Dimerization facilitates the conformational transitions for bacterial phosphotransferase enzyme I autophosphorylation in an allosteric manner

The bacterial phosphotransferase system is central to sugar uptake and phosphorylation. Enzyme I (EI), the first enzyme of the system, autophosphorylates as a dimer using phosphoenolpyruvate (PEP), but it is not clearly understood how dimerization activates the enzyme activity. Here, we show that EI dimerization is important for proper conformational transitions and the domain association required for the autophosphorylation. EI(G356S) with reduced dimerization affinity and lower autophosphorylation activity revealed that significantly hindered conformational transitions are required for the phosphoryl transfer reaction. The G356S mutation does not change the binding affinity for PEP, but perturbs the domain association accompanying large interdomain motions that bring the active site His189 close to PEP. The interface for the domain association is separate from the dimerization interface, demonstrating that dimerization can prime the conformational change in an allosteric manner.

The bacterial phosphotransferase system is central to sugar uptake and phosphorylation. Enzyme I (EI), the first enzyme of the system, autophosphorylates as a dimer using phosphoenolpyruvate (PEP), but it is not clearly understood how dimerization activates the enzyme activity. Here, we show that EI dimerization is important for proper conformational transitions and the domain association required for the autophosphorylation. EI(G356S) with reduced dimerization affinity and lower autophosphorylation activity revealed that significantly hindered conformational transitions are required for the phosphoryl transfer reaction. The G356S mutation does not change the binding affinity for PEP, but perturbs the domain association accompanying large interdomain motions that bring the active site His189 close to PEP. The interface for the domain association is separate from the dimerization interface, demonstrating that dimerization can prime the conformational change in an allosteric manner.
Enzyme I (EI) is the first protein of the bacterial phosphotransferase system, which catalyzes the sugar transport and phosphorylation [1,2]. EI catalyzes an Mg 2+ -dependent autophosphorylation reaction using phosphoenolpyruvate (PEP) as a substrate, and a phosphoryl transfer reaction to histidine-containing phosphocarrier protein, HPr. EI consists of an Nterminal domain (EIN) comprising an HPr binding subdomain (EINa), and a catalytic phosphohistidine subdomain (EINab), and a C-terminal dimerization domain (EIC) for PEP binding [3,4]. EI switches between open and closed conformational states via large domain motions for its autophosphorylation reaction. Free EI adopts an open state that is relevant to a phosphoryl transfer reaction between EI and HPr [5,6], and PEP binding to EI induces a closed state that enables the autophosphorylation reaction [7]. Transition between two conformational states involves a hinge motion of EINa and a swivel motion of EINab (Fig. 1). Once the hinge motion disengages EINa from EINab, the active site His189 of EINab can be brought to PEP bound on EIC without steric clash by the swivel motion.
EI forms a dimer via EIC, and the dimerization of EI and isolated EIC has been extensively investigated by sedimentation equilibrium experiments [8,9]. PEP binding largely increased EI dimerization, inducing a compact EI dimer formation. It is known that an EI dimer is capable of the autophosphorylation reaction, but how the dimerization makes EI functionally competent is not clearly understood [10,11]. Previously, EI with a defect in dimerization and autophosphorylation activity was found in Salmonella typhimurium strains, and a G356S mutation was linked to the reduced EI activity [12,13]. When the G356S mutation was introduced to EI of Escherichia coli, EI(G356S) also exhibited significantly weaker dimerization and reduced autophosphorylation activity [13]. Here, we used EI (G356S) of E. coli to address the mechanistic link between EI dimerization and autophosphorylation activity. We demonstrate that EI dimerization facilitates the conformational transitions between open and closed states required for the autophosphorylation reaction.

Cloning, protein expression, and purification
Full-length and domain deletion mutants of EI(1-575) were cloned into a pET11 or a pET15b vector with an Nterminal His 6 tag. EI A denotes the active site H189A mutation in the ab subdomain of EI, and EIC(G356S) denotes the G356S mutation in EIC(231-575). EINab    15 NH 4 Cl as the sole nitrogen source. The culture was induced at an A 600 of~0.8 by the addition of 1 mM isopropyl-b-D-thiogalactopyranoside, and harvested by centrifugation after 4 h of induction. The cell pellet was resuspended in 50 mL (per liter of culture) of 50 mM Tris/ HCl, pH 7.4, 200 mM NaCl, 2 mM b-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride, and 1 tablet of protease cocktail inhibitor (S8830 SIGMAFAST; Sigma-Aldrich, St. Louis, MO, USA). The suspension was lysed by three passages through Emulsiflex (Avestin, Ottawa, ON, Canada) after homogenizing and was centrifuged at 24 000 g for 20 min at 4°C. The supernatant fraction was filtered and loaded onto a DEAE column or a HisTrap column (GE Healthcare, Chicago, IL, USA). The fractions containing the proteins were purified by a Super-dex200 column or a Superdex75 column (GE Healthcare) equilibrated with 20 mM Tris/HCl, pH 7.4, 200 mM NaCl, and 2 mM b-mercaptoethanol and were then further purified by monoQ anion-exchange column (8 mL; GE Healthcare) with a 160-mL gradient of 1 M NaCl. All proteins were dialyzed against 20 mM Tris/HCl, pH 7.4, 100 mM NaCl, 2 mM b-mercaptoethanol, and 4 mM MgCl 2 (buffer A) for further analysis.

Circular dichroism
Circular dichroism (CD) spectroscopy was conducted at 25°C using a Chirascan TM -plus CD spectrometer. Wave scans were acquired by sampling data at 1-nm intervals between 200 and 250 nm for far UV CD measurement. Far UV CD spectroscopy was carried out with 10 lM of proteins in buffer A using a 0.5-mm quartz cuvette. Each far UV CD spectrum was obtained from an average of three scans, and the results were presented as mean residue ellipticity (deg cm 2 Ádmol À1 ) at each wavelength.

Multiangle light scattering
Purified proteins were characterized by multiangle light scattering (MALS) following the size exclusion chromatography. Two hundred micromolar proteins was injected onto a WTC-0303 column (Wyatt Technology, Santa Barbara, CA, USA) equilibrated with buffer A in the presence and absence of 10 mM PEP. The chromatography system was connected to an 18-angle light scattering detector (DAWN HELEOS II; Wyatt Technology), a dynamic light scattering detector (DynaPro Nanostar; Wyatt Technology), and a refractive index detector (Optilab t-rEX; Wyatt Technology). Data were collected every 1 s at a flow rate of 0.5 mLÁmin À1 at 25°C. Data analysis was carried out using the software package ASTRA 6 (Wyatt Technology) to determine the molar mass and mass distribution of the sample. NMR spectroscopy NMR spectra were recorded at 25°C on Bruker Avance 600-and 900-MHz spectrometers equipped with x, y, zshielded or z-shielded gradient triple resonance probes. To examine the binding interface of isolated EINab for isolated EIC(G356S), two-dimensional 1 H-15 N heteronuclear single quantum correlation spectra of 0.3 mM 15 N-labeled EINab(H189A) were obtained titrating with unlabeled EIC (G356S) in the presence and absence of 10 mM PEP in buffer A. The backbone chemical shifts of EINab(H189A) were obtained by the comparison with previously assigned chemical shifts of EIN, and confirmed by three-dimensional triple resonance through-bond scalar correlation CBCA-CONH and HNCACB experiments. NMR spectra were processed using the NMRPipe [14] program and analyzed using the PIPP [15] and the NMRView [16] programs.

Impact of G356S mutation on EI and EIC dimerization
We employed the active site mutant EI A (His189 replaced with Ala) in this study, so that PEP binding to EI would not proceed with the autophosphorylation reaction. The circular dichroism spectra were very similar between EI A and EI A (G356S), and also between EIC and EIC(G356S), indicating that the G356S mutation did not perturb the secondary structures (Fig. 2). We examined the dimerization states of the mutants using MALS data. EI A (~63.5 kDa) exhibited 119.5 AE 1.2 kDa at 50 lM of the elution concentration, indicating~90% of a dimer (Fig. 3A). The elution concentration was obtained from the UV 280 absorption at the peak height. On the other hand, EI A (G356S) exhibited 90.1 AE 4.1 kDa at 50 lM of the elution concentration, indicating~40% of a dimer (Fig. 3A). The estimated equilibrium dissociation constant (K D ) of an EI A dimer from the dimer fraction was~0.4 lM that was in good agreement with K D~0 .6 lM from the sedimentation velocity data of wild-type EI [5]. The K D value of EI A (G356S) was calculated as~30 lM, indicating a~75-fold reduction in the dimerization affinity. PEP binding, however, largely restored the EI dimerization, and both EI A and EI A (G356S) formed predominantly a dimer (> 90%) upon PEP binding (Fig. 3B). EIC (~38.5 kDa) exhibited 73.1 AE 1.3 kDa at 50 lM of the elution concentration, and mostly existed as a dimer (Fig. 3C), which was consistent with K D~6 nM from sedimentation velocity data [9]. EIC (G356S) eluted as two peaks of 65.2 AE 4.8 kDa and 42.5 AE 1.6 kDa with a 1 : 2 ratio, both of which corresponded to a mixture of a monomer and a dimer (Fig. 3C). The separate elution of two peaks suggests a possible conformational heterogeneity of EIC (G356S), which was not evident in EI A (G356S). The lower bound of K D value of EIC(G356S) was estimated as 5 lM from the higher molecular weight fraction, indicating that the G356S mutation caused a significantly larger impact on the dimerization of isolated EIC than that of EI. PEP binding increased the dimerization of EIC(G356S) such that EIC(G356S): PEP appeared as a single peak of 67.1 AE 3.8 kDa, indicating~80% of a dimer (Fig. 3D). Taken together, the G356S mutation considerably reduced the dimerization of both EI A (G356S) and EIC (G356S), but PEP binding largely restored the dimerization affinity, especially for EI A (G356S). Although the estimated dimerization constants from the MALS data are only semiquantitative, they are consistent with the reported values of EI A and EIC from previous analytical centrifugation experiments [5,9]. As EI A (G356S) formed a tight dimer that was comparable to EI A , we examined the impact of G356S mutation on the conformational transitions of EI during the phosphoryl transfer reaction.

Impact of G356S mutation on PEP binding and conformational transitions of EI
It has been reported that EI(G356S) suffers from significantly reduced autophosphorylation activity with only~4% of V max of autophosphorylation compared to EI [13]. As EI(G356S) formed a weaker dimer than EI, the reduced activity was supposed to originate from an EI(G356S) monomer. Our study, however, showed that EI(G356S) could form a tight dimer in the presence of PEP, suggesting that the reduced activity could be intrinsic to the EI(G356S) dimer. We investigated the impact of G356S mutation on PEP binding and conformational transitions of EI using EI A (G356S) and its domain deletion mutants by calorimetry and NMR spectroscopy. We recall that EI switches from an open state to a closed state upon PEP binding, which involves a hinge motion of EINa and an association of EINab and EIC by a swivel motion [5,7]. The apparent free energy (DG app ) of PEP binding to EI A is comprised of the free energy contributions from intrinsic PEP binding of EIC (DG PEP ) and accompanying conformational transitions (DG hinge and DG asso ), such that DG app (EI A ) = DG PEP + DG hinge + DG asso , as previously described [17]. When the EINa subdomain is removed from EI A , EI A DEINa binds to PEP and switches to the closed state without the hinge motion, yielding DG app (EI A DEINa ) = DG PEP + DG asso . The PEP binding (DG PEP ) can be directly measured using isolated EIC that binds to PEP without domain motions [18]. Linear combinations of measured DG app (EI A ), DG app (EI A DEINa ), and DG PEP values can then provide individual DG PEP , DG hinge , and DG asso of EI A . Recently, NMR residual dipolar coupling and small-angle X-ray scattering data have revealed that EI(H189A) forms a mixture of partially closed and closed states in the presence of PEP [19]. The partially closed state represents the intermediate state between the open and closed states, where the domain orientation of EINa and EINab resembles the closed state conformation, but EINab is not fully engaged with EIC to catalyze the in-line phosphoryl transfer reaction with PEP. Thus, it should be mentioned that DG asso in our study includes the domain association to form the partially closed state in addition to the closed state observed in the crystal structure. Given that buried accessible surface in the closed state is 3.4 times larger than that in the partially closed state, we speculate a larger contribution of the closed state to the measured DG asso [19].
We first examined whether the G356S mutation would affect the intrinsic PEP binding of EIC. The K D value of PEP binding to EIC(G356S) was measured as 310 AE 100 lM (Fig. 4A), which was comparable tõ 260 AE 80 lM obtained for EIC [18]. Thus, G365S mutation did not perturb the intrinsic PEP binding affinity of EIC. We then examined the conformational transitions of EI A (G356S) upon PEP binding. The DG app value for overall PEP binding and the conformational transitions of EI A (G356S) was measured as À6.6 AE 0.1 kcalÁmol À1 (Fig. 4B), which was À2.1 kcalÁmol À1 smaller than that of EI A (DG app = À8.7 AE 0.2 kcalÁmol À1 ). The decreased free energy change suggests that the conformational transitions of EI A (G356S) upon PEP binding were less efficient than EI A . We further examined whether the hinge motion of EINa (DG hinge ) or the association between EINab and EIC (DG asso ) by the swivel motion was affected by the G356S mutation (Fig. 4C). DG hinge was obtained by subtracting DG app (À8.0 AE 0.1 kcalÁmol À1 ) of EI A -D EINa(G356S) from DG app (À6.6 AE 0.1 kcalÁmol À1 ) of EI A (G356S). DG hinge of EI A (G356S) was 1.4 AE 0.1 kcalÁmol À1 , which was largely the same as that of EI A (Table 1). On the other hand, the G356S mutation significantly reduced the domain association between EINab and EIC. DG asso of EI A (G356S) was  Table 1. À3.3 AE 0.2 kcalÁmol À1 , whereas that of EI A was À5.2 AE 0.2 kcalÁmol À1 , revealing that the G356S mutation reduced the domain association by À1.9 AE 0.3 kcalÁmol À1 ( Table 1). The hinge motion separates EINab and EINa within EIN and exposes hydrophobic interfacial residues between the subdomains, resulting in unfavorable entropic cost ( Table 1). The energetic penalty of the hinge motion is largely compensated by the association between EINab and EIC accompanying the swivel motion, which buries wide hydrophobic interaction surfaces [7]. The enthalpic and entropic changes upon the hinge motion were very similar between EI A and EI A (G356S), indicating that the G356S mutation had little impact on the hinge motion ( Table 1). The G356S mutation, however, profoundly influenced the thermodynamics of the association between EINab and EIC. EI A (G356S) exhibited much weaker domain association that was attributed to high enthalpic cost outweighing entropic gains from the association. The origin of large changes in enthalpy and entropy in EI A (G356S) is not clear, but the large difference in entropic contributions may reflect a less compact dimeric state of EI A (G356S).

Interfaces for the association between EINab and EIC
We further investigated the impact of the G356S mutation on the interface 15 N-EINab for EIC by NMR titration. We monitored the 1 H-15 N HSQC spectra of 15 N-EINab titrating with EIC or EIC(G356S). Neither of the titration experiments showed any change in the absence of PEP, indicating that EINab did not interact with EIC without PEP. This is consistent with previous data that EI adopts predominantly an open state in the absence of PEP [5]. When 0.3 mM 15 N-EINab was complexed with 0.45 mM EIC in the presence of 10 mM PEP, several residues exhibited severe line broadening (Fig. 5A). Residues with the largest line broadening were I5, A7, A102, L149, T164, T168, G178, G184, G185, S188, T190, R195-E198, G206, and N230, and most of them were located at the interaction surface for EIC (Fig. 5C). On the contrary, a similar titration of 0.3 mM 15 N-EINab with 0.45 mM EIC(G356S) in the presence of PEP showed little changes, which is consistent with the weaker interaction between EIC and EINab of EI A (G356S) (Fig. 6). Increasing the concentration of EIC(G356S) up to 0.9 mM resulted in modest changes for a few residues that exhibited line broadening as well in the titration with EIC (Fig. 5B). Thus, EINab employs similar binding interfaces for both EIC(G365S) and EIC in the presence of PEP, albeit much weaker affinity for EIC(G365S).
The dimer interface of EI is mainly comprised of b3a3 loop (333-366) and b6a6 loop (453-477) of EIC. We suppose that the PEP binding, EI dimerization, and domain association events are connected by an intricate signaling network that is allosterically regulated by the b3a3 loop. The b3a3 loop not only constitutes the dimer interface, but also contains Arg358 that interacts with the phosphoryl group of PEP, propagating conformational changes between the dimerization interface and the PEP binding site (Fig. 5D). Initial PEP binding to EI likely adjusts the b3a3 loop conformation to provide optimal dimerization interfaces. A compact EI dimer, once formed, suppresses conformational dynamics prevailing in EIC to facilitate the EIC-EINab domain association, as was demonstrated by NMR relaxation dispersion [20,21]. The sequential signaling cascade from PEP binding to compact dimerization to domain association triggers the overall open-to-closed conformational transition.  Residues of EINab A that exhibited chemical shift changes or line broadening upon the titration are annotated by the residue types and numbers in the spectra. Side and top views (PDB code 2hwg) of (C) EINab associated with EIC and (D) the EIC dimer as a cartoon diagram. EINab is colored in cyan and EIC in red. EINab residues with line broadening are shown as yellow spheres. Individual subunits of the EIC dimer are colored in yellow and red, and Gly356 residues are shown as space-filling models enclosed by dashed circles in orange in (D). EINab in cyan is added on the right panel to illustrate that the interaction surface of EIC for EINab is distant from the dimerization interface. The active site phospho-His189 and oxalate bound on EIC are shown as space-filling models as a visual guidance.
EI are sensitive to a single G356S mutation. We speculate that the hydroxyl side chain of Ser356 might form hydrogen bonds with neighboring Glu350 and Asn352 to distort the b3a3 loop conformation, which could be deleterious to the conformational transition. We note that the dynamic nature at the active site extends to the intermolecular interaction between EI and HPr, suggesting that the conformational plasticity may finetune protein-protein interactions in general [22].
In summary, we demonstrate that the dimerization of EI facilitates the conformational transitions required for the autophosphorylation reaction in an allosteric manner. The loop b3a3 of EIC is in the center for the allosteric regulation between PEP binding, compact dimerization, and conformational transitions of EI. A single G356S mutation in the loop b3a3 was enough to perturb the communication between the dimerization and the conformational transition, leading to a defect in EI autophosphorylation. The mechanistic link between protein dimerization and allosteric regulation may be general in other multidomain multimeric enzymes [23,24].