Báo cáo khoa học: Catalytic activation of human glucokinase by substrate binding – residue contacts involved in the binding of D-glucose to the super-open form and conformational transitions ppt

Although X-ray crystallographic studies have revealed the a-d-glucose binding residues in the closed state, the con-tact residues that make essential contributions to its binding to the

binding – residue contacts involved in the binding of

transitions

Janne Molnes1,2,3, Lise Bjørkhaug1,2, Oddmund Søvik1, Pa˚l R Njølstad1,4and Torgeir Flatmark3

1 Department of Clinical Medicine, University of Bergen, Norway

2 Center for Medical Genetics and Molecular Medicine, Haukeland University Hospital, Bergen, Norway

3 Department of Biomedicine, University of Bergen, Norway

4 Department of Pediatrics, Haukeland University Hospital, Bergen, Norway

Glucokinase (GK), ATP : d-hexose

6-phosphotransfer-ase (EC 2.7.1.1), catalyses the phosphorylation of

a-d-glucose (Glc) to form a-d-glucose 6-phosphate, using

MgATP2)as the phosphoryl donor It is a key

regula-tory enzyme in the pancreatic b-cell [isoform 1 of human glucokinase (hGK)] [1] and plays a crucial role

in the regulation of insulin secretion, and has been termed the glucose sensor of the b-cell [2] GK is also

Keywords

glucokinase; hysteresis; intrinsic tryptophan

fluorescence

Correspondence

T Flatmark, Department of Biomedicine,

University of Bergen, N-5009 Bergen,

Norway

Fax: +47 55586360

Tel: +47 55586428

E-mail: torgeir.flatmark@biomed.uib.no

(Received 19 December 2007, revised 7

March 2008, accepted 10 March 2008)

doi:10.1111/j.1742-4658.2008.06391.x

a-d-Glucose activates glucokinase (EC 2.7.1.1) on its binding to the active site by inducing a global hysteretic conformational change Using intrinsic tryptophan fluorescence as a probe on the a-d-glucose induced conforma-tional changes in the pancreatic isoform 1 of human glucokinase, key resi-dues involved in the process were identified by site-directed mutagenesis Single-site Wfi F mutations enabled the assignment of the fluorescence enhancement (DF⁄ F0) mainly to W99 and W167 in flexible loop structures, but the biphasic time course of DF⁄ F0 is variably influenced by all trypto-phan residues The human glucokinase–a-d-glucose association (Kd= 4.8 ± 0.1 mm at 25C) is driven by a favourable entropy change (DS = 150 ± 10 JÆmol)1ÆK)1) Although X-ray crystallographic studies have revealed the a-d-glucose binding residues in the closed state, the con-tact residues that make essential contributions to its binding to the super-open conformation remain unidentified In the present study, we combined functional mutagenesis with structural dynamic analyses to identify residue contacts involved in the initial binding of a-d-glucose and conformational transitions The mutations N204A, D205A or E256A⁄ K in the L-domain resulted in enzyme forms that did not bind a-d-glucose at 200 mm and were essentially catalytically inactive Our data support a molecular dynamic model in which a concerted binding of a-d-glucose to N204, N231 and E256 in the super-open conformation induces local torsional stresses at N204⁄ D205 propagating towards a closed conformation, involving struc-tural changes in the highly flexible interdomain connecting region II (R192-N204), helix 5 (V181-R191), helix 6 (D205-Y215) and the C-terminal helix 17 (R447-K460)

Abbreviations

PDB, protein databank.

expressed in the liver (hGK isoforms 2 and 3) [3] and

in the central nervous system (hGK isoform 1) [4],

where the enzyme has a similar important function in

glucose metabolism In humans, a number of naturally

occurring mutations in the GK gene (GCK) have been

detected in patients suffering from familial, mild

fast-ing hyperglycemia (maturity-onset diabetes of the

young type 2; MODY2), persistent hyperinsulinemic

hypoglycemia of infancy and permanent neonatal

diabetes mellitus [5–7]

Although GK is a monomeric enzyme, it shows

non-hyperbolic (sigmoidal) dependence on Glc concentration

in steady-state enzyme kinetics [8,9] However, the

equi-librium binding of Glc alone is characterized by a

hyper-bolic binding isotherm, as first determined by intrinsic

tryptophan fluorescence (ITF) spectroscopy of rat liver

glucokinase [10] This enzyme [10] and the recombinant

human enzyme [11] are both activated in vitro by

incu-bation with Glc, and the process has been described as a

reversible transition from an inactive, low affinity state

to a high activity, higher affinity state [10,11] Crystal

structure analyses of the unliganded and Glc-bound

hGK [12] have confirmed the biochemical and

biophysi-cal studies by demonstrating that the binding of Glc at

the active site indeed induces a large-scale domain

move-ment that closes the active site cleft and creates the

stereochemical environment for binding of the

cosub-strate (MgATP2)) and thus catalysis Moreover, the

maximal activation of rat liver glucokinase by Glc and

the related overall conformational transition, as

fol-lowed in real-time by ITF spectroscopy, was shown to

be a relatively slow process [10] characteristic of a

hys-teretic enzyme [13] Although X-ray crystallographic

studies have revealed the structures of the unliganded

(super-open) and Glc-bound (fully-closed) states of

hGK [12], the residue contacts that make essential

tributions to the binding of Glc to the super-open

con-formation have not been identified The characterization

of these residues is important for our understanding of

how substrate binding is coupled to the global

confor-mational transition and catalytic activation

In the present study, recombinant wild-type hGK

and selected mutant forms were isolated aiming: (a) to

examine the contribution of its three tryptophan

resi-dues (Fig 1A) to the multiphasic fluorescence

enhance-ment induced by Glc binding; (b) to identify the active

site residues involved in the binding of Glc to the

super-open state (Fig 1C) of this two domain [large

(L) and small (S)] enzyme, and thus the site of

initia-tion of the global conformainitia-tional transiinitia-tion; and (c) to

gain some insight into how the local torsional stresses

at the contact residues in the super-open state

propa-gate through the structure towards cleft closure and a

catalytically competent conformation To explore these aspects, we used a combined approach of molecular dynamics studies by real-time ITF spectroscopy, struc-tural dynamic analyses and functional mutagenesis Our findings provide new insight into the catalytic acti-vation of hGK by substrate binding that will be valu-able in studies of human diseases associated with mutations in the GCK gene, notably in some mutations

in which the molecular mechanism is not yet under-stood

W257 Glc

CompA

T168

K169

D205

N204 N231 E256 E290

E290

E256

N231

N204

D205

A

B

C

Fig 1 (A) Localization of tryptophan residues in the 3D structure

(Glc) and the allosteric activator compound A (PDB identity: 1v4s) (B) The Glc contact residues in the substrate-bound state (PDB identity: 1v4s) All residues were individually mutated (C) The spa-cial proximity of active site residues in the L-domain and connecting

in the super-open state (PDB identity: 1v4t) The structural images

Tryptophan residues in wild-type hGK

The crystal structures of hGK [12] have identified the

positions and the interactions of its three tryptophans in

the absence (super-open state) and in the presence

(fully-closed state) of Glc and

2-amino-4-fluoro-5-(1-methyl-1H-imidazol-2-ylsulfanyl)-N-thiazol-2-yl-benzamide, a

synthetic allosteric activator termed compound A

(Fig 1A) Based on the coordinates of the two

struc-tures [protein databank (PDB) identity: 1v4t and 1v4s],

molecular motion analyses (http://molmovdb.mbb

yale.edu/cgi-bin/morph.cgi?ID=496337-23316) revealed

a change in the backbone dihedral torsion angle

(Du + Dw) for W99, W167 and W257 to be 110.5,

26.3 and )0.2, respectively It should be noted that residues 157–179, unassigned in the electron density map of the super-open structure (1v4t), were ‘repaired’

by the molmovdb algorithm

Steady-state kinetics of wild-type hGK and the

Wfi F mutant forms

As previously reported [14,15], the wild-type hGK and wild-type glutathione S-transferase (GST)-hGK dem-onstrated the same steady-state kinetic parameters as well as the Kd value for Glc in the ITF binding assay (Fig 2C), and the GST fusion proteins were therefore mostly used in the kinetic analyses of mutant proteins The wild-type GST-hGK revealed a positive kinetic cooperativity with Glc [Hill coefficient (nH) = 1.7

4.2 4.4 4.6 4.8 5.0 5.2 5.4 5.6 5.8

0 20 40 60 80

100

Feq

0 20 40 60 80 100 120 140 160

Kd

[r2 = 0.96]

0.0 0.2 0.4 0.6 0.8 1.0 [r2 = 0.99]

[Glc] (m M )

x10 3 (K –1 ) 1

T

338.6 nm

340.5 nm

Wavelength (nm)

320 340 360 380 400 420 440

340.3 nm

341.0 nm

Wavelength (nm)

320 340 360 380 400 420 440

by increasing concentrations of Glc The solid lines represent the fit of the data to two hyperbola as obtained by nonlinear regression

was 295 nm and excitation and emission slits were 3 and 7 nm, respectively (A, B), or 4 and 7 nm, respectively (C, D).

± 0.1] with a substrate concentration yielding

half-maximum saturation ([S]0.5) of 8.4 ± 0.2 mm, whereas

all three Wfi F mutants demonstrated a reduced

catalytic activity (Table 1) W167F-hGK showed a

pronounced reduction in ‘catalytic efficiency’

(24-fold), with an approximate three-fold reduction in

Vmax and a six-fold increase in the [S]0.5value for Glc,

and the Hill coefficient was reduced to nH=

1.12 ± 0.08 The W257F mutant also revealed a

slightly reduced affinity for Glc, whereas W99F

showed a small increase in both affinity and

‘cata-lytic efficiency’ A normal positive kinetic

cooper-ativity was observed for the W99F and W257F mutant

forms

Correlation of tryptophan environments with

fluorescence properties

The static solvent accessibility of W99, W167 and

W257 in the super-open⁄ fully-closed state was

calcu-lated [16] as 27⁄ 45, X ⁄ 4.6 and 0.8 ⁄ 0.0%, respectively

W167 and W257 are ‘buried’ tryptophans, whereas

W99 is a surface residue with a high degree of

expo-sure to aqueous solvent in both states, notably in the

Glc-bound state Note that no number exists for W167

in the super-open state because residues 157–179 are

unassigned in the electron density map [12] Figure 3A

(kex= 295 nm) of wild-type hGK at pH 7.0 with

kmax 340.5 nm (kmax 341 nm for wild-type

GST-hGK; Fig 2B), consistent with the solvent accessibility

of the three tryptophans On denaturing with 6 m

gua-nidium hydrochloride, a red shift was observed

(kmax 357 nm), close to the spectrum for free

trypto-phan (data not shown) On increasing the temperature

from 7C to 37 C, an approximately 25% decrease in

the fluorescence intensity at kmaxand an approximately

5.1 nm red shift in the kmax were observed (Fig 3B)

These changes suggest a complex effect of temperature

on the conformational substates of the apo enzyme,

presumably with a more solvent-exposed W99 at the

higher temperature Moreover, in rapid mixing

experi-ments a temperature change from 1C to 39 C resulted in a time dependent quenching of the fluores-cence within a time scale of approximately 6 min (Fig 3C) A semi-log plot (Fig 3C, inset) revealed a biphasic time course with a relatively fast phase (t1 ⁄ 2 11 s) and a slow phase (t1 ⁄ 2 64 s)

Effect ofD-glucose on the equilibrium fluorescence of wild-type hGK and its W fi F mutant forms

From the equilibrium fluorescence spectra of wild-type hGK (Fig 2A), it is seen that, upon the addition

of 200 mm Glc, fluorescence (DFeq⁄ F0) increases by approximately 60%, with 1.9 nm blue shift in kmax

A similar effect was seen with the wild-type GST-hGK fusion protein (Fig 2B) The increase in fluores-cence units was comparable without (DFeq) F0 33) and with (DFeq) F0 30) fusion partner, considering the experimental error in determining the absorption coefficients at 280 nm for the two proteins No effect

of Glc was observed on the fluorescence spectrum of the isolated GST protein (data not shown) The pro-tein with and without fusion partner also shows the same time-dependence of the fluorescence enhance-ment (see below) and revealed identical hyperbolic binding isotherms for Glc with Kd values of 4.8 ± 0.1 and 4.9 ± 0.1 mm, respectively, at 25C (Fig 2C) These data demonstrate that the fusion partner at the N-terminal does not perturb the sub-strate-induced conformational changes of hGK, and the GST fusion proteins were therefore used alterna-tively in the studies of mutant proteins due to their potentially higher in vitro stability To better under-stand the driving force of the hGK–Glc interaction, the temperature dependence was determined in the range 7–32 C (Fig 2D) A least-square linear fit (r2= 0.96) yields a DHvan’t Hoff of 32 ± 3 kJÆmol)1 from the slope and a DS of 150 ± 10 JÆmol)1ÆK)1 from the y-axis intercept Thus, the favourable DS overcomes the unfavourable DH and drives the association between hGK and Glc

Table 1 Steady-state kinetic parameters of wild-type GST-hGK and its Trp mutant forms.

Ha

To simultaneously demonstrate the effect of Glc on

both the fluorescence enhancement and the spectral

shifts, the +Glc⁄)Glc fluorescence difference spectra

were recorded for wild-type and Wfi F mutant forms

The difference spectrum of wild-type GST-hGK

(DFeq⁄ F0 27%) revealed a kmax 334 nm (Fig 4A)

compatible with an additive contribution of the three

tryptophans to the Glc-induced fluorescence

enhance-ment At identical protein concentrations, all the

Wfi F mutant forms resulted in a decreased

Glc-induced fluorescence enhancement, being most

pronounced for W99 (DFeq⁄ F0 10%) (Fig 4B) and

W167 (DFeq⁄ F0 6%) (Fig 4C), whereas a DFeq⁄ F0

value of approximately 19% was observed for the

W257F mutant form (Fig 4D) The W167F and

W257F mutants revealed only an approximate 2 nm

shift in kmax in the +Glc⁄)Glc difference spectra compared to wild-type GST-hGK, whereas the W99F mutant demonstrated an approximately 11 nm blue shift, as expected from the low solvent accessibility of the remaining W167 and W257 residues For wild-type GST-hGK and the two ‘buried’ tryptophan mutants (W167F and W257F), a close correlation (r2= 0.99) was observed between the DFeq⁄ F0value and the cata-lytic activity at 200 mm Glc (Fig 5 and supplementary Table S1) This correlation is presumably related to a variable perturbation of the overall structural dynam-ics in the W167F and W257F mutant forms that affects the time-dependent Glc-induced conformational change (supplementary Fig S1) and catalytic activa-tion to the same extent This is in contrast to the W99F mutant form because W99 is more solvent

110 115 120 125 130 135

338.6 nm

343.7 nm

340.5 nm

Ft

Ft= 0

7 °C

37 °C

320 340 360 380 400 420 440

Wavelength (nm) Wavelength (nm)

Time (s)

0 20 40 60

0

20 10

40 30

50 60

80 100 120 140 160

0.2 0.4 0.6 1.0 1.2 1.6 1.8 2.2

Time (s)

10 15 20 25 30 35 40 110

120 130 140 150

Temp ( °C)

B A

C

expected decrease in fluorescence intensity at k = 340 nm, with an end point at approximately 6 min The semi-log plot (inset) shows a

fluores-cence intensity of 20 data points (2 s) (first phase) or 100 or 200 data points (10 s or 20 s, respectively) (second phase) The data were

emission slits were 3 and 7 nm in (A, B) or 4 and 7 nm in (C), respectively.

exposed in a highly flexible surface loop The far-UV

CD spectrum of wild-type hGK revealed negative

bands at 208.5 and 222 nm (data not shown)

charac-teristic of a protein predominated by a-helical

second-ary structure, with an apparent a-helical content of

approximately 31% W167F-hGK revealed a similar

CD spectrum, but with an estimated slightly reduced

a-helical content The thermal denaturation profiles of

the two proteins, as measured at 222 nm in the

pres-ence of 50 mm Glc, gave Tm values of 44.2C and

42.4C for wild-type hGK and W167F-hGK,

respec-tively These data demonstrate that the secondary

structure and conformational stability of the W167F

mutant is relatively well preserved, and support the

conclusion that the functional effects of the Wfi F

mutation are presumably mainly related to a structural

perturbation due to its localization next to T168 and

K169 whose side-chains normally form hydrogen bond

interactions with Glc in the fully-closed conformation

D-glucose-induced conformational dynamics The time course of the fluorescence enhancement induced by Glc was followed on a second-to-minute time scale As shown in Fig 6, a rapid initial phase (0–5 s) represented approximately 80% of the total increase in fluorescence of wild-type hGK (Fig 6A) and wild-type GST-hGK (Fig 6B), and includes the two phases observed by transient kinetics [11], but the equilibrium level (DFeq⁄ F0) was not reached until approximately 3 min at 25C The data in Fig 6C refer to the total fluorescence change, DFeq⁄ F0, (black bars) or the amplitude of the fast phase, DFinitial⁄ F0 (gray bars) or the slow phase, DFslow⁄ F0, (open bars), all relative to the baseline value F0 A biphasic time course was also observed for the W99F and W257F mutant forms (supplementary Fig S1B,D) although the total amplitude at equilibrium and the relative pro-portion of the two phases varied (Fig 6C), and the time required to reach the equilibrium value increased

By contrast, in the W167F mutant form the rapid phase dominated, with a scarcely detectable slow phase, and the overall amplitude was markedly reduced (Fig 6C and supplementary Fig S1C) This may be related to the loss of kinetic cooperativity of Glc binding (mH= 1.12 ± 0.08; Table 1)

a-d-Mannoheptulose (MH) is a nonmetabolized competitive inhibitor of GK This Glc analogue has been proposed to bind at the catalytic site in the closed conformation of GK [17,18] with a 50% inhibition at

336.4 nm

334.3 nm

323.4 nm

∼332 nm

320 340 360 380 400 420

Wavelength (nm)

320 340 360 380 400 420 Wavelength (nm)

320 340 360 380 400 420

Wavelength (nm)

320 340 360 380 400 420 Wavelength (nm) 0

10

20

30

0

10

20

30

0 10 20 30

0 10 20 30

forms The Glc-induced fluorescence changes of the GST fusion

fluorescence difference spectra Each spectrum was obtained by

subtracting the signal averaged spectra obtained in the absence of

Glc from the spectra obtained in the presence of Glc (A) Wild-type

GST-hGK; (B) W99F GST-hGK; (C) W167F GST-hGK and (D) W257F

wavelength of 295 nm and excitation and emission slit widths of 3

and 7 nm, respectively.

WT

W99F W257F

W167F

[r2 = 0.99]

Feq

5 10 15 20 25 30

Relative catalytic activity

induced fluorescence enhancement of wild-type GST-hGK and

val-ues for fluorescence intensity (Table S1) A linear correlation of

mutant forms Graphic points including error bars represent the mean ± SD of three or four measurements.

approximately 2 mm [19] From supplementary

Fig S2A, it is seen that MH binds to the super-open

conformation and induces an equilibrium enhancement

of hGK fluorescence similar to Glc and with a similar

biphasic time dependency From the hyperbolic

bind-ing isotherm (supplementary Fig S2B), a Kd of

8.0 ± 0.7 mm was calculated at 25C

Functional mutation analysis of Glc contact residues at the active site

The 3D structure of the closed state of hGK (PDB identity: 1v4s) has revealed that Glc is hydrogen bonded to amino acids in the L-domain (residues N204, D205, N231, E256 and E290) and the S-domain (residues T168 and K169) (Fig 1B) To identify the contact residues involved in the initial binding of Glc

to the super-open state (Fig 1C) of this two domain hinge-bending enzyme, all the actual residues were individually mutated (supplementary Table S2) The mutant forms were expressed as GST-fusion proteins and subjected to steady-state enzyme kinetics and Glc-induced fluorescence enhancement analysis (Table 2) The main results of this screen (Table 2) are alternatively shown in supplementary Fig S3, includ-ing the ‘catalytic efficiency’ (kcat⁄ [S]0.5) (black bars) and the fluorescence enhancement at 200 mm Glc, (DFeq⁄ F0)max (gray bars) The mutations in the L-domain (N204A, D205A and E256A⁄ K) resulted in enzyme forms that did not give any fluorescence enhancement by Glc and they were essentially catalyti-cally inactive at a Glc concentration of 200 mm N231A gave a DFeq⁄ F0 response of approximately 6% versus wild-type and no measureable activity By con-trast, the mutations in the S-domain (T168G and K169N) experienced a variable partial loss ( 20– 40%) of Glc-induced fluorescence enhancement, with

an increased Kd value and reduced catalytic activity (Table 2) The titration curves for the mutants T168G, K169N and Q287V all revealed clear hyperbolic bind-ing isotherms for Glc (r2= 0.99) (data not shown) For the mutant N231A, the accuracy of the experi-ments was hindered by the low fluorescence response

to Glc (DFeq⁄ F0 at 200 mm Glc 6%), but the data were fitted to a hyperbolic binding curve (r2= 0.91) (data not shown)

Structural dynamic analyses 3D structural analyses of the Glc-induced conforma-tional changes [12] revealed that the enzyme is a very dynamic structure with a high conformational flexibil-ity The crystallographic B factor values for Ca car-bons (Fig 7A), demonstrating the freedom and restriction for various sites, revealed low values (£ 30 A2) for the Glc-interacting residues in the unli-ganded state, except for T168 and K169 The confor-mational fluctuations, computed by the Gaussian network model (GNM) [20,21], revealed similar sites (minima) of low translation mobility compatible with N204, D205, N231 and E256 (Fig 7B) as potential

0

5

10

15

20

25

30

Time (s)

Time (s)

50

60

70

80

90

100

A

B

C

Fluorescence intensity 100

110

120

130

140

mutant forms (A, B) The time course for the Glc-induced

fluores-cence enhancement of wild-type hGK (A) and wild-type GST-hGK

comparison of the time-dependent fluorescence enhancement in

values listed in supplementary Table S1 for the total change in ITF

of 295 nm and excitation and emission slit widths of 4 and 7 nm,

respectively Each column represents the mean ± SD of three

mea-surements.

ligand binding sites The binding of Glc changes not only the tertiary structure (large scale domain motion), but also the secondary structure and side-chain posi-tioning⁄ interactions Thus, 17 helices were identified in the unliganded super-open state (PDB identity: 1v4t) versus 19 helices in the Glc and allosteric activator-bound closed state (PDB identity: 1v4s) The changes

in the backbone and side-chain dihedral angles for the Glc contact residues are shown in Table 3

1 Æs

] 0.5

k cat

S0.5

1 Æs

(nH

Feq

⁄F0 ) max

Kd

a Catalytic

nH

] 0.5

0.000 0.002 0.004 0.006 0.008 0.010 0.012 0.014

N204

T168 K169

E256

N231

Q287 E290

T168

N204 D205

N231

E256 Q287 E290

0 20 40 60 80

100 A

B

2 )

Residues

Residues

mobilities in the global modes (B) for the unliganded state of wild-type hGK (PDB identity: 1v4t) The residue fluctuations (B) were predicted by the GNM [20,21], and the profile represents the slow-est frequency mode As indicated, the residues 157–179 of the S-domain are unassigned in the electron density map.

Table 3 Changes in backbone and side-chain dihedral angles of Glc contact residues in wild-type hGK on binding Glc Values are calculated from the coordinates of the super-open (PDB identity: 1v4t) and fully-closed (PDB identity: 1v4s) state.

It should be noted that the GK activator

(pound A; Fig 1A) binds to the binary hGKÆGlc

com-plex, but it is not known in what way its binding

perturbs the structure [12]

Discussion

The multiphasic global conformational transition

and the kinetic cooperativity ofD-glucose binding

Based on enzyme kinetic studies [8,9], crystal structure

analyses [12] and real-time ITF spectroscopy

[10,11,22], there appears to be broad agreement that

the catalytic activation of monomeric GK by its

sub-strate Glc can be presented by the equation:

GKþ Glc $k1

k 1

GK Glc $k2

k 2

GK Glc ð1AÞ where GK represents the ligand-free, inactive state of

the enzyme, GK Æ Glc its binary low-activity

(low-affinity) enzyme–substrate complex and GK* Æ Glc the

binary complex of the high-activity (higher affinity)

state of the enzyme in which a relatively slow

confor-mational change (isomerization) has occurred,

charac-teristic of a hysteretic enzyme [13] In transient kinetic

analyses of the Glc-induced enhancement of ITF with

wild-type hGK [11] and seven activating mutations

[22], a biphasic time course was observed suggesting

two kinetically distinguishable events within the time

scale of 0–5 s The observed rate constant for the first

phase, kobs1, was linearly dependent on the Glc

con-centration, whereas the second-phase rate constant,

kobs2, exhibited a hyperbolic dependence on the

sub-strate concentration The amplitude of the first and

second phase represented approximately 25% and

75%, respectively, of the total fluorescence

enhance-ment of wild-type hGK [11] Based on these analyses,

it was concluded that the positive cooperativity of GK

observed in enzyme kinetics [8,9] is a kinetic behavior

that is mediated by the Glc-induced conformational

change with intermediate stable states of different

affinity for Glc

In previous transient kinetic analysis [11], the time

scale was 0–5 s, whereas, in the present study (Fig 6),

the equilibrium enhancement (DFeq⁄ F0) of ITF is not

reached until approximately 3 min at 25C in

wild-type hGK (and wild-wild-type GST-hGK) and found to be

very temperature dependent (data not shown) This

suggests that more than two discernible consecutive

steps (Eqn 1A) may accompany Glc recognition and

binding at the active site, a conclusion that is further

supported by the molecular dynamics and targeted

molecular dynamics simulations on the enzyme in its

transition from the fully-closed to the super-open state [23] The simulations indicate that the overall confor-mational transition includes three likely stable interme-diate states with variable degrees of cleft opening Our results provide an additional framework to understand the Glc-induced enhancement in ITF of wild-type hGK First, the Wfi F mutation analyses reveal that all the three tryptophans contribute to the overall enhancement of ITF induced by Glc, with major contributions of W99 and W167 (Figs 4 and 5 and supplementary Table S1), both located in highly flexible loop structures W99, located in one of the regions connecting the L- and S-domains, undergoes a

[(Du + Dw) of 110] upon Glc binding, whereas the corresponding change in the microenvironment is more uncertain for the buried W167 (residues 157–179 not assigned in the super-open state) Second, the hGK– Glc association (Kd= 4.8 ± 0.1 mm at 25C) is dri-ven by a favourable entropy change (DS = 150 ±

10 JÆmol)1ÆK)1), which overcomes an unfavourable enthalpy change (DHvan’t Hoff= 32 ± 3 kJÆmol)1) The relatively large positive DS is in keeping with that

an increase in protein dynamics plays a dominant role

in the interaction, with large scale domain movement and cleft closure (desolvation) as well as changes in peptide backbone conformation⁄ side-chain rotameric states [12] Finally, the temperature induced ( 1 C to

39C) reversible quenching of ITF (Fig 3C) is consis-tent with a slow conformational isomerization, and the biphasic time course (t1⁄ 2 11 s and t1⁄ 2 64 s) sug-gests the presence of a relatively stable intermediate in the transition This reversible isomerization (‘thermal hysteresis’) is reminiscent of the Glc-induced confor-mational isomerization, and supports the existence of

an equilibrium between conformational substates in the apo enzyme Interestingly, recent pre-steady-state analyses of Glc binding to wild-type hGK [24] have provided evidence that the substrate-free enzyme in solution is in a preexisting equilibrium between at least two conformers (i.e super-open and closed) which differ in their affinity for Glc as presented by the equation:

GKþ Glc $ GK  Glc $ GK Glc l

GKþ Glc $ GK Glc

ð1BÞ

where the binding of Glc shifts the equilibrium towards the high activity (higher affinity) closed state

of the enzyme (GK*)

In the present study, the time course and equilib-rium (DFeq⁄ DF0) fluorescent enhancement induced by Glc were studied in the absence of MgATP2) because

such data are not complicated by turnover conditions

with the formation of glucose 6-phosphate The

possi-bility, therefore, arises that the ITF responses and the

related structural conformational changes measured

for the GKÆ Glc binary complex may be different in

the GKÆ Glc Æ MgATP ternary complex The question

was recently addressed by Kim et al [24] using a

non-hydrolyzable ATP analogue (PNP-AMP) Their

tran-sient kinetic analyses suggested that PNP-AMP may

change the equilibrium between the two proposed GK

conformers (Eqn 1B), but the accuracy of the

experi-ments was hindered by the low signal amplitude [24]

Therefore, they also studied the equilibrium binding of

PNP-AMP to the enzyme, and reported a relatively

large decrease in fluorescence (DFeq⁄ F0) which was

interpreted as a nucleotide induced conformational

change However, because no corrections were made

for the large inner-filter effect due to the significant

absorbance of the nucleotide at the selected excitation

wavelength of 285 nm, further fluorescence analyses in

which proper corrections are made for the inner-filter

effect are required before any conclusions can be

drawn Additionally, Heredia et al [11] have

per-formed differential scanning calorimetry of wild-type

hGK and concluded that 10 mm MgATP2), in contrast

to 100 mm Glc, did not have any significant effect on

the Cpexc(kcalÆmol)1ÆC)1) and thermal midpoint

tran-sition temperature, further supporting the conclusion

that more studies are required to settle the issue of

a possible MgATP2) induced conformational change

in GK

Residues involved in the binding ofD-glucose to

the super-open conformation

The 3D structure has revealed that GK is a typical

two-domain enzyme and, in the unliganded state, the

L- and S-domains are far apart and bisected by a wide

open, solvent-accessible cleft [12] In the Glc-bound

fully-closed state, the two domains are in close

proxim-ity, and the desolvated ligand is engulfed in the cleft

and held in place by extensive hydrogen-bonding

inter-actions with residues in the L-domain (residues N204,

D205, N231, E256 and E290) and the S-domain

(resi-dues T168 and K169) (Fig 1B) In the super-open

state, however, these contact residues are too far apart

to simultaneously interact with Glc Those residues

involved in the first binding of Glc to the super-open

conformation and the subdomain in which the global

conformational transition is initiated have not yet been

experimentally identified

Our point mutation analysis provides experimental

evidence that Glc binds first to residues in the

L-domain and subsequently (after closure) to residues

in the S-domain The mutations N204A, D205A and E256A⁄ K resulted in enzyme forms that did not bind Glc at all as measured by ITF, and they were essen-tially catalytically inactive (Table 2 and supplementary Fig S3), whereas N231A gave a DFeq⁄ F0 response of approximately 6% versus wild-type and no measurable activity By contrast, in the mutations of the S-domain (T168G and K169N), Glc induced a significant fluorescence enhancement ( 60% and 80% versus wild-type), and with reduced affinity (Table 2 and supplementary Fig S3) In the 3D structure of the super-open state, residues 204, 205, 231 and 256 dem-onstrate spatial proximity, with N204, N231 and E256

in the most favourable positions (Cc-Cc⁄ Cd distances

in the range of 5.1–5.4 A˚) and side-chain orientations for a concerted interaction with Glc (Fig 8B) The side-chain of D205, suggested to be the triggering tar-get in Glc binding [12], is, however, in a more unfa-vourable orientation and forms a salt-bridge with R447 in helix 17 (Fig 9) This stabilizing salt-bridge is broken upon Glc-binding (Fig 8C), and the side-chain

of D205 is reorientated [the side-chain dihedral angles change by 117.7 (v1) and 16.5 (v2)] (Table 3), whereas the (Du + Dw) value is changed by only 0.3 Thus, D205 subsequently interacts with Glc In the D205A mutant form, there is no salt bridge, and Ala does not function as a contact residue, offering an explaination for the Glc nonbinding effect of the D205A mutation

Local torsional stresses induced by Glc binding and propagation of the conformational transition From the 3D structures of hGK [12], the overall molecular motion induced by Glc binding is character-ized by a complex shear⁄ sliding and hinge type of movements, as previously described for the structurally related hexokinase I [25,26] The core region (middle and outer layers) of the S-domain is rotated by approximately 99 as a rigid body compared to 12 for hexokinase I Whereas three regions, connecting the L- and S-domains, were assigned as hinge regions in hexokinase I [25] no hinge regions were defined for hGK [12] Using the Hinge Master algorithm for pre-diction of hinge regions in the closed conformation (PDB identity: 1v4s), the highest score was obtained for residues in two of the regions connecting the L- and the S-domains [i.e in connecting region (CR) I (residues 62–73) and in CR II (residues 192–204)] and the crystal structures of the two conformational states demonstrate large changes in the main-chain torsion angles of both regions on Glc binding CR II, which