Proc. Nati. Acad. Sci.
USA
Vol. 86, pp. 9380-9384, December 1989
Genetics
Deletion of stably integrated DNA is suppressed by
cadmium and zinc
(recombination/heavy metals/transfection/mutagenesis)
JOHN M. ABRAMS* AND ROBERT T. SCHIMKEt
Department of Biological Sciences, Stanford University, Stanford, CA 94305
Contributed by Robert T. Schimke, July 20, 1989
ABSTRACT The bacterial xanthine-guanine phosphori-
bosyltransferase (GPT) gene was fused to a metal-responsive
promoter and transfected into a murine cell line. Clonal
transformants harboring metal-responsive or nonresponsive
GPT genes (using a thymidine kinase promoter) were then
studied for the loss of transfected gene function either during
periods of constitutive expression or during periods of induced
activity. Nontoxic levels of cadmium and zinc markedly re-
duced the frequency of mutagenesis in all transfected lines
irrespective of transcriptional status. A survey of 17 GPT-
clones derived from two original transfectants showed partial
or complete excisions of the transfected gene in every case.
These studies show that quantities of cadmium and zinc that
induce metallothioneins also suppress the incidence of deletions
in murine cells.
That deletions are a common form of mutation is clearly
indicated by the number of human defects and cancer-related
mutational events resulting from genetic deficiencies. The
existence of fragile sites (1) and "hot spots" for genetic
rearrangements (2) suggests a nonuniform distribution of
recombination in the eukaryotic genome, which, in part, may
be ascribed to the transcriptional status of a particular genetic
domain (3-5).
Given the capacity for these recombinational episodes to
favor transcribing DNA, we pursued mutagenic studies by
using mouse cell lines (6) that carry the bacterial guanine
phosphoribosyltransferase (GPT) gene under the control of an
inducible metallothionein promoter. Because these lines are
also deficient for hypoxanthine phosphoribosyltransferase
(HPRT), we can score for the loss of transfected GPT gene
function under conditions of induced or constitutive transcrip-
tional activity by using 6-thioguanine (6-TG), a cytotoxic agent
that is selective against guanine salvage (7). From a study of
nine transformed lines, we observed that cadmium and zinc
(Cd/Zn) consistently and sometimes dramatically reduced the
frequency of conversion to the GPT- phenotype by deletion.
This mutagenic suppression occurs regardless of whether or
not the GPT genes are transcriptionally responsive to metals.
We conclude that levels of Cd/Zn that are sufficient to induce
metallothioneins can also suppress the frequency of deletions
at sites of integrated DNA.
MATERIALS AND METHODS
DNA Constructions. pHSIGPT is a plasmid that carries the
bacterial GPT gene immediately downstream of the human
metallothionein I"a promoter and was constructed by inserting
a 1.7-kilobase (kb) Bgl II/BamHI fragment derived from
pRSVGPT (8) into a BamHI-digested plasmid harboring the
human metallothionein Ila promoter region (9). The resulting
The publication costs of this article were defrayed in part by page charge
payment. This article must therefore be hereby marked "advertisement"
in accordance with 18 U.S.C. §1734 solely to indicate this fact.
9380
plasmid contains 840 base pairs (bp) of human metallothionein
Ila 5' DNA (including 70 bp of metallothionein untranslated
sequence) fused to the bacterial GPT gene. TKGPT (obtained
from S. Tilghman, Princeton University) is a similar vector
that positions the bacterial GPT gene and polyadenylylation
signals derived from the bovine growth hormone gene down-
stream from the herpesvirus thymidine kinase (TK) promoter
(10).
Cell Culture and Transfections. Mouse A9 cells (6) carrying
a mutation for the X chromosome-linked HPRT gene were
carried in Dulbecco's modified Eagle's medium supple-
mented with 10% fetal bovine serum and gentamicin (10
mg/ml). Plasmid DNA (5-30 gg) was typically used to
transfect 1 x 106 A9 cells by calcium phosphate coprecipi-
tation without glycerol shock (11); 24 hr later, transfected
cells were placed under selection in medium containing
hypoxanthine (3 mg/ml), aminopterin (400 ,ug/ml), and thy-
midine (2 mg/ml) (HAT). Clonal cell lines, derived either by
dilution or with the use of cloning rings, are referred to in a
manner consistent with the plasmid they carry and were
carried under selective conditions prior to use.
Assay for GPT Activity. GPT enzyme levels were deter-
mined by monitoring the accumulation of xanthine mono-
phosphate from 4 jig of cell extract (12). Protein determina-
tions were carried out as described (13). To quantify GPT
activity, appropriate spots corresponding to the sample of
interest were removed and counted in scintillation fluid.
6-TG Selections. Suicide selections against transfected
GPT activity were performed as described in Results (see
Fig. 1). All mutagenic assays requiring metal induction used
a combination of 1 ,uM CdCl2 and 100 ,uM ZnCl2 unless
otherwise specified. At the time of 6-TG addition (5 ,ug/ml),
three determinations of cell number were averaged from
parallel plates of similarly treated cells. After 9-14 days in
6-TG, resistant colonies were either stained with crystal
violet or expanded for genomic DNA preparations.
Analysis of Genomic DNA and RNA. Genomic DNA was
prepared, blotted, and hybridized as described (13) to a30 ng
(-107 cpm) of a hexamer-labeled GPT-specific BamHI/Bgl II
fragment from pRSVgpt. Cytoplasmic RNA was prepared,
electrophoresed, blotted, and hybridized as described (13) to
a hexamer-primed EcoRI/HindIII metallothionein I gene
fragment (14).
RESULTS
Fig. 1 diagrams our strategy for quantifying the mutability of
transfected GPT genes during periods of induced or consti-
Abbreviations: GPT, bacterial xanthine-guanine phosphoribosyl-
transferase; HPRT, hypoxanthine phosphoribosyltransferase; 6-TG,
6-thioguanine; HAT, hypoxanthine/aminopterin/thymidine; TK,
thymidine kinase.
*Present address: Department of Brain and Cognitive Science,
Massachusetts Institute of Technology, Cambridge, MA 02139.
tTo whom reprint requests should be addressed.
 Genetics: Abrams and Schimke Proc. Natl. Acad. Sci. USA 86 (1989) 9381

Forward over a 4-day period gave cloning efficiency values that

Selection Qi ) A9 cells HATS, 6-TG r differed by only 8% between nontreated and metal-treated
cells (54% without metal exposure and 46% with metal
exposure). Thus, the level of metals used in these experi-
X - Transfect GPT Genes ments does not significantly affect the recovery of colonies
after exposure. At the end of the neutral period (HT medium
with or without metals), parallel plates were harvested for
OE ) - Isolate HAT r tester clones determinations of cell number and GPT activity, while the
_~~~~~~~~~~~~----
remaining plates received new medium supplemented with 5
gg of 6-TG per ml (see Fig. 1). This agent selectively kills cells
Reverse
DAY 0
that can salvage hypoxanthine and guanine for DNA synthe-
Selection - Seed tester clones in HT sis. Mutant cells that have lost GPT activity will replicate and
will thus survive when challenged with this cytotoxic agent.
DAY 1
- Add Cd and Zn In essence, then, this strategy assays the incidence of spon-
taneous GPT mutation during the 4- to 5-day neutral period
k
after release from positive selection (removal of HAT) until
RESPONDERS NON-RESPONDERS they are challenged with negative selection (6-TG).
We performed cell mixing experiments to assess both the
stringency and fidelity of this 6-TG selection protocol. Our
results showed that the recovery of hprt- cells within a vast
metals control metals control excess of dying wild-type cells is highly efficient (76-84%).
Other controls (not shown) confirmed that resistance to 6-TG
DAY 5
and growth in HAT are mutually exclusive phenotypes
- Count and harvest parallel plates (6-TG-resistant variants not representing a deficiency of GPT
-Add 6-TG to remaining cultures activity, if they occur at all, must arise at a frequency of
<1000).
To demonstrate that heavy metals are sufficient to produce
inductive conditions in these cells, cytoplasmic RNA from
Mutant cells HAT s, 6-TG r
one transfected clone (HSIgpt clone I) was analyzed for
endogenous metallothionein synthesis after exposure to
these metals. After 28 hr of exposure to a combination of 1
DAYS 14- 19 ,4M Cd and 100 ,uM Zn, we observed a 12-fold induction of
metallothionein mRNA (Fig. 2) with a size range consistent
- Count # 6-TG r colonies with those that have been previously reported (15). In con-
FIG. 1. Schematic diagram of experimental strategy. Forward
selection of a murine HPRT- line in HAT medium identifies GPT
transformants. After clonal transformed lines were characterized,
each was subjected to reverse selection against GPT. Parallel plates
were seeded in HT medium at day 0. At 24 hr, heavy metals were
added to experimental plates. At 48 hr, the HT medium of all plates
was appropriately replaced with metals in experimental plates or
without metals in control plates. This medium change was repeated
at % hr. At 120 hr the medium was again replaced with fresh medium
containing the negative selecting agent 6-TG. Mutagenic episodes
were scored 9-14 days later by the incidence of 6-TGr colonies.
Graphic symbol represents GPT activity in responder (metal induc-
ible) and nonresponder cell lines. Italics refer to selective pheno-
types. HATS, HAT sensitive.
tutive genetic activity. Clonal transfectants harboring metal-
responsive (HSIgpt) or nonresponsive GPT constructs (driv-
en by the herpes TK promoter, TKgpt) were initially seeded
in parallel plates at predetermined densities (typically 1 x 103 -l700 bp
cells per cm2) in medium supplemented with hypoxanthine - 450 bp
and thymidine but lacking aminopterin (HT medium). Re-
lease from HAT selection requires a period of weaning in HT
medium to allow for sufficient salvaging of nucleotides while
residual levels of aminopterin that would otherwise prevent
immediate resumption of de novo pathways are cleared.
Removal of these transformants from the forward selection
for GPT is thus generally accomplished over a 4- to 5-day
period (see Fig. 1).
Twenty-four hours after plating cells in HT, heavy metals 1 2
were added to experimental plates and, twice more during
this nonselective neutral period of growth in HT, the medium FIG. 2. Induction of murine metallothionein I RNA in a clonal
was replaced with or without a heavy metal supplement (see GPT transformant. Cytoplasmic RNA (30 ,ug) from control (lane 1)
or metal-induced (lane 2) HSIgpt clone I cells was electrophoresed,
Fig. 1 and legend). Parallel cultures of clonal transformants blotted onto a nylon membrane, and probed with a labeled metal-
were thus grown in HT medium for 5 days and challenged (or lothionein I-specific fragment. Cells were harvested 28 hr after heavy
not) with heavy metals for 4 days. In control experiments, metal treatment and show a 12-fold induction of metallothionein I
cells seeded in HT medium-and metal-treated in this fashion RNA. The position of relevant size markers is shown.
9382 Genetics: Abrams and Schimke Proc. Natl. Acad. Sci. USA 86 (1989)
CONTROL METALS transfected clones. For each mutagenic determination, par-
allel plates were assayed in the presence or absence of heavy
metals for determinations of both cell number and GPT
activity at the time of 6-TG addition. The number of cell
doublings during the neutral period (nonselective HT me-
dium) was derived for each clonal line according to the
1,f U.,
formula noted in the table legend. Although heavy metals do
not significantly affect proliferation in about half of the clones
examined, the growth of some clones is retarded by these
ions (e.g., clones HSIgpt E and HSIgpt N). The incidence of
mutant colonies divided by the number of cells that were
challenged with 6-TG is presented as 6-TGr frequency in
Table 1. Spontaneous rates of GPT inactivation ranged over
3 orders of magnitude, from 2 to 2000 mutants in 106 cells. For
all clonal lines examined, it is consistently evident that
exposure to heavy metals during the neutral period signifi-
cantly lowers the incidence of 6-TGr mutants. The extent of
this suppression is numerically reflected as the suppression
ratio, which varies from 2- to 14-fold (Table 1). Because many
clones show doubling numbers (Table 1) that are unaffected
or only slightly influenced by the presence of metals, the
suppressed mutagenic frequencies are not attributable to
direct effects on cell proliferation nor are they attributable to
FIG. 3. Colony formation of 6-TGr GPT mutants from three affect on colony formation (pHSIgpt clone I shows 54%
clonal HSIgpt lines. Note the suppressed incidence of mutant cloning efficiency in the absence of metals and 46% plating
colonies after exposure to metals in every case. Top row, HSIgpt efficiency after metal exposure). Repeat selections (given in
clone T; middle row, HSIgpt clone I; bottom row, HSIgpt clone V.
parentheses in Table 1) show that suppression of 6-TG
trast, the inducibility of the transfected constructs (see Table resistance after preexposure to metals is reproducible, al-
1) was always far less dramatic than that observed for the though the absolute values for mutagenic frequencies can
vary by moderate amounts. Furthermore, although GPT
endogenous metallothionein I gene (Fig. 2). This discrepancy activity is not inducible in clones HSIgpt E, TKgpt A, and
might result either from the heterologous nature of the TKgpt E, mutation frequencies of these genes are also
metallothionein promoter itself (human regulatory sequences substantially reduced by exposure to heavy metal ions. These
in a mouse background) or from integration site effects that findings demonstrate that mutagenic episodes at transfected
tend to preclude regulation. loci can be suppressed by extracellular metals irrespective of
Fig. 3 shows the results from one set of reverse selections whether any transcriptional induction had occurred.
after exposure to conditions that are constitutive or inducing Table 1 also gives estimated gene dosages for six of the
for the transfected genes. For each clonal line shown, the clones based on Southern blot analyses of genomic DNA
incidence of spontaneous 6-TG-resistance (6-TG0 mutants is digested with an enzyme that cleaves once in the transfected
far less among those cells that had been exposed to heavy plasmid. The hybridization pattern of one clone, HSIgpt T,
metals during the neutral period. Table 1 compiles the indicates that it carries a tandem array of 300-500 copies of
regulatory and mutagenic data from similar assays of nine the plasmid, probably arranged in head-to-tail configuration
Table 1. Clonal expression of GPT transformants and suppression of 6-TG resistance by heavy metals
Cell
GPT Basal Induction, doublings 6-TGr frequency Suppression
Clone gene, no. activity -fold - + - + ratio
HSIGPT clones
HSIgpt I 2-4 52 3.5 4.8 4.9 1.9 x 10-4 3.3 x 10-5 5.8
(HSIgpt I) 4.4 4.2 3.1 x 10-4 8.1 x lo- 3.8
HSIgpt T 300-500 119 2.4 4.2 4.1 7.2 x 10-' 1.4 x 10-5 5.1
HSIgpt B 2-3 69 4.5 3.0 3.1 5.4 x 1o-4 1.6 x 10-4 3.3
HSIgpt V 7-10 79 3.9 4.7 3.9 3.2 x 10-4 1.7 x 10-4 1.9
HSIgpt U 2-4 24 3.5 2.9 2.8 3.1 x 10-4 4.8 x 10-5 7.1
HSIgpt N 3 27 2.9 2.6 1.7 2.1 x 10-3 2.5 X 10-4 8.4
HSIgpt E* ND 20 1.2 3.1 1.5 3.0 x 10-5 2.1 x 10-6 14.3
TKGPT clones
TKgpt A ND 35 0.9 2.7 2.2 5.3 x 10-4 1.4 x 10-4 3.8
TKgpt E ND 37 1.2 3.6 3.1 1.4 x 10-5 3.9 x 10-6 3.6
Gene dosages were estimated from EcoRV restriction profiles and densitometry of genomic DNAs probed with
hexamer-labeled GPT DNA. Basal level of GPT enzyme activity is given in units of product (XMP) per min per 100 ng of
protein. Induction (-fold) is calculated by dividing the basal GPT activity into the activity levels induced by metals. The
number of cell doublings during the period of growth in HT medium without metals (-) or with metals (+) was derived by
solving the formula 2x = Y, where Y is the -fold increase in cell number during the mutagenic period (in HT medium) and
X is the number of cell doublings during this same period. 6-TGr frequencies are shown of control cultures (-) or of cultures
exposed to metals during the mutagenic period (+). Suppression ratio indicates ratio of 6-TGr frequency in the absence of
metals divided by the frequency observed in the presence of metals. HSIgpt clone I in parentheses refers to a second
mutagenic trial. ND, not determined. GPT activity in this clone was assayed in the presence of 5 mM CdCl2. In all other
cases, + refers to a supplemental mixture of 1 mM CdCl2 and 100 mM ZnCI2, whereas - refers to no addition.
 Genetics: Abrams and Schimke Proc. Natl. Acad. Sci. USA 86 (1989) 9383

HSI gpt T HSlgpt B

V 2 3 4 5 6 7 8 9 10 11 12 13 V 14 15 16 17 18

*_i
5.2 kb
-S
I.

FIG. 4. Analysis of GPT genes from a total of 17 6-TGr mutants derived from 2 HSIgpt clones. Genomic DNA from HSIgpt clone T or HSIgpt
clone B is denoted by arrowheads. These DNAs and similar preparations from 12 HSIgpt clone T-derived mutants (lanes 2-13) and 5 HSIgpt
clone B-derived mutants (lanes 14-18) were digested with EcoRV, electrophoresed, and then stained in ethidium bromide to confirm equal DNA
quantitations. The blotted membrane was hybridized with a GPT-specific probe. HSIgpt clone T-derived mutants in lanes 9-13 were selected
after metal exposure during the neutral period, whereas lanes 2-8 represent HSIgpt clone T mutants selected without metal exposure during
this period. All HSIgpt clone B mutants shown (lanes 14-18) were selected without metal exposure. The migration of a unit length 5.2-kb plasmid
is indicated. Note that all mutants are complete or partial gpt gene deletions.
(see Fig. 4). Given that the levels of transforming enzyme in or partial deletions can account for the loss of GPT activity.
HSIgpt clone T are not inordinately high relative to cell Deletions thus constitute a predominant mode of spontane-
clones harboring much lower gene dosages (e.g., HSIgpt I, ous mutation at transfected sequences not only in these
HSIgpt B), it is likely that most of these genes are not transformed cells but in other culture systems as well (7, 17,
functional. 18). In contrast, studies of spontaneous mutagenesis at
A total of 17 mutants derived from two original transfec- hemizygous or X-chromosome linked genes have yielded
tants (HSIgpt B and HSIgpt T) were analyzed to determine widely varying results with regard to the nature of genetic
the nature of the genetic lesions responsible for 6-TG resis- alteration. Whereas 83% of spontaneous tk- lesions surveyed
tance (Fig. 4). Al S mutants derived from HSIgpt B no longer were shown to carry complete deletions of the functional
carry any evidence of the transfected gene (signal in lane 18 allele (21), similar studies at the adenine phosphoribosyl-
is nonspecific background hybridization). Twelve mutants transferase (aprt) (19) and hprt (22) loci found a predom-
derived from HSIgpt clone T were analyzed, 5 of which were inance (67% and 90%, respectively) of base substitutions
exposed to metals during the neutral period and 7 of which among the total spontaneous mutants scored. This variable
were selected without preexposure to metals. Whereas 2 propensity for deletion events among endogenous genes may
mutants selected without metal preexposure have deleted the reflect variability among flanking sequence structures that
entire array of transfected genes, 5 of the 7 HSIgpt T mutants can influence recombinagenic potential. Hence, the deletion
selected under these conditions still retain a small fraction bias of transfected DNA could result from preferential in-
(1-25%) of the transfected sequences. The partial genetic sertions within highly recombinagenic domains.
remnants of this tandem array that remains in these 5 mutants An unexpected and intriguing result of our studies con-
must be either silent or mutated to allow for growth in 6-TG. cerns the general suppression of 6-TG resistance after expo-
In contrast, all 5 of the mutants selected under conditions of sure to heavy metals. This phenomenon is not attributable to
metal preexposure underwent complete deletions for the differential proliferation rates because most of the surveyed
entire array of transfected genes (Fig. 4). Thus, from a total clones show little or no effects with regard to cell doubling
of 17 mutants analyzed, all clones show evidence for major numbers when exposed to these ions (Table 1). Additional
deletions of the transfected gene; 13 have undergone com- controls demonstrated that the concentration of metals used
plete deletions and 5 have deleted most of the genes from in these experiments also has little if any effect on cloning
within a large tandem array that likely includes all functional efficiency (see Results). We thus conclude that the deletion
templates within this transfected domain. The remaining of transfected genes is suppressed by heavy metals. Given
nonfunctional genes in these 5 mutants might carry identical these findings, we are currently using mouse L cells in a
small alterations (e.g., point mutations) that occurred either similar protocol to examine the effects of metals on the
before or after stable integration (16) or, alternatively, they mutagenic propensity of an endogenous locus (HPRT).
may have been silenced by epigenetic determinants such as Although the metal treatment used during these experi-
methylation (17). ments is highly inductive for endogenous metallothionein
RNA (Fig. 2), the mechanism(s) whereby Cd and Zn mediate
DISCUSSION a suppression of deletions is not clear. One possible expla-
nation might involve suppressive effects on the enzymes that
The elevated mutability of transfected genes, as illustrated by mediate spontaneous rearrangments. Metal-induced inhibi-
these experiments, has been previously observed in a wide tion of recombination/deletion may thus constitute a protec-
variety of cell culture systems using various recombinant tive "stress" response that both facilitates DNA repair and
genes (7, 17, 18). Consistent with these reports, we also find avoids strand-exchange processes that might otherwise be
that losses of transfected gene function (occurring at frequen- lethal. Consistent with this line of reasoning, a variety of
cies of 103-106) are far more common than lesions at native pretreatments (Mn, Zn, and surgical lesions) that induce
loci (which reportedly arise at frequencies of 105-108; see metallothionein proteins are reported to enhance the surviv-
refs. 19 and 20). For all 17 mutants analyzed here, complete ability of irradiated mice (23) and increase the tolerance of
9384 Genetics: Abrams and Schimke
cultured cells to a variety of DNA-damaging agents (24). In
light of our results, these reports suggest that one parallel
consequence of elevated metallothioneins in vivo or in cul-
tured cells is a greater capacity to tolerate DNA damage via
some effect on recombination machinery.
Alternatively, the ability of metals to suppress the excision
of transfected DNA may relate to subtle effects on the control
of proliferation in cells that are adapted for growth in culture.
Cellular transformation is well correlated with higher fre-
quencies of DNA amplification (25) and these events may, in
fact, constitute only one possible end product of the same
molecular processes that mediate deletions and other chro-
mosomal rearrangements (26, 27). A relatively unstable ge-
nome is indeed one prevalent property associated with a
classically transformed phenotype (28) and perhaps it is this
condition that is at least partially reversed by Cd/Zn. Thus,
an alternative approach toward understanding the suppres-
sion of deletions by metals focuses on the connection be-
tween genetic malleability and the control of cell cycle
progression.
1. Laird, C., Jaffe, E., Karpen, G., Lamb, M. & Nelson, R. (1987)
Trends Genet. 3, 274-281.
2. Roeder, G. S., Keil, R. L. & Voelkel-Meiman, K. A. (1986) in
Current Communications in Molecular Biology: Mechanisms
of Yeast Recombination, eds. Klar, A. & Strathern, J. (Cold
Spring Harbor Lab., Cold Spring Harbor, NY), pp. 29-33.
3. Keil, R. L. & Roeder, G. S. (1984) Cell 39, 377-386.
4. Thomas, B. J. & Rothstein, R. (1989) Cell 56, 619-630.
5. Yancopoulos, G. D. & Alt, F. W. (1986) Annu. Rev. Immunol.
4, 339-368.
6. Watson, B., Gormley, I. P., Gardener, S. E., Evans, H. J. &
Harris, H. (1972) Exp. Cell Res. 75, 401-409.
7. Tindall, K. R., Stankowski, L. F., Machanoff, R. & Hsie,
A. W. (1984) Mol. Cell. Biol. 4, 1411-1415.
8. Gorman, C. M., Merlino, G. T., Willingham, M. C., Pastan, I.
& Howard, B. H. (1982) Proc. NatI. Acad. Sci. USA 79,
6777-6781.
Proc. Natl. Acad. Sci. USA 86 (1989)
9. Karin, M., Haslinger, A., Holtgreve, H., Cathala, G., Slater, E.
& Baxter, J. D. (1984) Cell 36, 371-379.
10. McKnight, S. L. (1980) Nucleic Acids Res. 8, 5949-5964.
11. Davis, L. G., Dibner, M. D. & Battey, J. F. (1986) Basic
Methods in Molecular Biology (Elsevier, New York), pp.
286-289.
12. Chu, G. & Berg, P. (1985) Nucleic Acids Res. 13, 2921-2930.
13. Abrams, J. M., Thorpe, S. M. & Schimke, R. T. (1989) J. Biol.
Chem. 264, 14016-14021.
14. Searle, P. F., Davison, B. L., Stuart, G. W., Wilkie, T. M.,
Nordstedt, G. & Palmiter, R. D. (1984) Mol. Cell. Biol. 4,
1221-1230.
15. Fornace, A. J., Schalch, H. & Alamo, I. (1988) Mol. Cell. Biol.
8, 4716-4720.
16. Roberts, J. M. & Axel, R. (1982) Cell 29, 109-119.
17. Gebara, M. M., Drevon, C., Harcourt, S. A., Steingrimsdottir,
H., James, M. R., Burke, J. F., Arlett, C. F. & Lehman, A. R.
(1987) Mol. Cell. Biol. 7, 1459-1464.
18. Thacker, J., Debenham, P. G., Stretch, A. & Webb, M. (1983)
Mutat. Res. 111, 9-23.
19. Nalbantoglu, J., Phear, G. & Meuth, M. (1987) Mol. Cell. Biol.
7, 1445-1449.
20. Thacker, J., Stretch, A. & Brown, R. (1982) Mutat. Res. 103,
371-378.
21. Yandell, D. W., Dryja, T. P. & Little, J. B. (1986) Somatic Cell
Mol. Genet. 12, 255-264.
22. Fuscoe, J. C., Fenwick, R. G., Ledbetter, D. H. & Caskey,
C. T. (1983) Mol. Cell. Biol. 3, 1086-1096.
23. Matsubara, J., Tajima, Y. & Karasawa, M. (1987) Mutat. Res.
111, 267-275.
24. Endreson, L. & Rugstad, H. E. (1987) Experientia Suppl. 52,
595-602.
25. Otto, E., McCord, S. & Tisty, T. D. (1989) J. Biol. Chem. 264,
3390-3396.
26. Schimke, R. T. (1988) J. Biol. Chem. 263, 5989-5992.
27. Carroll, S. M., DeRose, M. L., Gaudray, P., Moore, C. M.,
Needham-Vandevanter, D. R., Von Hoff, D. D. & Wahl,
G. M. (1988) Mol. Cell. Biol. 8, 1525-1533.
28. Nowell, P. C. (1982) in Tumour Cell Heterogeneity, Origins
and Implications, eds. Owens, A. H., Coffey, D. S. & Baylin,
S. B. (Academic, New York), pp. 351-369.