[CANCER RESEARCH
38, 431-438, February
1978]
Detection of Mutagenic Impurities in Carcinogens and Noncarcinogens by
High-Pressure Liquid Chromatography and the Sa/mone//a/Microsome
Test
E. Victor Donahue,
Joyce McCann,1 and Bruce N. Ames2
Department of Biochemistry, University of California, Berkeley, California 94720
ABSTRACT
We have used high-pressure liquid chromatography
and the Sa/mone//a/microsome mutagenicity test to look
for mutagenic impurities in 11 carcinogens and noncarcinogens. Because of the million-fold range in mutagenic
potency observed in the Salmonella test, even trace
amounts of potent mutagenic impurities in a nonmutagenic compound could be detected. The mutagenicity of
7-hydroxy-2-acetylaminofluorene, a noncarcinogen in the
standard animal carcinogenicity tests, is shown to be due
to a small amount of impurity, which is probably the potent
carcinogen 2-acetylaminofluorene. This is discussed in
relation to the statistical limitations of animal carcino
genicity tests. We also discuss the role of mutagenic im
purities in assessing the mutagenicity of environmental
(and industrial) chemicals with high-sensitivity mutagenic
ity assays, such as the Sa/monef/a/microsome test.
INTRODUCTION
We have previously described a rapid and sensitive bac
terial test for the detection of environmental mutagens
(reviewed in Refs. 4 and 26). The test measures reversion
to prototrophy of a special set of histidine mutants of
Salmonella typhimurium and incorporates an aspect of
mammalian metabolism by utilizing homogenates of rat (or
human) liver for activation of chemicals to their mutagenic
forms. Recently, we have validated the test for the detection
of carcinogens as mutagens (25, 27). Of 175 carcinogens
tested, spanning a wide variety of chemical classes, 90%
were mutagenic in the test. Despite the difficulties inherent
in defining noncarcinogenicity (25, 26), few of the 107
noncarcinogens tested showed any degree of mutagenicity.
Despite the fact that chemicals of the highest purity availa
ble were used, in a few cases we found that the classifica
tion of a chemical as to mutagenicity or nonmutagenicity
was complicated by the presence of mutagenic impurities.
In the past few years, the technique of HPLC3 has been
developed to a high level of sophistication (13). The tech' Senior Fellow of the California Division of the American Cancer Society.
2 Recipient of Contract AT(04-3)34PA156 from the Energy Research and
Development Administration. To whom requests for reprints should be
addressed.
3 The abbreviations used are: HPLC, high-pressure liquid chromatogra
phy; 1-AA, 1-aminoanthracene; 2-AB, 2-aminobiphenyl; 2-AAF, 2-acetylami
nofluorene; 1-NA, 1-naphthylamine; bis-2,7-AAF, bis-2,7-acetylaminofluorene; 3-OH-B(a)P, 3-hydroxybenzo(a)pyrene; B(e)P, benzo(e)pyrene; 7MB(a)A, 7-methylbenz(a)anthracene; 7-OH-2-AAF, 7-hydroxy-2-acetylaminofluorene; 5-OH-2-AAF, 5-hydroxy-2-acetylaminofluorene; 4-AAF, 4-acetylaminofluorene; TCE, trichloroethylene.
Received April 21, 1977; accepted November 11, 1977.
FEBRUARY
nique is similar to normal liquid column chromatography
except that the eluting solvent is pushed through the column
by a high-pressure pump. This method can be quite useful
as a tool when looking for mutagenic impurities: (a) the
relative speed of the method makes it particularly adaptable
for use with a rapid in vitro screening method like the
Salmonella test; (b) the sensitivity and degree of resolution
possible are sufficient to permit separation of chemicals
from very tiny amounts of mutagenic impurities that might
be present; (c) one can readily obtain amounts of chemicals
from the columns sufficient for mutagenesis testing; (d)
there are a number of different types of columns and
solvents that may be used in HPLC, permitting its applica
tion to the separation of a wide range of different types of
chemicals; and (e) the breakdown of relatively unstable
chemicals can be minimized because of short retention
times and because the column operates at ambient temper
atures.
We describe here the use of HPLC, in combination with
the Sa/mone//a/microsome test, to detect small amounts of
mutagenic impurities. We also discuss the problems that
mutagenic impurities can create in mutagenesis testing.
MATERIALS
AND METHODS
Chemicals. Chemicals were obtained as follows. When
possible, chemicals of the highest purity commercially avail
able were used, and purity, when known, is indicated in
parentheses. 1-AA (90%), 2-AB (>97%), 2-AAF (>97%), and
benzo(a)pyrene (>99%) were from Aldrich Chemical Co.,
Inc., Milwaukee, Wis. 1-NA and 2-naphthylamine (>99%)
were from Sigma Chemical Co., St. Louis, Mo. bis-2,7AAF, 2-aminoanthracene, and 4-aminobiphenyl (99%) were
from Schuchardt, Munich, West Germany. 3-OH-B(a)P and
B(e)P (highly purified) were from the National Cancer Insti
tute, Bethesda, Md. Samples of 7-MB(a)A were gifts of P.
Grover and J. Flesher. Samples of 7-OH-2-AAF were gifts of J.
and E. Miller and E. Weisburger, and 5-OH-2-AAF was a gift
of E. Weisburger. 4-AAF (>98%) and 3-methoxy-4-aminoazobenzene were gifts of V. Simmon.
Solvents. Solvents were from either Mallinckrodt Chemi
cals, St. Louis, Mo., or Matheson, Coleman and Bell,
Cincinnati, Ohio, and were spectroquality grade. All sol
vents were degassed on the house vacuum prior to use.
Columns. Columns (purchased prepacked) were, for the
benzopyrenes and 7-MB(a)A, a reverse-phase Micro C,«
column (10-^m particles coated with octadecyl silane), 4
mm (inside diameter) x 30 cm (Part 27324; Waters Associ
ates, Milford, Mass.), and, for all other chemicals, a Lichrosorb silica column, 4.5 mm (inside diameter) x 25 cm,
1978
Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1978 American Association for Cancer Research.
431
E. V. Donahue
et al.
packed with 10-¿imLichrosorb Si60 particles (Part 254-04; icals that we examined are not highly soluble in this solvent
Altex Scientific Co., Berkeley, Calif.)- Void volumes of both system, which limits the amount of chemical that may be
columns were about 2.5 ml.
injected onto the column. For example, 6 injections were
necessary to obtain sufficient 7-OH-2-AAF for mutagenesis
Chromatography. Chromatography was at room temper
ature (-23°). In most experiments we used a Spectra-Phys
testing. A variety of 2-AAF metabolites and other aromatic
ics high-pressure liquid Chromatograph, Model 3500B amines have been resolved on HPLC (15, 18, 39, 43).
(Spectra-Physics Co., Santa Clara, Calif.). In a few cases a Polycyclic hydrocarbons were chromatographed in the re
recently acquired Waters Associates Model ALC/GPC 244 verse-phase mode, using an acetonitrile/water solvent sys
high-pressure liquid Chromatograph equipped with a Model tem. Typically, methanol/water instead of acetonitrile/water
6000A solvent delivery system was used. Absorbance was has been used as a mobile phase (14, 21, 33, 34, 42). Both
at either 254 or 280 nm using a Spectra-Physics Model 230 solvent systems permit comparable resolution, and methadual-channel absorbance detector or a Waters Associates nol is less expensive than acetonitrile. However, acetonitrile
has the advantage of permitting faster separations for 2
Model 440 absorbance detector. Flow rates were 2 ml/min
except for B(e)P (3 ml/min) and 7-MB(a)A (4 ml/min).
reasons: (a) acetonitrile has a much lower viscosity than
Eluates were collected in glass vials and dried under a flow does methanol, and thus one can run aqueous acetonitrile
of nitrogen gas. Recovery of sample was essentially 100% solutions at a faster flow rate without exceeding the pres
from the column. This was determined spectrophotometricsure limit of the column; and (b) acetonitrile has a greater
ally in several cases by reinjecting collected material back ability to solubilize many polycyclic hydrocarbons than
does methanol (35), and so greater amounts can be loaded
onto the column.
into the system per injection.
Details of the conditions used for the Chromatographie
Mutagenesis Testing. Mutagenesis testing was as previ
separations are in the legend to Chart 1. In general, for
aromatic amines the running solvents were either isooctane ously described (4). All results reported are taken from the
(2,2,4-trimethylpentane)/methylene
chloride or, for the linear region of dose-response curves. Chemicals were
more polar compounds such as 2-AAF, chloroform/methaprotected from light with aluminum foil. Purified chemicals
nol [Matsushima et al. (23) have used a similar solvent not immediately used for mutagenicity testing were dried
system for thin-layer Chromatography of 2-AAF derivatives]. and stored frozen, in sealed vials under nitrogen.
One disadvantage of chloroform/methanol is that the chemA. 2-tacro-
6 a-Nophltiyloniix
C r-Hy*», D S-Hydrair-2-AAF
E 4-ÕAF
RESULTS
"I
F
l-
H 27-»j-«F
G Bfl(uo(f)prrm
2O
40
I 7-IMhrlbefiionthroctA« JJ. 3-Hydroiy-b«i»o(o)pTfeoe
3O
«0
90
K 3-UelKoiy-4-omtnootob«nitnc
Chart 1. HPLC profiles. For A to G the position of the close relative is
indicated by an arrow. Injections are marked by negative deflections of the
base line. All tracings shown were at 254 nm except for 2,7-bis-AAF and 3methoxy-4-aminoazobenzene. which were at 280 nm. Running solvents (v/v)
were ¡sooctane/methylene chloride (7/6) (A, B, F); chloroform/methanol
(97.5/2.5) (H); CHCIj/methanol (9/1) (C); CHCI3/methanol (100/1) (D); CHCI3/
méthylène
chloride (2/1) (E); acetonitrile/H20 (3/2) (/); isooctane/methylene
chloride (3/2) (K); and acetonitrile/HjO (1/1) (G, J). Horizontal axis, ml;
vertical axis, absorbance. In all cases the volume injected was 0.1 ml.
Aromatic amines were injected in the running solvent, and the polycyclic
hydrocarbons were injected in acetomtrile. Concentrations of the injection
solutions were 0.1 mg/ml (H), 0.2 mg/ml (/), 0.5 mg/ml (D, J), 1 mg/ml (G,
K), 2.5 mg/ml (C), 5 mg/ml (£),10 mg/ml (fl, F), and 20 mg/ml (A).
432
We have used HPLC to obtain highly purified samples of
a number of carcinogens and noncarcinogens (8 aromatic
amines and 3 polycyclic hydrocarbons) for mutagenesis
testing. All chemicals were chromatographed under a vari
ety of solvent conditions and retention times. HPLC tracings
for the 11 chemicals are shown in Chart 1. The tracings
shown are those from the preparative runs used to obtain
the purified samples for which data are reported in Tables 1
and 2.
Before purification all 11 chemicals were weakly mutagenie in the Salmonella test, and for several reasons (see
below) we suspected that mutagenic impurities might be
responsible for all or part of the mutagenic activity.
Even though, in general, chemicals were of the highest
purity commercially available, we found a number of impur
ities in all chemicals examined (see Chart 1) with only 1
exception, B(e)P. In almost all cases the impurities repre
sented only a very small fraction of the total absorbing
material. However, since ng amounts of the most potent
mutagens will cause a significant mutagenic response in
the Salmonella test (25-27), only very low levels of potent
mutagenic impurities would, in theory, be required to cause
a mutagenic response (see "Discussion").
We compared the mutagenicity of each of the 11 chemi
cals before and after purification on HPLC (Tables 1 and 2
and Chart 2), and in 4 of the 11 cases examined we found a
significant change in the mutagenic activity after purifica
tion. In 1 case (7-OH-2-AAF, Table 1 and Chart 2), the
purified chemical was virtually nonmutagenic; in 2 cases
[bis-2,7-AAF and 7-MB(a)A, Table 2], direct mutagenic activ
ity (not requiring the rat liver "S-9 Mix") was lost on
CANCER
RESEARCH
VOL. 38
Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1978 American Association for Cancer Research.
Impurities and the Salmonella/Microsome
Test
Table 1
Mutagenicity of close relatives of potent carcinogens before and after purification by HPLC
Activity"Before
Mutagenic
HPLCRevenants
HPLCRevertants
per:Pfl121/40121/50239/20059/46047/50023
per:Chemical2-Naphthylamine1-NA4-Aminobiphenyl2-AB2-AA
impurity''0.21.49ND0.080.250.2NDRatio192535715302552362HQ252/4133/405069/50202/5048
F4-AAF5-OH-2-AAF7-OH-2-AAF2-Aminoanthracene1-AABenzo(a)pyrene''B(e)PCarcinogenicity6+cO+?+00cO+w+d+w+%
" Strain TA100 and S-9 Mix with 20 /il of Aroclor-induced S-9 were used except for the aminobiphenyls (50 /il of S9), the acetylaminofluorene derivatives (TA1538; 150 /il of phénobarbitalS-9), and the aminoanthracenes (TA98; 50 /il
of phenobarbital-induced S-9). All results are single points from the linear region of dose-response curves; e.g., 252/4
= 252 revertants/4 /ig. Spontaneous revertants have been subtracted and were approximately 160 (TA100), 25
(TA1538), and 40 (TA98) per Petri plate. The ratio compares the mutagenic activities of the potent carcinogens to
those of each weak relative on a nmole basis; e.g., 4-AAF was 357 times less mutagenic than was 2-AAF before HPLC.
* +, carcinogen; 0, noncarcinogen; w*, weak carcinogen; cO, noncarcinogen in most studies, with some reports of
weak or marginal activity; ?, generally assumed a noncarcinogen, based on structural considerations, but has not
been tested for carcinogenicity. Carcinogenicity references are cited in Ref. 27.
' Percentage of impurity (by weight) was calculated based on the absorption of the suspected potent relative at 254
nm. determined in a parallel run. ND, not detected.
" 1-AA has been considered ai borderlini
borderline carcinogen (6).
' A benzo(a)pyrene impurity in B(e)P would be less expected than a number of other potentially mutagenic relatives,
because of the usual route of synthesis of B(e)P.
Table 2
Mutagenicity of several carcinogens before and after HPLC
S-9 Mix with 20 /il of Aroclor-induced S-9 was used except for bis-2,7-AAF (150 /il of
phenobarbital-induced S-9). Strains were TA98 [7-MB(a)A without S-9 Mix, 3-OH-B(a)P], TA1538
(bis-2,7-AAF, 3-methoxy-4-aminoazobenzene), and TA100 [7-MB(a)A + S-9 Mix].
HPLCRevertants
per:S-9
HPLCRevertants
per:/¿g
nmole4851/10
/ig
Chemicalbis-2,7-AAF7-MB(a)A"3-OH-B(a)P"3-Methoxy-4-aminoazobenzeneBefore
nmole4/10
135.8+
<0.12338/10
9.512/20
<0.15505/20
2.7+
260/5
12.6372/2
6.1127/5
49.8+
6.8396/5
96.7+ 20590/10''
361/1
21
467After .215193/10
345
" 7-MB(a)A from a second source did not appear to contain the impurity.
Results were variable from experiment to experiment.
c Dose-response curves for 3-methyoxy-4-aminoazobenzene were nonlinear (27).
564/5
.6221/20
purification,
while activity in the presence of the S-9 Mix
was retained; in 1 case (2-AB) the mutagenic
potency
decreased, but did not disappear. Each of these cases is
discussed briefly below.
Close Relatives of Potent Mutagens [2-AB, 1-NA, 7-OH2-AAF, 5-OH-2-AAF, 4-AAF, 1-AA, and B(e)P]. We were
particularly
interested in Chemicals A to G in Chart 1
because these had a combination
of properties that sug
gested, not only that mutagenic impurities might be pres
ent, but that in many cases we might be able to predict
what the impurities were: (a) they had all been reported to
31
be either very weakly carcinogenic
or noncarcinogenic
in
the standard animal carcinogenicity
tests; (b) they were all
close relatives (in most cases isomers) of chemicals that
are much more potent carcinogens in animal carcinogenic
ity tests (these are listed in Table 1 in italics) and are also
potent mutagens in the Salmonella test; and (c) they were
all very weakly mutagenic in the Salmonella test compared
to their potent relatives (Table 1).
Since purification
procedures
commonly employed in
the preparation
of chemicals may not efficiently resolve
substances that are isomerie or that differ only slightly in
FEBRUARY 1978
Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1978 American Association for Cancer Research.
433
E. V. Donahue
et al.
700
Before HPLC
500
£ 300
I
100
100
300
500
Micrograms
Chart 2. Mutagenicity of 7-OH-2-AAF before and after purification on
HPLC. Each dose level was incorporated into top agar along with bacterial
tester strain TA1538 and S-9 Mix (see Table 1). Histidine revenants were
scored after 48 hr of incubation (4).
structure, it seemed probable that at least some of the
mutagenic activity of these weak mutagens could be due to
the potent relative, present as an impurity. Also, even
though in almost every case the chemicals used were quite
pure, in most cases only a very small amount of impurity of
the potent relative would be necessary to cause the weak
mutagenic activity.
In all but 2 [4-AAF and B(e)P] of the 7 chemicals exam
ined, we detected some impurity with the retention time
expected for the potent relative [marked with an arrow on
each tracing (A to G) in Chart 1]. Although
no 2-AAF
impurity peak was detected in 4-AAF, the minimum amount
of 2-AAF required to account for the observed mutagenicity
of 4-AAF is so small that it would not have been visible in
the trailing edge of the main peak. The purified sample did
not contain the trailing edge. Although we did not chemi
cally characterize
the impurities,
Chromatographie
runs
were carried out under conditions that varied the retention
times, and in all cases each impurity had the retention time
expected for the potent relative.
In each case the amount of potent relative was estimated
from the tracing, based on its known absorption, and the
percentage of impurity was calculated (Table 1). This rough
estimate seemed to exclude the potent relative as respon
sible for the mutagenic activity of 1-NA, 1-AA, and B(e)P,
since not enough of the potent relative was present to
account for the observed mutagenicity of the weak relative
(Table 1). However, the amount of the potent relative
present in 7-OH-2-AAF, 5-OH-2-AAF, and 2-AB was of the
correct order of magnitude to account for the activity of
these compounds.
For all 7 weak relatives, we collected a sufficient amount
of the purified chemical for a mutagenicity test and com
pared directly, in each case, the impure and pure weak
relatives. In 2 of the 3 cases in which the amount of potent
relative had appeared, from the tracing, to be sufficient to
account for the mutagenicity,
we found the purified weak
relative significantly less mutagenic. (5-OH-2-AAF is difficult
to interpret because the mutagenic activity of the impure
sample is so weak.) For the other 4 weak relatives, there
was no substantial
change in activity upon purification.
434
These results are shown in Table 1.
7-OH-2-AAF was the most dramatic example, losing al
most all mutagenic activity upon purification (Chart 2), and
it serves as an illustration.
If one assumes that the 0.25%
impurity (Chart 1; Table 1) is all due to 2-AAF, then there
was about 0.25 ¡¿g
of 2-AAF present in each 100 ¿¿g
of
unpurified
7-OH-2-AAF tested. From the data shown in
Table 1, one can then calculate that a 2-AAF impurity could
have accounted
for about (0.0025)(250)(4802/10)
= 300
revenants, which is very close, given that this calculation
is only an estimate, to the 436 revertants actually observed
for 250 /¿gof unpurified 7-OH-2-AAF.4 As additional evi
dence that the impurity was most probably 2-AAF, we
collected pure 7-OH-2-AAF and the impurity peak sepa
rately. We found that the mutagenic activity of impure 7OH-2-AAF was restored by addition of the pure impurity
peak to pure 7-OH-2-AAF (data not shown), although limited
amounts of 7-OH-2-AAF prevented a repetition of this exper
iment.
The same kind of calculation may be made for 2-AB. It
appears probable that about half of the mutagenic activity
of unpurified 2-AB was due to a mutagenic impurity, which
is very probably 4-aminobiphenyl.
Pure 2-AB retained a
significant amount of mutagenicity,
and we believe that it
is a weak mutagen.
Chemicals Thought to Require Metabolic Activation for
Mutagenic Activity, Which Were Active Directly [bis-2,7AAF, 7-MB(a)A, and 3-OH-B(a)P]. These 3 chemicals are
carcinogens
(6, 12, 17, 27) and either require metabolic
activation as carcinogens or are closely related to carcino
gens and mutagens that must be activated (1). All 3 chemi
cals showed a significant
amount of mutagenic activity
without activation (Table 2), especially bis-2,7-AAF, which
was extremely
active even after recrystallization
from
ethanol (2, 3). They also contained several impurities (Chart
1). 3-OH-B(a)P, used as a standard in enzyme assays for
aryl hydrocarbon
hydroxylase, contained at least 14 impu
rity peaks, including 2 major impurities (Chart 1). At higher
sensitivity settings we could detect as many as 22 impuri
ties. After purification
on HPLC bis-2,7-AAF and 7-MB(a)A
lost all direct mutagenic activity. Not surprisingly,
since
both are carcinogens (6), they retained mutagenic activity
in the presence of S-9 Mix.
The mutagenic impurity in bis-2,7-AAF is of some interest
because it must be a quite potent mutagen. Since none of
the impurities accounted for more than 1% of the total UVabsorbing material (assuming that the impurities have a
similar extinction coefficient to that of bis-2,7-AAF), only
on the order of 0.1 /¿gof a single mutagenic impurity
would have caused the almost 5000 revertants recorded in
Table 2. Such a potent mutagen may be of theoretical
interest, and in fact new classes of unusually potent muta
gens could be discovered initially as mutagenic impurities.
Because it seemed possible that such a potent impurity
might have also influenced the results of animal carcinogenicity studies done with this compound
(see "Discus
sion"), we also tested a sample of bis-2,7-AAF (kindly
4 This calculation assumes a linear dose-response curve in the low-dose
region, which is in general correct (27) and appeared so in this case in 1
experiment with small amounts of 2-AAF in the presence of excess 7-OH-2AAF, although limited material prevented more thorough analysis. In any
case, the calculation is an estimate of the maximum reversion expected.
CANCER
RESEARCH
VOL. 38
Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1978 American Association for Cancer Research.
Impurities and the Salmonella/Microsome
supplied by E. Weisburger) used by Morris ef al. (30) in
extensive animal carcinogenicity studies. This material
showed no direct mutagenic activity (data not shown) at
dose levels (up to 20 /u.g) at which the material from
Schuchardt was strongly mutagenic (Table 2).
Results for 3-OH-B(a)P were quite erratic from experiment
to experiment, and we cannot conclude whether the muta
genic activity is due to impurities or to 3-OH-B(a)P itself.
The pattern of impurity peaks on HPLC changed markedly
as the injection solution aged over 1 to 2 hr (data not
shown). This suggests that the variable mutagenic activity
could have been caused by oxidation or breakdown prod
ucts of 3-OH-B(a)P or by a co-eluting impurity. The known
oxidation products of 3-OH-B(a)P (40) do not appear to be
mutagenic (41).
3-Methoxy-4-amînoazobenzene. 3-Methoxy-4-aminoazobenzene, a carcinogen (6), was purified on HPLC because
of its unusually great mutagenic potency (467 revertants/
nmole) compared to its close relative 4-aminoazobenzene
(0.29 revertants/nmole) (27). The dose-response curve for
3-methoxy-4-aminoazobenzene was also extremely nonlin
ear (data not shown) (27). Purification on HPLC confirmed
that essentially all of the mutagenic activity was due to 3methoxy-4-aminoazobenzene. The HPLC tracing is shown
in Chart 1, and results of the mutagenicity test are shown
in Table 2.
DISCUSSION
We have reported here on the combined use of HPLC
and the Salmonella /microsome test for rapid purification
and detection of mutagenic impurities in 11 chemicals.
These (most are quite pure already) were selected from
almost 200 mutagens (27), as especially likely (see "Re
sults" for discussion) to contain impurities. We have shown
that 6 of these did in fact contain mutagenic impurities,
and we have discussed each of these under "Results."
Impurities and Mutagenic Potency
We detected mutagenic impurities present in as small
amounts as about 0.1% (Table 1). We also detected an
extremely potent mutagenic impurity (present at about 1%)
when only 10 fj.g of the impure chemical were tested (Table
2). That mutagenic impurities present in such very low
amounts are detected in the Salmonella test is not surpris
ing in view of the extremely wide range of the mutagenic
potency of chemicals [over a million-fold (25-28)] in this
test. As an extreme example, mg dose levels are required
to detect the weakest mutagens, and this defines the usual
upper limit of the amount of a chemical tested in a standard
test protocol (about 10 mg). On the other hand, the most
potent mutagens can be detected at doses as great as
1,000,000 times less, in the ng range. Thus, in an extreme
hypothetical case, a very potent mutagenic impurity, pres
ent in a nonmutagen at as little as 0.0001%, could con
ceivably cause a mutagenic response. And if this were
present as a 1% impurity (not an unusual amount in most
commercial-grade chemicals), it could cause a mutagenic
response when 1 /¿gof the nonmutagenic, unpurified
chemical was tested.
FEBRUARY
Test
The Problem of Mutagenic Impurities in Mutagenesis
Testing
Obviously, in evaluating the role that mutagenic impuri
ties will actually play in mutagenicity testing, of primary
concern is the frequency with which they are likely to
occur and how to recognize the problem when it does
occur. Mutagenic impurities are probably relatively uncom
mon in highly purified commercial chemicals for 2 reasons.
(Industrial chemicals may present more problems and are
discussed separately.) (a) Most of the 136 mutagens for
which we have reported quantitative mutagenicity data (27)
were readily detected at dose levels of less than 10 ^.g, and
almost all (86%) were detected at doses of less than 100
ng. Only 8 mutagens required doses of 1 mg or more for
detection. Thus, one does not usually require high doses
(where impurities would most probably present the greatest
problem) to detect mutagenic activity. This strongly sug
gests that there were very few cases of mutagenic impurities
among these 136 mutagens. (b) A mutagenic impurity must
be quite a potent mutagen to be detected in the test. In our
experience, testing a large number of mutagens, less than
5% of all the mutagens that we have detected are potent
enough to be detected at the 0.1-/ng level (1% of a 10-/¿g
sample). However, about one-third can be detected at or
below the 1-/¿g
range. If this distribution of mutagens along
the potency scale approximates the distribution in the real
world, then when a mutagenic impurity is present (a rarity
to begin with) about 30% of the time it would be detectable
when 100 ^g (for a 1% impurity) of the impure chemical
are tested.
Mutagenic Impurities in Industrial Chemicals
The presence of mutagenic impurities in substances of
high-grade purity is not likely to be a common occurrence.
For industrial chemicals, which often contain fairly large
amounts of impurities, mutagenic impurities may be more
common. The Salmonella test can be used in the design of
industrial syntheses and as a batch process monitor, to
minimize the introduction of mutagenic impurities. At the
Agricultural Division of American Cyanamid Co., Princeton,
N. J., a research compound with potential utility in animals
was identified as weakly mutagenic in Salmonella on
TA1538 and TA98 (R. Gustafson, personal communication).
About 30 samples from a variety of batches were tested,
and a mutagenic impurity was suspected because a few of
the batches were nonmutagenic. It was subsequently found
that the mutagenic activity could be removed from the
active batches by recrystallization. By testing of various
process intermediates, the synthetic step that introduced
the mutagenic impurity was also identified. Introduction of
a charcoal adsorption and filtration step in the synthetic
process removed the impurity. The Salmonella test was
also used to determine that the final batch process chosen
to synthesize this potentially useful chemical did not intro
duce any mutagenic impurities.
At Merck Sharp and Dohme, Rahway, N.J. (P. R. Vagelos,
personal communication), certain batches of an important
chemical were very weakly mutagenic (about 0.007 re
vertants/nmole), and the mutagenic activity was found to be
due to an impurity. The Salmonella test is now being used
1978
Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1978 American Association for Cancer Research.
435
E. V. Donahue et al.
at Merck to ensure that small changes introduced in largescale production do not introduce mutagenic impurities.
The Detection of the Carcinogen 2-AAF as an Impurity in
7-OH-2-AAF by Salmonella but Not by Animal Cancer
Tests
We have shown and have discussed under "Results" that
the weak mutagenicity of a highly purified sample of 7-OH2-AAF is probably due to a small amount of 2-AAF impurity,
which is itself a potent carcinogen. The sample of 7-OH-2AAF that we tested was actually used in thorough animal
tests, which were substantially negative (Ref. 29; E. Weis
burger, personal communication).5 Thus, this may be an
example of the bacterial test being more efficient than the
animal carcinogenicity tests in detecting small amounts of
a potent carcinogen.
A comparison between the amounts of 2-AAF detected
as mutagenic in the bacterial test with the minimum amount
detectable by standard animal cancer tests is instructive.
The amount of 2-AAF in the sample of 7-OH-2-AAF that we
examined here was about 0.25% (Table 1). The carcino
genicity sudies on 7-OH-2-AAF (8, 20, 29) were feeding
studies in which the chemical was fed at concentrations on
the order of 0.06%. Thus, in these studies the 2-AAF im
purity was being fed at about 0.0002% in the diet. This is far
below the amount usually administered in 2-AAF carcino
genicity studies [about 0.03% in the diet (6)], and the size
of a rodent bioassay required to detect the much lower
incidence of tumors that might be expected from a 0.0002%
dose of 2-AAF would be impractically large. The size of the
"megamouse" experiments currently in progress on 2-AAF
at the National Center for Toxicological Research, Jeffer
son, Ark., (32) are illustrative. These studies have been
designed to detect carcinogenicity of 2-AAF in the lowdose range (0.003 to 0.015% in the diet), and more than
24,000 mice are being used, with about 3,000 mice/dose
level tested. The study has been designed to detect a 1%
tumor incidence.
If this kind of sensitivity difference between the bacterial
and animal tests carries over to the detection of weak
carcinogenic activity in general, then this kind of explana
tion may also apply to some other "false positives" in the
Salmonella test. This may actually be the case for some of
the "noncarcinogens" that are weak mutagens in the bac
terial test. These could be weak carcinogens difficult for
the animal carcinogenicity tests to detect. [We have dis
cussed this elsewhere (25, 26).]
Impurities and Animal Carcinogenicity Tests
As discussed above, the bacterial test may be more
sensitive to the presence of mutagenic (or carcinogenic)
impurities than are the animal cancer tests. Although we
know of no case in which a carcinogenic impurity caused a
false positive response in an animal cancer test, there are
some examples where it seems a possibility.
5Two other studies (8, 20) also were substantially negative. These used
preparations of 7-OH-2-AAF that, because of their routes of synthesis, were
likely to contain a greater amount of 2-AAF as an impurity. One of these (20)
actually reported a weak carcinogenic response. However, the increase in
tumors occurred very late in the study, and all controls were killed earlier.
During the time that the controls were alive, the study was negative.
436
When Highly Impure Chemicals Have Been Tested. We
have discussed the highly impure dye auramine in this
respect (25). Recently, auramine has been found to be
weakly mutagenic for Salmonella (26). It would be of inter
est to determine whether the mutagenicity could be due to
an impurity.
When a Chemical Contains an Unusually Potent Impu
rity. bis-2,7-AAF, compared to 2-AAF and its other relatives,
has a very broad tissue specificity and has been referred to
as a ubiquitous carcinogen (6). This is a characteristic
more often associated with direct-acting carcinogens, such
as nitrosomethylurea, than with those requiring metabolic
activation, as in the case of bis-2,7-AAF. We reported here
(Table 2) the presence of a very potent direct-acting muta
genic impurity in 1 sample of bis-2,7-AAF (see "Results"
for further discussion), which could conceivably have influ
enced the cancer tests. As discussed in "Results," another
sample actually used in 1 of the many animal cancer tests
on bis-2,7-AAF did not contain the impurity.
When a Chemical Has Been Administered to Animals in
Unusually Large Amounts. On a per kg of body weight
basis, most carcinogens are active when given to rodents
in daily p.o. doses on the order of mg/day, and some
potent carcinogens can be detected in the /xg/kg dose
range. The weakest carcinogens that have been detected
in animals have been given in daily doses on the order of
g/kg. In such cases a potent carcinogen, present as an
impurity, could conceivably cause a carcinogenic response.
TCE and saccharin are possible examples, and we discuss
these below. [We have discussed 2 other possible exam
ples, thiourea and thioacetamide, elsewhere (25, 26)].
TCE, a major industrial solvent, is also used in an extrac
tion step in the production of decaffeinated coffee, and the
Food and Drug Administration has recently proposed that
this use be discontinued because of a report that it is a
weak carcinogen in mice (10). In this study daily doses of
over 1 g/kg were administered by stomach tube for most of
the lifetime of the animals. The lowest total dose that
produced a response was almost 1 kg. TCE used in the
carcinogenicity tests, although quite pure (>99%), con
tained a number of low-level impurities, including the car
cinogen (19) epichlorohydrin (0.09%) and the possible car
cinogen 1,2-epoxybutane (0.19%). The daily doses of these
2 impurities in the TCE carcinogenicity study were 2.2 to
4.4 mg of epoxybutane and 1 to 2 mg of epichlorohydrin
per kg. These doses do not seem insignificant when com
pared to daily doses of known carcinogens that elicit a
significant tumor response in rodents when administered
p.o. [e.g., propane sultone (<1 mg/kg), nitrosomethylurea
or MNNG (4 mg/kg), and a single 10-mg/kg dose of nitrosoethylurea (22)] .6
Saccharin, the artificial sweetener, causes bladder can* A role for epoxybutane is unlikely. Epoxybutane, although its structure
and weak mutagenicity (27) suggest that it could be a carcinogen, was
negative in 2 quite limited carcinogenicity tests (19). The only conclusion
possible from such limited studies is that, if epoxybutane is a carcinogen, it
is not likely to be very potent and, therefore, would probably not have been
a factor in the TCE carcinogenicity test. Epichlorohydrin, on the other hand,
is a known carcinogen (19) and is a potent mutagen (Ref. 31; V. Simmon,
manuscript in preparation). It cannot be ruled out as a contributor to the
TCE carcinogenicity test result. TCE (containing epichlorohydrin and epox
ybutane impurities) has recently been reported to be weakly mutagenic in
the Salmonella test (36), and it would be of interest to determine whether
highly purified TCE is mutagenic.
CANCER
RESEARCH
VOL. 38
Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1978 American Association for Cancer Research.
Impurities and the Salmonella/Microsome
cer in rats ingesting large doses (more than 2 g/kg/day)
over their lifetimes (7). Based on this finding the Food and
Drug Administration has recommended that saccharin be
banned as a food additive. Saccharin used in these most
recent animal cancer tests, although quite pure, contained
small amounts (about 20 ppm) of as yet unidentified impur
ities. The daily dose of impurities to the rats was about 50
/xg/kg/day. This is quite a low dose, but some of the more
potent rodent carcinogens (such as aflatoxin and 3-methylcholanthrene) can be detected in the standard animal
cancer tests at doses in this range (22). Thus, it is possible
that the carcinogenicity of saccharin could be due to a
potent carcinogenic impurity. This possibility has recently
been given more credence by the finding that impurities in
saccharin are mutagenic in the Salmonella test (Ref. 37;
also see discussion in Ref. 24).
Other Examples of Mutagenic Impurities
Several other chemicals of importance in terms of human
exposure have been reported to contain mutagenic or
carcinogenic impurities, (a) Quinaldic acid, a tryptophan
metabolite, commercially obtained at 98% purity, was rather
strongly mutagenic (26 revertants/nmole; M. Hollstein, R.
Talcott, and E. Wei, personal communication, and Ref. 5),
but after purification it was nonmutagenic (5). (b) The
flame retardant Tris (9) contains the mutagen and known
carcinogen 1,2-dibromo-3-chloropropane. (c) Chlordane, a
pesticide banned by the Environmental Protection Agency
for most uses, but retained for use in termite control,
contains mutagenic impurities (38). (d) Methoxyclor, sug
gested as a substitute for dichlorodiphenyltrichloroethane,
dieldrin, and chlordane (11), contains a weakly mutagenic
impurity (16). It is currently under test for carcinogenicity
at the National Cancer Institute.
ACKNOWLEDGMENTS
We thank E. Weisburger, E. Miller, and J. Miller for chemicals and helpful
discussion; J. Neilands for the use of his Spectra-Physics Chromatograph;
P. Champlin, M. Flynn, and W. Joyce from Waters Associates; and E.
Yamasaki for the dose-response curves for 7-MB(a)A.
ADDENDUM
D. Henschler, E. Eder, T. Neudecker, and M. Metzler (Arch. Toxicol.,
37: 233-236, 1977) have also suggested that impurities could have caused
the carcinogenic effect of trichloroethylene.
REFERENCES
1. Ames, B. N., Durston, W. E., Yamasaki, E., and Lee, F. D. Carcinogens
Are Mutagens; A Simple Test System Combining Liver Homogenates for
Activation and Bacteria for Detection. Proc. Nati. Acad. Sei. U. S., 70:
2281-2285, 1973.
2. Ames, B. N., Gurney, E. G., Miller, J. A., and Bartsch, H. Carcinogens
as Frameshift Mutagens: Metabolites and Derivatives of 2-Acetylaminofluorene and Other Aromatic Amine Carcinogens. Proc. Nati. Acad. Sci.
U. S., 69. 3128-3132, 1972.
3. Ames, B. N., Lee, F. D., and Durston, W. E. An Improved Bacterial Test
System for the Detection and Classification of Mutagens and Carcino
gens. Proc. Nati. Acad. Sei. U. S., 70: 782-786, 1973.
4. Ames, B. N., McCann, J., and Yamasaki, E. Methods for Detecting
Carcinogens and Mutagens with the Sa/mone//a/mammalian-microsome
Test. Mutation Res., 31: 347-364, 1975.
5. Andrews, A. W., Silverman, S. J., and Valentine, C. E. The Identification
of Endogenous and Exogenous Mutagenic Compounds. In: National
FEBRUARY
Test
Cancer Institute Fourth Annual Collaborative Conference, Orlando, Fla.
1976, p. 82. Washington, D. C.: U. S. Department of Health, Education,
and Welfare, 1976.
6. Arcos, J. C., and Argus. M. F. Chemical Induction of Cancer, Vol. 2A
(pp. 30-32 and 158-159) and 2B (pp. 15, 51 and 174-175). New York:
Academic Press. Inc., 1974.
7. Arnold, D. L., Moodie, C. A., Stavric, B., Stoltz, D. R., Grace, H. C., and
Munro. I. C. Canadian Saccharin Study. Science, 197: 320, 1977.
8. Bielschowsky, F. The Carcinogenic Action of 2-Acetylaminofluorene
and Related Compounds. Brit. Med. Bull.. 4: 382-385. 1947.
9. Blum, A., and Ames, B. N. Flame-Retardant Additives as Possible
Cancer Hazards. The Main Flame Retardant in Children's Pajamas Is a
Mutagen and Should Not Be Used. Science, 195: 17-23, 1977.
10. Carcinogenesis Bioassay of Trichloroethylene. NCI Carcinogenesis
Technical Report Series No. 2, 1976.
11. Chem. Eng. News, 53. 15-16, 1975.
12. Cook, J. W., and Schoental, R. Carcinogenic Activity of Metabolic
Products of 3:4-Benzpyrene: Application to Rats and Mice. Brit. J.
Cancer, 6: 400-406, 1952.
13. Dixon, P. F., Gray, C. H., Linn, C. K., and Stoll, M. S. (eds.). High
Pressure Liquid Chromatography in Clinical Chemistry. New York:
Academic Press, Inc., 1976.
14. Freudenthal, R. l., Leber, A. P., Emmerling, D., and Clarke, P. The Use
of High Pressure Liquid Chromatography to Study Chemically Induced
Alterations in the Pattern of Benzofajpyrene Metabolism. Chem.-Biol.
Interactions, 11: 449-458, 1975.
15. Fullerton, F. R., and Jackson, C. D. Determination of 2-Acetylaminofluorene and Its Metabolites in Urine by High Pressure Liquid Chromatog
raphy. Biochem. Med., 16: 95-103, 1976.
16. Grant, E. L., Mitchell, R. H., West, P. R., Mazuch, L., and AshwoodSmith. M. J. Mutagenicity and Putative Carcinogenicity Tests of Several
Polycyclic Aromatic Compounds Associated with Impurities of the Insec
ticide, Methoxychlor. Mutation Res.. 40: 225-228. 1976.
17. Grover, P. L., Sims, P., Mitchley, B. C. V., and Roe. F. J. C. The
Carcinogenicity of Polycyclic Hydrocarbon Epoxides in Newborn Mice.
Brit. J. Cancer, 31: 182-188, 1975.
18. Gutmann, H. R. Isolation and Identification of the Carcinogen W-Hydroxy-2-fluorenylacetamide and Related Compounds by Liquid Chro
matography. Anal. Biochem., 58. 469-478, 1974.
19. Hartwell, J. L. Survey of Compounds Which Have Been Tested for
Carcinogenic Activity, USPHS Publication No. 149. Washington, D. C.:
through 1972-1973.
20. Hoch-Ligeti, C. Effect of Feeding 7-OH-2-Acetaminofluorene to Albino
Rats. Brit. J. Cancer, 7: 391-396, 1947.
21. Holder, G., Yagi, H., Dansette, P.. Jerina, D. M., Levin, W., Lu, A. Y. H.,
and Conney, A. H. Effects of Inducers and Epoxide Hydrase on the
Metabolism of Benzo(a)pyrene by Liver Microsomes and a Reconstituted
System: Analysis by High Pressure Liquid Chromatography. Proc. Nati.
Acad. Sei. U. S., 71: 4356-4360, 1974.
22. lARC Monograph on the Evaluation of Carcinogenic Risk of Chemicals
to Man. IARC (Intern. Agency Res. Cancer) Monographs, 7-72: 19721976.
23. Matsushima, T., Grantham, P. H., Weisburger, E. K., and Weisburger,
J. H. Phenobarbital-Mediated Increase in Ring- and N-Hydroxylation of
the Carcinogen N-2-Fluorenylacetamide, and Decrease in Amounts
Bound to Liver Deoxyribonucleic Acid. Biochem. Pharmacol., 27: 20432051, 1972.
24. McCann, J. Results of a Battery of Short-Term Tests on Highly Purified
Saccharin. In: Cancer Testing Technology and Saccharin, Appendix 2,
pp. 91-108. Washington, D. C.: Office of Technology Assessment,
Congress of the United States, 1977.
25. McCann, J., and Ames, B. N. Detection of Carcinogens as Mutagens in
the Sa/mone//a/microsome Test: Assay of 300 Chemicals. Discussion.
Proc. Nati. Acad. Sei. U. S., 73: 950-954, 1976.
26. McCann, J., and Ames, B. N. The Sa/mone//a/microsome Mutagenicity
Test: Predictive Value for Animal Carcinogenicity. In H. Hiatt, J. D.
Watson, and J. A. Winsten (eds.). Origins of Human Cancer, pp. 14311450. New York: Cold Spring Harbor Laboratory, 1977.
27. McCann, J., Choi, E.. Yamasaki, E., and Ames, B. N. Detection of
Carcinogens as Mutagens in the Sa/mone//a/microsome Test: Assay of
300 Chemicals. Proc. Nati. Acad. Sei. U. S., 72: 5135-5139, 1975.
28. Meselson, M., and Russell, K. Comparisons of Carcinogenic and Muta
genic Potency. In: H. Hiatt, J. D. Watson, and J. A. Winsten (eds.),
Origins of Human Cancer, pp. 1473-1481. New York: Cold Spring
Harbor Laboratory. 1977.
29. Morris, H. P., Velat, C. A., Wagner, B. P., Dahlgard, M., and Ray, F. E.
Studies of Carcinogenicity in the Rat of Derivatives of Aromatic Amines
Related to N-2-Fluorenylacetamide. J. Nati. Cancer Inst., 24: 149-180,
1960.
30. Morris, H. P., Wagner, B. P., Ray, F. E., Stewart, H. L., and Snell, K. C.
Comparative Carcinogenic Effects of W,/V'-2,7-Fluorenylene-Ã-)Ã-'s-Acetamide by Intraperitoneal and Oral Routes of Administration to Rats with
Particular Reference to Gastric Carcinoma. J. Nati. Cancer Inst., 29:
977-1011, 1962.
1978
Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1978 American Association for Cancer Research.
437
E. V. Donahue et al.
31. Mukai, F. H., and Hawryluk, I. The Mutagenicity of Some Halo-Ethers
and Halo-Ketones. Mutation Res.. 21: 228, 1973.
32. Scientific Report, 1974/1975. Jefferson, Ark.: National Center for Toxicological Research.
33. Selkirk, J. K., Croy. R. G., and Gelboin, H. V. Benzo(a)pyrene Metabo
lites: Efficient and Rapid Separation by High-Pressure Liquid Chromatography. Science, 184: 169-171, 1974.
34. Selkirk. J. K., Croy, R. G.. Whitlock. J. P.. Jr., and Gelboin, H. V. In
Vitro Metabolism of Benzo(a)pyrene by Human Liver Microsomes and
Lymphocytes. Cancer Res., 35. 3651-3655, 1975.
35. Separation of Polynuclear Aromatics by Liquid Chromatography. Milford, Mass.: Waters Associates, Inc., 1973.
36. Simmon, V. F., Kauhanen, K., and Tardiff, R. G. Mutagenic Activity of
Chemicals Identified in Drinking Water. In D. Scott, B. A. Bridges, and
F. H. Sobéis(eds.), Progress in Genetic Toxicology, pp. 249-258.
Amsterdam: Elsevier/North-Holland BiomédicalPress, 1977.
37. Stolz, D.. Stavric. B., Klassen, R., Bendali, R. D., and Craig, J. The
Mutagenicity of Saccharin Impurities. I. Detection of Mutagenic Activity.
J. Environ. Pathol. Toxicol.. 1: 139-146. 1977.
38. Tardiff, R. G., Carlson, G. P.. and Simmon, V. F. Halogenated Organics
in Tap Water: Toxicological Evaluation. In: R. L. Jolley (ed.). Proceed
ings of the Conference on the Environmental Impact of Water Chlorina-
438
39.
40.
41.
42.
43.
tion, July 1976, Conference No. 751096, pp. 213-227. Available from
National Technical Information Service, U. S. Department of Commerce,
Springfield, Va.
Thorgeirsson, S. S., and Nelson, W. L. Separation and Quantitative
Determination of 2-Acetylaminofluorene and Its Hydroxylated Metabo
lites by High Pressure Liquid Chromatography. Anal. Biochem., 75:
122-128, 1976.
Weibel, F. J. Metabolism of Monohydroxybenzo(a)pyrenes by Rat Liver
Microsomes and Mammalian Cells in Culture. Arch. Biochem. Biophys.,
768. 609-621, 1975.
Wislocki, P. G., Wood, A. W., Chang, R. L., Levin, W., Yagi, H.,
Hernandez, 0., Dansette, P. M., Jarina, D., and Conney, A. H. Mutagen
icity and Cytotoxicity of Benzo(a)pyrene Arene Oxides, Phenols, Qui
ñones,and Dihydrodiols in Bacterial and Mammalian Cells. Cancer
Res., 36. 3350-3357, 1976.
Yang, S. K., Selkirk, J. K., Plotkin, E. V., and Gelboin, H. V. Kinetic
Analysis of the Metabolism of Benzo(a)pyrene to Phenols, Dihydrodiols,
and Quiñonesby High-Pressure Chromatography Compared to Analysis
by An/I Hydrocarbon Hydroxylase Assay, and the Effect of Enzyme
Induction. Cancer Res., 35. 3642-3650, 1975.
Young, P. R., and McNair, H. M. High-Pressure Liquid Chromatography
of Aromatic Amines. J. Chromatography, ÃŽ79.569-579, 1976.
CANCER
RESEARCH
VOL. 38
Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1978 American Association for Cancer Research.
Detection of Mutagenic Impurities in Carcinogens and
Noncarcinogens by High-Pressure Liquid Chromatography and
the Salmonella/Microsome Test
E. Victor Donahue, Joyce McCann and Bruce N. Ames
Cancer Res 1978;38:431-438.
Updated version
E-mail alerts
Reprints and
Subscriptions
Permissions
Access the most recent version of this article at:
http://cancerres.aacrjournals.org/content/38/2/431
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 16, 2017. © 1978 American Association for Cancer Research.