Báo cáo khoa học: Importance of the gating segment in the substrate-recognition loop of pyranose 2-oxidase pptx

Although the position and orientation of the slow substrate 2-deoxy-2-fluoro-glucose when bound in the active site of pyra-nose 2-oxidase variants is identical to that observed earlier, t

substrate-recognition loop of pyranose 2-oxidase

Oliver Spadiut1,*, Tien-Chye Tan1,2,*, Ines Pisanelli3, Dietmar Haltrich3and Christina Divne1,2

1 KTH Biotechnology, Royal Institute of Technology, Stockholm, Sweden

2 Department of Medical Biochemistry and Biophysics, Karolinska Institutet, Stockholm, Sweden

3 Food Biotechnology Laboratory, Department of Food Science and Technology, BOKU – University of Natural Resources and Applied Life Sciences, Vienna, Austria

Keywords

active-site loop; alanine-scanning

mutagenesis; crystal structure; pyranose

2-oxidase; site-saturation mutagenesis

Correspondence

C Divne, KTH Biotechnology, Royal Institute

of Technology, Albanova,

Roslagstullsbacken 21, SE-106 91

Stockholm, Sweden

Fax: +46 8 5537 8468

Tel: +46 8 5537 8296

E-mail: divne@biotech.kth.se

or D Haltrich; Food Biotechnology

Laboratory, Department of Food Science

and Technology, BOKU – University of

Natural Resources and Applied Life

Sciences, A-1190 Vienna, Austria

Fax: +43 1 47654 6251

Tel: +43 1 47654 6140

E-mail: dietmar.haltrich@boku.ac.at

*These authors contributed equally to this

work

Database

The atomic coordinates and structure

factors for the models are available in the

Protein Data Bank database under the

accession numbers 3K4J (H450Q),

3K4K (F454N), 3K4L (F454N 2FG ),

3K4M (Y456W 2FG ) and 3K4N

(F454A ⁄ S455A ⁄ Y456A)

(Received 22 February 2010, revised 2 May

2010, accepted 6 May 2010)

doi:10.1111/j.1742-4658.2010.07705.x

Pyranose 2-oxidase from Trametes multicolor is a 270 kDa homotetrameric enzyme that participates in lignocellulose degradation by wood-rotting fungi and oxidizes a variety of aldopyranoses present in lignocellulose to 2-ketoaldoses The active site in pyranose 2-oxidase is gated by a highly conserved, conformationally degenerate loop (residues 450–461), with a conformer ensemble that can accommodate efficient binding of both elec-tron-donor substrate (sugar) and electron-acceptor substrate (oxygen or quinone compounds) relevant to the sequential reductive and oxidative half-reactions, respectively To investigate the importance of individual resi-dues in this loop, a systematic mutagenesis approach was used, including alanine-scanning, site-saturation and deletion mutagenesis, and selected variants were characterized by biochemical and crystal-structure analyses

We show that the gating segment (454FSY456) of this loop is particularly important for substrate specificity, discrimination of sugar substrates, turn-over half-life and resistance to thermal unfolding, and that three conserved residues (Asp452, Phe454 and Tyr456) are essentially intolerant to substitu-tion We furthermore propose that the gating segment is of specific impor-tance for the oxidative half-reaction of pyranose 2-oxidase when oxygen is the electron acceptor Although the position and orientation of the slow substrate 2-deoxy-2-fluoro-glucose when bound in the active site of pyra-nose 2-oxidase variants is identical to that observed earlier, the substrate-recognition loop in F454N and Y456W displays a high degree of confor-mational disorder The present study also lends support to the hypothesis that 1,4-benzoquinone is a physiologically relevant alternative electron acceptor in the oxidative half-reaction

Abbreviations

ABTS, 2,2¢-azinobis(3-ethylbenz-thiazolinesulfonic acid; 2FG, 2-deoxy-2-fluoro- D -glucose; BQ, 1,4-benzoquinone; Fc+, ferrocenium ion; P2O, pyranose 2-oxidase; PDB, Protein Data Bank; TLS, translation, libration, screw-rotation.

Pyranose 2-oxidase (P2O; pyranose:oxygen

2-oxidore-ductase; EC 1.1.3.10) from Trametes multicolor

(syno-nym Trametes ochracea) is a 270 kDa, 8a-(N3)-histidyl

flavinylated, homotetrameric flavoprotein oxidase

found in the fungal hyphal periplasmic space [1–5]

P2O has been suggested to perform a dual function

during lignin degradation by wood-rotting fungi, by

providing H2O2 for ligninolytic enzymes [6,7] and

reducing quinones as part of the extracellular quinone

redox cycling machinery [8] In some white-rot

basid-iomycetes, P2O participates in a secondary

meta-bolic pathway in which d-glucose is first converted

to 2-keto-d-glucose (2-arabino-hexos-2-ulose;

d-gluco-sone; (Scheme 1) and then further oxidized by

aldos-2-ulose dehydratase to the b-pyrone antibiotic

cortalcerone [9,10]

During the reductive half-reaction, P2O catalyses

the oxidation at the C2 of several aldopyranoses

(Scheme 1) present in lignocellulose to the

correspond-ing 2-ketoaldoses, accompanied by electron transfer to

FAD, yielding the reduced flavin, FADH2 [11] In the

oxidative half-reaction, the cofactor is re-oxidized by

an electron acceptor (e.g O2 or quinone compounds),

producing either H2O2 or reduced quinone [1,8] The

reaction mechanism is of the Ping Pong Bi Bi type [12],

which is common in flavoprotein oxidoreductases

[13,14] In analogy with the hydride-transfer

mecha-nism proposed for flavoproteins in general [15,16], and

in particular for the reductive half-reaction of Phanero-chaete chrysosporium cellobiose dehydrogenase [17,18], which is the closest relative of P2O among glucose-methanol-choline family members, His548 in P2O is suitably positioned to act as a general base to deproto-nate the equatorial substrate 2-OH group, with support from an O2–Asn593 Nd2 hydrogen bond, accompanied

by transfer of the axial 2-hydrogen atom as a hydride from C2 to the flavin N5 atom [5,19] The two half-reactions catalysed by P2O (i.e oxidation of electron donor and reduction of electron acceptor) have differ-ent prerequisites with respect to the local chemical and structural environment Our crystal structures of a closed state of wild-type P2O with the competitive inhibitor acetate bound (Fig 1A) [5], and an open state with a slow sugar substrate bound (Fig 1B) [19], have shown that a conserved substrate-recognition loop

450HRDAFSYGAVQQ461is highly dynamic and offers

a conformational gating mechanism to P2O

In T multicolor P2O, b-d-glucose is oxidized regi-oselectively at C2 without traces of 3-keto product For the related Peniophora gigantea P2O, however, oxidation at C3 can take place as a side reaction when the glucose 2-OH group is either absent (e.g 2-deoxy-d-glucose) or modified (e.g 2-keto-d-glucose) [11] This was also observed for the T multicolor P2O

Scheme 1 Substrates (electron acceptors

and donors) discussed in the text.

mutant H167A with the slow substrate

2-deoxy-2-flu-oro-d-glucose (2FG) (Scheme 1, Fig 1B) [19] In the

H167A variant, the flavinylation ligand His167 has

been mutated to an alanine and the FAD is

noncova-lently bound It was also shown that the flavin

reduc-tion was significantly slower than in the wild-type,

which enabled structure determination of an ordered

complex with 2FG [19] The ability of the enzyme to

oxidize some substrates at both C2 and C3 requires

that the sugar can bind in two productive binding

modes On the basis of the in silico modelling of

glu-cose, we have suggested that these two binding modes

are related by a 180 rotation about an axis running

through two points in the pyranose ring (one point

between C5 and O5, and the other between C2 and

C3), producing almost isosteric substrate-binding

modes [19] Although d-galactose (Scheme 1) is a

relatively poor substrate for wild-type P2O (6%

relative activity compared to d-glucose) [8], it is

struc-turally very similar to d-glucose, differing only by the

axial C4 hydroxyl group (equatorial in glucose), and

thus, the C4 position must be important for the

substrate selectivity mechanism in P2O We have

shown that d-galactose cannot be well accommodated

in the P2O active site as a result of possible steric hindrance between the axial O4 and the side chain of Thr169 [20,21] Engineering of P2O for improved

d-galactose turnover is also highly relevant for indus-trial purposes because C2 oxidation of d-galactose yields 2-keto-d-galactose, which can be further reduced at C1 to give the low caloric sweetener

d-tagatose [22]

In the present study, we report the results obtained from a systematic mutagenesis approach that aimed to investigate the importance of key amino acids in the substrate-recognition loop of T multicolor P2O by means of alanine-scanning, site-saturation and deletion mutagenesis, as well as the characterization of mutants

by biochemical and crystal-structure analysis We dis-cuss the catalytic competence and stability of the mutants and their preference for electron donors and electron acceptors in light of the steady-state kinetics, stability and structural data presented The finding of the present study demonstrate that the gating segment

of the substrate-recognition loop is of particular importance for P2O function and substrate specificity,

as well as for catalytic and thermal stability at elevated temperatures

Fig 1 Overall P2O subunit structure of the closed and open state Subunit structure of (A) wild-type P2O in complex with acetate (PDB code 1TT0) with the substrate-recognition loop closed [5] and (B) the P2O variant H167A in complex with 2FG (PDB code 2IGO) where the loop is completely open [19] The H167A mutant has the flavinylation ligand His167mutated to alanine, and the FAD is noncovalently bound The Rossmann domain is coloured pink, and the substrate-binding domain is shown in light blue Loops are shown in beige and the FAD cofactor is shown in yellow The ‘head domain’, which is not present in related glucose-methanol-choline members, is shown in light green The substrate-recognition loop (residues 450–461) is highlighted as a purple coil, with the side chains that form the gating segment (i.e Phe 454 , Ser 455 and Tyr 456 ) shown as stick representations The ligands (coloured bright green), acetate in the closed form and 2FG in the open state, are bound at the re face of the isoalloxazine ring.

Site-saturation, alanine-scanning and deletion

mutagenesis

Site-saturation mutagenesis was employed to generate

a library of T multicolor P2O variants targeting the

substrate-recognition loop, combined with

high-throughput screening of mutants in a 96-well plate

format A procedure for generating and screening P2O

variants targeting position 450 has been described

previously [23] We used this approach to generate

enzyme variants by targeting the positions 452, 454

and 456 The P2O mutant library covered > 95% of

all possible combinations of variants at the selected

positions (His450, Asp452, Phe454and Tyr456), and

high-throughput screening included 360 colonies for each

position, which were tested for activity towards two

electron-donor substrates: d-glucose and d-galactose

The statistics show that a large number of mutations

at these positions result in inactive P2O variants: 60%,

44%, 56% and 48% inactive variants for His450,

Asp452, Phe454 and Tyr456, respectively (site-saturation

mutagenesis at position 450 has been reported

sepa-rately) [23] This demonstrates the importance of the

targeted loop residues for enzymatic activity and⁄ or

proper folding and stability In addition,

alanine-scan-ning and site-directed mutagenesis were performed

A set of mutants that displayed interesting

characteris-tics with respect to d-glucose or d-galactose turnover

were selected and subjected to more detailed analysis

These included the single-substitution variants H450Q,

F454P, F454N and Y456W; two multiple-alanine

mutants targeting the gating segment (F454A⁄ Y456A

and F454A⁄ S455A ⁄ Y456A); and one deletion mutant

lacking the FSY segment (D454–456) Typical yields of

mutant P2Os were in the range 15–30 mgÆL)1 culture

medium, although the D454–456 and alanine mutants were expressed at significant lower levels (0.2– 0.8 mgÆL)1) and overexpression was accompanied by

an increase in inclusion-body formation

Kinetic characterization of mutants Apparent kinetic constants were determined for the loop variants as a preliminary means to assess the effect

of these mutations on the P2O-catalysed reaction; accordingly, one of the substrates of P2O (either the electron acceptor or donor) was fixed, with the other one being varied, and the obtained data were fit to the Michaelis–Menten equation for a single substrate For all variants, the apparent kcatapp values with d-glucose and O2(fixed at a concentration of 256 lm, air satura-tion) as substrates, kcat[Glc⁄ O2], decreased dramatically compared to the wild-type enzyme, and Km[Glc]values increased by a factor in the range 1.9–3.3 (Table 1), resulting in decreased catalytic efficiency constants (kcat[Glc⁄ O2]⁄ Km[Glc]) The only exceptions are Y456W, a conservative mutation with one bulky hydrophobic amino acid replacing another one, which shows a simi-lar kcat[Glc⁄ O2], albeit with a doubled Km[Glc] and an associated three-fold decrease in kcat[Glc⁄ O2]⁄ Km[Glc], as well as H450Q, where kcat[Glc⁄ O2] was decreased by a factor of 2 Consistently, turnover numbers for the elec-tron donor⁄ acceptor substrate pair d-galactose ⁄ O2 for the variants H450Q and Y456W were comparable to that of the wild-type Other variants, in particular the Ala mutants and D454-456, showed three- to 12-fold lower kcat[Gal⁄ O2], and up to six-fold elevated Km[Gal] values (Table 1)

Mutations in the gating segment affected sugar substrate specificity significantly The wild-type enzyme displays a clear preference for d-glucose over d-galac-tose, as indicated by a selectivity ratio of  160, with

Table 1 Apparent steady-state kinetic constants of T multicolor P2O wild-type and mutants with D -glucose (0.1–50 m M ) or D -galactose (0.1–200 m M ) as electron donor and O2(air) under saturation as electron acceptor.

Variant

Km[Glc]

(m M )

kcat[Glc ⁄ O 2 ] (s)1)

kcat⁄ K m [Glc]

( M )1Æs)1)

Km[Gal]

(m M )

kcat[Gal ⁄ O 2 ] (s)1)

kcat[Gal ⁄ O 2 ] ⁄ K m [Gal] ( M )1Æs)1)

F454A ⁄ S455A ⁄ Y456A 2.1 ± 0.3 0.20 ± 0.01 0.094 · 10 3

13 ± 3 0.14 ± 0.01 0.011 · 10 3

the ratio of the specificity constants [24] for the

two substrates being [(kcat[Glc⁄ O2]⁄ Km[Glc])⁄ (kcat[Gal⁄ O2]⁄

Km[Gal])] Most of the mutations in the gating segment

reduced the selectivity ratio The lowest ratio [(kcat[Glc ⁄

O2]⁄ Km[Glc])⁄ (kcat[Gal ⁄ O2]⁄ Km[Gal])] of 8.4 was observed

for the triple-alanine mutant, which, however, still

retains some preference for d-glucose over d-galactose

By contrast, the replacements F454N or Y456W show

increased [(kcat[Glc⁄ O2]⁄ Km[Glc])⁄ (kcat[Gal ⁄ O2]⁄ Km[Gal])] values

of 180 and 290, respectively This effect on substrate

selectivity is even more pronounced when considering

the disaccharide melibiose (D-Gal-a(1fi 6)-D-Glc;

Scheme 1) Melibiose is a rather poor substrate for

P2O, mainly because of its very high Kapp

m value (Km[Mel]= 1530 mm and kcat= 7.6 for wild-type;

Table 2), and the selectivity ratio [(kcat[Glc⁄ O2]⁄ Km[Glc])⁄

(kcat[Mel⁄ O2]⁄ Km[Mel])] for wild-type is  8700,

indicat-ing very strong discrimination of P2O in favour of the

monosaccharide substrate d-glucose over the

disaccha-ride melibiose, with oxygen as acceptor Again,

muta-tions in the gating segment of the substrate loop

reduced the substrate selectivity for all of the variants

to values in the range 12–1630 The lowest [(kcat[Glc⁄ O2]⁄ Km[Glc])⁄ (kcat[Mel⁄ O2]⁄ Km[Mel])] ratios of 93 and 12 were found for the F454P and triple-Ala mutant, respectively

Furthermore, apparent kinetic constants were deter-mined for the reduction of the two-electron acceptor 1,4-benzoquinone (BQ) to hydroquinone (Scheme 1, Table 3), and for the reduction of the 1-electron accep-tor ferrocenium (Fc+) to ferrocene (Scheme 1, Table 4), with either d-glucose or d-galactose at satu-rating concentrations Variant Y456W is characterized

by improved BQ and Fc+ binding, as indicated by lower Kapp

m values, and increased kappcat values, resulting

in an  2.5-fold higher kcat[BQ⁄ Glc]⁄ Km[BQ]and kcat[BQ⁄ Gal]⁄ Km[BQ] relative to wild-type The triple-Ala and loop-deletion mutants also show considerably lower

Km[BQ] with Glc as electron donor, although with an associated decrease in kcat[BQ ⁄ Glc]values Similar results were obtained for the reduction of BQ with d-galac-tose, and for Fc+with d-glucose or d-galactose as sat-urating electron-donor substrates In addition, H450Q, F454N and Y456W show improved kinetic properties

Table 2 Apparent steady-state kinetic constants of T multicolor P2O wild-type and mutants with melibiose (5.0–500 m M ) as electron donor and O2(air) under saturation as electron acceptor.

Variant

Km[Mel] (m M ) kcat[Mel ⁄ O 2 ] (s)1)

k cat [Mel ⁄ O 2 ] ⁄ K m [Mel]

( M )1Æs)1)

(k cat [Glc ⁄ O 2 ] ⁄ K m [Glc]) ⁄ (kcat[Mel ⁄ O 2 ] ⁄ K m [Mel])

Table 3 Apparent steady-state kinetic constants of T multicolor P2O wild-type and mutants with BQ (0.01–1.5 m M ) as electron acceptor and D -glucose or D -galactose at saturation (100 m M each) as electron donor.

Variant

Km[BQ]

(l M )

kcat[BQ ⁄ Glc]

(s)1)

kcat[BQ ⁄ Glc] ⁄ K m [BQ]

( M )1Æs)1)

Km[BQ]

(l M )

kcat[BQ ⁄ Gal]

(s)1)

kcat[BQ ⁄ Gal] ⁄ K m [BQ] ( M )1Æs)1)

5.2 ± 0.8 2.7 ± 0.1 0.52 · 10 6

F454A ⁄ S455A ⁄ Y456A 29 ± 11 15 ± 1 0.52 · 10 6

8.9 ± 1.0 1.2 ± 0 0.13 · 10 6

for Fc+ (lower Km and higher kcat values) both with

glucose and galactose as saturating substrate compared

to wild-type Both BQ and Fc+are considerably larger

molecules compared to O2, and, most likely,

shorten-ing the loop or introducshorten-ing smaller side chains

pro-motes the reaction with the larger electron-acceptor

substrates

Heat inactivation, pH optima, UV-visible spectra

ThermoFAD analysis

The half-life of P2O activity (i.e the time during which

the enzyme remains active) was measured for wild-type

and mutants at constant pH (6.5) at 60 or 70C The

inactivation constant, kin, and the half-life of activity,

s1 ⁄ 2, were determined (Table 5) On the basis of

[ln(residual activity) versus time] plots, all mutants

show first-order inactivation kinetics (Fig 2) H450Q

shows pronounced destabilization (Fig 2A), whereas

the other variants show similar or improved stability

(Fig 2A,B) The substitutions Phe454fi Asn, and

Tyr456fi Trp result in a 29- and 34-fold increase in

s1 ⁄ 2values at 60C, respectively Interestingly, the ala-nine-substituted variants are also more stable at 60C, with a 12-fold and 23-fold increase in s1 ⁄ 2 for F454A⁄ Y456A and F454A⁄ S455A ⁄ Y456A, respec-tively Some stabilization is also seen for D454–456 (four- to five-fold increase) at 60C A similar trend

of increased heat inactivation half-life was observed for all variants at 70C (Fig 2C)

All variants show pH optima at pH 6.5 (data not shown), suggesting that the altered kinetics is not inti-mately correlated with changes in pH profile All enzymes also display typical flavoprotein UV-visible spectra with absorption maxima kmax at 345 and

456 nm (data not shown), and reduction of the enzymes with d-glucose and sodium dithionite in the absence of oxygen resulted in the disappearance of the absorption peak at 456 nm that was expected for the fully reduced state FAD was not released upon trichloroacetic acid treatment, demonstrating that, despite extensive mutagenesis in the vicinity of the FAD-binding pocket, the mutants remain properly fla-vinylated (not shown) The thermal stability was inves-tigated using the ThermoFAD technique to derive thermal unfolding transition values (Tm) The Tm val-ues are summarized in Table 6 Of the variants ana-lyzed, all but two mutants show slightly decreased Tm values (1–3C) Y456W and F454A ⁄ Y456A show improved Tmvalues by 5.5 and 1.5C, respectively

Overall monomer structure of loop variants The mutants analyzed structurally include the unbound forms of H450Q, F454N and F454A⁄ S455A ⁄ Y456A, and F454N or Y456W with bound 2FG Data for the triple-Ala mutant were obtained to medium-low resolu-tion (2.75 A˚), and the model is included mainly to evalu-ate the backbone conformation of the substrevalu-ate loop

Table 4 Apparent steady-state kinetic constants of T multicolor P2O wild-type and mutants with Fc + (0.005–1.5 m M ) as electron acceptor and D -glucose or D -galactose under saturation (100 m M each) as electron donor.

Variant

Km[Fc + ] (l M )

kcat[Fc + ⁄ Glc]

(s)1)

kcat[Fc + ⁄ Glc] ⁄ K m [Fc + ] ( M )1Æs)1)

Km[Fc + ] (l M )

kcat[Fc + ⁄ Gal]

(s)1)

kcat[Fc + ⁄ Gal] ⁄ K m [Fc + ] ( M )1Æs)1)

16 ± 7 2.9 ± 0.3 0.18 · 10 6

Table 5 Heat inactivation half-life of T multicolor P2O wild-type

and mutants at 60 and 70 C k in , inactivation constant; s1⁄ 2,

half-life; ND, not determined.

Variant

kin(min)1)

(min)

kin (min)1)

(min)

F454A ⁄ S455A ⁄

Y456A

All structures show satisfactory model statistics

(Table 7), and are very similar overall to the previously

determined crystal structures of T multicolor P2O

[5,19,25,26], demonstrating that the mutations do

not result in significant structural changes beyond the

targeted region In the unbound form of variants H450Q, F454N and F454A⁄ S455A ⁄ Y456A, the sub-strate loop is in the open conformation Therefore, structural comparisons are made with H167A rather than the wild-type because H167A has the substrate-recognition loop in the fully open conformation [19] This is in agreement with our earlier observation that the substrate loop tends to be in the open state either when the active site is unoccupied, or when sugar (elec-tron donor) is bound However, in the former case, we typically observe varying degrees of disorder of the loop, whereas the loop becomes well defined when sugar substrate is bound [19,25,26] The same is observed in the present study where the unbound variants display varying degrees of fluctuation of the open state, which is manifested as weak but interpretable electron density indicative of multiple conformers Furthermore, the two 2FG-bound F454N and Y456W models show the open state of the substrate-recognition loop

Structure of the Y456W-2FG complex Despite the larger tryptophan side chain, the substrate loop in Y456W2FG (Fig 3A) assumes the same open state as observed in H167A2FG [Protein Data Bank (PDB) code 2IGO] [19] However, the electron density

is weak for the Trp456indole ring, as well as for the sub-strate loop, suggesting that the larger side chain induces local disorder and suboptimal side-chain packing Despite this local disorder, the 2FG molecule is orderly bound in an orientation identical to that in H167A2FG, corresponding to the C3-oxidation binding mode (Fig 3B) (i.e oriented for oxidation at the substrate C3 atom) The only notable difference is that the side chain

of Asp452 assumes a different conformation, offering the possibility of a hydrogen bond between its Od2 carboxylic oxygen and the glucosyl O1 of 2FG, thus replacing the 2FG O1-Gln448Ne2 interaction observed

in H167A2FG [19] The Tyr456 side chain of the open state in H167A2FG is located some 13 A˚ from the

A

B

C

Fig 2 Inactivation kinetics of P2O wild-type and mutants

Inactiva-tion at 60 C (pH 6.5) (A) , wild-type; , H450Q; , F454P; ,

F454N; , Y456W (B) , wild-type; , F454A ⁄ Y456A; , F454A ⁄

S455A ⁄ Y456A; , D454–456 (C) Same as in (A) and (B), but at

70 C.

Table 6 Melting temperature T m of T multicolor P2O wild-type and mutants ND, not determined.

21

21

21

21

Rsym

Rfactor

b ⁄work

Rfree

c favoured

a Rsym

Rhkl

Ri

Ri

Rfactor

Rhkl

|Fc

|Fo

c Ramachandran

2FG molecule, where it forms hydrogen bonds to Gln365Ne2 and one water molecule In the open state

of Y456W2FG, the Trp456 side chains adopts the same position and conformation as the tyrosine, and also makes the same edge-to-face ring stacking interaction with Phe454 as observed in H167A2FG The tryptophan side chain is unable to form the tyrosine Og hydrogen bonds, although a new hydrogen bond is possible between Trp456Ne1 and Asp101 Od2 (Fig 3C) The observation that the loop is highly ordered in the H167A2FG complex was attributed to two principal factors: first, that the H167A variant is redox impaired (i.e removal of the covalent His167-FAD bond reduces the oxidative power of FAD) and, second, that 2FG

is a very slow substrate for P2O [i.e reduction by 2FG

of wild-type P2O (kobs= 0.0064 min)1) and of H167A (kobs= 0.000027 min)1)] [19] Because H167A2FG and Y456W2FGbind the same slow substrate, the difference

is therefore mainly attributed to the introduction of the larger tryptophan side chain at position 456, which may alter the conformational ensemble accessible to the loop, and to the intact His167-FAD bond, which retains the oxidative power of FAD in Y456W to allow higher 2FG turnover rates

Structure of the F454N-2FG complex

In the F454N2FG complex, more pronounced changes occur in the substrate-recognition loop (Fig 4) The electron density is of very high quality for the overall protein, including the 2FG molecule bound in the same C3-oxidation mode as in H167A2FG [19] and

A

B

C

Fig 3 Active-site structure of the Y456W mutant with bound 2FG.

(A) Active-site loop conformation in Y456W2FG (yellow)

superim-posed onto H167A 2FG (green) [19] The comparison of the mutant

is made with H167A rather than the wild-type because the

sub-strate-recognition loop is open in H167A and the same ligand is

bound The 2FG molecule is bound for oxidation at C3 (B) Close-up

of region around 2FG Superposition of models as in (A) For clarity,

no water molecules are shown (C) Details of the interactions made

by Trp 456 (Y456W) and Tyr 456 (H167A) The overall loop

conforma-tion, the bound ligand, the flavin cofactor and most active-site

resi-dues are strikingly similar in the two complexes Small backbone

changes at position 452 induce a different conformation of the Asp

side chain, and the interactions made by Tyr456are abolished by

the mutation The Trp side chain is instead stabilized by a hydrogen

bond to Asp 101

Fig 4 Active-site structure of the F454N mutant with bound 2FG Loop conformation in F454N2FG (yellow) superimposed onto H167A 2FG (green) [19] The 2FG molecule is bound for oxidation at C3 The Phe fi Asn replacement at position 454 induces significant changes in the 452–454 backbone and side chains without affecting the position or orientation of the bound 2FG molecule.

Y456W2FG (present study) Nonetheless, the substrate

loop in F454N2FG is disordered beyond the mutated

residue 454, lacking interpretable electron density for

the segment 455–460 Thus, the Phe454fi Asn

replace-ment appears to have a large influence on the local

conformation of the loop compared to other variants

for which structural data have been analyzed We have

reported previously that, in the open loop

conforma-tion in the C3 oxidaconforma-tion mode (H167A2FG), residues

454 and 456 do not form specific interactions with the

sugar substrate but, rather, they are folded away from

the active site [19] It is therefore particularly

interest-ing to note that Asn454 assumes a position not

previ-ously observed in 2FG complexes of P2O, and that its

side-chain amide group approaches the exocylic O6

hydroxyl group of the substrate, but does not come

close enough to form a hydrogen bond (Fig 4) In a

structure superposition, the distance between the Ca

position of Asn454 in F454N2FG and that of Phe454 in

the closed wild-type acetate complex (WTACT; PDB

code 1TT0) [5] is 1.5 A˚ The lack of density beyond

position 454 does not allow further analysis of this

loop conformer, indicating that, even in the presence

of orderly bound substrate analogue, the loop

under-goes significant dynamic fluctuations Despite the

unu-sual position of Asn454, Asp452 is positioned as in the

Y456W mutant, forming a potentially tight interaction

with 2FG O1 (distance 2FG O1–Asp452Od2, 2.6 A˚)

Structure of H450Q

In the WTACT complex, Arg451, the residue

immedi-ately following the mutated histidine in H450Q has

two well defined alternative side-chain conformations,

each of which appears appropriately stabilized In the

first conformation, two interactions are possible:

Arg451Ng1–Asp470Od2 and Arg451Ng2–water In the

second conformation, the possible interactions are:

Arg451Ng1–Ser465 O and Arg451Ng2–water Thus, it

appears that the arginine can alternate between these

two alternative conformations when the loop is in the

fully closed state In the open loop state of H167A2FG,

the latter conformation is prevalent (with a Ser465

interaction) H450Q assumes the same open loop

con-formation as H167A2FG; however, in H450Q, we

observe backbone displacements of 1 A˚ at positions

451 and 452, and the Arg451 side chain assumes a

dif-ferent conformation than those observed in the fully

closed and open states, participating in a different set

of side-chain interactions (Fig 5): Arg451Ng1–Asp101

Od1, Arg451Ng1–Tyr456 Og and Arg451Ng1–water

The changes introduced in the region 450–452 are

also manifested as a somewhat weak density for the

side-chain carboxylate group of Asp452, most likely because the carboxylate group lacks interaction possi-bilities in H450Q In both the WTACT and H167A2FG complexes, Asp452can participate in side-chain interac-tions with nearby residues and solvent (WTACT: Asp452 Od2–Lys91Nf, Asp452 Od2–Ala453N, Asp452 Od1–two water molecules; H167A2FG: Asp452 Od1–Asp470 Od2, Asp452Od1–Arg472Ng1)

Comparison of the active-site volumes

To determine whether there are changes in the volume

of the active site as a result of mutations that may explain the ability to accommodate substrates other than glucose, the cavity volumes were calculated for the mutant models, and compared with those of the closed state in WTACT (PDB code 1TT0) [5] and the open state in H167A2FG (PDB code 2IGO) [19] The substrate-recognition loop in F454N with bound 2FG is heavily disordered in the region 455–460 and, because these residues were not modelled, the volume could not be calculated for this mutant In all models, the substrate-recognition loop is in the open state, which means that the gorge leading to the active site is open to the large internal void at the homotetramer centre from which the active sites are accessible This complicates any attempt at computation of the

Fig 5 Active-site structure in the H450Q mutant without ligand Changes of the 450–452 backbone region (in particular, Asp452and Arg 451 ) in H450Q (yellow) compared to H167A2FG (green) [19] Hydrogen bonds formed by Arg 451 are coloured yellow in H450Q, and green in H167A The mutation at position 450 leads to back-bone perturbation accompanied by compensatory stabilizing interac-tions formed by Arg 451 Although no sugar is bound in H450Q, the overall structure of the substrate-recognition loop is similar to that

of the open sugar-binding state of H167A2FG.