Báo cáo khoa học: The effect of pH on the initial rate kinetics of the dimeric biliverdin-IXa reductase from the cyanobacterium Synechocystis PCC6803 pptx

The initial rate demonstrates a linear dependence on enzyme concentration from 0.5–5 lg⁄ mL when assayed at pH 5 with NADPH or NADH as cofactor.. Initial rate experiments with NADPH or N

biliverdin-IXa reductase from the cyanobacterium

Synechocystis PCC6803

Jerrard M Hayes and Timothy J Mantle

School of Biochemistry and Immunology, Trinity College, Dublin, Ireland

Introduction

Cyanobacteria utilize linear tetrapyrroles as

light-har-vesting pigments that are found covalently attached to

phycobiliproteins in the ‘light pipes’ known as

phyco-bilisomes [1] Two genera, Synechococcus and

Prochlo-rococcus, are suggested to be responsible for 25% of

global photosynthesis [2] and, although some strains of

Prochlorococcus do not express phycobiliproteins (e.g

MED4), others (e.g SS120) express a phycourobilin-containing type-III phycoerythrin [3] Linear tetrapyr-role metabolism in cyanobacteria is therefore a major physiological pathway Cyanobacteria express ferre-doxin-dependent bilin reductases (PcyA, PebA and PebB) that synthesize phycocyanobilin and phycoery-throbilin from biliverdin-IXa [4] These linear

tetrapyr-Keywords

biliverdin reductase; compulsory ordered

mechanism; dimer; pH; Synechocystis

Correspondence

J M Hayes, School of Biochemistry and

Immunology, Trinity College, Dublin 2,

Ireland

Fax: +353 677 2400

Tel: +353 895 1612

E-mail: jehayes@tcd.ie

(Received 23 April 2009, revised 9 June

2009, accepted 11 June 2009)

doi:10.1111/j.1742-4658.2009.07149.x

Biliverdin-IXa reductase from Synechocystis PCC6803 (sBVR-A) is a stable dimer and this behaviour is observed under a range of conditions This is in contrast to all other forms of BVR-A, which have been reported to behave

as monomers, and places sBVR-A in the dihydrodiol dehydrogenase⁄ N-ter-minally truncated glucose–fructose oxidoreductase structural family of dimers The cyanobacterial enzyme obeys an ordered steady-state kinetic mechanism at pH 5, with NADPH being the first to bind and NADP+the last to dissociate An analysis of the effect of pH on kcatwith NADPH as cofactor reveals a pK of 5.4 that must be protonated for effective catalysis Analysis of the effect of pH on kcat⁄ KmNADPH identifies pK values of 5.1 and 6.1 in the free enzyme Similar pK values are identified for biliverdin binding to the enzyme–NADPH complex The lower pK values in the free enzyme (pK 5.1) and enzyme–NADPH complex (pK 4.9) are not evident when NADH is the cofactor, suggesting that this ionizable group may inter-act with the 2¢-phosphate of NADPH His84 is implicated as a crucial resi-due for sBVR-A activity because the H84A mutant has less than 1% of the activity of the wild-type and exhibits small but significant changes in the protein CD spectrum Binding of biliverdin to sBVR-A is conveniently monitored by following the induced CD spectrum for biliverdin Binding of biliverdin to wild-type sBVR-A induces a P-type spectrum The H84A mutant shows evidence for weak binding of biliverdin and appears to bind a variant of the P-configuration Intriguingly, the Y102A mutant, which is catalytically active, binds biliverdin in the M-configuration

Abbreviations

hBVR-B, human biliverdin-IXb reductase; HSA, human serum albumin; sBVR-A, biliverdin-IXa reductase from Synechocystis PCC6803; GST, glutathione S-transferase.

roles are then incorporated into the phycobilisome

complex Intriguingly, some strains of cyanobacteria

express biliverdin-IXa reductase (BVR-A), which

catal-yses the pyridine nucleotide-dependent reduction of

biliverdin-IXa to bilirubin-IXa The first report of a

cyanobacterial BVR-A was from Synechocystis

PCC6803 (sBVR-A) [5] and BVR-A-like sequences are

also clearly identifiable in Gleobacter, Anabena, Nostoc

and Trichodesmium (E Franklin & T J Mantle,

unpublished results) Schluchter and Glazer [5]

reported on the unusual acidic pH optimum for

sBVR-A They also describe features of a Synechocystis

PCC6803 strain lacking sBVR-A, which they interpret

as indicating that the reaction product, bilirubin-Ixa,

plays a role in phycobiliprotein biosynthesis [5] We

have been intrigued that sBVR-A can potentially divert

flux from phycobilin biosynthesis and also potentially

reduce the phycobilins to the corresponding rubin, a

reaction clearly catalysed in vitro [5], albeit with a

rela-tively high Kmfor phycocyanobilin [5]

Questions on the possible function of BVR-A in

cyanobacteria parallel a major re-evaluation of the

function of BVR-A in mammals Once considered to

play a role solely in the elimination of excess haem, it

is now implicated in the maintenance of a major

anti-oxidant, bilirubin-IXa [6] At high doses, biliverdin

appears to tolerize the immune system of recipients

undergoing organ transplantation in animal studies

[7,8], although it is presently unclear whether this

effect is caused by biliverdin-IXa or bilirubin-IXa

Because BVR-A is reponsible for the production of

bil-irubin-IXa in neonates at birth, it is also a

pharmaco-logical target for treating neonatal jaundice [9] The

rat [10,11] and human enzymes [12] have been

crystal-lized; however, although there are complexes with

NAD+ [11] and NADP+ [12], little is known about

the biliverdin binding site Although the mammalian

enzymes have received the most attention, comparative

studies on the salmon and Xenopus tropicalis enzymes

are available [13,14] In this respect, the enzyme from

Synechocystis is of considerable interest because it

exhibits a narrow acidic pH optimum compared to the

broad range of pH that can support activity for the

mammalian enzymes [5,14] The cyanobacterial enzyme

is also refractory to activation by inorganic phosphate

when NADH is the cofactor [14] In preliminary

exper-iments, we observed that sBVR-A is not subject to the

potent substrate inhibition observed with the

mamma-lian enzymes and is therefore the first candidate,

among all BVR-A forms studied to date, where a

com-plete initial rate study can be comcom-pleted in the absence

of a biliverdin-binding protein as well as at the

opti-mum, presumably physiological, pH In preliminary

gel filtration experiments, we have also shown that the Synechocystis enzyme behaves as a dimer and such studies are extended to include the light-scattering and analytical ultracentrifugation studies described here

We report a complete initial rate study, including the effect of pH on the kinetic parameters and site-directed mutagenesis studies, to gain an understanding of the function of sBVR-A in cyanobacteria and also to increase our knowledge of the mechanism of an enzyme closely related to a pharmacological target for neonatal jaundice

Results

The expression vector pETBVR-A allowed us to rou-tinely prepare 20 mg of electrophoretically homoge-nous sBVR-A from 4 L of culture using Escherichia coli BL21 (DE3) cells Using this approach, the enzyme has two additional residues at the N-terminus (Ser-Gly) but lacks the His-tag in the preparation reported earlier [5] The enzyme was colourless; how-ever, the UV spectrum revealed significant absorbance

at 260 nm Analysis of the protein sample by HPLC revealed that, in addition to the protein, there was one major and two minor peaks that absorbed at 254 nm The major peak, which eluted at 38 min, was identified

as NADPH by its retention time, fluorescence emission spectrum and UV absorbance spectrum We have not pursued the identity of the two minor peaks All three compounds were released from the enzyme when it was bound to 2¢,5¢-ADP-sepharose and, under these conditions, the enzyme was eluted without contamina-tion In preliminary experiments, we observed that sBVR-A eluted just before BSA on gel filtration in

25 mm sodium citrate pH 5 (the optimum pH for activity; see below) and, by comparison with the elu-tion volume of standard proteins, this is consistent with a molecular mass of 69 kDa at 20C and 74 kDa

at 4C (Table 1) Although the enzyme is less active

at pH 7.5, gel filtration was carried out at this pH and

at 20C as well as 4 C and, under all these condi-tions, the molecular mass of the enzyme corresponds

to that of the dimer (Table 1) This result is novel because all BVR-As described to date have been reported to behave as monomers [15–17] To confirm that sBVR-A is a dimer, we examined the native molecular mass using light-scattering and analytical ultracentrifugation (both sedimentation velocity and sedimentation equilibrium) and the results obtained are provided in Table 1 These confirm the results of the gel filtration and are consistent with the dimeric nature of sBVR-A because several purified prepara-tions have been shown to run with a subunit molecular

mass of 34 kDa on SDS⁄ PAGE Prior to a detailed

kinetic analysis, the stability of sBVR-A at a range of

pH values was determined Over the pH range 5–7, the

enzyme did not lose any activity when pre-incubated

for 180 min The enzyme was unstable outside this

range, being particularly unstable below pH 4.5 At

pH 4, the half life was 30 s At pH 8, the enzyme

started to lose activity after 60 min and, at pH 9,

retained 75% of the activity after 60 min of

incuba-tion The initial rate demonstrates a linear dependence

on enzyme concentration (from 0.5–5 lg⁄ mL) when

assayed at pH 5 with NADPH or NADH as cofactor

All initial rate experiments were conducted within this

range of enzyme concentration In a preliminary set of

experiments, there was no substrate inhibition up to a

biliverdin concentration of 50 lm, in clear distinction

to the mammalian enzymes

Initial rate experiments with NADPH or NADH as

the variable substrate were carried out by working at

various fixed concentrations of biliverdin-IXa and

varying the concentration of NADPH from 5–100 lm

or NADH from 50–1000 lm The ‘fixed’

concentra-tions of biliverdin were also varied from 0.5–10 lm to

yield the plot for NADPH as the variable substrate

shown in Fig 1A The apparent Vmax values for

NADPH as the variable substrate were then replotted

against the biliverdin concentration (Fig 1B) to yield

the true Vmax and true Km values for biliverdin with

NADPH as cofactor (Table 2) The initial rate

mea-surements also yielded linear double-reciprocal plots

(not shown) that intersected to the left of the

recipro-cal initial rate axis, suggesting that the mechanism was

sequential However, these experiments could not iden-tify which of the substrates bound first or whether there was any particular order in their binding A simi-lar pattern was obtained with NADH as the variable substrate (data not shown)

Initial rate experiments with biliverdin-IXa as the variable substrate were carried out similarly to those described for NADPH but using various ‘fixed’ con-centrations of NADPH and varying the biliverdin-IXa concentration in the range 0.5–10 lm The data were

Table 1 Relative molecular mass of native sBVR-A AUC, area

under the curve.

pH

Temperature (C)

MW (kDa)

AUC equilibrium

A

B

Fig 1 Initial rate kinetics of sBVR-A with NADPH as the variable substrate (A) The reaction was conducted in 100 m M sodium cit-rate buffer (pH 5) and the reaction was initiated by the addition of sBVR-A (5 lg) The concentrations of NADPH are indicated and the concentrations of biliverdin-IXa were 0.5 l M ( ), 1 l M ( ), 2 l M

(.), 5 l M (r) and 10 l M (•) Each point represents the mean and the error bars represent the standard deviation of triplicate values The curves are least squares fits to a rectangular hyperbola (B) A replot of the apparent V max from (A) against the concentrations of biliverdin-IXa The curve is a least squares fit to a rectangular hyper-bola and the error bars are the standard errors from the fits in (A).

fitted to a rectangular hyperbola (Fig 2A) The true

Vmax and Km for NADPH were calculated by

replot-ting the apparent Vmax values (obtained from the fits

in Fig 2A) against the NADPH concentration

(Fig 2B) and the kinetic constants are shown in

Table 2 The initial rate data with biliverdin-IXa as

the variable substrate also yielded linear intersecting

double-reciprocal plots that were consistent with a

sequential mechanism (data not shown) Although

these data sets indicate that the enzyme obeys a

sequential mechanism, product inhibition patterns are required to distinguish between steady-state ordered, random sequential and Theorell–Chance mechanisms NADP+ inhibition with NADPH as the variable substrate was carried out at saturating (10 lm) levels

of biliverdin The inhibitory concentrations of NADP+ were in the range 0–100 lm, whereas the concentration

of NADPH varied in the range 5–100 lm Curves were again fitted to the initial rate data set and used to yield double-reciprocal plots (Fig 3A) The pattern of the double-reciprocal plots shows that NADP+ exhibits competitive kinetics against NADPH When the slope values of the double-reciprocal plots were replotted against the inhibitor concentration, a linear relation-ship was obtained (Fig 3B) and was used to determine the inhibitor constant Kisfor NADP+, which is shown

in Table 3 NADP+ inhibition was also carried out with biliverdin as the variable substrate These experi-ments were performed as described for NADPH but keeping the NADPH concentration constant at nonsat-urating (10 lm) and satnonsat-urating (1 mm) levels of NADPH and varying the biliverdin concentration in the range 0.5–10 lm A concentration of 1 mm was used for NADPH to ensure saturation NADP+ showed mixed inhibition against biliverdin at nonsatu-rating levels of NADPH When the experiment was repeated at saturating levels of NADPH, no inhibition

A

B

Fig 2 Initial rate kinetics of sBVR-A with biliverdin-IXa as the

vari-able substrate (A) The reaction was conducted in 100 m M sodium

citrate buffer (pH 5) and the reaction was initiated by the addition of

sBVR-A (5 lg) The concentrations of biliverdin-IXa are indicated and

the concentrations of NADPH were 5 l M ( ), 10 l M ( ), 20 l M (.),

50 l M (r) and 100 l M (•) Each point represents the mean and the

error bars represent the standard deviation of triplicate values The

curves are least squares fits to a rectangular hyperbola (B) A replot

of the apparent V max from (A) against the concentrations of NADPH.

The curve is a least squares fit to a rectangular hyperbola and the

error bars are the standard errors from the fits in (A).

Table 2 Kinetic parameters for wild-type and mutant forms of sBVR-A.

sBVR-A

Variable substrate

V max (lmolÆmin)1Æ

mg)1)

Km (l M )

kcat (s)1) Wild-type NADPH 0.78 ± 0.06 10.78 ± 3.2 0.44 ± 0.034

Biliverdin 0.79 ± 0.07 2.32 ± 0.59 0.45 ± 0.02 NADH 0.29 ± 0.04 207 ± 66 0.17 ± 0.023 Biliverdin 0.24 ± 0.027 1.6 ± 0.55 0.15 ± 0.015 Y102A NADPH 0.33 ± 0.06 3.55 ± 3.2 0.18 ± 0.034

Biliverdin 0.33 ± 0.022 16.19 ± 1.62 0.18 ± 0.012 R185A NADPH 0.089 ± 0.006 4.54 ± 1.4 0.05 ± 0.034

Biliverdin 0.089 ± 0.01 3.22 ± 0.9 0.05 ± 0.006 H129A NADPH 0.68 ± 0.01 4.53 ± 0.32 0.39 ± 0.006

Biliverdin 0.68 ± 0.055 1.3 ± 0.34 0.39 ± 0.03 H126A NADPH 0.62 ± 0.1 23.3 ± 10 0.35 ± 0.06

Biliverdin 0.64 ± 0.05 4.67 ± 0.72 0.36 ± 0.03 H97A NADPH 0.58 ± 0.05 5.81 ± 2.3 0.33 ± 0.03

Biliverdin 0.58 ± 0.026 2.26 ± 0.28 0.33 ± 0.015

Biliverdin

0.008 0.008

Unable to calculate E101A NADPH 0.27 ± 0.034 8.8 ± 4 0.15 ± 0.02

Biliverdin 0.25 ± 0.22 21.66 ± 24.55 0.14 ± 0.13 D285A NADPH 0.1 ± 0.006 1 ± 0.73 0.057 ± 0.01

Biliverdin 0.11 ± 0.013 1.60 ± 0.6 0.062 ± 0.007

was observed (data not shown) The inhibition

con-stants (Kisfrom the slope replot and Kiifrom the

inter-cept replot) for NADP+with biliverdin as the variable

substrate are shown in Table 3 NAD+ showed

com-petitive kinetics against NADH and was a mixed

inhibitor against biliverdin at nonsaturating (100 lm)

levels of NADH (Table 3)

Bilirubin inhibition with biliverdin as the variable

substrate was conducted at saturating (100 lm) levels

of NADPH and revealed that bilirubin is a mixed

inhibitor against biliverdin at saturating levels of

NADPH Product inhibition experiments with bilirubin

as an inhibitor were also conducted with NADPH as the variable substrate (5–100 lm) at nonsaturating (1 lm) and saturating (10 lm) levels of biliverdin Ini-tial rate data for nonsaturating levels of biliverdin show that bilirubin exhibits mixed inhibition kinetics

at nonsaturating levels of biliverdin and, when the experiment was repeated at saturating levels of biliver-din, the inhibition becomes uncompetitive (Fig 4) Bilirubin exhibits mixed inhibition against NADH at nonsaturating levels of biliverdin and mixed inhibition against biliverdin at saturating levels of NADH (data not shown) Inhibition constants are shown in Table 3 These product inhibition patterns are entirely consis-tent with sBVR-A obeying a steady-state ordered mechanism at pH 5, with NADPH being the first to bind and NADP+the last to dissociate

Inorganic phosphate anion has been shown to be an activator of human BVR-A [14] Increasing amounts

of sodium phosphate (0–100 mm) were added to the sBVR-A assay using both NADH and NADPH as cofactor and at pH 5 and pH 7 The pH was moni-tored before and after the assay to ensure that it did not change significantly when adding increasing amounts of phosphate The effect of ionic strength was found to be minimal Inorganic phosphate was found

to have no effect on sBVR-A activity with either cofac-tor at either pH This is a major discriminating feature between the cyanobacterial enzyme and the vertebrate BVR-A family members

It is often the case that, when determining the effect

of pH on the kinetic parameters of a two-substrate enzyme, one substrate is held at 10· Km(91% saturat-ing) and the variation of initial rate with the concen-tration of the second substrate is then used to estimate

kcat and the Km for the variable substrate However, the assumption that a concentration that saturates at one pH will saturate at all the pH values under investi-gation is not without risk All the initial rate parame-ters reported in the present study were determined in accordance with the classic analysis of Florini and Ves-tling [18] to calculate Km and Vmax The effect of pH

on kcat was investigated over the pH range 4.25–7.0 with both NADPH and NADH The values measured for kcatare shown when the data set is described with NADPH or biliverdin as the variable substrate The same kcat profile should be obtained (irrespective of which substrate is held as the variable) and this is clearly seen in Fig 5A Evidently, there is a pK at 5.4 for the ‘less acidic’ limb of the pH curve defining a side chain that must be protonated for catalysis to occur There is no co-operativity for this protonation because the plot of log kcatversus pH gives a slope of approximately –1 (Fig 5B) On the ‘more acidic’ limb

A

B

Fig 3 Product inhibition by NADP+with NADPH as the variable

substrate The reaction was conducted in 100 m M sodium citrate

buffer (pH 5) and the reaction was initiated by the addition of

sBVR-A (5 lg) Biliverdin-IXa was held constant (10 l M ) at

saturat-ing levels and the levels of NADPH are indicated The

concentra-tions of NADP + were 0 l M ( ), 10 l M ( ), 20 l M (.), 50 l M (r)

and 100 l M (•) (A) The data are represented as a double-reciprocal

plot and (B) a slope replot (apparent K m ⁄ V max from fits to a

rectan-gular hyperbola) against the concentration of NADP +

of this curve, there is a second pK (4.7) and

proton-ation of this group reduces the kcat(but only by 50%)

Great care has to be taken because the enzyme is

highly unstable at the ‘more acidic’ pH values Initial

rates were obtained at pH 4 during the first few

sec-onds of the reaction under conditions when at least

90% of the activity was retained; however, these are

clearly not ideal conditions With NADH as cofactor,

the pK on the ‘less acidic’ limb is clearly not

co-opera-tive (data not shown) and has a similar value (5.7) to

that observed with NADPH (5.4) It is intriguing that

the kcat on the ‘more acidic’ limb shows very little

dependence on pH with NADH as the cofactor

The effect of pH on the kcat⁄ Kmvalues for NADPH and NADH was also analysed The log plot for

kcat⁄ Km is shown for the NADPH (Fig 6A) and bili-verdin (Fig 6B) data sets and for the NADH (Fig 6C) and biliverdin (Fig 6D) data sets With NADPH as

Table 3 Initial rate kinetic parameters for product inhibition studies of sBVR-A.

Fig 4 Product inhibition by bilirubin-IXa with NADPH as the

vari-able substrate at saturating levels of biliverdin (A) The reaction

was conducted in 100 m M sodium citrate buffer (pH 5) and the

reaction was initiated by the addition of sBVR-A (5 lg)

Biliverdin-IXa was held constant at saturating levels (10 l M ) and the

concen-trations of NADPH are indicated The concenconcen-trations of bilirubin-IXa

were 0 l M ( ), 1 l M ( ), 2 l M (.), 5 l M (r) and 10 l M (•) The

data are represented as a double-reciprocal plot.

A

B

Fig 5 Effect of pH on kcat with NADPH as cofactor: 25 m M

sodium citrate (pK values of 3.13, 4.76 and 6.4) was used as buffer over the entire pH range studied (A) Values for k cat were obtained with NADPH as the variable substrate (•) and with biliverdin-IXa as the variable substrate ( ) (B) The log ⁄ log plot for (A) is shown.

cofactor, the kcat⁄ Kmdata reveal two pK values of 5.1 and 6.1 with NADPH as the variable substrate and 4.9 and 5.6 with biliverdin as the variable substrate This

is consistent with two ionizing groups in the free enzyme with pK values of 5.1 and 6.1 that define NADPH binding Interestingly, with NADH (Fig 6C) and biliverdin (Fig 6D) as the variable substrates, there is only a single pK (i.e 5.3 for NADH binding

to the free enzyme and 5.5 for biliverdin binding to the enzyme–NADH complex) The only difference between NADPH and NADH is the 2¢-phosphate on NADPH and it is tempting to suggest that there may be a disso-ciable group with a pK of 5.1 in the free enzyme (4.9

in the enzyme–NADH complex) that is not involved in binding NADH This pK may be associated with an ionizing residue that is involved in binding the 2¢-phos-phate of NADPH

The effect of pH on the initial rate kinetics is consis-tent with two ionizing groups in the enzyme active site involved in binding NADPH, which may be perturbed slightly in the enzyme–NADPH complex but which are both required for binding biliverdin In the ternary com-plex, a group with a pK of 5.4 must be protonated for efficient catalysis with NADPH (in the case of the ter-nary complex with NADH as cofactor, this pK is 5.7) The nature of the second pK in the ternary complex with NADPH (4.7) is unclear There is no analogous

pK in the binding of NADH and it is not readily appar-ent in the ternary complex with NADH as cofactor

To identify the ionizing residues, we have attempted

to crystallize sBVR-A, so far without success We have therefore built a model using the rat enzyme as a tem-plate and this is shown in Fig 7 In this model, we have highlighted residues from the sBVR-A model that are candidates for the ionizing residues These include four His (84, 97, 126 and 129) one Glu (101), one Asp (285) and one Tyr (102) residue All were mutated to Ala residues and the sequences confirmed The gluta-thione S-transferase (GST) fusions were purified, the GST domain cleaved and removed by affinity chroma-tography and the mutant sBVR-As analysed in terms

of CD spectra, induced CD spectra for biliverdin and initial rate kinetic parameters The kinetic parameters

of all the mutants are shown in Table 2 This clearly rules out His97, His126 and His129, which have kcat and Km values that are very close to those displayed

by the wild-type enzyme In addition these three His

to Ala mutants show CD spectra and induced CD spectra for biliverdin that are very close to those exhibited by the wild-type enzyme (Fig 8A) However,

a clear candidate for a key active site residue is His84 The specific activity of the H84A mutant is 1% of the wild-type and is so low that we were unable to

A

B

C

D

Fig 6 log kcat⁄ K m versus pH for NADPH and NADH as the variable

substrates (A) Log kcat⁄ K m with NADPH as the variable substrate.

(B) Log k cat ⁄ K m with biliverdin as the variable substrate and NADPH

as cofactor (C) Log kcat⁄ K m with NADH as cofactor (D) Log kcat⁄ K m

with biliverdin as the variable substrate and NADH as cofactor.

determine the kinetic parameters with confidence.

Examination of the CD spectrum of H84A protein

reveals that it is close to, but not identical with, that

of the native enzyme (data not shown) We suggest

that H84A may be the residue responsible for

proton-ating the pyrollic nitrogen prior to hydride transfer

(see Discussion); however, we cannot discount a

mod-est global structural change having some role in the

decreased catalytic activity It should be noted that, in

this respect, the H84A mutant was isolated with bound

nucleotide and is clearly able to bind biliverdin-IXa

(Fig 8B), suggesting that both substrates are able to

bind to the H84A mutant

The binding of biliverdin to wild-type sBVR-A

stabi-lizes the helical P-configuration (Fig 8A) of the linear

tetrapyrrole, also known as the ‘lock washer’ [19] and,

by this criteria, biliverdin can be seen to bind weakly

to the H84A mutant (Fig 8B), albeit with a

broaden-ing of the positive ellipticity into a peak at 400 nm

and a significant shoulder at 325 nm Two of the

mutants (R185A and D285A) have kcatvalues that are

only 10% of wild-type (Table 2) The D285A mutant

has modest changes in the Kmvalues and the induced

CD spectrum for biliverdin is similar to wild-type The

R185A mutant also shows similar Km values to the

wild-type; however, the positive ellipticity of the

induced CD for biliverdin shows a sharp peak at

400 nm (the wild-type shows a broad peak centred at

390 nm) and a clear minor peak at 325 nm (Fig 8C)

In the case of the E101A mutant, there is a significant increase in the Km for biliverdin (nine-fold) and the

kcat is reduced to a third of that of the wild-type (Table 2) The induced CD spectrum for biliverdin bound to the E101A mutant shows a considerably reduced amplitude, with the positive ellipticity split into two peaks at 325 nm and 400 nm (Fig 8D) In this case, the trough is centered at 580 nm (compared

to 700 nm in the wild-type) Intriguingly the Y102A mutant exhibits CD behaviour that reflects the M-con-figuration (Fig 8E) The ability of this mutant to stabilize the opposite enantiomer is associated with a modest (50%) drop in the kcat and a seven-fold increase in the Kmfor biliverdin

Discussion

All mammalian forms of BVR-A are reported to behave as monomers These include the enzymes from pig spleen and rat liver [15], human liver [16] and ox kidney [17] We have artificially created a dimer of rat BVR-A by using fused GST domains as sites for dimerization [20] The Synechocystis enzyme is there-fore the first natural dimer reported for BVR-A We were careful to use a range of techniques to measure the native molecular mass of sBVR-A and to conduct these experiments under a range of conditions, includ-ing temperature, pH and the presence or absence of phosphate, because this has such a pronounced activat-ing effect on the mammalian enzymes with NADH as cofactor [14] Under all of these conditions, the native enzyme exhibits a molecular mass of 66–80 kDa and, because the molecular mass is 34 kDa as measured by SDS⁄ PAGE, we conclude that sBVR-A is a stable dimer In light of the recent reports on the structures

of monkey dihydrodiol dehydrogenase [21] and the N-terminally truncated dimeric form of glucose–fruc-tose oxidoreductase [22], we propose that sBVR-A joins this small family of pyridine nucleotide-depen-dent oxidoreductases that dimerize via the C-terminal b-sheet domain It is intriguing that we purify sBVR-A with bound pyridine nucleotide because this is also a feature of the glucose–fructose oxidoreductase enzyme

In addition to its unique quaternary structure, sBVR-A also exhibits a sharp pH optimum, which we reproducibly measured as pH 5 This behaviour is in contrast to that displayed by the mammalian BVR-A monomers, which show activity over a broad range of

pH values in the range 5–9 [14] The cyanobacterial BVR-A is not subject to the potent substrate inhibition observed with the mammalian forms and this has

Fig 7 A model for sBVR-A sBVR-A model (green) superimposed

on the rat BVR-A crystal structure (grey) The amino acid residues

that mutated and their positions within the sBVR-A model are

shown The numbers indicated represent the amino acid residues

of sBVR-A.

allowed us to complete a full initial rate study on the

Synechocystis enzyme and to rigorously establish that

it obeys an ordered steady-state mechanism The effect

of pH on the initial rate parameters has allowed us to

identify two ionizing groups in the free enzyme that

are required in the unprotonated (pK 5.1) and

proton-ated forms (pK 6.1), respectively, for binding NADPH

The protonation state of the lower pK does not affect

the binding of NADH This is consistent with an

ioniz-ing residue, pK 5.1, in the free enzyme which, in the

deprotonated state, may promote interaction with the

2¢-phosphate group of NADPH but plays no

signifi-cant role in binding NADH In the case of human

BVR-A, we have suggested that the protonation state

of Glu75, may effect the interaction with Arg44, which

is a key residue involved in binding the 2¢-phosphate

of NADPH [14] Further work is required to identify

possible analogous candidates in sBVR-A There is

clearly a pK of 5.4 in the ternary complex that is

required to be protonated for efficient catalysis with

NADPH and which, with NADH as cofactor, exhibits

a pK of 5.7 Our mutagenesis studies tentatively

iden-tify this residue as His84 We have recently discussed

the possibility that a His residue may be responsible

for supplying a proton to the pyrrole nitrogen atom of

biliverdin-IXb prior to hydride transfer in the case of

human biliverdin-IXb reductase (hBVR-B) Although

structurally distinct to BVR-A, BVR-B is a good model for mechanistic studies on the reduction of the linear tetrapyrrole ‘C10’ position by hydride BVR-B is

a ‘non-Ixa’ biliverdin reductase [23] and is unable to accommodate biliverdin-IXa in a productive orienta-tion, although we have shown that it clearly binds, albeit rotated by 90 [24], when compared with the bil-iverdin isomers that are substrates (i.e the IXb, IXd and IXc isomers) Mutagenesis studies on BVR-B have indicated that a solvent hydroxonium ion may be the source of the proton and this was found to be consis-tent with quantum mechanical⁄ molecular mechanical calculations [25] However, our studies with BVR-B as

a model have demonstrated that there is a requirement for proton transfer to the pyrrole nitrogen atom prior

to hydride transfer in the hBVR-B reaction co-ordinate [25] and we suggest that His84 is a good candidate for this function in sBVR-A The second ‘more acidic’ pK

in the kcat data set (pK 4.7) is also prominent with NADPH but less so with NADH

We have taken advantage of the induced CD spectra

of biliverdin when enantiomeric forms are stabilized by binding to proteins, including serum albumins [26] and, as reported in the present study, sBVR-A In solution, these chiral forms are clearly in equilibrium

so that no CD spectrum is seen Bilirubin adopts two enantiomeric ‘ridge tile’ configurations [19,27], whereas

Fig 8 Induced CD spectra of biliverdin-IXa bound to sBVR-A and various mutants sBVR-A and the mutants indicated (all at 29 l M ) were incubated with biliverdin-IXa (30 l M ) and NADP + (100 l M ) (A) Wild-type, (B) H84A, (C) D285A, (D) E101A and (E) Y102A.

biliverdin is suggested to oscillate between two helical

‘lock washer’ configurations, and one of these, the

P-configuration, is clearly stabilized in a

biliverdin–myo-globin complex [28] Although it is tempting to suggest

that trapping oscillations between two helical forms is

the phenomenon responsible for the P(lus) and

M(inus) spectra of biliverdin when bound to human

serum albumin (HAS) and BSA respectively [26], this

remains to be confirmed A recent X-ray structure of

HSA with bilirubin bound [29] shows a ZZE

configu-ration (not a ridge tile), whereas, in solution, HSA

stabilizes a P-type induced CD spectrum so that, until

we have CD spectra of the appropriate crystals,

abso-lute assignments will not be possible The wild-type

and most mutants that we have studied show

P-behav-iour [by convention the sign of the longer wavelength

defines (P)lus or (M)inus] However the Y102A mutant

appears to stabilize the inverted chiral M-form This

mutant exhibits catalytic activity (kcatis approximately

50% of wild-type), albeit with a seven-fold increase in

the KmBILIVERDIN Because the KmNADPH is very

simi-lar to wild-type, this suggests that hydride transfer

from the C4 of the nicotinamide ring can be

accom-plished relatively efficiently, even with variable

configu-rations of biliverdin bound at the active site The

Y102A mutant therefore accomodates a variant

config-uration of biliverdin to the wild-type enzyme but

retains the ability to catalyse the transfer of hydride

from both pyridine nucleotides As discussed

previ-ously [30], the most likely model for the biliverdin

binding site can accommodate a number of

conforma-tions of biliverdin, including the various locked

iso-mers that have been shown to bind productively in

hBVR-A [30] and the two helical P- and M-conformers

described in the present study

The description of a functional BVR-A in some

cyanobacteria introduces an important issue with

regard to the subcellular localization of both PcyA and

sBVR-A These two enzymes potentially compete for

substrate and different subcellular localizations would

provide a way out of this hypothetical dilema The

opti-mum pH for activity for sBVR-A at acid pH values is

consistent with the hypothesis that sBVR-A may be

localized in the lumen of the thylakoid [5], which is

reported to maintain a pH in the range 5.5–5 [31] As a

result of the low abundance of this protein, we have not

been able to confirm this using immunogold labelling

(L Weaver, J M Hayes & T J Mantle, unpublished

results) The enzymes responsible for the synthesis of

the light-harvesting pigments phycocyanobilin and

phy-coerythrobilin (PcyA, PebA and PebB) are all

ferre-doxin-dependent and their reaction product is destined

for incorporation into the phycobilisomes that decorate

the cytosolic side of the thylakoid membrane The bilin reductase PcyA exhibits a pH optimum of 7.5 [32], whereas PebA and PebB are assayed at pH 7.5 [4], con-sistent with a distinct subcellular localization to

sBVR-A and most likely the cytosol, which has been reported

to maintain a pH in the range 6.8–7.2 [31] Further work is required to resolve this important question

Experimental procedures

The protein coding DNA for sBVR-A was amplified from Synechocystis PCC6803 genomic DNA using forward (5¢-CGCGGATCCCATGTCTGAAAATTTTG-3¢) and reverse (5¢-CGCCTCGAGCTAATTTTCAACTATATC-3¢) primers containing BamH1 and Xho1 sites, respectively, to allow directional cloning into a modified pET41a expression vec-tor (Novagen, Madison, WI, USA) The GST-sBVR-A fusion protein expressed from pETBVR-A in E coli BL21 (DE3) cells was purified on glutathione-sepharose (Chroma-trin Ltd, Dublin, Ireland) cleaved with thrombin (Sigma– Aldrich, St Louis, MO, USA) and the GST fragment removed by affinity chromatography on glutathione-sepha-rose Prior to HPLC analysis, the purified protein was incubated in 6 m urea at 95C for 2 min, centrifuged at

16 000 g for 2 min and immediately loaded onto a Supelco Discovery C18 reversed phase HPLC column (Supelco, Bellefonte, PA, USA) (25· 4 mm) at a flow rate of

1 mLÆmin)1 The HPLC column was equilibrated in

100 mm potassium phosphate (pH 6) and elution was achieved using a linear gradient of 0–40% methanol Size-exclusion chromatography was conducted using

1· 100 cm Sephacryl 200 HR (Sigma–Aldrich) columns equilibrated at pH 5 (25 mm sodium citrate, 100 mm NaCl) and pH 7.5 (25 mm Tris⁄ HCl, 100 mm NaCl) at both 4 C and 20C The calibration proteins used [b-amylase (200 kDa), alcohol dehydrogenase (150 kDa), BSA (66 kDa), carbonic anhydrase (29 kDa) and cytochrome c (12.4 kDa)] were individually applied to the column and their elution volumes used to construct a standard curve of log molecular mass versus elution volume

Light scattering was performed on sBVR-A at pH 5 and

pH 7.5 at 20C Protein samples (0.25 mgÆmL)1) were clar-ified using a 0.22 lm filter and applied to an S-200 Super-dex HR gel-filtration column connected to an AKTA FPLC system (Amersham Biosciences, Little Chalfont, UK) The column was run at 20C and a flow rate of 0.5 mgÆmL)1 in the desired equilibration buffer (25 mm sodium citrate, pH 5, 100 mm NaCl or 25 mm Tris⁄ HCl,

pH 7.5, 100 mm NaCl) The gel-filtration column was con-nected online to a miniDawn Tristar light-scattering detec-tor (Wyatt Technology, Santa Barbara, CA, USA) and an Optilab rEX Rayleigh interference detector (Wyatt Tech-nology) The weight-average molar mass of sBVR-A was calculated using the software astra (Wyatt Technology)