side chain of Tyr-96, producing p-coumarate. The hydroxyl group on this Tyr-96 is within a hydrogen bond distance from the backbone amide nitrogen of Gly-103 and nearby water molecules, as in the case of RsTAL (Louie et al., 2006). Thus, it is plausible that the hydroxyl group of Tyr-96 is connected to a proton network and has reduced pK_a, relaying a proton back and forth from the bulky solvent. The amine group is released subsequently from MIO (Fig. 8B). In this reaction, His-123 of SbPAL1 provides a differential positioning of the two substrates, l-Phe and l-Tyr, which was identified previously in bacterial PAL (Louie et al., 2006; Moffitt et al., 2007) and confirmed by our kinetic and molecular docking assays.

Substrate Specificity and PAL Isozymes in Sorghum

As suggested by previous studies (Watts et al., 2006) and our results, Phe/His at residue 123 serves as the factor determining PAL or TAL activity, with Phe-123 stipulating PAL activity and His-123 TAL activity. As our molecular docking results indicated, an apparent establishment of a hydrogen bond between the nitrogen atom on the His-123 side chain and the hydroxyl group of the l-Tyr substrate might orient its amine closer to the methylidene carbon of the MIO group, leading to an N-MIO intermediate followed by an E₂-like transition state. Quantum mechanics/molecular mechanics calculations in RsTAL determined that the hypothetical Friedel-Crafts route for ammonia elimination from l-Tyr was less likely due to the high energy of Friedel-Craft-type intermediates (Pilbák et al., 2012).

Without this hydrogen bond, as is the case in the H123F-SbPAL1 mutant, the ortho-carbon of the phenyl ring is located closer to the MIO, leading to a deamination reaction of the Friedel-Crafts type. Our enzyme kinetic data indicate that SbPAL1 effectively deaminates both l-Phe and l-Tyr, but PAL activity displays higher catalytic efficiency. The H123F-SbPAL1 mutant displayed only PAL activity. Consistent with these observations, the monocotyledonous ZmPAL, BdPAL1, and SbPAL1, all of which have both PAL and TAL activities, contain His at this position, whereas dicotyledonous PALs, such as the thoroughly investigated PcPAL, have Phe and display only PAL activity. Changing the corresponding residue from Phe to His, however, causes PALs to gain TAL activity without complete loss of PAL activity (Röther et al., 2002; Watts et al., 2006), whereas changing the corresponding residue from His to Phe causes RsTAL to use l-Phe as its preferred substrate (Louie et al., 2006). Thus, there could be additional residues besides Phe/His-123 responsible for the substrate specificity, as indicated in PALs from Bambusa oldhamii (Hsieh et al., 2010). Sequence alignment of PAL, PAM, TAL, and HAL of various organisms shows that Lys-443 also is conserved in l-Phe-specific enzymes (PAL and PAM). However, TAL and HAL have Met at the corresponding position. Our docking results show that Lys or Met at residue 443 could form cation-π or sulfur-π interactions, respectively, with the aromatic ring of the bound substrate. To determine this plausible effect of Lys-443, the enzyme kinetics of the K443E-SbPAL1 mutant were examined. As expected, the K443E mutant became inactive for both l-Phe and l-Tyr, probably due to electrostatic repulsion between the anionic carboxylate and electron-rich π system. Thus, Lys-443 seems to be critical for binding and positioning of the substrate aromatic moiety.

A similar phenomenon was observed with the two assayed SbPAL isozymes (Table II). As expected, Sb06g022740, which has His at residue 123, exhibited PAL and TAL activity, whereas Sb04g026520, which has Phe at this position, displayed only PAL activity. In addition, the turnover rate (k_cat) values of both isozymes were comparable to those of the wild type and H123F-SbPAL1. However, both Sb06g022740 and Sb04g026520 showed significantly decreased substrate affinity. The K_m values of Sb06g022740 were 12- and 3.5-fold higher for l-Phe and l-Tyr, respectively, compared with that of wild-type SbPAL1. Similarly, the K_m value of Sb04g026520 for Phe was 25-fold higher than that of H123F-SbPAL1. Despite the complete conservation in their constituting residues critical for catalysis and substrate binding (except for Phe/His-123), the noticeable differences in K_m between SbPAL1 and these two isozymes are significant, meaning that the Phe/His-123 residue is not the only factor that influences the enzyme activity and substrate selectivity. This result also correlates with our aforementioned hypothesis of the long-distance secondary effect onto the catalytic residues, as even small changes in noncritical residues affected binding of the substrate significantly.

Another residue that may contribute to the differential activity is Phe-102, located on the inner lid-loop and six residues apart from Tyr-96. Due to the folding of the inner lid-loop, the side chains of the two residues, Tyr-96 and Phe-102, were 4.46 Å apart from each other. Thus, the side chain of Phe-102 could interact with the bound substrate. As the phenolic side chain of Tyr-96 acts as a general base during the deamination reaction of l-Tyr, the Tyr side chain, as in the F102Y-SbPAL1 mutant, may provide an alternative catalytic base in a deamination reaction for Tyr, leading to higher TAL efficiency. Supporting this hypothesis is the observation that the F102Y-SbPAL1 mutant showed an improved k_cat for l-Tyr. Conversely, the unaffected k_cat for l-Phe reflects that the Friedel-Crafts-type deamination reaction for l-Phe is not impacted by another catalytic base (Tyr-102) from the inner lid-loop. However, F102Y-SbPAL1 displayed increased K_m values for both l-Tyr and l-Phe. Thus, the F102Y-SbPAL1 mutant has an increased turnover rate for l-Tyr without alteration for that of l-Phe, but the efficiency (k_cat/K_m) for Phe was diminished.

Tyr-96 acts as a base in the E₂-type deamination of l-Tyr, but it does not have any role during the Friedel-Crafts-type deamination of l-Phe. However, the Y96F-SbPAL1 mutant showed significant loss of activity for both of the substrates (Table II; Fig. 4B) and no measurable affinity to either trans-cinnamate or p-coumarate. The area surrounding the Tyr-96 side chain is hydrophobic due to the presence of Phe-102, Leu-120, and Leu-243. Thus, it is plausible that Tyr-96 orients the carboxyl group of substrates in addition to acting as a general base for l-Tyr. To support this notion, Tyr-96 also is highly conserved among both ammonia-lyases and aminomutases (Fig. 6).

CONCLUSION

In the United States, sorghum biomass (stalks and leaves) serves as an important forage crop for livestock. In addition, sorghum is being developed as a bioenergy crop for cellulosic biofuels. Second-generation biofuels are produced from monomeric sugars derived from cellulose and hemicellulosic polysaccharides in plant biomass. The presence of lignin makes plant cell walls resistant to breakdown either in livestock digestive systems or in the biomass conversion process that occurs in biorefineries. A key challenge in the development of next-generation bioenergy sorghums is the balance between agroindustrial needs and plant fitness (Casler et al., 2002). While transgenic down-regulation of genes encoding enzymes in monolignol and lignin biosynthesis has been successful in improving biomass conversion (Xu et al., 2011; Jung et al., 2013), there is a risk of increased susceptibility to biotic and abiotic stresses. This is illustrated by the RNA interference-mediated down-regulation of BdPAL1 and BdPAL2 activity in brachypodium. While this led to increased efficiency of enzymatic saccharification of the cell wall polysaccharides, the transgenic plants displayed delayed development and reduced root growth and became more susceptible to two fungal pathogens. In contrast, tolerance to an insect pest, drought, or UV radiation was not altered. A more detailed understanding of the structure and catalysis of enzymes involved in monolignol and lignin biosynthesis can provide additional tools to tailor cell wall composition (Walker et al., 2013, 2016; Green et al., 2014; Jun et al., 2017; Moural et al., 2017; Sattler et al., 2017) by providing targets for mutations that can be identified in natural or mutagenized populations through forward or reverse genetics approaches (Xin et al., 2008; Jiao et al., 2016) or that can be introduced using genome-editing tools such as TALEN (Cermak et al., 2011) or CRISPR/Cas9 technology (Jinek et al., 2012; Jiang et al., 2013).

The first enzyme of the general phenylpropanoid pathway, PAL, is an attractive target for this approach, given its pivotal role in generating precursors for both lignin and various defense-related compounds. Through crystal structure, molecular docking, mutagenesis, kinetic analyses, and phylogenetic analyses, we have identified that SbPAL1 and other genes encoding putative PAL isozymes in the genome of sorghum can be classified into three groups. The first group, which includes SbPAL1 and Sb06g022740, has both PAL and TAL activity. The second group of five PAL isozymes are dedicated PALs. Sb01g014020, which does not fit in either group, is expressed at low levels in roots and may not play a significant role in phenylpropanoid metabolism. Changing key features of these enzymes altered their preference for substrate and product. Thus, this study reveals targets for genome-editing approaches aimed at tailoring lignin levels in plants to improve conversion biomass into biofuels or the forage utilization of livestock without negatively affecting plant growth and responses to biotic and abiotic stresses.

While similarities in genome organization and individual sequences among grasses have often enabled results from one grass species to be translated to a related grass species, in the case of PAL, apparent differences in enzyme characteristics between maize, brachypodium, and sorghum suggest the existence of species-specific differences that enable optimal performance for the environments these species have adapted to and that necessitate some caution in extrapolating data from one grass species to another.

MATERIALS AND METHODS

Chemicals and General

Analytical-grade chemicals were obtained from Sigma-Aldrich, Thermo Fisher, and Alfa-Aesar. Screening solutions for crystallization were obtained from Hampton Research. The compound AIP was generously provided to us by Dr. John Ralph and Dr. Hoon Kim at the University of Wisconsin, Madison.

Recombinant Enzyme Expression and Purification

The PAL1 cDNA corresponding to the sorghum (Sorghum bicolor) gene Sb04g026510 was cloned into pET30a vector for overexpression. For the expression of recombinant SbPAL1, 200 mL of Luria-Bertani medium containing 100 µg mL⁻¹ kanamycin and 34 µg mL⁻¹ chloramphenicol was inoculated with a freezer stock of Escherichia coli Rosetta cells (EMD Millipore) containing the pET30a-SbPAL1 construct and grown overnight at 37°C while shaking. This culture was used to inoculate 3 L of Luria-Bertani medium, which was grown to an OD₆₀₀ of 0.4 at 37°C with shaking. The cells were then brought to 18°C with continuous shaking, and isopropyl β-thiogalactopyranoside was added to a final concentration of 0.2 mm. The culture was grown at 18°C while shaking for an additional 24 h. Cells were collected by centrifugation at 5,000 rpm for 20 min at 4°C. The cell pellet was resuspended in 40 mL of lysis buffer (50 mm sodium phosphate, pH 8, 300 mm NaCl, and 15 mm imidazole) and was sonicated five times with 15-s pulses (model 450 sonifier; Branson Ultrasonics). The lysate was cleared by centrifugation at 16,000 rpm for 25 min. Cleared supernatant was applied to 15 mL of nickel-nitrilotriacetate agarose (Qiagen), equilibrated with lysis buffer, and placed into a gravity-flow column. The column was washed with 20 column volumes of washing buffer (50 mm sodium phosphate, pH 8, 300 mm NaCl, and 25 mm imidazole), and protein was eluted with elution buffer (50 mm sodium phosphate, pH 8, 300 mm NaCl, and 250 mm imidazole). Column fractions containing SbPAL1 were desalted and concentrated into buffer A (20 mm Tris, pH 8, and 5% [v/v] glycerol) using an Amicon 8050 ultrafiltration cell with a 10-kD cutoff membrane (Millipore). Concentrated protein was applied to a Mono-Q column (GE Healthcare) that was preequilibrated with buffer A using a flow rate of 2 mL min⁻¹. SbPAL1 was eluted from the column with a linear NaCl gradient. The fractions containing SbPAL1 were pooled and buffer exchanged into 20 mm Tris, pH 8. Reducing agents such as βME were not used due to the formation of an adduct with MIO. All expression and purification of isozymes and mutants were performed in an identical manner to SbPAL1, with SDS-PAGE to check the presence and purity of enzymes after each purification step. Protein concentrations were determined by using the BCA Assay Kit (Thermo Fisher).

Site-directed mutations were created in the SbPAL1 coding region by PCR-based amplification using Phusion High-Fidelity DNA polymerase (New England Biolabs). The amplification was performed using complementary plus- and minus-strand oligonucleotides containing the target mutations and was followed by DpnI (New England Biolabs) digestion to degrade the template prior to transformation of competent E. coli Rosetta cells (EMD Millipore). Both mutations were confirmed by DNA sequencing (GENEWIZ).

Crystallization and Structure Determination

Crystals of SbPAL1 were grown through the hanging-drop, vapor-diffusion method. The purified SbPAL1 was concentrated to 10 mg mL⁻¹ in 20 mm Tris, pH 8, and mixed in a 1:1 volume with a reservoir solution that contained 200 mm ammonium formate, pH 6.6, and 20% (w/v) polyethylene glycol 3,350. Crystals of SbPAL1 appeared within 10 d. Adequate cryoprotection was achieved by passing crystals through a small drop of storage buffer/mother liquor mixture, which was brought to a final concentration of 50% (w/v) polyethylene glycol 3,350. SbCAD4 was crystallized in the space group I4₁22 and had unit cell dimensions of a = b = 126.304 Å, c = 337.477 Å, α = β = γ = 90°. Data were collected to 2.5 Å at the Berkeley Advanced Light Source (Beamline 5.0.2), with an exposure time of 2 s and a detector distance of 380 mm at 100 K. Diffraction data were scaled using the program HKL2000 (Otwinowski and Minor, 1997).

Phasing and Refinement

Initial phasing of diffraction data was performed by molecular replacement with the Phaser in the PHENIX package (Adams et al., 2010) using the coordinate of PcPAL (PDB identifier 1W27; Ritter and Schulz, 2004) as a search model. The following conformation and position were refined further using PHENIX and manually adjusted with the software COOT (Emsley et al., 2010). The final R_work was 16.6%, R_free was 20.4%, and root mean square deviations from ideal geometry of the model were 0.004 Å for bonds and 0.599 for angles. The statistics for the diffraction data are listed in Table I. The coordinates and diffraction data have been deposited to the PDB with identifier 6AT7.

Enzyme Kinetics

The kinetic parameters of wild-type and mutant SbPAL1 and its isozymes were determined by measuring reaction product (cinnamate, p-coumarate, or urocanate) formation. Enzyme kinetic assays were performed in a 1-mL reaction volume of 100 mm Tris buffer, pH 8, containing 133.3 nm of the purified enzyme. Substrate (l-Phe, l-Tyr, l-His, or l-DOPA) concentration was varied from 0 to 10 mm. Reaction mixture without substrate was incubated at 30°C, and the reaction was initiated by adding the substrate to a proper final concentration. Product formation was observed over 5 min at 275, 310, 280, or 350 nm for cinnamate, p-coumarate, urocanate, or caffeate, respectively. Kinetic parameters were calculated with Origin 92 (OriginLab). For the inhibition kinetics, 5 and 20 µm AIP were added to the reaction mixture before incubation. Inhibition rate also was tested in the presence of 2, 5, and 10 mm βME.

Molecular Docking of Substrate

In spite of our numerous attempts to obtain complex crystals for SbPAL1 with both approaches of cocrystallization and crystal soaking, we were not able to obtain any suitable complex crystal. In addition, the affinity of SbPAL1 for the typical inhibitor compound, AIP, was too low to be used for the complex crystallization. Therefore, to examine the mode of substrate binding and the effect of His/Phe/Tyr at position 123, in silico substrate-docking experiments were performed with AutoDock Vina (Trott and Olson, 2010) for l-Phe and l-Tyr. Prior to a docking calculation, ammonia adduct bound to the methylidene carbon of MIO was removed from the SbPAL1 crystal structure and in silico mutation of H123F- and H123Y-SbPAL1 was performed with the COOT software. Once the substrate-binding site was confirmed by a blind search where the search box contained a whole wild-type SbPAL1 tetramer, both binding affinity and modes of the substrates were determined by an exhaustive search within the active-site pocket of the wild type, H123F-SbPAL1, and H123Y-SbPAL1. For each search, an ensemble of 10 docking solutions, where the predicted energy of binding ranged from −6.4 to −5.7 kcal mol⁻¹, were generated. A solution of binding pose with the lowest predicted energy was presented.

ITC

ITC reactions were carried out in a MicroCal iTC200 isothermal titration calorimeter (Malvern). All titrations were performed at 25°C and stirred at 1,000 rpm. The calorimetric cell contained 0.06 mm SbPAL1 in 20 mm MOPS, pH 7.7, into which an initial 0.8 µL of ligand solution was injected, followed by 19 subsequent 2-µL injections of 0.6 mm reaction products (cinnamate, p-coumarate, or caffeate) that were also in 20 mm MOPS, pH 7.7. Both the enzyme and reaction products were titrated against the same buffer (20 mm MOPS, pH 7.7), prior to the ITC experiment, to account for the heat of dilution. The same calorimetric method was employed for inhibitor (AIP) titration for the wild-type, H123F-SbPAL1, and Y96F-SbPAL1. The Origin 7 MicroCal Data Analysis software analysis package (GE Healthcare) was used for ITC curve fitting. A one-set-of-sites model was employed. Curve-fitting equations can be found in the MicroCal iTC200 User Manual appendix.

Phylogenetic Analysis

Phylogenetic analyses of amino acid sequences of eight SbPAL isoenzymes were conducted with the MEGA7 software package (Kumar et al., 2016). The evolutionary history was inferred using the minimum evolution method (Rzhetsky and Nei, 1994). The evolutionary distances were computed using the Poisson correction method (Zuckerkandl and Pauling, 1965) and are in units of the number of amino acid substitutions per site. The minimum evolution tree was searched using the close-neighbor-interchange algorithm (Nei and Kumar, 2000) at a search level of 1. The neighbor-joining algorithm (Saitou and Nei, 1987) was used to generate the initial tree. All positions containing gaps and missing data were eliminated. There were a total of 689 positions in the final data set.

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. Relative activity of SbPAL1 in the absence of reducing agent and in the presence of βME, tris-2-carboxyethylphosphine, and reduced glutathione.