In vitro reconstitution and characterization of pyruvate dehydrogenase and 2‐oxoglutarate dehydrogenase hybrid complex from Corynebacterium glutamicum

Abstract Pyruvate dehydrogenase (PDH) and 2‐oxoglutarate dehydrogenase (ODH) are critical enzymes in central carbon metabolism. In Corynebacterium glutamicum, an unusual hybrid complex consisting of CgE1p (thiamine diphosphate‐dependent pyruvate dehydrogenase, AceE), CgE2 (dihydrolipoamide acetyltransferase, AceF), CgE3 (dihydrolipoamide dehydrogenase, Lpd), and CgE1o (thiamine diphosphate‐dependent 2‐oxoglutarate dehydrogenase, OdhA) has been suggested. Here, we elucidated that the PDH‐ODH hybrid complex in C. glutamicum probably consists of six copies of CgE2 in its core, which is rather compact compared with PDH and ODH in other microorganisms that have twenty‐four copies of E2. We found that CgE2 formed a stable complex with CgE3 (CgE2‐E3 subcomplex) in vitro, hypothetically comprised of two CgE2 trimers and four CgE3 dimers. We also found that CgE1o exists mainly as a hexamer in solution and is ready to form an active ODH complex when mixed with the CgE2‐E3 subcomplex. Our in vitro reconstituted system showed CgE1p‐ and CgE1o‐dependent inhibition of ODH and PDH, respectively, actively supporting the formation of the hybrid complex, in which both CgE1p and CgE1o associate with a single CgE2‐E3. In gel filtration chromatography, all the subunits of CgODH were eluted in the same fraction, whereas CgE1p was eluted separately from CgE2‐E3, suggesting a weak association of CgE1p with CgE2 compared with that of CgE1o. This study revealed the unique molecular architecture of the hybrid complex from C. glutamicum and the compact‐sized complex would provide an advantage to determine the whole structure of the unusual hybrid complex.


| INTRODUC TI ON
Pyruvate dehydrogenase (PDH) and 2-oxoglutarate dehydrogenase (ODH) are critical enzymes in central carbon metabolism and generate vital high energy compounds acetyl-CoA and succinyl-CoA, respectively; both of which are major CoA derivatives in cell and serve as substrates for protein acetylation and succinylation (Komine-Abe et al., 2017;Mizuno et al., 2016;Nagano-Shoji et al., 2017). PDH and ODH are large enzyme complexes (∼10 MDa) composed of three subunits; E1, E2, and E3. E1 (EC 1.4.2.1) is involved in thiamine pyrophosphate-dependent oxidative decarboxylation of 2-oxoacid (pyruvate or 2-oxoglutarate) with the concomitant transfer of the corresponding acyl group to a lipoyl group attached on a lysine residue of E2. E2 (EC 2.3.1.12) is a dihydrolipoyl acyltransferase that catalyzes the transfer of the acyl group to CoA. E3 (EC 1.8.1.4) is a dihydrolipoyl dehydrogenase that is responsible for FAD-dependent reoxidation of dihydrolipoamide to lipoamide on E2 to generate NADH (Perham, 2000;Patel, Nemeria, Furey, & Jordan, 2014). The complexes are organized in both structural and functional aspects using E2 as the core. E2 protein consists of three different types of domains: one to three N-terminal lipoyl domains (~80 amino acids each), a peripheral subunit-binding domain (PSBD, ~45 amino acids), and a core-forming C-terminal catalytic acyltransferase domain (~250 amino acids). As demonstrated by the crystal structures of PDH and ODH from several organisms (Izard et al., 1999;Knapp et al., 1998;Mattevi et al., 1992Mattevi et al., , 1993Wang et al., 2014), E2 trimer is the principal building block to be assembled into an octahedral (24-mer) or icosahedral (60-mer) symmetrical shape to form the inner core. Both E1 and E3 exist as a dimer and are noncovalently tethered to the PSBD of E2. In many organisms, including Escherichia coli, the E3 protein is shared by PDH and ODH, while specific E1 and E2 proteins are utilized for each enzyme complex.

PDH and ODH from the order Corynebacteriales including
Corynebacterium glutamicum have several unique features. First, there is only a single gene encoding E2 in the genome, and the same E2, as well as E3, proteins are used by PDH and ODH.
Second, there exists a variant protein of E1o with an extra E2 acyltransferase domain in the N-terminal region, which corresponds to OdhA (CgE1o, NCgl1084) in C. glutamicum (Usuda et al., 1996) and Kgd or SucA in Mycobacterium (Tian et al. 2005;Wagner et al. 2019). The deletion of odhA resulted in the loss of ODH activity, and OdhA certainly functions as E1o ).
Mycobacterium SucA functions as a multifunctional enzyme capable of reductive succinyl-transfer to a lipoyl residue via the canonical dehydrogenase reaction (Wagner et al. 2011), nonreductive decarboxylation of α-ketoglutarate (2-oxoglutarate) to produce succinic semialdehyde (Tian et al., 2005;Wagner et al., 2011), and carboligation with glyoxylate to give 2-hydroxy-3-oxoadipate (de Carvalho et al., 2010;Wagner et al., 2011). Third, a small regulatory protein (OdhI in C. glutamicum and GarA in Mycobacterium) interacts with the E1o variant to control the enzymatic activity (Niebisch et al., 2006;Nott et al., 2009). OdhI is composed of two domains: an N-terminal domain, containing phosphorylation sites, and a C-terminal forkhead-associated (FHA) domain, recognizing phosphorylated serine/threonine. The FHA domain also establishes the binding to CgE1o (Krawczyk et al., 2010). When the relevant N-terminal sites are phosphorylated, OdhI undergoes a conformational change through the interaction of the FHA domain with the phosphorylated sites, releasing ODH from the inhibition by OdhI (Barthe et al., 2009).
In a previous study that investigated the subunit organization of C. glutamicum, ODH, CgE2 (AceF, NCgl2126), and CgE3 (Lpd, NCgl0355) were copurified along with a tagged CgE1o as expected.
Interestingly, CgE1p (AceE, NCgl2167) was also copurified with the ODH complex (Niebisch et al., 2006). Furthermore, when a tagged CgE1p was used as prey, all the components of the ODH complex were copurified (Niebisch et al., 2006). This was the first evidence to suggest the presence of a PDH-ODH hybrid complex in C. glutamicum. Further enzymatic and genetic studies revealed that each E2 acyltransferase domain of CgE2 and CgE1o specifically catalyzes the transacetylase and transsuccinylase reaction, respectively, suggesting that the E2 catalytic domain of CgE2 and CgE1o is responsible for PDH and ODH activities . It was also revealed that CgE1o and also CgE2 are required for the ODH activity   (Wagner et al., 2014;Wagner et al., 2019).
All the pieces of evidence strongly support the presence of the PDH-ODH hybrid complex. However, the molecular architecture, including the subunit stoichiometry and arrangement of the hybrid complex, remains to be elucidated. In the present study, we aimed to characterize the C. glutamicum PDH-ODH complex in vivo and in vitro and show the unique molecular architecture of the hybrid complex.

| Bacterial strains and culture conditions
Corynebacterium glutamicum ATCC13869 was used as the wild-type strain. All strains and plasmids used in this study are shown in Table 1.
Corynebacterium glutamicum cells were grown in 25 ml of glutamate production medium (Mizuno et al., 2016) at 31.5°C in 500-ml baffled flasks on a rotary shaker at 100 rpm (70 mm of the shaking width, Takasaki Kagaku, Japan). Tween 40 (1.5 g/l) was added after 3 h of cultivation to induce l-glutamate production, and the same volume of distilled water was added for control (uninduced). Nine-hour cultivated cells were used for the lysate preparation.

| Ultracentrifuge analysis
Corynebacterium glutamicum cells were lysed in TESG15 buffer (50 mM TES-NaOH [pH 7.6] and 15% glycerol) containing 1 mM PMSF, 1 mM DTT, 10 µg/ml DNase, and 10 µg/ml RNase at high pressure using EmulsiFlex-B15 (Avestin Inc., Ottawa). After cell debris removal by centrifugation (7000 g, 10 min, 4°C), the cleared lysate containing 1.2 mg of protein was layered onto a 10%-30% or 15%-45% sucrose density gradient in 50 mM TES-NaOH (pH 7.6). The gradient was centrifuged at 36,000 rpm (max. 230,000 g) in a HITACHI P40ST rotor for 17 h at 4°C. After centrifugation, the gradient was fractionated using a Piston Gradient Fractionator (BioComP Instruments, Fredericton) into 21 fractions of 0.5 ml and the bottom fraction. Escherichia coli DH5α cells were grown in 2 × YT medium until the OD 660 reached 3.5 and processed using the same ultracentrifuge protocol. Aliquots (10 μl) were separated by SDS-PAGE and subjected to Western blot analysis to detect each subunit of PDH and ODH. Denatured samples were prepared by boiling for 5 min in the presence of 1% SDS before ultracentrifugation.

| Preparation of recombinant proteins
Recombinant CgE1o protein was expressed as a C-terminal Histagged protein using the pET-21a vector (pAA62) (Komine-Abe et al.,

2017). Recombinant CgE1p protein was expressed as a C-terminal
His-and Flag-tagged protein using the pET-21a vector (pKH4).

| In vitro reconstitution of PDH and ODH
Dialyzed or gel filtration purified CgE2-E3 proteins were mixed with CgE1p and/or CgE1o in TESG10N buffer and incubated on ice for 30 min. Then, the sample was used for enzyme assays or gel filtration chromatography.
Chromatography was performed at a flow rate of 1.0 ml per min for This study TA B L E 1 Bacterial strains and plasmids used in this study 0.2 ml of sample was applied. Chromatography was performed at a flow rate of 0.2 ml per min for 250 min, and 0.5 ml fractions were collected.

| Enzyme assays
PDH and ODH activities were assayed as described previously Co., Ltd., Japan), and 6 mM pyruvate (for PDH) or 1 mM 2-oxoglutarate (for ODH). Assays were performed at 30°C, and the initial rate of APAD + reduction was monitored using spectrophotometry at 365 nm. Endogenous APAD + reduction in the absence of the substrates was subtracted from the APAD + reduction in the presence of the substrates. One unit of activity was defined as the reduction of 1 nmol of APAD + (ε = 9.1 l μmol −1 cm −1 ) per min.  (Koike et al., 1960). Structural studies revealed that in both complexes, 24 copies of E2 subunit form an octahedral inner core, and up to 24 dimers of either E1 or E3 associate with the E2 core (Knapp et al., 1998;Murphy & Jensen, 2005;Reed et al., 1975). In our ultracentrifuge analysis of E. coli lysates using a 15%-45% sucrose density gradient (Figure 1a), the PDH subunit proteins AceE (E1p), AceF (E2p), and Lpd (E3) were detected in fractions 18-19, representing a PDH complex. The three PDH subunits were also detected in fractions 8-9. Previous studies reported that the PDH complex was present in multiple forms, a major 56-60 S component (a full complex) as well as minor 86-90 S and 20-26 S (a partial complex) components (Danson et al., 1979;Koike et al., 1960). The PDH complexes in fractions 18-19 and 8-9 were assigned as the full and partial complexes of PDH, respectively. Meanwhile, SucA (E1o), SucB (E2o), and Lpd (E3) of the ODH complex were detected in fractions 15-16 and 6-7, which were likely the full and partial complexes of ODH, corresponding to major 37 S and minor 22 S components, according to a previous sedimentation analysis ( Koike et al., 1960), respectively.

| Detection of the PDH-ODH hybrid complex in C. glutamicum lysates using ultracentrifuge analysis
We also detected the ribosomal large subunit protein L3 (RplC) in fractions 15-17, representing ribosomes (an approx. molecular mass of 2.7 MDa). Thus, our centrifugation system could assign PDH, ODH, and ribosomes in fractions with the expected molecular sizes and confirmed that PDH and ODH are different complexes in E. coli.
These results assured that our ultracentrifuge system was reliable to analyze the molecular architecture of large protein complexes like PDH and ODH.
We next analyzed C. glutamicum lysate samples prepared from cells grown in glutamate production medium without Tween 40 (as an inducer) using ultracentrifugation in a 10%-30% sucrose density gradient. CgE1p, CgE1o, and CgE2 proteins were detected in the "light" fraction 4 (and also 5-6) ( Figure 1b). CgE2 was migrated to the lighter fractions (2 and 3) in SDS-treated denatured samples.
We thus speculated that CgE1p, CgE1o, and CgE2 formed a PDH-ODH hybrid complex which was eluted in fraction 4, although the fractions containing CgE3 could not be determined due to the unavailability of the specific antibody. The molecular mass of a hybrid PDH-ODH complex could be much smaller than that of the E. coli PDH or ODH complex (Figure 1b, Figure A1). Interestingly, CgE1p and CgE1o migrated to higher molecular mass fractions (fraction 7 and more) which did not contain CgE2. This may suggest that the two E1 proteins interact with unknown proteins or are multimerized.
We also performed a similar analysis using lysates from cells grown in Tween 40-triggered glutamate-producing conditions and compared the sedimentation patterns of the PDH and ODH subunit proteins. The CgE2 was constantly detected in higher molecular mass fractions in glutamate-producing conditions compared with nonproducing conditions (Figure 1b).

| Reconstitution of the PDH and ODH activities in vitro
We tried to reconstitute PDH and ODH activities in vitro using subunit proteins prepared from E. coli. We successfully copurified CgE2 both CgE1p and CgE1o were added to a reaction mixture, PDH and ODH activities were detected. These results indicated that PDH and ODH activities were successfully reconstituted in vitro.
Next, we analyzed the reconstituted sample containing CgE1o, CgE1p, and CgE2-E3 using gel filtration chromatography. Three peak fractions (a, b, and c) were observed in the chromatogram. Western blot analysis using specific antibodies revealed that the a fraction contained CgE1o, CgE2, and CgE3 proteins, the b fraction contained CgE1o and CgE3, and the c fraction majorly contained CgE1p F I G U R E 1 Detection of PDH and ODH in vivo. (a) Ultracentrifuge analysis of Escherichia coli lysates. The total lysate (TL) was sedimented using a 15%-45% sucrose density gradient. Aliquots (10 μl) of 21-fractionated and the bottom (b) samples were separated using 8% or 10% SDS-PAGE, followed by Western blot analysis to detect each subunit. Black bars indicate the positions of PDH, ODH, and ribosomes detected. PDH p and ODH p represent partial complexes of PDH and ODH, respectively. A representative data set of two biologically independent experiments is shown. (b) Ultracentrifuge analysis of Corynebacterium glutamicum lysates. The TL was sedimented using a 10%-30% sucrose density gradient. Uninduced, conditions without Tween 40 as an inducer; induced, Tween 40-induced glutamate-producing conditions; denatured, denatured samples of uninduced conditions. Detection of each subunit was performed as shown in (a). Black bars indicate the positions of Escherichia coli PDH, ODH, and ribosomes detected in the same conditions (see Figure A1). A representative data set of three biologically independent experiments is shown (Figure 2d, left). ODH activity was detected in the a fraction, indicating that the peak a represents the ODH complex, in which CgE1o was associated with the CgE2-E3 subcomplex. On the other hand, no PDH activity was detected in any peak fraction (Figure 2d, right).
Also, when CgE2-E3 was incubated only with CgE1p, CgE2-E3 and CgE1p were eluted in separate fractions ( Figure A2). These results suggested that the association of CgE1p with CgE2-E3 was too weak to be retained during gel filtration.

| Subunit stoichiometry of the CgE2-E3 subcomplex and ODH complex
The subunit stoichiometry of CgE2-E3 and ODH complexes was analyzed using gel filtration chromatography with molecular size standards. Although the CgE2-E3 subcomplex was eluted outside the range of the size standards, the molecular mass of the subcomplex was estimated to be 780 kDa (Figure 3a). CgE2 and CgE3  Figure A3). Since both CgE1o-containing fractions exhibited ODH activity when incubated with CgE2-E3, they were not inactive aggregates but functional multimers (Figure 3b).
Therefore, we assumed that CgE1o is present at equilibrium between the two multimer forms in solution.
Next, we tried to determine the molecular mass of the ODH complex. The mixture sample containing CgE1o and CgE2-E3 provided a new peak fraction with a higher molecular mass than the individual CgE1o and CgE2-E3 samples (Figure 3c). Since the new peak fraction contained CgE1o, CgE2, and CgE3 ( Figure 3c) and exhibited ODH activity, the active ODH complex was indeed reconstituted in vitro.
The exact molecular mass could not be estimated for the ODH complex because it ranged out of the molecular size standards. However, it was estimated to be approximately 940 kDa.
As described above, the full PDH complex, consisting of CgE1p, CgE2, and CgE3, was not detected in our gel filtration chromatography. The CgE1p (a theoretical molecular mass of 102,826 Da) was mainly eluted in a peak fraction with the estimated molecular size of 170 kDa, which likely represents a dimeric form (theoretically 206 kDa) in solution ( Figure A2).

| Kinetic analysis of in vitro reconstituted PDH and ODH
We performed kinetic analyses of the reconstituted PDH and ODH complexes. For this analysis, the affinity-purified proteins were further purified using gel filtration chromatography to remove contaminated proteins ( Figure A4). As a result of the enzyme assays using a mixture of CgE2-E3 and CgE1p or CgE1o, shown in Figure 4, the apparent K m value of the reconstituted PDH complex for pyruvate was 3.5 ± 0.2 mM with a V max of 0.48 ± 0.01 nmol min −1 mg −1 , and the apparent K m of the reconstituted ODH complex for 2-oxoglutarate was 0.010 ± 0.001 mM with a V max of 0.48 ± 0.00 nmol min −1 mg −1 . Our apparent K m value of the PDH complex for pyruvate was comparable to that reported previously using C. glutamicum lysates (1.7 mM), while our apparent K m of the ODH complex for 2-oxoglutarate was one order smaller than that determined using lysates (0.13 mM) .

| The effect of OdhI
OdhI inhibits ODH by binding to CgE1o (Krawczyk et al., 2010;Niebisch et al., 2006). However, it is unclear whether OdhI just associates with CgE1o in the ODH complex or elicits the dissociation of CgE1o from the full ODH complex to reduce the ODH activity.
We examined the effect of OdhI on in vitro reconstituted ODH complex using gel filtration chromatography. It was observed that OdhI induced a peak shift in the gel filtration chromatogram and a substantial portion of OdhI coeluted with CgE1o as well as CgE2-E3 ( Figure 6). This result suggested that OdhI did not dissociate CgE1o from the ODH complex.

| DISCUSS ION
In the present study, we performed a detailed molecular characterization of PDH and ODH from C. glutamicum in vivo and in vitro. Using ultracentrifuge analysis, we detected the PDH and ODH of C. glutamicum in the fractions much smaller than those of E. coli (Figure 1).

It was shown that the catalytic domains of Azotobacter vinelandii
and E. coli E2p, as well as E. coli E2o, are assembled in 24-mer structures using gel filtration (Knapp et al., 1998;Schulze et al., 1991).
On the other hand, CgE2 likely existed in a hexamer in the CgE2-E3 subcomplex, which is one fourth of the octahedral PDH and ODH complexes, according to our gel filtration analysis (Figure 3). The in vivo evidence and in vitro evidence indicate that PDH and ODH of C. glutamicum are small and compact compared with those of other

species.
A current schematic model of the CgPDH-ODH hybrid complex is shown in Figure 7. The core-forming CgE2-E3 subcomplex (the predicted mass of 831 kDa) is hypothetically composed of two CgE2 trimers and four CgE3 dimers, in which six PSBD sites are available for E1 and E3 subunit binding. Four PSBD sites of CgE2 are occupied by the CgE3 dimer, and the remaining two sites are available for the CgE1 binding. Though we were not successful in detecting a PDH complex using gel filtration analysis, it seems reasonable that PDH activity was saturated with 150 pmol of CgE1p and 18 pmole of CgE2-E3 (Figure 5a). Considering that 18 pmol of CgE2-E3 contains 36 pmol of vacant PSBD sites, to which 72 pmol of CgE1p monomer (36 pmol of the dimer) can bind, our experimental data support our hypothesis on CgE2-E3 stoichiometry. On the other hand, we successfully reconstituted the active ODH complex (Figure 2). Unlike EcE1p, EcE1o directly interacts with the E2 core, not the PSBD, via its N-terminal extended region (Frank et al. 2007;Packman & Perham, 1986). The N-terminal region of CgE1o has a structural similarity to F I G U R E 4 Kinetic analysis of PDH (a) and ODH (b) activities. A mixture of CgE1p (15 μg) or CgE1o (20 μg) with CgE2-E3 (15 μg) was incubated on ice for 30 min for reconstitution and then subjected to enzyme assays. Dots show individual data of triplicate assays the N-terminal region of EcE1o, suggesting its key function in the association of CgE1o with the hybrid complex Niebisch et al., 2006). From the estimated molecular mass (940 kDa) of the ODH complex, it is likely speculated that one or two molecules of CgE1o (theoretical molecular mass of 135 kDa) associate with a single CgE2-E3 subcomplex at most. CgE1o alone was present mainly in a hexamer in solution and equilibrium between multimer forms ( Figure 3b). Mycobacterium smegmatis SucA (KGD), which has a high similarity to CgE1o, is reported to be a dimer in the crystal structure (Wagner et al., 2011(Wagner et al., , 2014. Taking the above observations into consideration, it can be speculated that CgE1o associates with the CgE2 core as a dimer to form the active ODH complex. The ODH activity was not saturated even when 600 pmol of CgE1o (300 pmol of CgE1o dimer) was added to 18 pmol of CgE2-E3 ( Figure 5a); an excess amount of CgE1o dimer were bound to the binding sites on the CgE2 core. We currently do not know the reason for the unsaturated ODH activity with the amount of CgE1o and need to study this further to understand the relationship between the molecular assembly and activity of CgODH.
We observed CgE1o-and CgE1p-dependent decrease in PDH and ODH activities, respectively (Figure 5b,c). This evidence is consistent with the formation of the PDH-ODH hybrid complex, as suggested previously Niebisch et al., 2006), in which both CgE1p and CgE1o subunits associate with the single CgE2-E3 subcomplex. As mentioned above, in E. coli, E1p and E1o utilize different sites (PSBD and the core, respectively) to bind to the cognate E2 (Packman & Perham, 1986;Frank et al., 2007;Arjunan et al., 2014), and CgE1p and CgE1o would not compete for binding to CgE2. Both CgE1p and CgE1o utilize lipoyl residues on CgE2 for tethering the acetyl and succinyl group derived from pyruvate and 2-oxoglutarate, respectively. We thus speculated that CgE1p and CgE1o might compete for the availability of lipoyl residues on CgE2.
Since the active ODH complex was retained during gel filtration, but PDH was not ( Figure 2, Figure A2), we assumed that the association of CgE1o with CgE2-E3 was more stable than that of CgE1p with CgE2-E3. Since the association of E1p with E2 is retained in gel filtration in other bacteria (Lessard et al., 1998;Schulze et al., 1993), CgPDH is likely more fragile compared to other PDHs. C. glutamicum PDH showed a K m value of 3.5 mM for pyruvate (Figure 4), which is comparable to the intracellular concentration (7.5 mM) of pyruvate reported in E. coli (Yang et al., 2001). Since the CgPDH complex seemed more fragile compared to CgODH, and PDH activity was more strongly affected by CgE1o than ODH activity was by CgE1p, the PDH activity might be altered in response to the pyruvate levels and the expression levels of CgE1o. Conversely, the K m value of CgODH for 2-oxoglutarate (0.01 mM) was much lower than the intracellular concentration of 2-oxoglutarate (0.44 mM) (Bennett et al., 2009)   . We assumed that it is because we used the NAD + analog (APAD + ) in enzyme assays, which, in some cases, give different specific activities from those with NAD + (Kavanagh et al., 2007).  (Kim et al., 2010;Schultz et al., 2007), and the decrease in the PDH activity is observed in glutamate-producing conditions (Hasegawa et al., 2008). Considering two enzyme activities on a single complex, the change in the subunit ratio of CgE1p and CgE1o would affect the balance of PDH and ODH activities. In this regard, it was intriguing that the shift of CgE2 toward higher molecular mass fractions was observed in glutamate-producing conditions ( Figure 1b). This may reflect the association of OdhI with the ODH (or hybrid) complex ( Figure 6). Alternatively, the change in the molecular assembly and/or stoichiometry of the hybrid complex could occur, which would contribute to the control of PDH and ODH activities during glutamate overproduction.
We found that CgE1o alone formed multimers, mainly a hexamer, which were ready to form an active ODH complex (Figure 3). The in vitro multimerization of CgE1o agreed with the CgE1o detected in higher molecular mass fractions, independent of CgE2, in vivo ( Figure 1b). The E1o component is usually a dimer (Frank et al., 2007;Wagner et al., 2011). However, considering that CgE1o has an extra E2-catalytic domain capable of forming a trimer (Izard et al., 1999;Knapp et al., 1998;Mattevi et al., 1992Mattevi et al., , 1993Wang et al., 2014), following possibilities for the detected hexamer were considered: three CgE1o dimers and two CgE1o trimers, and we prefer to consider the possibility of two CgE1o trimers assembled via the extra E2 domain (Figure 7). A structural study needs to be conducted to reveal the subunit assembly of the CgE1o hexamer. Meanwhile, CgE1p existed majorly as a dimer in vitro ( Figure A2), and it also migrated to higher molecular mass fractions independent of CgE2 in vivo (Figure 1b). It might be speculated that CgE1p interacted with unknown proteins in vivo. has not been reported. The compact size of the hybrid complex would provide an advantage to determine its whole structure and understand the molecular mechanism to balance the PDH and ODH activities, which should be critical to control the metabolic flux in C. glutamicum.

ACK N OWLED G M ENTS
We thank Takashi Hirasawa for providing the anti-CgOdhA antibody.
We thank Kan Tanaka for providing anti-EcAceE, anti-EcAceF, anti-EcSucA, anti-EcSucB, and anti-EcLpd antibodies. We thank Fujio Kawamura for providing the anti-BsRplC antibody. This work was partially supported by JSPS KAKENHI (grant no. 20K05803) and Noda Institute for Scientific Research.

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R CO NTR I B UTI O N S
Hirokazu Kinugawa: Formal analysis (equal); investigation (equal).

E TH I C S S TATEM ENT
None required.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data generated or analyzed during this study are included in this article.  . Aliquots (5 μl) of fractions (5 ml each) were analyzed using 10% SDS-PAGE and the CBB staining gel image is shown (lower). An estimated molecular mass of the peak fraction containing CgE1p was 170 kDa F I G U R E A 3 Gel filtration chromatography of the CgE1o460k fraction. The fraction containing CgE1o460k were collected and subjected to gel filtration chromatography, which provided two peaks of 750 kDa and 460 kDa again. The chromatography pattern of CgE1o in Figure 3a is shown with a broken line as a reference

F I G U R E A 4
The purity of gel-filtration purified His-tagged CgE1p, CgE1o, and CgE2-E3 proteins. Purified samples (2 μg) were separated using 10% SDS-PAGE and detected using CBB staining. These protein samples were used for kinetic analysis (Figure 4)