The yeast ALA synthase C‐terminus positively controls enzyme structure and function

Abstract

5‐Aminolevulinic acid synthase (ALAS) is a pyridoxal 5′‐phosphate (PLP)‐dependent enzyme that catalyzes the first and rate‐limiting step of heme biosynthesis in α‐proteobacteria and several non‐plant eukaryotes. All ALAS homologs contain a highly conserved catalytic core, but eukaryotes also have a unique C‐terminal extension that plays a role in enzyme regulation. Several mutations in this region are implicated in multiple blood disorders in humans. In Saccharomyces cerevisiae ALAS (Hem1), the C‐terminal extension wraps around the homodimer core to contact conserved ALAS motifs proximal to the opposite active site. To determine the importance of these Hem1 C‐terminal interactions, we determined the crystal structure of S. cerevisiae Hem1 lacking the terminal 14 amino acids (Hem1 ΔCT). With truncation of the C‐terminal extension, we show structurally and biochemically that multiple catalytic motifs become flexible, including an antiparallel β‐sheet important to Fold‐Type I PLP‐dependent enzymes. The changes in protein conformation result in an altered cofactor microenvironment, decreased enzyme activity and catalytic efficiency, and ablation of subunit cooperativity. These findings suggest that the eukaryotic ALAS C‐terminus has a homolog‐specific role in mediating heme biosynthesis, indicating a mechanism for autoregulation that can be exploited to allosterically modulate heme biosynthesis in different organisms.

Short abstract

PDB Code(s): 8EIM;

1. INTRODUCTION

Pyridoxal 5′‐phosphate, the active form of vitamin B₆, is an indispensable cofactor central to amino acid metabolism in all organisms (Snell, 1953). PLP‐dependent enzymes catalyze a multitude of reactions, such as transamination, decarboxylation, racemization, and α/β eliminations (Eliot & Kirsch, 2004; Toney, ^{2005, 2011}). As a result, the functional diversity of these enzymes makes up approximately 4% of all classified enzymatic activities (Percudani & Peracchi, 2003). Initiation of enzymatic reactions involving PLP is universal, starting with formation of a Schiff base between a conserved active site lysine and PLP (Fasella, 1967). The versatility of PLP originates after this step, where the type of reaction that ultimately occurs is dependent on the protein moiety (Metzler & Greeff, 1954). Enzyme structure therefore dictates reaction and substrate specificity (Toney, 2005, 2011). Because of their ubiquity and chemical diversity, investigation of PLP‐dependent enzymes represents a significant opportunity to understand varied mechanisms of metabolic control that can be exploited in downstream applications.

5‐aminolevulinic acid synthase (ALAS, EC 2.3.1.37) is a PLP‐dependent enzyme that belongs to the Fold‐Type I (FT I) CoA‐dependent acyltransferase family, and it catalyzes the condensation of glycine and succinyl‐CoA to form aminolevulinic acid (ALA) (Christen & Mehta, 2001; Shemin & Kumin, ¹⁹⁵²; Shemin & Rittenberg, ¹⁹⁴⁶). ALAS is the first and rate‐limiting enzyme controlling heme biosynthesis in α‐proteobacteria and non‐plant eukaryotes (Gibson et al., 1958; Kikuchi et al., ¹⁹⁵⁸); however, some reports indicate this may not hold true for certain yeast and fungal species (Hoffman et al., 2003). Due to the importance of heme for all organisms to mediate essential pathways including oxygen transport, electron transport, and transcriptional regulation, it is important to understand both the conserved and divergent mechanisms that influence the initiation of heme production across species (Ponka, 1997).

In eukaryotes, ALAS localizes to the mitochondrial matrix and is subject to regulation at multiple steps. These include transcriptional regulation (Gotoh et al., 2011; Surinya et al., ¹⁹⁹⁸), mitochondrial import (Kikuchi & Hayashi, 1981; Munakata et al., ²⁰⁰⁴), cofactor binding (Kardon et al., 2015), protein–protein interactions (Bishop et al., 2012; Burch et al., ²⁰¹⁸; Furuyama & Sassa, ²⁰⁰⁰; Medlock et al., ²⁰¹⁵), and degradation (Kubota et al., 2016; Tian et al., ²⁰¹¹). Although lower eukaryotes typically contain one homologous ALAS gene, vertebrates contain two ALAS isoforms with different expression patterns (Bishop et al., 1990). ALAS1 is a housekeeping form that is ubiquitously expressed (Cotter et al., 1995). ALAS2 is the erythroid‐specific isoform responsible for producing ~90% of the total heme pool in humans and plays an essential role in red blood cell development (Ponka, 1997). While all eukaryotic ALAS enzymes share high sequence similarity in their prototypical PLP FT I catalytic cores, there is greater variation among their N‐ and C‐terminal extensions, which are absent from bacterial ALAS orthologs (Whatley et al., 2008). The variations in feedback mechanisms, tissue expression, and structure may provide distinct means of regulation that is specific to each eukaryotic niche.

An emerging mechanism for controlling eukaryote ALAS activity is through changes in tertiary structure. All ALAS enzymes exist as homodimers with two active sites buried at the dimer interface. It was previously reported that the ClpX unfoldase remodels Saccharomyces cerevisiae ALAS (Hem1) to open the active sites for accelerated cofactor binding (Kardon et al., 2015, 2020). Whether this function holds true for vertebrates and under what conditions is unclear (Rondelli et al., 2021). Furthermore, several reports indicate that the conformational flexibility of the ALAS active site loop is important for regulating ALA product release in all organisms (Astner et al., 2005; Lendrihas et al., ²⁰¹⁰). Finally, a recent study highlighted the necessity for conformational rearrangement of the human ALAS2 C‐terminus to allow for succinyl‐CoA substrate binding (Bailey et al., 2020). In humans, the C‐terminal extension serves an autoinhibitory role, and inherited genetic mutations underlie various blood disorders due to aberrant heme biosynthesis (Balwani & Desnick, 1993; Bottomley et al., ¹⁹⁹⁵; Kadirvel et al., ²⁰¹²). One disorder, X‐linked protoporphyria (XLP), results from enzyme hyperactivity upon C‐terminal truncation (Balwani et al., 2013; Ducamp et al., ²⁰¹³; Whatley et al., ²⁰⁰⁸). Previous ALAS structures and kinetic studies suggest the eukaryotic C‐terminal extension employs differential regulatory mechanisms (Bailey et al., 2020; Brown et al., ²⁰¹⁸; Hunter & Ferreira, ²⁰²²; Na et al., ²⁰¹⁸).

Although multiple clinical and biochemical reports established an autoinhibitory function for the vertebrate ALAS2 C‐terminal extension, recent evidence revealed this function is not conserved across eukaryotes, as C‐terminal deletions in Hem1 lead to decreased enzyme activity in vitro and in vivo rather than hyperactivity underlying XLP (Brown et al., 2018). Motivated by this work that highlighted a unique role for the ALAS C‐terminal extension among eukaryotes, we determined the molecular mechanism of the Hem1 C‐terminus in modulating enzyme activity. Using X‐ray crystallography, enzymatic assays, and mass spectrometry, we report that the yeast C‐terminal extension serves to organize the Hem1 active site to allow for proper cofactor and substrate binding, and C‐terminal deletion causes enzyme instability and dysregulated flexibility. This work uncovers the structural basis of ALAS regulation meditated by the non‐conserved C‐terminus in S. cerevisiae, thus opening the door to understanding how this conserved enzyme is regulated across species. This information can be leveraged to uncover novel regulatory mechanisms to target related PLP‐dependent enzymes for industrial or pharmacological purposes.

2. MATERIALS AND METHODS

2.1. Protein expression and purification

S. cerevisiae ALAS (Hem1) constructs were expressed and purified as described (Brown et al., 2018; Kardon et al., ²⁰¹⁵). Each was cloned into a modified pET28b vector containing an N‐terminal hexahistidine tag followed by the SUMO fusion protein (His₆‐SUMO). The plasmids were transformed into Escherichia coli BL21‐Codonplus (DE3)‐RIL cells (Agilent Technologies) and cultivated in LB medium containing 25 μg/mL chloramphenicol and 50 μg/mL kanamycin at 37°C to an OD₆₀₀ of ~0.6. Protein expression was induced with 0.5 mM IPTG. The cultures were shaken for 4 h at 22°C, followed by cell harvesting.

Cells were resuspended in 25 mM HEPES‐KOH (pH 8.0), 100 mM KCl, 400 mM NaCl, 2 mM MgCl₂, 20 mM imidazole, 10% glycerol, 1 mM DTT, and 20 μM PLP (lysis buffer) with EDTA‐free protease inhibitor (Thermo Fisher Scientific) and lysed by high‐pressure homogenization. Cell debris was removed by centrifugation and the supernatant was incubated with Ni²⁺‐NTA agarose resin (Qiagen) at 4°C for 1 h. The resin was washed with lysis buffer, and the tagged protein was eluted with 250 mM imidazole. The His₆‐SUMO tag was cleaved overnight at 4°C with SUMO protease and dialyzed into 25 mM HEPES‐KOH (pH 7.6), 100 mM KCl, 10% glycerol, 1 mM DTT, and 20 μM PLP. This was followed by a second Ni²⁺‐NTA purification step to remove the affinity tag. The protein was then isolated via gel‐filtration chromatography (HiLoad 16/600 Superdex 200 pg) and simultaneously buffer exchanged into 25 mM HEPES‐KOH (pH 7.6), 100 mM KCl, 10% glycerol, and 0.5 mM tris(2‐carboxyethyl)phosphine (TCEP; storage buffer). The purified protein was concentrated using a 30 kDa molecular weight cutoff centrifugal filter (Amicon) to 40 mg/mL, flash‐frozen in liquid nitrogen, and stored at −80°C.

2.2. Crystallization

Purified Hem1 ΔCT was diluted to 5 mg/mL in storage buffer. Crystals were obtained by the hanging‐drop, vapor‐diffusion method at 25°C by adding 1 μL protein to 1 μL of reservoir solution (0.09 M magnesium acetate [pH 7.9] and 20% PEG 3350). Fully grown crystals were formed in 3 days and harvested after 1 month, in which the harvested crystals were briefly cryoprotected in reservoir solution containing 30% glycerol and then flash‐frozen in liquid nitrogen.

2.3. Data collection, structure determination, and substrate modeling

A native diffraction data set was collected at the Life Sciences Collaborative Access Team beamline 21‐ID‐F at the Advanced Photon Source, Argonne National Laboratory at a wavelength of 0.97872 Å at 100 K using a Rayonix MX‐300 detector. The data were indexed, integrated, and scaled using the XDS program package (Kabsch, 2010). POINTLESS was used to confirm the space group of P2₁2₁2₁ (Evans, 2011). Molecular replacement was performed in PHASER (McCoy et al., 2007) using one subunit of wildtype Hem1 (PDB 5TXR) as the search model, yielding a single unique solution containing two molecules (one biological homodimer) in the asymmetric unit. Automatic model building and initial refinement were done with ModelCraft (Bond & Cowtan, 2022). Rounds of iterative model building and refinement were performed using Coot and REFMAC5 in the CCP4 program suite (Emsley et al., 2010; Murshudov et al., ²⁰¹¹; Winn et al., ²⁰¹¹). TLS groups used during latter stages of refinement were assigned using TLSMD server (Painter & Merritt, 2006). Model coordinates and structure factors have been deposited in the Protein Data Bank with accession code 8EIM. All structure figures were generated using PyMOL (Schrödinger, LLC, 2010). Structural biology applications used in this project were compiled and configured by SBGrid (Morin et al., 2013).

Substrate modeling was performed by aligning both subunits of WT Hem1 (PDB 5TXR) and Hem1 ΔCT (PDB 8EIM) with chain A of substrate‐bound Rhodobacter capsulatus ALAS containing pyridoxyl‐glycine (PDB 2BWP) or succinyl‐CoA (PDB 2BWO) in PyMOL (Schrödinger, LLC, 2010). The resulting pyridoxyl‐glycine and succinyl‐CoA substrate orientations were then modeled into the Hem1 structures to identify potential interacting residues.

2.4. ALA product release measurement

Enzyme activity was measured by a discontinuous colorimetric activity assay using a protocol adapted from Shoolingin‐Jordan et al. and Bailey et al. with modifications (Bailey et al., 2020; Shoolingin‐Jordan et al., ¹⁹⁹⁷). Reaction mixtures contained 50 mM potassium phosphate (pH 7.4), 1 mM DTT, 20 mM MgCl₂, various concentrations of glycine and succinyl‐CoA, and 5 μg/mL purified enzyme (175 μL total). Reactions were incubated at 30°C for 7 min and subsequently terminated with 100 μL 10% trichloroacetic acid. Protein was removed by centrifugation at 13,000g for 5 min, and 240 μL of the resulting supernatants were added to 240 μL of 1 M sodium acetate (pH 4.4). Next, 20 μL of acetylacetone was added to each of the mixtures, and the samples were boiled for 10 min at 100°C to derivatize ALA. Samples were cooled for 15 min, and three 100 μL aliquots were reacted with equal amounts of modified Ehrlich's reagent in a clear 96‐well plate. The absorbance was monitored at 553 nm for 20 minutes with a microplate reader (Thermo Fisher Scientific). The endpoint readout was obtained by using the peak absorbance values and converting them to molar quantities of ALA (extinction coefficient of 60,400 M⁻¹ cm⁻¹). Kinetic parameters (V _max, K _M, and k _cat) were determined by titrating glycine (0.5–50 mM glycine with 200 μM succinyl‐CoA) or succinyl‐CoA (10–200 μM succinyl‐CoA with 50 mM glycine) into reaction mixtures and fitting the resulting kinetic curves with Michaelis–Menten or allosteric sigmoidal models in Prism GraphPad 9. Each datapoint represents three (glycine) or six (succinyl‐CoA) biological replicates with three technical replicates each.

2.5. Hydrogen–deuterium exchange mass spectrometry

Wildtype and mutant proteins were prepared at 20 pmol/μL. Labeling occurred in phosphate‐buffered saline (PBS) pH 7.4 in D₂O at 20°C for 0, 10, 100, 1000, and 10,000 s. The reaction was quenched in 500 mM TCEP, 4 M guanidinium in 200 mM phosphate buffer, pH 2.4, 0°C. Samples were injected into a nano‐ACQUITY UPLC system with HDX technology. Online digestion was performed at 20°C and 7600 psi at a flow of 150 μL/min of 0.1% formic acid in H₂O (3500 psi at 40 μL/min), using an immobilized‐pepsin column. Peptides were identified in un‐deuterated samples using Waters ProteinLynx Global Server 3.0.3 software with non‐specific protease, minimum fragment ion matches per peptide of three, and oxidation of methionine as a variable modification. Deuterium uptake was calculated and compared to the non‐deuterated sample using DynamX 3.0 software. Criteria were set to minimum intensity 5000, minimum products 4, and the sequence length 5–25. Results were averaged across replicate analyses, at a given time point and the standard deviation determined.

2.6. Spectral scans

UV‐Visible absorption spectra of Hem1 proteins at a concentration of 40 μM were obtained with a Molecular Devices SpectraMax M2 microplate reader with a 1 cm quartz cuvette at room temperature. Fluorescence emission spectra of purified Hem1 diluted in PBS to a concentration of 40 μM were obtained with a microplate reader using a black 384‐well plate. Samples were excited at a wavelength of 330 nm, and the emission wavelength ranged from 350 to 600 nm. Blank spectra were collected from samples containing all components except for enzyme immediately prior to enzyme readings. Experimental scans were normalized by subtracting buffer‐only spectra.

3. RESULTS

3.1. The Hem1 C‐terminus influences PLP cofactor microenvironment

To elucidate the mechanism of the Hem1 C‐terminal extension in the autoregulation of enzyme function, we produced mature, WT Hem1 (residues 58–548) and Hem1 lacking the terminal 14 amino acids (Hem1 ΔCT, residues 58–534). Previous results determined from a coupled spectroscopic assay showed that Hem1 ΔCT displayed an approximate 25% decrease in activity compared to WT (Brown et al., 2018). This evidence supported a divergent role for the Hem1 C‐terminus in altering enzyme function compared to higher eukaryotes, since it is well‐established that vertebrate ALAS2 C‐terminal deletions lead to enzyme hyperactivity (Bailey et al., 2020; Bishop et al., ²⁰¹³; Ducamp et al., ²⁰¹³; Fratz et al., ²⁰¹⁵; Kadirvel et al., ²⁰¹²).

Our initial comparisons of protein oligomerization and stability show that C‐terminal truncation does not affect dimerization but increases the rate of protein unfolding (Figure S1). To delve into how the C‐terminal extension mediates Hem1 enzyme activity, we examined several molecular factors that affect the active sites, including perturbations of cofactor and/or substrate binding. Since correct binding and orientation of PLP in the active sites are necessary for efficient catalysis, we first determined the mode of PLP cofactor binding to Hem1 using UV–Visible absorbance and fluorescence spectroscopies. ALAS holoenzymes contain an internal aldamine, consisting of a covalent Schiff base between the ε‐amino group of the active site lysine and the formyl group of PLP (Fasella, 1967; Jenkins & Sizer, ¹⁹⁵⁷), which can tautomerize into enolimine, substituted aldamine, and ketoenamine species; this equilibrium depends on the active site environment of each enzyme (Figure 1a) (Davis & Metzler, 1972; Heinert & Martell, ¹⁹⁶²; Ikushiro et al., ¹⁹⁹⁸; Soniya & Chandra, ²⁰¹⁸). Importantly, the presence of the protonated ketoenamine species facilitates the formation of the external aldamine produced after nucleophilic attack by the amine group of the glycine substrate (Weeks et al., 2009). An absorbance peak at 330 nm corresponds to either the substituted aldamine or enolimine species, while a peak at 420 nm corresponds to the active ketoenamine state (Metzler, 1957). The WT Hem1 spectra showed two absorbance peaks, indicating an equilibrium between PLP binding modes (Figure 1b), which has been noted for other ALAS orthologs (Fratz et al., 2015; Zhang et al., ²⁰⁰⁵). Conversely, Hem1 ΔCT displayed a decrease in the 330 nm absorbance peak with an increase in 420 nm absorbance, indicating a higher proportion of the ketoenamine species.

FIGURE 1.

The Hem1 PLP microenvironment is altered by the C‐terminal extension. (a) Tautomeric states of the Hem1 internal aldamine formed between Lys337 and PLP. The ketoenamine and enolimine species include intact imine bonds that are essential for the reaction mechanism. Conversely, the substituted aldamine is catalytically inactive because of a loss of the imine bond through attack by a nearby nucleophile. (b) UV–Visible absorbance spectra for WT (solid line) and Hem1 ΔCT (dashed line). (c) Fluorescence emission spectra of WT (solid line) versus ΔCT (dashed line) at an excitation wavelength of 330 nm.

To parse apart the two species associated with absorbance at 330 nm, a fluorescence emission spectrum was collected after excitation at 330 nm. The presence of the enolimine yields a fluorescence emission band at ~520 nm, and the substituted aldamine exhibits emission at ~375 nm (Figure 1a). Analysis of both WT and ΔCT showed two resulting peaks corresponding to the substituted aldamine and enolimine species (Figure 1c). The substituted aldamine species predominated in WT Hem1, while there were equivalent tautomeric amounts in Hem1 ΔCT. Thus, deletion of the Hem1 C‐terminus alters the PLP microenvironment, resulting in a change in cofactor tautomerization.

3.2. Enzyme kinetics are controlled by the C‐terminal extension

Next, in vitro activity of both enzymes was assessed by measuring the rate of ALA product release (Figure 2a), which is the rate‐limiting step of the ALAS reaction cycle. These results confirm an approximate 30% decrease in Hem1 enzymatic function upon C‐terminal truncation. To determine the molecular basis for the diminished activity, Hem1 enzyme kinetics were determined under steady‐state conditions with varying glycine or succinyl‐CoA concentrations (Figure 2b,c, Table 1). WT Hem1 bound glycine with an apparent affinity of 0.73 ± 0.09 mM compared to the significantly weaker affinity of Hem1 ΔCT (6.37 ± 0.50 mM) combined with a lower V _max, suggesting that the Hem1 C‐terminus substantially stabilizes glycine binding. Catalytic efficiency (k _cat/K _M) of Hem1 ΔCT also decreased ~12‐fold with respect to glycine turnover. Although both enzymes displayed comparable succinyl‐CoA K _M values (26 ± 2 μM vs. 33 ± 5 μM for WT and ΔCT, respectively), the V _max for succinyl‐CoA activity was also decreased for ΔCT compared to WT. Additionally, while glycine binding exhibited classical Michaelis–Menten kinetics for both enzymes, succinyl‐CoA binding kinetics differed between the variants. WT Hem1 displayed a sigmoidal curve indicative of positively cooperative binding (n _Hill = 1.4 ± 0.2), a behavior seen for other ALAS orthologs (Bishop et al., 2012, 2013). In contrast, Hem1 ΔCT did not display positive cooperativity but rather fit best with a Michaelis–Menten model. This ablation of positive cooperativity points toward the Hem1 C‐terminus maintaining inter‐subunit communication between succinyl‐CoA binding sites.

FIGURE 2.

Hem1 ΔCT demonstrates altered enzyme kinetics. (a) Overall rates of ALA product release for WT Hem1 versus Hem1 ΔCT using saturating substrate concentrations. Rates are in units of nmol of ALA per mg of enzyme over time. (b, c) The rate of ALA product release was measured as a function of increasing glycine concentration and 200 μM succinyl‐CoA (b) or increasing succinyl‐CoA concentration and 50 mM glycine (c) for WT Hem1 (closed circles) and Hem1 ΔCT (open circles). Data are an average of 3 or 6 biological replicates and error bars represent SEM. (b) Glycine kinetic data were fit with a Michaelis–Menten model for WT and ΔCT (solid and dashed lines, respectively). (c) Succinyl‐CoA kinetic data were modeled by an allosteric sigmoidal model for WT Hem1 with a Hill coefficient n _H = 1.4 but a Michaelis–Menten model for Hem1 ΔCT. The inset shows a magnified view of curve fits at lower succinyl‐CoA concentrations.

TABLE 1.

Hem1 kinetic parameters.

	Glycine				Succinyl‐CoA
	V max (nmol mg−1 min−1)	K M (mM glycine)	k cat (min−1)	k cat/K M (min−1 M−1)	V max (nmol mg−1 min−1)	K M (μM succinyl‐CoA)	k cat (min−1)	k cat/K M (min−1 M−1)
WT	95 ± 2	0.73 ± 0.09	2.24 ± 0.05	3.09 × 103	121 ± 5	26 ± 2	2.9 ± 0.1	1.11 × 105
ΔCT	71 ± 2	6.37 ± 0.50	1.64 ± 0.04	2.57 × 102	87 ± 5	33 ± 5	2.0 ± 0.1	6.00 × 104

3.3. Truncation of the C‐terminal extension has a global effect on Hem1 structure

Our in vitro data suggest the Hem1 C‐terminus mediates the proper arrangement and formation of the enzyme active site. The Hem1 C‐terminal domain is a long, flexible extension that predominantly wraps around the cis subunit and packs in an extended hydrophobic groove that buries ~1400 Ų of solvent‐accessible surface area (SASA; Brown et al., ²⁰¹⁸). The terminal 14 amino acids of this extension are solvent‐exposed and important for making direct contacts with the opposite active site. To understand how this regulatory region communicates with residues at the active site to stabilize cofactor and substrate binding, the crystal structure of Hem1 ΔCT bound to PLP was determined at 2.1 Å resolution (Table 2). Hem1 ΔCT maintains the dimeric architecture reported for WT Hem1 and other ALAS orthologs (Figure 3a, regions from alternate subunit connoted with *). The global structure is similar to WT Hem1 (1.2 Å RMSD), but key regions exhibit significant deviations. First, there is asymmetry in the two active sites with respect to PLP binding (Figure 3b). In one subunit, the PLP molecule is covalently bound to the catalytic lysine (K337), forming the intact internal aldamine. However, the PLP molecule at the adjacent site is bound non‐covalently and could only be modeled at 80% occupancy, indicating greater dynamics at this position (Figure S2).

TABLE 2.

Crystallographic data collection and refinement statistics.

	Hem1 ΔCT (8EIM)
Data collection
Space group	P212121
Unit‐cell parameters
a, b, c (Å)	79.27, 99.53, 120.23
α, β, γ (°)	90, 90, 90
Wavelength (Å)	0.97872
Resolution (Å)	47.91–2.1 (2.22–2.10)
Z (molecules per ASU)	2
Total/unique reflections	364,598/56,191
Completeness (%)	99.7 (98.5)
Multiplicity	6.5 (6.3)
Average I/σ (I)	11.80 (1.27)
CC1/2 (%)	99.8 (59.4)
Rsym on I	10.9 (157.2)
Wilson B‐factor	39.5
Model refinement
Resolution (Å)	47.91–2.1
No. of reflections (total)	56,174
No. of reflections (test)	2809
R work	0.1878
R free	0.2244
Root‐mean‐square deviations
Bond lengths (Å)	0.007
Bond angles (°)	1.316
Ramachandran Plot
Favored (%)	97
Allowed (%)	3
Outliers (%)	0
Average B‐factors (Å2)	58.63
Protein	32.4
Ligand	56.77
Water	49.6

FIGURE 3.

The Hem1 ΔCT crystal structure shows perturbations to overall Hem1 structure and conserved structural motifs. (a) Cartoon representation of the Hem1 ΔCT crystal structure (PDB 8EIM). One protomer is colored pink (Subunit A) and the second protomer is colored tan and labeled with an asterisk (Subunit B*). PLP cofactor shown as yellow spheres. (b) Active site residues and PLP coordination for each Hem1 ΔCT active site. Subunit A contains a non‐covalently bound PLP molecule at 80% occupancy and Subunit B contains a covalent pyridoxyl‐lysine moiety. Blue and gray mesh represents 2F_o‐2F_c electron density maps contoured to 1.0σ for the catalytic lysine and PLP and the surrounding residues, respectively. PLP omit maps are shown in Figure S2. (c) The position and dynamics of key structural and catalytic motifs are altered in the absence of the Hem1 C‐terminal extension. The Hem1 ΔCT protomers are colored light and dark gray with relevant motifs colored as indicated. The domain architecture of WT Hem1 is shown below for reference. NT: N‐terminal domain, β‐arm (cyan), GR: glycine‐rich motif (magenta), K337: pyridoxyl‐lysine (yellow), ASL: active site loop (orange), CT: C‐terminal domain (dark green).

In addition to changes in PLP binding, there were also significant structural perturbations due to the deletion of the C‐terminal extension that resulted in large regions of disorder in the electron density map or altered conformations. Interestingly, there is more disorder in chain B compared to chain A for Hem1 ΔCT (Figure 3c); however, we are unable to identify a correlation between the degree of PLP binding and structural disorder. The first three disordered regions found in both ΔCT subunits include the extreme N‐terminus (NT), a portion of helix α1 and the following 4‐strand β‐sheet (β‐arm), and a catalytic loop termed the glycine‐rich motif (GR) buried in the enzyme core. These three disordered segments partially overlap among the two Hem1 ΔCT chains, but the fourth disordered region (E496–E512) is unique to Subunit B and corresponds to a surface‐exposed loop in the extended C‐terminus. A shorter portion of the fourth region was also disordered in one subunit of WT Hem1 (G505–F509). As a result of this overall structural flexibility, there is an increase in solvent accessibility of the enzyme active sites located at the dimer interface. In WT Hem1, there is 9456 Ų of buried SASA, which is ~22% of the total surface area. Hem1 ΔCT buries only 7161 Ų, or 17%, at the dimer interface. Notably, this change in buried SASA cannot be attributed solely to deletion of the last 14 amino acids, which does not impact the overall percent of buried protein surface. One implication of this increased active site accessibility and disorder could be the measured changes in PLP and substrate binding seen in the in vitro assays.

3.4. Deletion of the Hem1 C‐terminus contributes to flexibility of conserved structural and catalytic motifs

Due to the conformational flexibility of Hem1 ΔCT, there are likely multiple substrate‐binding residues impacted by C‐terminal truncation. Modeling Hem1 with its substrates using bacterial ALAS as a template revealed substantial differences between WT and ΔCT substrate interactions (Table 3; Astner et al., ²⁰⁰⁵). The most significant difference between the two enzymes is the absence of succinyl‐CoA interactions with conserved motifs (Figure 3c) in Hem1 ΔCT. Modeling WT Hem1 with succinyl‐CoA shows protein‐ligand interactions with the β‐arm (R91), GR (T150), and the active site loop (ASL; P451, T452). Comparatively, Hem1 ΔCT has fewer contacts overall and only demonstrates a potential interaction between the ASL and succinyl‐CoA via residue T452.

TABLE 3.

Hem1 active site interactions between cofactor and substrates. Substrates modeled using chain A of 2BWP and 2BWO for pyridoxyl‐glycine and succinyl‐CoA, respectively.7 ^a

Enzyme	Active site	PLP bond	PLP	Pyridoxyl‐glycine (2BWP)	Succinyl‐CoA (2BWO)
WT	A	PLP	C182, Y183, H209, H284, T334, T366*, T367*	N121, C182, Y183, H209, H284, T334, G343, T366*, T367*	S204, D205, H209, K223, P451, T150*
	B	PLP	C182*, Y183*, H209*, H284*, T334*, T366*, T367	N121*, C182*, Y183*, H209*, H284*, T334*, G343*, T366, T367	R91*, S204*, D205*, E206*, H209*, K223*, T452*, T150
ΔCT	A	PLP	C182, Y183, H209, H284, T334, T367*	N121, C182, Y183, H209, H284, S256, T334, T366*, T367*	S204, D205, H209, K223, T452
	B	LLP	C182*, Y183*, H209*, H284*, T334*, T366, T367	N121*, H209*, H284*, S256*, T334*, T366, T367	S204*, D205*, K223*

Note: Bolded residues are part of conserved catalytic or structural motifs. Asterisks denote Subunit B.

^a Underlined residues indicate those that differ between active sites for the same enzyme.

Of the regions that are disordered upon C‐terminal deletion, the GR and β‐arm motifs represent highly‐conserved catalytic and structural segments. To confirm that the dynamics of these key motifs are not impacted by crystallization conditions or artifacts, we carried out hydrogen‐deuterium exchange mass spectrometry (HDX‐MS) to identify flexible protein regions in solution. The relative uptake of deuterium was monitored over time for both WT and Hem1 ΔCT (Figures 4 and S3). Comparison of these data with the Hem1 ΔCT crystal structure confirms that the disordered regions in the crystal structure are also flexible in solution, as indicated by differential deuterium uptake.

FIGURE 4.

The Hem1 C‐terminal extension affects enzyme conformational dynamics. (a) Deuterium uptake of WT compared to Hem1 ΔCT over a time course of 0.167–167 min was measured by hydrogen‐deuterium exchange mass spectrometry (HDX‐MS). The relative fractional deuterium uptake is represented as a heat map with red corresponding to higher uptake in WT Hem1 and blue indicating higher uptake in Hem1 ΔCT. Data are plotted for a set of linear peptides, and gray regions indicate areas lacking significant deuterium uptake based on indicated criteria (raw HDX‐MS data are shown in Figure S3). The domain architecture for Hem1 ΔCT is shown above the heat map, and the disordered regions identified in the Hem1 ΔCT crystal structure are shown below (Chain A residues 58–67, 105–112, and 153–155; Chain B residues 58–68, 86–112, 151–155, and 496–512). (b) Relative deuterium uptake mapped onto one WT Hem1 protomer (PDB 5TXR) over time. Colors correspond to the heat map in panel A. N: N‐terminus; C: C‐terminus. (c) Relative deuterium uptake mapped onto WT Hem1 homodimer at 16.67 min.

The β‐arm is an N‐terminal segment comprised of residues 83–118 that consists of the last portion of the helix α1 followed by three antiparallel strands of a 4‐strand β sheet. This structural motif is largely conserved among FT I PLP‐dependent enzymes within the CoA and aminotransferase II (AT II) subfamilies. Since the C‐terminal extension makes multiple contacts with the β‐arm in WT Hem1, truncation of the extension ablates these interactions and is coupled with a mostly disordered β‐arm (Subunit B) or a β‐arm region largely rotated away from the active site (Subunit A) in the Hem1 ΔCT crystal structure (Figures 5a and S4). Because of this conformational change, the motif also adopts a new secondary structure topology as it packs against the underside of the catalytic core. The WT β1–β2 (residues 97–108) motif is mostly buried and interacts with the GR and ASL. The new β1–β2 (residues 89–102) motif in Hem1 ΔCT doubles in SASA by undergoing a rearrangement, displacing R91 by 21 Å in Hem1 ΔCT. We cononfirmed the rearrangement was not due to crystal packing artifacts. It was previously observed that R91 of the β‐arm motif interacted with T452 in the WT Hem1 ASL. The same ASL threonine in the R. capsulatus crystal structure was shown to be necessary for loop closure and coordination of the succinyl‐CoA substrate (Astner et al., 2005). Intriguingly, the ASL was captured in the open conformation in both subunits, regardless of the position of the β‐arm motif or R91, thus confirming the Hem1 C‐terminus likely does not affect the dynamics of the yeast ASL. However, the ASL plays a critical role in mediating ALA product release among all ALAS orthologs, and its conformation is controlled by the C‐terminal extension in vertebrate ALAS2 enzymes, supporting a divergent role for the Hem1 C‐terminus in controlling enzyme function compared to higher eukaryotes (Bailey et al., 2020; Fratz et al., ²⁰¹⁵; Hunter & Ferreira, ¹⁹⁹⁹). Although there is no indication the yeast C‐terminus affects the ASL conformation in the crystal structures, we cannot rule out the possibility that the Hem1 ASL may become flexible in the presence of substrate.

FIGURE 5.

C‐terminal truncation reveals a conformational change to the conserved 3β‐strand motif, altering interactions between other conserved motifs proximal to the active site. (a) Crystal structure of WT Hem1 (PDB 5TXR) with protomers colored dark gray and light gray, PLP colored yellow, and β‐arm motif colored blue. The β‐arm motif of Hem1 ΔCT (cyan) was superimposed onto WT Hem1. Insets of subunit A show interactions between conserved motifs (colored as per Figure 3c). (b) Comparison of C‐terminal extension interactions between the β‐arm (β‐arm–CT) and GR (GR–CT) for WT Hem1 versus Hem1 ΔCT. Interacting residues shown in stick representation with asterisks denoting subunit B. Black dashed lines indicate hydrogen bonding or pi–pi interactions. Dashed lines in secondary structure cartoons indicate disordered residues. (c) Superposition of WT Hem1 (gray) and Hem1 ΔCT (magenta) GR motif showing the displacement of T150*.

The GR (residues 143–156) is an established ALAS catalytic motif that is largely recalcitrant to mutations and mediates cofactor and substrate binding (Gong & Ferreira, 1995). Deletion of the invariant catalytic arginine in this region (R151) resulted in an approximate 80% decrease in in vitro Hem1 activity (Brown et al., 2018). Importantly, a previous Hem1 crystal structure demonstrated that GR disorder correlated with a loss of PLP binding (Brown et al., 2018). Our HDX‐MS data show that Hem1 ΔCT exhibits significantly greater deuterium exchange at the GR segment at all time points compared to the WT, confirming greater GR mobility in the absence of the extreme C‐terminus (Figure 4). The GR was disordered to varying extents in both Hem1 ΔCT subunits, but the ordered segments were displaced significantly compared to WT Hem1 (Figure 5b). The GR adopts a new helical topology in which T150 involved in succinyl‐CoA binding is displaced by 10 Å (Figure 5c). Since both subunits in the Hem1 ΔCT crystal structure contain PLP, our findings suggest that the flexibility of the GR is controlled primarily by the C‐terminal extension rather than the presence of cofactor.

4. DISCUSSION

4.1. The regulatory role of the C‐terminal extension in yeast

It was previously noted that ALAS interconverts between open and closed conformations, which would impact the accessibility of the ALAS active sites buried at the dimer interface (Hunter & Ferreira, 1999). Our spectroscopic results establish that initiation of the Hem1 reaction is controlled by the C‐terminal extension through mediating structural rearrangements that alter active site exposure. The shift to a higher proportion of the protonated ketoenamine in Hem1 ΔCT is likely due to the higher solvent accessibility of the enzyme active sites compared to WT Hem1. In contrast, deletions in the human ALAS2 C‐terminal extension resulted in a shift from obligate ketoenamine to a mixture of ketoenamine and substituted aldamine (Fratz et al., 2015). Since the overall ALAS reaction mechanism and active site are highly conserved, the observation that Hem1 C‐terminal deletion has the opposite effect on cofactor binding compared to human ALAS2 confirms that the C‐terminal extension exhibits an ortholog‐specific function.

Our kinetic studies emphasize the importance of the Hem1 C‐terminal extension in maintaining rather than inhibiting enzyme activity. The higher solvent accessibility of the Hem1 ΔCT active sites affects substrate binding and potential conformational changes induced by substrate binding, which leads to diminished catalytic efficiency. The decrease in catalytic efficiency in Hem1 ΔCT may stem from the inability to consistently generate the reactive pyridoxyl‐glycine external aldamine, evidenced by the nearly eight‐fold decrease in glycine binding in the presence of saturating succinyl‐CoA. Furthermore, previous studies in ALAS2 determined that succinyl‐CoA binds in a positively cooperative manner (Bishop et al., 2012), but Hem1 ΔCT does not exhibit any cooperativity for substrate binding. Although bacterial ALAS was crystallized in a symmetric conformation containing covalently bound PLP in both half‐sites (Astner et al., 2005), the asymmetry with respect to cofactor binding in eukaryotic ALAS has been captured in a previous Hem1 structure (PDB 5TXT), indicating potential differences in the homodimer active sites (Brown et al., 2018). Kinetic studies have also alluded to inter‐subunit communication in vertebrate ALAS2 in regard to substrate binding (Bishop et al., 2013; Hunter & Ferreira, ¹⁹⁹⁹; Zhang & Ferreira, ²⁰⁰²).

This loss in cooperativity paired with increased solvent accessibility at the active sites upon C‐terminal truncation suggests that the Hem1 C‐terminal extension is responsible for maintaining inter‐subunit communication. Although the exact molecular mechanism is still unknown, the overall greater solvent accessibility in Hem1 ΔCT dimer implies the extension induces or initiates a disorder‐to‐order transition of global enzyme topology (Figure 6a). This broad change propagates to the active site by directly affecting two major regions: the structural β‐arm motif that makes up a wall of the active site and the conserved catalytic glycine‐rich loop. Paired with the differences in PLP occupancy at the two active sites, the differences in disorder of the Hem1 ΔCT subunits at the β‐arm and GR point toward structural asymmetry that extends past cofactor binding and instead could be an inherent property of Hem1 that is controlled by the C‐terminal extension. This asymmetry was also alluded to in bacterial and vertebrate ALAS enzymes (Bailey et al., 2020; Na et al., ²⁰¹⁸).

FIGURE 6.

The ALAS C‐terminal extension is a structurally divergent regulatory region. (a) Model of the mechanism by which the yeast C‐terminal extension controls ALAS structure and function. The extension wraps around its originating subunit to interact with the β‐arm of the opposite subunit in WT Hem1. Upon truncation of the 14‐terminal amino acids, the interaction between the CT and β‐arm is ablated, leading to increased structural flexibility and solvent accessibility, altered cofactor binding, and decreased enzyme activity. (b) Heat map depicting percent sequence identity of the C‐terminal extension across ALAS homologs. Full amino acid sequences were first aligned using Clustal Omega (Sievers et al., 2011). The C‐terminal extension regions were then extracted from this alignment and compared using SIAS to calculate pairwise sequence identity and similarity. ALAS isoforms are grouped as non‐vertebrate Hem1, vertebrate housekeeping ALAS1, and vertebrate erythroid‐specific ALAS2. Species included: S. cerevisiae, Yarrowia lipolytica, Schizosaccharomyces pombe, Ashbya gossypii, Neurospora crassa, Agaricus bisporus, Homo sapiens, Mus musculus, Rattus norvegicus, Bos taurus, Delphinapterus leucas, Gallus gallus, Opsanus tau, and Danio rerio. (c) Structures of eukaryotic ALAS with the C‐terminal extensions highlighted green. Shown are the crystal structures of WT Hem1 (PDB 5TXR), H. sapiens ALAS2 (hALAS2, PDB 6HRH), and the AlphaFold model H. sapiens ALAS1 (hALAS1, AF‐P13196; Jumper et al., ²⁰²¹).

4.2. Sequence and functional divergence of the ALAS C‐terminal extension

Multiple sequence alignment of the eukaryotic C‐terminal extension highlights its variability among homologs (Figure 6b). Much of the investigation into human ALAS has focused on ALAS2 since it is responsible for producing ~90% of the total heme pool and is subject to disease‐relevant mutations. XLP is caused by human ALAS2 hyperactivity resulting from perturbations specifically in the ALAS2 C‐terminal extension. However, separate point mutations in this region also cause decreased enzyme activity, leading to diminished heme production and X‐linked sideroblastic anemia. The crystal structure of ALAS2 shows the C‐terminal extension wrapping around the cis subunit toward the dimer interface and occluding the active site, thus inhibiting substrate binding and product release (Figure 6c). This is in direct contrast to the Hem1 C‐terminus, which adopts a distinct conformation to regulate enzyme activity in the opposite manner. Although there is no experimental structure of the housekeeping vertebrate isoform ALAS1, the predicted model of human ALAS1 features C‐terminal extensions that pack on the outside of the cis subunit in a distal position relative to the two active sites (Figure 6c; Jumper et al., ²⁰²¹). However, the prediction confidence for this region is low (average pLDDT ~51). Compared to ALAS2, ALAS1 contains a longer N‐terminal extension that could interact with the mostly unstructured and flexible C‐terminus. The growing knowledge about the functional impacts of both the primary and tertiary structure of this divergent region could illuminate new details about enzyme function that govern diverse mechanisms for regulating eukaryotic heme biosynthesis.

4.3. The β‐arm motif represents an opportunity for allosteric regulation of PLP‐dependent enzymes

The ability of Hem1 to mediate the positioning of the β‐arm via its C‐terminal extension is an important aspect of S. cerevisiae‐specific regulation. The β‐arm motif was discussed previously as the possible locus of Hem1‐ClpX interaction (Kardon et al., 2020). ClpX is an AAA+ unfoldase that participates in mitochondrial protein quality control (Fischer et al., 2012). In S. cerevisiae, it acts as a chaperone for cofactor binding by partially unfolding Hem1 and accelerating PLP insertion (Kardon et al., 2015). A region spanning helix α1 to strand β2 was proposed as the initial ClpX binding site, in which unfolding starting at α1 restructures the β‐arm to open the active site (Kardon et al., 2020). Our truncation of the Hem1 C‐terminal extension mimics this mechanism by disrupting the same structured motif, thus allowing for bound PLP to exist in both active sites compared to the wildtype Hem1 structure (Brown et al., 2018). However, unlike ClpX regulation, the C‐terminal truncation negatively affects Hem1 activity due to an inability to refold the necessary catalytic regions.

The function of the β‐arm and its location in the conserved catalytic core begins to highlight its importance within other PLP‐dependent enzymes. Among these enzymes, only the FT I CoA‐acyltransferase and aminotransferase II subfamilies feature the β‐arm based on known structural homology (Table S1; Holm & Sander, ¹⁹⁹⁵). Understanding how the structural dynamics of this region are controlled may yield insight into the regulation of a broader group of enzymes. Our current study uncovered an opportunity to exploit both conserved (i.e., β‐arm) and divergent (i.e., C‐terminal extension) structural segments for allosteric regulation of ALAS and potentially many more PLP‐dependent enzymes mediating diverse functions. Further investigation into FT I CoA and AT II structural differences will be the necessary foundation for targeting this broader group of proteins for industrial and pharmaceutical use.

AUTHOR CONTRIBUTIONS

Jenny U. Tran: Conceptualization (equal); data curation (lead); formal analysis (lead); investigation (lead); methodology (lead); visualization (lead); writing – original draft (equal); writing – review and editing (equal). Breann L Brown: Conceptualization (equal); data curation (supporting); formal analysis (supporting); funding acquisition (lead); investigation (supporting); methodology (supporting); project administration (lead); supervision (lead); visualization (supporting); writing – original draft (equal); writing – review and editing (equal).

CONFLICT OF INTEREST STATEMENT

The authors declare no conflict of interest.

Supporting information

DATA S1. Supporting Inforamtion

Tran JU, Brown BL. The yeast ALA synthase C‐terminus positively controls enzyme structure and function. Protein Science. 2023;32(4):e4600. 10.1002/pro.4600

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are openly available in the RCSB Protein Data Bank at https://doi.org/10.2210/pdb8EIM/pdb, or are available from the corresponding author upon reasonable request.