2674
`
`Biochemistry 1991, 30, 2674-2684
`
`Crystal Structures of Cytochrome P-450cAM Complexed with Camphane,
`Thiocamphor, and Adamantane: Factors Controlling P-450 Substrate
`Hydroxylation t.*
`
`Reetta Raag§ and Thomas L. Poulos*
`Center for Advanced Research in Biotechnology of the Maryland Biotechnology Institute, University of Maryland at Shady
`Grove, 9600 Gudelsky Drive, Rockville, Maryland 20850, and the Department of Chemistry and Biochemistry, University of
`Maryland, College Park, Maryland 20742-5115
`Received September 26, 1990; Revised Manuscript Received December 4, 1990
`
`ABSTRACT: X-ray crystal structures have been determined for complexes of cytochrome P-450cAM with the
`substrates camphane, adamantane, and thiocamphor. Unlike the natural substrate camphor, which hydrogen
`bonds to Tyr96 and is metabolized to a single product, camphane, adamantane and thiocamphor do not
`hydrogen bond to the enzyme and all are hydroxylated at multiple positions. Evidently the lack of a
`substrate-enzyme hydrogen bond allows substrates greater mobility in the active site, explaining this lower
`regiospecificity of metabolism as well as the inability of these substrates to displace the distal ligand to the
`heme iron. Tyr96 is a ligand, via its carbonyl oxygen atom, to a cation that is thought to stabilize the
`camphor-P-450cAM complex [Poulos, T. L., Finzel, B. C., & Howard, A. J. (1987) J. Mol. Bioi. 195,
`687-700]. The occupancy and temperature factor of the cationic site are lower and higher, respectively,
`in the presence of the non-hydrogen-bonding substrates investigated here than in the presence of camphor,
`underscoring the relationship between cation and substrate binding. Thiocamphor gave the most unexpected
`orientation in the active site of any of the substrates we have investigated to date. The orientation of
`thiocamphor is quite different from that of camphor. That is, carbons 5 and 6, at which thiocamphor is
`primarily hydroxylated [Atkins, W. M., & Sligar, S. G. (1988) J. Bioi. Chern. 263, 18842-18849], are
`positioned near Tyr96 rather than near the heme iron. Therefore, the crystallographically observed thio(cid:173)
`camphor-P-450cAM structure may correspond to a nonproductive complex. Disordered solvent has been
`identified in the active site in the presence of uncoupling substrates that channel reducing equivalents away
`from substrate hydroxylation toward hydrogen peroxide and/or "excess" water production. A buried solvent
`molecule has also been identified, which may promote uncoupling by moving from an internal location to
`the active site in the presence of highly mobile substrates.
`
`he cytochrome P-450 superfamily of enzymes catalyzes
`
`many different types of oxidative reactions involved in steroid
`hormone biosynthesis, fatty acid metabolism, and detoxifica(cid:173)
`tion of foreign compounds (Nebert et al., 1981; Nebert &
`Gonzalez, 1987; Anders, 1985). Xenobiotic-metabolizing
`P-450s generally oxidize substrates to more soluble forms,
`facilitating their excretion. Occasionally these products linger
`in the cytoplasm as "activated" electrophilic compounds, many
`of which are mutagens and/or carcinogens (Heidelberger,
`1975; Sato & Omura, 1978; Anders, 1985; Ortiz deMon(cid:173)
`tellano, 1986; Wolf, 1986). Due to the broad substrate spe(cid:173)
`cificity of this superfamily, and its ability to catalyze multiple
`types of reactions, there is much interest in structure-function
`relationships of P-450s. Ultimate goals include designing
`compounds to selectively inhibit individual P-450s and engi(cid:173)
`neering novel P-450s to facilitate detoxification of specific
`environmental contaminants.
`The best characterized P-450, and the only one for which
`a crystal structure is known, is the bacterial camphor hy(cid:173)
`droxylase P-450cAM (Gunsalus et al., 1974; Debrunner et al.,
`
`tsupported in part by NIH Grant GM 33688.
`!Crystallographic coordinates have been submitted to the Brookhaven
`Protein Data Bank under the following file names: 4CPP, cytochrome
`P-450cAM-adamantane; 6CPP, cytochrome P-450cAM--camphane; 8CPP,
`cytochrome P-450cAM-thiocamphor.
`*Correspondence should be addressed to T.L.P. at CARB.
`§University of Maryland.
`
`1978; Gunsalus & Sligar, 1978; Ullrich, 1979; Wagner &
`Gunsalus, 1982; Poulos eta!., 1985, 1987). The reaction cycle
`of cytochrome P-450cAM is shown in Figure I. Besides hy(cid:173)
`droxylating camphor, P-450cAM will also hydroxylate various
`other compounds. We have determined the X-ray crystal
`structures of ferric cytochrome P-450cAM complexed with
`different substrates and inhibitors, as well as in the ferrous
`carbon monoxide and camphor bound form (Raag & Poulos,
`1989a,b, 1990; Raag et a!., 1990). These structures, together
`with data on substrate-dependent parameters and site-directed
`mutagenesis of P-450s (White eta!., 1984; Fisher & Sligar,
`1985; Atkins & Sligar, 1988a,b), have enabled us to better
`understand factors that influence regiospecificity and efficiency
`of P-450 reactions. Here we extend these studies to include
`three additional substrates. All substrate-P-450cAM coordi(cid:173)
`nates have been submitted to the Brookhaven Protein Data
`Bank (Bernstein et a!., 1977).
`
`MATERIALS AND METHODS
`
`Thiocamphor synthesis was according to Scheeren et a!.
`(1973) with the exception that P2S5 (FLUKA, Ronkonkoma,
`NY) was used in place of P4S 10• Samples were analyzed
`(Galbraith Laboratories, Knoxville, TN) for C, H, 0, and S
`to confirm that the correct compound had been prepared.
`P-450cAM was crystallized according to our earlier procedure
`(Poulos et al, 1982). To prepare the various substrate-P-
`450cAM complexes, crystals were soaked in a mother liquor
`
`0006-2960/91/0430-2674$02.50/0
`
`© 1991 American Chemical Society
`
`MYLAN - EXHIBIT 1009
`
`

`

`P-450cAM-Substrate Crystal Structures
`
`Biochemistry, Vol. 30, No. 10, 1991 2675
`
`8
`
`0
`
`F 2+
`:~02fe
`:
`o·
`'o
`Fe 3 •
`l~Lo~y/--e- SH
`~ Fe 2
`•
`I
`\
`SH
`H20
`2H•
`FIGURE I: P-450 reaction cycle [modified from Atkins and Sligar
`( 1988a) ). SH and SOH represent substrate and oxidized substrate,
`respectively. "Uncoupling~ reactions compete with substrate ~y­
`droxylation.
`"Efficiency~ refers to the percentage of reducmg
`equivalents utilized toward substrate oxidation, as opposed to hydrogen
`peroxidej"excess" water production.
`
`0
`
`+3
`
`Table 1: Summary of Substrate-P-450 Data Collection
`substrate
`adamantane
`thiocamphor
`camphane
`2.11 A
`!.91 A
`2.09 A
`max rcsolutn
`total observatns
`144 737
`133 336
`I 08 560
`0.079
`0.058
`0.067
`Rsym0
`3.79 A, 100%
`3.47 A, 100%
`3.82 A, 100%
`%data collected to
`2.76 A, 100%
`3.04 A, 100%
`3.01 A, 100%
`2.63 A, 100%
`2.41 A, 100%
`2.65 A, 100%
`2.19 A, 100%
`2.41 A, 100%
`2.39 A, 100%
`2.03 A, 72%
`2.24 A, 100%
`2.22 A, 100%
`1.91 A, 45%
`2.11 A, 73%
`2.09 A, 54%
`2.19 A, 1.91
`2.41 A, 2.50
`2.13 A, 2.15
`2.11 A, 2.13
`2.03 A, 1.08
`2.24 A, 1.46
`1.91 A, o.ss
`2.11 A, 0.12
`2.09 A, 0.24
`• Rsym = L:l/1 -
`(11)J/L;l1 where / 1 -
`intensity of the ith observation
`and (/1) = mean intensity.
`
`lju(l)
`
`consisting of 40% saturated ammonium sulfate, 0.05 M po(cid:173)
`tassium phosphate, and 0.25 M KCI at pH 7.0, with saturating
`amounts of camphane, adamantane, or thiocamphor. Soak
`times were about three to four days. X-ray diffraction data
`were collected from single crystals of the various substrate(cid:173)
`P-450cAM complexes by using a Siemens area detector /Rigaku
`rotating anode and processed by using the XENGEN program
`package (Howard et al., 1987) on a Digital Equipment Cor(cid:173)
`poration Microvax II. Data collection statistics are presented
`in Table I.
`Substrates were initially sketched by using the Chemnote
`two-dimensional molecular construction facility in the mo(cid:173)
`lecular modeling package QUANTA (Polygen Corp., Waltham,
`MA), installed on a Silicon Graphics IRIS workstation.
`Following two-dimensional model building, substrate coor(cid:173)
`dinates were energy minimized, again through QUANTA, using
`CHARMm steepest descents and Newton-Raphson energy
`minimization procedures. Substrate van der Waals volumes
`were calculated with QUANTA as well. Thiocamphor was
`modeled by substituting sulfur for oxygen in camphor, with
`the sulfur-carbon bond length maintained at the corresponding
`value for the oxygen-carbon bond. Such a model for thio(cid:173)
`camphor should be adequate since the bond orders of C=S
`and C=O bonds are similar (Demarco et al., 1969).
`Crystallographic refinement was carried out by using the
`restrained parameters-least squares package of programs
`
`Table II: Summary of Substrate-P-450 Crystallographic Refinement
`adaman-
`thio-
`tane
`camphor
`camphane
`substrate
`resolutn range (A)
`I O.Q--2.1
`I O.Q--2.1
`10.0-1.9
`20 548
`22 650
`retlectns measured
`27 786
`15174
`19565
`retlectns used•
`20 585
`R factorb
`0.184
`0.175
`0.190
`rms deviation of bond dist (A)
`0.019
`0.020
`0.020
`0.033
`0.033
`bond angles (A)
`0.032
`0.037
`0.036
`dihedral angles (A)
`0.036
`• Reflections with I > 2u(l): I = intensity. b R = L:JFo - F,l/L:Fo.
`
`(Hendrickson & Konnert, 1980) and is summarized in Table
`II. Initial F0 - Fe 1 and 2F0 - Fe difference Fourier maps were
`based on structure factor calculations using coordinates from
`the 1.7-A refined camphor-P-450cAM structure (Poulos et al.,
`1987) and diffraction data obtained from the substrate-P-
`450cAM complexes. Camphor coordinates were not included
`in the initial structure factor calculations. F0 - Fe maps were
`contoured at ±3u and 2F0 - Fe maps were contoured at +0.5
`and + 1 u ( u is the standard deviation calculated over an entire
`asymmetric unit of the electron density map). Substrates were
`positioned into the F0 - Fe maps and refined together with the
`protein, with substrate temperature factors initially starting
`at 19-20 A 2, or near the mean temperature factor for all
`protein and heme atoms in the camphor-P-450cAM structure.
`Structures were judged to have refined sufficiently once F0
`- Fe maps showed little or no interpretable density when
`contoured at 3u. Initial F0 - Fe and final 2F0 - Fe maps are
`shown in Figures 2 and 3. Refined models were subjected
`to additional refinement without bond, angle, or nonbonded
`contact distance restraints to better estimate active site dis(cid:173)
`tances. Comparison of both coordinate and temperature factor
`shifts was carried out as described elsewhere (Poulos &
`Howard, 1987).
`
`RESULTS
`
`Figures 2 and 3 show the initial F0 - Fe and final 2F0 - Fe
`maps of the camphane-, adamantane-, and thiocamphor-P-
`450cAM complexes. Modeling of camphane and adamantane
`bound to the enzyme was relatively straightforward. However,
`neither adamantane nor camphane is able to hydrogen bond
`to P-450cAM and so we cannot be sure that these substrates
`do not occupy multiple orientations. Since models with single
`orientations for these substrates were successful in eliminating
`most of the difference electron density from the active site
`region, we take this as evidence that at least the major binding
`orientations of these substrates have been identified.
`Camphane. Despite the similarity in structure and binding
`orientation between camphane and camphor, the atomic tem(cid:173)
`perature factors of camphane refined to values (30 A2) about
`twice those of camphor (16 A 2). This indicates that camphane
`is highly mobile when bound to P-450cAM· One factor that
`could artificially raise temperature factors is an inaccurate
`model. We modeled the observed active site electron density
`with a single, fully occupied camphane molecule, but such a
`model would be inaccurate if the camphane occupancy in the
`crystal was incomplete. This is feasible since neither camphane
`nor adamantane was especially soluble in the crystallization
`mother liquor. However, since we were able to successfully
`model these substrates with full occupancy, the contribution
`
`1 Abbreviations: F,, calculated structure factors; F0 , observed struc(cid:173)
`ture factors; L6, sixth or distal ligand to heme iron; R factor, L:IFo -
`F,J/L:Fo.
`
`

`

`2676 Biochemistry, Vol. 30, No. /0, /99/
`A
`
`B
`
`D
`
`FIGt:RE 2: Initial F0 - F. difference electron density maps for P-450CAM
`complexed with camphane (A), adamantane (B), and thiocamphor
`(C. D). Maps were calculated with diffraction amplitudes from
`substrate-P-450cAM complexes and phases from camphor-P-450cAM
`coordinates (Poulos eta!.. 1987). Maps are contoured at ±3u with
`negative and positive density depicted as dotted and solid lines, re(cid:173)
`spectively. Substrate coordinates were not included in phase calcu(cid:173)
`lations for maps A, B, and C, but camphor coordinates were included
`in the calculation of map D. Map D, with the camphor carbonyl
`oxygen in negative density and with positive density between the
`substrate and heme. indicates that thiocamphor binds "upside down"
`in the active site, compared with camphor. Note the electron density
`corresponding to the distal ligand to iron in all rna ps.
`
`from the substrate-free structure is probably minimal (less than
`20%). The presence of minor, unmodeled, binding orientations
`due to the lack of an enzyme-substrate hydrogen bond also
`could artificially raise substrate temperature factors.
`With these caveats in mind, we still believe the high tem(cid:173)
`perature factor of camphane is genuine. One reason is that
`
`Raag and Poulos
`
`A
`
`B
`
`c
`
`FIGURE 3: Final 2F0 - Fe electron density maps for P-450cAM com(cid:173)
`plexed with camphane (A). adamantane (B), and thiocamphor (C).
`Substrate and distal ligand coordinates were omitted from phase
`calculations for the maps shown. Note the electron density corre·
`spending to the distal ligand to iron in all maps.
`
`the temperature factors of Tyr96 side-chain atoms in the
`camphane-P-450cAM structure are in the vicinity of 28 A2 and
`only about 13 A2 in the camphor-P-450cAM complex, sug(cid:173)
`gesting that the high camphane atomic temperature factors
`are reaL
`Adamantane. Temperature factors of Tyr96 side-chain
`atoms were about 19 A 2 in adamantane-bound and 10 A2 in
`adamantanone-bound P-450cAM structures. These values are
`consistent with the lower average temperature factor of ada(cid:173)
`mantane (24 A2) than camphane and indicate that adaman(cid:173)
`tane is less mobile than camphane when bound by P-450cAM·
`We were concerned about this implication, considering that
`adamantane is highly symmetric and smaller than camphane
`and that neither substrate is able to hydrogen bond to the
`enzyme. Therefore we performed several refinement exper·
`iments.
`Using the refined coordinates for adamantane-P-450cAM•
`with an R factor of 18.1 %, we repositioned adamantane in the
`substrate electron density in two different ways: (I) by ap(cid:173)
`proximately switching the locations of secondary and tertiary
`carbons and (2) by approximately switching the locations of
`atoms and bonds. Both of these repositionings resulted in a
`somewhat poorer fit of the substrate to the electron density.
`Next, 10 cycles of refinement were conducted in which tern·
`
`

`

`P-450cAM-Substrate Crystal Structures
`
`Biochemistry, Vol. 30, No. 10, 1991 2677
`
`perature factors of all atoms and occupancies of solvent atoms
`were refined alternately. In both experiments, the average
`temperature factor of adamantane increased by about 2.5 A2
`(to 26.1 and 25.9 A2, respectively). Ten control refinement
`cycles increased the temperature factor of the best-fit ada(cid:173)
`mantane model by approximately 1.0 A2 to 24.7 A2• In all
`three cases, the R factor dropped only 0.1% to 18.0%. How(cid:173)
`ever, electron density maps calculated with both sets of re(cid:173)
`positioned coordinates contained substantial amounts of ±3u
`F0 - Fe difference electron density, indicating that the substrate
`was incorrectly positioned.
`Our next two experiments involved fixing the temperature
`factors of all adamantane atoms arbitrarily at 16.0 and 32.0
`A 2 and calculating electron density maps to determine if there
`would be any observable effects. No differences were found
`either in R factor or in electron density maps when refined
`adamantane temperature factors (23.6 A2 average) or arbitrary
`ones of 32.0 A2 were used. However, when adamantane at(cid:173)
`omic temperature factors were set to 16.0 A2, although the
`R factor remained at 18.0%, positive 3u difference electron
`density appeared around the substrate in the F0 - Fe map,
`suggesting that the new temperature factors were incorrect.
`In the final adamantane refinement experiment, the ada(cid:173)
`mantanc model considered best fit to the electron density was
`again used but all substrate atomic temperature factors were
`started at 32.0 A2. After 10 refinement cycles, the average
`atomic temperature factor for adamantane rose insignificantly
`to 32.1 A2, rather than dropping toward the previously de(cid:173)
`termined value of 23.6 A2. Once again, the R factor remained
`at \8.0% and difference electron density maps showed no
`indication that substrate temperature factors might be in(cid:173)
`correct. On the basis of these refinement experiments, we
`cannot give a definitive value for the adamantane temperature
`factor, but we regard it as being somewhere in the neigh(cid:173)
`borhood of 25-32 A2.
`- Fe map,
`In the initial thiocamphor F0
`Thiocamphor.
`which was based on camphor-P-450cAM coordinates without
`camphor included, the substrate appeared to be binding "upside
`down" with respect to camphor (Figure 2C). Thus a second
`F0 - Fe map was calculated, again based on the camphor-P-
`450cAM coordinates, but this time including camphor coor(cid:173)
`dinates, in their original orientation, in the phase calculation.
`When this map was contoured at ±3u, the camphor carbonyl
`oxygen was found to occupy a region of negative difference
`electron density and a large region of positive difference
`electron density remained between the substrate and heme
`(Figure 20). Although it seemed apparent that thiocamphor
`and camphor do not bind to P-450cAM in the same orientation,
`another map was calculated, based on coordinates in which
`thiocamphor occupied the same orientation as camphor and
`in which an additional water molecule was included to occupy
`the positive density near the distal heme ligation site. As
`expected, the thiocamphor sulfur atom was surrounded by
`negative density and the water ligand was insufficient to ac(cid:173)
`count fully for the positive density between thiocamphor and
`the heme. Next, the distal water ligand was removed and
`thiocamphor was rotated (by approximately 180° in the plane
`of Figures 2 and 3) so that the thiocarbonyl was no longer
`directed toward Tyr96 but was directed toward the heme iron.
`Maps based on this model contained much less difference
`electron density but some positive difference density still re(cid:173)
`mained between the sulfur and the heme iron. After including
`the distal ligand again, with thiocamphor in the new "upside
`down" orientation (Figure 4A), only a small amount of positive
`and negative difference density remained to either side of the
`
`A
`
`~2 p ='
`
`FIGURE 4: Comparison of the two different binding orientations
`determined crystallographically for thiocamphor. The presence of
`the distal ligand is only compatible sterically with the minor (30%)
`orientation shown on the right.
`
`sulfur atom: positive between the substrate and porphyrin ring;
`negative between the substrate and distal helix. Since a ro(cid:173)
`tation of the sulfur to better accommodate this residual density
`resulted in a much deteriorated fit of the thiocamphor methyl
`groups to the electron density, we attempted to fit thiocamphor
`to the density in yet a third orientation. This orientation again
`had the sulfur directed toward the heme, but now the six(cid:173)
`membered ring of the substrate was essentially parallel to the
`porphyrin plane (Figure 4B), rather than perpendicular as it
`had been previously (and as it is when camphor binds). The
`new F0 - Fe map based on this thiocamphor orientation had
`considerable difference electron density, indicating that this
`was not the major binding orientation of thiocamphor.
`Finally, we were best able to minimize the difference
`electron density by including thiocamphor in both orientations
`(Figure 4), neither corresponding to that of camphor. The
`occupancy of the first orientation was estimated and fixed at
`70%. In this position the sulfur atom approaches to within
`2.35 A of the distal ligand and most likely displaces it. The
`occupancy of the second orientation was fixed at 30%, and in
`this position the sulfur atom is 3.80 A from the distal ligand.
`Curiously enough, when the occupancy of the distal ligand was
`fixed at 0.30 (to correspond with the second thiocamphor
`orientation) and only its temperature factor allowed to refine,
`although the temperature factor dropped to quite a low value
`(I 0 A 2), positive difference density still remained around this
`ligand in F0 - Fe maps. Nor did subsequent release of the fixed
`occupancy of the distal ligand, during refinement, remove the
`difference density, despite the fact that the occupancy climbed
`to 0.51 (after 20 refinement cycles; the temperature factor also
`dropped to 8.3 A2 during this time). These results suggest that
`occupancies and temperature factors are unable to recover,
`in a reasonable number of refinement cycles, from initial poor
`estimates of these values. They also confirm that though
`highly correlated, occupancies and temperature factors are not
`entirely interchangeable as modeling parameters.
`We were only able to eliminate the difference electron
`density from around the distal ligand by including it initially
`with full occupancy and a temperature factor of 20.0 A2, near
`the mean temperature factor for all protein and heme atoms,
`and allowing both occupancy and temperature factor of this
`ligand to refine. In the final model, the distal ligand has a
`temperature factor of 19.6 A2 and an occupancy of 0.90 and
`F0 - Fe maps calculated with these coordinates show no re(cid:173)
`sidual difference density around the sixth ligation position. The
`temperature factor of thiocamphor itself refined to about 23.5
`A2, or about 50% higher than that of camphor bound in the
`active site. This higher mobility for thiocamphor, a larger
`substrate than camphor, suggests that thiocamphor may oc(cid:173)
`cupy additional minor orientations, possibly the one corre(cid:173)
`sponding to camphor bound to the enzyme. Minor thio(cid:173)
`camphor orientations which we have not modeled could po(cid:173)
`tentially account also for the discrepancy between the occu(cid:173)
`pancies of the distal ligand and of thiocamphor orientation 2
`(Figure 4B). Another influential feature could be the fact that
`
`

`

`2678 Biochemistry, Vol. 30, No. 10, 1991
`
`Table Ill: Various Substrate-Dependent ParametersK
`
`camphor
`
`adamantanone
`
`adamantane
`
`thiocamphor
`
`camphor/Y96F
`
`Raag and Poulos
`
`otD ®
`
`norcamphor
`
`ca!)lphane
`
`molec vol
`hydrogen bond to Y96
`no. of iron ligands
`rcd~x pot. Fel+ / F e2+
`high-spin%
`regiospecif of substr hydroxylatn
`
`300 A3
`3151\3
`yesb
`yes•
`s•
`5b
`-175mV'
`-170 mY'
`96-98%'·d
`94-97%""'
`5-exo (I 00% )d-f 5 (IOO%)d
`
`293 A3
`no
`6
`
`322 A3
`no
`6
`
`315 A3
`no
`
`99%d
`65%'
`I (100%)d 5-exo (64%)'
`6-exo (34%)
`3-exo (2%)
`
`59%'
`5-exo (92% )•I
`4 (1%)
`6-exo (2-4%)
`3-exo (Q-:.4%)
`9 (<1%)
`
`yes
`100%'
`
`309 A 3
`no
`6
`
`236 A3
`yesh
`6b
`-206 mY'
`46%'
`46%'
`5-exo (45%Y 5-exo (90% )'
`6-exo (47%)
`6-exo (10%)
`3-exo (8%)
`
`33.5 A 2b
`yes
`12o/J
`
`30.1 A2
`no
`8%'
`
`3.0 Ab
`
`2.88 A
`
`substr temp factor (Fe3+)
`substr hydrophilic groups
`hydroxylatn
`"efficiency"
`L6-substr dist
`
`16.2 A 2 •
`yes
`100%•1
`
`16.5A2 b
`yes
`
`24.7 A2
`no
`
`23.5 A 2
`yes
`98%'
`
`NA
`
`NA
`
`2.63 A
`
`2.35 A (70%)
`3.35 A (30%)
`1.73 Ab
`1.67 A
`1.35 A
`1.95 A
`NA
`NA
`L6-iron dist
`0.97b
`1.00
`0.90
`1.00
`NA
`NA
`L6 occupancy
`7.7 A2
`3.8 A 2 b
`19.6 A 2
`14.3 A 2
`NA
`NA
`L6 temp factor
`l.OOb
`l.OOb
`0.72
`0.91
`0.89
`1.00•
`cation occupancy
`10.0 A2 b
`14.2 A 2
`21.7 A 2
`7.9 A2 h
`15.5 A2
`12.1 A 2•
`cation temp factor
`• Poulos et al. (1985, 1987). b Raag and Poulos (1989a). 'Fisher and Sligar (1985). dWhite et al. (1984). 'Atkins and Sligar ( 1988b). f Atkins
`and Sligar ( 1989). KCarbon numbering for each substrate begins with C-1 at the top of the six-membered ring which is in the plane of the table.
`Numbering proceeds counterclockwise such that the carbonyl carbon is C-2 and C-5 is in the lower right-hand portion of the ring. Note that C-5 is
`a secondary carbon in some substances and a tertiary carbon in others.
`
`the distal ligand is located between the heme iron and the
`substrate sulfur atom. These two neighbors could conceivably
`interact via the distal ligand and increase electron density at
`this location, which could be reflected in an anomalously high
`ligand occupancy.
`Although initial occupancy estimates for the two thio(cid:173)
`camphor orientations were successful in eliminating sub(cid:173)
`strate-associated difference electron density, we decided to
`explore other occupancy combinations because of the dis(cid:173)
`crepancy between the refined occupancy of the distal ligand
`(0.90) and the estimated occupancy of the thiocamphor ori(cid:173)
`entation (0.30), which would be sterically compatible with the
`presence of the ligand. After calculating and examining maps
`based on occupancy combinations ranging from 0.30/0.70 to
`0.80/0.20 in increments of 0.10, we concluded that the relative
`occupancies of thiocamphor orientations I and 2 (parts A and
`8 of Figure 4, respectively) are probably around 65% and 35%,
`respectively, with an error of roughly 10%.
`
`DISCUSSION
`Substrate Hydroxylation Profiles
`Camphane. Although camphane is incapable of hydrogen
`bonding with Tyr96, its similarity to camphor in overall shape
`and size causes it to be bound in a nearly identical position
`in the P-450cAM active site. The methyl groups of camphane
`and camphor interact with the same active site features. As
`with camphor, the 5-carbon atom of camphane is the nearest
`to the heme iron atom, explaining the observed preference
`(90% of products) for 5-exo hydroxylation of this substrate
`(Atkins & Sligar, 1988b ). That I 0% of the products are 6-exo
`hydroxylated (Atkins & Sligar, 1988b) can be attributed to
`the enhanced mobility of camphane in the P-450cAM active
`site (Table I II).
`Hydroxylation profiles and crystallographic data on nor(cid:173)
`camphor- and camphane-P-450cAM complexes, in comparison
`with camphor complexes, demonstrate that two features, a
`hydrogen bond to the enzyme and complementary van der
`
`Waals interactions, are both necessary to lower the mobility
`of a substrate. Low substrate mobility, as revealed by crys(cid:173)
`tallographic temperature factors, appears to be critical for high
`regiospecificity of substrate metabolism (Table III).
`Adamantane. Adamantane is the only substrate we have
`investigated, in this study, that is metabolized to a single
`product despite having a relatively high active site mobility.
`The single product can be attributed to the existence of only
`two types of unique carbon atoms in adamantane, together
`with the greater reactivity of tertiary versus secondary carbons
`(White et a!., 1984).
`Thiocamphor. Thiocamphor binds to P-450cAM in two
`orientations, both of which are different from that preferred
`by camphor and both of which have sulfur as the substrate
`atom nearest to iron. A priori, the proximity of the sulfur atom
`to the heme suggests that the thiocamphor hydroxylation
`mechanism might involve an initial single electron transfer
`from sulfur to heme instead of, or in competition with, initial
`hydrogen abstraction, as is thought to occur with camphor
`(Ortiz de Montellano, 1986). However, the major products
`of thiocamphor metabolism are 5- and 6-exo hydroxylated,
`and these substrate atoms are among the farthest substrate
`atoms from the active oxygen location in our thiocamphor(cid:173)
`P-450cAM model. Modeling of thiocamphor in the orientation
`preferred by camphor (with full occupancy) resulted in dif(cid:173)
`ference electron density maps strongly suggesting that such
`a model was incorrect. Nevertheless, the products of thio(cid:173)
`camphor hydroxylation imply that this substrate is only me(cid:173)
`tabolized when it adopts a camphor-like orientation in the
`active site. These data lead us to conclude that the conformers
`seen in the crystal structure are nonproductive and that the
`camphor-like conformer is fractionally occupied and crys(cid:173)
`tallographically unobservable. Although thiocamphor appears
`to make a snug van der Waals fit with P-450cAM• it may be
`possible for it to occassionally rotate within the active site to
`yield a camphor-like complex, as suggested by molecular dy(cid:173)
`namics simulations of the Tyr96Phe mutant-camphor complex
`
`

`

`P-450cAM-Substrate Crystal Structures
`
`Biochemistry, Vol. 30, No. 10, 1991
`
`2679
`
`(Richard Ornstein and Mark Paulsen, personal communica(cid:173)
`tion).
`
`Distal Aqua Ligand (L6)
`Camphor and adatnantanone both have low temperature
`factors and displace the distal ligand, while all of the other
`substrates investigated here are more loosely bound and do
`not displace the ligand. As with norcamphor (Raag & Poulos,
`I989a), the distal ligand in the presence of camphane has full
`occupancy and a very iow temperature factor (7.7 A2), indi(cid:173)
`cating that the 46% high-spin percentage of the camphane(cid:173)
`bound enzyme is probably not due to partial occupancy of the
`distal ligation site. Rather, by displacing most of the active
`site solvent of the substrate-free enzyme (Poulos eta!., 1986),
`camphane increases the active site hydrophobicity, thereby
`shifting the OW /H 20 equilibrium toward H 20 and yielding
`an increase in redox potential. Partial protonation of the distal
`ligand decreases its ligand field strength, which is responsible
`for the increase in high-spin percentage. This argument is
`supported by the observation that only high-spin P-450cAM
`is protonated (Sligar & Gunsalus, 1979).
`The adamantane-P-450cAM complex is especially interesting
`because, although the small size and relatively high mobility
`of this substrate allow the heme to remain hexacoordinate, this
`complex is fully high-spin like the pentacoordinate camphor(cid:173)
`and adamantanone-P-450cAM complexes (Table III). There
`are two factors that could account for the high-spin nature
`of the adamantane complex: a relatively long Fe-L6 distance
`and a high distal ligand mobility (Table IIJ). Of all the
`substrate-P-450cAM structures we have determined, the
`iron-sixth ligand distance is at its longest ( 1.96 A) in the
`presence of adamantane. Both the long bond length and
`greater ligand mobility could result from the relatively close
`approach of adamantane to the distal ligand: approximately
`0.4 A closer than norcamphor or camphane. This short sub(cid:173)
`strate-ligand distance may also promote protonation of the
`ligand, resulting in the observed high-spin hexacoordinate
`complex.
`Thiocamphor does not fit the shorter iron-L6 distance/lower
`spin pattern. It induces a spin state (65% high spin) inter(cid:173)
`mediate between those induced by norcamphor and camphane
`( 46%) and adamantane (I 00% ), but in the presence of thio(cid:173)
`camphor the iron-L6 distance is only 1.35 A, the shortest we
`have seen. However, the thiocamphor complex is not directly
`comparable to the other substrate complexes since at least two
`orientations are observable. It may be that a population in
`which thiocamphor bound exclusively in the ligand-allowing
`orientation (Figure 4b) would have a greater low-spin com(cid:173)
`ponent than the 65% quoted in Table III, which presumably
`arises from a population with mixed binding modes. In ad(cid:173)
`dition, in the presence of thiocamphor the distal ligand appears
`to be linking iron and the substrate sulfur atom and its ligand
`field strength may thus be different from that of a simple
`OW /H 20 ligand.
`A final point regarding the almost unbelievably short
`iron-L6 distance in the presence of thiocamphor should be
`made. Recall that only the minor crystallographically observed
`thiocamphor orientation is sterically compatible with the
`presence of a distal ligand. However, the electron density
`against which our model is refined represents contributions
`from all thiocamphor orientations in all protein molecules in
`the crystal lattice. This averaged electron density, which for
`the most part lacks a distal ligand contribution, probably
`obscures the