PCBs: structure-function relationships and mechanism of action.

Numerous reports have illustrated the versatility of polychlorinated biphenyls (PCBs) and related halogenated aromatics as inducers of drug-metabolizing enzymes and the activity of individual compounds are remarkably dependent on structure. The most active PCB congeners, 3,4,4',5-tetra-, 3,3',4,4'-tetra-, 3,3',4,4',5-penta- and 3,3',4,4',5,5'-hexachlorobiphenyl, are substituted at both para and at two or more meta positions. The four coplanar PCBs resembled 3-methylcholanthrene (3-MC) and 2,3,7,8-tetrachlorodibenzo-p-dioxin (2,3,7,8-TCDD) in their mode of induction of the hepatic drug-metabolizing enzymes. These compounds induced rat hepatic microsomal benzo(a)pyrene hydroxylase (aryl hydrocarbon hydroxylase, AHH) and cytochromes P-450a, P-450c and P-450d. 3,4,4',5-Tetrachlorobiphenyl, the least active coplanar PCB, also induced dimethylaminoantipyrine N-demethylase and cytochromes P-450b+e and resembled Aroclor 1254 as an inducer of the mixed-function oxidase system. Like Aroclor 1254, all the mono-ortho- and at least eight di-ortho-chloro analogs of the coplanar PCBs exhibited a "mixed-type" induction pattern and induced microsomal AHH, dimethylaminoantipyrine NM-demethylase and cytochromes P-450a-P-450e. Quantitative structure-activity relationships (QSARs) within this series of PCBs were determined by comparing their AHH induction potencies (EC50) in rat hepatoma H-4-II-E cells and their binding affinities (ED50) for the 2,3,7,8-TCDD cytosolic receptor protein. The results showed that there was an excellent correlation between AHH induction potencies and receptor binding avidities of these compounds and the order of activity was coplanar PCBs (3,3',4,4' -tetra-, 3,3',4,4',5-penta- and 3,3',4,4',5,5'-hexachlorobiphenyls) greater than 3,4,4',5-tetrachlorobiphenyl approximately mono-ortho coplanar PCBs greater than di-ortho coplanar PCBs. It was also apparent that the relative toxicities of this group of PCBs paralleled their biological potencies. The coplanar and mono-ortho coplanar PCBs also exhibit differential effects in the inbred C57BL/6J and DBA/2J mice. These compounds induce AHH and cause thymic atrophy in the former "responsive" mice whereas at comparable or higher doses none of these effects are observed in the nonresponsive DBD/2J mice. Since the responsiveness of these two mice strains is due to the presence of the Ah receptor protein in the C57BL/6J mice and its relatively low concentration in the DBA/2J mice, the results for the PCB cogeners support the proposed receptor-mediated mechanism of action.

The mechanism of action of the toxic halogenated aromatics has primarily been derived by investigating the effects of 2,3,7,8-TCDD and related PCDD isomers and congeners and several highly toxic halogenated aromatics that are approximate isostereomers of 2,3,7,8-TCDD (11,12,(35)(36)(37)(38). Several significant correlations were noted for the PCDD series of compounds, including: (1) the most active PCDDs are substituted at the 2,3,7, and 8 lateral positions (i.e., 2,3,7,8-TCDD), and Cl substitution or removal of the lateral Cl groups gives compounds with diminished activity; (2) there is an excellent correlation between the toxicity of several PCDDs, their AHH-inducing potency and their relative affinities for a high affinity, low capacity hepatic cytosolic receptor protein; (3) the differential effects of the PCDDs and related toxic halogenated aromatics on genetically inbred nonresponsive DBA/2J and responsive C57BL/6J mice suggest that receptor protein binding by these toxic ligands is the first critical step which initiates the complex sequence of events leading to the observed biologic and toxic responses. In C57BL/6J mice the soluble binding protein (i.e., the Ah receptor) also plays a role in mediating the activities of several aromatic hydrocarbons such as 3-MC and benzo(a)pyrene.
The most toxic PCB congeners, namely, 3,3',4,4'tetra-, 3,3',4,4',5-penta-and 3,3',4,4',5,5'-hexachlorobiphenyl can assume coplanar conformations and are approximate isostereomers of 2,3,7,8-TCDD. These three PCBs are potent AHH inducers (39)(40)(41)(42)(43), elicit biologic and toxic effects comparable to those reported for 2,3,7,8-TCDD, and the results support a common mechanism of action for the toxic PCBs and PCDDs. Based on PCB structure-activity relationships (SARs), it was concluded that the most active compounds must be substituted at the para and at least one meta position of both phenyl rings and must not contain any orthochloro substituents (40,41). Unfortunately, the proposed PCB SARs define a subset of three compounds that are highly active but are present as only trace components in the commercial PCB mixtures (44). It is conceivable that the observed biologic and toxic effects of PCB formulations may either be due to the trace levels of the coplanar PCBs, and/or the presence of highly toxic PCDF impurities (45,46) or the presence of other toxic PCBs which have not been defined by the proposed structure-activity correlations (40,41). Studies in our laboratory with Aroclor 1254 showed that after Florisil column chromatography, the cleaned-up (PCDF-free) commercial PCBs and the crude mixture exhibited comparable AHH induction potencies. This suggested that unidentified toxic PCBs may be present in the commercial mixtures and a more comprehensive PCB structure-activity study was undertaken to resolve this problem.
The di-ortho coplanar PCBs exhibited low binding affinities for the cytosolic receptor protein (52) and were relatively inactive as AHH/EROD inducers in rat hepatoma H-4-II-E cells (51). Thus, no meaningful quantitative SAR were derived for these compounds.
The toxicity of the di-ortho coplanar PCBs has not been systematically investigated; however, two members of this group, 2,2',3,3',4,4'-and 2,2',3,4,4',5'-hexachlorobiphenyl are porphyrinogenic in rats after longterm feeding studies (64). Both of these compounds are among the most active di-ortho coplanar PCB inducers of rat hepatic microsomal AHH and cytochromes P-450c. Based on the relatively low activities of this group of compounds, future toxicologic research should focus on the effects of 2,2',3,4,4',5'-hexachlorobiphenyl since this isomer is a major component of commercial PCBs and preferentially bioconcentrates in human blood, adipose tissue and breast milk (32)(33)(34)61,62). Differential Effects on Genetically Inbred C57BL/6J and DBA/2J Mice This paper has focused on the remarkable effects of structure on the biologic and toxic potencies of PCBs and the parallel between the effects of PCBs and the related toxic halogenated aromatics. It is evident that for both PCBs and PCDDs the most active compounds are approximae isostereomers of 2,3,7,8-TCDD and there is an excellent correlation between the cytosolic receptor binding affinities, AHH induction potencies and toxicities of these halogenated aromatics. Further evidence for a common mechanism of action of PCBs and PCDDs is suggested by the differential effects of these compounds on the responsive C57BL/6J and nonresponsive DBA/2J inbred mice (38,40,42,(68)(69)(70). Administration of the coplanar PCBs or 2,3,7,8-TCDD to C57BL/6J mice results in the induction of hepatic microsomal AHH, immunotoxicity and weight loss whereas the nonresponsive mice are much less susceptible to the effects of these compounds (e.g., Fig. 5). 2,3,4,4',5-Pentachlorobiphenyl, a mono-ortho coplanar PCB induces AHH and causes thymic atrophy in the responsive C57BL/6J mice whereas no significant AHH induction or thymic atrophy is observed in the DBA/2J mice (62). Comparable results have been obtained for most of the monoortho coplanar PCBs (70) and Aroclor 1254 and this suggests that this group of compounds and their coplanar PCB precursors all act through a common mechanism. Support for the receptor-mediated mechanism of action of 2,3,7,8-TCDD has been derived from comparable studies with the two inbred mice strains (36)(37)(38). The responsive C57BL/6J mice contain relatively high levels of the Ah receptor in hepatic and extrahepatic tissues whereas relatively low levels ofthis protein have been detected in the nonresponsive DBA/2J mice (71). Therefore the differential effects of coplanar and mono-ortho coplanar PCB congeners in the two strains of mice lend further support to the proposed receptormediated mechanisms of action for the toxic halogenated aromatics.
Receptor: Ligand Interactions: QSAR The qualitative and quantitative SARs observed for PCB isomers and congeners also support the proposed mechanism of action for the toxic halogenated aromatics (Fig. 6). This model is based on the mechanism of action of steroid hormones in which the process is initiated by the noncovalent interaction between a ligand (i.e., a coplanar toxic halogromatic) and a receptor protein. The ligand: receptor binding complex is then translocated into the nucleus, and presumably interacts with a specific region of the nuclear DNA. These events trigger the de novo synthesis of m-RNA and protein which ultimately leads to the pleiotypic responses observed in the host animals. Although there is considerable evidence for a receptor-mediated mechanism of action for PCBs, the mechanistic details for the individual steps involved in this process are not well defined. This review has focused on two aspects of this model, namely, the receptor:ligand interaction and the induction of cytochrome P-450-dependent monooxygenases.
McKinney and co-workers (72,73) have suggested that an important factor which facilitates the ligand:receptor protein interaction is the molecular polarizability of the lateral chloro substituents. A more comprehensive understanding of ligand:receptor interactions can be obtained by determining quantitative SARs within a series of structurally related compounds (74)(75)(76)(77). This approach is routinely used in planning the synthesis of new agricultural chemicals and drugs and requires an active model chemical substrate that is amenable to structural manipulation at an active site. We have chosen 4'-substituted 2,3,4,5-tetrachlorobiphenyls (Fig. 7) as our first group of structurally related ligands for the following reasons (78,79): (1) 4'-substituted-2,3,4,5-tetrachlorobiphenyls contain a single variable substituent located at a critical lateral (para) position in the biphenyl nucleus, and the unsubstituted compound (Fig. 7, X = H) is relatively nontoxic and does not induce hepatic microsomal AHH; (2) like other toxic halogenated aromatics, the parent PCB congener, 2,3,4,4',5-pentachlorobiphenyl, induces microsomal AHH and cytochromes P-450c and P-450d, binds to the hepatic cytosolic receptor protein and causes thymic atrophy and weight loss in male Wistar rats; 2,3,4,4',5-pentachlorobiphenyl elicits similar effects in the responsive C57BL/6J mice but is inactive in the nonresponsive DBA/2J mice (Fig.  5); (3) this series of compounds can be synthesized by the Cadogan coupling of 4-substituted anilines and 1,2,3,4-tetrachlorobenzene to give a single biphenyl coupling product. Figure 7 illustrates the structurally diverse 4'-substituted-2,3,4,5-tetrachlorobiphenyls which were used as ligands for the competitive receptor binding assays. The results obtained for 13 different halogenated biphenyls that differ only in the structure of the 4'-lateral substituent is summarized in Figure 8. Multiple linear regression analysis were performed with a FACOM M200 computer at the Data Processing Center of Kyoto University. Hydrophobic (,n), electronic (5) and hydrogen-binding (HB) accepting parameters were assigned for each substituent.
Hammett constants for the substituents were obtained from the literature (75) and values were calculated as described (77). The HB parameters for each substituent were assigned values of zero (for nonhydrogen bonders) or one (for hydrogen bonders) (76). Preliminary analysis of the receptor binding data indicated that three compounds with bulky 4'-substituents (X = phenyl, n-butyl and t-butyl) did not fit the equation. One apparent reason for their lack of fit is the van der Waals volume of the bulky substituents (> 35 cm/mole for these groups) which introduces a steric factor into the calculation. If the results for the bulky 4'-n-butyl-, 4'-t-butyl and 4'-phenyl-2,3,4,5-tetrachlorobiphenyls are omitted, thus multiple regression analysis of the data gave log (1/EC.) = 139 u + 1.31 ar + 1.2 HB + 4.20 for n = 15, SD = 0.31, r = 0.916, where r is the correlation coefficient and SD the standard deviation. This equation was recalculated by omitting the receptor binding time for the 4'-nitroand 4'-N-acetylamino-2,3,4,5-tetrachlorobiphenyls. The latter compound did not fit the equation and there were problems in assigning an HB value for the nitro-substituted biphenyl due to possible in-plane and out-of-plane conformations. The recalculated equation for 13 compounds was log (1/EC50) = 1.53 uF + 1.47 -r + 1.09 HB + 4.08 for n = 13, SD = 0.13, r = 0.978. The excellent correlation between binding affinities and substituent physical chemical parameters thus confirms the importance of steric, electronic, hydrophobic and hydrogen bonding factors in facilitating the interactions between toxic haloaromatics and the active binding state of the cytosolic receptor protein. The validity of the QSAR approach has been confirmed by determining the AHH/ EROD induction potencies of the 4'-substituted-2,3,4,5tetrachlorobiphenyls using rat hepatoma H-4-II-E cells in culture. With one exception, the order of induction potency for these compounds was identical to their order of binding affinities; the linear regression equation for this correlation is being calculated. 4'-Iso propyl-2,3,4,5tetrachlorobiphenyl was much less active as an inducer than predicted from receptor binding data and this may be related to problems in cell uptake.
Preliminary results have shown that other substituted biphenyls, dibenzo-p-dioxins and dibenzofurans can also be used to probe in vitro QSARs and this data khc,bng potenty 9.6xlo-Hl will be used to further define the important interactions between halogenated aromatics and the 2,3,7,8-TCDD receptor protein. It is also apparent from in vivo studies that halogenated aromatics designed for in vitro QSAR studies can be used to investigate regulatory mechanisms for cytochrome P-450 isozymes and other drugmetabolizing enzymes. This research is now in progress.
The financial assistance of the National Institutes of Health (ROl-ES02798), the Natural Sciences and Engineering Research Council of Canada, the Center for Comparative Medicine and the Texas Agricultural Experiment Station (Grant No. 6376) is gratefully acknowledged.