Evolution of lysine and arginine biosynthesis revealed by substrate specificity of lysine biosynthetic enzymes in Thermus thermophilus

TL;DR

This study shows that two enzymes in Thermus thermophilus can work on both lysine and arginine pathways, suggesting a shared evolutionary origin for these amino acid synthesis routes.

Contribution

The study reveals substrate promiscuity in lysine biosynthetic enzymes and suggests a shared evolutionary origin with arginine biosynthesis.

Findings

LysZ and LysY enzymes can act on intermediates from the arginine biosynthetic pathway.

LysZ activity on arginine intermediates is about 60% of its activity on lysine intermediates.

Phylogenetic analysis suggests a shared ancestral pathway for lysine and arginine biosynthesis.

Abstract

Metabolic pathways are considered to originate from broad‐specificity ancestors that later diverged into specialized routes. Thermus thermophilus possesses an unusual amino group carrier protein (AmCP)‐mediated lysine biosynthetic pathway alongside a canonical arginine biosynthetic pathway. Although each route is considered specific to its cognate amino acid, several lysine biosynthetic enzymes have been shown to accept arginine intermediates. We herein investigated [LysW]‐aminoadipate kinase (LysZ; EC:2.7.2.17) and [LysW]‐L‐2‐aminoadipate 6‐phosphate reductase (LysY; EC:1.2.1.103), which catalyze the second and third steps, respectively, in the conversion of α‐aminoadipate (AAA) to lysine using amino group carrier protein LysW (AmCP), to define their specificity and evolutionary origin. To examine the potential promiscuity, we engineered LysX variants capable of synthesizing LysW‐Glu, an artificial LysW‐bound analogue that mimics an arginine pathway intermediate. LysZ exhibited activity for LysW‐Glu that was approximately 60% of the original activity for LysW‐AAA. The activity of LysY for LysW‐Glu phosphate was estimated to be approximately 15–20% of that observed with LysW‐AAA phosphate. The present study revealed that both enzymes can also act on an arginine biosynthetic intermediate, but with distinct degrees of efficiency. Phylogenetic reconstructions further suggested that an AmCP‐mediated biosynthetic pathway represents a primitive route for the synthesis of lysine and arginine in a primordial cell. More generally, the results obtained herein will contribute to a more detailed understanding of the evolutionary strategies employed by nature to specialize and expand metabolic pathways and adjust enzyme promiscuity.

Two enzymes from the Thermus thermophilus lysine biosynthetic pathway, LysZ and LysY, unexpectedly exhibit substrate promiscuity. They are able to recognize the LysW intermediates utilized by the ArgW‐mediated arginine route, even though the bacterium already possesses the canonical arginine biosynthetic pathway. This promiscuity suggests an evolutionary link between lysine and arginine biosynthesis, shedding new light on the evolution of essential amino acid pathways.

Figures

Designed LysX variants. Substrate binding sites in (A) LysX/ArgX of T. kodakarensis, (B) ArgX of Sulfurisphaera tokodaii, and (C) LysX of T. thermophilus. In (A), the substrates, AAA and attached C‐terminal E53 residue from LysW from T. kodakarensis are shown in the yellow stick models. In (B), bound glutamate is shown in the yellow stick models. Structural representations were generated using pymol (version 3.1, Schrödinger, LLC, New York, NY, USA). (D) Amino acid sequence alignment of LysX family proteins. Residues related to substrate specificity are highlighted by red rectangles, and the residue numbers of LysX of T. thermophilus are described above. TtLysX, DgLysX, and StLysX are LysX from T. thermophilus, Deinococcus geothermalis, and S. tokodaii, respectively; TK0278 and PH1721 are LysX/ArgX from T. kodakarensis and Pyrococcus horikoshii, respectively; SaArgX, MsArgX, and StArgX are ArgX from S. acidocaldarius, Metallosphaera sedula, and S. tokodaii, respectively. Substrate recognition motifs are indicated on the right. The multiple amino acid sequence alignment was prepared by clustalw [15]. The aligned sequences were retrieved from UniProt with the following accession numbers: TtLysX (Q72HE6), DgLysX (Q1IZ86), StLysX (Q976J9), TK0278 (Q5JFW0), PH1721 (O59396), SaArgX (Q4J8E7), MsArgX (A4YI79), and StArgX (Q970U6). (E) Enzyme activities of LysX variants toward glutamate and AAA. The names of LysX variants are shown on the left. Residues of wild‐type LysX are indicated by blue squares, while those substituted with ArgX‐type and LysX/ArgX‐type residues are shown in orange and green, respectively. The horizontal axis indicates the conversion efficiency, calculated as [product/(substrate + product)] × 100%. The bars filled with orange and blue indicate activities toward AAA and glutamate, respectively. Vertical dashed lines indicate the activities of wild‐type LysX from T. thermophilus for glutamate and AAA. Data represent the mean ± SD of three independent experiments.

Production of LysW‐AAA and LysW‐Glu. (A) Tricine SDS/PAGE analysis of purified LysW‐AAA and LysW‐Glu, produced in E. coli harboring pET11b‐lysWX and pET11b‐lysWX6, respectively. Red arrows indicate the bands corresponding to LysW‐AAA or LysW‐Glu. Lanes S, soluble fraction of the cell lysate; H, soluble fraction after the heat treatment; FT, flow‐through fraction; E1‐5, eluate fractions with buffer A containing 0–1 m NaCl. The gel image shown is representative of three independent experiments (n = 3), which yielded similar results. (B) LC–MS analysis showing the production of LysW‐Glu and LysW‐AAA, indicated by the orange and blue peaks, respectively. A dashed line indicates the peak of LysW (theoretical mass: 5812.48 Da), which was not converted into LysW‐AAA or LysW‐Glu. The spectrum shown is representative of three independent experiments (n = 3), all of which yielded similar results.

LysZ activities for LysW‐AAA and LysW‐Glu. (A) Scheme of the LysZ reaction and hydroxylamine derivatization using LysW‐AAA and LysW‐Glu. (B) Detection of the product of the LysZ reaction on tricine SDS/PAGE using LysW‐AAA and LysW‐Glu as substrates after 15‐, 30‐, and 60‐min incubations. The fast‐migrating band (black arrow) indicates the substrate, LysW‐AAA/Glu, while the slow one (red arrow) represents the product derivative, LysW‐AAA/Glu‐hydroxamate. The gel image shown is representative of three independent experiments (n = 3), which yielded similar results. Band densities were calculated by imagej and transformed into the line chart on the right. LysZ activities toward LysW‐AAA and LysW‐Glu are represented by the orange and blue lines, respectively. (C) LC–MS analysis of the LysZ reaction using LysW‐Glu as the substrate. The upper panel shows the spectrum before the reaction, where a black peak corresponding to LysW‐Glu was detected (theoretical mass: 5941.59 Da; observed: 5941.0 Da). The lower panel shows the spectrum after the reaction, where an orange peak corresponding to LysW‐Glu‐hydroxamate was observed (theoretical mass: 5956.61 Da; observed: 5956.1 Da). The spectrum shown is representative of three independent experiments (n = 3), all of which yielded similar results. (D) LC–MS analysis of the LysZ reaction for LysW‐AAA as the substrate. The upper panel shows the spectrum before the reaction, where a black peak corresponding to LysW‐AAA was detected (theoretical mass: 5955.62 Da; observed: 5955.1 Da). The lower panel shows the spectrum after the reaction, where a blue peak corresponding to LysW‐AAA hydroxamate was observed (theoretical mass: 5970.64 Da; observed: 5970.0 Da). The spectrum shown is representative of three independent experiments (n = 3), all of which yielded similar results.

Coupling assay of LysZ and LysY. (A) Scheme of the LysZ and LysY reactions. (B) Activity measurements using LysW‐AAA and LysW‐Glu as the initial substrate. Activities were shown as specific activity per 1 mg LysY. Bars represent mean values with overlaid data points from three independent experiments. (C) LC–MS analysis of the LysZ and LysY coupling reaction using LysW‐Glu (left) and LysW‐AAA (right) as substrates. To facilitate comparison, the upper panels (before reaction) are reproduced from Fig. 3C,D. The upper panel shows the spectrum before the reaction, where black peaks corresponding to unmodified and LysW‐Glu (theoretical mass: 5941.59 Da; observed: 5941.0 Da) and LysW‐AAA (theoretical mass: 5955.62 Da; observed: 5955.1 Da) were detected. The lower panel shows the spectrum after the reaction, where an orange peak corresponding to LysW‐Glu semialdehyde (theoretical mass: 5925.60 Da; observed: 5925.1 Da) and a blue peak corresponding to LysW‐AAA semialdehyde (theoretical mass: 5939.47 Da; observed: 5939.0 Da) were detected. These results indicate that both substrates were converted to their corresponding semialdehyde forms by the coupled LysZ and LysY reaction. The spectrum shown is representative of three independent experiments (n = 3), all of which yielded similar results.

Phylogenetic analysis of lysine biosynthetic enzymes. (A) LysZ. (B) LysY. The evolutionary history was inferred using the Maximum Likelihood method implemented in mega11 [17], based on 24 amino acid sequences and the Le_Gascuel_2008 model. Details of alignment and tree construction are described in the Materials and methods section. The tree with the highest log likelihood is shown. Bootstrap support values (percentage of replicate trees in which associated taxa clustered together) were calculated from 1000 replicates and are shown next to the branches. Evolutionary rate variation among sites was modeled with a discrete Gamma distribution (five categories, parameters are 2.6897 and 2.5327 for LysZ and LysY, respectively) and the percentage of invariable sites (+I is 1.63% and 2.29% for LysZ and LysY, respectively). The tree is drawn to scale, with branch lengths measured as the number of substitutions per site. Key ancestral nodes are indicated by red dots, with those labeled with 1 and 2 representing ancestors 1, the last common ancestors of LysZ /LysY and ArgW‐ArgZ/ArgY types, and ancestors 2, the last common ancestors of all AmCP‐type enzymes, respectively. The last common ancestors of LysZ/ArgZ/ArgB and LysY/ArgY/ArgC are indicated by blue dots. Functional subgroups are color‐coded as indicated in the figure. The lysine and arginine biosynthetic pathway contexts for each species are summarized in Table S1. (C) Evolutionary model of the lysine and arginine biosynthetic pathways. Each symbol represents a complete biosynthetic pathway for lysine or arginine, classified based on its substrate specificity. Colors and shapes indicate the type of substrate utilized by the pathway: both LysW‐AAA and LysW‐Glu (light green circle), only LysW‐AAA (blue circle), only LysW‐Glu (orange circle), or N‐acetylglutamate (orange square). The branching structure reflects a maximum likelihood phylogenetic tree constructed from the LysZ and LysY sequences, which exhibit nearly identical topologies, along with analyses of additional lysine biosynthetic enzymes. This model supports the hypothesis that the lysine and arginine biosynthetic pathways both evolved from an ancestral bifunctional AmCP‐mediated pathway (light green) with broad substrate recognition.

Tables

Table 1. List of LysX variants engineered in the present study.

Name	Mutations	Type	Motif
LysX WT	LysX WT	LysX type	F‐I‐T‐NS
LysX1	LysX F182I	ArgX type	I‐I‐T‐NS
LysX2	LysX I192Y	ArgX type	F‐Y‐T‐NS
LysX3	LysX T203V	ArgX type	F‐I‐V‐NS
LysX4	LysX NS258259GF	ArgX type	F‐I‐T‐GF
LysX5	LysX F182I/I192Y	ArgX type	I‐Y‐T‐NS
LysX6	LysX F182I/I192Y/T203V	ArgX type	I‐Y‐V‐NS
LysX7	LysX F182I/I192Y/T203V/NS258259GF F182I/I192Y/T203V/NS258259GF	ArgX type	I‐Y‐V‐GF
LysX8	LysX F182I/T203V	ArgX type	I‐I‐V‐NS
LysX9	LysX F182I/NS258259GF	ArgX type	I‐I‐T‐GF
LysX10	LysX I192Y/T203V	ArgX type	F‐Y‐V‐NS
LysX11	LysX I192Y/NS258259GF	ArgX type	F‐Y‐T‐GF
LysX12	LysX T203V/NS258259GF	ArgX type	F‐I‐V‐GF
LysX13	LysX F182I/I192Y/NS258259GF	ArgX type	I‐Y‐T‐GF
LysX14	LysX F182I/T203V/NS258259GF	ArgX type	I‐I‐V‐GF
LysX15	LysX I192Y/T203V/NS258259GF	ArgX type	F‐Y‐V‐GF
LysX16	LysX F182Y	LysX/ArgX type	Y‐I‐T‐NS
LysX17	LysX NS258259NA	LysX/ArgX type	F‐I‐T‐NA
LysX18	LysX F182Y/NS258259NA	LysX/ArgX type	Y‐I‐T‐NA

Table 2. Primers used for the site‐directed mutagenesis of lysX.

Primer name	Sequence
LysXF182I_Fw	CGGGACATCCGGGTCATCGTGGTGGGGGAAAGG
LysXF182I_Rv	CGGGACATCCGGGTCATCGTGGTGGGGGAAAGG
LysXI192Y_Fw	AGGGCCATCGCCGCCTACTACCGCAGAAGCGCC
LysXI192Y_Rv	GGCGCTTCTGCGGTAGTAGGCGGCGATGGCCCT
LysXT203V_Fw	CACTGGATCACCAACGTCGCCCGGGGCGGGCAG
LysXT203V_Rv	CTGCCCGCCCCGGGCGACGTTGGTGATCCAGTG
LysXNS258259GF_Fw	ACCATGGAGTTTAAAGGCTTCGTGCACACCACCGGG
LysXNS258259GF_Rv	CCCGGTGGTGTGCACGAAGCCTTTAAACTCCATGGT
biLysXF182Y_Fw	CGGGACATCCGGGTCTACGTGGTGGGGGAAAGG
biLysXF182Y_Rv	CCTTTCCCCCACCACGTAGACCCGGATGTCCCG
biLysXS259A_Fw	ATGGAGTTTAAAGGCGTCGTGCACACCACCGGG
biLysXS259A_Rv	CCCGGTGGTGTGCACGACGCCTTTAAACTCCAT

Full text

Introduction

Biological adaptation and the transition from simple to complex life forms are intimately connected with the evolution of amino acid metabolic pathways. Among 20 proteinogenic amino acids, several biosynthetic pathways are catalyzed by enzymes that have evolved from common ancestors [1, 2, 3]. In bacteria, the arginine biosynthetic pathway typically begins with the acetylation of the α‐amino group of glutamate. Subsequent steps in arginine biosynthesis include phosphorylation, reduction, transamination, and the removal of the protecting acetyl group from the α‐amino group [4]. The enzymes ArgB, ArgC, and ArgE in the arginine pathway share structural similarities with the enzymes LysC, Asd, and DapE in the diaminopimelate (DAP) pathway of lysine biosynthesis, and they catalyze analogous chemical reactions [5]. Furthermore, ArgD in arginine biosynthesis is a bifunctional enzyme that functions as DapC in lysine biosynthesis in Escherichia coli [6]. These findings suggest an evolutionary relationship between lysine and arginine biosynthesis. An important example is the lysine biosynthetic pathway in Thermus thermophilus. In the latter half of lysine biosynthesis of T. thermophilus, α‐aminoadipate (AAA) is loaded via the α‐amino group on the amino group carrier protein (AmCP) called LysW, and subsequent reactions occurring on AmCP proceed similarly to those of the arginine biosynthetic pathway [7].

The AmCP‐mediated lysine biosynthetic pathway has also been shown to function in ornithine and/or arginine biosynthesis in some microorganisms. The hyperthermophilic archaeon, Thermococcus kodakarensis, possesses a set of enzymes in AmCP‐mediated pathways that are capable of utilizing both AAA and Glu as substrates to synthesize lysine and ornithine, respectively [8]. Similarly, in Sulfolobus, most enzymes in AmCP‐mediated lysine biosynthesis function as a bifunctional enzymes for arginine biosynthesis, except for the enzymes LysX and ArgX, which catalyze the reaction in AmCP‐mediated lysine and arginine biosynthesis, loading AAA and glutamate, respectively, to AmCP [9]. LysX and ArgX are paralogous enzymes that presumably evolved from a common ancestral enzyme with dual functions similar to LysX/ArgX in T. kodakarensis. These findings strongly suggest that the lysine and arginine biosynthetic pathways share a common ancestral pathway. The AmCP‐mediated pathway may represent one of the earliest evolutionary developments in these pathways, which supports the ‘patchwork’ hypothesis that early life forms possessed minimal genomes and relied on multifunctional enzymes to conduct diverse biochemical reactions [10, 11]. It is important to note that in the haloarchaeon, Natrinema gari, lysine is biosynthesized via the DAP pathway, while arginine is biosynthesized through an ArgW (a homologous protein of LysW in arginine biosynthesis)‐mediated pathway [12]. Therefore, studies on AmCP‐mediated lysine and arginine biosynthetic pathways provide an excellent model for investigating how evolutionary pressures have guided the specialization and retention of these pathways in various forms, thereby meeting distinct biochemical needs and facilitating adaptation to diverse environments through the diversification of metabolic pathways.

T. thermophilus possesses both lysine and arginine biosynthesis pathways: AmCP (LysW)‐mediated lysine biosynthesis and typical arginine biosynthesis with N‐acetyl modification. However, the enzymes aminotransferase (LysJ) and hydrolase (LysK), which are responsible for the last two steps in the in LysW‐mediated lysine biosynthetic pathway, exhibit the ability to catalyze similar reactions in typical arginine biosynthesis that use one methylene shorter substrates with N‐acetyl modifications [13, 14]. This finding suggests the substrate promiscuity of enzymes for the current lysine biosynthetic pathway in T. thermophilus. In contrast, LysX and ArgX in Sulfolobus load only AAA and glutamate on AmCP, respectively, serving as a key determinant specifying whether the pathway synthesizes lysine or ornithine [9]. The substrate specificities of the remaining lysine biosynthetic enzymes, LysZ and LysY, of T. thermophilus have yet to be investigated. Therefore, further studies on the substrate specificities of these two enzymes will provide insights into enzyme promiscuity and the evolution of the lysine/arginine biosynthetic pathway in T. thermophilus.

In the present study, we designed variants of LysX from T. thermophilus and screened for variants that acquired the ability to load glutamate on LysW in order to produce LysW‐Glu, a potential substrate in arginine biosynthesis. We then characterized the substrate specificities of LysZ and LysY from T. thermophilus using LysW‐Glu and original LysW‐AAA. We found that LysZ and LysY catalyzed the expected reactions in LysW‐mediated arginine biosynthesis. These results demonstrate that most of the current LysW‐mediated lysine biosynthetic enzymes have the potential to catalyze arginine biosynthetic reactions. We also discuss the enzyme promiscuity and evolution of lysine/arginine biosynthetic enzymes.

Results

Design of LysX variants to exhibit activity that produces LysW‐Glu

To investigate the potential of LysZ and LysY to function in arginine biosynthesis, it was necessary to prepare LysW‐Glu, a possible intermediate in AmCP‐mediated arginine biosynthesis. Since LysX from T. thermophilus has been shown to exhibit strict specificity to recognize AAA as a substrate, we needed to change the substrate specificity of LysX toward that of ArgX or a LysX/ArgX‐type bifunctional enzyme that efficiently recognizes glutamate as a substrate. Based on our previous structural analysis and sequence comparison of LysX from T. thermophilus, ArgX from Sulfolobus acidocaldarius, which is specialized for arginine biosynthesis, and bifunctional LysX/ArgX from T. kodakarensis, which act in both lysine and arginine (ornithine) biosynthesis, we proposed a signature motif comprising five amino acid residues that affect the substrate specificity of LysX‐related proteins [8] (Fig. 1A–D). We engineered 18 variants of LysX by replacing the residues in this signature motif with corresponding residues from monofunctional ArgX or bifunctional LysX/ArgX (Table 1). To examine how substitutions in the 5‐amino acid signature motif affect substrate specificity, we measured the activity of each variant to conjugate LysW with either AAA or glutamate by monitoring the products, LysW‐AAA and LysW‐Glu, using liquid chromatography‐mass spectrometry (LC–MS). Although we previously detected only the trace activity of LysX to conjugate glutamate with LysW [7], an overnight reaction led to the identification of LysW‐Glu with a conversion rate that was at least 10‐fold lower than that for LysW‐AAA (Fig. 1E). All single‐site variants at 182, 192, 203, and 258/259 retained the ability to conjugate the native substrate AAA with LysW; however, the LysX1 variant carrying the single‐site replacement F182I also loaded glutamate on LysW. Among the 18 variants constructed, the LysX15 variant with I192Y, T203V, and N258G/S259F, which was expected to have a smaller and more hydrophobic binding pocket, exhibited the most enhanced activity for glutamate with reduced activity for the larger substrate, AAA. The effects of the amino acid replacement in the 5‐amino acid signature motif were complex, markedly affecting the substrate specificity of LysX to a different extent.

Designed LysX variants. Substrate binding sites in (A) LysX/ArgX of T. kodakarensis, (B) ArgX of Sulfurisphaera tokodaii, and (C) LysX of T. thermophilus. In (A), the substrates, AAA and attached C‐terminal E53 residue from LysW from T. kodakarensis are shown in the yellow stick models. In (B), bound glutamate is shown in the yellow stick models. Structural representations were generated using pymol (version 3.1, Schrödinger, LLC, New York, NY, USA). (D) Amino acid sequence alignment of LysX family proteins. Residues related to substrate specificity are highlighted by red rectangles, and the residue numbers of LysX of T. thermophilus are described above. TtLysX, DgLysX, and StLysX are LysX from T. thermophilus, Deinococcus geothermalis, and S. tokodaii, respectively; TK0278 and PH1721 are LysX/ArgX from T. kodakarensis and Pyrococcus horikoshii, respectively; SaArgX, MsArgX, and StArgX are ArgX from S. acidocaldarius, Metallosphaera sedula, and S. tokodaii, respectively. Substrate recognition motifs are indicated on the right. The multiple amino acid sequence alignment was prepared by clustalw [15]. The aligned sequences were retrieved from UniProt with the following accession numbers: TtLysX (Q72HE6), DgLysX (Q1IZ86), StLysX (Q976J9), TK0278 (Q5JFW0), PH1721 (O59396), SaArgX (Q4J8E7), MsArgX (A4YI79), and StArgX (Q970U6). (E) Enzyme activities of LysX variants toward glutamate and AAA. The names of LysX variants are shown on the left. Residues of wild‐type LysX are indicated by blue squares, while those substituted with ArgX‐type and LysX/ArgX‐type residues are shown in orange and green, respectively. The horizontal axis indicates the conversion efficiency, calculated as [product/(substrate + product)] × 100%. The bars filled with orange and blue indicate activities toward AAA and glutamate, respectively. Vertical dashed lines indicate the activities of wild‐type LysX from T. thermophilus for glutamate and AAA. Data represent the mean ± SD of three independent experiments.

Preparation of LysW‐Glu in vivo

Since no variants completely converted LysW to LysW‐Glu in the in vitro overnight reaction described above, we attempted to produce LysW‐Glu in E. coli cells that co‐produced LysW and each of the LysX variants, LysX1, LysX6, and LysX15, with high activity toward glutamate. After recombinant E. coli cells were cultured in 2 × YT medium supplemented with 5 mm glutamate, LysW derivatives were purified by anion‐exchange chromatography and subjected to an LC–MS analysis (Fig. 2A). The results obtained indicated that the LysX1, LysX6, and LysX15 variants successfully converted the majority of LysW to LysW‐Glu (Fig. 2B). In parallel, LysW‐AAA was produced using E. coli cells co‐producing LysW and wild‐type LysX in a similar culture supplemented with 5 mm AAA, giving a conversion rate of nearly 100% (Fig. 2B).

Production of LysW‐AAA and LysW‐Glu. (A) Tricine SDS/PAGE analysis of purified LysW‐AAA and LysW‐Glu, produced in E. coli harboring pET11b‐lysWX and pET11b‐lysWX6, respectively. Red arrows indicate the bands corresponding to LysW‐AAA or LysW‐Glu. Lanes S, soluble fraction of the cell lysate; H, soluble fraction after the heat treatment; FT, flow‐through fraction; E1‐5, eluate fractions with buffer A containing 0–1 m NaCl. The gel image shown is representative of three independent experiments (n = 3), which yielded similar results. (B) LC–MS analysis showing the production of LysW‐Glu and LysW‐AAA, indicated by the orange and blue peaks, respectively. A dashed line indicates the peak of LysW (theoretical mass: 5812.48 Da), which was not converted into LysW‐AAA or LysW‐Glu. The spectrum shown is representative of three independent experiments (n = 3), all of which yielded similar results.

Bifunctionality of LysZ: insights into potential arginine biosynthesis

LysZ phosphorylates the carboxy group of the AAA moiety in LysW‐AAA using ATP to form LysW‐AAA phosphate (LysW‐AAAP) (Fig. 3A). Based on structural similarities between LysW‐Glu and LysW‐AAA with a small difference of only a single carbon unit, LysZ was expected to exhibit promiscuous activity in order to recognize both LysW‐Glu and LysW‐AAA, similar to LysJ and LysK. By monitoring a band shift by the phosphorylation of LysW‐Glu/AAA followed by hydroxamate derivatization with hydroxylamine on tricine SDS/PAGE, we assessed the activity of LysZ for LysW‐Glu (Fig. 3B). The results obtained revealed that wild‐type LysZ catalyzed the phosphorylation of LysW‐Glu as well as LysW‐AAA; however, activity for LysW‐Glu was slightly lower (approximately 60%) than that for the native substrate LysW‐AAA (Fig. 3B). The formation of LysW‐AAAP and LysW‐GluP was further verified by the detection of their hydroxamate derivatives using LC–MS (Fig. 3C,D).

LysZ activities for LysW‐AAA and LysW‐Glu. (A) Scheme of the LysZ reaction and hydroxylamine derivatization using LysW‐AAA and LysW‐Glu. (B) Detection of the product of the LysZ reaction on tricine SDS/PAGE using LysW‐AAA and LysW‐Glu as substrates after 15‐, 30‐, and 60‐min incubations. The fast‐migrating band (black arrow) indicates the substrate, LysW‐AAA/Glu, while the slow one (red arrow) represents the product derivative, LysW‐AAA/Glu‐hydroxamate. The gel image shown is representative of three independent experiments (n = 3), which yielded similar results. Band densities were calculated by imagej and transformed into the line chart on the right. LysZ activities toward LysW‐AAA and LysW‐Glu are represented by the orange and blue lines, respectively. (C) LC–MS analysis of the LysZ reaction using LysW‐Glu as the substrate. The upper panel shows the spectrum before the reaction, where a black peak corresponding to LysW‐Glu was detected (theoretical mass: 5941.59 Da; observed: 5941.0 Da). The lower panel shows the spectrum after the reaction, where an orange peak corresponding to LysW‐Glu‐hydroxamate was observed (theoretical mass: 5956.61 Da; observed: 5956.1 Da). The spectrum shown is representative of three independent experiments (n = 3), all of which yielded similar results. (D) LC–MS analysis of the LysZ reaction for LysW‐AAA as the substrate. The upper panel shows the spectrum before the reaction, where a black peak corresponding to LysW‐AAA was detected (theoretical mass: 5955.62 Da; observed: 5955.1 Da). The lower panel shows the spectrum after the reaction, where a blue peak corresponding to LysW‐AAA hydroxamate was observed (theoretical mass: 5970.64 Da; observed: 5970.0 Da). The spectrum shown is representative of three independent experiments (n = 3), all of which yielded similar results.

Catalytic potential of LysY in arginine synthesis

LysY catalyzes the NADPH‐dependent reduction of LysW‐AAAP to produce LysW‐AAA semialdehyde (LysW‐AASA), following the LysZ reaction in lysine biosynthesis (Fig. 4A). Due to the difficulties associated with preparing the unstable reactive phosphate, LysW‐AAAP, LysY activity was measured using a LysZ‐LysY coupling assay at an elevated temperature of 70 °C by monitoring a decrease in the absorbance of NADPH at 340 nm. The apparent oxidation of NADPH was observed when Lys‐Glu or Lys‐AAA was added as the initial substrate (Fig. 4B). The oxidation of NADPH was markedly lower with LysW‐Glu as the initial substrate than with LysW‐AAA. However, it is important to note that this was the coupling reaction with LysZ. The marked decrease observed in NADPH oxidation activity was the sum of the reduction in the activities of LysZ and LysY toward LysW‐Glu derivatives. An LC–MS analysis confirmed the formation of LysW‐AASA and LysW‐glutamate semialdehyde (GluSA), the expected products of the LysZ and LysY coupling reactions; however, the amount of LysW‐GluSA produced was low (Fig. 4C).

Coupling assay of LysZ and LysY. (A) Scheme of the LysZ and LysY reactions. (B) Activity measurements using LysW‐AAA and LysW‐Glu as the initial substrate. Activities were shown as specific activity per 1 mg LysY. Bars represent mean values with overlaid data points from three independent experiments. (C) LC–MS analysis of the LysZ and LysY coupling reaction using LysW‐Glu (left) and LysW‐AAA (right) as substrates. To facilitate comparison, the upper panels (before reaction) are reproduced from Fig. 3C,D. The upper panel shows the spectrum before the reaction, where black peaks corresponding to unmodified and LysW‐Glu (theoretical mass: 5941.59 Da; observed: 5941.0 Da) and LysW‐AAA (theoretical mass: 5955.62 Da; observed: 5955.1 Da) were detected. The lower panel shows the spectrum after the reaction, where an orange peak corresponding to LysW‐Glu semialdehyde (theoretical mass: 5925.60 Da; observed: 5925.1 Da) and a blue peak corresponding to LysW‐AAA semialdehyde (theoretical mass: 5939.47 Da; observed: 5939.0 Da) were detected. These results indicate that both substrates were converted to their corresponding semialdehyde forms by the coupled LysZ and LysY reaction. The spectrum shown is representative of three independent experiments (n = 3), all of which yielded similar results.

Bioinformatics investigation of LysZ and LysY promiscuities

The promiscuous activities of LysZ and LysY toward LysW‐Glu derivatives further support the hypothesis that LysW‐mediated lysine biosynthesis evolved from a bifunctional lysine/arginine biosynthetic system [8, 9, 16]. To examine the evolutionary relationships of these homologous enzymes, we constructed two phylogenetic trees for LysZ family enzymes with aspartate kinase as the outgroup (Fig. 5A) and LysY family enzymes using aspartate semialdehyde dehydrogenase as the outgroup (Fig. 5B).

Phylogenetic analysis of lysine biosynthetic enzymes. (A) LysZ. (B) LysY. The evolutionary history was inferred using the Maximum Likelihood method implemented in mega11 [17], based on 24 amino acid sequences and the Le_Gascuel_2008 model. Details of alignment and tree construction are described in the Materials and methods section. The tree with the highest log likelihood is shown. Bootstrap support values (percentage of replicate trees in which associated taxa clustered together) were calculated from 1000 replicates and are shown next to the branches. Evolutionary rate variation among sites was modeled with a discrete Gamma distribution (five categories, parameters are 2.6897 and 2.5327 for LysZ and LysY, respectively) and the percentage of invariable sites (+I is 1.63% and 2.29% for LysZ and LysY, respectively). The tree is drawn to scale, with branch lengths measured as the number of substitutions per site. Key ancestral nodes are indicated by red dots, with those labeled with 1 and 2 representing ancestors 1, the last common ancestors of LysZ /LysY and ArgW‐ArgZ/ArgY types, and ancestors 2, the last common ancestors of all AmCP‐type enzymes, respectively. The last common ancestors of LysZ/ArgZ/ArgB and LysY/ArgY/ArgC are indicated by blue dots. Functional subgroups are color‐coded as indicated in the figure. The lysine and arginine biosynthetic pathway contexts for each species are summarized in Table S1. (C) Evolutionary model of the lysine and arginine biosynthetic pathways. Each symbol represents a complete biosynthetic pathway for lysine or arginine, classified based on its substrate specificity. Colors and shapes indicate the type of substrate utilized by the pathway: both LysW‐AAA and LysW‐Glu (light green circle), only LysW‐AAA (blue circle), only LysW‐Glu (orange circle), or N‐acetylglutamate (orange square). The branching structure reflects a maximum likelihood phylogenetic tree constructed from the LysZ and LysY sequences, which exhibit nearly identical topologies, along with analyses of additional lysine biosynthetic enzymes. This model supports the hypothesis that the lysine and arginine biosynthetic pathways both evolved from an ancestral bifunctional AmCP‐mediated pathway (light green) with broad substrate recognition.

The phylogenetic trees of LysZ and LysY were similar to each other, exhibiting similar overall topologies, with differences observed only in specific clades (Fig. 5A,B). Some of the closest relatives to LysZ and LysY from T. thermophilus were orthologs in lysine biosynthesis from Deinococcus radiodurans [18]. In the tree for LysZ family enzymes, the Deinococcus–Thermus clade shares a common ancestor with a LysZ‐like enzyme, which may be called ArgZ and plays a role in arginine biosynthesis using the AmCP‐mediated system found in haloarchaea [12], as strongly supported in the LysZ phylogeny (bootstrap value of 93%). In contrast, the corresponding node in the LysY phylogeny had bootstrap support of 46%. Despite attempting various methods to construct the phylogenetic tree and testing different evolutionary models, the topology at this point remained stable, suggesting some evolutionary fluctuations in LysY that differed from LysZ. Nevertheless, these results are consistent with the promiscuous enzyme activities of LysZ and LysY, thereby supporting the hypothesis that LysW‐mediated lysine biosynthesis in D. radiodurans and T. thermophilus originated and evolved from a hypothetical bifunctional enzyme, ancestor 1, which was presumably involved in both lysine and arginine biosynthesis. This hypothesis is further supported by ancestor 2 diverging into the Deinococcus–Thermus node, which comprised bifunctional enzymes capable of recognizing both LysW‐AAA and LysW‐Glu. Furthermore, ancestor 2 and ArgB, the canonical arginine biosynthetic enzyme, were derived from a more distant common ancestor. The same conclusion may be drawn for the LysY tree in which LysY and a LysY‐like enzyme, which may be called ArgY, for arginine biosynthesis using the AmCP‐mediated system found in haloarchaea shared an ancestor with bifunctional enzymes capable of recognizing both LysW‐AAAP and LysW‐GluP. Based on our phylogenetic analyses of LysZ and LysY, as well as previous studies on LysJ, LysK, and LysX [9, 13, 14], an evolutionary model emerged for the lysine and arginine biosynthetic pathways (Fig. 5C). The ancestral system likely included LysW‐dependent enzymes with broad specificity capable of processing both LysW‐AAA and LysW‐Glu. An early gene duplication event gave rise to two paralogous copies of this pathway. In a lineage, one copy diverged toward the exclusive recognition of N‐acetylglutamate, giving rise to the canonical arginine biosynthetic route, while the other either retained bifunctionality or later specialized to recognize only LysW‐AAA or LysW‐Glu, albeit with varying degrees of residual substrate promiscuity.

Discussion

The in‐depth analysis conducted in the present study revealed that LysX recognized AAA as a preferable substrate but exhibited measurable activity for glutamate at less than 10% of the activity for AAA. However, since this activity was too low to prepare a sufficient amount of LysW‐Glu for an analysis of the ability of LysZ and LysY to recognize LysW‐Glu, we introduced amino acid replacements into the signature motif composed of five amino acid residues of LysX in various combinations and observed shifts in substrate specificity. The LysX1 variant with F182I exhibited increased activity toward glutamate, forming LysW‐Glu with similar activity to that for LysW‐AAA, which remained retained. The LysX6 variant with the additional replacements of I192Y and T203V exhibited strong activity for glutamate and slightly reduced activity for AAA. This result demonstrates the critical and synergistic roles of residues at positions 192 and 203 in assessments of the substrate preferences of AAA and glutamate. Among them, LysX15 with the I192Y, T203V, N258G, and S259F replacements exhibited the highest activity for glutamate and markedly decreased activity for AAA. In contrast, the LysX7 variant, which possesses all the amino acid residues characteristic of the ArgX‐type enzyme, showed the pronounced loss of activity for both substrates, suggesting that simultaneous mutations in residues at positions 182 and 258/259 caused steric hindrance, leading to a decrease in activity decrease in the LysX framework.

As discussed above, the AmCP‐mediated lysine and arginine biosynthetic pathways presumably diverged from a bifunctional enzyme system for the biosynthesis of both lysine and arginine. In such a bifunctional enzyme system, the concentrations of the final products must be determined based on the concentration of AAA or glutamate in cells. Since the amino groups of most amino acids are transferred from glutamate, the basal concentration of AAA must be lower than that of glutamate in cells. Therefore, in a single lysine/arginine bifunctional biosynthesis system, glutamate was an inevitable competitive inhibitor of AAA for lysine biosynthesis. Therefore, a lysine‐specific biosynthetic system was presumably required to respond to environmental changes and induce the immediate biosynthesis of lysine. In T. thermophilus, after the duplication of genes for bifunctional enzymes, one still used the AmCP system for the amino group modification of AAA, while the other developed to use the acetyl group for the amino group modification of glutamate. Since the specificity of LysX to AAA is a crucial factor for lysine biosynthesis, evolutionary pressure likely favored enhanced specificity for AAA and reduced activity toward glutamate, resulting in the current form of LysX in T. thermophilus.

In the present study, we characterized the substrate specificities of LysZ and LysY and found that these enzymes exhibited promiscuous activities for both LysW‐AAA and LysW‐Glu derivatives, the latter of which were potential substrates in arginine biosynthesis. This result indicates that the lysine biosynthetic pathway in T. thermophilus must have originated from a bifunctional AmCP‐mediated pathway involved in lysine and arginine/ornithine biosynthesis. It is important to note that LysZ exhibited activity for LysW‐Glu that was approximately 60% of that for the natural substrate, LysW‐AAA, while the combined activity of LysZ and LysY was estimated to show activity for LysW‐GluP that was approximately 10% of that of the native substrate, LysW‐AAAP; however, these enzymes evolved under the same environment. The promiscuous activity of LysZ surpassed the usual range for promiscuous activity, typically representing less than 1–10% of the native reaction rate [19]. Obata et al. [20] proposed that the level of promiscuous activity may vary and an enzyme may exhibit strong promiscuous activity by chance if that activity is not disadvantageous. LysZ may have been subjected to negligible or no selective pressure to be specialized for lysine biosynthesis and lost its activity for LysW‐Glu, because evolved LysX did not produce LysW‐Glu in sufficient quantities to affect lysine biosynthesis.

T. thermophilus possesses a canonical type of the arginine biosynthetic pathway. As an extremophile, T. thermophilus relies on arginine to produce unique branched polyamines [21]. These compounds are crucial for stabilizing nucleic acid conformations and protecting them from depurination in extremely high‐temperature environments [22, 23]. Notably, these unique polyamines are abundant in cells grown at extremely high temperatures, such as 80 °C, but are found in smaller amounts in cells grown at lower temperatures, such as 65 °C [24]. A similar requirement for branched polyamines has been reported for the hyperthermophilic archaeon, T. kodakarensis. A T. kodakarensis mutant lacking a biosynthetic gene for branched polyamines had a decreased thermal limit for growth [25]. In addition, the RNA polymerase of T. kodakarensis exhibited lower transcription activity in the absence of branched polyamines at temperatures higher than 80 °C in vitro [26]. Therefore, branched polyamine biosynthesis is indispensable for crucial cellular events in extremely high or hyperthermophilic environments. T. thermophilus may have acquired the canonical arginine synthesis pathway to produce large amounts of arginine and to finely regulate its biosynthesis [27]. Newly acquired arginine biosynthesis enhances fitness under extremely high temperature conditions and thereby facilitates the expansion of its ecological niche. Although the promiscuous activities of LysZ and LysY as well as LysJ and LysK [13, 14] may be remnants of a bifunctional biosynthetic system, we assume that these properties may be advantageous to re‐evolve the current lysine biosynthesis pathway to the lysine/arginine bifunctional biosynthesis pathway, thereby supporting the high‐level production of polyamines when needed.

In conclusion, the present study revealed the promiscuous activities of the enzymes involved in the lysine biosynthetic pathway of T. thermophilus. The results obtained support the patchwork hypothesis, suggesting that an AmCP‐mediated biosynthetic pathway represents an early, primitive route for the slow and small‐scale synthesis of lysine and arginine (Fig. 5C) and provides a more detailed understanding of the evolutionary strategies employed by nature to specialize and expand metabolic pathways and adjust enzyme promiscuity. Further studies on the mechanisms by which enzymes evolve in biosynthetic pathways will ultimately enhance our ability to engineer enzymes and metabolic pathways for industrial and medical applications and also lead to the secrets of the origin and diversity of life being revealed.

Materials and methods

Protein expression and purification

Recombinant proteins of LysW, LysX, LysZ, and LysY were prepared as described in our previous studies [8, 28, 29].

Preparation of LysX variants

Plasmids to produce the LysX variants, which were used in activity assay experiments, were constructed by site‐directed mutagenesis using pET26b‐lysX as a template for the PCR reaction. The primers used for PCR are listed in Table 2. These LysX variants were purified in the same manner as wild‐type LysX.

In vivo

LysW‐AAA and LysW‐Glu production

LysW‐AAA recombinant proteins were prepared using E. coli BL21(DE3) harboring pET11b‐lysWX, which encodes LysW and wild‐type LysX. LysW‐Glu recombinant proteins were prepared using E. coli BL21(DE3) strains harboring pET11b‐lysWX1, pET11b‐lysWX6, or pET11b‐lysWX15, each encoding LysW and the corresponding LysX variants (LysX1, LysX6, and LysX15, respectively). They were cultured in the same manner as LysW. Gene expression was induced by adding a final concentration of 0.1 mm IPTG to the culture, followed by 5 mm AAA or glutamate. The culture was continued at 30 °C for an additional 12–14 h. Protein purification was performed using the same method as that for LysW. LysW‐AAA and LysW‐Glu were eluted with 20 mm Tris/HCl pH 8.0 (buffer A) containing 300 mm NaCl and 500 mm NaCl, respectively, from the anion‐exchange column (DE52, Cytiva, Tokyo, Japan). Purified proteins were concentrated by VIVASPIN 20 (MWCO 3000) (Sartorius, Göttingen, Germany) and used for enzyme assays.

Enzyme assays

LysX catalyzes the conjugation between LysW and AAA. LysX activity assays were conducted in 50‐μL reaction mixtures containing 1 mm LysW, 2.6 μm LysX variants, 5 mm glutamate or AAA, 1 mm MgSO_4_, 5 mm ATP, and 100 mm HEPES‐NaOH, pH 8.0, and then incubated at 60 °C overnight. Reaction products were analyzed by the high‐resolution Triple TOF 5600 system (ABSCIEX, Tokyo, Japan) equipped with the UPLC Nexera system (Shimadzu, Kyoto, Japan) with the Proteonavi S5 column, 2.0 × 50 mm (Osaka Soda, Osaka, Japan). The mobile phases consisted of (A) water with 0.1% formic acid and (B) acetonitrile with 0.1% formic acid. Separation was performed under the following gradient: 20% B for 1 min, a linear increase to 70% B over 5 min, hold at 70% B for 1 min, decrease to 20% B over 1 min, and equilibration at 20% B for 4 min. The flow rate was 0.4 mL·min^−1^. MS analyses were performed using electrospray ionization in the positive mode. The multiply charged ions observed for LysW, LysW‐AAA, and LysW‐Glu were deconvoluted using peakview (ABSCIEX, version 2.2) to obtain the monoisotopic molecular weight of the intact protein species. LysX activity was semi‐quantitatively evaluated by comparing the peak areas of LysW (substrate) and the products (LysW‐AAA or LysW‐Glu), calculated using peakview.

The LysZ activity assay method using tricine SDS/PAGE was performed as previously established [28]. During the activity assay, labile LysW‐AAAP and LysW‐GluP were converted to stable hydroxamate derivatives by adding hydroxylamine. Since the derivative of LysW‐AAAP/LysW‐GluP migrated slower than that of LysW‐AAA/LysW‐Glu on tricine SDS/PAGE, we evaluated the activities of LysZ by analyzing the densities of the bands derived from LysW‐AAA/LysW‐Glu (substrate) and LysW‐AAA‐hydroxamate/LysW‐AAA‐hydroxamate (product). After running tricine SDS/PAGE, gels were scanned, and images were analyzed with imagej software [30]. Band densities were quantified by converting images to grayscale, subtracting the background, and measuring the integrated density of each band corresponding to the substrate (LysW‐AAA or LysW‐Glu) and the product (LysW‐AAA‐hydroxamate or LysW‐Glu‐hydroxamate). The activity of LysZ was shown as the band ratio of the product to total.

LysY catalyzes the reduction of LysW‐AAAP to LysW‐AAA semialdehyde (LysW‐AASA). The reductase activity of LysY toward LysW‐AAAP and LysW‐GluP was assayed in 1 mL of a reaction mixture containing 100 mm HEPES‐NaOH, pH 8.0, 5 mm MgSO_4_, 1 mm ATP, 40 μm LysW‐AAA or LysW‐Glu, 0.2 mm NADPH, 170 nm LysZ, and 26 nm LysY at 70 °C. Reactions were initiated after 5 min of preheating at 70 °C by the addition of LysZ, and the oxidation of NADPH was monitored for 30 s using a UV2600‐UV–visible spectrophotometer (Shimadzu) to measure decreases in absorbance at 340 nm.

Phylogenetic analyses and topology test

The sequences of LysZ and LysY homologous proteins were retrieved by Blast searches at NCBI and UniProt. The enzymes catalyzing similar reactions (phosphorylation and reduction) in the DAP pathway, aspartokinase and aspartate semialdehyde dehydrogenase from E. coli, were included as outgroups. Additionally, experimentally verified LysZ, ArgW‐ArgZ, LysZ/ArgB, and N‐acetylglutamate kinase (NAGK) were manually included. Twenty‐four sequences were selected for each enzyme, including LysZ and LysY. These homologous sequences were aligned using t‐coffee (slow mode) [31] in jalview [32]. The phylogenetic tree was reconstructed using the maximum likelihood model and Le_Gascuel_2008 model [33] in mega11 [17], with the phylogenetic analysis conducted using 1000 bootstrap replicates. All positions with less than 50% site coverage were eliminated, that is, fewer than 50% alignment gaps, missing data, and ambiguous bases were allowed at any position (partial deletion option). The detailed lysine and arginine biosynthetic pathway contexts for the species used in the phylogenetic analysis are summarized in Table S1.

Conflict of interest

The authors declare no conflict of interest.

Author contributions

WS, AY, and MN designed experiments. WS performed experiments. WS, AY, SK, and MN analyzed data. The manuscript was written through the contributions of all authors. AY and MN led the study.

Supporting information

Table S1. Overview of lysine and arginine biosynthetic pathways in representative organisms used in the phylogenetic analyses.