Báo cáo khoa học: H NMR study of the molecular structure and magnetic properties of the active site for the cyanomet complex of O2-avid hemoglobin from the trematode Paramphistomum epiclitum pdf

The distal Tyr66E7 is found oriented out of the heme pocket in solution as found in both crystal structures.. Comparison to the alternate crystal structures The pattern of NOESY cross pe

H NMR study of the molecular structure and magnetic properties

Weihong Du1, Zhicheng Xia1, Sylvia Dewilde2, Luc Moens2and Gerd N La Mar1

1

Department of Chemistry, University of California, Davis, CA, USA;2Department of Biomedical Sciences,

University of Antwerp, Wilrijk, Belgium

The solution molecular and electronic structures of the active

site in the extremely O2-avid hemoglobin from the trematode

Paramphistomum epiclitum have been investigated by 1H

NMR on the cyanomet form in order to elucidate the distal

hydrogen-bonding to a ligated H-bond acceptor ligand

Comparison of the strengths of dipolar interactions in

solution with the alternate crystal structures of

methemo-globin establish that the solution structure of wild-type Hb

more closely resembles the crystal structure of the

recom-binant wild-type than the true wild-type met-hemoglobin

The distal Tyr66(E7) is found oriented out of the heme

pocket in solution as found in both crystal structures

Ana-lysis of dipolar contacts, dipolar shift and paramagnetic

relaxation establishes that the Tyr32(B10) hydrogen proton

adopts an orientation that allows it to make a strong H-bond

to the bound cyanide The observation of a significant iso-tope effect on the heme methyl contact shifts confirms a strong contact between the Tyr32(B10) OH and the ligated cyanide The quantitative determination of the orientation and anisotropies of the paramagnetic susceptibility tensor reveal that the cyanide is tilted
10
from the heme normal

so as to avoid van der Waals overlap with the Tyr32(B10)

Og The pattern of heme contact shifts with large low-field shifts for 7-CH3and 18-CH3is shown to arise not from the 180
rotation about the a-c-meso axis, but due to the
45
rotation of the axial His imidazole ring, relative to that in mammalian globins

Keywords: hemoglobin; trematode; H-bonding; dipolar shift; NMR

Globins (hemoglobin, Hb, and myoglobin, Mb) are ferrous

heme-containing, O2-binding proteins found widespread in

nature [1,2] They exhibit an extraordinary range of ligation

rates and affinities, as well as autoxidation rates (conversion

to the nonfunctional ferric hemin) in spite of a highly

conserved folding topology (the Mb fold) The majority of

globins, which consists of
150 residues, are arranged in a

compact globule consisting of eight (A–H) helices, with the

heme wedged between the E and F helices A completely

conserved His F8 (eighth residue on helix F) provides the

only covalent bond to the protein, although a conserved

aromatic ring (CD1) (first residue and the loop between

helices Cand D) provides considerable stabilization by

p-stacking on the heme [1–3] Some recently discovered

truncated (
100–120 residues) globins from bacteria

exhibit the general Mb fold but retain only four of the

helices, leaving a largely conserved active site with respect to more conventional globins [4,5], and one has an unprece-dented Tyr (CD1) [3] The modulation of the extreme range

of ligation rates in monomeric globins appears to be controlled primarily by limited sets of residues on the distal (opposite side to the proximal His F8) side of the heme, which determine the distal pocket polarity, provide stabi-lizing H-bonds to ligands and/or sterically interfere with ligand binding [6] Among the extensively studied mono-meric mammalian Mbs, the key residues have been identi-fied at positions E11 and E7 (generally His, but occasionally Gln [7]), where the latter provides the crucial H-bond to stabilize O2 Val E11 makes van der Waals contact with the ligand and, in certain cases, may sterically destabilize ligands [6,7] The B10 residue is generally a Leu, except in elephant

Mb [8,9], where a Phe at B10 (and a Gln at E7) results in an

Mb with a relatively reduced autoxidazibility but conserved

O2affinity relative to other mammalian Mbs

The globins from invertebrates exhibit much more vari-ability in both the nature of the distal residues that provide the stabilizing H-bond to the O2and their positions in the globin [10–17] Thus the sea hare Aplysia limacina possesses

a Val E7, but H-bond stabilization of O2is provided by Arg E10 [18] The monomeric Hb from Glycera dibranchiata possesses a Leu E7, and the absence of an alternate H-bond donor leads to rapid O2off-rates [19] Some of the most unusual Hbs characterized to date are those from nematodes and trematodes (mammalian parasites) such as the nematode Ascaris suum(As) [11,12] and the trematode Paramphisto-mum epiclitum (Pe) [14,15], which exhibit extraordinarily

Correspondence to G N La Mar, Department of Chemistry,

University of California, One Shields Avenue, Davis, CA 95616, USA.

Fax: + 1 530 752 8995, Tel.: + 1 530 752 0958,

E-mail: lamar@indigo.ucdavis.edu

Abbreviations: Hb, hemoglobin; Mb, myoglobin; WT, wild type; rWT,

recombinant WT; metHbCN, cyanide ligated ferric hemoglobin;

NOE, nuclear Overhauser effect; DSS,

2,2-dimethyl-2-silapentane-5-sulfonate; WEFT, water-eliminated Fourier transform;

Pe, Paramphistomum epiclitum.

Note: a website is available at http://www.chem.ucdavis.edu/faculty/

(Received 20 December 2002, revised 24 March 2003,

accepted 28 April 2003)

high O2affinity via an exceptionally slow rate Their

off-rates are too slow to allow these globins to participate in

aerobic respiration, and their exact physiological role is

openly debated [14,15,20] The nematode/trematode globins

share the property of a Tyr at position B10, which in

conjunction with or without the E7 H-bond donor, provides

an extremely strong H-bond to bound O2[11,12,16] It must

be noted, however, that a Tyr B10 occurs in other globins

such as native Lucina pectinata (Lp) Hbs [21], with
ordinary

O2affinity, where the B-helix is too far from the heme to

allow a strong H-bond A Tyr B10 has been incorporated

into a mammalian Mb variant, where its relatively minor

influence on O2affinity was attributed to the B-helix being

too close to the heme to allow a robust H-bond to O2[22]

The extremely O2-avid monomeric Pe Hb is unique in that it

possesses Tyr at both the B10 and E7 positions [14] Solution

1H NMR on Pe WT HbO2[23] had found two labile protons

in the vicinity of the bound O2, both arising from Tyr,

implying that both residues are oriented into the heme

pocket However, mutagenesis on Pe Hb [16] has shown that

substituting Tyr E7 does not, while substituting Tyr B10

does, strongly reduce O2 affinity via a faster O2 off-rate

These studies firmly establish that the O2avidity does not

require an H-bond by Tyr E7 However, these results do not

alone establish whether the Tyr E7 is oriented into or out of

the heme pocket

While crystals of Pe HbO2 have not been prepared to

date, the oxidized Pe wild type (WT) and recombinant

wild-type (rWT) metHb have crystallized in two different

forms [16,17], and the detailed structures provide

import-ant information on the novel Hb The Tyr66(E7) ring is

oriented out of the heme pocket in both forms with the

Tyr32(B10) in one of the structures [16,17] serving as a

H-bond acceptor to a ligated water molecule The two

structures of WT metHb and rWT metHbH2O, exhibit

significant differences in the interaction of the FG loop

with the heme and in the position of the B-helix, with

Tyr32(B10) closer to the iron by 1–2 A˚ in WT than rWT

metHb, such that the ligated water is lost [16,17] The

sizable structural accommodation to crystal forms for Pe

metHb is unprecedented and reflects a surprising

plasti-city whose functional relevance is unknown The

differ-ences in the two structures could be rationalized by

interactions between two molecules in the unit cell in one,

but not the other, crystal form [16,17]

Spectral congestion precludes more definitive1H NMR

structural studies of diamagnetic Pe WT HbO2at present

[23], and the molecule does not crystallize Hence, we have

embarked on a1H NMR study of the paramagnetic Pe WT

metHbCN form whose ligand, like O2, is an (albeit weaker)

H-bond acceptor [24–27], and whose tilting from the heme

normal reflects steric repulsive and/or H-bonding attractive

interactions in the distal pocket [24,28–30] The large

hyperfine shifts, moreover, provide highly enhanced

resolu-tion to the active site residue protons that greatly facilitates

their assignment, and whose hyperfine shifts contain a

wealth of information on the detailed molecular structure

not readily obtained in an analogous diamagnetic complex

[30,31] At the same time, the paramagnetic-induced

relax-ation is sufficiently weak so as not to interfere with effective

conventional 2D NMR experiments [32] that will confirm

whether the active site structure is closer to one or the other

crystal forms, or distinct from both The hyperfine shifts of the heme are sensitive to the presence of an H-bond to ligated cyanide [24–26,33], and hence provide direct infor-mation on distal H-bonding interactions Lastly, a relatively robust interpretive basis of the hyperfine shift of the heme and active site residue for low-spin ferric globins in terms of distal ligand tilt [28–30] and axial His orientation [26,34–38] has been established on the basis of characterizing a variety

of hemoproteins with different properties The Pe metHb provides a novel His F8 [16] orientation that will allow further testing of this procedure

Experimental procedures Protein samples

The monomeric wild-type (WT) hemoglobin, labeled WT

Hb, from the trematode Paramphistomum epiclitum (Pe) and Isoparorchis hypselbagi (Ih) were isolated and purified

as described previously [14] The cyanomet complexes were prepared by adding approximately five molar equivalents of KCN to the air-oxidized Hb The final concentration of Pe metHbCN complex was
2 mMand that of Ih metHbCN was 0.2 mM The1H2O solution was subsequently converted

to 2H2O solution using an Amicon ultrafiltration cell Solution pH was adjusted with NaO1H (NaO2H) or1HCl (2HCl) solution

NMR spectroscopy

1H NMR data were collected on a Brucker AVANCE 600 spectrometer, operating at 600 MHz for protein samples in both1H2O and2H2O over the temperature range 15–35
C,

at a repetition rate of 1 s)1with presaturation of the solvent signal Water-eliminated Fourier transform (WEFT) [39] spectra were recorded to detect broad, strongly relaxed proton signals Chemical shifts were referenced to 2,2-dimethyl-2-silapentane-5-sulfonate (DSS) through the water resonance Non-selective T1s, with ± 15% uncertainty, were determined for the resolved strongly relaxed protons at 25
Cfrom the initial magnetization recovery of a standard inversion-recovery pulse sequence The distance of proton H (with

T1) from the iron, RFeH, was estimated from the relation:

RFeH¼ RFeH½T1=T1i1=6 ð1Þ where R*Fe-His the distance for a reference proton with T1* Using both the heme 18-CH3 for H* (R*Fe¼ 5.88 A˚,

T1*¼ 140 ms) and F8 NdH (R*Fe¼ 5.01 A˚,

T1*¼ 37 ms) as reference The 1 : 1 [40] and steady-state NOE spectra were recorded at 30
Cas described previously

in detail [41] NOESY [42] and TOCSY [43] spectra were collected (512t1blocks, 2048t2points) at 25
C , 30
Cand

35
Cin order to identify scalar and dipolar connectivities among heme and amino acid residues Spectral widths of

13 kHz and mixing times of 80 ms for NOESY and 50 ms for TOCSY were used Scans (192) were collected for each block with a repetition time of 1.0 s)1 Two-dimensional data sets were processed byXWINNMRsoftware on a Silicon Graphics Indigo workstation Both NOESY and TOCSY spectra were processed by a 30
shifted sine bell squared apodization, and zero-filled to 2048· 2048 points prior to Fourier transformation

Magnetic axes determination

Experimental dipolar shifts for the structurally conserved

residues and backbone protons were used as input to search

for the Euler rotation angles a, b and c These transform the

molecular pseudosymmetry coordinates x¢, y¢ and z¢ (Fig 1)

obtained from crystal coordinates [16,17] into the magnetic

axes x, y and z, by minimizing the error function according

to the following equation [28–30,44]:

½jddip(obs) ddip(calc)Fða; b; cÞj2 ð2Þ where

ddipðcalcÞ ¼ ð12plNoÞ1½2Dvaxð3cos2h0 1ÞR3

þ 3ðDvrhsin2h0cos 2X0ÞR3 ð3Þ

The observed dipolar shifts are given by:

ddipðobsÞ ¼ dDSSðobsÞ  dDSSðdiaÞ ð4Þ where ddip(obs) and dDSS(dia) are the chemical shifts (in p.p.m.) referenced to DSS, for the paramagnetic Pe metHbCN complex and an isostructural diamagnetic com-plex, respectively Limited dDSS(obs) are available from Pe HbO2 [23]; for other residues dDSS(dia) may be reliably estimated from the available molecular structure [8,28,29]

as follows:

dDSSðdiaÞ ¼ dtetrþ dsecþ drc ð5Þ where dtetr, dsecand drcare the chemical shifts of an unfolded tetrapeptide relative to DSS [45], the effect of secondary

Fig 1 Schematic representation of the heme pocket of Pe Hb as found in the crystal structure of Pe metHbH 2 O [16]and as confirmed herein by solution 1 H NMR of Pe metMbCN Proximal and distal residues are shown as rectangles and circles, respectively, and arrows connecting heme substituents and residues, and between residues, reflect NMR observed, and crystallographically expected, contacts The reference, x¢, y¢, z¢, as well

as the magnetic coordinate systems, x, y, z, are shown, where b represents the tilt of the major magnetic axis, z, away from the heme normal (z¢ axis),

a is the angle between the projection of the tilt of the z axis on the x¢, y¢ plane (defined direction of tilt of z and the x¢ axis), and j
a + c defines the location of the rhombic (x, y) axes The orientation of the axial His imidazole plane relative to the heme is given by /.

structure [46] and the heme-induced ring current shift [47],

respectively

Structural modeling

Protons were added to the crystal coordinates of

recom-binant WT Pe metHbH2O [16] and WT metHb [17] using

the program INSIGHT-II (Accelrys) This provided unique

coordinates for all protons of interest except the Tyr32(B10)

hydroxyl proton, as its position was determined from the1H

NMR spectral parameters

Results The1H NMR spectra of Pe WT metHbCN in1H2O and

2H2O are illustrated in Fig 2A,B A WEFT spectrum [39] designed to emphasize strongly relaxed signals is shown in Fig 2C Comparison of the traces in Fig 2A,B reveals the presence of two strongly relaxed labile protons at 32.5 p.p.m (T1
10 ms) and 17.5 p.p.m (T1
35 ms),

as well as a weakly relaxed one at 11.4 p.p.m and several inconsequentially relaxed peaks in the 11–9 p.p.m win-dow The2H2O WEFT trace in Fig 2Clocates a broad

Fig 2.1H NMR spectra (600 MHz) of Pe WT metHbCN at 30
C, pH » 7.0 (A) Relaxed (repetition rate 1 s)1) reference trace in1H 2 O; (B) relaxed, reference trace in 2 H 2 O; (C) WEFT spectrum (relaxation delay 30 ms, repetition rate 10 s)1) in 2 H 2 O which allows detection of strongly relaxed, broad signs at 9 p.p.m.; and (D) steady-state NOE difference trace at 35
Cupon saturating the Tyr32(B10) OH signal (vertical arrow) Heme resonances are labeled as shown in Fig 1, and residues are labeled by position numbers and protons.

(
300 Hz) and very strongly relaxed (T1< 5 ms) peak

on the low-field shoulder of the diamagnetic envelope

Heme pocket residue assignments were pursued by

backbone connectivities as standard in a diamagnetic

protein [48], with the remainder of the target residues

assigned by detection of NOESY residue–heme and

interresidue cross-peaks solely by the standard Mb fold,

and the observation of relaxation effects and/or sufficient

TOCSY cross-peaks to identify the side chain uniquely

The identification of hyperfine shifted and relaxed

reso-nances was greatly facilitated by variable temperature

studies to define unique scalar/dipolar connectivities as

described previously [49] Assignments deduced herein are

given by the heme-labeling scheme shown in Fig 1 and

the residue numbers and proton The chemical shifts for

the heme and axial His98(F8) are listed in Table 1, and

those for assigned residues with significant dipolar shifts

are listed in Table 2 T1 values for predominantly

paramagnetically influenced protons are given in

paren-theses

Heme assignments

TOCSY spectra (not shown) identify four (two three-spin

and two four-spin) hyperfine shifted and relaxed spin

systems with dipolar contacts (not shown) to four strongly

temperature-dependent methyl peaks [two resolved (Curie

behavior) and two nonresolved methyl peaks (antiCurie

behavior)], that uniquely identify the pyrrole substituents

[30] Dipolar contacts to adjacent meso-Hs (5-H, 10-H,

15-H, 20-H), with their unique low-field intercept at

T)1¼ 0, locate the four meso-Hs (as listed in Table 1)

Sequence-specific assignments The detection of the Ni–Ni+1, ai–Ni+1, bi–Ni+1, ai–Ni+3 and/or ai–bi+3 NOESY connection diagnostic of helical fragments [48] with limited (but sufficient) TOCSY-identi-fied side chains leads to the identification of the six segments labeled I–VI (Fig 3) Fragment I is represented by

Zi-AMXi+1-Glyi+2-Zi+3-AMXi+4-AMXi+5-, where Z is

>4 spin side chain, and AMXi+1and AMXi+4 exhibit significant low-field dipolar shifts The relaxed, low-field labile proton at
17 p.p.m exhibits a NOE to the AMXi+5 and its peptide NH that is unique for the proximal His98(F8), and AMXi+1is in dipolar contact with a two-spin aromatic ring, as expected for Tyr94(F4); this identifies

I as Gln93–His98 of the proximal F-helix(F3–F8) Further backbone (nonhelical) dipolar connections (Fig 3) allow the adjacent assignments of Thr99–Val103, the residues that constitute the FG corner (FG1–FG5), with expected dipolar contacts to pyrroles B and C(Fig 1) and residues in the Cand G helices (see below) Fragment II is represented

by AMXi-Alai+1-Zi+2-Thri+3-Leui+4-Zi+5-Zi+6-Alai+7, which the sequence uniquely identifies as the expected E-helix (E7–E16) segment Tyr66–Val75, with the ThrE10 and LeuE11 side chains exhibiting weak-to-moderate upfield shifts The detection of an inconsequentially shifted, two-spin aromatic ring in contact with AMXi (Fig 4E) confirms Tyr66(E7) Key dipolar contacts that define the orientation of the Tyr66(E7) ring include those to the 13-propionate Hbs (Fig 4B,C) and to the Phe46(CD1) ring, and the NHs of Arg48(CD3) and Leu49(CD4) (Fig 4G)

It was not possible to detect a labile proton in dipolar contact with the Tyr66(E7) CeHs signal that would identify the side chain hydroxyl proton The observed heme-residue contacts are characteristic of the general Mb fold (Fig 1)

Fragment III is detected as Glyi-Zi+1-Xi+2-Ilei+3 -AMXi+4-Thri+5-Zi+6-AMXi+7-AMXi+8 (Fig 3) where contacts of three-spin aromatic rings to AMXi+4 and AMXi+8 and a two-spin aromatic ring to AMXi+7 uniquely identify Gly111–Phe119 on the G-helix (G8– G16) An additional AMX spin-system connected to a hyperfine shifted aromatic ring identifies a Phe, and its backbone exhibits the ai-3-Nicross peak to AMXiin III (Gly111); this identifies it as Phe108(G5) The side chains exhibit the expected NOESY cross peaks to the pyrrole A/B junction and to the E-helix as depicted in Fig 1 Frag-ment IV, Vali-AMXi+1-Xi+2-Zi+3-Zi+4-Alai+5-Alai+6

-Zi+7-Zi+8(Fig 3), is unambiguously identified as Val139– Ile147 on the H-helix (H15–H23) The dipolar contacts of Phe140(H16) and Met143(H19) to pyrrole A, as well as Ile147(H23) to the axial His98(F8), and the interresidue contacts to the G-helix (Fig 1) confirm their locations in

a standard H-helix

The helical fragment V is represented by Zi-AMXi+1

-Zi+2-Zi+3-AMXi+4 (Fig 3), where dipolar contacts of a two-spin aromatic ring to AMXi+1(and to the 7-CH3and 8-vinyl; Fig 1) identifies the Gln41–His45 fragment on the C-helix (C3–C7), with the expected moderate dipolar shifts, and which is in contact with the pyrrole B/Cjunction (Fig 1) Backbone NOESY connections allow the extension

of sequential assignment of fragment V to include AMXi+5 -Ser that must arise from Phe46(CD1) and Ser47(CD2)

Table 1.1H NMR spectral parameters for the heme and His98(F8)

signals in Pe metHbCN Chemical shifts in p.p.m are referenced to

DSS in 1 H 2 O 100 m M phosphate, pH 6.8, 30
C Non-selective T 1 , in

ms, in square brackets for resolved resonances.

Protons d DSS (obs) [T 1 ]

7-CH 3 23.64 [88]

12-CH 3 7.97 18-CH 3 14.10 [143]

3-H a 12.40 [182]

3-H b s )4.88 [253], )4.23 [235]

8-H a 8.83 8-H b s )1.22 [186], 0.29 13-H a 14.32 [110], 5.44 13-H b s )3.34 [152], )2.63 [165]

17-H a s 13.29 [92], 4.86 17-H b s )3.80 [130], )169 [155]

His98(F8) NH 12.81 [129]

C a H 8.05

C b H 8.67

C b H 10.59 [84]

N d H 17.51 [37]

Table 2. 1H NMR spectral parameters for strongly dipolar shifted

active site residues in Pe metHbCN Observed chemical shifts,

d DSS (obs), in p.p.m., are referenced to DSS in 1 H 2 O, 100 m M

phos-phate, pH 6.8 at 30
C Diamagnetic chemical shifts, d DSS (dia),

cal-culated via Eqn (5) using the WT Pe metHb H 2 O crystal coordinates

[16] na, Not assigned.

Residue Proton d DSS (obs) d DSS (dia)

C e Hs 11.79 5.48

C b Hs 3.43, 3.32 3.12, 2.80

C b H 3.66 3.78, 3.36

C b H 1.95 3.21, 2.98

C b H 2.15 2.75, 2.67

C b H 3 2.37 1.06

C c H 3 0.15 1.05

C d H 3 2.63 )0.81

C d H 3 ¢ 2.09 )1.10

C b H 3 )0.03 1.70

Table 2 (Continued).

Residue Proton d DSS (obs) d DSS (dia)

C b H¢ 3.25 2.76

C d Hs 7.40 7.71

C e Hs 6.57 7.16

C a H¢ 6.80 2.26

C b H 2.90, 2.79 1.62, 1.49

C c Hs 2.13 1.09

C d Hs 2.30 1.47

C e Hs 2.47 2.73

C b H¢ 6.54 1.50

C b H¢ 10.59 0.67

N d H 17.51 6.42

C c H 3 2.12 0.68

C c H 3 0.54 )0.09

C c H 3 ¢ )0.05 1.21

C d Hs 4.22 5.63

C a H¢ 2.92 3.50

C b H 3.01, 2.82 3.69

C b H¢ 2.82 3.31

C d Hs 6.52 7.10

C e Hs 6.82 6.96

C b H 2.16 3.01, 3.31

C d Hs 6.19 7.39

C e Hs 4.95 7.46

C b Hs 2.28 0.98, 1.34

C e H 3 0.48 1.27 Ile147(H23) C a H 4.38 3.15

C c Hs 2.78 0.18

C c H¢ 1.22 )0.15

C d H 3 0.54 )2.44

Dipolar contact to a strongly relaxed and moderately

dipolar shifted aromatic spin-system that is in contact with

pyrrole C(Fig 4B,C) confirm both the assignment as

Phe(CD1) and the orientation of the heme, as depicted in

Fig 1 and as found in the crystal structure The CfH of

Phe46(CD1) exhibits the strong relaxation (T1
20 ms)

characteristic for this residue

The remaining helical segment VI, Zi-Thri+1-Glyi+2

-Zi+3-Glyi+4-Alai+5-AMXi+6-AMXi+7-Alai+8-Zi+9

-AMXi+10-Thri+11-Alai+12 (Fig 3), is unique to residues

Glu26–Ala38 on the B-helix (B4–B16) The dipolar contact

of a weakly shifted, three-spin aromatic ring to AMXi+10,

and that of a strongly hyperfine shifted two-spin aromatic

ring to AMXi+6 confirm the assignments of Phe36(B14)

and the key Tyr32(B10) These side chains do not exhibit

NOESY cross peaks to the heme (as expected), but exhibit

the expected contacts to helix E (Tyr32(B10) to Leu70(E11),

Fig 4D; Ala67(E8), Fig 4F; and Tyr66(E7), Fig 4G, as

depicted in Fig 1 A strong NOE to the assigned Tyr32(B10

CeHs signal, upon saturating the extreme low-field, strongly

relaxed (T1
10 ms) labile proton signal (Fig 2D), locates

the residue side chain hydroxyl proton

Comparison to the alternate crystal structures The pattern of NOESY cross peaks of the Tyr66(E7) ring

to the heme (Fig 4A–C), Tyr32(B10) (Fig 4F), and in particular, to the Phe46(CD1) backbone (Fig 4E), as summarized in Fig 1, unequivocally establish that the Tyr66(E7) ring is oriented out of the heme pocket, exactly

as found in both Pe metHb crystal structures [16,17] The position is further supported by the calculated and observed small ddip (and hence, negligible temperature-dependence to its shifts) for the crystallographic orienta-tion of the Tyr66(E7) ring (see below) While its hydroxyl proton could not be located by its characteristic strong NOE to the definitively assigned CeHs, most likely due to its lability, there is no orientation of the OH group that can bring it close enough (>5 A˚) to interact with the bound cyanide

The interresidue and residue-heme NOESY cross peak pattern that led to the schematic representation of the Pe metHbCN heme cavity structure in Fig 1 is equally consistent with qualitative expectations of either of the two Pe metHb crystal structures [16,17] It is only upon quantitative consideration of cross peak intensities that such detailed structural distinctions can be made The two X-ray structures, one of WT and the other of rWT metHb, exhibit differences that include important por-tions of the protein that we have characterized above Thus parts of the FG corner move away from the heme and the B-helix moves closer to the heme in the WT metHb than in the rWT metHbH2O crystal structure, with the result that the ligated water is lost in the latter complex [16,17]

Prior to determining the magnetic axes, which will allow

us to elaborate the tilt of the ligated cyanide and charac-terize the H–bonding interaction of Tyr32(B10) with its ligand, it is necessary to establish which crystal structure better represents the solution structure This distinction can

be made on the basis of three NMR observations, the NOESY cross peak intensities between proton i and j (/ rij)6), the paramagnetism-induced relaxation,

T11i / R6Fei;, of proton i, and the magnetic axes themselves (see next section)

Inspection of the two sets of crystal coordinates identifies

a series of proton pairs whose separations differ significantly between the two structures [16,17] and which we have been able to identify unambiguously The contacts involve the

FG corner and the position of helix B relative to the heme and E-helix backbone Table 3 lists the alternate rijfor eight sets of proton pairs in the two structures, as well as the observed NOESY cross peak intensity (s, < 2.5 A˚; m, 2.5– 4.6; w > 4.0 A˚) In each case, the distance in bold is the one

in better agreement with the experiment, and each of the four distances dictate that the solution structure of WT metHbCN is consistent only with the crystal structure of rWT metHbH2O [16] (except for a labile proton on Tyr32(B10), see below) The alternate structures predict characteristic relaxation time differences for several proton sets in the alternate crystal structure, i.e Tyr32(B10), Thr99(FG1), Val103(FG5), but in only one case is the key resonance resolved so that its T1can be quantitated Thus the movement of the B-helix towards the heme in the WT metHb relative to that in the rWT metHbHO crystal

Fig 3 Schematic representation of the sequential NOESY cross peak

pattern for the six characterized helical fragments I–VI that identify key

sections of the F, E, G, H, C and B helices, respectively.

structure leads to the reduction of the RFe-ifor the two Tyr

B10 CeHs of 5.8 and 7.9 A˚ in rWT metHbH2O (with

expected T1
20–30% shorter than that for a heme

methyl), to 4.5 and 5.8 A˚ in WT metHb such that a T1is

expected closer to that of a meso-H (T1
50 ms) The

observed T1(Tyr32(B10) (CeH)
100 ms, is consistent with

the former, but not the latter distances, such that the

relaxation effects similarly confirm a WT metHbCN

solution structure similar to the rWT metHbH2O, but not

the WT metHb crystal structure

Magnetic axes The orientation of the magnetic axes was determined by using the ddip(obs) via Eqns (4) and (5) for Pe metHbCN The anisotropies at 30
C, which have been shown to be highly conserved in a wide variety of cyanomet globins [8,26,28–30], are Dvax 8m3ms1 and Dvrh¼

8m3ms1, as reported for sperm whale met-MbCN The coordinates that determine R, h¢ and W¢ in Eqn (3) were taken alternatively from the Pe rWT

Fig 4 Portions of the 600 MHz1H NOESY spectra (mixing time 80 ms) of Pe metHbCN in 1

H 2 O 100 m M in phosphate, pH 7.0 at 35
C Dipolar contacts are illustrated, involving key distal residues Phe46(CD1) and helical ring cross peaks for Tyr32(B10) (F), Phe36(B14) (G), Phe46(CD1) (E), Tyr66(E7) (G) and Leu70(E11) (E), and to NHs of heme-residue contacts Tyr66(E7) and Phe46(CD1) to have propionates (A, B, C) and interresidue con-tacts from Tyr66(E7) to Phe46(CD1) and NHs

of Arg48(CD3) Leu49(CD4) (G), and Tyr32(B10) (D) contacts to Ala67(E8) (F) The cross peak between Tyr32(B10) C e Hs and Tyr66 (E7) C d Hs is observed only at a lower contour level.

metHbH2O [16] (case I) or the WT metHb [17] (case II)

crystal structures In order to utilize the information in ddip

for distinguishing between the two crystal structures, the

experimental shifts and crystal coordinates initially used to

determine the magnetic axes were only for those protons

where the residue exhibited the same position in the

alternate crystal structures The results lead to equally

well-determined orientations of a¼ 203 ± 10, b ¼ 9
± 1,

j¼ 50 ± 10 and residual F/n ¼ 0.14 p.p.m.2 for case I,

and a¼ 206 ± 10, b ¼ 10 ± 1, j ¼ 40 ± 10
and

resid-ual F/n¼ 0.20 p.p.m.2for case II The plot of ddip(obs) vs

the ddip(calc) (,j) for each set of magnetic axes are given

in Fig 5A (case I) and 5B (case II), and each represents a

good fit The differences in b do not reflect differences in tilt

of the axis so much as a small difference in the reference coordinate system x¢, y¢ and z¢ in the two structures (due to different nonplanarity of the heme) The ddip(obs) and

ddip(calc) for those protons whose coordinates differed significantly in the two structures are shown as s and h For the most part, in particular Thr99(FG1) and Val103(FG5), residues with different geometries exhibit reasonable fits for both cases in Fig 5, in part because ddipis small, but also because their position is not very sensitive to dipolar shift However, it is noted that the Tyr32(B10) CeHs exhibit a very reasonable fit for case I (Fig 5A), but an unacceptable fit for case II (Fig 5B) Hence the magnetic axes completely concur with the results of both NOESY intensity analysis and paramagnetic relaxation effects in finding the Pe metHbCN active site solution structure to coincide with the crystal structure of rWT metHbH2O [16] but not WT metHb [17]

Redetermination of the magnetic axes orientation (a, b, c), from a large variety of available input data using only the pertinent Pe rWT metH2O crystal structure [16] led to

a¼ 202 ± 10
, b ¼ 9 ± 1
and j ¼ 52 ± 10
for a three-parameter search using the sperm whale metMbCN anisotropies [29], and yielded minimally changed orienta-tion, a¼ 202 ± 10, b ¼ 9 ± 1 and j ¼ 51 ± 10 for the five-parameter search that yielded Dvax¼ 2:36 

m3mol1 which are within the uncertainties of the respective determinations (not shown) The tilt of the major magnetic axes is correlated with Fe-CN tilt [8,28,30] (with the negative z axis), and indicates that the cyanide is tilted
10
in the direction of the 5-H position The rhombic axes are defined by j
50
in Fig 1 The difference in the overall shift dispersion pattern of Pe metHbCN relative to, for example, any of the mammalian metMbCN where both the FG corner and PheCD1 residues exhibit large upfield and downfield shifts, respect-ively, is due to the smaller tilt, b

Fig 5 Plot of the d dip (obs) vs d dip (calc) for the

magnetic axes of Pe WT metHbCN as based on

the crystal coordinates of rWT metHbH 2 O;

and WT metHb (A) rWT metHbH 2 O; and (B)

WT metHb, using as input only the d dip (obs)

for protons whose positions are the same

in the two crystal structures [16,17], with

Dv ax ¼ 2.48 · 10)8m 3 Æmol)1and

Dv rh ¼ )0.58 · 10 )8 m3Æ mol)1as reported

for sperm whale metMbCN [29] The solid

markers represent the input data for the

structurally conserved protons, while open

markers are for those protons whose positions

differ significantly in the two crystal structures.

Table 3 Comparison of predicted and observed NOESY cross peak

intensity for the two crystal structures of Pe metHb Inter-proton

separation r ij (A˚) Pe rWT metMbH 2 O crystal structure [16], Pe WT

metHb crystal structure [17] Observed NOESY cross peak intensities,

s (strong, r ij < 2.5 A˚), m (moderate, 2.5 < r ij < 4.0 A˚), weak (weak,

4.0 < r ij < 5.0 A˚) Distances in bold are in agreement with the NMR

observations.

r ij (A˚) rWT metHbH 2 O WT metHb NOE F-helix/FG-corner

NH(FG1)-C a H(F8) 3.49 2.45 m

C a H(FG1)-C a H(H23) 2.55 5.92 mÆs)1

C a H(FG1)-C c H(H23) 3.47 6.22 m

C b H(FG1)-C a H(F6) 3.80 7.02 m

B-helix

C a H 2 (B10)-C a H(E8) 2.22 3.11 s

C e H2(B10)-C b H 3 (E8) 4.33/4.86 5.33 m

C b H 2 (B10)-C b H 1 (E7) 2.91 2.52 m

C e H 2 (B10)-C b H 2 (E11) 3.14 2.38 m

Structural simulation of Tyr32(B10)

The good correlation between ddip(obs) and ddip(calc) for

the Tyr32(B10) ring in the magnetic axes based on the

rWT metHbH2O crystal structure [16] indicates that the

ring (and hence B-helix) occupies the same position as in

the crystal This conclusion is supported by the relaxation

properties of the CeHs signal (see above) The hydroxyl

proton position is not directly determined in the crystal

structure, but can be inferred by the position of other

H-bond acceptor/donors in the immediate vicinity The

proposal that the Tyr32(B10) hydroxyl proton acted as a

donor to the carbonyl of Tyr66(E7) in the rWT

metHbH2O crystal structure [16] places it
5.9 A˚ from

the iron with an angle of
17
with the Tyr32(B10)

Ce-Cf-O-H plane; we define this angle w¼ 0 In the

rWT metHbH2O crystal structure [16], the heme ligand

(water molecule) is an H-bond donor, while in

metH-bCN, it (cyanide) is an H-bond acceptor, so that a

significantly different OH orientation can be expected A

plot of the effect of the angle, w, between the Tyr32(B10)

Cf-OH and ring planes, on the three distinctive variables

that depend critically on the orientation of the OH group

is illustrated in Fig 6 The shaded areas correspond to

the observed values of ddip(calc) (Fig 6A), distance to the

iron, RFe¼ 4.0–4.5 A˚ (Fig 6B), as indicated by

T1
10 ms, and Tyr32(B10) OH to Tyr66(E7) CdH

distance, rij of Tyr66(E7), Fig 6C, as indicated by a

weak-to-moderate NOESY cross peak intensity

Inspec-tion of Fig 6 reveals a single orientation,

w¼)140 ± 20
, that essentially quantitatively and

sim-ultaneously accounts for the three observations The

effect on the Tyr32(B10) O-H.N(cyanide) angle on Y is

illustrated in Fig 6D The position of the Tyr32(B10) Og

relative to the cyanide ligand tilted by
10
in the

direction of the 10-H position is illustrated in Fig 7 and

reveals a van der Waals contact between Tyr32(B10) and

the cyanide

Discussion

Active site structure

The combination of NOESY cross peak intensities,

paramagnetic relaxation and the magnetic axes data

provide compelling evidence that WT Pe metHbCN

much more closely resembles the structure found in the

crystal structure of rWT Pe metHbH2O [16] than of WT

PemetHb [17] Because the WT Pe metHbCN active site

structure is essentially the same as rWT metHbH2O, and

different from WT metHb, the present data support the

interpretation that the structural differences between rWT

and WT Pe metHb in crystals result from the extensive

interaction between the two molecules in the unit cell for

WT metHb, rather than from significant structural

differences between isolated WT and rWT Pe Hb

molecules [17] The distal Tyr66(E7) ring was found

oriented out of the heme pocket in both rWT metHbH2O

and WT metHb [16,17] Our NMR data on Pe metHbCN

confirm that Tyr66(E7) is similarly oriented away from

the heme iron in a position essentially the same as in the

crystal structure with its O H much too far removed

(>6 A˚) from the cyanide to provide a H-bond The failure to resolve the OgH signal for Tyr66(E7) can be attributed to its expected rapid exchange with solvent While cyanide is a H-bond acceptor and a weak mimic of

O2, it does not induce a rearrangement of the Tyr66(E7) ring into the heme pocket relative to the high-spin, metHb complexes

1H NMR data on Pe WT HbO2had shown that there are two interacting labile protons from two Tyr in the distal pocket capable of interacting with the bound O2 [23] Moreover, NOESY cross peaks between the Tyr66(E7) ring and the terminus of Leu70(E11) indicated that the Tyr66(E7) ring is oriented into the heme pocket There appears to be no obvious rationalization for these contra-dictory results It has been demonstrated that Tyr66(E7) can

be substituted without significantly affecting the extreme O2 ligation dynamics/thermodynamics [16] However, these results do not alone determine that Tyr66(E7) is not oriented into the heme pocket, they only demonstrate that any interaction of the Tyr66(E7) ring with O2 does not incrementally increase the H-bond stabilization of bound

O2 A crystal structure or solution NMR structure of Pe HbO is clearly important

Fig 6 Plots of d dip (calc) derived from optimized magnetic axes, the distance to iron via paramagnetic relaxation R Fe , distance to Tyr66(E7)

C d H(via NOESY cross peak intensity) and /O-H-N angle as a function

of the C f O-H to aromatic plane dihedral angle, w, for the Tyr32(B10) hydroxyl group, with the ring position as defined in the rWT metHbH 2 O crystal structure [16]and confirmed for the WT metMbCN solution structure described here The shaded portions represent the observed values (and their uncertainties) of the three variables Note that a simultaneous fit for all three variables occur only for w
)140 ± 20
.