A Heuristic Model of the Oxygenation

of Hydrocarbons by Cytochrome P-450

AUG 22 1973
SO°r33

LL

Larry Hjelmeland
a

 
Since 1959, a considerable amount of research effort has been
directed towards the development of a non-enzymatic model of the
monooxygenases. These enzymes are also known as the mixed function
oxygenases and their general function is appropriately described by

the following equation.

RH + 02 + DHy —> ROH + HLO + D
RH is an organic substrate and DHz is a reduced electron donor which
serves as an essential cofactor for this reaction.

The monooxygenases are especially important to the field of
pharmacology since the major component of the drug metabolizing
enzymes of the human liver, cytochrome P-450, is a member of this
class. Thus although the non-enzymatic model systems were devised
to study the chemical activation of molecular oxygen, they also serve
the purpose of demonstrating how organic substrates might be meta-
bolized by the human liver. This paper will explore the possibility of
using a heuristic computer program to serve as a model of enzyme
P~450.

Before exploring the actual operation of the program, though,

it is essential to discuss some of the assumptions which are made
2.
in using this method and also to show that these assumptions are
justified at least in part by the information which is currently
available on enzyme P-450. The first section of this paper will
therefore be a short discussion.of the factors which affect chemical
reactivity and how they are represented in non-chemical systems.

The second section will relate these factors of reactivity to the
function of the enzyme. Section III will describe the heuristic
model and the fourth section will evaluate the performance of the

model on a simple class of compounds.
I. Factors Affecting Reactivity

The factors which affect enzymatic catalysis are thought to be
more limited than those which affect chemical reactivity in general.
Most simple discussions of reactivity are centered on steric quali-
fication of the substrate with respect to the active site with the
additional factors of electrostatic interaction and Van der Waals
forces also receiving some consideration. A more general perspective
of reactivity will be given here, though, since cytochrome P-450
does not follow our usual ideas of enzyme substrate interaction and
in addition it is of some value to be comprehensive in an attempt
to show which factors may be modeled in a heuristic program such as
the one proposed later in this paper. This general treatment roughly

follows the same section in (11).

Steric and Strain Effects
In the reaction of Phco + SbF, with a series of alkylbenzenes,
the effect of steric hinderance in the transition state may be seen

in the distribution of isomers.

B Gis
chy CH CH3 Oh CH; CH, _t — CH;
1o°3 B.L 0.6
of \2 7
8b-8 42-7 96-7

Lesomer distriloution is he Friedel -Cratts
ben2zoylation: ot alkylwenzenes
4.

The amount of orthosubstitution clearly decreases with increasing
size of the alkyl substituent. This is directly due to steric
interference. Steric interaction is usually sited as the major
factor involved in the lotk and key fit required for the high speci-
ficity of most enzymes.

The relief of angle-strain may also direct the course of a
reaction. When a bond is strained 9ut of its normal geometry, an
increase in the ground state energy of the molecule is also seen.
Relief of this strain energy may direct a reaction such as the

hydrolysis of lactones.

Oo toni

1g a
HO S33 path le C.0O%
o-——> O Q Hy ——7

pho Ce only ag

Lactone h Ard|ysis not co"on
loy O-M Kut cleavage

HO”

A general pathway for lactone hydrolysis is via the cleavage
of the O-Acyl bond after nucleophilic attack at the carbonyl carbon.
In the example shown above, though, the relief of angle strain in
the transition state directs the reaction towards o-alkyl cleavage,
as is conclusively demonstrated by the distribution of 180 in the

product.
Stereoelectronic Effects

 

The degree of overlap between two orbitals is of crucial impor-
tance in determining the energy of activation for a reaction which
involves either bond breaking or bond making between those orbitals.
This degree of overlap depends not only on the distance between the
two, but also on the angular disposition of the bonding lobes. An
important example is the deprotonation of triptycene and triphenyl

methane by strong bases.

K
Tripty cene Tripty ceny lide. anion

 

 

 

L_ _
Tripveny) metnone Triphewu| metnulide
AWto Yi
6.
In the former, no delocalization of charge to the phenyl groups
in the transition state is possible since the TY electron clouds
are orthogonal to the lone pair electrons of the carbanion. This
situation achieves minimum overlap between the lone pair and TI clouds.
The triphenylmethylide anion is considerably stabilized, though,
since maximum overlap is achieved due to the paralell disposition
of the bonding lobes. The deprotonation of triphenylmethane proceeds

about 1/2 million times faster than triptycene accordingly.

Electronic Effects

Substituents in organic molecules can markedly affect chemical
reactivity due to electronic effects. These electronic effects
are transmitted across the molecules by polar and conjugative mechanisms.
The well known change in acidity due to addition of electronegative
substituent is demonstrated by the pk of the following carboxylic

acids in water at 25° c.

Values of ek tor carboxylic acids

Acid pk

 

C\CH, COOH a.B3b6
ClCH COOH | 1 30,

Cl2C- COOH 0.65
7.

The polar effect consists of two seperate components. The first
is a polarization of bonds between the reaction center and the sub-
stituent which is normally termed an inductive effect. The second
is called the field effect and involves the direct electrostatic
interaction of the substituent and reaction center through space.
Experiments to clarify the relative strengths of these components
in the overall substituent effect have proved difficult to interpret,
but it is important to note that most authors agree that the field
effect is more often the most significant.

The other mechanism for the transmission of electronic effects
is by the conjugative effect or delocalization. This is achieved
by the overlap of a p-orbital at the reaction center with some
TI-electron system which may then overlap with p-electrons of the
substituents. The increased acidity of toluene with respect to
metnane is due to a conjugative effect on the benzyl anion.
This lowers the energy difference between the charged and neutral

species and hence increases the acidity. ,
Analysis of Reactivity in Model Systems

In order to analyze the factors that affect reactivity in model’
systems, it is first essential to discuss the space which is used
to build the model. For example -- if an analysis of a molecule
by the molecular orbital method is being done then the space we are
dealing with is geometric. There is a measure of distance between
points in this space and all arguments which pertain to the geometry
of a molecule are therefore applicable. Unfortunately, it is often
difficult or even impossible to construct models and perform an
analysis even with empirical molecular orbital methods.

A simpler space in which molecules may be represented is a
topological space. This space has no measure of distance or geometry,
but only of connectedgess. Given an atom in a molecule, for example,
it is only possible to ascertain what atoms it is connected or bonded
to. Without a measure of distance, it is not possible to say how
far away these atoms are. This greatly simplifies the mechanics of
the representation. Molecules become labelled graphs and are subject
to manipulation by fairly simple computer programs, as opposed to those ©
which deal with geometric representations (14). This simplicity is
bought at the price of a tremendous restriction of any analysis of
chemical reactivity. Steric and strain effects are clearly related
to a notion of geometry and distance as are obviously all stereoelectronic
effects. This means we may not incorporate these ideas into any
pvogram which deals with a topological representation. In a similar
fashion, the field component of the substituent effect is also
disallowed. Conjugative effects, on the other hand, are possible to

represent, but will not always give satisfactory results due to their
9.

obvious relation to stereoelectronic effects. Any usefulness of
conjugative mechanisms will depend on a fortuitous situation involving
free rotation about a bond connecting the p-orbital of the reaction
center involved and the Tl-electron system.

If for example, the TT-system is locked in a position orthogonal to
the reaction center, then no conjugative effect will be seen. This
is precisely the type of question which is indeterminable in a
topological space, since no notion of orthogonality exists without
geometry.

The only effect which may be completely represented with no notion
of geometry is the induetive component of the field effect. It is
extremely important to realize this in dealing with the topological

representation of molecules.

II. The Mechanism of P-450 Oxygenations

With this brief discussion of reactivity im mind, I would now
like to explore the mechanism of P-450 oxygenations and relate that
mechanism to the factors discussed so far.

Perhaps one of the most remarkable facts to emerge about this
enzyme is its surprising lack of molecular specificity. Where as
most enzymes are specific for only a handful of molecules, P-450
appears to be capable of binding and introducing an atom of oxygen
into almost any lipophilic substrate (5). The details of the exact
sequence of events at P-450 are now commonly accepted to correspond

to the following sequence (3).
FEN x "

e ~s
Ke tcox © o\ e
4.0 6) Fett ¢
3) O2
Fors
o= Fett

B+ 7 ©
Fe%ts| @ pete *

\ he
o, — Oo,
\. Su stvate binds +o Levi torm of P-¥s0

Z. Ferric sub strote complex reduced to ferrous form.
3. ferrous suloshule complex binds molecular Oxyqen

4.Oxy genared terrous complex veduced

5.0ne atow ot Oxyom inserted iN subsite , the
other ts released aS HO

6 Hycrroxulerted sulostate released, reaenevativ
fErnic. Poeun ot P-450. y 4

In the rate limiting step of this sequence, the second electron
is added to further reduce the P-450/substrate/05 complex. Model
systems such as 2-Mercaptobenzoic Acid/Fe (II) /0, have been studied
in order to learn more about the species of oxygen which is incor-
porated into the molecule (1). From these studies and similar
experiments with liver microsomes, the active form of oxygen has
been deduced. Since the typical reactions of this complex are carbon-
hydrogen bond insertion and eloctrophilic addition to double bounds,
it was reasoned thai{the active oxygen was a neutral oxygen atom or
oxene, a species isoelectic to carbpenes which show exactly the

same reaction products. Experimental evidence supports this conclusion (2).
11.

Oxygenation proceeds with the retention of a deuterium label, which
would be expected for the insertion reaction but not for a mechanism
involving displacement by a hydroxyl ion. In addition, this reaction
shows an isotope effect when hydrogen is replaced by deuterium in
those C-H bonds involved. This indicates that the rate limiting
step of the overall reaction involves C-H bond breaking and again supports
the insertion mechanisn.

It is now appropriate to discuss those factors which might
affect the reactivity of various substrates with respect to the
oxene electrophilic mechanism. Steric effects seem to play a small
role as mentioned earlier. That fact that such a variety of molecules
may be bound to P-450 does not exclude the possibility of steric
effects, in fact experimental evidence does support some steric
interaction, particularly for larger molecules (2). On the other
hand, conjugation seems to be very important. Benzylic and allylic
positions are commonly activated towards oxygenation by P-450 (2).
Inductive effects are also present. Carbon~hydsogen bonds which
are adjacent to heteroatoms are often oxygenated. This effect is
commonly seen in the dealkylation of ethers and thioethers and the
dealkylation of secondary and tertiary amines (2). In a series
of studies on the simple alkanes, it was also determined that the
reactivity of carbon-hydrogen bonds in these compounds followed

the order of bond weakness, that is: 3°92°31° (6,7,8,9).

III. The Heuristic Model .

brie¢

The bewbet survey of the factors affecting reactivity of substrates
12.
for cytochrome P-450 given in the previous section reveals an interesting
trend. All the important factors (ignoring steric interactions for
large molecules) are electronic in nature and are precisely those
which admit to being represented in a topological rather than geo-
metrical space. To test this possibility, I have constructed a
simple set of heuristics to be incorporated into a model which would
select the best sites for P-450 oxygenations within any candidate
molecule. The heuristic is simple: choose those C-H bonds, in order,
which are weakest as the best sites. First year organic chemistry

textbooks provide relative strengths for the following groups:

Strenotus ot C-H bounds in hudrocarlons

Benzglic c ao og 2° 2 4°
ANulic

a

The actual computer program to be‘used is the Dendral Predictor (14).
This is essentially a graph manipulation program based on the situation
action rule. The program accepts a candidate molecule and a list of
situation - action rules. The rules are applied to the candidate
in the following fashion:
1. Search the candidate for any subgraph which might match
the situation given in the first rule.
2. Ifsuch a match is found, perform the action indicated in
the second part of the rule. An example will illustrate this
process. Suppose the rule is composed of the following

situation and action.
Exawple ot Stuation Achon Rule, 2»:

S A
€ Nord | Replace H by on?

Then, given the following candidate molecule as input, the pro-

gram would produce the following output.

Input Out put

The tacit assumption of this model is that carbon-hydrogen bond
strength can be roughly estimated by looking at ‘the neighboring
groups according to the simple heuristic which I have provided. Again
the important distinction to be made is that this model is purely

topological.
IV. Evaluation
To test the efficacy of the model, the following results are

reviewed to demonstrate the similarities and differences of predicted

and observed results. The discussion is limited to hydrocarbon
14,

molecules.

Alkanes

2,2 dimethylpropane, 2 - methylbutane, isobutane, n - pentane,
n- butane, and n- heptane, the observed results are in accordance
with the programs prediction except that the primary alcohols of n -
pentane, n - butane, and isobutane were not sergeted (6,7,8,9).

This confirms the order of reactivity for 3° “vrs 1° carbon hydrogen

bonds.

Alicyclic Rings

Methylcyclohexane is metabolized in agreement with the program.
The tertiary bond is favored as a site of attack (9). Trans and cis
decalin are oxygenated at the 2 position, in disagreement with the

program, which predicts attack at the 9 and 10 positions.

ey

Benzylic Positions

Ethylbenzene, propylbenzene, butylbenzene and p - ¢xylene are

all attacked at the benzylic postion as would be expected (2).

Other Results

While experiments with hydrocarbons are difficult to find, many
hydrocarbon like molecules which contain one or two functional groups
have been experimented with. The data here suggests, for example, that
steric effects play an important part for larger molecules. Thus
most steroids are attacked only at axial groups (2B, 6B, 7% 118,

and 16e) (2). This appears to be a general rule for larger hydrocarbon =
15.
like alicyélic ring compounds.
The metabolism of 1 THC demonstrates the validity of this

method for large molecules (12, 13).

ke AA-THE Mayer Metalolites

+ « OW

se
om

ay

— > Ivdicates srtes preclicteol
— Indicates sites actually found

Of the twelve positions available for attack by P-450, 3 of the
4 sites selected by the model as being most important are found to be
the metabolites produced in greatest quantity. The next observed
netabolitesbre hydroxylations on the side chain.’ Again, the failure

of the program in this case is most likely due to some steric effect.
V. Conclusion

This paper is intended only to provide an introductory survey
of the heuristic technique for predicting the metabolites of P-450
reactions. Further research on the exact mechanism of monooxygenations
might allow the development of better and more sophisticated heuristics
and aconsajuent improvement of performance. The inherent limitations

of the representation will always cause problems such as the avoidance
16.
of steric interactions. Possibilities for futher study include the
expansion of the heuristics to include functional molecules and more

primitive methods of estimating relative bond strengths.
3.

4.

6.

7.

106

REFERENCES

Ullich V., Staudinger Hj.: Model Systems in Studies of the
Chemistry and the Enzymatic Activation of Oxygen. In: Handbuch
der Experimentelleu Pharmakologie, XXVIII/2, Springer Verlag.
Berlin-Heidleberg-New York (1971).

Daly, J.: Enzymatic Oxidation at Carbon. In: Handbuch der
Experimentelleu Pharmakologie, XXVIII/2, Springer Verlag, Berlin-
Heidleberg-New York (1971).

Estabrook, R.W., et. al.: Studies on the Molecular Function of
Cytochrome P-450 during Drug Metabolism. Drug Met. and Disp.
1, 98 (1972).

Ullrich, V.: Enzymatic Hydroxylations with Molecular Oxygen.
Angew. Chem. Internat. Edit. 11, 701 (1972).

Frommer U., Ullrich V.: Model Hydroxylation Reactions. Z.
Naturforsch. 26b, 322 (1971).

Frommer U., et. al.: Hydroxylation of Aliphatic Compounds by
Liver Microsomes. Hoppe-Seyler's Z. Physiol. Chem. 351, 903
(1970).

Frommer U., et. al.: Hydroxylation of Aliphatic Compounds
by Liver Microsomes. Hoppe-Seyler's Z. Physiol. Chem. 351,
913 (1970). .

Frommer U., et. al.: The Monoxygenation of N-Heptane by Rat
Liver Microsomes. Biochim. et Biophys. Acta 280, 487 (1972).

Frommer U., et. al.: Hydroxylation of Aliphatic Compounds
by Liver Microsomes. Archiv Fir Pharm. 266, 328 (1970).

Ullrich, V.: Oxygen Activation by the Iron (II) -2- Mercaptobenzoic
Acid Complex. A Model for Microsomal Mixed Function Oxygenases. Z.
Naturforsch. 24b, 699 (1969).

 
ll.

12.

13.

14.

Alder, R.W., et.al. Mechanism in Organic Chemistry. Wiley-
Interscience, London-New York-Sydney-Toronto, 1971.

Augrell, S., et.al.: Metabolic Fate of Tetrahydrocannabinol.

In: Cannabis and its Derivatives.
London 1972.

Oxford University Press,

Burnxtein, S., et.al.: The Urinary Metabolites of D1 THC.

In: Cannabis and its Derivatives.
London 1972.

Mass Spectrum Prediction In Dendral.
tation.)

Oxford University Press,

(Dendral Group documen-