From a dimer to a monomer: Construction of a chimeric monomeric isocitrate dehydrogenase

Abstract

Many isocitrate dehydrogenases (IDHs) are dimeric enzymes whose catalytic sites are located at the intersubunit interface, whereas monomeric IDHs form catalytic sites with single polypeptide chains. It was proposed that monomeric IDHs were evolved from dimeric ones by partial gene duplication and fusion, but the evolutionary process had not been reproduced in laboratory. To construct a chimeric monomeric IDH from homo‐dimeric one, it is necessary to reconstitute an active center by a duplicated region; to properly link the duplicated region to the rest part; and to optimize the newly formed protein surface. In this study, a chimeric monomeric IDH was successfully constructed by using homo‐dimeric Escherichia coli IDH as a start point by rational design and site‐saturation mutagenesis. The ~67 kDa chimeric enzyme behaved as a monomer in solution, with a K _m of 61 μM and a k _cat of 15 s⁻¹ for isocitrate in the presence of NADP⁺ and Mn²⁺. Our result demonstrated that dimeric IDHs have a potential to evolve monomeric ones. The evolution of the IDH family was also discussed.

1. INTRODUCTION

Isocitrate dehydrogenases (IDHs) are widely distributed in bacteria and eukaryote that play key roles in energy metabolism and biosynthesis. IDHs belong to β‐decarboxylating dehydrogenase superfamily that catalyze the oxidative decarboxylation of isocitrate to produce 2‐ketoglutarate and CO₂ with the reduction of NAD(P)⁺. ¹ , ² The product 2‐ketoglutarate is also a key precursor for biosynthesis of glutamate. ³ Recently, it was also found that some mutant IDHs catalyze the production of 2‐hydroxyglutarate, which is an onco‐metabolite that promotes a number of cancers, from 2‐ketoglutarate. ⁴ , ⁵ , ⁶ , ⁷

Based on the coenzyme specificity, IDHs can be categorized into two groups, NAD⁺‐dependent IDH (NAD‐IDH) and NADP⁺‐dependent IDH (NADP‐IDH). NAD‐IDHs provide NADH for ATP synthesis, while NADP‐IDHs generate NADPH for biosynthesis. ⁸ , ⁹ It was demonstrated that NAD‐IDH is an ancient phenotype, and NADP‐IDH arose during adaptive evolution. ¹⁰ The oligomeric structures of IDHs are also diverse. Many IDHs, such as Escherichia coli IDH (EcIDH), are homo‐dimeric enzymes whose catalytic sites are composed of the residues from both subunits. ¹¹ Homo‐tetrameric and hetero‐octameric IDHs are found in some bacteria and eukaryotic mitochondria, but their structural building blocks are also dimers. ¹² , ¹³ , ¹⁴ , ¹⁵ Monomeric IDHs, for example, Azotobacter vinelandii IDH (AvIDH), occur in a few species of bacteria and alga which perform the same catalytic function as the dimeric IDHs with single polypeptide chains. ¹⁶ , ¹⁷

The evolution of monomeric IDHs remains elusive. Phylogenetic analyses suggest that IDHs can be divided into three groups: Type I IDHs, Type II IDHs, and monomeric IDHs. ¹⁸ , ¹⁹ , ²⁰ The monomeric IDHs have low sequence identities (<10%) compared with the other two types of IDHs. ¹⁸ Nevertheless, the structural alignments showed that the key residues involved in substrate, metal ion, and coenzyme binding are highly conserved in all types of IDHs. In addition, their topological structures are surprisingly similar. ¹⁶ , ¹⁷ , ²¹ Based on the sequential and structural information, it was proposed that monomeric IDHs might be evolved from dimeric IDHs by the partial gene duplication and fusion. ¹⁶ , ¹⁷ , ²¹ However, there was no experimental evidence to support this hypothesis.

In this study, the topological structures of homo‐dimeric EcIDH and monomeric AvIDH were analyzed and compared in detail. Rational designs were proposed to construct chimeric monomeric IDHs from EcIDH. Chimeric monomeric IDHs were constructed by inserting a duplicated region into the original EcIDH sequence with two linkers. The linkers and a newly formed protein surface were optimized by site‐saturation mutagenesis. The finally constructed chimeric monomeric IDH showed high catalytic activity for isocitrate in the presence of NADP⁺ and Mn²⁺. This result provides experimental evidence to support that dimeric IDHs have a potential to evolve monomeric ones. The natural evolutionary pathways of IDHs were also discussed based on an extensive phylogenetic analysis.

2. RESULTS

2.1. The rational designs of chimeric monomeric IDHs

EcIDH contains two identical subunits of ~45 kDa. Each subunit includes three domains: a large domain with the N and C termini (domain A), a clasp‐like domain (domain B), and a small domain (domain C). Two catalytic sites are located at the interface between domains A and B′‐C′, and domains A' and B‐C, respectively (the prime sign indicates the second subunit) (Figures 1a and S1). ¹¹ AvIDH has only a single polypeptide chain of ~80 kDa with two domains: a small domain with the N and C termini (domain I) and a large domain (domain II). One catalytic site is located at the cleft between domains I and II (Figures 1b and S1). ¹⁶

FIGURE 1.

The structural alignment of Escherichia coli isocitrate dehydrogenase (EcIDH) and Azotobacter vinelandii IDH (AvIDH) and the designed structure of a chimeric monomeric isocitrate dehydrogenase (IDH). (a) The structure of EcIDH with isocitrate and Mg²⁺ (Protein Data Bank [PDB] ID: 1CW7). The two subunits were showed in deep blue and cyan, respectively. The four‐helix bundle formed by both subunits was colored in pale yellow. The U‐turn loops were showed in orange and highlighted by yellow outlines. The lower is the scheme of the structure of EcIDH. (b) The structure of AvIDH with isocitrate and Mn²⁺ (PDB ID: 1ITW). The four‐helix bundle was colored in salmon. The antiparallel linkers I and II were showed in red and highlighted by yellow outlines. The lower is the scheme of the structure of AvIDH. (c) The structures of EcIDH and AvIDH were aligned according to several conserved residues. The domain A of EcIDH and domain I of AvIDH, and domains B‐C and B′‐C′ of EcIDH and domain II of AvIDH were, respectively, superimposed after alignment. The unaligned domain A′ of EcIDH was colored in magenta. The unique gate‐like regions of AvIDH in front of catalytic center were depicted by red circles. (d) The designed structure of the chimeric monomeric IDH. The two linkers L1 and L2 to be constructed were illustrated by curve red arrows with yellow outlines

Despite of the different oligomeric states and the low sequence identity, the structures of EcIDH and AvIDH can be well aligned according to several conserved residues in the catalytic sites (Figure 1c). Domain A of EcIDH and domain I of AvIDH are nearly superimposed after alignment. Nevertheless, domain II of AvIDH contains structural features such as a four‐helix bundle and a clasp‐like region which are similar to those of EcIDH formed by domains B‐C and B′‐C′. Therefore, AvIDH may acquire the catalytic ability by fusing domain B′‐C′ of the second subunit to the first subunit. ¹⁶ In each subunit of EcIDH, there is a U‐turn loop that links domains A and B‐C (Figure 1a). In AvIDH, however, two antiparallel linkers (linkers I and II) occur at the same position that link the putative duplicated region with the remaining regions (Figure 1b).

To construct a chimeric monomeric IDH from EcIDH, we need to break the U‐turn loop of one subunit, and then insert a duplicated fragment corresponding to domain B′‐C′ with two proper linkers (L1 and L2) (Figures 1d, S1, and S2). The resulted protein should contain: N terminal part of domain A (domain A_N) → L1 → domain B′‐C′ → L2 → domain B‐C → C terminal part of domain A (domain A_C). The design of the linkers is very important because it is directly related to the folding and the activity of the chimeric enzyme.

We focused the aligned structures around the U‐turn loops of EcIDH and linker I of AvIDH. In AvIDH, a short β‐sheet (Ser143 to Arg146) before linker I was superimposed upon a fragment (Cys127 to Pro130) before the U‐turn loop of the first subunit in EcIDH. Between the β‐sheet and linker I, there is a short α‐helix (Leu149 to Lys157) in AvIDH, which is absent in EcIDH. The start points of the U‐turn loop of the first subunit (Gly136 in EcIDH) and linker I (His158 in AvIDH) are nearly overlapped but their extension directions are different. The U‐turn loop bends back but linker I extends forward. Linker I and the U‐turn loop of the second subunit converge at a position corresponding to Ala164 in AvIDH and His143′ in EcIDH (Figure 2a). According to this information, the first linker L1 was designed as: (a) replacing the fragment from Pro130 to His143 in EcIDH (14 residues) by the fragment from Arg146 to Ala164 in AvIDH (19 residues) and (b) replacing the fragment from Gly136 to His143 in EcIDH (8 residues) by the fragment from His158 to Ala164 in AvIDH (7 residues, HPHKMGA).

FIGURE 2.

Rational designs of the linkers and the optimization of the newly formed protein surface of the chimeric monomeric isocitrate dehydrogenase (IDH). (a,b) The aligned structures around the U‐turn loops of Escherichia coli IDH (EcIDH) (deep blue and cyan) and linkers I and II of Azotobacter vinelandii IDH (AvIDH) (gray). Based on the alignment, the first linker L1 was designed as the fragment from Arg146 to Ala164 in AvIDH to replace the fragment from Pro130 to His143 in EcIDH (in Mono1), or the fragment from His158 to Ala164 in AvIDH to replace the fragment from Gly136 to His143 in EcIDH (in Mono2); and the second linker L2 was designed as the fragment from Gln395 to Asp400 in AvIDH. (c) The scheme of the construction of the chimeric monomeric IDH. A_N, N terminal part of domain A; U, U‐turn loop; B‐C, domain B‐C; A_C, C terminal part of domain A; L1, the first linker; L2, the second linker. (d) The newly formed protein surface of the chimeric monomeric IDH. The residues on the surface that had been subjected to random mutagenesis were showed in yellow stick representations. The mutant with highest activity (Mono7) has only four missense mutations (Arg208′Leu, Tyr216′Leu, Asp311′Ala, and Vla317′Ser) which were marked with red circles

An α‐helix (Ala383 to His396) before linker II is a member of the four‐helix bundle in AvIDH, which is well aligned with an equivalent α‐helix (Asn303′ to Val317′) in EcIDH. There are two glycines, Gly318′ and Gly319′, following the α‐helix in EcIDH. In contrast, the α‐helix in AvIDH has an additional half‐turn composed of two bulky residues, Gln395 and His396, which guarantee linker II extends in an antiparallel direction of linker I. Linker II and the U‐turn loop of the first subunit converged at Pro401 in AvIDH and Pro144 in EcIDH (Figure 2b). Therefore, the duplicated region of the chimeric IDH corresponding to domain B′‐C′ was confined from Pro144′ to Val317′ (Figure 2c), and the second linker L2 was designed as the fragment from Gln395 to Asp400 in AvIDH (7 residues, QHGAFD) to join Val317′ and Pro144.

2.2. Construction of the chimeric monomeric IDHs

Two chimeric proteins were constructed based on the rational design, each one has the same duplicated region (Pro144′ to Val317′, domain B′‐C′) and the second linker L2 (QHGAFD). One of them (Mono1) has a longer L1 linker composed of 19 residues in AvIDH (RAPLSVKNYARKHPHKMGA) that substitutes for the fragment from Pro130 to His143. The other protein (Mono2) has a shorter L1 linker composed of seven residues in AvIDH (HPHKMGA) that substitutes for the fragment from Gly136 to His143. The genes of them were cloned into the pTrcHisA vector, and transformed into a recombination and IDH deficient strain JMT1 (recA, Δicd::kan; derived from JM109). The transformed strains were streaked onto minimal medium agar plates supplemented with 1 mM IPTG. If a gene did not complement the IDH deficiency, the transformed strain with this gene would not growth due to the lack of glutamate synthesis. ³ The control strain with wild‐type E. coli icd gene formed colonies after 2 days incubation at 37°C. After 4 days incubation, the strain with the gene of Mono2 began to form colonies. The strain with the gene of Mono1 did not growth at all (Figure S3).

The genes of Mono1 and Mono2 were subcloned into the pET22b vector, and the recombinant proteins were overexpressed and purified. The gel filtration experiments showed that ~80% of Mono2 was in monomeric states. Mono2 had a K _m of 121 μM and a k _cat of 0.08 s⁻¹ for isocitrate in the presence of NADP⁺ and Mn²⁺ (Table 1), whereas Mono1 had no detectable activity. These results revealed that the short α‐helix (Leu149 to Lys157) in AvIDH is not compatible in EcIDH and not necessary for the catalysis. The following work was based on Mono2, and Mono1 was not investigated further.

TABLE 1.

Kinetic parameters of the chimeric proteins (Mono2‐7) and wild‐type EcIDH

Enzyme	Linker	Substrate	Mg2+		Mn2+
			K m (μM)	k cat (s−1)	K m (μM)	k cat (s−1)
Mono2	L1: HPHKMGA	NADP+	244.8 ± 128.1	0.027 ± 0.007	151.7 ± 28.0	0.087 ± 0.017
	L2: QHGAFD	ICT	212.3 ± 14.5	0.036 ± 0.011	120.5 ± 74.2	0.076 ± 0.002
Mono3	L1: PCSFNVR	NADP+	251.9 ± 67.6	0.034 ± 0.014	757.4 ± 388.4	0.182 ± 0.050
	L2: DSGANC	ICT	240.7 ± 38.6	0.053 ± 0.041	185.3 ± 27.5	0.142 ± 0.057
Mono4	L1: SANRFTM	NADP+	126.5 ± 18.5	0.132 ± 0.007	266.8 ± 152.2	0.437 ± 0.146
	L2: LVRNIN	ICT	228.2 ± 57.5	0.128 ± 0.013	138.4 ± 21.1	0.351 ± 0.169
Mono5	L1: SACNFRL	NADP+	89.2 ± 3.8	0.207 ± 0.021	106.1 ± 6.6	1.205 ± 0.228
	L2: RDFHFA	ICT	654.5 ± 21.2	0.393 ± 0.004	184.8 ± 25.6	1.279 ± 0.177
Mono6	L1: GLAALFF	NADP+	37.1 ± 6.9	0.821 ± 0.007	62.7 ± 10.1	3.686 ± 0.056
	L2: VRNVHR	ICT	248.2 ± 53.8	1.076 ± 0.089	82.8 ± 2.8	4.030 ± 0.269
Mono7 a	L1: GLAALFF	NADP+	25.5 ± 3.5	2.890 ± 0.349	50.7 ± 5.4	15.989 ± 1.770
	L2: VRNVHR	ICT	84.9 ± 10.9	2.512 ± 0.095	60.7 ± 4.7	14.860 ± 0.566
EcIDH	—	NADP+	10.7 ± 1.0	47.632 ± 1.742	5.7 ± 0.7	27.980 ± 3.521
		ICT	4.4 ± 1.1	32.741 ± 1.533	3.2 ± 0.4	25.740 ± 0.870

Abbreviation: EcIDH, Escherichia coli isocitrate dehydrogenase.

^a In Mono7, the residues on the newly formed protein surface were optimized based on Mono6.

To avoid possible recombination inside the gene and facilitate PCR‐based mutagenesis, the nucleotide fragment encoding the duplicated region (Pro144′ to Val317′, domain B′‐C′) was chemically synthesized with some codons being silently changed (Figure S2), and inserted into the original place of the gene encoding Mono2. Then, the L1 and L2 linkers were optimized by site‐saturation mutagenesis using primers with random nucleotides, and screening on the minimal medium agar plates. After three rounds of selections, we obtained four clones that formed colonies after incubation at 37°C for 3–4 days. The chimeric proteins expressed by these clones were named as Mono3‐6, which have different L1 and L2 sequences and increasing IDH activities (Table 1).

During the construction of the chimeric protein, the duplicated region (domain B′‐C′) was inserted which enable it to form catalytic site with a single polypeptide chain. However, this operation also created a newly formed protein surface, which might have negative influence on the stability and activity of the protein (Figure 2d). To obtain a better chimeric monomeric IDH, this surface was optimized by site‐saturation mutagenesis and screening. Mono6, the chimeric monomeric IDH that had the highest activity among Mono2‐6, was used as a prototype. Fifteen residues (Val150′, Phe152′, Arg153′, Glu164′, Arg208′, Ala212′, Tyr216′, Asp307′, Asp311′, and Val317′ in domain B′‐C′; and Lys230, Lys235, Phe236, Asp283, and Leu291 in domain B‐C; all numbering refers to EcIDH) on the surface were subjected to mutagenesis. The first colony appeared after 2 days incubation on the minimal medium agar plate was picked and sequenced (Figure S3). To our surprise, there were only four missense mutations (Arg208′Leu, Tyr216′Leu, Asp311′Ala, and Vla317′Ser) and a synonymous mutation (Ala212′, GCG to GCC) in the resulting protein (hereinafter called Mono7) (Figures 2d and S2). And the mutation directions of the three of four mutations were from polar to apolar, which were unexpected. The lower mutation frequencies of the residues on the surface than those in the loop regions suggested that the residues had close relationships with each other and any substitution may have large effect on the protein.

2.3. Characterization of the chimeric monomeric IDHs

The purified recombinant Mono7 protein gave a single band on SDS polyacrylamide gel electrophoresis (PAGE) with an apparent molecular mass of ~60 kDa (Figure 3a). MALDI‐TOF mass spectrometry showed that the actual molecular mass of Mono7 was 66.8 kDa (Figure 3b), which was consistent with the theoretical value calculated from the protein sequence. Gel filtration experiments revealed that ~85% of Mono7 in solution was eluted as a peak with a molecular mass of ~60 kDa, which indicated that most of the Mono7 protein was in monomeric state (Figure 3c). The oligomeric states of Mono7 were also analyzed by analytical ultracentrifugation. The sedimentation velocity experiments showed that the major peak (~88.2%) of a Mono7 sample (~14.7 μM) was monomeric (Figure 3d). And the sedimentation equilibrium experiments gave a relatively high dissociation constant of Mono7 (~62 μM) between dimer and monomer (Figure S4). The purified Mono7 protein was subjected to native PAGE and the gels were stained either by Coomassie brilliant blue (CBB) or for the IDH activity. The major protein band was monomeric which migrated fastest. And the corresponding band also showed high activity (Figure 3e). This result indicated that the oligomerization of Mono7 was scarce, and had no contribution to its activity. Therefore, we concluded that Mono7 is indeed a monomeric IDH.

FIGURE 3.

Characterization of the chimeric monomeric isocitrate dehydrogenase (IDH) Mono7. (a) Analysis of protein expression by SDS polyacrylamide gel electrophoresis (PAGE). Lane M: protein marker; lane 1: crude extract of the cells overexpressed Mono7; lane 2: purified Mono7. (b) MALDI‐TOF mass spectra of Mono7. (c) Determination of the oligomeric states of Mono7 in solution by gel filtration on a Superdex 200 Increase 10/300 GL column (GE Healthcare). The calculated protein mass of the main peak (85% of the total protein) was ~60 kDa. And that of the smaller peak was ~130 kDa. (d) The sedimentation velocity ultracentrifugation analysis of Mono7. The sedimentation coefficient distribution c(s) was illustrated for a sample at 14.7 μM. The estimated molecular masses were shown above the peaks. (e) Mono7 samples on native polyacrylamide gels were stained by Coomassie brilliant blue (CBB) or for the IDH activity. Protein ladders were loaded to serve as negative controls (NC)

The kinetic parameters of the recombinant Mono7 as well as Mono2‐6 were determined in the presence of Mg²⁺ or Mn²⁺ (Table 1). Interestingly, each chimeric protein required Mn²⁺ instead of Mg²⁺ for maximum activity, although wild‐type EcIDH has no this metal preference. The similar metal preference for Mn²⁺ was also observed in some natural monomeric IDHs. ¹⁸ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ It suggested that Mn²⁺ may be more helpful to maintain the activity of monomeric IDHs. In each condition, Mono2‐7 exhibited gradually increasing k _cat values but fluctuating K _m values for NADP⁺ or isocitrate, which implied that the screening method we used might be more powerful for optimization of the enzyme activity rather than the affinity for substrates. Mono7 gave a k _cat value of ~15 s⁻¹ for isocitrate in the presence of Mn²⁺ that was approximately half of that of EcIDH (k _cat ~26 s⁻¹). Considering that Mono7 has only one catalytic center per protein while EcIDH has two, the construction of Mono7 almost reached the maximum of the catalytic velocity based on the structure of EcIDH. Among the chimeric proteins, Mono7 had the lowest K _m values for NADP⁺ (51 μM) and for isocitrate (61 μM), respectively. However, these values were still ~10‐fold higher than those of wild‐type EcIDH. Further experiments are needed to construct better chimeric monomeric IDHs with higher affinity for the substrates.

3. DISCUSSION

The above result showed that chimeric monomeric IDHs could be constructed from homo‐dimeric EcIDH by domain duplication and site‐saturation mutagenesis. It also provided a valuable strategy for converting the oligomeric states of other proteins. The constructed chimeric monomeric IDH with highest catalytic activity was Mono7. As some natural monomeric IDHs, ¹⁸ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ it was also found that the activity of Mono7 was higher in the presence of Mn²⁺ (k _cat ~ 15 s⁻¹) instead of Mg²⁺ (k _cat ~ 2.7 s⁻¹). These could be interpreted by that the stability of Mn²⁺‐protein complexes is higher than that of Mg²⁺‐protein complexes, ²⁸ the former may help the monomeric IDHs to keep more closed structures.

However, the K _m values of Mono7 for substrates were still ~10‐fold higher than those of wild‐type EcIDH. We speculated that the lower affinities of Mono7 for substrates might be due to the absence of domain A' of the second subunit. Without the supporting force provided by this domain, the protein might tend to stay in the open state that lowered the affinities for the substrates. The natural monomeric IDHs may compensate the loss of domain A' by another strategy. A closer inspection of the AvIDH structure showed that except for the large duplicated region, there are also several small insertions in the protein sequence. Some of these insertions form gate‐like regions in front of the catalytic center (Figure 1c), which might help the enzyme to close the catalytic center after binding the substrates by interface interaction. To obtain a better chimeric monomeric IDH with higher affinities for substrates, the gate‐like regions of the natural monomeric IDHs should be added. But a preliminary attempt to insert the gate‐like regions into Mono7 was not successful. To achieve this goal, extensive optimization of the interfaces between the inserted regions and the original protein will be necessary.

EcIDH is a homo‐dimeric Type I NADP‐IDH. Although our experiments in this paper demonstrated that EcIDH has a potential to evolve a monomeric NADP‐IDH, it does not mean that the natural monomeric IDHs are evolved in the exactly same pathway. To explore the evolution of monomeric IDHs, we made an extensive phylogenetic analysis using sequences of 98 IDHs of various types and 27 isopropylmalate dehydrogenases (IMDHs) that belong to the same superfamily (Table S1). The cluster formed by the IMDH sequences was set as an outgroup. IDHs were divided into two major clusters. One cluster consists of Type I IDHs, whereas another cluster contains both Type II IDHs and monomeric IDHs (Figure 4a).

FIGURE 4.

Evolutionary relationships of the isocitrate dehydrogenases (IDHs). (a) A maximum‐likelihood phylogenetic tree of IDHs. The root of the tree was placed on the branch linking the IDHs to the isopropylmalate dehydrogenases (IMDHs) based on biochemical evidence. ¹⁰ The NAD‐IDHs and the NADP‐IDHs were depicted in red and blue, respectively. The common ancestor of all types of the IDHs was marked by an asterisk. The places of Escherichia coli IDH (EcIDH) and Azotobacter vinelandii IDH (AvIDH) were denoted by a circle and a diamond, respectively. (b) A deduced evolutionary map of IDHs based on the phylogenetic analysis. The proposed natural evolutionary pathways were depicted with black arrows. The gray curved arrow depicted the conversion of the Type I dimeric EcIDH to the chimeric monomeric ones in this study, which is potential but might not be historical

The phylogenetic tree revealed that Type I IDHs and monomeric IDHs as well as Type II IDHs have a common ancestor. It was proposed that the ancestor of IDHs is a NAD‐IDH, and the occurrences of the NADP‐IDHs are ancient adaptation events. ¹⁰ The early‐found monomeric IDHs such as AvIDH are NADP‐IDHs. However, we have recently discovered that there are monomeric NAD‐IDHs in a few prokaryotic species. ¹⁸ Ancient monomeric IDHs might be NAD‐IDHs that evolved from the ancient dimeric or tetrameric NAD‐IDHs, and monomeric NADP‐IDHs might be evolved later in an adaptation event.

Due to the higher structural similarities such as the clasp‐like regions and the cofactor‐recognition helixes, it had been speculated that the monomeric IDHs might evolved from homo‐dimeric Type II IDHs. ¹⁷ However, our phylogenetic analysis does not support so. The monomeric IDHs are closer to the root than Type II IDHs. Therefore, it is more probable that Type II IDHs evolved from the monomeric IDHs rather than the inverse. Two new subcluster of Type II IDHs, prokaryotic homo‐hexameric NAD‐IDHs and eukaryotic homo‐dimeric NAD‐IDHs, were recently discovered by our laboratory. ¹⁸ , ¹⁹ , ²⁰ A prokaryotic homo‐hexameric NAD‐IDH from Congregibacter litoralis KT71 has a high sequence similarity to homo‐dimeric Type II IDHs except for a 167 aa insert. ¹⁹ Meanwhile, this insert also well aligned to the region of the monomeric IDHs corresponding to domain B‐C of our construction (Figure S5). Therefore, the homo‐hexameric NAD‐IDHs might be evolutionary intermediates between the monomeric and the homo‐dimeric Type II IDHs. And the homo‐dimeric Type II IDHs might be evolved from the homo‐hexameric IDHs that lost the region corresponding to domain B‐C (and retained the region corresponding to domain B′‐C′). Then, an independent adaptation event caused the cofactor dependence of Type II IDHs to be changed from NAD⁺ to NADP⁺ (Figure 4b). A possible evolutionary pathway of Type II IDHs is proposed: ancient dimeric or tetrameric IDH → monomeric IDHs → homo‐hexameric IDHs → homo‐dimeric Type II IDHs (Figure 4b, right), just like a Chinese proverb said, “A long‐lasting separation is always followed by a reunion, and vice versa.” Further investigation on the interconversion of IDHs in different oligomeric states may be helpful in understanding the structures and the full evolutionary map of the IDHs, which would also provide useful information for discovering new anti‐cancer drugs and strategies.

4. MATERIALS AND METHODS

4.1. Bioinformatic analysis

All protein sequences were downloaded from the NCBI web site (http://www.ncbi.nlm.nih.gov). Crystal structure coordinates of AvIDH (PDB ID: 1ITW) and EcIDH (PDB ID: 1CW7) were obtained from the Protein Data Bank (PDB) database (http://www.rcsb.org). To align these two structures, the main chain positions of Ser113, Arg119, Arg129, Tyr160, Asp 307, Asp311, Lys230′, and Asp283′ in EcIDH (the prime sign indicates the second subunit) were respectively aligned with those of Ser132, Arg139, Arg145, Tyr420, Asp548, Asp552, Lys255, and Asp350 in AvIDH using pair fitting function in PyMOL (Delano Scientific). The chimeric monomeric IDHs and their linkers were designed based on the structural alignment.

Phylogenetic analysis was performed with sequences of 98 of various types IDHs and 27 IMDHs (Table S1). These sequences were aligned using ClustalW, ²⁹ and a bootstrapped maximum‐likelihood tree was constructed using MEGA 7.0. ³⁰ The root of the tree was placed on the branch linking the IDHs to the IMDHs based on biochemical evidence. ¹⁰

4.2. Bacterial strains, gene construction, and screening

The recombination and IDH deficient E. coli strain JMT1 (recA, Δicd::kan) was constructed by inserting a kanamycin resistance gene (kan) into the icd gene of JM109 using the λ Red recombinase system. ³¹ Plasmid pZGP3 ¹⁰ was used as the PCR template to amplify the icd gene of E. coli (encoding EcIDH). To construct the gene encoding Mono1, three fragments corresponding to domain A_N (Met1 to Arg129), domain B‐C (Pro144 to Val317), and domain B′‐C′ plus domain A_C (Pro144 to Met416) of EcIDH were amplified by PCR with the primer pairs: EcIDHF (AGCGGATCC ATG GAA AGT AAA GTA GTT GTT CCG) and loop1R1 (CGG ATG TTT ACG CGC ATA GTT TTT CAC GCT CAG CGG CGC ACG ACG CAG GCA GAT GTA GAG AT); loop1F1 (GTG AAA AAT TAT GCG CGT AAA CAT CCG CAT AAA ATG GGC GCG CCT GAA CTG ACC GAT ATG GTT AT) and loop2R (AAA GGCGCC GTG CTG AAC CTG CGC TGC CAG GGC GT); loop2F (CAC GGCGCC TTT GAT CCT GAA CTG ACC GAT ATG GTT AT) and EcIDHR (CCGAAGCTTA CAT GTT TTC GAT GAT CGC). These primers were designed to give an overlapped linker L1 (encoding Arg146 to Ala164 in AvIDH) between the first and the second fragments; a half of linker L2 (encoding Gln395 to Ala398 in AvIDH, QHGA) at the end of the second fragment; and another half of linker L2 (encoding Gly397 to Asp400 in AvIDH, GAFD) at the head of the third fragment. The first and the second fragments were fused using fusion PCR. The resulted fragment could not be directly fused to the third fragment using fusion PCR because they have long identical sequences. Instead, KasI restriction sites (GGCGCC, encoding a glycine and an alanine, underlined above in the L2 encoding sequences) introduced by the primers were used to ligate them together. The gene encoding Mono2 was constructed similarly, but the first fragment corresponding to Met1 to Gln135 of EcIDH was amplified using the primer pair EcIDHF and loop1R2 (CGC GCC CAT TTT GTG CGG GTG CTG ATA GTA ACG TAC CGG ACG); and loop1F2 (CAC CCG CAC AAA ATG GGC GCG CCT GAA CTG ACC GAT ATG GTT) and loop2R. The loop1F2 and loop1R2 were designed to give a different linker L1 that encoding His158 to Ala164 in AvIDH (HPHKMGA).

The two constructed genes were digested with BamHI and HindIII and cloned into pTrcHisA (Invitrogen) to give pTrcMono1 and pTrcMono2 plasmids. The two plasmids were transformed into the JMT1 strain, and streaked onto the plates with the minimal agar medium (K₂HPO₄ 7 g, KH₂PO₄ 2 g, (NH₄)₂SO₄ 10 g, sodium citrate 0.5 g, agar 15 g/L) and 100 mg/ml ampicillin, 10 mg/ml kanamycin, and 1 mM IPTG. The plates were incubated at 37°C.

To avoid possible recombination inside the Mono2 gene and facilitate PCR‐based mutagenesis, the second fragment encoding Pro144 to Val317 in EcIDH was chemically synthesized with some codons being silently changed (Figure S2). The synthetic fragment was fused to the first and the third fragments using the same linkers as Mono2. The resulted gene also encodes the Mono2 protein, but the long identical sequences were eliminated. This gene was cloned into pTrcHisA to give pTrcMono2M plasmid. The L1 and L2 linkers in pTrcMono2M were optimized by the QuikChange multisite‐directed mutagenesis kit (Agilent) using two primers with random nucleotides at the linker positions: loop1Fr (CGT CCG GTA CGT TAC TAT CAG NNN NNN NNN NNN NNN NNN NNN CCG GAG TTA ACG GAC ATG GTG ATT T), and loop2Fr (GAT GCG CTG GCG GCC CAA GTG NNN NNN NNN NNN NNN NNN CCT GAA CTG ACC GAT ATG GTT ATC). The mutant plasmids obtained were transformed into the JMT1 strain and screened on minimal medium agar plates supplemented with IPTG. The colonies formed were picked and sequenced. The genes encoding Mono3‐6 were obtained by this procedure.

To optimize the residues on the newly formed protein surface of Mono6, five primers with random nucleotides at the places that encode the residues were used: Sense1 (A ACG GAC ATG NNN ATT NNN NNN GAG AAT AGC GAG GAT ATC TAC GCC GGC ATT NNN TGG AAG GCG GAT AG), Sense2 (GAG GAG GGT ACG AAG NNN CTG GTG CGC NNN GCG ATT GAG NNN GCG ATC GCG AAT GAC CG), Sense3 (ATG AAT CTG AAT GGC NNN TAT ATC AGC NNN GCG CTG GCG GCC CAA NNN GTG CGT AAT GTC C), Sense4 (TG ACT CTG GTG CAC NNN GGC AAC ATC ATG NNN NNN ACC GAA GGA GCG), and Sense5 (A GAC GTG ATT GCT NNN GCA TTC CTG CAA CAG ATC CTG NNN CGT CCG GCT GAA TAT G). The mutagenesis and screening methods were as above. The gene encoding Mono7 was obtained by this procedure (Figure S2).

4.3. Protein expression and purification

The gene to be expressed was subcloned into pET22b(+) (Novagen), and the resulted plasmid was transformed into the BL21(DE3) strain. The transformant was picked and cultured in LB medium with 100 mg/ml ampicillin at 37°C until A₆₀₀ reached 0.5–0.8. Then, the culture was induced by 0.5 mM IPTG at 20°C for 20 hr. The cells were harvested, resuspended with LEW buffer (50 mM Tris–HCl, pH 7.5, 300 mM NaCl, and 10% glycerol), and disrupted by sonication on ice. The lysate was centrifuged at 10,000g for 20 min. The supernatant was loaded onto a Talon cobalt affinity column (Clontech) and eluted with Elution buffer (LEW buffer with additional imidazole of 150 mM).

4.4. Electrophoreses and staining

The molecular weights of the proteins were estimated by 12% SDS PAGE and CBB staining. The electrophoresis buffer contained 25 mM Tris–HCl, pH 8.3, 250 mM glycine, and 1% SDS. To evaluate the oligomeric states and the IDH activity of the proteins, the samples were subjected to 8% native PAGE. The native electrophoresis buffer was as above but SDS was omitted. The gels were stained by CBB, or using 35 mM Tris–HCl (pH 7.5), 2 mM MgCl₂, 0.01% (wt/vol) phenazine methosulfate, 0.03% (wt/vol) thiazolyl blue tetrazolium bromide, 1 mM NADP⁺, and 1 mM isocitrate to verify the IDH activity.

4.5. Gel filtration experiments

The oligomeric states of the proteins in solution were analyzed by gel filtration. A 500‐μl aliquot of the sample was loaded and separated on a Superdex 200 Increase 10/300 GL column (GE Healthcare) equilibrated with 50 mM NaH₂PO₄, pH 7.5, and 300 mM NaCl. The column was precalibrated using these protein standards: Ovalbumin (45 kDa), Conalbumin (75 kDa), Aldolase (158 kDa), Ferritin (440 kDa), and Thyroglobulin (669 kDa).

4.6. Mass spectrometry

Mass spectrometry analyses were conducted on a MALDI TOF‐TOF 5800 analyzer (AB SCIEX) in linear high mass positive‐ion mode. Sinapinic acid was used as the matrix and trifluoroacetic acid was used as an ionization auxiliary reagent. The data of MS spectra were analyzed using Series Explorer (AB SCIEX).

4.7. Analytical ultracentrifugation

Analytical ultracentrifugation experiments were performed on a ProteomeLab XL‐I analytical ultracentrifuge equipped with an AN‐60Ti rotor (Beckman Coulter). For sedimentation velocity experiments, conventional double‐sector aluminum centerpieces of 12 mm optical path length were used. Protein samples of 380 μl were loaded together with 400 μl buffer, centrifuged at 46,000 rpm and 20°C, using continuous scan mode and radial spacing of 0.003 cm. Scans were collected at 280 nm in 3 min intervals. The absorbance and the cell radius datasets were fitted by using SEDFIT software (https://sedfitsedphat.nibib.nih.gov/software/default.aspx) with a continuous sedimentation coefficient distribution c(s) model.

For sedimentation equilibrium experiments, protein samples of 110 μl and the reference buffer of 120 μl were loaded into a six‐sector Epon charcoal‐filled centerpiece at three concentrations (1.2, 2.5, and 3.0 μM). Samples were spun at 8°C at 9,000; 15,000; and 20,000 rpm until sedimentation equilibrium was achieved. The data were analyzed using SEDPHAT software (https://sedfitsedphat.nibib.nih.gov/software/default.aspx) with a monomer‐dimer self‐association model.

4.8. Enzyme assay and kinetic analysis

Activity of wild‐type EcIDH and the chimeric proteins was measured in 1‐ml cuvettes at 20°C. The reaction mixture included 35 mM Tris–HCl, pH 7.5, 2 mM MnCl₂ or MgCl₂, 1 mM isocitrate, and 1 mM NADP⁺. After adding the protein sample, the producing of NADPH was monitored at 340 nm (ε ₃₄₀ = 6,220 M⁻¹ cm⁻¹) with a thermostated Cary 300 UV–vis spectrophotometer (Agilent). To measure the kinetic parameters of the samples, the concentrations of isocitrate or NADP⁺ were fixed at 1 mM with varying concentrations of the other one. The initial velocities of the reactions were measured, and the apparent K _m and V _max values for NADP⁺ or for isocitrate were obtained by nonlinear regression using Prism 5.0 (Prism). The enzyme concentrations were measured by the Bio‐Rad protein assay kit (Bio‐Rad). The k _cat values were calculated as V _max/[E]₀, where [E]₀ was the initial protein concentration in the reaction mixture.

CONFLICT OF INTEREST

The authors declare no conflicts of interest.

AUTHOR CONTRIBUTIONS

Changqing Tian: Data curation, investigation, writing‐original draft. Bin Wen: Data curation, investigation, writing‐original draft. Mingjie Bian: Investigation. Mingming Jin: Investigation. Peng Wang: Formal analysis, investigation, writing‐review and editing. Lei Xu: Conceptualization, data curation, investigation, writing‐review and editing. Guoping Zhu: Conceptualization, funding acquisition, project administration, resources, writing‐review and editing.

Supporting information

Appendix S1: Supporting Information

Tian C, Wen B, Bian M, Jin M, Wang P, Xu L, et al. From a dimer to a monomer: Construction of a chimeric monomeric isocitrate dehydrogenase. Protein Science. 2021;30:2396–2407. 10.1002/pro.4204