Báo cáo khoa học: Spectroscopic and DNA-binding characterization of the isolated heme-bound basic helix–loop–helix-PAS-A domain of neuronal PAS protein 2 (NPAS2), a transcription activator protein associated with circadian rhythms docx

In the present study, we compared the optical absorption spectra, resonance Raman spectra, heme-binding kinetics and DNA-binding characteristics of the isolated fragment containing the N

isolated heme-bound basic helix–loop–helix-PAS-A domain

of neuronal PAS protein 2 (NPAS2), a transcription

activator protein associated with circadian rhythms

Yuji Mukaiyama1, Takeshi Uchida2*, Emiko Sato1, Ai Sasaki1, Yuko Sato1, Jotaro Igarashi1,

Hirofumi Kurokawa1, Ikuko Sagami1†, Teizo Kitagawa2and Toru Shimizu1

1 Institute of Multidisciplinary Research for Advanced Materials, Tohoku University, Sendai, Japan

2 Okazaki Institute for Integrative Bioscience, National Institutes of Natural Sciences, Okazaki, Japan

In animals, biological rhythms are coordinated in

adaptive synchrony by the brain, specifically by the

suprachiasmatic nucleus of the hypothalamus The

suprachiasmatic nucleus is a major coordinator of

internal circadian organization and is itself

synchron-ized with the day–night cycle by direct neural input

from specialized retinal photoreceptors [1–6] In its simplest form, the molecular clockwork consists of autoregulatory transcriptional and translational feed-back loops that have both positive and negative ele-ments [1–6] The positive components are two transcription factors, CLOCK and mouse brain and

Keywords

circadian rhythms; DNA binding;

heme-sensor protein; PAS domain; resonance

Raman spectroscopy

Correspondence

T Shimizu, Institute of Multidisciplinary

Research for Advanced Materials, Tohoku

University, 2-1-1 Katahira, Aoba-ku,

Sendai 980-8577, Japan

Fax: +81 22 217 5604 ⁄ 5390

Tel: +81 22 217 5604 ⁄ 5605

E-mail: shimizu@tagen.tohoku.ac.jp

Present address

*Division of Chemistry, Graduate School of

Science, Hokkaido University, Kita-ku,

Sapporo, Japan

†

Graduate School of Agriculture, Kyoto

Prefectural University, Shimogamo,

Sakyo-ku, Kyoto, Japan

(Received 23 February 2006, accepted

4 April 2006)

doi:10.1111/j.1742-4658.2006.05259.x

Neuronal PAS domain protein 2 (NPAS2) is a circadian rhythm-associated transcription factor with two heme-binding sites on two PAS domains In the present study, we compared the optical absorption spectra, resonance Raman spectra, heme-binding kinetics and DNA-binding characteristics of the isolated fragment containing the N-terminal basic helix–loop–helix (bHLH) of the first PAS (PAS-A) domain of NPAS2 with those of the PAS-A domain alone We found that the heme-bound bHLH-PAS-A domain mainly exists as a dimer in solution The Soret absorption peak of the Fe(III) complex for bHLH-PAS-A (421 nm) was located at a wave-length 9 nm higher than for isolated PAS-A (412 nm) The axial ligand trans to CO in bHLH-PAS-A appears to be His, based on the resonance Raman spectra In addition, the rate constant for heme association with apo-bHLH-PAS (3.3· 107mol)1Æs)1) was more than two orders of magni-tude higher than for association with apo-PAS-A (< 105mol)1Æs)1) These results suggest that the bHLH domain assists in stable heme binding to NPAS2 Both optical and resonance Raman spectra indicated that the Fe(II)–NO heme complex is five-coordinated Using the quartz-crystal microbalance method, we found that the bHLH-PAS-A domain binds spe-cifically to the E-box DNA sequence in the presence, but not in the absence, of heme On the basis of these results, we discuss the mode of heme binding by bHLH-PAS-A and its potential role in regulating DNA binding

Abbreviations

bHLH, basic helix–loop–helix; BMAL1, mouse brain and muscle Arnt-like 1; CooA, CO-sensing heme-bound transcriptional regulator from Rhodospirillum rubrum; Ni-NTA, Ni 2+ -nitrilotriacetic acid; NPAS2, neuronal PAS domain protein 2; QCM, quartz-crystal microbalance method.

muscle Arnt-like 1 (BMAL1), both of which contain a

basic helix–loop–helix (bHLH) domain and two PAS

domains [7–10] These two transcription factors form a

heterodimer that binds to the E-box sequence, which

drives the transcription of three Period genes

(designa-ted mper1, mper2 and mper3 in the mouse) and two

cryptochrome genes (mcry1 and mcry2) The mPER

and mCRY proteins appear to act as negative

compo-nents of the feedback loop [1–10]

The mammalian forebrain expresses neuronal PAS

domain protein 2 (NPAS2), which is homologous to

CLOCK [7–10] NPAS2 plays an important role in

maintaining circadian behaviors in normal light–dark

and feeding conditions, and it is critical for adaptation

to food restriction [11–14] NPAS2 also forms a

het-erodimer with BMAL1, and the hethet-erodimer activates

the transcription of per and cry, whereas it represses

BMAL1 gene expression Like CLOCK and BMAL1,

NPAS2 has bHLH, PAS-A and PAS-B domains in its

N-terminal region, but in contrast to CLOCK, both

the PAS-A and PAS-B domains of NPAS2 contain a

heme-binding site, and CO binding to the heme

inhib-its the DNA-binding activity of the NAPS2–BMAL1

heterodimer [15]

The bHLH-PAS proteins are critical regulators of

gene expression networks underlying a variety of

essen-tial physiological and developmental processes [7,16]

In many cases, bHLH proteins dimerize to form

func-tional DNA-binding complexes, whereas bHLH-PAS

proteins are distinct from other members of the

broader bHLH superfamily due to the dimerization

specificity conferred by their PAS domains The

bHLH-PAS proteins tend to be ubiquitous latent

sig-nal-regulated transcription factors that often recognize

variant forms of the classic E-box enhancer sequence

bound by other bHLH proteins [7,16] Because CO

binding to the heme causes dissociation of NPAS2

from BMAL1 and the E-box sequence, the bound

heme itself may affect the DNA-binding properties of

NPAS2 by interacting with the bHLH region in a

direct or indirect manner Therefore, it is worth

study-ing how the bHLH domain contributes to heme

bind-ing to the PAS-A domain in NPAS2 and how the

bound heme participates in the binding of NPAS-2 to

the E-box DNA sequence

In the present study, we investigated the role of the

bHLH domain by characterizing the isolated

heme-bound bHLH-PAS-A and heme-bound PAS-A

domains of NPAS2 using optical absorption

spectros-copy, resonance Raman spectrosspectros-copy, and

heme-bind-ing kinetic analyses We found that the bHLH domain

significantly affects the spectra of the heme-bound

PAS-A domain and appears to assist in stable heme

binding, stabilizing the protein molecule We also dem-onstrated specific binding of the bHLH-PAS-A protein

to the E-box of DNA using the quartz-crystal micro-balance (QCM) method

Results

Protein expression and purification

We attempted to express the isolated bHLH-PAS-A (amino acids 1–240) and PAS-A (amino acids 78–240) domains of NPAS2 as described previously (see Experimental procedures) [22] The bHLH-PAS-A domain was properly overexpressed and folded in Escherichia coli and displayed sufficient heme-binding ability SDS⁄ PAGE followed by staining with Coo-massie Brilliant Blue revealed that the purified His-tagged bHLH⁄ PAS-A and PAS-A domains of NPAS2 were more than 95% homologous (supplementary Fig S1) To remove the His tag, the His-tagged bHLH-PAS-A and PAS-A domains were treated with thrombin and then purified using a Ni2+ -nitrilotriace-tic acid (Ni-NTA) agarose column The yields of the bHLH-PAS-A and PAS-A domains were 8.0 and 9.0 mgÆL)1of culture, respectively

Size exclusion chromatography Gel filtration analysis of a solution of bHLH-PAS-A protein using Superdex 75 (Amersham Biosciences, Uppsala, Sweden) revealed that the solution contains one major peak (more than 70%) and two minor peaks (supplementary Fig S2A) The heme-reconstitu-ted bHLH-PAS-A domain was the major species and had a molecular mass of nearly 60 kDa (supplement-ary Fig S2B) Because the monomer has a predicted molecular mass of 28 kDa, this suggests that the major species of the bHLH-PAS-A domain exists as a dimer

We collected only the major fraction and reapplied it

to the Superdex 75 column The chromatographic pro-file was the same, suggesting that the fraction was in equilibrium between monomer, dimer (majority) and trimer forms

Optical absorption spectra of Fe(III), Fe(II) and Fe(II)–CO complexes

To understand the heme environment of the bHLH-PAS-A domain of NPAS2, we obtained optical absorption spectra of the overexpressed and purified bHLH-PAS-A domain Optical absorption spectra of the Fe(III), Fe(II) and Fe(II)–CO complexes of the bHLH-PAS-A domain of NPAS2 are shown in

Fig 1A and are summarized in Table 1 For the

Fe(III) form, the absorption maxima were located at

421 nm and 543 nm at pH 8.0 The absorption

max-ima of the Fe(II) species were at 426, 530 and 559 nm,

and those of the Fe(II)–CO complex were at 422, 538

and 566 nm On the other hand, optical absorption

spectra of the Fe(III), Fe(II) and Fe(II)–CO complexes

of the PAS-A domain of NPAS2 were different from

those of the bHLH-PAS-A domain (Fig 1B and

Table 1) Namely, in the Fe(III) form, the absorption

maxima were located at 412 and 538 nm at pH 8.0 The absorption maxima of the Fe(II) species were at

423, 530 and 558 nm, and those of the Fe(II)–CO complex were at 420, 530 and 568 nm It is likely that Fe(II) and Fe(II)–CO complexes of the bHLH-PAS-A domain have mostly six-coordinate low-spin heme, whereas the Fe(III) complex has five-coordinate high-spin heme as a minor component in addition to the major six-coordinate low-spin heme [17]

Effect of pH on the Fe(III) complex

To identify the axial ligand of the bHLH-PAS-A domain of NPAS2, we examined the effect of modula-ting pH on the Fe(III) complex We did not find any remarkable pH-dependent spectral changes between

pH 6 and pH 11

Resonance Raman spectra of Fe(III), Fe(II) and Fe(II)–CO complexes

To further examine the nature of the heme environ-ment of the bHLH-PAS-A domain of NPAS2, we ana-lyzed the resonance Raman spectra of the Fe(III), Fe(II) and Fe(II)–CO complexes The spectra of the Fe(III) and Fe(II) complexes in the high-frequency region are shown in Fig 2 and summarized in Table 2

4

.

0

3

.

0

2

.

0

1

.

0

0

.

0

0 7 0 6 0 5 0 4 0

3

) m n ( h t g e l e v W

3

×

) I I ( e F ) I ( e F O C -I ( e F

8

.

0

6

.

0

4

.

0

2

.

0

0

.

0

0 7 0 6 0 5 0 4 0

3

) m n ( h t g e l e v W

5

×

) I I ( e F ) I ( e F O C -I ( e F

A

B

Fig 1 Optical absorption spectra of the isolated heme-bound basic

helix–loop–helix (bHLH)-PAS-A (A) and PAS-A (B) domains Fe(III)

(solid line), Fe(II) (dotted line) and Fe(II)–CO (dashed line)

com-plexes are shown Spectra were obtained in a buffer consisting of

100 m M Tris (pH 7.5) All proteins were His-tag-free.

Table 1 Optical absorption spectra of isolated basic helix–loop–

helix (bHLH)-PAS-A and PAS-A domains of neuronal PAS domain

protein 2.

Soret

(nm)

Visible (nm)

Soret (nm)

Visible (nm)

0 7 0 6 0 5 0 4 0 3 0 2

t h s n a m a

1555 1605 1618

) I I ( e F

) I ( e

Fig 2 Resonance Raman spectra of the high-frequency region of the Fe(III) (lower) and Fe(II) (upper) complexes of the basic helix– loop–helix (bHLH)-PAS-A domain.

Bands at 1374 and 1359 cm)1for the Fe(III) and Fe(II)

complexes were assigned as the redox-sensitive m4 band

[17,32] For the Fe(III) complex, the spin-coordination

and coordination-state marker bands (m3) were located

at 1470, 1492 and 1504 nm, corresponding to the

six-coordinate high-spin, five-six-coordinate high-spin and

six-coordinate low-spin states, respectively [17,32] The

shoulder at 1492 cm)1was ascribed to a five-coordinate

high-spin complex observed as a minor component in

the Fe(III) complex For the Fe(II) complex, on the

other hand, m3 bands were observed at 1471 and

1493 cm)1, representing the five-coordinate high-spin

and six-coordinate low-spin states, respectively

Because of the sensitivity of the Fe–CO and C–O

stretching frequencies to the heme environment

(i.e electrostatic and steric interactions with surround-ing groups), spectra of CO adducts of heme proteins provide valuable information about the heme pocket [17,32] Low-frequency and high-frequency regions of the resonance Raman spectra of the Fe(II)–CO complex

of the bHLH-PAS-A domain are shown in Fig 3A,B Isotope-sensitive lines were observed at 486 cm)1for the Fe–CO stretching mode (mFe–CO) and at 1919 cm)1 for the C–O stretching mode (mC–O) when we used13C18O Finally, we assigned the 495 and 1962 cm)1 bands to

mFe–COand mC–O, respectively (Table 3)

The inverse relationships between the mFe–CO and

mC–O frequencies are used to determine the axial ligand

of the Fe(II) heme iron (Fig 4) [17,32] The frequen-cies for the bHLH-PAS-A domain corresponded to Fe–His but not Fe–S coordination Because the mFe–CO stretching frequency (495 cm)1) is lower than those of heme complexes in a polar environment [17,32], it appears that the CO is located in a somewhat hydro-phobic environment

Optical absorption and resonance Raman spectra

of the Fe(II)–NO complex Since NO may be involved in transcription control, we next obtained optical absorption and resonance

Table 2 Resonance Raman spectra of the basic helix–loop–helix

(bHLH)-PAS-A domain of neuronal PAS domain protein 2 Excitation

was at 413.1 nm 5cHS, five-coordinate high spin; 6cLS,

six-coordi-nate low spin.

Complex m 2 (cm)1) m 3 (cm)1) Coordination

m c ( 1 )

0 7 0 6 0 5 0 4 0

3

m c ( f i h n a m a

A

0 0 0 0 0 9 0 9 0 8 0 8 0 7

t i h n a m a R

B

Fig 3 Effects of isotopically labeled CO molecules on resonance Raman spectra of the Fe (II)–CO complexes of basic helix–loop–helix (bHLH)-PAS-A in the low-frequency (A) and high-frequency (B) regions The bottom lines show the12C16O complex, the middle lines the

13 C 18 O complex, and the top lines the difference spectra between the 12 C 16 O and 13 C 18 O complexes.

Raman spectra of the Fe(II)–NO complex As shown

in Fig 5, the Fe(II)–NO complex had a Soret

absorp-tion peak at 394 nm, which is characteristic of a

five-coordinated Fe(II)–NO heme complex [23–27,32] In

addition, the characteristics of the resonance Raman

spectra of the Fe(II)–NO complex (Fig 6) were very

similar to those of the five-coordinated Fe(II)–NO

complex [24,32] Isotope-sensitive resonance lines were

observed at 514 cm)1 for the Fe–NO stretching mode

(mFe–NO) and at 1670 cm)1 for the N–O stretching

mode (mN–O) when we used15N16O

Heme-binding kinetics

In order to understand the heme-binding character

of the bHLH-PAS-A domain, we examined the

association rate constants (kon) for binding of the

Fe(II)–CO heme complex to the apo-bHLH-PAS-A

and apo-PAS-A domains of NPAS2, using a

modification of the methods described by Hargrove

et al [18] Fe(II)–CO heme (1.0 lm) was mixed with

2, 3 or 4 lm bHLH-PAS-A domain or 16 lm

apo-PAS-A domain using a stopped-flow apparatus at

25
C There was a concomitant increase in the

absorbance at 421 and 419 nm and a decrease at

408 nm, which correspond to the binding of Fe(II)–

CO heme to the apo-bHLS-PAS-A domain and the apo-PAS-A domain, respectively This shows that Fe(II)–CO heme bound strongly to the apo-bHLH-PAS-A domain (Fig 7A) The time-dependent increase in absorbance accompanying Fe(II)–CO heme binding to the apo-bHLH-PAS-A domain was composed of only a single phase (Fig 7A, inset), and the rate of association was dependent on the apoprotein concentration (Fig 7C) As summarized

in Table 4, the rate constant for association with the apo-bHLH-PAS-A domain was 3.3· 107mol)1Æs)1

In contrast, Fe(II)–CO heme binding to the apo-PAS-A domain was very slow and did not saturate under our experimental conditions (Fig 7B) There-fore, the kon value of the PAS-A domain should be less than 105mol)1Æs)1

We also determined the rate constant for the dissoci-ation of heme from the holo-bHLH-PAS-A domain of NPAS2 by mixing it with an excess of the H64Y⁄ V68F apomyoglobin mutant The reaction was monitored

by following the increase in absorbance at 410 nm (Fig 8), which accompanies the formation of Fe(III) heme-bound myoglobin The observed koff was 5.3· 10)3s)1 for the bHLH-PAS-A domain Table 4 summarizes koff as well as the calculated Kdvalues for the bHLH-PAS-A domain and those of other heme-binding proteins The Kd value of heme was estimated

to be 1.6· 10)10m for the bHLH-PAS-A domain

Analysis of DNA binding by the QCM method The QCM is a very sensitive device for the detection

of DNA–protein and protein–protein interactions in solution, which are monitored by the linear decreases

of the emitted frequency with increasing mass present

0 0 8 0 6 0 4 0 2 0 0 0

0 7 0 6 0 5 0 4 0 3

) m n ( h t g e l e v W

3

×

) I ( e F O N -I ( e F

Fig 5 Optical absorption spectra of the Fe(II)–NO (bold line) and Fe(II) (thin line) heme complexes of basic helix–loop–helix (bHLH)-PAS-A The position of the Soret band for the Fe(II)–NO complex (394 nm) suggests that it is a five-coordinated NO–heme complex.

Table 3 Resonance Raman spectra of Fe(II)–CO complexes of the

basic helix–loop–helix (bHLH)-PAS-A and PAS-A domains of

neuron-al PAS domain protein 2.

m Fe–CO

(cm)1)

m C–O

(cm)1) References

Fig 4 Inverse correlations between frequencies of mC–Oand mFe–CO

of resonance Raman spectra for the Fe (II)–CO complex The axial

ligand trans to CO of basic helix–loop–helix (bHLH)-PAS-A appears

to be His.

on the QCM electrode [19,20] It appeared to be worth

examining the interaction between the bHLH-PAS-A

domain and DNA To understand whether the

bHLH-PAS-A domain of NPAS2 can bind to DNA with the

E-box sequence, we injected the heme-bound

bHLH-PAS-A domain onto the E-box-bound sensor chip,

and the decrease in frequencies was observed with time

(Fig 9A) This confirms that, under the experimental

conditions used here, the heme-bound bHLH-PAS-A

domain bound to the DNA containing the E-box

sequence To examine whether the heme is required for

binding to the E-box sequence, we injected heme-free

bHLH-PAS-A onto the E-box-bound sensor chip In

this case, a decrease in the frequencies was not

observed (Fig 9B) To confirm that the binding to the

E-box sequence was specific, we examined the binding

of a mutant E-box sequence (wild type: CACGTG:

mutant GACGTC) Essentially no frequency shift was

observed when either the heme-bound or heme-free

domains were applied to the mutant E-box

(Fig 9C,D) Similarly, the PAS-A domain without the

bHLH domain did not bind to the E-box sequence

Also, the heme-bound bHLH-PAS-A domain did not

bind to a random DNA sequence (not shown)

Collec-tively, these results show that the bHLH-PAS-A

domain specifically binds to DNA containing an E-box

sequence under the experimental conditions used Because the bHLH-PAS-A forms mainly a dimer in solution (supplementary Fig 2S), it may bind to the E-box as a dimer

Discussion

The findings from the current study suggest that the bHLH domain assists and stabilizes heme binding by the isolated bHLH-PAS-A domain of NPAS2 In addi-tion, specific binding of the isolated bHLH-PAS-A domain to the E-box was observed only when it was bound to heme

The optical absorption spectra revealed that Fe(III), Fe(II) and Fe(II)–CO complexes of bHLH-PAS-A and PAS-A were six-coordinate low spin The Soret peaks

of Fe(III), Fe(II) and Fe(II)–CO complexes of bHLH-PAS-A, however, were red-shifted compared to those

of PAS-A, suggesting that there are direct or indirect interactions between the bHLH domain and the heme environment in the PAS-A domain In addition, the Soret absorption of PAS-A had a shoulder at approxi-mately 370–380 nm, which probably corresponds to free heme Therefore, it appeared that the affinity of the isolated PAS-A domain for heme is lower than that of the bHLH⁄ PAS-A domain This idea was

0 7 0 6 0 5 0 4 0

3

t i h n a m a

0 7 0 6 0 5 0 4 0 3

t i h n a m a

1361 1376

1508 1584

Fig 6 Effects of isotopically labeled NO molecules on resonance Raman spectra of the Fe(II)–NO complexes of basic helix–loop–helix (bHLH)-PAS-A in low-frequency (A) and high-frequency (B) regions The bottom lines show the 14 N 16 O complex, the middle lines the 15 N 16 O complex, and the top lines the difference spectra between the 14 N 16 O and 15 N 16 O complexes.

ported by the kinetic studies of heme binding, which

revealed that heme binding to the isolated PAS-A is

very slow and did not saturate under our experimental

conditions (Fig 7, Table 4)

The resonance Raman spectra showed that the heme

coordination states of Fe(III) complexes of the isolated

bHLH-PAS-A and the PAS-A domains were a mixture

of five-coordinated high spin and six-coordinated low spin Similarly, for the Fe(II) complexes, both are a mixture of five-coordinated high spin and

six-coordina-Table 4 Association and dissociation rate constants and equilib-rium parameters for heme binding to the basic helix–loop–helix (bHLH)-PAS-A and PAS-A domains and other heme proteins.

Proteins

k on

(mol)1Æs)1)

k off

(s)1)

K d

bHLH-PAS-A 3.3· 10 7 5.3 · 10)3 1.6· 10)10 This work

< 10 5 3.0 · 10)4

1.5 · 10)3 1.4· 10)10 Unpublished

observations b

SOUL 1.9· 10 6 6.1 · 10)3 3.2· 10)9 [17]

P22HBP 1.0· 10 8

4.4 · 10)3 4.4· 10)11 [17]

Sw Mb 7.6· 10 7 8.4 · 10)7 1.3· 10)14 [18]

a

Spectral changes observed for both association and dissociation were composed of two phases b Data for heme-regulated eIF2a kinase (HRI) were taken from the PhD thesis of J Igarashi (Tohoku University, Sendai, Japan).

0 0

0 0

0 0

0 4 0 4 0 4 0 4 0 3

) m n ( h t g n e l e v W

A

B 6 0

4 0

2 0

0 0 0

0 0

0 0

0 0

) s ( e m i T

Fig 8 Optical absorption spectral changes accompanying heme dissociation from basic helix–loop–helix (bHLH)-PAS-A and associ-ation to H64Y ⁄ V68F apomyoglobin mutant (A) and the spectral change upon Fe(III)–myoglobin formation as monitored at 410 nm with a mixture of Fe(III) bHLH-PAS-A and H64Y ⁄ V68F apomyoglobin (B).

B

C

0

x

0

0

0

0 4 0 4 0 4 0 4 0 4

0

3

0 5 0 5

0 2 5 1 0 1 5 0 0 0

) s ( e m i T

0

x

0

0

0

0 4 0 4 0 4 0 4 0 4 0

3

Wavelength (nm) Wavelength (nm)

0

0

0

0

0

0

0

0

kob

0 x

2 - 6 5

1 0

1 5

0 0

.

0

) M ( A -S A P / H L H b [

Fig 7 Optical absorption spectral changes accompanying

assoc-iation of Fe(II)–CO heme with heme-free basic helix–loop–helix

(bHLH)-PAS-A (A) and PAS-A (B) after mixing using a stopped-flow

spectrometer The inset in (A) shows the spectral change at

421 nm, which was composed of only a single phase The

correl-ation between kobs and the concentration of heme-free

bHLH-PAS-A is shown in (C).

ted low spin (Table 2) In addition, the inverse

correla-tion between mFe–COand mC–Ofrequencies revealed that

the ligand trans to CO in the bHLH-PAS-A domain

and PAS-A domain is His These spectral findings are

the same as those reported for the isolated PAS-A

domain [22]

In contrast to the effect of CO, the effect of NO on

the transcription activity of the NPAS2–BMAL1

heterodimer has not been previously reported We

obtained a typical optical absorption spectrum with a

Soret band at 394 nm and resonance Raman spectra

corresponding to the five-coordinated NO–Fe(II) heme

complex of bHLH-PAS-A (Figs 5 and 6) Some of

the heme-sensor proteins, including soluble guanylate

cyclase [23], CO-sensing heme-bound transcriptional

regulator from Rhodospirillum rubrum (CooA) [24],

cystathionine b-synthase [25], heme-regulated inhibitor

[26], and cytochrome c¢ [27], form five-coordinate NO–

heme complexes, and this modifies their function

Thus, it seems likely that NO binding to the heme

affects the DNA-binding properties of NPAS2

The rate constant for heme association (kon) with the isolated bHLH-PAS-A domain of NPAS2 was of the same order as those for heme association with the heme-regulated kinase inhibitor (unpublished observa-tions) and sperm whale myoglobin [18] Also, the rate constant for the association of heme with isolated bHLH-PAS-A of NPAS2 was much higher than that for the isolated PAS-A domain This further supports the idea that the bHLH region assists with the stable binding of heme to the PAS-A domain in the isolated bHLH-PAS-A protein

We also determined the rate constant for the dissoci-ation of heme (koff) from the isolated

holo-bHLH-PAS-A domain We found that the rate constant was similar

to those of other heme proteins (Table 4) On the basis

of these association and dissociation rate constants, we estimated the heme dissociation equilibrium constant (Kd) The apparent Kd value of heme for the isolated bHLH-PAS-A domain was much higher than that of sperm whale myoglobin, but comparable to that of heme-regulated kinase inhibitor (unpublished

∆F (Hz

s) ( e m i

s) ( e m i T s)

( e m i T

2 S A P N -o l o h

2 S A P N -o l o h

2 S A P N -o l o h

2 S A P N -o l o h

2 S A P N -o l o h

A

S

B

2 S A P N -o p a A S

2 S A P N -o p

a S 2 A P N -o p a 2 S A P N -o p a

B A

2 S A P N -o l o h

2 S A P N -o l o h

0

1

-0

-0

0

1

0 0 0 5 0 0 0 5 0 0 0 5 0

0

2 S A P N -o l o

h S 2 A P N

-o

l

o

h

0

A

S

B

2 S A P N -o l o h C

0 0

-0 5 -0

0 5

0 0

0 5 0

0 0

5 0

0

A S

B a p o - N P A S 2

2 S A P N -o p a

2 S A P N -o p a 2 S A P N -o p

D

Fig 9 Quartz crystal microbalance (QCM) analyses for the binding of holo (heme-bound)-basic helix–loop–helix (bHLH)-PAS-A to the E-box sequence (A), apo (heme-free)-bHLH-PAS-A to the E-box sequence (B), holo-bHLH-PAS-A to the mutant E-box sequence (C), and apo-bHLH-PAS-A to the mutant E-box sequence (D) Aliquots (5 lL) of holo-bHLH-apo-bHLH-PAS-A (2.55 lgÆlL)1) or apo-bHLH-PAS-A (1.80 lgÆlL)1) were added stepwise to 2 mL of buffer bathing the chip, allowing time for the frequency change to stabilize between each step Addition of the PAS-A domain lacking the bHLH domain did not change the frequency The DNA sequence containing the E-box was 5¢-GGGGCGCCACGTGA GAGG-3¢, and that containing the mutant E-box was 5¢-GGGGCGCGACGTCAGAGG-3¢ (E-box regions are underlined).

tions) Because the heme-regulated kinase inhibitor

responds to the heme concentration in cells by switching

the kinase reaction on or off, heme must bind to the

pro-tein reversibly Therefore, it is possible that in NPAS-2,

heme reversibly binds to the PAS-A domain Note that

the spectral change accompanying both association and

dissociation of heme for the isolated PAS-B domain of

NPAS2 was composed of two phases (Table 4)

The QCM data demonstrated that the isolated

bHLH-PAS-A domain binds to the E-box DNA

sequence under specific conditions In contrast, the

bHLH-truncated PAS-A domain did not bind to the

E-box, and the isolated bHLH-PAS-A domain did not

bind to the mutated E-box, indicating that the binding

of the isolated bHLH-PAS-A domain to the E-box

sequence is specific In addition, a previous study

showed that full-length human NPAS2 alone, without

a partner such as BMAL1 or ARNT, is unable to bind

to the E-box sequence [13] and that mouse NPAS2

alone does not activate circadian rhythm-associated

transcription [7] Thus, truncation of the C-terminal

domain and the binding of heme may affect the

DNA-binding properties of the bHLH-PAS-A fragment

We further examined the binding of CO to the Fe(II)

bHLH-PAS-A domain (5 lm heme) of NPAS2, but the

increase of the Soret absorption upon CO binding was

composed of more than two phases (data not shown)

Also, the initial phase was not dependent on the CO

concentration at high concentrations (200–500 lm)

Taking into account only initial CO-dependent (10–

90 lm CO) CO binding, which is a straight line, we

estimated that the CO binding rate is 0.13 lmol)1Æs)1

This value is slightly smaller than that reported by

Dioum et al (0.37 lmol)1Æs)1) [15] Resonance Raman

spectroscopy showed that the heme coordination state

is a mixture of five-coordinate and six-coordinate and

that the protein exists in solution in an equilibrium

between the monomer, dimer and trimer These factors

may contribute to the complicated kinetic behavior

Further studies are required to address this issue

Based on resonance Raman spectral studies of

His119fiAla, His138fiAla, His171fiAla and

Cys170fiAla mutants of the isolated PAS-A domain, it

has been suggested that His119 and Cys170 are the axial

ligands for the Fe(III) complex, whereas His119 and

His171 are the axial ligands for the Fe(II) complex [22]

Note that some sensor proteins, including PAS proteins,

are known to have substantial flexibility [28–31]

In summary, the present study suggests that the

bHLH domain plays an important role in assisting and

stabilizing heme binding to the PAS-A domain in the

isolated bHLH-PAS-A domain of NPAS2 The QCM

data indicated that the isolated bHLH-PAS-A domain

specifically binds to the E-box DNA sequence Further studies using both NPAS2 and BMAL1 are needed to elucidate the mechanism of DNA binding by NPAS2

Experimental procedures

Materials Oligonucleotides (18 bp; E-box, random) and 5¢-biotinylated oligonucleotides (18 bp) were synthesized by the Nippon Gene Institute (Sendai, Japan) Restriction enzymes and modification enzymes were purchased from Takara Bio (Otsu, Japan), Toyobo (Osaka, Japan), New England Bio-labs (Beverly, MA), and Nippon Roche (Tokyo, Japan) Other chemicals weer of the highest grade available and were purchased from Wako Pure Chemicals (Osaka, Japan)

Plasmid construction

To construct the plasmid coding for the N-terminal domain, the cDNA of NPAS2 was generated by RT-PCR using RNA isolated from mouse forebrain [21,22] The sequences of the PCR products were confirmed by deter-mination of the nucleotide sequence by Sanger’s method using a DSQ-2000 L automatic sequencer (Shimadzu Co., Kyoto, Japan) The 6xHis-tagged isolated bHLH-PAS-A domain (amino acids 1–240) and isolated PAS-A domain (amino acids 78–240) of NPAS2 were created by subcloning into the NdeI and SalI sites of the of pET-28a(+) expres-sion vector (Novagen, Madison, WI) E coli strain BL21 (DE3) codon plus RIL was transformed with the expression vectors pET28a-PAS-A or pET28a-bHLH-PAS-A

Protein expression and purification

pET28a-bHLH-PAS-A were incubated in Terrific Broth

600 nm reached 0.6 Expression of the isolated bHLH-PAS-A and bHLH-PAS-A domains of NPAS2 was then induced

isopropyl-1-b-thiogalactoside and mild shaking The E coli cells were then suspended in buffer A (50 mm sodium phosphate (pH 7.8), 50 mm NaCl, 10% glycerol, 1 mm EDTA, 1 mm

pepstatin A, 2 mm 2-mercaptoethanol) and lysed by pulse sonication (three times for 2 min each with 2 min inter-vals) on ice using an Ultrasonic Disruptor UD-201 (Tomy Seiko, Tokyo, Japan) Hemin (100 lm final concentration) dissolved in 0.01 m NaOH was added to this lysate, and the mixture was allowed to equilibrate on ice for 30 min After centrifugation at 35 000 g for 30 min, the

superna-tant was adjusted to 20% saturated ammonium sulfate

and incubated for 30 min on ice After centrifugation at

35 000 g, the supernatant was adjusted to 70% saturated

ammonium sulfate Precipitates from the 70% saturated

solution were collected by centrifugation and dissolved in

buffer A The excess ammonium sulfate was removed and

the buffer was exchanged by applying the solution to

Sephadex G-25 (100 mL) that had been pre-equilibrated

with buffer B (50 mm sodium phosphate (pH 7.8), 50 mm

NaCl, 10% glycerol, 2 mm dithiothreitol) The resulting

solution was applied to an Ni-NTA column (Qiagen,

Hil-den, Germany) pre-equilibrated with buffer C (50 mm

sodium phosphate (pH 7.8), 50 mm NaCl, 2 mm

2-merca-ptoethanol) The column was washed stepwise with buffer

C containing 0, 20 and 50 mm imidazole The protein

eluted at 100 and 150 mm imidazole was collected and

concentrated To remove the excess imidazole and to

exchange the buffer with buffer D (100 mm Tris⁄ HCl

applied to a HiTrap desalting column (Amersham

Bio-sciences, Uppsala, Sweden)

Removal of the His tag

Purified His-tagged protein in 100 mm Tris⁄ HCl buffer

(pH 8.0) was equilibrated for 1 h on ice with 1–2 equivalents

of hemin dissolved in 0.01 m NaOH Excess hemin was then

removed using a Bio-Gel P-6 column (Bio-Rad, Hercules,

CA) with the same buffer Thrombin protease (10 units⁄ mg)

was added to the heme-saturated His-tagged protein in

50 mm Tris⁄ HCl (pH 8.0), 150 mm NaCl and 2.5 mm

was applied to an Ni-NTA column pre-equilibrated with

buffer D, and proteins were eluted with the same buffer

The purified protein was rapidly frozen in liquid nitrogen

checked by Coomassie Brilliant Blue R250 dye binding

Gel filtration

To determine the molecular mass, gel filtration was carried

out using an AKTA liquid chromatography apparatus

equipped with a Superdex75 HR 10⁄ 30 column (Amersham

Biosciences) The buffer used for gel filtration was 100 mm

lar mass was estimated by correlation between the

molecu-lar mass and the elution volumes for the following standard

proteins: albumin (67 kDa), ovalbumin (43 kDa),

chymo-trypsinogen A (25 kDa), and ribonuclease A (13.7 kDa)

Optical absorption spectra

Spectral experiments were carried out under aerobic

condi-tions using a Shimadzu UV-2500 spectrophotometer

spectral experiments were conducted using a Shimadzu

reduced by sodium dithionite, the sample solution was sat-urated with argon gas

Resonance Raman spectra The bHLH-PAS-A domain of NPAS2 (35 lm in 100 mm

air-tight spinning cell with a rubber septum and reduced by the addition of sodium dithionite (10 mm final concentration) 12

Laboratories, Andover, MA) gas was introduced into the Raman cell with an airtight syringe Raman scattering was excited at 413.1 nm with a Kr ion laser (BeamLok 2060, Spectra-Physics, Mountain View, CA) The excitation light was focused into the cell at a laser power of 5 mW for the Fe(III) and Fe(II) complexes For the CO–Fe(II) complexes,

to avoid photolysis, the laser power was 0.1–0.2 mW Raman

SPEX750M single polychromator (Jobin Yvon, Longjum-eau, France) Raman shifts were calibrated with indene,

Heme-binding kinetics Heme association experiments were carried out using an

and 50 mm NaCl and was purged with nitrogen gas for

30 min The buffer was then saturated with CO gas CO–

Reactions between the protein and CO–Fe(II) hemin were monitored at 421 nm [18]

Heme dissociation experiments were conducted using a

with a temperature controller Dissociation of heme from the Fe(III) bHLH-PAS-A domain of NPAS2 was examined

as Fe(III) myoglobin formation by monitoring the increase

in absorbance at 410 nm upon mixture of the Fe(III) bHLH-PAS-A domain of NPAS2 (3 lm) with a 10-fold

phosphate buffer (pH 7.0) containing 0.6 m sucrose at

CO-binding kinetics

To measure the CO association rates, the bHLH-PASA domain of NPAS2 (10 lm) was reduced with sodium

solu-tion was then rapidly mixed with controlled CO-saturated buffer (c 1 mm CO) using a stopped-flow