Subscriber access provided by NEW YORK UNIV

Article

Conformational Landscape and the Selectivity of Cytochrome P450cam Edward John Basom, James W Spearman, and Megan C Thielges J. Phys. Chem. B, Just Accepted Manuscript • DOI: 10.1021/acs.jpcb.5b03896 • Publication Date (Web): 08 May 2015 Downloaded from http://pubs.acs.org on May 12, 2015

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

The Journal of Physical Chemistry B is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 23

The Journal of Physical Chemistry

1 3

2

Conformational Landscape and the Selectivity of Cytochrome P450cam 4 6

5

Edward J. Basom, James W. Spearman, and Megan C. Thielges* 7 8 9

Department of Chemistry, Indiana University, Bloomington, Indiana 47405, United States 10 1 12 13 14 15 16 17 18 19 20 21 2 23 24 25 26 27 28 29 30 31 32 3 34 35 36 37 38 39 40 41 42 43 4 45 46 47 48 49 50 51 52 53 54 5 56 57 58 59 60 ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 3

2

ABSTRACT: Conformational heterogeneity and dynamics likely contribute to the remarkable 5

4

activity of enzymes, but are challenging to characterize experimentally. These features are of 6 8

7

particular interest within the cytochrome P450 class of monooxygenases, which are of great 10

9

academic, medicinal, and biotechnological interest as they recognize a broad range of substrates, 1 12

such as various lipids, steroid precursors, and xenobiotics, including therapeutics. Here, we use 13 15

14

linear and 2D IR spectroscopy to characterize the prototypical P450, cytochrome P450cam, 17

16

bound to three different substrates, camphor, norcamphor, or thiocamphor, which are 18 19

hydroxylated with high, low, and intermediate regioselectivity, respectively. The data suggest 20 2

21

that specific interactions with the substrate drive the population of two different conformations, 24

23

one that is associated with high regioselectivity, and another associated with lower 25 27

26

regioselectivity. Although Y96 mediates a hydrogen bond thought necessary to orient the 29

28

substrate for high regioselectivity, the population and dynamics of the conformational states are 31

30

largely unaltered by the Y96F mutation. 32

This study suggests that knowledge of the

34

3

conformational landscape is central to understanding P450 activity, which has important 36

35

practical ramifications for the design of therapeutics with optimized pharmacokinetics, and the 37 38

manipulation of P450s, and possibly other enzymes, for biotechnological applications. 39 40 41 42 43 4 45 46 47 48 49 50 51 52 53 54

KEYWORDS: 2D IR, nonlinear spectroscopy, energy landscape, enzyme specificity, 5 57

56

cytochrome P450s 58 59 60 ACS Paragon Plus Environment

Page 2 of 23

Page 3 of 23

The Journal of Physical Chemistry

1 3

2

INTRODUCTION 5

4

Cytochrome P450s are heme-dependent monooxygenases that catalyze the hydroxylation of 6 7

hydrocarbons in the synthesis of various steroids and lipids.1-3 In addition to their central role in 10

9

8

the metabolism of many xenobiotics, making their activity a central component in the 1 12

pharmokinetics of many therapeutics, and their potential use in a variety of biotechnological 13 15

14

applications, such as bioremediation, the P450s have received a great deal of attention due to 17

16

their ability to hydroxylate unactivated carbon centers with high regio- and enantioselectivty.1-3 18 19

However, the means by which this activity is manifested has remained unclear. 20 2

21

Cytochrome P450cam (P450cam) catalyzes the regioselective hydroxylation of d24

23

camphor to 5-exo-hydroxycamphor, and has emerged as a model system for the study of the 26

25

cytochrome P450 family of enzymes.4,5 27

Previous structural,6-9 spectroscopic,10-12 and

29

28

computational studies13,14 of P450cam-substrate complexes have suggested that the high 31

30

specificity of camphor hydroxylation results, at least in part, from the formation of a tight 32 34

3

enzyme-substrate complex mediated by packing interactions that restrict the motion of the 36

35

substrate relative to the reactive oxygen center. Consistent with this conclusion, previously 37 38

reported studies of CO-bound P450cam (P450camCO)-substrate complexes with linear and 2D 39 41

40

IR spectroscopy revealed that the CO experiences different environments when bound to 43

42

camphor or its analogs camphane or norcamphor,11,12,15 which are hydroxylated with similar and 45

4

lower regioselectivity, respectively.16,17 46

Interestingly, the norcamphor-induced environment

48

47

showed faster dynamics, which were suggested to reflect less substrate restriction and to lead to 50

49

the lower regioselectivity.15 However, while the data clearly indicate that the CO experiences 51 53

52

different environments when P450camCO is bound to the different substrates, it was not possible 5

54

to determine whether the differences result from the population of different conformations or 56 57 58 59 60 ACS Paragon Plus Environment

The Journal of Physical Chemistry

Page 4 of 23

1 3

2

from altered electrostatic and packing forces associated with the different substrates within a 5

4

single conformation. 6 7 8

A second model explaining the high regioselectivity of camphor hydroxylation is based 9 1

10

on structural studies that reveal the presence of a hydrogen bond (H-bond) between the camphor 13

12

ketone and the side chain of active site residue Y96, which might also contribute to fixing the 14 15

orientation of the substrate within the active site (Figure 1A).5,6 Consistent with this suggestion, 16 18

17

thiocamphor (Figure 1B), which is expected to form a weaker H-bond, is hydroxylated with a 20

19

lower regioselectivity.16 Nonetheless, sulfur is also significantly larger and more polarizable 21 23

2

than oxygen, suggesting that packing interactions could also differentiate the substrates. 25

24

Moreover, the fact that P450cam hydroxylates thiocamphor with a regioselectivity lower than 27

26

camphor, but higher than norcamphor17 (which retains the putatively critical H-bond acceptor, 28 30

29

but within a less encumbered scaffold) makes the relative contributions of packing, dynamics, 32

31

and H-bonding unclear. 3 34

To 35

investigate

how

conformational

36 37

heterogeneity and dynamics contribute to P450cam 38 39

activity, 40

we

characterized

the

complex

of

41 42

P450camCO 43 4

with

camphor,

norcamphor,

and

thiocamphor with linear and 2D IR spectroscopy. 45 46

We also prepared and characterized the substrate 47 49

48 Figure 1. Structures of P450cam and substrates. (A) Structure of the P450camCO complex with camphor in the immediate vicinity of the CO ligand (PDB entry 1T87). Shown are the heme, the CO distal ligand, the C357 proximal ligand, the bound camphor substrate, and the Y96 residue. (B) Structures of substrates camphor (top), norcamphor (middle), and thiocamphor (bottom).

56

5

54

53

52

51

50

complexes of Y96F P450camCO to assess the contribution of the intermolecular H-bond. We find that the binding of a substrate to P450cam selectively stabilizes one of two distinct conformations, one that

57 58 59 60

ACS Paragon Plus Environment

Page 5 of 23

The Journal of Physical Chemistry

1 3

2

has a local energy landscape associated with high regioselectivity of hydroxylation, the other 5

4

with low regioselectivity. The intermediate regioselectivity of thiocamphor hydroxylation arises 6 8

7

from simultaneous population of the two conformations. 10

9

Finally, the crystallographically

observed intermolecular H-bond involving Y96 influences the local electrostatics of the active 1 12

site, but has little effect on either the relative populations of conformational states or the nature 13 15

14

of the energy landscapes within the conformations. 16 17 18 19

MATERIALS AND METHODS 20 2

21

P450cam Sample Preparation. 24

23

P450cam was produced by recombinant expression with

plasmid pDNC334A (kindly provided by Thomas Pochapsky, Brandeis University).18 25 27

26

Experiments were performed with P450cam C334A, which has been shown to display activity 29

28

identical to wild-type but decreased propensity for aggregation.19 The mutation Y96F was 31

30

introduced into the P450cam gene sequence using site-directed mutagenesis (Stratagene) and 32 34

3

confirmed by sequencing. Both proteins were expressed and purified as previously described 36

35

(see also SI).20,21 37 38

Camphor-bound samples were equilibrated in 50 mM potassium phosphate, pH 7 with 39 41

40

100 mM KCl, 20% glycerol, and 5 mM d-camphor, followed by spin-concentration to 1-2 mM. 43

42

All other samples were first passed over a 10 cm Sephadex G25 column (GE Life Sciences) in 4 45

50 mM Tris-Cl, pH 7.4 to remove camphor from the storage buffer. Norcamphor-bound samples 46 48

47

were subsequently equilibrated into 50 mM potassium phosphate, pH 7 with 100 mM KCl, 20% 50

49

glycerol, and 85 mM norcamphor (Sigma Aldrich) before spin-concentration to 1-2 mM.10 51 53

52

Thiocamphor (Apollo Scientific) was first dissolved in ethanol and then added to a concentrated 5

54

substrate-free protein sample such that the final solution contained 8 mM thiocamphor and less 56 57 58 59 60 ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 3

2

than 5% ethanol. The dissociation constant for the Y96F P450cam complex with thiocamphor 5

4

was determined by a spectroscopic titration (SI). All samples with thiocamphor were gently 6 8

7

rocked overnight at 4°C. The concentrated protein samples were gently purged under Ar(g) and 10

9

CO(g) for several minutes, reduced with 15 equivalents of sodium dithionite, and again purged 1 12

with CO(g) for several minutes in order form the P450camCO complexes. The CO-bound 13 15

14

samples were loaded between CaF2 windows with a 38.1 µm Teflon spacer. Visible spectra of 17

16

all CO-bound samples contained the characteristic band at 446 nm and no evidence for a band at 18 19

420 nm. 20 21 2 24

23

FT IR and 2D IR Spectroscopy. All FT IR spectra were recorded at 2 cm-1 resolution with an 25 27

26

Agilent Cary 670 FT IR spectrometer and N2(l)-cooled MCT detector. The instrument was 29

28

purged with dry N2(g) prior to data collection. Additional details are available in SI. 31

30

2D IR experiments were performed as previously described in the literature,22 and are 32 3

described in greater detail in SI. Briefly, three ~120 fs pulses centered at ~1950 cm-1 generated 36

35

34

with a mid-IR laser system consisting of a Ti:Sapphire oscillator/regenerative amplifier-pumped 37 38

optical parametric amplifier were applied to the sample in a boxcar geometry. To generate a 39 41

40

single 2D IR spectrum, the time between the first two pulses (τ) was scanned, while time 43

42

between the second and third pulses, the waiting time (Tw), was held constant. At a time ≤τ after 4 45

the application of the third pulse, the third-order polarization generated in the sample leads to 46 48

47

emission of a vibrational echo signal, which was combined with a fourth pulse, the local 50

49

oscillator (LO) to amplify the signal and obtain phase information. The heterodyned signal was 51 52

dispersed by a spectrograph and detected on an MCT array to generate the vertical, ωm axis of 5

54

53

the 2D spectrum. At each pixel of the array, interferograms acquired along τ were Fourier 56 57 58 59 60 ACS Paragon Plus Environment

Page 6 of 23

Page 7 of 23

The Journal of Physical Chemistry

1 3

2

transformed to produce the horizontal, ωτ axis of the 2D spectrum. 2D spectra are acquired at a 5

4

variety of Tw times. The time-dependent changes in the 2D spectra provide the dynamical 6 8

7

information about the system. 9 10 1 12

Data Analysis. The center-line slope (CLS) method applied to the Tw-dependent 2D spectra, 13 14

which has been described in detail elsewhere (see also SI),23 in combination with fitting to the 17

16

15

linear spectra, was used to obtain the frequency-frequency correlation function (FFCF), a 19

18

quantitative description of the spectral dynamics.22 The FFCF was described by a sum of fast 20 2

21

(homogeneous) and slower (inhomogeneous) dynamics: 23 25

24 26

𝐹𝐹𝐶𝐹 = 28

27

𝑡 𝛿 − + ∑ ∆2𝑖 𝑒 𝜏𝑖 𝑇2

The inhomogeneous dynamics are described by the sum of exponential decay terms, 30

29

where ∆𝑖 describes the frequency fluctuation amplitude associated with the dynamics, and 𝜏𝑖 31 3

32

describes the timescale. The Akaike information criterion was used to determine the number of 34 35

parameters included in the fits (SI). 36

The

37 39

38 40

contribution to the dynamics, where

1 𝑇2

𝛿 𝑇2 1

term describes the homogeneous dephasing 1

1

= 𝑇 ∗ + 2𝑇 + 3𝑇 . 2

1

𝑜𝑟

The homogenous dephasing

41

contribution accounts for very fast dynamics in which the frequency fluctuation amplitude and 42 43

timescale cannot be separated, where ∆2𝑖 𝜏𝑖 < 1. The orientational lifetime 𝑇𝑜𝑟 is assumed to be 46

45

4

on the order of several ns for proteins, and so the term is neglected. 𝑇1 is the vibrational lifetime, 47 48

set to 19 and 27 ps for the camphor and norcamphor complexes of P450camCO, respectively.15 51

50

49

1

𝑇2∗ is the pure dephasing time and is related to the homogeneous line width by Γ = 𝜋𝑇 ∗. 52

2

54

53

The 2D spectra of the thiocamphor complexes, which exhibited multiple component 56

5

bands, were also analyzed via a procedure described by Fenn and Fayer24 to extract the FFCF 57 58 59 60 ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 3

2

from one component when the FFCF of the second component and the relative populations of the 5

4

associated states are known. The center frequencies along the ωτ axis from 1D slices of the 2D 6 7

IR spectra at each ωm (the center line data) were extracted from the spectra of the norcamphor 10

9

8

complexes. Time-dependent fractional contributions of the two components were determined 1 12

from the relative areas of the associated bands in the linear spectra and the vibrational lifetimes 13 14

of CO in the complexes with camphor and norcamphor.15 The center line data obtained from the 17

16

15

2D IR spectra of the thiocamphor complex were then fit to two components, fixing the relative 18 19

contributions of the components and setting the high frequency component to the center line data 20 2

21

extracted from norcamphor complex, to extract the center line data for the second component. 24

23

The FFCF of the second component was then determined from Tw-dependent slopes of the center 25 27

26

line data. See SI for additional details. 28 29 31

30

RESULTS 32 34

3

The CO absorption of wild-type and Y96F P450camCO were first characterized when bound to 36

35

camphor, norcamphor, or thiocamphor by FT IR spectroscopy (Figure 2). As expected, the 38

37

spectra all show one or more bands around 1950 cm-1 due to the stretching vibration of the heme39 41

40

bound CO. The spectra were fit to a sum of Gaussian functions to determine the frequencies, 43

42

line widths, and relative amplitudes of absorption bands (Table 1). For the wild-type protein, the 4 45

spectrum of the complex with the native substrate, camphor, shows a single absorption band at 46 48

47

1939.4 cm-1 with a line width of 13 cm-1. Compared to the camphor complex, the CO band of 50

49

the norcamphor complex is ~7 cm-1 higher in frequency and ~4.5 cm-1 narrower, in accord with 51 52

previous observations.12,15 53

Unlike those of the camphor and norcamphor complexes, the

5

54

spectrum of the thiocamphor complex required three Gaussian functions for adequate fitting, 56 57 58 59 60 ACS Paragon Plus Environment

Page 8 of 23

Page 9 of 23

The Journal of Physical Chemistry

1 3

2

indicating the presence of three distinct CO vibrational bands (Figure 2). The relative areas of 5

4

the three bands suggest that the two low frequency bands (1938.4 and 1945.8 cm-1, respectively) 6 7 8

reflect states that contribute to 38% and 49% 9 10

(A)

(C)

of the total population, respectively, while

20

19

18

17

16

15

14

13

Normalized Absorbance

12

1

the remaining 13% is associated with the band at the highest frequency (~1958 cm-1). (B)

(D)

Interestingly, the frequency and line width of the lowest frequency band are similar to

21 2 Wavenumber (cm-1) Figure 2. FT IR spectra of (A) P450camCO complexes with camphor (blue) and norcamphor (red), (B) P450camCO complex with thiocamphor (black line) and individual Gaussian components (blue, red, and green lines), (C) Y96F P450camCO complexes with camphor (blue) and norcamphor (red), (D) Y96F P450camCO complex with thiocamphor (black line) and individual Gaussian components (blue, red, and green lines).

31

30

29

28

27

26

25

24

23

those of the single band observed for the camphor complex, while the frequency and line width of the other dominant band are similar to those of the single band observed for the norcamphor complex.

32 34

3

The CO absorptions of the Y96F P450camCO-substrate complexes show similar numbers 36

35

of bands, line widths, and relative intensities as observed for the wild-type protein (Figure 2 and 37 38

Table 1). A single, relatively narrow band was observed for the Y96F P450camCO-norcamphor 39 41

40

complex. The spectrum of the Y96F P450camCO-camphor complex is also dominated by a 43

42

single band, although a very small (~2 % integrated area) band is observed at 1962 cm-1. Like 4 45

the wild-type protein, the Y96F P450cam-CO complex with thiocamphor shows an IR spectrum 46 48

47

composed of three bands with relative integrated areas indistinguishable from those of the 50

49

corresponding wild-type spectrum. However, removal of the Y96 hydroxyl group does generally 51 53

52

result in a shift of the CO absorption to higher frequency. The shift was greater for the camphor 54 5 56 57 58 59 60 ACS Paragon Plus Environment

The Journal of Physical Chemistry

Page 10 of 23

1 3

2

complex (6.3 cm-1) than for the norcamphor and thiocamphor complexes (~3 cm-1); the minor 5

4

band of the thiocamphor complex was not shifted by the mutation. 6 7 8

2D IR spectroscopy was then employed to characterize and compare the spectral 10

9

diffusion within the CO absorption bands of the camphor, norcamphor, and thiocamphor 1 12

complexes. Examples of time-dependent 2D IR spectra acquired for the P450camCO-substrate 13 15

14

complexes are shown in Figure 3A-C. The CLS method applied to the Tw-dependent 2D spectra 17

16

in combination with fitting to the linear absorption bands was used to extract the FFCFs (Figure 18 19

4, Tables 1 and S1). 20 21 2 24

23

Table 1. Absorption Spectra Parameters for P450camCO Complexes 25 26 27 substrate

no. bands

ν (cm-1)

FWHM (cm-1)

rel. area (%)

Δ1 (cm-1)

τ1 (ps)

Δ2 (cm-1)

wt

camphor

1

1939.4 ± 0.2

12.9 ± 0.5

-

3.0 ± 0.1

20.7 ± 2.8

4.1 ± 0.1

regioselectivity* (%) 100a

wt

norcamphor

1

1946.2 ± 0.1

9.7 ± 0.4

-

2.7 ± 0.03

19.5 ± 2.5

2.6 ± 0.1

45a

wt

thiocamphor

3

1938.4 ± 1.1

14.4 ± 0.8

38 ± 1

3.4

23.3

4.6

64b

1945.8 ± 0.2

9.2 ± 0.3

49 ± 2







1957.6 ± 1.2

14.3 ± 0.3

13 ± 3

-

-

-

28

32

31

30

29

34

3 35 36 37

4

43

42

41

40

39

38 Y96F

camphor

1

1945.7 ± 0.6

13.6 ± 0.1

-

3.5 ± 0.1

21.7 ± 4.3

4.5 ± 0.8

92a

Y96F

norcamphor

1

1949.0 ± 0.3

11.4 ± 0.2

-

2.9 ± 0.1

13.7 ± 1.3

3.1 ± 0.2

36a

Y96F

thiocamphor

3

1942.7 ± 0.3

12.2 ± 0.5

37 ± 3

3.1

23.5

3.9

-

1948.8 ± 0.2

8.3 ± 0.7

50 ± 6







1958.0 ± 1.1

12.3 ± 0.3

13 ± 4

-

-

-

45 46 47 48 *

49

5-exo product obtained. †Assumed to be identical to the corresponding norcamphor complex a Reference 16. b Reference 17.

52

51

50

53 54 5 56 57 58 59 60 ACS Paragon Plus Environment

Page 11 of 23

The Journal of Physical Chemistry

1 2 3

0.25 ps

4 5 6

1950

7

8 ps

48 ps

0.25 ps

(A)

1955

9

1930

10

1955

15

14

13

1940 1930

17

16

21

20

19

18

1950

(C)

1945 1935 1955

1940

1945

1930

1935 1930 1940 1950

2

(E)

-1

-1

ωm (cm )

ωm (cm )

1950

12

(D)

1935

(B)

1

1930 1940 1950 -1

1930 1940 1950

(F)

1935 1945 1955 1935 1945 1955

ωτ (cm )

23

48 ps

1945

1940

8

8 ps

-1

1935 1945 1955

ωτ (cm )

Figure 3. 2D IR spectra of wild-type P450camCO complexes with (A) camphor, (B) norcamphor, (C) thiocamphor, and Y96F P450camCO complexes with (D) camphor, (E) norcamphor, and (F) thiocamphor at three different waiting times (Tw): 0.25, 8, and 48 ps.

27

26

25

24

The FFCFs of both the camphor and norcamphor complexes show decays on the ~20 ps 28 29

timescale and a large constant offset reflective of dynamics slower than the timescale of the 30 32

31

experiment (~100 ps, limited by the ~20 ps vibrational lifetime of CO). 34

3

The frequency

fluctuation amplitude associated with the ~20 ps timescale dynamics is similar for the camphor 36

35

and norcamphor complexes (∆1 = 3.0 and 2.7 cm-1, respectively), while significantly greater 39

38

37

amplitude is associated with the slowest timescale motions for the camphor (∆2 = 4.1 cm-1) 41

40

compared to the norcamphor (∆2 = 2.6 cm-1) complex. Thus, when bound in the camphor 42 4

43

complex, the CO shows greater heterogeneity in frequency than in the norcamphor complex, and 46

45

the additional frequency fluctuation amplitude is sampled on the slowest timescales. We note 47 48

that while the FFCF decays measured here for the camphor and norcamphor complexes show a 49 51

50

53

52

somewhat slower absolute timescale than those reported previously,15 which may be due to differences in experimental set-up or sample preparation, the differences between the complexes 54 5

are well reproduced. Likely due to the slower timescales, we could not justify fitting a time 56 57 58 59 60 ACS Paragon Plus Environment

The Journal of Physical Chemistry

Page 12 of 23

1 3

2

constant to the slowest component, as was done in the previous study, but rather the slowest 5

4

component was modeled as a constant offset. If the timescale of the slowest component is varied 6 8

7

in the multiexponential fits of the FFCFs, the obtained time constants for the camphor complex 10

9

report slower dynamics than the norcamphor complex (Table S2), consistent with the previous 1 12

study. 13 14 15

Comparison of the 2D spectra of the thiocamphor complex at short and long Tw reveals 17

16

that the amplitude along the diagonal decays more quickly at low compared to high frequencies, 18 19

consistent with its being comprised of two frequency components (Figure 3C). Temporarily 20 2

21

treating the 2D spectra as a single band when applying the CLS analysis, the obtained FFCF 23 24

shows a decay intermediate between that of the

32

31

30

29

28

27

Normalized FFCF

26

25

camphor and norcamphor complexes (Figure 4), likely reflecting the combined presence of the rapidly and slowly sampled component bands.

3 34

To analyze the individual components of the 35

43

42

41

40

39

38

37

36

Tw (ps) Figure 4. FFCF decay curves (symbols) and fits (lines) for the wild-type P450camCO complexes with camphor (black squares), norcamphor (red diamonds), thiocamphor (light blue triangles), and Y96F P450camCO complexes with camphor (blue circles), norcamphor (green triangles), and thiocamphor (purple asterisks).

band, further analysis was performed using a procedure outlined previously24 to determine the FFCFs for multi-component 2D spectra when the

fractional contributions of the component bands and the FFCF of one component are known. 4 45

Because the linear spectrum of the thiocamphor complex was well fit by a sum of bands with 46 48

47

frequencies and line widths that correspond to those of the camphor and norcamphor complexes 50

49

(Table 1), we anticipated that the Tw-dependent line width changes in the 2D spectra likely arise 51 53

52

from the contributions from the FFCFs for the camphor- and norcamphor-like bands. Thus, we 5

54

used the FFCF determined for the norcamphor complex as one component, along with the 56 57 58 59 60 ACS Paragon Plus Environment

Page 13 of 23

The Journal of Physical Chemistry

1 3

2

fractional amplitudes of the two states from the fit to the linear spectrum and their vibrational 5

4

lifetimes,15 to determine the FFCF of the second component. An overlay of the FFCF obtained 6 8

7

in this manner with the FFCF determined for the camphor complex shows that they are 10

9

indistinguishable (Figure 5). Moreover, an exponential fit to the determined FFCF yielded 1 12

amplitudes and time constants similar to those measured for the camphor complex (Table 1). 13 14 15

2D IR spectroscopy was also used to

23

2

21

20

19

18

Normalized FFCF

17

16

characterize the Y96F P450camCO complex with camphor, norcamphor, and thiocamphor (Figure 3D-F).

24

Although the introduction of

Y96F leads to significant changes in the center 25

31

30

29

28

27

26

Tw (ps) Figure 5. FFCF decay curves (symbols) and fits (lines) for the wild-type P450camCO complex with camphor (black squares) and the lower frequency component extracted from the 2D spectra of the thiocamphor complex (red asterisks).

frequency of the observed CO absorption bands, the FFCFs determined for the bands were very

similar to those measured for the wild-type protein (Figure 4), including the FFCFs of the 32 34

3

thiocamphor complex, which were determined as described above (Figure S5). One difference 36

35

observed was a slightly faster time constant for the faster component of the FFCF for the Y96F37 38

norcamphor complex compared to wild-type. However, the mutation did not appear to result in a 39 41

40

similar change for the camphor complex, although a small effect might have been masked by the 43

42

greater error in the data for this complex. Regardless, in comparison to the influence of the 4 45

bound substrate, perturbation to the protein dynamics by Y96F is generally minor. 46 47 48 50

49

DISCUSSION 51 52

As observed in previous studies of P450cam and many other enzymes,11,12,15,25-28 the IR spectra 53 5

54

of CO bound to P450cam is sensitive to the local environments of the ligand. In agreement with 56 57 58 59 60 ACS Paragon Plus Environment

The Journal of Physical Chemistry

Page 14 of 23

1 3

2

previous reports,12,15 the spectra of the camphor and norcamphor complexes show single bands 5

4

that differ in frequency and line width, suggesting that the CO experiences single, but unique 6 8

7

environments in the two complexes. While these different environments have been interpreted 10

9

as being associated with unique conformations of the active site, it is also possible that they are 1 12

associated with specific aspects of the substrate bound. 13

In contrast to the camphor and

15

14

norcamphor complexes, the spectrum of the thiocamphor complex shows three distinct bands. 17

16

Since the three bands reflect three different environments, their observation within a single 18 19

complex reveals the presence of three conformations. Based on the relative integrated areas of 20 2

21

the bands, the two conformations corresponding to the lowest frequency bands make up the 24

23

majority of the population (38 and 49%, Table 1), while the highest frequency band is associated 25 27

26

with a minor conformation (13% population). 29

28

2D IR spectroscopy provides more rigorous analysis of the line widths than possible by 31

30

linear spectroscopy. The line widths for CO bound to all substrate complexes are dominated by 32 34

3

inhomogeneous broadening, which, for a protein-bound oscillator, primarily reflects the 36

35

frequency distribution due to environmental heterogeneity from the ensemble of populated 37 38

substates within the conformation associated with each band. The 2D spectra report on the 39 40

correlation of the CO vibrational frequencies in the P450camCO ensemble before (horizontal, 𝜔𝜏 43

42

41

axis) and after (vertical, 𝜔𝑚 axis) a set waiting time (Tw). At short Tw (left panels, Figure 3), the 4 46

45

2D spectra appear highly elongated along the diagonal, reflecting correlation between the initial 48

47

and final frequencies because the protein is given insufficient time to sample different 50

49

conformational substates. As the Tw time is increased, the protein has more time to sample its 51 53

52

different substates, which leads to changes in the local environment around the CO. This causes 5

54

the initial and final frequencies to lose their correlation, and the 2D spectra appear less elongated 56 57 58 59 60 ACS Paragon Plus Environment

Page 15 of 23

The Journal of Physical Chemistry

1 3

2

(right panels, Figure 3). The Tw-dependent changes in the 2D line shapes may be analyzed to 5

4

determine the FFCF (Figure 4, Table 1), a quantitative description of the spectral dynamics that 6 8

7

reflects the interconversion among substates and thus illuminates the nature of the energy 10

9

landscape within a specific conformation. 1 12

A multi-exponential fit was used to extract the timescales over which the distribution of 13 15

14

substates is sampled within each complex, where each exponential term is assumed to reflect a 17

16

tier of substates in the protein energy landscape. The FFCFs of the CO bands in the P450camCO 18 19

complexes show motion on three timescales: homogeneous dynamics (which are very fast on the 20 2

21

timescale of this measurement), dynamics on an intermediate timescale (~20 ps), and dynamics 24

23

slower than the timescale of the experiment (~100 ps), reflecting three tiers of conformational 26

25

substates. As observed in previous studies,15 the FFCF of the norcamphor complex decayed 27 29

28

more quickly than that of the camphor complex (Figure 4). 31

30

That the fit to the linear spectrum of the thiocamphor complex yielded two dominant 32 34

3

bands with frequencies and line widths corresponding to those of the single bands observed for 36

35

the camphor and norcamphor complexes suggests that the thiocamphor complex populates each 37 38

of the conformations individually populated by the camphor and norcamphor complexes. 39 41

40

Correspondingly, when the 2D spectra of the thiocamphor complex were analyzed as single 43

42

bands, an FFCF decay intermediate between that of the camphor and norcamphor complexes was 4 45

determined (Figure 4), which likely reflects the simultaneous population of the states with more 46 48

47

(norcamphor-like) and less (camphor-like) rapidly decaying FFCFs. We more rigorously tested 50

49

this possibility by analyzing the 2D IR spectra of the thiocamphor complex using the FFCF 51 53

52

determined for the norcamphor complex as one component, in conjunction with the linear FT IR 5

54

spectra and vibrational lifetimes, to extract the FFCFs of the second component (see also SI).24 56 57 58 59 60 ACS Paragon Plus Environment

The Journal of Physical Chemistry

Page 16 of 23

1 3

2

The FFCF component obtained in this manner was virtually identical to that obtained for the 5

4

camphor complex (Figure 5). Thus, not only do the frequencies and line widths of the two 6 8

7

dominant thiocamphor states appear to correspond to those of the norcamphor- and camphor10

9

bound conformations, the dynamics are also strikingly similar. This indicates that the local 1 12

energy landscapes within the conformations are also the same, and strongly supports the 13 15

14

assignment of the two bands of the thiocamphor complex to the individual conformations 17

16

populated in the camphor and norcamphor complexes. 18 19

The difference in frequency between the CO bands in the two observed states implies 20 2

21

differences in the local electrostatics at the active site. A standard interpretation of frequency 24

23

changes in hemeprotein-bound CO attributes an increase (decrease) in vibrational frequency to 25 27

26

an increase (decrease) in negative electron density near the CO that causes a decrease (increase) 29

28

in the extent of Fe d-orbital back-bonding to the π* antibonding orbital of CO.28 For example, 31

30

the higher frequency of the CO band of the norcamphor compared to camphor complex has been 32 34

3

attributed to increased hydration of the active site, presumably counteracting an otherwise 36

35

positive potential.12 This interpretation is supported by crystal structures that show a greater 38

37

number of active site water molecules in the norcamphor complex than in the camphor complex.7 39 41

40

If the higher frequency band is a result of greater active site hydration, the observation that the 43

42

same band appears in the spectrum of the thiocamphor complex indicates that a small substrate 4 45

like norcamphor is not required to accommodate the additional water molecules, as has been 46 48

47

previously suggested.7,10 Similarly, the even higher frequency of the third, minor band observed 50

49

in the thiocamphor complex could reflect an active site with even more water content. In line 51 53

52

with this, a band of comparably high frequency is observed in the spectrum of the substrate-free 5

54

enzyme,11,15 which might reflect the population of a state similar to that characterized by x-ray 56 57 58 59 60 ACS Paragon Plus Environment

Page 17 of 23

The Journal of Physical Chemistry

1 3

2

crystallography for the substrate-free P450camCO that shows an “open” conformation with a 5

4

channel of water molecules that reaches into the active site.29 6 8

7

The introduction of the Y96F mutation, which ablates the H-bond observed in the crystal 10

9

structures of the camphor and norcamphor complexes,6,7 results in CO absorptions that are 12

1

approximately 3-6 cm-1 shifted to higher energy. Based on the model described above, these 13 15

14

shifts suggest that the mutation increases the electron density proximal to the CO ligand, perhaps 17

16

allowing for the accommodation of more water molecules. That the shift is somewhat smaller 18 19

for the thiocamphor complex than the camphor complex may suggest that the reduced H-bonding 20 2

21

potential or the increased size of the sulfur atom reduces the number of water molecules 24

23

introduced, or perhaps that it prevents them from close approach to the CO ligand. However, a 25 27

26

similar shift was observed with the norcamphor complex, which one would expect to be able to 29

28

accommodate at least as many additional water molecules as the camphor complex. It seems 31

30

likely that while changes in active site solvent accommodation do contribute to the spectral 32 34

3

shifts, additional factors likely also contribute. 36

35

For example, both the smaller core of

norcamphor and the increased size of sulfur may induce reorganization of the substrate within a 37 38

given active site conformation, and thus reduce the change in electrostatics induced by the Y96F 39 41

40

mutation. Finally, the minor high frequency absorption of the thiocamphor complex is an 43

42

exception, as its frequency was not affected by the mutation. As discussed above, it is likely that 4 45

this conformation is already associated with more water molecules, which may make it less 46 48

47

sensitive to the precise number of water molecules or to subtle orientational differences of the 50

49

substrate. 51 53

52

Although the Y96F mutation shifts the frequencies of all the major CO bands, somewhat 5

54

surprisingly, it does not alter their relative integrated areas, indicating that the mutation does not 56 57 58 59 60 ACS Paragon Plus Environment

The Journal of Physical Chemistry

Page 18 of 23

1 3

2

affect the populations of conformations. The H-bonding interaction therefore does not appear to 5

4

differentially stabilize the conformations populated in the thiocamphor complex, and by 6 8

7

extension is not likely critical for stabilizing the corresponding conformations populated by the 10

9

camphor and norcamphor complexes. Furthermore, comparison of the FFCFs for the wild-type 1 12

and Y96F complexes show little differences (Figure 4), which indicates that Y96F induces little 13 15

14

change in the local energy landscapes within the populated conformations. One exception is a 17

16

slightly faster time constant obtained for the faster component of the FFCF for the Y96F18 19

norcamphor complex compared to wild-type. This might reflect greater mobility of substrate or 20 2

21

an active site group from either removal of a H-bonding interaction with Y96 or reduction in the 24

23

steric hindrance of the hydroxyl group. The mutation did not appear to result in the same change 25 27

26

in the FFCF for the camphor complex, although a small effect might have been masked by the 29

28

greater error in the data for this complex. In general, however, the change in the protein 31

30

dynamics by Y96F is small compared to the influence of the bound substrate. Thus, the H-bond 32 34

3

is unlikely to contribute substantially to regiocontrol of hydroxylation. 36

35

Importantly, the linear and 2D IR data provide insight into the potential molecular origins 37 38

of P450cam regioselectivity. The high selectivity appears to be correlated with the population of 39 41

40

the camphor-like state and lower selectivity is associated with the population of the norcamphor43

42

like state. Moreover, the regioselectivity associated with a given state appears to be correlated 45

4

with the nature of the FFCFs. As observed previously,15 the FFCF for the camphor-like states 46 48

47

(including the camphor-like state observed with the thiocamphor complex) show less decay on 50

49

the longest timescale than the norcamphor-like states (including the norcamphor-like state 51 53

52

observed with the thiocamphor complex). Analysis of the FFCFs with a multi-exponential 5

54

model suggests that the differences arise from greater frequency fluctuation amplitude associated 56 57 58 59 60 ACS Paragon Plus Environment

Page 19 of 23

The Journal of Physical Chemistry

1 3

2

with the slowest timescale motions. This in turn suggests that a greater distribution of slowly 5

4

sampled frequencies is associated with the higher regioselectivity of substrate hydroxylation. It 6 8

7

is counterintuitive that increased regioselectivity would result from the population of a greater 10

9

number of states; however, it is possible that the greater frequency heterogeneity results from an 1 12

increase in coupling of the CO to active site groups associated with the structure of the camphor13 15

14

like state. While previous studies of P450cam complexes suggested that higher regioselectivity 17

16

was associated with slower active site motions, we could not justify fitting a timescale for the 18 19

slowest protein motions here, as discussed in the Results section. However, an equivalent 20 2

21

analysis of the FFCF reproduces the previously reported results (Table S2). Regardless of the 24

23

fitting details, the 2D data show that the regioselectivity of the enzyme appears to depend on the 25 27

26

nature of the energy landscape within the conformation induced by the substrate. 28 29 31

30

CONCLUSIONS 32 3 34

This study reveals that the intermediate specificity of thiocamphor hydroxylation by P450cam 35 37

36

arises from the population of multiple conformations, and furthermore, links hydroxylation 39

38

regioselectivity to the nature of the local energy landscape within the populated conformation, 40 42

41

and not the average electrostatic environment. Interestingly, the local energy landscape within 4

43

each state appears to be independent of the bound substrate, and is rather a function of which 45 46

conformation is populated, with the role of the substrate being only to favor the population of 47 49

48

one or the other state. The crystallographically observed Y96-substrate H-bond makes only 51

50

small contributions to the dynamics observed within each complex, and no contribution to the 52 53

relative populations of the conformations. In contrast, the populations of the two states likely 54 56

5

depend on enzyme-substrate packing interactions. For example, the data is consistent with the 57 58 59 60 ACS Paragon Plus Environment

The Journal of Physical Chemistry

Page 20 of 23

1 3

2

methyl groups of camphor and thiocamphor preferentially stabilizing the more catalytically 5

4

selective state, and the increased size and/or polarizability of the sulfur substituent of 6 8

7

thiocamphor stabilizing the catalytically more promiscuous state. While earlier studies suggest 10

9

an “imperfect” lock-and-key mechanism, with imperfections arising from the inability of a 12

1

substrate to fully engage packing or H-bonding interactions with the enzyme,5-14 the current data 13 15

14

are strongly indicative of a conformational selection or induced-fit mechanism of substrate 17

16

recognition. 18

Regardless, the data suggest that understanding the selectivity of P450cam

19

hydroxylation requires knowledge of its conformational landscape, and correspondingly, that 20 2

21

additional characterization will further our understanding of how to predict P450 activity with a 24

23

given substrate (or drug) or how to optimize them for synthetic or biotechnological applications. 25 26 27 29

28

ASSOCIATED CONTENT 30 32

31

Supporting Information Available: Additional details about expression and purification of 3 34

wild-type and Y96F P450cam, determination of the Kd of thiocamphor to Y96F P450cam, 35 37

36

preparation of samples for IR experiments, experimental procedures for linear and 2D IR 39

38

spectroscopy, and analysis of the 2D IR spectra, including extraction of FFCFs from multi40 42

41

component 2D spectra, are provided. UV/visible spectra, binding titration data for thiocamphor 4

43

with Y96F P450cam, alternate Gaussian fits to the linear spectrum of the thiocamphor complex, 45 46

and the extracted FFCF of low-frequency component of the Y96F P450cam-thiocamphor 47 49

48

complex are also provided. 51

50

This material is available free of charge via the Internet at

http://pubs.acs.org. 52 53 54 5 57

56

AUTHOR INFORMATION 58 59 60 ACS Paragon Plus Environment

Page 21 of 23

The Journal of Physical Chemistry

1 3

2

Corresponding Author 4 5

*Email: [email protected] 6 7 9

8

Notes 10 12

1

The authors declare no competing financial interests. 13 15

14

ACKNOWLEDGMENT 16 18

17

This work was supported by Indiana University. E.B. was also supported by the Graduate 19 20

Training Program in Quantitative and Chemical Biology (T32 GM109825). 21 2 23 24 26

25

REFERENCES 27 28

(1) Guengerich, F. P., Cytochrome P450 and Chemical Toxicology. Chem. Res. Toxicol. 2008, 21, 7029 30

83. (2) Urlacher, V. B.; Girhard, M., Cytochrome P450 Monooxygenases: An Update on Perspectives for Synthetic Application. Trends Biotechnol. 2012, 30, 26-36. (3) Bernhardt, R., Cytochromes P450 as Versatile Biocatalysts. J. Biotechnol. 2006, 124, 128-145. (4) Wong, L.-l.; Westlake, A.; Nickerson, D., Protein Engineering of Cytochrome P450cam. Struct. Bond. 1997, 88, 175-206. (5) Poulos, T. L.; Raag, R., Cytochrome P450cam: Crystallography, Oxygen Activation, and Electron Transfer. FASEB J. 1992, 6, 674-679. (6) Poulos, T. L.; Finzel, B. C.; Howard, A. J., High-Resolution Crystal Structure of Cytochrome P450cam. J. Mol. Biol. 1987, 195, 687-700. (7) Raag, R.; Poulos, T. L., The Structural Basis for Substrate-Induced Changes in Redox Potential and Spin Equilibrium in Cytochrome P-450cam. Biochemistry 1989, 28, 917-922. (8) Raag, R.; Poulos, T. L., Crystal Structures of Cytochrome P-450cam Complexed with Camphane, Thiocamphor, and Adamantane: Factors Controlling P-450 Substrate Hydroxylation. Biochemistry 1991, 30, 2674-2684. (9) Nagano, S.; Tosha, T.; Ishimori, K.; Morishima, I.; Poulos, T. L., Crystal Structure of the Cytochrome P450cam Mutant That Exhibits the Same Spectral Perturbations Induced by Putidaredoxin Binding. J. Biol. Chem. 2004, 279, 42844-42849. (10) Schulze, H.; Hui Bon Hoa, G.; Jung, C., Mobility of Norbornane-Type Substrates and Water Accessibility in Cytochrome P450cam. Biochim. Biophys. Acta 1997, 1338, 77-92. (11) O'Keefe, D. H.; Ebel, R. E.; Peterson, J. a.; Maxwell, J. C.; Caughey, W. S., An Infrared Spectroscopic Study of Carbon Monoxide Bonding to Ferrous Cytochrome P-450. Biochemistry 1978, 17, 5845-5852. (12) Jung, C.; Hoa, G. H.; Schröder, K. L.; Simon, M.; Doucet, J. P., Substrate Analogue Induced Changes of the CO-Stretching Mode in the Cytochrome P450cam-Carbon Monoxide Complex. Biochemistry 1992, 31, 12855-12862. 57

56

5

54

53

52

51

50

49

48

47

46

45

4

43

42

41

40

39

38

37

36

35

34

3

32

31

58 59 60 ACS Paragon Plus Environment

The Journal of Physical Chemistry

Page 22 of 23

1 3

2

(13) Collins, J. R.; Loew, G. H., Theoretical Study of the Product Specificity in the Hydroxylation of Camphor, Norcamphor, 5,5-Difluorocamphor, and Pericyclocamphanone by Cytochrome P-450cam. J. Biol. Chem. 1988, 263, 3164-3170. (14) Paulsen, M. D.; Filipovic, D.; Sligar, S. G.; Ornstein, R. L., Controlling the Regiospecificity and Coupling of Cytochrome P450cam: T185f Mutant Increases Coupling and Abolishes 3Hydroxynorcamphor Product. Protein Sci. 1993, 2, 357-365. (15) Thielges, M. C.; Chung, J. K.; Fayer, M. D., Protein Dynamics in Cytochrome P450 Molecular Recognition and Substrate Specificity Using 2D IR Vibrational Echo Spectroscopy. J. Am. Chem. Soc. 2011, 133, 3995-4004. (16) Atkins, W. M.; Sligar, S. G., The Roles of Active Site Hydrogen Bonding in Cytochrome P450cam as Revealed by Site-Directed Mutagenesis. J. Biol. Chem. 1988, 263, 18842-18849. (17) Atkins, W. M.; Sligar, S. G., Molecular Recognition in Cytochrome P-450: Alteration of Regioselective Alkane Hydroxylation Via Protein Engineering. J. Am. Chem. Soc. 1989, 111, 859-861. (18) Ouyang, B.; Pochapsky, S. S.; Pagani, G. M.; Pochapsky, T. C., Specific Effects of Potassium Ion Binding on Wild-Type and L358p Cytochrome P450cam. Biochemistry 2006, 45, 14379-14388. (19) Nickerson, D.; Wong, L.-l., The Dimerization of Pseudomonas Putida Cytochrome P450cam: Practical Consequences and Engineering of a Monomeric Enzyme. Protein Eng. 1997, 10, 1357-1361. (20) Gunsalus, I. C.; Wagner, G. C., Bacterial P-450cam Methylene Monooxygenase Components: Cytochrome M, Putidaredoxin, and Putidaredoxin Reductase. Methods Enzymol. 1978, 58, 166-188. (21) Unger, B. P.; Gunsalus, I. C.; Sligar, S. G., Nucleotide Sequence of the Pseudomonas Putida Cytochrome P-450cam Gene and Its Expression in Escherichia Coli. J. Biol. Chem. 1986, 261, 11581163. (22) Park, S.; Kwak, K.; Fayer, M. D., Ultrafast 2D-IR Vibrational Echo Spectroscopy: A Probe of Molecular Dynamics. Laser Phys. Lett. 2007, 4, 704-718. (23) Kwak, K.; Park, S.; Finkelstein, I. J.; Fayer, M. D., Frequency-Frequency Correlation Functions and Apodization in Two-Dimensional Infrared Vibrational Echo Spectroscopy: A New Approach. J. Chem. Phys. 2007, 127, 124503. (24) Fenn, E.; Fayer, M. D., Extracting 2D IR Frequency-Frequency Correlation Functions from Two Component Systems. J. Chem. Phys. 2011, 135, 074502. (25) Nagano, S.; Shimada, H.; Tarumi, A.; Hishiki, T.; Kimata-Ariga, Y.; Egawa, T.; Suematsu, M.; Park, S.-Y.; Adachi, S.-i.; Shiro, Y.; et al., Infrared Spectroscopic and Mutational Studies on Putidaredoxin-Induced Conformational Changes in Ferrous CO-P450cam. Biochemistry 2003, 42, 1450714514. (26) Sun, Y.; Zeng, W.; Benabbas, A.; Ye, X.; Denisov, I.; Sligar, S. G.; Du, J.; Dawson, J. H.; Champion, P. M., Investigations of Heme Ligation and Ligand Switching in Cytochromes P450 and P420. Biochemistry 2013, 52, 5941-5951. (27) Phillips Jr., G. N.; Teodoro, M. L.; Li, T.; Smith, B.; Olson, J. S., Bound CO is a Molecular Probe of Electrostatic Potential in the Distal Pocket of Myoglobin. J. Phys. Chem. B 1999, 103, 8817-8829. (28) Spiro, T. G.; Wasbotten, I. H., CO as a Vibrational Probe of Heme Protein Active Sites. J. Inorg. Biochem. 2005, 99, 34-44. (29) Lee, Y. T.; Wilson, R. F.; Rupniewski, I.; Goodin, D. B., P450cam Visits an Open Conformation in the Absence of Substrate. Biochemistry 2010, 49, 3412-3419. 48

47

46

45

4

43

42

41

40

39

38

37

36

35

34

3

32

31

30

29

28

27

26

25

24

23

2

21

20

19

18

17

16

15

14

13

12

1

10

9

8

7

6

5

4

49 50 51 52 53 54 5 56 57 58 59 60 ACS Paragon Plus Environment

Page 23 of 23

The Journal of Physical Chemistry

1 3

2

TOC graphic: 4 5 6 7 8 9 10 1 12 13 14 15 16 17 18 19 20 21 2 23 24 25 26 27 28 29 30 31 32 3 34 35 36 37 38 39 40 41 42 43 4 45 46 47 48 49 50 51 52 53 54 5 56 57 58 59 60 ACS Paragon Plus Environment

Conformational landscape and the selectivity of cytochrome P450cam.

Conformational heterogeneity and dynamics likely contribute to the remarkable activity of enzymes but are challenging to characterize experimentally. ...
1MB Sizes 1 Downloads 9 Views