ORIGINAL RESEARCH PAPER

Detection of glucose-induced conformational changein hexokinase II using fluorescence complementation assay

Eun-Ju Jeong Æ Kyoungsook Park ÆHyou-Arm Joung Æ Chang-Soo Lee ÆDai-Wu Seol Æ Bong Hyun Chung Æ Moonil Kim

Received: 24 October 2006 / Revised: 22 December 2006 / Accepted: 23 December 2006 /Published online: 24 February 2007� Springer Science+Business Media B.V. 2007

Abstract Conformational changes in hexokinase

are induced by its binding to glucose, thus

providing an excellent example of an ‘induced

fit’ model. To observe glucose-induced fluores-

cence restoration in hexokinase II using split-

enhanced, green fluorescent protein (EGFP) in a

process involving the reconstitution of split

EGFP, E. coli cells expressing the chimeric NEG-

FP:HXK:CEGFP recombinant protein were trea-

ted with glucose and visualized via fluorescence

read-outs. The reconstituted EGFP generated a

strong fluorescence upon glucose stimulation of

the bacteria. Moreover, the fluorescence intensity

became stronger with increasing glucose up to

10 mM, with a maximum being observed after

60 min in a time- and concentration-dependent

manner. Conformational changes associated with

glucose-induced fit in human hexokinase II can

thus be monitored successfully in vivo via fluores-

cence reconstitution assays, coupled with a quick

and easy fluorescent read-out protocol.

Keywords Conformational change �Fluorescence complementation assay �Glucose-induced fit � Hexokinase II �Split EGFP

Introduction

Hexokinases not only catalyze the first step in the

glycolytic pathway, but also senses glucose levels

in most organisms (Stulke and Hillen 1999;

Rolland and Sheen 2005). Mammalian genomes

encode for one low-affinity glucokinase (type IV)

and three high-affinity hexokinases (type I, II, and

III) (Wilson 2003). Given the amino acid similar-

ity of the type I–III isozymes between the N- and

C-terminal regions, the evolutionary hypothesis

of mammalian hexokinases that the type I–II

isozymes resulted from duplication and fusion of

a gene encoding an 50-kDa type IV hexokinase

have been suggested. For instance, the type I

hexokinase (hexokinase I) is a 100 kDa protein

with its C-terminal region possessing the catalytic

function, whereas the N-terminal was associated

with a regulatory role (Arora et al 1993).

Many lines of evidence suggest that conforma-

tional changes in hexokinase are induced by the

E.-J. Jeong � K. Park � H.-A. Joung � C.-S. Lee �B. H. Chung (&) � M. KimBioNanotechnology Research Center, KoreaResearch Institute of Bioscience and Biotechnology,P.O. Box 115, Yuseong, Daejeon 305-600,Republic of Koreae-mail: chungbh@kribb.re.kr

M. Kim (&)e-mail: kimm@kribb.re.kr

D.-W. SeolDepartment of Surgery, University of PittsburghSchool of Medicine, Pittsburgh, PA 15261, USA

123

Biotechnol Lett (2007) 29:797–802

DOI 10.1007/s10529-007-9313-x

binding of glucose (Aleshin et al 1998; Kamata

et al 2004), mechanistically caused by the closing of

the cleft between the lobes into which the glucose

binds. This is an excellent example of the ‘induced

fit’ model. Recently, human hexokinase I has been

determined to adopt a dimeric form in the crystal

structure (Mulichak et al 1998; Rosano et al 1999).

Upon the binding of glucose to hexokinase, the

attendant movement and rotation of the small and

large subdomains cause the enzyme to adopt its

closed conformation. From the reported 3-D

structures of hexokinase I, a homology modeling

of the structure of human hexokinase II has been

also suggested, thereby showing that the N-terminal

half from one subunit parallel the C-terminal

domain of the second subunit in the dimeric

conformation (Pastorino and Hoek 2003).

In this study, in order to investigate in detail

the glucose-induced structural transition of hexo-

kinase II on the basis of its antiparallel dimeric

structure model, we employed a split-enhanced

green fluorescent protein (EGFP) complementa-

tion assay. Our results demonstrate that glucose-

induced conformational change in the dimeric

hexokinase II can be successfully monitored

in vivo using fluorescence reconstitution assays,

with a rapid and simple detection method.

Materials and methods

Construction of the pNEGFP:HXK:CEGFP

plasmid

In order to construct the pNEGFP:HXK:CEGFP

plasmid, the partial gene encoding for the EGFP

N-terminal fragment (1–157 amino acids) was

initially amplified with the 5¢ primer (CGG GAT

CCA TGG TGA GCA AGG GCG GAG) and the

3¢ primer (GGA ATT CTC CTC CTC CTC CGC

GCT TCT CGT TGG GGT C) via polymerase

chain reaction (PCR). The 5¢ and 3¢ termini were

designed to harbor the BamHI and EcoRI restric-

tion enzyme cleavage sites, respectively. In order to

generate the CEGFP fragment, the partial gene

encoding for the GFP C-terminal fragment (158–

238 amino acids) was amplified via PCR with the 5¢primer (CCC AAG CTT GGA GGA GGA GGA

GAT CAC ATG GTC CTG CTG) and the 3¢

primer (ATA AGA ATG CGG CCG CCA CAT

TGA TCC TAG CAG AG) using the HindIII and

NotI restriction enzyme cleavage sites, respec-

tively. The human type II hexokinase gene was

then PCR-amplified using the 5¢ primer (CGG

AAT TCA TGA TTG CCT CGC ATC TGC) and

the 3¢ primer (ACT AAG CTT TCG CTG TCC

AGC CTC ACG) with the EcoRI and HindIII

restriction enzyme cleavage sites, respectively. The

PCR products were then purified using a DNA

purification kit (Qiagen), and digested with the

indicated restriction enzymes. The resultant DNA

fragments were then ligated with the pET 28a

vector, using a ligation kit (Takara, Japan)

(pNEGFP:HXK:CEGFP), and cloned into the

pET 28 a vector using the BamHI and NotI sites,

thereby generating the NEGFP:HXK:CEGFP in-

frame fusion. The NEGFP:HXK:CEGFP fusion

gene was verified via DNA sequencing.

Expression of the recombinant

NEGFP:HXK:CEGFP protein

The pNEGFP:HXK:CEGFP construct was trans-

formed into E. coli DH5a. The ampicillin-resistant

colonies were screened via the DNA isolation of

3 ml overnight cultures, followed by restriction

mapping. Plasmid DNA was prepared and purified

using a QIAprep spin miniprep kit (Qiagen,

Germany). The plasmids were then transferred to

the expression host, E. coli BL21 (DE3) (Strata-

gene, CA), then plated onto LB plates. A single

colony from a fresh plate was picked and grown at

37�C in 3 ml Luria–Bertani (LB) broth containing

30 mg kanamycin/ml, to an OD600 of 0.6. This was

then inoculated into 100 ml LB broth containing

kanamycin. The cells were grown at 37�C with

shaking to an OD600 of 0.6. The cells were induced

with 1 mM IPTG, and grown at 18�C for 12 h with

shaking. The cells were then harvested by centri-

fugation (6,000 g for 10 min at 4�C).

SDS-PAGE and Western blotting analysis

The E. coli cells were harvested and suspended in

50 mM Tris/HCl buffer (pH 8.0). After cell

disruption by sonication, the soluble and insoluble

fractions were separated by centrifugation

(10 min at 10,000 g). The total cell lysates and

798 Biotechnol Lett (2007) 29:797–802

123

soluble and insoluble fractions were resolved on

10% (v/v) SDS gel and transferred onto nitrocel-

lulose membranes. Western blotting analysis was

conducted using antibodies against human Type

II hexokinase (Santa Cruz, CA).

In vivo visualization of glucose-induced

conformational changes in hexokinase II

For the detection of fluorescence, 3 ll of E. coli

cells expressing NEGFP:HXK:CEGFP protein,

either with or without IPTG treatment, were

spread onto a glass slide stuck to sticky tape, with

wells approx 3 mm diam and 100 lm deep to

compartmentalize to E. coli culture. The samples

were then treated with 10 mM glucose for 1 h at

room temperature. After 1 h, the fluorescence

images of the E. coli cells generating the chimeric

NEGFP:HXK:CEGFP protein were acquired

using a GenePix 4200A 488 nm laser, a micro-

array scanner (Axon Inc., CA).

Results and discussion

The rationale for the split fluorescent protein-

based monitoring of glucose-induced

conformational change in hexokinase II

Several lines of evidence appear to suggest that

the generation of fluorescence from partial GFP

fragments, which is attributed to reconstituted

GFP, might also be used to evaluate protein–

protein interactions (Ozawa et al 2001). Based on

this split GFP technology, we previously

described a split GFP reconstitution assay which

could be utilized for the monitoring of conform-

ationally-altered proteins (Jeong et al 2006). In

that approach, we elected to utilize the maltose

binding protein (MBP) as a model protein for

conformational change. The split GFP method is

principally predicated on the restoration of fluo-

rescence effected when split fluorescent proteins

are brought back into contact with each other.

Upon stimulation with maltose, the two lobes of

MBP are brought closer together, which results in

the complementation of the two split GFP

fragments attached to the N- and C-termini of

MBP. This split GFP system is contingent on the

reconstitution of the NGFP and CGFP fragments,

which results in recovery of fluorescence.

In accordance with the same strategy, we have

attempted herein to evaluate the possibility of

monitoring the conformational alterations atten-

dant to the induced fit of human hexokinase II in

response to glucose stimulation, via the split

fluorescence protocol. This glucose-induced con-

formational alteration is generally referred to as

an ‘embracing mechanism’, which may require

hexokinase to embrace or surround its substrate,

such that the only way in which the substrate can

enter and leave the active site is for the enzyme to

open and close (Blow and Steitz 1970). In

particular, we employed a GFP-modified version,

with the alterations Ser 65–Thr and Phe 64–Leu,

referred to as enhanced green fluorescent protein

(EGFP), as a reporter. This selection was due to

the fact that EGFP produces a fluorescence 35

times more intense than that of wild-type GFP

(Kain and Ma 1999). The split EGFP strategy for

the detection of glucose-induced structural tran-

sition from open form to closed form in hexoki-

nase was based on the introduction of EGFP, as a

visible reporter molecule, which splits into two

non-fluorescent fragments, which are linked to

the N- and C-termini of hexokinase, respectively

(Fig. 1). As shown in Fig. 1, we hypothesized that

the exogamous chromophores would move closer

together when glucose bound to the dimeric

hexokinase II, thereby resulting in the comple-

mentation of the two exogamous split EGFP

fragments. In order to evaluate the veracity of this

hypothesis, the first step was to prepare the fusion

construct of two non-fluorescent fragments,

NEGFP (1–157 amino acids) and CEGFP (158–

238 amino acids), each of which are linked

genetically to the N- and C-terminal halves of

hexokinase II as a single structural molecule, as

described in Materials and Methods.

Expression of the chimeric

NEGFP:HXK:CEGFP protein

In order to monitor the glucose-induced confor-

mational alterations in hexokinase via the split

EGFP method, we initially constructed the

pNEGFP:HXK:CEGFP plasmid to express the

chimeric NEGFP:HXK:CEGFP recombinant pro-

Biotechnol Lett (2007) 29:797–802 799

123

tein (Fig. 2A). As a consequence of the construc-

tion of the chimeric NEGFP:HXK:CEGFP gene,

NEGFP was then fused to the N-terminal end of

hexokinase II, and CEGFP was attached to its

C-terminal end (Fig. 2A). Glucose binding would

then bring the exogamous N- and C-termini of

hexokinase II in a dimeric structure into closer

physical proximity, thereby inducing a subsequent

change in the intensity of fluorescence. For the

expression of the recombinant NEGFP:HXK:CE-

GFP protein, the pNEGFP:HXK:CEGFP plas-

mids were then transformed into the expression

host, E. coli BL21 (DE3). As shown in Fig. 2B,

after the induction of IPTG, we noted an obvious

extra band, which corresponded to a molecular

mass of approximately 126 kDa, consistent with

the expected molecular weight of NEG-

FP:HXK:CEGFP, as compared to the bands

observed with the E. coli strain harboring the

pET 28a control plasmid. The expressed protein

was then validated via Western blotting using the

polyclonal antibody of hexokinase II (Fig. 2C).

Then, E. coli cells expressing the chimeric

NEGFP:HXK:CEGFP recombinant protein were

treated with glucose, and visualized further using

a fluorescence read-out for the detection of

structural alterations associated with the glu-

cose-induced fit in hexokinase II.

glucose

Closed conformation

Open conformation

C2

N2

L

N1

L

C1

NEGFP NEGFP

CEGFP

Glycinelinker(GGGG)

C2

N2

L

C1

N1

L

CEGFP

EGFP EGFP

Fig. 1 Schematic diagram of glucose-induced conforma-tional changes in hexokinase II via split EGFP comple-mentation assay. The N- and C-terminal halves of the splitEGFP were fused to the N- and C-termini of the dimerichexokinase II molecule. Upon glucose binding, theexogamous chromophores moved closer together, result-ing in the recovery of EGFP fluorescence. Peptide glycinelinker sequences are shown at the top. N1, N-terminaldomain 1; N2, N-terminal domain 2; C1, C-terminaldomain 1; C2, C-terminal domain 2; L, Linker domain

lacIkan

ori

f1 origin

pET28a-pET28a-NEGFP:HXK:CEGFPNEGFP:HXK:CEGFP

6×His

NEGFP

BamHI NotIGlycinelinker (GGGG)

CEGFPHexokinase

75 -

kDa M 1 2

100 -

150 -1 2

A

B C

Fig. 2 Construction of the pNEGFP:HXK:CEGFP plas-mid and expression of chimeric NEGFP:HXK:CEGFPrecombinant protein. (A) The map of recombinant plasmidpNEGFP:HXK:CEGFP for expression of NEGFP:HXK:CEGFP. kan, kanamycin resistant gene; ori, origin ofreplication; lac I, lactose repressor gene. (B) SDS-PAGEanalysis of recombinant NEGFP:HXK:CEGFP protein. M,protein marker; 1, total cell lysates of non-induced BL21(DE3); 2, total cell lysates of IPTG-induced BL21 (DE3).The arrow indicates the expressed NEGFP:HXK:CEGFPprotein. (C) This identity was confirmed via Westernblotting using the polyclonal antibody of hexokinase II

800 Biotechnol Lett (2007) 29:797–802

123

Fluorescence complementation assay for the

monitoring of conformational transitions in

hexokinase II in vivo

Visualization of glucose-induced conformational

changes in hexokinase II using fluorescence com-

plementation assay (see methods), is shown in

Fig. 3. Cells harboring the pNEGFP:HXK:CE-

GFP plasmid to express the chimeric protein

emitted a strong green fluorescence upon addition

of 10 mM glucose. Cells to which glucose had not

been added manifested no detectable fluores-

cence. In order to confirm whether the fluores-

cence observed in the glucose-treated cells was,

indeed, arising from the addition of glucose, the

transformed E. coli cells were treated with

10 mM maltose as a negative control. As is

shown in Fig. 3, although some fluorescence was

observed in response to maltose, this was a

minimally detectable signal (Fig. 3). This raises

the question as to whether perhaps a small

amount of glucose exists in the maltose solution

as an impurity.

Next, we attempted to assess glucose-depen-

dent alterations in the fluorescence intensity of

NEGFP:HXK:CEGFP-expressing E. coli cells

using a GenePix 4200A laser scanner (Axon

Inc., CA). As is shown in Fig. 4A, glucose-

induced conformational transitions in hexokinase

II caused the production of green fluorescent

protein, in a concentration-dependent fashion.

The observed increases in fluorescence intensity

appear to be the result of glucose-induced con-

formational changes associated with the induced

fit in hexokinase II, thereby indicating that the

complemented EGFP refolded correctly for the

formation of the EGFP fluorophore. As is shown

in Fig. 4A, a minimal increase in fluorescence

intensity was detected after treatment with 1 mM

glucose as the onset of the conformational change

associated with the glucose-induced fit in hexoki-

nase II. This indicated that low concentrations of

glucose quantitatively induce small alterations.

This increase was sustained with treatments of up

_+_

_

Non-induced

GlucoseInduced Maltose0.0

0.2

0.4

0.6

0.8

1.0

Rel

ativ

e fl

uo

resc

ence

in

ten

sity

(R

IU)

Fig. 3 In vivo visualization of glucose-induced conforma-tional changes in hexokinase II. For the detection offluorescence images, E. coli cells expressing chimericNGFP:HXK:CGFP protein, either with or without IPTGaddition, were initially spread onto glass slides stuck tosticky tape with 3 mm diameter wells in order to partitionthe bacterial cultures. Then, the fluorescence images of theNEGFP:HXK:CEGFP-expressing E. coli cells treatedwith glucose (or maltose as a control) were acquired witha GenePix 4200A laser scanner (Axon Inc., CA)

0 1 2 10Glucose

(mM)

0 30 60 90 120Time

(min)

0

1 10Glucose concentration (mM)

00.0

0.2

0.4

0.6

0.8

1.0

Rel

ativ

e fl

uo

resc

ence

in

ten

sity

(R

IU)

0.0

0.2

0.4

0.6

0.8

1.0

Rel

ativ

e fl

uo

resc

ence

in

ten

sity

(R

IU)

30 60 90 120Time (min)

5

2 5

A

B

Fig. 4 Structural transition associated with the glucose-induced fit in hexokinase II. Glucose-induced conforma-tional changes in hexokinase II were assessed viafluorescence analyses with a GenePix 4200A frombacterial cells generating NEGFP:HXK:CEGFP proteinsat differing (A) concentrations and (B) incubation timesusing glucose as a substrate

Biotechnol Lett (2007) 29:797–802 801

123

to 10 mM, at which point the solution is

completely saturated (Fig. 4A). The maximum

fluorescence was 5.2 times higher than that

observed in the absence of glucose which indi-

cates that glucose did, indeed, elicit a conforma-

tional change in hexokinase II, resulting in a

concomitant increase in the fluorescence inten-

sity. This demonstrates that the optimal glucose

concentration range for the induction of confor-

mational changes in hexokinase II is about

10 mM under our experimental conditions. In

order to detect the time-dependent changes

occurring in the fluorescence intensity, we treated

NEGFP:HXK:CEGFP-expressing bacterial cells

with glucose until the optimal time for glucose-

induced conformational change in hexokinase II

could be determined, at which time the fluores-

cence intensity would be maximally saturated. As

is shown in Fig. 4B, in the presence of 10 mM

glucose, the fluorescence intensity was 3.5 times

higher than that measured in initiation time point

of glucose treatment. After 60 min, as the tem-

poral threshold with regard to total image satu-

ration, the fluorescence intensity reached its

maximum, thereby indicating that the optimal

incubation time for glucose was somewhere

within the 60–90 min range.

In conclusion, we have successfully monitored

the glucose-mediated structural transition, which

is also referred to as the ‘glucose-induced fit’, in

human hexokinase II, via a split EGFP comple-

mentation technique. Upon binding to glucose,

hexokinase II manifests a structural change that

moves the exogamous chromophores closer

together. As a consequence of this glucose-

induced conformational change, the split EGFP

is reconstituted, and recovers its fluorescent

properties. The resultant activity of the recon-

stituted EGFP can be rapidly and directly

measured via fluorescence intensity measure-

ments. Collectively, the results of this study

showed that the conformational change of

dimeric hexokinase II associated with the glu-

cose-induced fit can be rapidly and directly

monitored in vivo via fluorescence complemen-

tation assays, predicated on the reconstitution of

split EGFP.

Acknowledgements This research was supported bygrants from the Nano/Bio Science & Technology Program(MOST, Korea), and the KRIBB Initiative ResearchProgram (KRIBB, Korea).

References

Aleshin AE, Zeng C, Bourenkov GP, Bartunik HD,Fromm HJ, Honzatko RB (1998) The mechanism ofregulation of hexokinase: new insights from thecrystal structure of recombinant human brainhexokinase complexed with glucose and glucose6-phosphate. Structure 6:39–50

Arora KK, Filburn CR, Pedersen PL (1993) Structure/function relationships in hexokinase. Site-directedmutational analyses and characterization of overex-pressed fragments implicate different functions for theN- and C-terminal halves of the enzyme. J Biol Chem268:18259–18266

Blow DM, Steitz TA (1970) X-ray diffraction studies ofenzymes. Annu Rev Biochem 39:63–100

Jeong J, Kim SK, Ahn J, Park K, Jeong EJ, Kim M, ChungBH (2006) Monitoring of conformational change inmaltose binding protein using split green fluorescentprotein. Biochem Biophys Res Commun 339:647–651

Kain SR, Ma JT (1999) Early detection of apoptosis withannexin V-enhanced green fluorescent protein. Meth-ods Enzymol 302:38–43

Kamata K, Mitsuya M, Nishimura T, Eiki J, Nagata Y(2004) Structural basis for allosteric regulation of themonomeric allosteric enzyme human glucokinase.Structure 12:429–438

Mulichak AM, Wilson JE, Padmanabhan K, Garavito RM(1998) The structure of mammalian hexokinase-1. NatStruct Biol 5:555–560

Ozawa T, Takeuchi TM, Kaihara A, Sato M, Umezawa Y(2001) Protein splicing-based reconstitution of splitgreen fluorescent protein for monitoring protein–protein interactions in bacteria: improved sensitivityand reduced screening time. Anal Chem 73:5866–5874

Pastorino JG, Hoek JB (2003) Hexokinase II: the integra-tion of energy metabolism and control of apoptosis.Curr Med Chem 10:1535–1551

Rolland F, Sheen J (2005) Sugar sensing and signalingnetworks in plants. Biochem Soc Trans 33:269271

Rosano C, Sabini E, Rizzi M, Deriu D, Murshudov G,Bianchi M, Serafini G, Magnani M, Bolognesi M(1999) Binding of non-catalytic ATP to humanhexokinase I highlights the structural componentsfor enzyme-membrane association control. Structure7:1427–1437

Stulke J, Hillen W (1999) Carbon catabolite repression inbacteria. Curr Opin Microbiol 2:195–201

Wilson JE (2003) Isozymes of mammalian hexokinase:structure, subcellular localization and metabolic func-tion. J Exp Biol 206:2049–2057

802 Biotechnol Lett (2007) 29:797–802

123