The epigenetic toxicity of pyrene and related ozonation byproducts containing an aldehyde functional group

Article abstract of DOI:10.1021/es0106117

Gap junction intercellular communication (GJIC) was used to assess the epigenetic toxicity of pyrene, pure byproducts of pyrene ozonation, and other compounds similar in chemical structure. Byproduct mixtures collected from HPLC were also evaluated using GJIC. Of the 11 pure compounds studied, five inhibited GJIC completely. Two inhibiting compounds contained four rings and were the only compounds studied with greater than three rings. The remaining three compounds contained either two or three rings, and all three contained an aldehyde group. Toxicological evaluation and GC/MS of impure byproduct mixtures showed that two common compounds were found in inhibiting fractions. These common compounds contained both a bay region and at least one aldehyde group.

Full text of DOI:10.1021/es0106117

Environ. Sci. Technol. 2001, 35, 3576-3583
translational level (modulating the stability of the gene’s
The Epigenetic Toxicity of Pyrene
and Related Ozonation Byproducts
Containing an Aldehyde Functional
Group
H O L L Y A . H E R N E R , † , §
J A M E S E . T R O S K O , ‡ A N D
S U S A N J . M A S T E N * , †
message), or the posttranslational level (modifying the gene
product or protein by phosphorylation, glycosylation, ni-
trosylation, etc.). At a dosage of 4.5 mol of ozone/ mol of
pyrene, all byproducts inhibitory to GJIC were eliminated
(4). It is apparent that successful remediation must be defined
by the removal of all potentially toxic compounds, not just
the parent compounds responsible for the original risk.
Upham et al. (9) showed that certain four- and five-ringed
PAHs were more inhibitory to GJIC than certain three-ringed
PAHs. In addition, compounds with a bay region were more
inhibitory than those compounds without a bay region (9).
The term bay region is used to describe the cove area created
when multiple benzene rings create a sterically hindered
C-shaped region. Phenanthrene, a three-ringed PAH with a
bay region, was more inhibitory to GJIC than anthracene, a
three-ringed PAH without a bay region (11). In another study,
the addition of a methyl group to anthracene, at a position
creating a bay-like region, caused inhibition of GJIC, whereas
adding the same functional group at another position on the
compound (without the bay-like region resulting) caused no
inhibition (10). A common theory links the bay region, which
many PAHs contain, to carcinogenesis (10-14). Functional
groups such as the hydroxy, the diol-epoxide, and the methyl
group that are adjacent to or part of the bay region have also
been shown to enhance the activity of PAHs (12, 15-18).
The ozonation of pyrene creates two types of byproducts:
phenanthrene and biphenyl-type compounds, with aldehyde,
carboxylic acid, and hydroxy functional groups. In the first
portion of the study, we used ozone to degrade pyrene and
to produce both phenanthrene and byphenyl-type byprod-
ucts. The epigenetic toxicity of pyrene, the byproducts of
pyrene ozonation, and six commerically available byproducts
similar to pyrene ozonation byproducts was evaluated using
in vitro bioassays of intercellular communication in rat liver
epithelial cell culture. The results of the toxicity evaluation
were compared to each compound’s chemical structure. The
first portion of the study focused on byproducts that
constituted a majority of the mass. In the second portion of
the study, impure byproduct fractions were evaluated to
determine whether a byproduct, constituting a small per-
centage of the total mass, was contributing to the increased
toxicity documented in previous studies. Two solutions
containing pyrene were ozonated at differing dosages and
fractionated. A toxicity evaluation and compound identifica-
tion using GC/ MS was performed for each fraction. The
chemical structures for each fraction were compared to the
respective toxicity. The objective was to determine whether
any trends existed between chemical structure and epigenetic
toxicity as measured by inhibition of GJIC.
Department of Civil and Environmental Engineering and
Department of Pediatrics and Human Development, Michigan
State University, East Lansing, Michigan 48824-1326
Gap junction intercellular communication (GJIC) was used
to assess the epigenetic toxicity of pyrene, pure byproducts
of pyrene ozonation, and other compounds similar in chemical
structure. Byproduct mixtures collected from HPLC were
also evaluated using GJIC. Of the 11 pure compounds studied,
five inhibited GJIC completely. Two inhibiting compounds
contained four rings and were the only compounds studied
with greater than three rings. The remaining three
compounds contained either two or three rings, and all
three contained an aldehyde group. Toxicological evaluation
and GC/MS of impure byproduct mixtures showed that
two common compounds were found in inhibiting fractions.
These common compounds contained both a bay region
and at least one aldehyde group.
Introduction
Polycyclic aromatic hydrocarbons (PAHs) are derived from
the incomplete combustion of organic matter and are found
in soil, air, and water (1-3). Sixteen PAHs have been identified
as priority pollutants and eight as carcinogens and possible
carcinogens (2, 3). These compounds are recalcitrant to
conventional treatment and persist in the environment (3).
However, some PAHs have been successfully degraded using
chemical and biological technologies (4-7).
Ozone has been used in the degradation of PAHs and
other compounds resistant to conventional treatment meth-
ods (3). Although ozone is capable of the degradation of PAHs
to carbon dioxide, water, and straight chain aliphatics, studies
have shown that certain byproducts of PAH ozonation can
be as or more harmful than the parent compounds them-
selves (4, 5, 8, 9). A dosage that will degrade the parent
compound and the harmful byproducts but still offers process
efficiency and low cost must be used.
Materials and Methods
Pyrene, a four-ringed PAH, was almost entirely degraded
(>90%) by ozone at a dosage of 1.6 mol of ozone/ mol of
pyrene (4). However, the mixture was still toxic as determined
using inhibition of gap junction intercellular communication
(GJIC) to monitor epigenetic toxicity (9). Epigenetic toxicity
is defined as the alteration of the expression of genes, either
at the transcriptional level (turning genes “on” or “off”), the
Chem icals. Pyrene, neutral red dye, and lucifer yellow dye
were purchased from Sigma Chemical Co. (St. Louis, MO).
Diphenic acid, 2-biphenyl carboxylic acid, 4-biphenyl car-
boxylic acid, 4-biphenyl carboxaldehyde, and 37% formal-
dehyde were all purchased from Aldrich Chemical Co.
(Milwaukee, WI). 4-Carboxy-5-phenanthrene carboxalde-
hyde, 1,2,3,4-tetrahydro-9-phenanthrene carboxaldehyde,
and 9-oxo-1-fluorene carboxaldehyde were purchased from
Sigma-Aldrich’s Library of Rare Chemicals (Milwaukee, WI).
Com pound Q, 4-carboxypenathrene, and 4-carboxy-5-
phenanthrene carboxaldehyde, byproducts of pyrene ozon-
ation (Figure 1), were generated in our laboratory. Acetonitrile
and sodium chloride were purchased from EM Science
* Corresponding author phone: 517-353-8539; fax: 517-355-0250;
e-mail: masten@egr.msu.edu.
† Department of Civil and Environmental Engineering.
‡ Department of Pediatrics and Human Development.
§
Present address: RMT Engineering, 1143 Highland Drive, Suite
B, Ann Arbor, MI 48108-2237.
9
3 5 7 6 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 17, 2001
10.1021/es0106117 CCC: $20.00
2001 Am erican Chem ical Society
Published on Web 07/31/2001

distance the dye traveled. The migration of the dye in the
PAH treated and untreated cells was reported as a fraction
of the distance the dye traveled in the control. The value of
the control was 1.00, and any value less than 0.3 was
considered completely inhibitory to GJIC. A value of 0.3 is
commonly used to represent completion inhibition and is
equal to the width of one row of cells with no dye having
migrated beyond its boundary. A value between 0.9 and 0.3
was considered partially inhibitory to GJIC since a value
greater than 0.9 is often difficult to distinguish statistically
from the controls.
Cell cultures were exposed to PAH stock solutions in the
range of 0 to 200 µM. The equivalent volumes of acetonitrile
added to cell cultures were used as vehicle controls. Vehicle
controls using acetonitrile were run in conjunction with all
cell culture experiments. The volume of acetonitrile in each
control was equivalent to the volume of acetonitrile required
to deliver the compound in the test cultures. All experiments
were conducted in triplicate (three plates per dosage or time
exposure), and values were reported as an average ( 1 SD.
A statistical analysis was performed to determine significance
between controls and test cultures. A t-test was performed
to determine whether target averages differed significantly
from that of the controls at p e 0.01.
FIGURE 1. Pyrene and laboratory-generated byproducts. Pyrene
was purchased commercially and used during ozonation experi-
ments for generation of pyrene ozonation byproducts. Compound Q,
4CP, and 4C5P were obtained by chromatographic separation of the
mixture of ozonation byproducts (ozonation conducted at pH ) 2
with ozone dosage of 1.6 mM ozone/mM pyrene, 0.5 M phosphate
buffer solution).
(Gibbstown, NJ). Sodium phosphate and ammonium per-
sulfate were purchased from Columbus Chemical Industries
(Columbus, WI) and Life Technologies (Gaithersburg, MD),
respectively. Phosphoric acid and sodium tetraborate were
purchased from J. T. Baker (Phillipsburg, NJ).
Ozonation. Ozone was generated by corona discharge
using a Polymetrics model T-408 ozone generator (San Jose,
CA). Oxygen was first dried using a molecular sieve. A 250-
mL gas washing bottle was used for the ozonation reactor.
Ozone was bubbled into 200 mL of an acetonitrile and water
(90:10 v/ v) solution containing 5 mM dissolved pyrene. In
an aqueous environment, water is a participating solvent in
the degradation of pyrene using ozone (22). A 10% water
solution ensures the reaction is not water limited but is dilute
enough to avoid solubility problems due to the nonpolar
characteristics of pyrene. The pH of the solution in the reactor
was acidified to approximately 2 using a phosphoric acid
solution (0.5 M) or increased to approximately 9.5 using a
borate buffer solution (16 mM) as prepared by Adams (23).
The flow of ozone was regulated at 100 mL/ min with a side
track flow controller (Sierra Instruments Inc., Monterey, CA).
The ozonated solutions were continuously mixed with a
magnetic stirrer and stir bar. Influent and effluent gaseous
ozone were monitored spectrophotometrically at 258 nm
using a UV-Vis spectrophotometer (model 1201 Shimadzu,
Scientific Instruments, Japan). Excess effluent gaseous ozone
was trapped in an aqueous 2% (w/ v) potassium iodide
solution.
Cell Culture Techniques. The GJIC assay was used
because it has been shown to be not only a reliable, simple,
and inexpensive in vitro assay, but also to be a very good
predictor of epigenetic or nonmutagenic toxicants (19). WB-
F344 rat liver epithelial cells were obtained from Dr. J. W.
Grisham and Dr. M. S. Tsao of the University of North Carolina
(Chapel Hill, NC) (13). The WB-344 cell line was used because
it is a diploid, nontumorgenic cell line, derived from the
strain of rat on which many of the environmental toxicants
and carcinogenic, or potentially carcinogenic, chemicals are
tested. Therefore, there would be a large amount of toxicity
and cancer data of in vivo tested chemicals with which our
results could be compared. While the use of human diploid,
nontumorgenic liver cells would have been ideal, primary
hepatocytes quickly lose their biochemical phenotypes in
vitro, and no normal human epithelial liver cells, equivalent
to these rat liver cells, exist.
Cells were cultured in 2 mL of D-medium (formula no.
78-5470EG, GIBCO Laboratories, Grand Island, NY) and were
supplemented with 5% fetal bovine serum (GIBCO Labo-
ratories, Grand Island, NY). Cells were prepared for experi-
mentation in 35 mm2 plastic Petri dishes (Corning, Corning,
NY). Cells were incubated at 37 °C in a humidified atmosphere
containing 5% CO2 and 95% air. Bioassays were conducted
on confluent cultures obtained after 2 days’ growth. Cell
cultures were photographed using a Nikkon Diaphot-TMD
epifluorescence phase-contrast microscope illuminated with
an Orsam HBO 200W lamp and equipped with a 35-mm FA
camera (Nikkon, Japan).
In Vitro Bioassays of GJIC. The scrape loading/ dye
transfer (SL/ DT) technique was adapted from El-Fouly et al.
(20). Cell cultures were incubated with target compounds
for various time periods with constant dosage or constant
time periods with varied dosage. GJIC assays were conducted
at noncytotoxic levels as determined by the neutral red dye
uptake assay (21). Following chemical treatment, cells were
rinsed five times with phosphate buffer solution (PBS).
Approximately, 2 mL of lucifer yellow dye were added to cell
cultures scraped using a steel surgical blade. Following a
3-min incubation, the dye was removed from cell cultures
that were again rinsed with PBS five times. Cell cultures were
fixed using 0.5 mL of 4% formalin. An inverted fluorescence
microscope equipped with a camera was used to record the
HPLC Analysis. Ozonated materials were evaporated in
vacuo (Buchii RE111 Rotovapor, Brinkman, Westbury, NY)
at 35 °C and weighed prior to separation by semipreparative
high-pressure liquid chromatography. Ozonation by-
product mixtures were fractionated using a semipreparative
HPLC (Perkin-Elmer, series 200, Cupertino, CA) using a
Phenomenex, Capcell Pak, C18 (5 µM) column (i.d. 10 × 250
mm). A gradient solvent system using acteonitrile and water
was employed with a flow rate of 1.5 mL/ min. A 70/ 30% (v/ v)
acetonitrile/ water solution at the time of injection was
ramped linearly to 75/ 25% (v/ v) acetonitrile/ water over 23
min. At 23 min, the solvent system was increased to 90/ 10%
(v/ v) acetonitrile/ water for 15 min. At 38 min after injection,
the solvent system was decreased to 70/ 30% (v/ v) acetonitrile/
water for 10 min to yield a total run time of 48 min. The
eluent was monitored at both 240 and 254 nm (Waters, model
2487, Milford, MA). Fractionated materials were dried in
vacuo at 35 °C and resuspended to make stock solutions
ranging in concentration from 0 to 20 mM as pyrene. Aliquots
ranging from 0 to 30 µL were added to the cell cultures used
for GJIC assays.
9
VOL. 35, NO. 17, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 3 5 7 7

curves were obtained for commercial and laboratory gener-
ated 4C5P, although commercial 4C5P was more inhibitory
than the laboratory isolated version. The commercially
obtained 4C5P was evaluated by HPLC and found to contain
impurities, one of which was pyrene. Residual pyrene, a
commercial compound impurity, is most likely responsible
for the increased activity noted in the 4C5P purchased
commercially.
Phenanthrene, which has the same base structure without
any functional groups, does completely inhibit GJIC (9). Both
4CP and 4C5P are phenanthrene-type compounds, each with
a bay region and an acid group. 4C5P also contains an
aldehyde group. However, 4C5P does exist in two configura-
tions, one of which is a pseudo-four-ringed structure where
the acid and aldehyde functional groups interact to create
a fourth nonaromatic ring (Figure 1). This alternate config-
uration may play a role in the lack of activity observed for
the 4C5P. Interestingly, 4C5P is the most abundant byproduct
of pyrene ozonation (24).
Of the laboratory-generated byproducts, only compound
Q (CMP Q), a three-ringed aromatic compound with a fourth
nonaromatic ring, almost completely inhibited GJIC but at
a higher dosage than pyrene (Figure 3). Although this
compound was inhibitory to GJIC, the results could not be
compared with pyrene from a parent-byproduct compound
standpoint since methylations encountered during purifica-
tion translated the structure of compound Q from its original
oxidized form. However, compound Q was one of the largest
PAHs studied (four rings) and had a unique structure. Toxicity
data for compound Q were evaluated and used as if it were
just another PAH, unrelated to pyrene.
FIGURE 2. Commercially obtained compounds. These seven
compounds, with structures similar to those generated during
ozonation of pyrene, were purchased and evaluated for their ability
to inhibit gap junction intercellular communication.
Previous studies have shown that some byproducts of
PAH ozonation can be as or more toxic than the parent
compound (4, 8, 9). Although none of the byproducts during
this study exhibited that characteristic, it is still likely to be
the case. Only a limited number of byproducts were isolated
due to the difficulty of separation of compounds with such
similar structures and due to the small masses generated for
many of the byproducts. It is very possible that one of the
byproducts which constitutes a small amount of the byprod-
uct mass is responsible for the more inhibitory results
observed in studies by Upham et al. (4). During byproduct
isolation studies, byproducts constituting a large percentage
of the total mass were the focus. In the second part of this
study, impure byproduct fractions characterized by GC/ MS,
were investigated to address this possibility.
Interest in the identification of relationships between
structure and chemical activity prompted the investigation
of six additional commercial compounds (Figure 2). The
breakdown of pyrene occurs with the formation of phenan-
threne-type compounds initially followed by their subsequent
breakdown and formation of biphenyl type compounds.
Therefore, two phenanthrene-type PAHs and four biphenyl-
type PAHs that were commercially available were evaluated
in the same manner. These six compounds resembled the
byproducts in ring number and functional groups. Both three-
ringed compounds, TPC and OFC, inhibited GJIC completely
at a low dose, with responses comparable to pyrene (Figure
4). Both compounds possess a bay-like region and an
aldehyde group but also have subtle differences (Figure 2).
Although, several explanations for the observed activity could
be proposed, the results strongly suggest that the aldehyde
group is at least partially responsible. In this study, OFC,
fluorene with an aldehyde group, completely inhibited GJIC,
whereas fluorene, studied previously, did not (9). And
although phenathrene is inhibitory to GJIC, TPC, phenan-
threne with an aldehyde group, was more inhibitory to GJIC
(9).
FIGURE 3. Dose response curves for pyrene and byproducts of
pyrene ozonation. Concentrations of chemicals ranging from 80 to
200 µM were applied to cell cultures. Cell cultures were exposed
to chemicals for 30 min. A GJIC value of 1 is indicative of total
communication, while a GJIC value of 0.3 is indicative of complete
inhibition of communication. Each data point represents an average
( 1 SD of the distance the dye traveled in three plates. PY and CMP
Q were found to differ significantly from the control at p e 0.01 as
determined by a two-tailed t-test.
Results and Discussion
Pure Com pound Evaluation. A series of compounds were
assessed for inhibition of GJIC. Pyrene, three byproducts of
pyrene ozonation, and seven commercially obtained com-
pounds similar to byproducts of pyrene ozonation were
studied. The compounds ranged from two to four rings and
contained a variety of functional groups. Figures 1 and 2
contain structures and abbreviations for all compounds.
Of the compounds studied, dose response experiments
showed that five compounds were strong inhibitors of GJIC,
and the remaining six were partial inhibitors. The parent
compound pyrene, the largest PAH in size investigated, was
a strong inhibitor, more so than pyrene ozonation byproducts
4CP and 4C5P (Figure 3). 4CP and 4C5P were both partial
inhibitors of GJIC. 4CP and 4C5P were not evaluated at higher
doses due to their aqueous insolubility. Similar dose response
The last four commercial compounds evaluated were all
two-ringed, biphenyl-type compounds with three containing
9
3 5 7 8 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 17, 2001

FIGURE 4. Dose response curves for pyrene and commercially
obtained compounds similar structurally to pyrene ozonation
byproducts. Concentrations of chemical ranging from 150 to 200 µM
were applied to cell culture. Cell cultures were exposed to chemicals
for 30 min. A GJIC value of 1 is indicative of total communication,
while a GJIC value of 0.3 is indicative of complete inhibition of
communication. Each data point represents an average ( 1 SD of
the distance the dye traveled in three plates. OFC, TPC, and 4BCH
were found to differ significantly from the control at p e 0.01 as
determined by a two-tailed t-test.
FIGURE 5. Time response curves for compounds inhibitory to GJIC.
The time of exposure varied from 2 to 350 min. The chemical dosage
for each compound was equal to the dosage at which GJIC was
completely inhibited during dose response experiments. A GJIC
value of 1 is indicative of total communication while a GJIC value
of 0.3 is indicative of complete inhibition of communication. Each
data point represents an average ( 1 SD of the distance the dye
traveled in three plates. All averages at 350 min (except PY) were
found to differ significantly from the averages at the inhibited state
at p e 0.01 as determined by a two-tailed t-test, indicating partial
recovery.
one or more acid functional groups and one compound
containing an aldehyde functional group. Of the four
biphenyl-type compounds, only 4BCH inhibited GJIC but at
a much higher concentration than pyrene (Figure 3). The
maximum concentration evaluated was dictated by solubility
for all compounds. Generally, biphenyl-type compounds are
not strong inhibitors of GJIC, which may explain the high
dosage of 4BCH required for inhibition of GJIC. Although
the concentrations examined were much higher than that
used for the three- and four-ringed compounds, the con-
centrations studied are not unreasonable since biphenyl-
type compounds are more soluble in aqueous solutions and
could be found in greater concentrations in the environment,
should they be produced.
For each of the five compounds that inhibited GJIC
completely, the inhibiting dose was applied to cell cultures
for varying time periods and GJIC observed (Figure 5). Pyrene
inhibited GJIC completely after approximately 20 min and
communication between adjacent cells remained blocked
after 70 min. Although compound Q inhibited GJIC after
approximately 20 min, partial restoration of communication
was observed after 60 min (Figure 5). Both TPC and OFC
blocked cell-cell communication within about 20 min and
also displayed partial recovery of communication after 2 h
(Figure 5). TPC regained almost 100% communication within
4 h of exposure. Although 4BCH required a high concentration
for inhibition of GJIC, the response was observed after only
1 min but recovered partial communication after only 45
min (Figure 5). Time response data for all other compounds
are reported in Herner (25). All five compounds that inhibited
GJIC, inhibited quickly within 30 min, but interestingly, all
five compounds also displayed partial recovery of GJIC
without the removal of the test compound.
kinase C (PKC) phosphorylates the gap junction protein
(connexin 43) and blocks cell-cell communication. At the
same time, however, PKC is also phosphorylating phos-
phatases which dephosphorylate the gap junction protein
and restore communication. There is a lag time before the
phosphatases are activated which allows the inhibition of
GJIC to be observed. If this process is indeed the mechanism
by which the chemical inhibits GJIC, the addition of a
phosphatase inhibitor would prohibit the recovery of GJIC.
It is also possible that the inhibiting compounds work in a
direct manner in their attack (26). Membrane fluidity and
other conditions of the cell’s homeostasis may be altered
which causes the initial inhibition of GJIC. The recovery of
GJIC may be due to the cell’s ability to adapt to the new
conditions. Finally, metabolism of the chemical may explain
the return of cell-cell communication. Certain cells produce
enzymes that can metabolize PAH compounds. However, it
is unlikely that the undifferentiated rat liver epithelial cells
used in this study produce such enzymes.
The indirect inhibition by PK production may also be the
proper explanation of the observed recovery. For example,
12-o-tetradecanoylphorbol-13-acetate (TPA), a known in-
hibitor of GJIC, works by this mechanism (27). In addition
to the dephosphorylation of the gap junction, phosphatase
dephosphorylates PKC, which initially causes inhibition of
GJIC (27). Studies showed that after communication was
restored, the addition of more TPA did not cause inhibition
of GJIC. In this case, the PKC has been completely inactivated
and the TPA cannot act again by the indirect mechanism to
inhibit GJIC until the PKC has been replenished, which takes
approximately 24 h (27).
Cell cultures exposed to compounds inhibitory to GJIC
completely recovered cell-cell communication when the
respective compound was removed from cell culture (Figure
6). In most cases, GJIC was restored completely after
approximately 2 h. This phenomena is consistent with the
Several possible explanations for the transient nature in
which the compounds blocked GJIC exist. It is possible that
the chemical indirectly inhibits GJIC by way of a protein
kinase (PK) activation (26). In this case, a PK such as protein
9
VOL. 35, NO. 17, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 3 5 7 9

FIGURE 6. Time of recovery curves for compounds inhibitory to
GJIC. Cell cultures were exposed to an inhibiting dosage of chemical
for 30 min and were observed up to 350 min after the chemical was
removed. A GJIC value of 1 is indicative of total communication,
while a GJIC value of 0.3 is indicative of complete inhibition of
communication. Each data point represents an average ( 1 SD of
the distance the dye traveled in three plates. All averages greater
than 340 min were found to differ significantly from the averages
at the inhibited state at p e 0.01 as determined by a two-tailed
t-test, indicating recovery.
theory that inhibition of GJIC by a tumor promoter is a
reversible process (28). In addition, the recovery of cell-cell
communication shows indirectly that the compounds tested
were not cytotoxic at the inhibiting doses. Cytotoxicity
experiments conducted for all compounds confirmed that
none of the compounds evaluated at any concentrations
studied were cytotoxic, even after 24 h of exposure (25). This
result proved that the inhibition caused by certain com-
pounds was not due to cell death.
Pyrene Byproduct Fraction Evaluation. Although the
isolated byproducts studied were not more inhibitory to GJIC
than the parent compound, pyrene; there is a possibility that
interaction between products might make the mixture more
toxic than pyrene. During the second phase of this study,
mixtures of byproducts generated by ozonation of pyrene at
low and high pH were fractionated using liquid chroma-
tography and tested for their ability to inhibit GJIC. Two
solutions of pyrene (5 mM), were ozonated until the majority
of the pyrene was oxidized (ozone dosage of approximately
1.6 mol of ozone/ mole of pyrene) (4). Byproducts of pyrene
ozonation were produced at both pH 2 and 9.5. Typically,
ozonation experiments are conducted at a low pH to ensure
that the majority of reactions are due to direct oxidation of
pyrene by molecular ozone. However, there are some
practical circumstances, such as in alkaline soils or water
treatment, where the pH could be much higher and should
be evaluated.
Dose response experiments for byproduct mixtures
generated at low and high pH showed that both mixtures
completely inhibited GJIC. Byproducts from ozonation at
low and high pH were fractionated by RP-HPLC into 10 and
six fractions, respectively (Figure 7). At low pH, a water-
soluble fraction was collected prior to fractionation using
the RP-HPLC. Because of the impurity of the fractions, the
bioassay concentrations were calculated based upon the
molecular weight of the parent compound, pyrene. All
fractions collected from RP-HPLC were evaluated for their
FIGURE 7. Chromatograms from separation of pyrene ozonation
byproducts. (A) Chromatogram from high-pressure liquid chro-
matographic separation of pure products for first portion of this
study. Details of the purification and solvent system are contained
in Herner (24). Peaks 9 and 10 are 4-carboxy phenanthrene and
4-carboxy-5-phenanthrene carboxaldehyde, respectively. Compound
Q was derived from peak 10 and methylated in the laboratory. (B)
Chromatogram from high-pressure liquid chromatographic frac-
tionation of byproduct mixture generated during ozonation of pyrene
at low pH. (C) Chromatogram from high-pressure liquid chromato-
graphic fractionation of byproduct mixture generated during
ozonation of pyrene at high pH.
ability to inhibit GJIC at 75 µM as pyrene, a concentration
at which pyrene completely inhibits GJIC. Fractions collected
from mixtures generated at both low and high pH were
characterized using GC/ MS, and results are shown in Tables
1 and 2. Figure 8 shows the chemical structures for the
identified byproducts.
9
3 5 8 0 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 17, 2001

TABLE 1. GC/MS Characteristics for HPLC Fractions and
Water-Soluble Fraction (low pH mixture)
peak no.
compound I.D. (Figure 8)a
water-soluble fraction
J , K, L
1
2
3
I, J , K, L
A, C, E, I, J
A, B, C, E, Gc, J , K, L, Rb
4
C
5
C, E
6
C, D, E
7
C
8
9
10
C
C, D, E, I, J , K, Lc
A
a
b
c
Mass spectra in ref 24. Mass spectrum located in ref 25. Com -
pound weakly present.
TABLE 2. GC/MS Characteristics for HPLC Fractions (high pH
mixture)
FIGURE 9. Dose response results for byproduct fractions generated
under low pH conditions during ozonation. Cell cultures were
exposed to a concentration of 75 µM (as pyrene) of each byproduct
fraction. Cell cultures were exposed to each byproduct fraction for
30 min. A GJIC value of 1 is indicative of total communication,
while a GJIC value of 0.3 is indicative of complete inhibition of
communication. Each data point represents an average ( 1 SD of
the distance the dye traveled in three plates. Fractions 1-3, 6, 9,
and 10 were found to differ significantly from the control at p e
0.01 as determined by a two-tailed t-test, indicating partial or full
inhibition of GJIC. Note: FW is the water-soluble fraction and has
a compound I.D. number of 1. Fractions are abbreviated consistent
with the following example: Fraction 1 is F1 and has a compound
I.D. number of 2.
peak no.
compound I.D. (Figure 8)a
1
2
3
4
5
6
Ac, C
C, Ec
B, C, D, E, Rb
Ec, Sb
C, Dc, Ec
A
a
b
c
Mass spectra in ref 24. Mass spectrum located in ref 25. Com -
pound weakly present.
FIGURE 10. Dose response curves for byproduct fractions F6 and
F9 generated under low pH conditions during ozonation. Cell cultures
were exposed to each byproduct fraction for 30 min at varying
dosages (concentration as pyrene). A GJIC value of 1 is indicative
of total communication, while a GJIC value of 0.3 is indicative of
complete inhibition of communication. Each data point represents
an average ( 1 SD of the distance the dye traveled in three plates.
Fractions 6 and 9 were found to differ significantly from the control
at p e 0.01 as determined by a two-tailed t-test. Note: Fraction 6
and fraction 9 are abbreviated as F6 and F9, respectively.
FIGURE 8. Previously identified byproducts of pyrene ozonation
(23). Tentative structures were confirmed using GC/MS.
Of the 11 fractions collected from the fractionation of the
pH 2 solution, only fraction 10, which contained residual
pyrene, was completely inhibitory to GJIC at 75 µM as pyrene
(Figure 9). Fractions 1-4, 6, and 9 partially inhibited GJIC.
Complete dose response experiments for fractions 6 and 9
showed these fractions were completely inhibitory to GJIC
at an increased concentration of 100 µM as pyrene (Figure
10). The complete dose response curve for pyrene, fraction
10, is shown in Figure 2. Fractions 6 and 9 both contained
three common compounds, C, D, and E. Fraction 10
contained residual pyrene. Pure compound C was studied
9
VOL. 35, NO. 17, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 3 5 8 1

FIGURE 11. Dose response results for byproduct fractions generated
under high pH conditions during ozonation. Cell cultures were
exposed to a concentration of 75 µM (as pyrene) of each fraction.
Cell cultures were exposed to each byproduct fraction for 30 min.
A GJIC value of 1 is indicative of total communication, while a GJIC
value of 0.3 is indicative of complete inhibition of communication.
Each data point represents an average ( 1 SD of the distance the
dye traveled in three plates. The MIX, PY, and fractions 3 and 6
were found to differ significantly from the control at p e 0.01 as
determined by a two-tailed t-test, indicating partial or full inhibition
of GJIC. Note: MIX is the byproduct mixture prior to fractionation
and has a compound I.D. number of 1. Fractions are abbreviated
consistent with the following example: fraction 1 is F1 and has a
compound I.D. number of 3.
FIGURE 12. Dose response curves for byproduct fractions F3 and
F6 generated under high pH conditions during ozonation. Cell cultures
were exposed to each byproduct fraction for 30 min at varying
dosages (concentration as pyrene). A GJIC value of 1 is indicative
of total communication, while a GJIC value of 0.3 is indicative of
complete inhibition of communication. Each data point represents
an average ( 1 SD of the distance the dye traveled in three plates.
Fractions 3 and 6 were found to differ significantly from the control
at p e 0.01 as determined by a two-tailed t-test. Note: Fraction 3
and fraction 6 are abbreviated as F3 and F6, respectively.
synergy between two or more compounds was the cause. It
is difficult to assess how a compound will behave when
present as a mixture with other compounds. In this case, the
behavior of many of these compounds individually was
unknown as was the exact concentration of each compound
with respect to the others in the mixture. To positively identify
compounds such as compounds B and D as the cause of the
inhibition of GJIC, additional studies of these individual
compounds would be required. Because these compounds
are not commercially available and difficult to isolate in the
laboratory, it will be necessary to synthesize these compounds
for individual study.
in detail and was not found to exhibit inhibitory character-
istics. Compound E was found in fractions other than 6 and
9, and these fractions were not active. Compound D,
containing both a bay region and an aldehyde group, was
present only in fractions 6 and 9 and appears to be a good
candidate for the inhibition noted.
In general, byproducts generated at high pH were
consistent with the byproducts generated at low pH. Of the
six fractions collected from the pH 9.5 solution, fractions 3
and 6 completely inhibited GJIC, and fractions 1, 2, 4, 5
partially inhibited GJIC at 75 µM as pyrene (Figure 11). The
mixture prior to RP-HPLC separation (crude) and pyrene are
also shown for reference. Complete dose response experi-
ments for fraction 3 and fraction 6 (containing residual
pyrene) showed that fraction 3 was slightly more epigeneti-
cally toxic than the parent compound, pyrene (Figure 12).
Fraction 3 contained compounds B, C, D, E, and R. From
test results of low pH fractions, compounds C and E may be
individually ruled out as the likely cause of the inhibition.
Compound R was unlikely as well since it was present only
in trace amounts. Compounds B and D appeared to be the
likely cause of the inhibition, although fraction 5, which was
not inhibitory to GJIC, contained a trace amount of com-
pound D. Compound D has both a bay region and an
aldehyde group. Compound B has the same basic structure
with the addition of two aldehyde groups instead of one. In
addition, commercial compounds, OFC, TPC, and 4BCH all
contained bay regions and aldehyde functional groups and
were inhibitory to GJIC where as their counterparts without
aldehyde groups were not. These results are consistent with
hypotheses drawn from the study conducted by Eberius et
al. (29).
Acknowledgments
We would like to acknowledge the following sources of
funding: NIESH-Superfund Grant 2P42ES04911-05 and the
National Science Foundation. We would also like to ac-
knowledge Dr. Long Lee, Dr. Muraleedharan Nair, Dr. Russel
Ramsewak, Dr. Jehng-Jyun Yao, Dr. Zhi-Heng Huang, Dr.
Beverly Chamberlain, and Alisa Rummel for their technical
assistance during this project.
Literature Cited
(1) Yoshikawa, T.; Ruhr, L. P.; Flory, W.; Giamalva, D.; Church, D.
F.; Pryor, W. Toxicol. Appl. Pharm. 1985, 79, 218-226.
(2) Sontag, J. M. In Carcinogens in Industry and the Environment;
Marcel Dekker: New York, 1981; p 169.
(3) Masten, S. J.; Davies, S. H. R. Environ. Sci. Technol. 1994, 28 (4),
181A-185A.
(4) Upham, B. L.; Yao, J. J.; Trosko, J. E.; Masten, S. J. Environ. Sci.
Technol. 1995, 29 (12), 2923-2928.
(5) Mahro, B.; Schaefer, G.; Kastner, M. In Bioremediation of
Chlorinated and PAH Compounds; Hinchee, R. E., Leeson, A.,
Semprini, L., Ong, S. K., Eds.; Lewis Publishers: Boca Raton,
1994; pp 203-217.
(6) Corless, C. E.; Reynolds, G. L.; Graham, N. J. D.; Perry, R. Wat.
Res. 1990, 24 (9), 1119-1123.
(7) Yang, Y.; Chen, R. F.; Shiaris, M. P. J. Bacteriol. 1994, 176 (8),
2158-2164.
Although individually it appears that byproduct com-
pounds containing both a bay region and an aldehyde group
caused the observed inhibition of GJIC, it is possible that
9
3 5 8 2 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 17, 2001

(8) Sasaki, J.; Arey, J.; Harger, W. Environ. Sci. Technol. 1995, 29,
1324-1335.
(9) Upham, B. L.; Masten, S. J.; Lockwood, B. R.; Trosko, J. E.
Fundam. Appl. Toxicol. 1994, 23, 470-475.
(10) Upham, B. L.; Weis, L. M.; Rummel, A. M.; Masten, S. J.; Trosko,
J. E. Fundam. Appl. Toxicol. 1996, 34, 260-264.
(11) Yamasaki, H. Carcinogenesis 1990, 11 (7), 1051-1058.
(12) Lehr, R. E.; Wood, A. W.; Levin, W.; Conney, A. H.; Thakker, D.
R.; Yagi, H.; Jerina, D. M. Polynuclear Aromatic Hydrocarbons:
Physical and Biological Chemistry, Sixth International Sympo-
sium; Battelle Press: Columbus, 1982; pp 21-37.
(13) Wood, A. W.; Levin, W.; Chang, R. L.; Yagi, H.; Thakker, D. R.;
Lehr, R. E.; Jerina, D. M.; J. M.; Conney, A. H. Polynuclear
Aromatic Hydrocarbons - Third Symposium on Chemistry,
Biology, Carcinogenesis, and Mutagenesis; Ann Arbor Science
Publishers: Ann Arbor, 1979; pp 531-551.
(14) Flesher, J. W.; Kadry, A. M.; Chien, M.; Stansburg, K. H.; Gairola,
C.; Sydnor, K. L. Polynuclear Aromatic Hydrocarbons: Formation,
Metabolism and Measurement; Seventh International Sympo-
sium; Battelle Press: Columbus, 1983; pp 505-515.
(15) Friesel, H.; Schope, K. B.; Hecker, E. Polynuclear Aromatic
Hydrocarbons: Formation, Metabolism and Measurement;
Seventh International Symposium; Battelle Press: Columbus,
1983; pp 515-530.
(16) Hoffmann, D.; Lavoie, E. J.; Hecht, S. S. Polynuclear Aromatic
Hydrocarbons: Physical and Biological Chemistry; Sixth Inter-
national Symposium; Battelle Press: Columbus, 1982; pp 1-19.
(17) Melikian, A. A.; Lavoie, E. J.; Hecht, S. S.; Hoffmann, D.
Polynuclear Aromatic Hydrocarbons: Formation, Metabolism,
and Measurement; Seventh International Symposium; Battelle
Press: Columbus, 1982; pp 861-875.
(18) Silverman, B. D.; Lowe, J. P. Polynuclear Aromatic Hydrocar-
bons: Physical and Biological Chemistry; Battelle Press: Co-
lumbus; pp 743-753.
(19) Rosenkranz, H. S.; Pollack, N.; Cunningham, A. R. Carcinogenesis
2000, 21, 1007-1011.
(20) El-Fouly, M. H.; Trosko, J. E.; Chang, C. C. Exp. Cell Res. 1987,
168, 422-430.
(21) Borenfreund, E.; Puerner, J. A. Toxicology Lett. 1985, 24, 119-
124.
(22) Bailey, P. S. Ozonation in Organic Chemistry, Volume II.
Nonolefinic Compounds; Academic Press: New York, 1982.
(23) Adams, V. D. In Water and Wastewater Manual; Lewis Publish-
ers: Chelsea, 1991.
(24) Yao, J. J. Ph.D. Dissertation, Michigan State University, 1997.
(25) Herner, H. A. Ph.D. Dissertation, Michigan State University,
1999.
(26) Trosko, J. E., personal communication; Michigan State Uni-
versity, February, 1999.
(27) Oh, S. Y.; Madhukar, B. V.; Trosko, J. E. Carcinogenesis 1988, 9
(1), 135-139.
(28) Trosko, J. E.; Madhukar, B. V.; Chang, C. C. Life Sci. 1993, 53,
1-19.
(29) Eberius, M.; Berns, A.; Schuphan, I. Fresenius J. Anal. Chem.
1997, 359, 2274-279.
Received for review February 5, 2001. Revised manuscript
received June 15, 2001. Accepted June 19, 2001.
ES0106117
9
VOL. 35, NO. 17, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 3 5 8 3