Reactivity Indices in Polycyclic Aromatic

0 downloads 0 Views 2MB Size Report
The results supportthe belief that specific. transformationsInvolvingdlhydrodiol,“bay-region― epox ide, and carbonium ion intermediates are responsible for.

Relationships between Carcinogenicity and Theoretical Reactivity Indices in Polycyclic Aromatic Hydrocarbons Iden A. Smith, Gregory D. Berger, Paul G. Seybold, et al. Cancer Res 1978;38:2968-2977.

Updated version

E-mail alerts Reprints and Subscriptions Permissions

Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/38/9/2968

Sign up to receive free email-alerts related to this article or journal. To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at [email protected] To request permission to re-use all or part of this article, contact the AACR Publications Department at [email protected]

Downloaded from cancerres.aacrjournals.org on June 2, 2013. © 1978 American Association for Cancer Research.

[CANCER RESEARCH 38, 2968-2977, September 1978] 0008-5472/78/0038-0000$02.00

Relationships between Carcinogenicity and Theoretical Reactivity Indices in Polycyclic Aromatic Hydrocarbons Iden A. Smith,1 Grsgory D. Berger,2 Paul G. Seybold,3 and M. P. Serve Departments of Chemlstryfl. A. S., G. D. B., P. G. S., M. P. S.Jand Biological Chemistry (P. G. S., M. P. S.J, Wright State University, Dayton, Ohio 45435

ABSTRACT

Theoreticalreactivftyindiceshave been used to exam inc the metabolic reactions presumed, on the basis of recent biochemicalevidence, to be responsiblefor the transformation of polycyclic aromatic hydrocarbon pre carcinogensto ultimate carcinogens.Of a large number of indices examined, several show strong correlations with carcinogenic

activfty in a set of 25 representative

compounds.The results supportthe belief that specific transformationsInvolvingdlhydrodiol,“bay-region― epox ide, and carbonium ion intermediates

are responsible for

the carcinogenicactivftyof these compounds.Additional implicationsof the results are discussed, includingthe suggestionthat this type of analysismightprovidea rapid and simplemeansfor prescreeningcompoundsfor poten tial carcinogens. INTRODUCTION

Models suggesting a link betweenthe carcinogenic activ ities of aromatic compounds and electronic properties dale back at least 40 years to the pioneering work of Schmidt (29). in the past quarter century, there have been proposed numerous theories attempting to correlate electronic mdi ces from molecular orbital theory with “carcinogenic achy ity.― A successful

theoretical

correlation

of this

nature

would be extremely important for several reasons: (a) it could provide important insight into the chemical mecha nism(s) by which these compounds cause cancer; (b) when perfected it could provide a rapid and simple screening procedure to supplement the lime-consuming and expen sive animal and bacterial experiments now required for testing of potential carcinogens; and (C) the information thus provided might eventually be used to guide the design of effective antitumor agents. Most theoretical models proposed thus far have focused attention on properties of the parent aromatic hydrocar bons, yielding in many cases quite interesting and sugges live results (6, 10, 12, 21-23, 26, 27). However, during the past severalyears there has been a considerable expansion in our experimental knowledge of the metabolism of PAH's,4and the results strongly suggest that the carcino genic process for these compounds involves a series of metabolic transformations that convert the relatively inert I Address

after

September

15,

1978:

Department

of Chemistry,

University

of California at San Diego, La Jolla, Calif. 92093. 2 Present

address:

Department

of

Chemistry,

Yale

University,

New

Haven,

Conn. 06520. 3 To whom requests 4 The abbreviations

for reprints should be addressed. used are: PAH, polycyclic aromatic

benzo(a)pyrene; BA, benz(a)anthracene. Received February 24, 1978; accepted June 16, 1978.

2968

hydrocarbon;

BP,

parent hydrocarbons into highly mutagenic and carcino genic metabolites (4, 5, 7, 8, 13-20, 24, 25, 32, 34—40). Thus, it is possible that the observed differences in carci nogenic activities among these compounds not only reflect characteristics of the parent hydrocarbons but also depend strongly or even crucially on properties of later metabolites and the propensity of these metabolites to undergo specific reactions. Because the nature of selectivity is central to understanding the mechanism of carcinogenesis in this important class of compounds, we have examined the metabolic reactions in question, using a large sample of representativecompounds and a variety of reactivity indices taken from molecular orbital theory. We shall show that strong correlations do exist between the properties of several putative metabolic intermediates and carcinogenic activities, and we shall distinguish those indices that appear most accurately to reflect the carcinogenic process. BiochemicalEvidence.The metabolictransformationsof PAH's presumedto be responsible for carcinogenic activity can be illustrated by the example of the “classic― carcino gen BP (Compound I), which has been the subject of extensive experimental examination. The reactions of BP can be regarded as typical for this class of compounds. As a result of a variety of studies, a primary path by which BP is metabolically activated and transformed in vivo from precarcinogen to ultimate carcinogen is believed to consist of the stages shown in Chart 1. These transformations can be regarded as consequences of metabolic attempts to render nonpolar compounds soluble so that they can be excreted. In Step a of Chart 1, BP is activated via a cellular monooxygenaseto form its 7,8-epoxide (ha). [The 4,5-bond in BP forms a classical “K-region― (27), and epoxide for mahionat this latter bond apparently forms a diversionary pathway from the principal route to carcinogenesis. Recent evidence (38) indicates that cellular epoxide hydrase may act in viva to detoxify reactants along this pathway.] In Step b, epoxide hydrase assists conversion of the 7,8-epoxide to

the 7,8-dihydrodiol (lIb). Experimentally,the 7,8-dihydrodiol is found to be more effectively converted to mutagenic and carcinogenic metaboliles than is BP itself or dihydrodiols formed at other positions (19, 32, 38). Saturation of the 7,8bond activates the 9,10-bond, leading to Step c in which Compound lib is transformed to the 7,8-dihydrodiol-9,10(bay-region5)epoxide (Ill) (34—40). Strong support for the importance of this step has come from the experimental observation that the product BP diol-epoxides have excep tionally high mutagenic activity for both bacterial and mam malian cells (16-18). In Step d the diol-epoxide Ill converts S A

“bay

region―

Is

a

concave

exterior

region

of

a

PAH

bordered

by

the

3

phenyl rings, at least one of which is a terminal ring. The prototypal example Is the region between carbon atoms 4 and 5 of phenanthrene.

CANCERRESEARCH VOL. 38

Downloaded from cancerres.aacrjournals.org on June 2, 2013. © 1978 American Association for Cancer Research.

@

@.R

Theoretical Reactivity Indices and Carcinogenicity Bay Region

2

:[email protected]@@(a) I

I Ia

7/

examined several potentially useful indices at each stage to find which of these, if any, can be used in the present model. Furthermore, although schemes more complete than the simple Hückel molecular orbital theory are availa ble, we see no reason to resort to these more complicated and time-consuming procedures at this early stage. For ease of analysis we shall identify certain key regions in the compounds under consideration. The K- and L-regions are as customarily defined (27), but it is necessaryto define two new regions, the A- and B-regions. The definitions of the regions are as follows: K-region, the electron-rich region typified by the 9,10-bond in phenanthrene (27)and contain ing the highest molecular ii- bond order; L-region, a region consisting of 2 pare-carbon atoms, as the meso-9,10-car bon atoms of anthracene [Thesecarbon atoms display the highest free-valence indices (27).]; A-region, the presump live initial epoxidation site, on the terminal ring of the bay region, in the metabolic pathway leading to carcinogenesis (This corresponds in most cases to what has sometimes been called the M-region.); B-region, the site of final epox idation on the terminal ring of the bay region on a bond adjacent to the bay region. These regions are illustrated in Chart 2 for some representativecompounds. To obtain a broadly based analysis, we have examined the properties of 25 representative PAH's and their pre sumptive metabolites, a set of compounds previously ex amined by Jerina et a!. (17, 18) in their study of carbonium ions. The compounds are illustrated in Chart 3. Relative values for carcinogenic activity have been estimated by using principally the report of Arcos and Argus (2) and are summarized in Table 1.

@ [email protected] _ @ HOJ'I1,III―@'@@@ (c @.f:[email protected][email protected]:I OH

@@-‘

HO

®

@/

OH

00

[email protected]@°

Chart 1. Presumedmetabolicstepsleadingto carcinogenesisby BP. For simplicity only 1 stereoisomeric form is illustrated.

@

@ @

@

@

@

spontaneously to the triol carbonium ion IV. It has been suggested (14) that carbonium ions such as IV act as ulli mate carcinogens via electrophilic attack on critical cellular nucleophiles, e.g. DNA. Evidence for this has come with the isolation and identification of specific adducts to gua nine (15, 20, 24, 25, 35). Jerina and Lehr (16), Jerina et a!. (17, 18), and Woodet a!. (38) have reviewed in much greater detail the experimental evidence supporting the importance of the transformation shown in Chart 1. This evidence has led them to propose the “bay-region theory― of carcinogenic activity for these compounds, which holds that diol-epoxides such as Ill, formed on saturated angular benzo rings, can be expected to display especially high carcinogenic activity (16-18). Theoretical calculations indicate that bay-region diol-epox ides should form carbonium ions more easily than other diol-epoxides (16-18). TheoreticalModel. In principleanyof the stepsshownin Chart 1 might be decisive in determining carcinogenicity in BP and other PAHcompounds. So too could be the relative efficiencies of competitive, noncarcinogenic metabolic pathways or factors (e.g. , size, solubilily) largely outside the present analysis. Nonetheless, indices from molecular orbital theory have been exceedingly instructive in our understanding of ordinary chemical reactions (33) and, in So far as the reactions under consideration are subject to the same constraints, such indices should be helpful in elucidating those reactivity differences that might influence carcinogenesis. Of course there are possible hazards in applying this kind of approach to complicated systems(30), but these hazards are not necessarily as great as has sometimes been feared, for reasons that we shall discuss later.

A key element in this approach is a proper choice of theoretical indices, since usually several alternatives are available for analyzing each reaction. We have therefore SEPTEMBER

1978

[email protected]@[email protected]

ap

PA

2.062

L932

2.010

i:@[email protected]@

(@@4 @A::[email protected]

[email protected]''.―'@[email protected]

2 37

2.314

HO'

848

@@.‘‘[email protected]

[email protected]:99

P994

2.333

@OH

2 00

2.27>

-OH

[email protected]

OH 03810

OH

OH OH

03637

0 4500 [email protected]

OH

10H [email protected]

Chart 2. Designated regions and calculated indices for BP, BA, and phenanthrene (PA).

2969

Downloaded from cancerres.aacrjournals.org on June 2, 2013. © 1978 American Association for Cancer Research.

I. A. Smith et a!. @

RESULTS

2.©@[email protected]

Examinationof PossibleIndices We now examine the individual stagesof metabolic Irans

6.©[email protected] 7.IgTgTgTg:1 formation (Chart 1) and associated theoretical indices.

5.©@©@O

CompoundI, the Parent Compoundor Precarcinogen. Prior theories of carcinogenesis have dealt largely with [email protected]@J electronic structures of the precarcinogenic parent com pounds. Because many of the parameters to be discussed [email protected] i'UTgTO@ s.TgTg:1 8W later bear some relation to molecular size, it seemeddesir able first to determine whether some simple feature related [email protected] this might not itself correlate with carcinogenic activity. [email protected] This seemed especially relevant since Arcos and Arcos (1) 1010I and Arcos and Argus (2) have shown that most PAHcarcin [email protected]'@ [email protected] [email protected]°@ l3.TQ] ogens (but also many noncarcinogens) have surface areas (“incumbrance areas―) within a certain range. One simple measure of molecular size (or mass) is the number of carbon atoms in the compound, and in Table 2 this number 16, [email protected] I7.5JQJOJ [email protected]@ is compared with carcinogenic activity. Apparently, al though most strong carcinogens are relatively large and the smallest compounds are not carcinogens, the correlation is far from complete. BP is a notable exception. Another gross molecular feature and one sometimes held 21.IOIOJ 22. JOIOJ 20.([email protected]@[email protected] [email protected] responsible for the action of certain drugs and for carcino genic activity (9, 11) is solubility. Experimentally, this is conveniently described in terms of the partition coefficient

[email protected]

@QJW

igxgxgx

[email protected]

P between a polar and a nonpolar phase. As demonstrated

IOTOT

by Rogers and Cammarata (28), this feature can be treated

[email protected]

within the framework Chart3. Compoundsexaminedin this study.

Numbering

of molecular

orbital theory in terms of

sums of atomic ir-eleclron densities Q,. (polar phase) and superdelocalizabililies Sr (nonpolar phase) for the mole

and characteristics

Table 1 of theompoundsexamined(see

3)Carcinogenicity Chart

K-L-Bay

and Ar-

indexArcosCorn

al.poundCompound

re

17)1Naphthalene———0nameregion?region?giongus

(2)

Jerina et (16,

—2Anthracene—+—0 —3Tetracene—+—0 —4Pentacene—+—0 —5Hexacene—+—?6BA+++5 +7Benzo(a)tetracene+++—8Phenanthrene+—+0 —9Benzo(c)phenanthrene+—+4 +10Chrysene+—+3 +11Benzo(b)chrysene+++—12Picene+—+0 —13Triphenylene——+0 —14Benzo(g)chrysene+—+17 ++1 +16Dibenz(a,j)anthracene+++4 5Dibenz(a,c)anthracene—++3 +17Dibenz(a,h)anthracene+++26

+18Naphtho(2,3-b)pyrene++a27

+

++19BP+—+7320

3322Dibenzo(a,i)pyrene+—+74 21Benzo(e)pyrene Dibenzo(a,l)pyrene+ ++++23Dibenzo(a,e)pyrene+—+50 +++24Dibenzo(a,h)pyrene+—+70 ++++25Tribenzo(a,e,i)pyrene——+16

+

—+

+—

+2

++ a This

compound

b Jerinaeta/.

2970

does

not

strictly

possess

(16, 17) have assigned

a bay

region

but

does

contain

a “pseudo'

‘ bay

region.

this as ++++.

CANCERRESEARCH VOL. 38 Downloaded from cancerres.aacrjournals.org on June 2, 2013. © 1978 American Association for Cancer Research.

Theoretical Reactivity Indices and Carcinogenicity Table 2

Comparison of carcinogenic potency and the number of carbon atoms

@ Compound

No. of carbon at oms

25

28

++

24

++++

Carcinogenicityindex

26

5

24 22 21

23 18 17 14 16 15 7 4 12 11 19 20

24

++++

24 24

++

+++

24

+ +

22 22

the K-, L-, and A-regions are shown in Table 4. The K- and L-region values agree with those given previously by [email protected] ster and Memory (21, 22), who suggest that a compound is expected to be a carcinogen if ‘K2.05 and ‘L2.30. Note that, for compounds with more than a single K-, L-, or Aregion, only the highest value is listed. II can be seen that in Table3 Comparisonof calculated partition coefficients and carcinogenic potencies

index531.40214.876?+2525.47912.321++—424.75812.010——1823.71611.561++—2423.41711.432++++—2222.8 In p(@Carcinogenicity

:::i

22

+CompoundSr-

22 22 22

22

+++++2122.04510.840+++2321.71910.699++++721.67210.679—+1120.50410.175—+1719.9819.949++—1619.9689.94 ++ 1.175+

22 20 20 18 18 18 18 18 14 14 10

6 10 9 13 3 8

2

3

cules in question. The atomic superdelocalizability at posi

6 10 9

lion r is defined as (10, 23)

13

[email protected]

2

where the sum is over occupied molecular orbilals, the [email protected] are carbon atom molecular orbital coefficients for molecu lar orbital j, and the molecular orbital energies are €@ =a+ mj3. Since all of the compounds examined here are alter

nanl aromatic hydrocarbons, their ir-electron charge densi ties are uniformly zero at all carbon positions. For this case Rogers and Cammarata(28) obtained the equation In P= 0.431

20

8 118.904 a In [email protected] = 0.4314

18.786

9.433

—

17.789

9.004

+

16.663 16.050 16.019 15.540

8.518 8.255 8.240

+

13.486 12.353 8.8769.485 @:Sr

+

— —

—

—

1.3297.

Table 4 Parent compound bond superdelocalizability indices and carcinogenic

potencies Carcino

genicity in

Corn

@: [email protected] 1.3297 [email protected]/IKdex22

182.206 2.1962.7241

++that

+

8.033 7.147 6.659 5.159++++

+

.943 2.0220.8808 0.9208+

+ + +

holds reasonably well for a large number of com + + + + 1 .945 0.9089 2.140 19 pounds. A comparison between carcinogenic activity and 0.8719 1.865 7 2.139 In P is shown in Table 3. No special relationship is evident ++ 0.9229 .951 21 2.114 from the results.24 1.8710.9131 0.8914++++ 2.5121.986 112.1752.0992.863 .8480.8889+the the many molecular orbital theories of carcinogenesis,62.0792.4381 + .848 17 K- and L-region approach of the Pullmans' (26, 27) has + 0.9142 1.875 2.2941 2.051 16 receivedthe widest attention. In this approach carcinogenic +++ 0.9385 1.908 23 2.033 activity is associated with a reactive K-region and a rela +lively .8130.8967 0.8940+ 202.0612.0282.292 unreactive L-region, both measured by indices de .8530.9228—rived 122.0081 + .904 from combinations of localization energies in the two + 0.9289 1.856 10 1.998 regions. The results are found in many although not all — 1.837 0.9213 1.994 8 cases to yield a suggestive relationship to carcinogenic + 0.9413 1.859 .975 9 activity (6, 26, 27). The Pullman indices are somewhat25 142.0041.9721 +complicated + 1 .8700.9501 0.9483+ .810+22) to calculate, and Mainsler and Memory (21,152.3061 have shown that a simpler index I, the sum of the two ‘@ 2.145 3.590 atomic superdelocalizabilities (10, 23) involved in a partic 2.083 3.010 3 ular regional bond, can be used to give comparable results.5 2.6262.192 1 .995— 24.010 2-center index would appear to be well suited for131 .803—estimating .867 reactivity with regard to epoxidation. Resultsfor11 —

@

—Of

@

@

—This

SEPTEMBER

1978

2971

Downloaded from cancerres.aacrjournals.org on June 2, 2013. © 1978 American Association for Cancer Research.

I. A. Smith et a!.

@

all cases > ‘A As a rule the most potent carcinogens lack L-regions and have ‘A values >1 .90. An exception is di benz(a,h)anlhracene (Compound 17). ir-Electron bond orders Pr, might also be a fair measure of 2-center attack and are shown in Table 5 for the K-, L-, and A-regions. Again, no strong correlation is evident. Most carcinogens have values for the energies of the highest occupied molecular orbital (in fJ units) between 0.300 and 0.500 (Table 6). CompoundlIb, the A-RegionDihydrodiol.Fromtheview point of elementary ir-electron theory, Compounds ha and lIb are equivalent since both are saturated at bond A. The efficacy of Step c depends on activation of the B-region bond, for which there are several possible measures. As before, a superdelocalizabilily index should be appropriate. We call this index lB', the prime indicating the dihydrodiol form . Elsewhere we have already demonstrated that a strong correlation exists between this index for the Bregion and carcinogenic activity in these compounds (3). Values of ‘B and ‘K' for the A-region dihydrodiol forms are summarized in Table 7. In all cases, formation of the A-re gion dihydrodiol significantly activates the B-region bond, as shown by the high value of index ‘BIn many cases,e.g., BP (Chart 2), the K-region bond is deactivated, accenting further the tendency toward reaction at the B-region. To see whether some other index might describe the bond activation of this stage as well as does ‘B― we examined the B-region ir-eleclron bond orders, as shown in Table 8. It can be seen that this index also correlates rather well with carcinogenic activity. CompoundIV, the CarboniumIon. Accordingto the pres ent picture, Irihydrotriol carbonium ions act as the final

Table 6

Comparisonof parent compound highest occupied molecular orbital energies (/3 units) and carcinogenic

potencies

occupied

molecularorbital

CompoundHighest activity130.6840—I0.6180—80.6052—90.5676+140.5319++100.5201+120.5019â energyCarcinogenic

Table 7 Bond superdelocalizability indices IB' and [email protected]'for A-region dihydrodiol forms

Compound‘B‘KCarcinogenicity index222.4072.180++++242.3902.174++++72.3752.196—252.3612.025++192.3582.09

Table5 Comparisonof parent compound mobile bond orders and carcinogenic potencies CompoundPAPlCCarcinogenicity index250.68310.7314++230.68540.7537+

+150.6864+200.68880.7754+130.6904—210.69180.7833++140.69310.7457++70.69810.7851—60.70050.7833+160.70270.7797+170.70280.7780++80.70680.7747—120 +

order.

metabolic species that attack critical cellular nucleophiles (presumablyDNA).Jerina et a!. (16—1 8) havepointed out the importance of bay-regioncarbonium ions and examined the ease of carbonium ion formation, using the change in de localization energy @[email protected]/f3, as obtained from a perturba lion calculation. The success of this index in correlating

2972

CANCER RESEARCH VOL. 38

a If there

given

is more

represents

than

1 possible

the K-region

K-region

or A-region

or A-region,

the

with the lowest

value

bond

Downloaded from cancerres.aacrjournals.org on June 2, 2013. © 1978 American Association for Cancer Research.

Theoretical Reactivity Indices and Carcinogenicity Table 8

Table9

Comparisonof ir-electron bond orders for the A-region dihydrodiols with carcinogenic potencies

Carbonium ion charge density Q,@ at the benzylic carbon position

and [email protected],@/f3 comparedwIth carcinogenic potencies Carcinogenic

Compound(from

Compound Chart @Ed,,@/f3b ityindex index220.89140.7450++++240.89230.7483++++190.89340.7675++++250.89420.7453++70.89510.7265—230.89550.7670+ 3)PB'@'PKCarcinogenicity ++-f:+ 22 0.3215 0.870 ,•‘@ (1

7

24 25

21

0.3333

0.846

0.3334 0.3494

0.845 0.818 0.808

0.3600

0.3637

19

++++ ++ ++

0.794

++++

+210.89570.7833++60.89610.7235+170.89680.7170++140.89700.7498++150.8974+160.89750.7185+200.89850.7763+80.89930.7123—130.8993—120.89970.6949—110.90030.6830—100.90040.6898+90.90300.6960+18 + 0.3712 0.782 5

23

0.3750 0.3810

0.778

6 17

0.3984

0.738

15

0.4083

16

0.4084

0.4103

14

20 4 18 13

12

more

than

1

B-region

(dihydrodiol)

exists,

the

value

listed is the lowest bond order.

+

++

0.722 0.722 0.728

+

+ ++

0.4138 0.4167 0.4286 0.4464

0.714 0.710

+

0.690

++

0.4476

0.662

0.664

8

0.4500

0.658

11 10

0.4580 0.4630

0.647 0.640

+

0.600

+

0.628

3

0.4706

9 2

0.5294

0.544

0.5714

0.488

a When a When

+++

0.766

0.4902

more

than

1 bay-region

carbonium

ion

is possible,

the

valuegivenrepresentsthecarboniumionwiththelowestQb. b FromJerinaet al. (16,17).

b By using the dihydrodiol with the lowest B-region bond order, the value given is the lowest K-region bond order.

data for the present set of compounds is impressive. A shortcoming is that this index cannot be used in certain cases (e.g., it does not distinguish methyl derivatives) (17), which has led us to examine alternative indices. One such index is the net ir-electron charge (i.e., Qb= I —

q5,

where

[email protected]

is

the

ir-electron

density)

at

the

Table 10 Comparison of carbonium ion free valence and atomic superdelocalizability indices with carcinogenic potencies

[email protected]'Carcinogenic index221.00850.844++++241.01200.819++++191.01790.713++++71.02010.913—251.020 ity

benzylic

carbon position b of the trihydrotriol carbonium ion. In Table 9 this index, which we call Qb, is compared with [email protected]/f3and carcinogenic

activity.

The index Qb is equally

successful with @[email protected]@/j3 in its correlation to carcinogenic ity, the two giving almost identical results. An advantageof the index Qb @5 that it is more generally applicable; e.g. , it can be used for distinctions among methyl derivatives. Two other indices, the free-valence index Fb and the atomic superdelocalizability Sbat the bay-region carbonium ion position also showed correlations with carcinogenic potency. (The free-valence index is Fr

1.732 —>;:Prs

where s refers to atoms bonded to the r atom.) Results for these indices appear in Table 10.

@ @

Composite Energy Indices Transformation energies can sometimes be instructive in kinetic analyses(witness the Jerina-Lehr energy index). We examined the following energies: the ir-energy loss in going from the parent compound to the A-region epoxide or dihydrodiol (Step a); the ir-energy loss in forming the dihydrodiol-epoxide from the A-region dihydrodiol (Step SEPTEMBER

If more than 1 bay-region carbonium ion is possible, the value given represents the carbonium ion with the highest value for the index. b For these compounds the bay-region carbonlum ion with the

[email protected]

1978

Downloaded from cancerres.aacrjournals.org on June 2, 2013. © 1978 American Association for Cancer Research.

2973

I. A. Smith et a!. C); and

@.EJ',the u--energy change in forming

the Irihydro

triol carbonium ion from the dihydrodiol-epoxide (Step d). The last of these is formally

@

identical

Table 12

Comparisonof the energy indices @EJ2 and [email protected]@ with carcinogenic potencies (see “Results―)

to the Jerina-Lehr

index @[email protected]//3, but it is calculated directly and not by using a perturbation technique. Table 11 shows that @[email protected] does not yield a sensible correlation. However, both and

Corn (fi)@(2b index220.8911—2.4577++++240.8824—2.4565+ [email protected]@(3)a (fi)Carcinogenic ity

[email protected]@ show a strong correlation with carcinogenic activity (Table12).

+190.8675—2.4550++++70.8666—2.4516—250.8614—2.4536++210.8563—2.4507+ + +

Many of these indices show strong correlations among themselves, as can sometimes be expected on theoretical grounds. Some of these relationships are illustrated in Charts 4 to 11. As seen in Chart 4, there is some tendency for ‘A and ‘K to increase together, but this does not always

+60.8480—2.4504+230.8457—2.451

9+

÷1

+

+140.8354—2.4494++150.8333—2.4484+160.8332—2.4486+200.8263—2.4466+80.8121—2.4462—130.811 70.8395—2.4495+

hold true. The Jerina-Lehr index [email protected]/f3correlates strongly with ‘B and almost exactly with the charge index Qb.

DISCUSSION The results reported here would appear to lend strong theoretical support to the idea that reactions such as those in Chart 1 are related to the carcinogenic activity of these compounds, i.e. , to the bay-region hypothesis (16-18). Jerina and Lehr (16) and Jerina et a!. (17, 18) in their study of carbonium ion delocalization energies were the first to demonstrate a convincing correlation between carcino genicily and properties of a melabolite. We have shown elsewherethat the index ‘B for the dihydrodiol intermediate also correlates strongly with carcinogenic potency (3). In the present much more extensive examination, ent that several reactivity indices representing

it is appar later meta

a If

more

than

1

bay-region

carbonium

ion

can

be

formed,

this

valuerepresentsthe carboniurnion with the largest @E,F13. 0 The

most

negative

AE,[email protected]

is

given

if

more

than

1

possible

B-

region is present.

bolic stages correlate strongly with carcinogenic activity. Becausethese indices can be sensibly associated with the

0

2.00 0

Table 11 @

@

0

C.mparison of theenergy index @[email protected]@1 with [email protected]@(1)(f3)aCarcinogenicity

(see “Results―)carcinogenic

1A

in

0

I.60

dex25

@

0

@

00 00

23

—3.3934

15

—3.3919

+

—3.3836

+

20

‘

13 21 14 7

—3.3799 —3.3681 —3.3671 —3.3427

6 -

16

—3.3292 —3.3292

8 12

—3.3185 —3.3052

10

—3.2992

11

++ + — —

—

19

—3.2611

++++

1

—3.2588

—

—3.2486

++++

—3.2072

++

2

—3.2023

—

3

—3.1859

—

4 5—3.4000

—3.1806 —3.1786++

—

more

than

1

A-region,

? the

region with the most negative [email protected],T―.

2974

+ +

‘

—3.2928

is

2.20

bons.Thecorrelationccefficient is r = 0.734.

—

++++

there

2.10

+

—3.2740

18

a If

++

22 24

‘ 2.00

‘K Chart 4. Correlation between indices ‘A and ‘K for the parent hydrocar

++

—3.2990

9

0

1.80

—

—3.3368

17

0

+++

value

listed

is

for

the

A-

reactions in Chart 1, the most reasonable interpretation is that such reactions are indeed involved in the carcinogenic process. It is interesting that properties of the later metabolic inter mediates give stronger correlations to carcinogenicily than do those of the parent compounds. In his valence bond study, Herndon (12) concluded tentatively that the parent aromatic hydrocarbons themselvesact as ultimate carcino gens. Most previous theories, similarly predating the cur rent metabolic evidence, have also operated at least im plicihlyfrom this assumption or at least the assumption that the properties of the parent compounds are determining for the carcinogenic phenomenon. The results obtained in this study lead us to conclude that properties of the later stages, although of course indirectly preordained by the natures of CANCER RESEARCH VOL. 38

Downloaded from cancerres.aacrjournals.org on June 2, 2013. © 1978 American Association for Cancer Research.

Theoretical Reactivity Indices and Carcinogenicity [email protected],

@

,

,

,

lower than the corresponding ‘K values. We interpret this to mean merelythat the reactions (Stepa) of Chart 1 represent 2.20 0 a minor chemical pathway compared to reaction at the Kregion, consistent with the experimental evidence (2). [This 2.15 does not necessarilyimply the samefor the in vivo situation in which enzymatic specificities can influence the relative ‘K 0 yields (13).] It is ironic that historically the well-known (both 2.10 0 theoretically and experimentally) high reactivities of K0 regions mayactually haveinhibited progress in understand 2.05 0 ing chemical carcinogenesis by PAH's,becauseit was natu 0 ral to focus attention on the predominant reactions rather 2.00 0 0 than on a minor pathway. An exception to the model also provides insight. Sr ‘ k Benzo(a)telracene(Compound 7) has a high ‘B' value and a 2.25 2.30 2.35 2.40 low Qbvalue, marking it as a potential carcinogen, but it is not found to be carcinogenic. An explanation may be that Chart5. Correlation between indices [email protected](parent hydrocarbon) and [email protected]' this compound has 2 active L-regions; the idea that the L (dihydrodiol). r = 0.587. region is a deactivating region appears to retain its useful ness in the present model, consistent with earlier theories (27). 0.55 00 Naphlho(2,3-b)pyrene(Compound 18) is apparently more active than predicted by the model, and we have no good 0.50 explanation for this case. However, the very qualitative 0 0 nature of the experimental data on carcinogenicity must be 0.75 recognized and mayprovide reasonnot to be overly worried 0 0

0

0

0

@

0

@

‘0

0

0

0

if complete

0

0

AEdeloc/@ 0.65

0 0

0.60

agreement

with the experimental

results is not

obtained at this stage. Development of a coherent, accu rate, and general numerical measure of carcinogenic po tency (if this ideal is even possible) would greatly facilitate

0.70

0 0,[email protected]@)[email protected]@00.89500

0.55

0

0.50 0

2.20

@

2.30

2.40

Chart 6. Correlation between the energy index

of Jerina et al.

and Lehr(16,17)andthe indexIs'. r = 0.981.

[email protected]‘‘

2.20

2.25

2.30

0

2.35

2.40

‘@

0.55

Chart 8. Correlation between indices [email protected]'and [email protected]'for the dihydrodiol. r = 0.909.

0

0.50

0 0'

0

‘

0.55

0b

00

0

0(@

0.40

0

0

0.50 0

9:,

0

00

@

,

00

0.45

0.35

0 0

2.20

2.25

2.30

2.35

00

0b 0A5

0

0

0

2.40

‘@ Chart 7. Correlation between indices [email protected]'(dihydrodiol) and [email protected](carbonium ion). r = 0.976.

@

@

the parent hydrocarbons, may in fact be more directly related to carcinogenic potency than are properties of the parent compounds. For the parent hydrocarbons the ‘A values are in all cases SEPTEMBER 1978

0.50

0.60

0.70

0.60

0.90

AEdSIOC

Chart9. [email protected]

for the carbonium

ion.r = 0.999.

2975

Downloaded from cancerres.aacrjournals.org on June 2, 2013. © 1978 American Association for Cancer Research.

I. A. Smith et al. 3

,

,0 0

the more fundamental sibly

alerting

biological

investigators

carcinogen

to

tests and pos

compounds

that

might

be

expected to be especially hazardous. @

000

If

ci:,

0.85 Op 0 (3) [email protected]

the

present

imenlal natural

0

as

metabolic

evidence to

inquire

successful

picture

appears as

as

it

to

why

was.

is

correct

(and

to be increasing the

That

original

is,

if

the

K-region

the

exper

that it is), it is theory

K-region

is

not

was on

the

main path to carcinogenesis and is in fact a diversionary 0

0.50

path,

it

is

hard

to

understand

why

reactivity

of

the

K-region

bond should correlate with carcinogenic potency. Jerina et a!. (18) have already pointed out that the angular ring

arrangement necessaryfor a K-region is also a prerequisite fora bay regionand thatcompounds thathave one almost

. always have theother. Also, asshown inChart 4,there isa

0.75 0

slight tendency for ‘A and ‘K to increasetogether. Both tend

@

I

2.20

‘

to

increase

with

molecular

size,

which

itself

correlates

2.25 2.30

2.35

2.40

somewhat

Finally, there is the interesting 0

enl

0 0

0.85 0

0

° °

analysis (30)

inherent

in

the

transport

are

as

has

complicated

dangers

060

works

Scribner

to

Still,

we

remain

puz

well

as

correctly

it pointed

application

of

biological

systems.

the

unknown

properties,

question

does.

In

a

out

some

molecular

of

orbital

catalytic

of

article the

dangers

techniques

Foremost

influences

and

of why the pres provocative

among solubilily

enzymes

on

these factors,

the

rates

of

the transformations under consideration. We admit to some degree of surprise ourselves at the extent of agreement

(3) 00

0.80

activity.

lionsthatwe have notyetfully perceived.

,

0

carcinogenic

zled and suspect that there are more fundamental connec

Chart 10. Correlationbetweenindices [email protected],,@'@and la'. r = 0.975.

@

with

0

°

0

achieved

0

carcinogenic

here

between activity.

the Taken

theoretical at

reactivity face

value

indices the

results

and imply

that solubility and transport considerations are not crucial to the carcinogenic potency of the compounds under study and that the enzymesoperating in these systemsare indeed largely nonspecific, as must be assumed for this sort of

0

0.75 0

analysis tobe valid. 0.50

0.60

0.70

0.80

0.90

@E deloc―@

@

Chart 11. Correlationbetweenindices

and

r = 0.930.

the development of a more quantitative theory. Which indices are “best― to use in future studies remains a somewhat subjective decision, but we offer the following comments basedon accuracy in representing carcinogenic potency, ease of calculation, general applicability, and physical interpretability. Clearly, for the dihydrodiol either ‘Bor PB' appears

adequate

for describing

activation

of the

B-region bond. For the carbonium ion we favor the index Qb because

it correlates

strongly

with

carcinogenicily,

is

The analysis presented here must be considered only as a crude first step toward a more complete theoretical examination of the metabolic transformations of these com pounds. That all of the subtleties and many of the specific details of the metabolic transformations are completely ignored should be recognized. The concept of carcinogenic activity itself for a given compound depends on a multitude of factors, including the nature of the organism tested, the mode of administration, the target organs examined, etc. The metabolic process occurs in the very complicated environment of a living cell. Almost certainly the metabolic pathway examined here is not the exclusive route to carci nogenicity for these compounds. The calculations ignore many features, including any influence of specific geomel rical configurations. The correlations observed apparently result becauseelectronic factors are of greater importance than any of these neglected features.

easily calculated, can be applied to substituent effects, and can be satisfyingly interpreted as a measure of carbo nium ion stability (lower positive charges indicate greater delocalization of electron density to this position and hence greater stability). In studies to be published elsewhere, we have found that ‘B and Qb are effective in explaining ACKNOWLEDGMENTS variations in carcinogenic activities among the methyl de we thankDr.Donald Jerinaforhiscomments onthisworkandfor rivatives of chrysene and benz(a)anthracene.Thus, we are sending us copies of several papers prior to publication. we also thank Professors Roland Lehr, J. D. Memory, and R. Daudel for sending us copies gaining confidence in the correctness of the present analy of their papersprior to publication. sis and the effectiveness of its indices. In fact we are now sufficiently confident to suggest that the present type of REFERENCES metabolic analysis appears suitable for use in selected I . Arcos, J. C., and Arcos, M. Molecular Geometryand Mechanismsof Action of Chemical Carcinogens. Progr. Drug Res., 4: 408-581 , 1962. cases as a rapid prescreening procedure, supplementing 2976

CANCERRESEARCH VOL. 38 Downloaded from cancerres.aacrjournals.org on June 2, 2013. © 1978 American Association for Cancer Research.

Theoretical Reactivity Indices and Carcinogenicity 2. Arcos, J. C., and Argus, M. F. MolecularGeometryand Carcinogenic Activity of Aromatic Compounds. New Perspectives. Advan. Cancer Rae., 11: 305-471,

1968.

3. Berger, G. D., Smith, I. A., Seybold, P. 6., and Servo, M. P. Correlation of an Electronic Reactivity Index with Carcinogenicity in Polycyclic Aromatic Hydrocarbons. Tetrahedron Letters, 231-234, 1978.

RNA,and Proteinof MouseSkinOccurswith HighStereoselectivity.

Science,199:778-781, 1978. 21. Mainster,M. A., and Memory,J. D. SuperdelocalizabilityIndicesandthe PullmanTheoryof ChemicalCarcinogens.Blochim. Biophys.Acta, 148: 605-608,1967. 22. Memory,J. D. ElectrophilicSuperdelocalizabllityand the PullmanThe

oryofCarcinogenesis. Inter. J. Quantum Chem. Q. B. Symp., 2: 179-180, 4. Borgen, A., Darvey,H., Castagnoli,N., Crocker,T. T., Rasmussen,R. 1975. E., and Wang,I. Y. MetabolicConversionof Benzo[ajpyreneby Syrian HamsterLiver Microsomesand Bindingof Metabolitesto Deoxyribonu 23. Nagata. C., Fukui, K., Yonezawa,T., and Tagashira, Y. Electronic cleic Acid. J. Med.Chem.,16:502-506,1973. Structure and CarcinogenicActivity of Aromatic Compounds.Cancer Res.,15:233-239,1955. 5. Boyland,E. ChemicalCarcinogenesisand ExperimentalChemotherapy 24. Nakanlshi,K., Kasal,H., Cho, H., Harvey,R. G.,Jeffrey,A. M., Jennette, of Cancer.YaleJ. Biol. Med.,20: 321-341, 1948. K. W., and Weinstein, I. B. Absolute Configuration of a Ribonucleic Acid 6. Coulson, C. A. Electronic Configuration and Carcinogenaeis.Advan. CancerRae.,1: 1-56, 1953. Adduct Formed in Wvo by Metabolism of Benzo(ajpyrene. J. Am. Chem. 7. Daudel, P., Duquesne, M., Vigny, P., Grover, P. L., and Sims, P. Fluorescence Spectral Evidence That Benzo(a)pyrene-DNA Products in Mouse Skin Arise from Diol-Epoxides. Federation European Biochem.

Soc. Letters,57: 250-253,1975. 8. Daudel, A. Recent Progress in Chemical Carcinogenesis.Intern. J. QuantumChem.Q. B. Symp.,4: 169-177,1977. 9. Franke, B. Structure-ActivityRelationshipsin PolycyclicAromatic Hy drocarbons:Inductionof MicrosomalAryl HydrocarbonHydroxylaseand Its PossibleImportancein ChemicalCarcinogenesis.Chem.-Blol.Inter actions, 6: 1-17, 1973.

10. Fukul, K., Yonezawa,T., and Nagata, C. Theory of Substitution in Conjugated Molecules. Bull. Chem. Soc. Japan, 27: 423-427, 1954. 11. Hansch, C., and Fujita, T. [email protected] A Method for the Correlation

of Biological Activity and ChemicalStructure.J. Am. Chem. Soc., 86: 1616-1626, 1964. 12. Herndon,W. C. Theoryof CarcinogenicActivity of Aromatic Hydrocar bons.Trans.N. Y. Acad.Sci., 36: 200-217,1974. 13. Holder,G.,Yagi,H.,Dansette,P.,Jerina,D.M.,Levin,w.,Lu,A.Y.H.,

and Conney,A. H. Effects of Inducers and Epoxide Hydraseon the Metabolism of Benzo[a]pyrene by Liver Microsomes and a Reconstituted System: Analysis by High Pressure Liquid Chromatography. Proc. Natl.

Acad.Sci. U. S., 71: 4356-4360,1974.

14. Hulbert, P. B., Carbonium Ion as Ultimate Carcinogen of Polycyclic Aromatic Hydrocarbons. Nature, 256: 146-148, 1975.

15. Jeffrey,A. M., Jennette,K. W., Blobstein,S. H., Weinstein,I. B., Beland, F. A., Harvey, R. G., Kasai, H., Mlvra, I., and Nakanishi, K. Benzo(a]pyrene-Nucleic Acid Derivative Found in VWo: Structure of a

Benzo[ajpyreneTetrahydrodiol Epoxide-GuanosineAdduct. J. Am. Chem.Soc.,98: 5714-5715,1976. 16. Jerina, D. M., and Lehr, R. E. The Bay-RegionTheory: A Quantum Mechanical Approach to Aromatic Hydrocarbon-Induced Carcinoge nicity. In: V. Ullrich, I. Roots. A. G. Hildebrant, A. W. Estabrook, and A.

H. Conney (ads.), Microsomes and Drug Oxidations. pp. 709-720. Oxford: Pergamon Press, Inc., 1977. 17. Jerina, D. M., Lehr, R., Schaefer-Ridder, M., Yagi, H., Karle, J. M..

Thakker,D. R., Wood,A. H., Lu, A. Y. H., Ryan,D., West,S., Levin, and Conney,A. H. Bay Region Epoxidesof Dihydrodiols:A Concept Explaining the Mutagenic and Carcinogenic Activity of Benzota]pyrene

and Benzo(ajanthracene.In: H. Hiatt,J. D.Watson,and I. Winstin(ads.), Origins of Human Cancer, pp. 639-658. Cold Spring Harbor, N. Y.: Cold Spring Harbor Laboratory, 1977.

18. Jerlna, D. M., Lehr, R. E., Yagi, H., Hemandez,0., Dansette,P. M., Wislocki, P. G., Wood,A. W., Chang,R. L., Levin,W., and Conney,A. H. Mutagenicity of Benzo[a]pyrene Derivatives and the Description of a

QuantumMechanicalModelWhich Predictsthe Easeof CarboniumIon Formation from Diol Epoxides. In: F. J. de Serres, J. R. Fouts, J. R.

Bend,and R. M. Philpot(eds.),In VitroMetabolicActivationIn Mutagen esisTesting,pp. 159-177.Amsterdam:Elsevier/NorthHollandBiomedi cal Press,1976. 19. Kapltulnik, J., Levin, W., Conney, A. H., Yagi, H., and Jerina, D. M. Benzo[a]pyrene 7,8-Dihydrodlol Is More Carcinogenic than Benzo[a]py

rene in NewbornMice. Nature,266: 378-380,1977. 20. Koreeda,M., Moore, P. D., Wislocki, P. G., Levin, W., Yagi, H., and Jerina, D. M. Bindingof Benzo(a]pyrene7,8-Dlol-9,10-Epoxides to DNA,

Soc., 99: 258-260, 1977.

25. Osborne, M. R.. Beland, F. A., Harvey,R. G., and Brooks, P. The Reaction of (±)-7a,[email protected],10fl-Epoxy-7,8,9,10-Tetrahydro benzo(a)pyrene with DNA. Intern. J. Cancer, 18: 362-368, 1976.

26. Pullman,A., StructureElectroniqueet Activité Cancérogene des Hydro carbures Aromatiques. Bull. Soc. Chim. France, 595-603, 1954.

27. Pullman, A., and Pullman, B. Electronic Structure and Carcinogenic Activity of Aromatic Molecules. New Developments. Advan. Cancer Rae..

3: 117-169,1955. 28. Rogers, K. S., and Cammarata,A. Superdelocalizabilityand Charge Density. A Correlation with Partftion Coefficients. J. Med. Chem., 12: 692-693, 1969.

29. Schmidt, 0. Die Characterisierungder einfachenund Krebserzeugen den aromatlschenKohlenwasserstoffedurch die Dlchtevertellungbee timmer Valenzelectronen. Z. Phys. Chem., 42: 83-110, 1939.

30. Scrlbner,J. D. MolecularOrbitalTheoryIn CarclnogenaeisResearch.J. Natl.CancerInst.,55: 1035-1038,1975. 31. Sims,P., and Grover,P. L. Epoxkiesin PolycyclicAromaticMetabolism and Carcinogenesis.Advan.CancerRae.,20: 165-274,1974. 32. Sims, P., Grover,P. L., Swaisland,A., Pal,K., and Hewer,A. Metabolic Activation of Benzo(ajpyrene Proceeds by a Diol-Epoxide. Nature, 252: 326-328, 1974.

33. Streitwieser,A., Jr. MolecularOrbitalTheoryfor OrganicChemists.New York,John Wiley& Sons,Inc., 1961. 34. Thakker, D. R., Yagi, H., Lu, A. Y. H., Levin, W., Conney,A. H., and Jerina, D. M. Metabolism of Benzo(a]pyrene. Conversion of (±)-trans 7,8-dihydroxy-7.8-dihydrobenzo(ajpyrene to highly mutagenic 7,8-diol 9,10-epoxides. Proc. NatI. Aced. Sci. U. S., 73: 3381-3385, 1976.

35. Weinstein,I. B., Jeffrey,A. M., Jennette,K. W., Blobstein,S. H., Harvey, R.G., Harris,C.,Autrup,H., Kasai,H.,andNakanishi,K. Benzo(ajpyrene Diol Epoxidae as Intermediates in Nucleic Acid Binding In vitro and In

Vivo.Science,193:593-595,1976. 36. Wood,A. W., Chang,R. L., Levin,W., Lehr, R. E., Schaefer-Ridder,M., Karle, J. M., Jerina, D. M., and Conney, A. H. Mutagenicity and Cytotoxicity of Benzla]anthracene Diol Epoxides and Tetrahydro-Epox Ides: Exceptional Activity of the Bay Region 1,2-Epoxldes. Proc. Natl. Aced. Sci. U. S., 74: 2746-2750, 1977.

37. Wood,A. W., Levin,W., Lu, A. Y. H., Ryan,D., West,S. B., Lehr, R. E., Schaefer-Ridder, M., Jerina, D. M., and Conney, A. H. Mutagenicity of Metabolically Activated Benzo[ajanthracene 3,4-Dihydrodiol: Evidence

for Bay Region Activation of Carcinogenic Polycyclic Hydrocarbons. Biochem. Biophys. Rae. Commun., 72: 680-686, 1976.

38. Wood,A. W., Levin,W., Lu, A. Y. H., Yagi,H., Hemandez,0., Jerina,D. M., and Conney,A. H. Metabolismof [email protected] and Benzo[a]pyrene Derivatives to Mutagenic Products by Highly Purified Hepatic Micro

somalEnzymes.J. Biol. Chem.,251:4882-4890,1976. 39. Yagi, H., Akagi,H.,Thakker,0. R., Mah,H. D., Koreeda,M., andJerina, D. M. AbsoluteStereochemistryof the Highly Mutagenic7,8-Diol9,10EpoxidesDerivedfrom the PotentCarcinogentrans-7,8-Dlhydroxy-7,8dihydrobenzo(ajpyrene.

J. Am. Chem. Soc., 99:2358-2359,1977.

40. Yang, S. K., McCourt, D. W., Leutz, J. C., and Gelboin, H. V. Benzo(a]pyrene Diol Epoxides: Mechanism of Enzymatic Formation and Optically Active Intermediates. Science, 196: 1199-1201 • 1977.

SEPTEMBER 1978

Downloaded from cancerres.aacrjournals.org on June 2, 2013. © 1978 American Association for Cancer Research.

2977