Tải bản đầy đủ - 0trang
216



Lim et al.



1.931 š 0.02 across these features, demonstrating that the transition

moments for the two A states are oriented at a similar angle. From the

measured polarization ratio, the equilibrium angle Âeq for the transition

moment was estimated to be Ä7° (51). The upper limit of 7° obtains in

the limit where the heme is a perfectly flat circular absorber, and this

value is in excellent agreement with static measurements of the polarization

anisotropy in ambient temperature MbCO crystals (52–54). The two spec˚ crystal structure of

troscopic results are significantly different from a 1.5 A

P21 MbCO, where the highly conserved distal histidine purportedly caused

CO to bind in a nonoptimal bent geometry of 39° and thereby inhibited the

˚ crystal structure of P6

binding of CO (55). In contrast to that result, a 2 A

MbCO reported an angle of 19° (56). Spiro and Kozlowski (18) suggested

that the discordant results between IR spectroscopy and x-ray crystallography might be rationalized by density functional theory, which was used to

explore the relationship between the direction of the CO transition moment

and the C–O bond axis when bound to Mb. They found that a 7° angle

between the heme plane normal and the transition moment of bound CO

could correspond to a C–O angle as large as 15° , but certainly not 39° .

The discrepancy between IR spectroscopy and x-ray crystallography

˚

became negligible when Bartunik and coworkers reported a 1.15 A

resolution structure of MbCO at ambient temperature and found the C–O

angle to be approximately 12° with respect to the heme plane normal (57).

The near-quantitative agreement obtained between IR spectroscopy and

atomic resolution x-ray crystallography leaves little doubt that the primary

source of ligand discrimination between CO and O2 is something other than

steric hindrance. Rather than suppression of the binding affinity of CO, it

has been suggested that ligand discrimination arises from enhancement of

the binding affinity of O2 by formation of a hydrogen bond with the distal

histidine (58).

The B -state spectra reveal two features, denoted B1 and B2 after

Ormos et al. (17), with B1 blue-shifted relative to B2 . According to Fig. 7,

the polarized absorbance ratio A? /Ajj for “docked” CO is much closer

to 0.75 than it is to 2, demonstrating that CO rotates substantially upon

dissociation from the heme iron. According to Fig. 7, the two B states

reveal a similar ratio, A? /Ajj D 0.856 š 0.03, which was found to be

consistent with Âeq ¾ 90° (11,51). X-ray structures of MbŁ CO at cryogenic temperatures reveal electron density assigned to unbound CO that

˚ from the binding site (59,60). The CO orientation from

is displaced Ä2 A

those structures is not inconsistent with the polarized IR results.



Copyright © 2001 by Taylor & Francis Group, LLC



Infrared Studies in Heme Proteins



217



Because the electrostatic field in the vicinity of the docking site is

anisotropic, the vibrational frequency of “docked” CO should be Stark

shifted and the direction of that shift should depend on the CO orientation.

Consequently, the two B-state features are interpreted as Stark-shifted

spectra with B1 and B2 corresponding to CO pointing in opposite directions.

Unlike the polarized absorbance ratio measured for the A states, the ratio

for the B states is vibration-frequency dependent with the polarization ratio

a maximum at the midpoint between the two features. Because the center

frequency of CO depends on its orientation, its vibrational frequency should

shift smoothly from one B state frequency to the other as it undergoes endto-end rotation. Consequently, one might expect the vibrational frequency

of CO near the transition state for end-to-end rotation to be centered

between the peaks of the two B states. At the midpoint, the ratio A? /Ajj

exhibits a maximum of ¾1, demonstrating that the trajectory for end-to-end

rotation passes through a transition state that has a component out of the

plane of the heme. If this trajectory were to lie in a plane, the transition

state would be oriented at least 55° from the heme plane normal (51). Such

a trajectory would maintain a ligand orientation far from that for bound

CO, even at the transition state for end-to-end rotation, thereby inhibiting

CO binding while permitting end-to-end rotation.

D. Ligand Translocation Trajectories



To experimentally probe the CO trajectory after dissociation, ultrafast

time-resolved polarized mid-IR spectra of photolyzed h-MbCO in G/W

were recorded (34), the results of which are plotted in Fig. 8A. This

study was performed in G/W primarily because the flatness of the solvent

absorbance spectrum near 2100 cm 1 minimizes temporal distortion of the

transmitted femtosecond IR probe pulse, thereby maximizing the effective

time resolution of the measurement. Two features are already apparent at

0.2 ps, the earliest time shown, and these features rapidly develop into

the “docked” states denoted B1 and B2 . The development of the “docked”

CO spectrum is further quantified by the time dependence of the polarization anisotropy, as defined in Equation (2). The B1 and B2 polarization

anisotropies, plotted in Fig. 8B, evolve exponentially with time constants

of 0.20 š 0.05 ps and 0.52 š 0.10 ps, respectively, and converge to the

same anisotropy of approximately 0.2. According to Fig. 8C, ligand translocation is accompanied by a 1.6 š 0.3 ps growth of the integrated isotropic B-state absorbance.



Copyright © 2001 by Taylor & Francis Group, LLC



218



Lim et al.



Figure 8 (A) Femtosecond time-resolved IR absorbance spectra of CO measured

after photodissociation from the heme of h-Mb in G/W. Spectra were recorded

with the photolysis and probe pulses polarized parallel Ajj and perpendicular

A? to one another. The polarized absorbance spectra reveal the time dependence

of the ligand orientation as well as the protein surroundings (see text). The

features labeled B1 and B2 (bottom) correspond to the least-squares fit of the

Ajj spectrum measured at 10 ps. For clarity, the background and hot band

contributions to the time-resolved spectra have been removed and the spectra

have been offset from one another. (B) Time dependence of the polarization

jj

?

B?

anisotropy, r t D [Bjji t

i t ]/[Bi t C 2Bi t ], after photodissociation from

the heme of Mb. Bi t represents the integrated absorbance under state i at time t

with the polarization denoted by a superscript. The polarization anisotropies of

B1 and B2 appear to evolve exponentially (solid lines) with time constants of

0.2 ps and 0.52 ps, respectively. (C) Time dependence of the isotropic B -state

absorbance (filled squares). The relative contribution of B1 to the total absorbance

is time independent out to 10 ps (open squares) and averages 56% (dashed

line). To generate these data, the isotropic absorbance was synthesized from

the polarized absorbance spectra according to the “magic” angle prescription:

AMA t D [Ajj t C 2A? t ]/3. (Adapted from Ref. 34.)



Copyright © 2001 by Taylor & Francis Group, LLC



Infrared Studies in Heme Proteins



219



How do we rationalize these observations? Photoexcitation of MbCO

renders the Fe–CO coordinate repulsive, causing CO to acquire translational

kinetic energy as it moves up and away from the heme iron. Before the

CO rotates, its polarization anisotropy should be approximately 0.19, and

its mid-IR absorbance should be confined to a single, broad feature. Upon

colliding with the surrounding protein, the ligand rebounds back toward the

heme and proceeds along one of two different trajectories. Because ligand

translation and rotation are not slow compared to 0.2 ps, the polarization

anisotropy measured at 0.2 ps is not the theoretical minimum of 0.19, nor

is the spectrum manifested as a single broad feature. As the CO translates

and rotates, the two trajectories become spectroscopically distinguishable,

owing to the vibrational Stark shift that arises from the electrostatic field

surrounding the ligand. The fact that B1 and B2 evolve at different rates

demonstrates that the trajectories leading to B1 and B2 are distinguishable

kinetically as well as spectroscopically. The polarization anisotropy for both

B1 and B2 converges within a few ps to ¾0.2. Had the polarization anisotropy

been measured using a wavelength where the heme is a perfectly flat circular

absorber, the polarization anisotropy would rise to only about 0.05. By

photolyzing at a wavelength where the heme is an elliptical absorber, the

range of the polarization anisotropy was enlarged, thereby improving the

signal-to-noise ratio of the measurement. Moreover, the fact that the polarization anisotropy exceeds 0.05 suggests that the major axis of the elliptical

heme absorber must be at least partially aligned with the C–O axis. If the

orientation of the major axis of the heme transition moment were known

as a function of wavelength, polarization anisotropy measurements at more

than one wavelength would permit a determination of both the azimuthal

CO orientation and its angle with respect to the heme plane normal.

The prompt appearance and independent development of the two B

states suggest that the two trajectories are deterministic in nature, with

the outcome (B1 or B2 ) established promptly after photodetachment. What

do the two limiting states correspond to structurally? From a kinematic

argument, it was rationalized that the faster B1 trajectory has CO sliding

into the docking site with the O end of C–O pointing toward the heme

iron (34). This structural assignment is supported by geminate rebinding

studies of photolyzed MbCO at 20 K, where B1 predominates: the geminate

rebinding of 13 C16 O was found to be slower than 12 C18 O, in spite of the

latter being heavier (61). Because geminate rebinding at 20 K is dominated

by tunneling (61) and the tunneling rate depends on distance as well as

mass, this surprising isotope effect can be rationalized by orienting C–O

such that the O end is pointing toward the heme iron. The orientation of the



Copyright © 2001 by Taylor & Francis Group, LLC



220



Lim et al.



CO in the docking site is too fine a detail to be extracted from the electron

density maps of photolyzed MbCO (59,60).

Interestingly, the proportion of the integrated absorbance ascribed

to B1 (plotted in Fig. 8B) remains largely unchanged during the protein

conformational reorganization. Assuming the integrated area under each

B state is proportional to its population, an assumption for which there

is experimental support (33), the partitioning between B1 and B2 is 1.3:1.

This ratio is not far from the statistical (1:1) distribution expected with

two possible CO orientations. The time independence of this ratio (out to

10 ps) provides added support for the suggestion that the ligand dissociation trajectories are deterministic. The ratio does not remain constant out to

longer times, however, but increases to 1.7:1 by 100 ps (51), demonstrating

that B1 is lower in energy than B2 and that end-to-end rotation between 10

and 100 ps renders the distribution thermodynamic rather than statistical.

If the integrated areas under B1 and B2 are indeed proportional to population, then the ratio of 1.7:1 corresponds to a free energy difference of

about 1.2 kJ/mol with B1 lower in free energy. Evidently, there is a modest

preference for the trajectory leading to B1 , which also turns out to be the

more stable state thermodynamically.

Because the interior of Mb is densely packed, ligand translocation

from the active binding site to the docking site requires some degree of

protein rearrangement, a process that should affect the vibrational spectrum

of “docked” CO. Moreover, the conformational response of the protein

should be more sluggish than the motion of the ligand. Might the 1.6 ps

growth of the integrated isotropic B -state absorbance be assigned to protein

rearrangement, or might it arise from other causes? One often equates

changes in integrated absorbance with changes in population, however, that

is not the case here: all CO produced photolytically is generated in less than

0.2 ps, the time resolution of the measurement. Might the growth be due

to thermal cooling of the CO and its environment? Because “docked” CO

is in contact with the heme, its kinetic temperature would be expected to

cool at a rate similar to the heme, which was found to thermally relax

with a time constant of 6.2 š 0.5 ps (22). Because the 1.6 ps growth of

the integrated B-state absorbance is longer than the 0.2 and 0.5 ps rotation times and shorter than the 6.2 ps cooling rate, it cannot be ascribed

to population or cooling dynamics. Rather, it most likely arises from reorganization of the neighboring protein residues about the nascent “docked”

CO. Recall that the integrated absorbance of CO is partitioned between

the narrow B states and a broad unresolved pedestal, the partitioning of

which is determined by the orientational constraints imposed on CO by the



Copyright © 2001 by Taylor & Francis Group, LLC



Infrared Studies in Heme Proteins



221



surrounding protein. The more constrained the CO, the smaller the amplitude of its librational motion, and the greater the integrated absorbance

measured under the narrow B-state features. A 1.6 ps time constant for

protein rearrangement that serves to constrain “docked” CO appears to be

quite reasonable. Evidently, this process helps to establish a steric barrier to

the reverse rebinding process. That this steric barrier arises before the electronic state of the heme relaxes to its ligand-receptive ground state (3.4 ps)

may explain the absence of ultrafast geminate ligand rebinding.

E. Origin of the Barrier to CO Rebinding



Photolysis of MbNO and MbO2 is followed by substantial geminate recombination (28,62–64) with NO rebinding on the sub-ns time scale. On the

other hand, geminate recombination of CO in Mb is minimal and occurs on

the few hundred ns time scale (65). The lack of significant CO rebinding is

˚ away from the

remarkable considering the ligand remains “docked” Ä2 A

binding site for several hundred ns. It had been suggested that the kinetic

differences among these ligands arise from differences in the electronic

barrier to binding, with CO having the highest electronic barrier and NO

having the lowest (66). The discovery of a ligand docking site that can

constrain the orientation of “docked” ligands as small as CO (11) lead us

to consider another possibility. The docking site might slow the rebinding

rate of CO by strongly hindering access to the transition state for CO

rebinding. Because O2 and NO both bind in a bent configuration, access to

their transition state for rebinding is far less hindered.

To explore the possibility that slow CO rebinding is a consequence

of a steric, not an electronic barrier, the geminate rebinding dynamics of

CO to Mb and microperoxidase were compared (67). Microperoxidase is

an enzymatically digested cytochrome c oxidase that consists of a heme

with a “proximal” histidine that is part of an 11-peptide fragment. This

peptide renders the heme soluble under the neutral conditions used in the

Mb studies. When reduced to Fe(II), 13 CO binds to microperoxidase and

a vibrational stretch near 1908 cm 1 appears. This frequency is virtually

identical to that found in Hb13 CO and is similar to the 1900 cm 1 transition

found in Mb13 CO. Consequently, extracting the heme out of the protein

appears to have only a minor effect on the heme-CO interaction and, one

might assume, the electronic barrier to ligand binding. Because the peptide

is not long enough to wrap around and fashion a docking site on the distal

side of the heme, photodissociated CO will be surrounded by disordered

solvent, not a highly organized docking site. Any differences in the rates

of geminate rebinding to Mb and microperoxidase might, therefore, be



Copyright © 2001 by Taylor & Francis Group, LLC



222



Lim et al.



ascribed primarily to the steric constraints imposed by the docking site

in Mb.

The geminate-rebinding dynamics measured after photolysis of

MbCO and microperoxidase-CO are shown in Fig. 9. The survival fraction

denotes the fraction of photolyzed hemes that remain in the deoxy form

after CO dissociation. The population was determined by measuring the

time dependence of the vibrational absorbance of bound CO. According to

Fig. 9, CO rebinds to microperoxidase much more rapidly than to Mb.



Figure 9 Geminate recombination after photolysis of MbCO (ž) and microperoxidase-CO ( ). The survival fraction refers to the population that remains unbound

after photolysis. The population was determined by measuring the IR absorbance

at frequencies corresponding to the peak of the bound CO stretch. (Adapted from

Ref. 67.)



Copyright © 2001 by Taylor & Francis Group, LLC