carc$$0101
Carcinogenesis vol.18 no.1 pp.121–125, 1997
Radon and lung carcinogenesis: mutability of p53 codons 249 and
250 to 238Pu α-particles in human bronchial epithelial cells
S.Perwez Hussain, Christopher H.Kennedy1,
Paul Amstad2, H.Lui2, John F.Lechner1 and
Curtis C.Harris3
Laboratory of Human Carcinogenesis, NCI, NIH, Bethesda,
MD 20892-4255, 1ITRI, Lovelace Biomedical and Environmental Research
Institute, Albuquerque, NM 87185 and 2Department of Pathology,
University of Maryland, Baltimore, MD 21201, USA
3To
whom correspondence should be addressed
Radon-222, a decay product of uranium-238 and a source
of high linear energy transfer (LET) α-particles, has been
implicated in the increased risk of lung cancer in uranium
miners as well as non-miners. p53 mutation spectrum
studies of radon-associated lung cancer have failed to show
any specific mutational hot spot with the exception of a
single study in which 31% of squamous cell and large cell
lung cancers from uranium miners showed a p53 codon
249 AGGarg → ATGmet mutation. Although the results of
laboratory studies indicate that double-strand breaks and
deletions are the principal genetic alterations caused by αparticles, uncertainty still prevails in the description of
DNA damage in radon-associated human lung cancer. In
the present study, we have evaluated the mutability of p53
codons 249 and 250 to α-particles in normal human
bronchial epithelial (NHBE) cells using a highly sensitive
genotypic mutation assay. Exposure of NHBE cells to a
total dose of 4 Gy (equivalent to ~1460 working level
months in uranium mining) of high LET α-radiation
induced codon 249 AGG → AAG transitions and codon
250 CCC → ACC transversions with absolute mutation
frequencies of 3.6 3 1027 and 3.8 3 1027 respectively. This
mutation spectrum is consistent with our previous report
of radon-associated human lung cancer.
Introduction
Mutation spectrum studies have revealed the presence of
characteristic patterns of DNA alteration induced by exogenous
and endogenous mutagens in cancer-related genes (1). p53, a
tumor suppressor gene which is mutated in about half of all
human cancers, shows specific mutation patterns for a number
of carcinogens (reviewed in 1,2). High levels of dietary
aflatoxin B1 are correlated with G:C → T:A transversion and
a serine substitution at residue 249 of p53 in hepatocellular
carcinoma (3–7); exposure to UV radiation is correlated with
transition mutations at dipyrimidine sites (8); cigarette smoking
is positively correlated with G:C → T:A transversions in lung
cancer (9). Recently, efforts have been made to define the p53
mutation spectrum in lung cancers from uranium miners (10–
13). The p53 mutation spectrum in lung cancer associated with
either radon exposure or tobacco smoking is presented in
*Abbreviations: LET, linear energy transfer; NHBE, normal human bronchial
epithelial; WLM, working level months; PACs, phenotypically altered cells;
MS, mutant standard.
© Oxford University Press
Figure 1. Taylor and colleagues (11) have reported the presence
of an AGGarg → ATGmet transversion at codon 249 of p53 in
31% of large cell and squamous cell cancers from uranium
miners. However, p53 mutation analyses in adenocarcinomas
by McDonald et al. (13), from the same cohort used by Taylor
et al. (11) and different lung cancers from uranium miners by
Vahakangas et al. (10) and Bartsch et al. (12), did not find
any cases with a codon 249 G → T mutation.
α-Particles, a type of high linear energy transfer (LET*)
radiation, are emitted by the short-lived radon-222 progeny
polonium-218 and polonium-214. Epidemiological studies
have implicated a high level of radon-222, a decay product of
uranium-238, in the high incidence of lung cancer among
uranium miners (14,15). A large proportion of available
literature on the biological lesions produced by α-particles
suggests that double-strand breaks, large deletions and sister
chromatid exchange are the major abnormalities (16–18).
In order to gain insight into the mutability of codons 249
and 250 of the p53 gene to α-particles, we have used a highly
sensitive genotypic mutation assay (19,20) to analyze αparticle-exposed normal human bronchial epithelial (NHBE)
cells.
Materials and methods
Cell culture
NHBE cells from a 15 year old male never smoker (strain 2129) and bronchial
epithelial cell growth medium were purchased from Clonetics (San Diego,
CA). These cells were determined to be negative for adenovirus, hepatitis
type B virus, human immunodeficiency virus, human papilloma virus and
mycoplasma (21). The cryopreserved cells (passage 1) were cultured and
expanded in plastic flasks (Corning, NY) pre-coated with FNC Coating Mix
(BRFF, Ijamsville, MD) (22).
Irradiation and post-irradiation cell culture
Confluent NHBE cells (passage 2), cultured on 1.5 µm thick Mylar film
(surface area 8 cm2), were exposed over a 17 day period to a total dose of
either 0 (unexposed control) or 4 Gy [equivalent to 1460 working level months
(WML)] of α-particles delivered in six equal, fractionated doses using a
stainless steel disc electroplated with sufficient 238Pu to provide 0.857 Gy/
min of α-particle energy (21). Culture medium was replenished three times a
week. The exposed cells capable of proliferation were allowed to grow to
confluence after each exposure (48–72 h) before initiating the next irradiation.
Starting 1 week after the final exposure, cultures of unexposed controls and
of samples exhibiting foci of phenotypically altered cells (PACs), were
subcultured to plastic tissue culture dishes pre-coated with FNC Coating Mix
(22). Cells were then passaged and split into several dishes as the culture
became confluent. This was repeated until the cultures underwent senescence.
Analysis of p53 codons 249 and 250 for base pair changes by HaeIII
RFLP-PCR
Cells from unexposed NHBE cultures (passage 3–5) and PAC cultures (passage
4–6) were pooled for the isolation of control and exposed DNA samples
respectively. DNA preparations containing 1.5 3 107 copies of p53 were
enriched in sequences with mutated HaeIII site spanning residues 14072–
14075 (codons 249 and 250) by exhaustive HaeIII restriction digestion and
size fractionation. Ten copies of an internal mutant standard (MS) were added
before gel isolation of a 100–200 bp fragment population containing a mutated
159 bp p53 segment, which extends from flanking 59 HaeIII site (residue
13981) to the flanking 39 HaeIII site (residue 14139) plus MS.
From the enriched DNA preparations described above, a 116 bp fragment
spanning residues 13999–14114 was amplified with Pyrococcus furiosus
121
S.P.Hussain et al.
Fig. 1. p53 mutation spectrum from lung cancer associated with radon exposure and tobacco smoking. (The pi diagram has been drawn using a p53 mutation
database [Hollstein et al. (34)] and is available on the Internet at http://www.nci.nih.gov/intra/LHC/LHCpage.htm.)
(Stratagene) DNA polymerase for 40 cycles. A 101 bp exon VII fragment
extending from residue 13999 to 14099 containing codons 249 and 250 was
then amplified using nested primers with EcoRI tails and Taq polymerase
(Perkin Elmer) in an additional 10 amplification cycles.
The RFLP-PCR products were cloned into λgt10 and the phage were plated
on Escherichia coli C600 Hfl. For each sample ~1000–1500 plaques on
different plaque screens were hybridized with 32P-labeled oligonucelotide
probes specific for each single base pair mutation, the wild-type and MS.
Selective washing temperatures were determined with λ constructs containing authentic mutant, wild-type and MS inserts. In each experiment
authentic mutant constructs were included to ascertain the selective hybridization condition. The absolute mutation frequencies were estimated from the
MS content of the RFLP-PCR product, the initial number of MS copies and
the number of copies of the p53 gene present at the time of the addition of
MS to the cellular DNA. (For a detailed description of the experimental
procedures see 19,20.)
Results
We have studied the mutability of codons 249 and 250 of the
p53 gene to α-particles in NHBE cells. Codons 249 and 250
fall in the HaeIII enzyme recognition site and are thus
accessible to our highly sensitive genotypic mutation analysis
system using the RFLP-PCR approach. Sequence information,
the location of the selected HaeIII site (residues 14072–14075),
the chosen amplimers and the structure of MS have been
described previously (19).
NHBE cells were exposed to fractionated doses of αparticles to achieve a total dose (4 Gy, ~1460 WLM) that is
comparable with the total occupational exposure received by
uranium miners. Irradiation resulted in the formation of foci
of PACs, which are characterized by morphologically distinct
cells with a high mitotic index. These cells also exhibited an
extended lifespan relative to unexposed controls, however,
none of the PACs acquired an indefinite population doubling
potential (21). Proliferating cultures of these PACs and unexposed controls were used to isolate DNA for analysis of p53
mutations.
Identification of specific mutant λ plaques was performed
by selective oligonucleotide hybridization (Figure 2). The
DNA that was exhaustively restricted with HaeIII contained
1.5 3 107 initial p53 copies from untreated control and α122
particle-treated cells alike. Samples of 1400–1500 plaques
each for control and treated samples on 10 Petri dishes were
analyzed by hybridization with 12 mutant-specific oligonucleotide probes as well as probes for wild-type sequence and MS.
Figure 2 shows the identified mutant λ plaques from only one
Petri dish each of exposed and non-exposed samples. Lambda
constructs, containing inserts with a single base pair mutation
in the restriction site of interest containing codons 249 and
250 were used as positive controls to ascertain the specificity
of the hybridization conditions in each filter analysis of an
RFLP-PCR product. Figure 3 shows the absolute mutation
frequencies at codons 249 and 250 in DNA from α-particleexposed and non-exposed NHBE cells. Exposure to α-particles
selectively induced a G (residue 14072) → A transition at the
second base of codon 249 and a C (residue 14074) → A
transversion at the first base of codon 250 with absolute
mutation frequencies of 3.6 3 1027 and 3.8 3 1027 respectively. We did not observe any G → T transversions in the
analyzed sequence. The untreated control sample did not
show any detectable background mutation. Absolute mutation
frequencies were calculated from the MS content of the RFLPPCR products, the initial number of MS copies and the number
of copies of the p53 gene presented at the time of the addition
of MS to the cellular DNA.
Discussion
A transition G → A at codon 249 (AGG → AAG) and a
transversion C → A at codon 250 (CCC → ACC) were
observed only in α-particle-exposed NHBE cells with absolute
mutation frequencies of 3.6 3 1027 and 3.8 3 1027 respectively
(Figures 2 and 3). We did not observe any G → T transversions,
including that reported by Taylor et al. (11), at the middle
base of codon 249 in 31% of squamous cell and large cell
lung cancers in uranium miners.
The presence of a G → A transition and a C → A
transversion and the absence of a G → T transversion in our
study are consistent with that of Vahakangas et al. (10), who
reported the presence of three C → A transversions and one
p53 mutability with
238Pu
Fig. 2. Identified mutant λ plaques for G → A transition and C → A transversion. 1400–1500 plaques each for control and treated samples on 10 Petri dishes
were analyzed by hybridization with 12 mutant-specific oligonucleotide probes, as well as probes for wild-type sequence and MS. Authentic mutant λ
constructs with mutated inserts were used as positive controls.
G → A transition and the absence of any G → T transversion
among the seven confirmed point mutations in lung cancer
from uranium miners. A recent study by Bartsch et al. (12)
have reported the absence of a codon 249 AGG → ATG
transversion in lung cancers from uranium miners in Saxony,
Germany, having almost the same cumulative radon exposure
as the miners from the Colorado plateau described in the
study of Taylor et al. (11). Another study, using 23 lung
adenocarcinomas from the same cohort used by Taylor and
colleagues (11), failed to show a codon 249 AGG → ATG
transversion, suggesting the possibility of a specific association
of codon 249 mutation with only squamous cell and large cell
carcinoma of the lung (13). However, nine squamous cell
carcinomas in the study by Bartsch et al. (12) and six
squamous cell carcinomas and two large cell carcinomas in
the Vahakangas et al. (10) studies failed to show this mutation.
It has also been proposed (23) that the presence of mycotoxins such as sterigmatocystin, a precursor of aflatoxin, in the
damp mines could be the basis of the codon 249 AGG → ATG
mutation in the study by Taylor et al. (11). Although aflatoxin
B1 and other carcinogens that form bulky DNA adducts are
expected to produce G → T transversions in general, aflatoxin
B1 preferentially induce an AGGarg → AGTser mutation at
codon 249 of p53 (19), which was observed only in one of
41 squamous cell carcinomas in the study by Taylor and
colleagues (11).
DNA damage caused by high LET α-particles predominantly
involves double-strand breaks and large deletions, but is also
reported to produce point mutations (24,25). One frequently
found mutation associated with α-particles is a G → A
transition. This base substitution was frequently found in codon
12 of the K-ras oncogene of lung tumors of rats that inhaled
plutonium (26). The hyperplastic lesion and adenomas in the
lungs of these animals also exhibited the same mutation,
suggesting that it arose early in the carcinogenic process.
Irradiation of human shuttle plasmid pZ189, containing the
E.coli supF gene as the mutational target, with α-particles
induced a high frequency of G → A transitions (25). Thus, it
is most likely that G → A transition is a consequence of αparticle exposure.
Because the primary NHBE cells used in this study were
Fig. 3. Absolute mutation frequencies of α-particle-induced base pair
changes in codons 249 and 250 of p53 in NHBE cells. Absolute mutation
frequencies were calculated from the percentage of the identified mutant λ
plaques by calibration with the MS contents of the RFLP-PCR products, the
initial number of p53 gene copies and the number of copies of mutant
standard added at the outset of the experiment (19,20).
from a single donor, the p53 mutations detected are not
necessarily representative of what would be found if NHBE
cells from several donors were irradiated and analyzed. Since
humans exhibit interindividual susceptibility to the development of cancer (27), it is predicted that some individuals will
show a higher level of unrepaired mutations after irradiation
relative to other individuals. This type of interindividual
variation was observed by Kadhim et al. (28), who detected
chromomsal aberrations in human hematopoietic cells from
only two out of four donors after exposure of the cells in vitro
to α-particles. Thus, the NHBE cells used in our study may
have come from a donor who was susceptible to the induction
of DNA damage by α-particles. Very different results may
have been obtained using cells from a different donor.
Although the exact mechanism of induction of point
mutations by high LET α-particles is not clear, the direct
123
S.P.Hussain et al.
deposition of energy in the DNA molecule, as well as free
radical-mediated indirect pathways, may be involved. The
premutagenic lesions causing G:C → A:T transitions may
originate either from guanine or cytosine. Deamination of
cytosine to uracil results in a C → T base substitution, as
proposed by Tindall et al. (29), to explain G:C → A:T
transition in λ phage by gamma irradiation. Alpha particleinduced transitions of the G → A type may arise from the
following: (i) misrepair of double-strand breaks in the DNA,
causing the formation of a base pair gap in one strand of the
rejoined DNA which, in the majority of cases, are filled with
an A as dictated by the ‘A rule’ (30); or (ii) by removal of a
damaged cytosine to create a non-informative, apyrimidinic
site, opposite of which an A is most often inserted, resulting
in a C → T transition on the transcribed strand and a G → A
base substitution in the non-transcribed strand (31).
The complete absence of G → T transversions in the present
study suggests that depurination of damaged guanine and
subsequent incorporation of adenine opposite the resulting
apurinic site do not occur in α-particle-induced mutagenesis.
The observed C → A transversions in the first position of
codon 250 can be produced by direct energy deposition into
the two neighboring C residues to form an intermediate,
analogous to a pyrimidine dimer leading to a C → A transversion. We have previously shown that UVB irradiation of
human fibroblasts leads to a C → A transversion in the first
position of codon 250 (32). H2O2 treatment of Simian cells
(CV-1) transfected with a pZ189 shuttle vector containing the
supF gene, induced a high frequency of G:C → A:T transitions
(33). Treatment of normal human fibroblasts with H2O2/FeCl3
induces a G → A transition at codon 248 of p53 and G → T
and C → A transversions in codons 249 and 250 of p53
(20). The p53 mutation spectrum induced by α-particles via
generation of free hydroxy radicals by radiolysis and the
spectrum generated by hydroxyl radicals which are produced
by a superoxide-driven Fenton-like reaction in a site-specific
manner may vary, as a difference in reactivity of the two
hydroxyl radicals generated through different pathways has
been suggested (25).
The mutation spectrum in a particular gene in a specific
cancer is a consequence of both mutability at the DNA level
and the selective growth advantage conferred by the altered
protein. A codon 249 AGG → AAG transition and a codon
250 CCC → ACC transversion result in the replacement of
arginine by lysine and proline by threonine respectively.
Moreover, high LET radiation efficiently induces an undefined
process that leads to genomic instability that is transmitted to
subsequent cell divisions (16,21). Consequently, an irradiated
clone of cells will experience delayed mutagenic effects for
several generations that may not be of the type directly
produced when the cell initially encountered the radiation.
Acknowledgements
The authors would like to acknowledge the excellent technical support of Ms
Jennifer Lane. Useful discussions with Drs Stefan Ambs and William P.Bennett
are also appreciated. We are thankful to Mohammed A.Khan for help with
the graphics and to Dorothea N.Dudek for editorial assistance. This research
was partly supported by the Office of Health and Environmental Research,
US Department of Energy under contract number DE-AC04-EV01013.
References
1. Greenblatt,M.S., Bennett,W.P., Hollstein,M. and Harris,C.C. (1994)
Mutations in the p53 tumor suppressor gene: clues to cancer etiology.
Cancer Res., 54, 4855–4878.
124
2. Hollstein,M., Sidransky,D., Vogelstein,B. and Harris,C.C. (1991) p53
mutations in human cancers. Science (Wash.), 253, 49–53.
3. Hsu,I.C., Metcalf,R.A., Sun,T., Welsh,J., Wang,N.J. and Harris,C.C. (1991)
p53 gene mutational hotspot in human hepatocellular carcinomas from
Qidong, China. Nature (Lond.), 350, 427–429.
4. Bressac,B., Kew,M.,. Wands,J. and Ozturk,M. (1991) Selective G to T
mutations of p53 gene in hepatocellular carcinoma from southern Africa.
Nature, 350, 429–431.
5. Soini,Y. et al. (1996) An aflatoxin associated mutational hotspot at codon
249 in the p53 tumor suppressor gene occurs in hepatocellular carcinomas
from Mexico. Carcinogenesis, 17, 1007–1012.
6. Ozturk,M. et al. (1991) p53 mutation in hepatocellular carcinoma after
aflatoxin exposure. Lancet, 338, 1356–1359.
7. Li,D., Cao,Y., He,L., Wang,N.J. and Gu,J. (1993) Aberration of p53 gene
in human hepatocellular carcinoma from China. Carcinogenesis, 14,
169–173.
8. Barsh,D.E., Rudolph,J.A., Simon,J.A., Lin,A., McKenna,G.J., Baden,H.P.,
Halperin,A.J. and Ponten,J. (1991) A role for sunlight in skin cancer: UVinduced p53 mutations in squamous cell carcinoma. Proc. Natl Acad. Sci.
USA, 88, 10124–10128.
9. Takeshima,Y., Seyama,T., Bennett,W.P., Akiyama,M., Tokuoka,S., Inai,K.,
Mabuchi,K., Land,C.E. and Harris,C.C. (1993) p53 mutations in lung
cancers from non-smoking atomic bomb survivors. Lancet, 342, 1520–
1521.
10. Vahakangas,K.H., Samet,J.M., Metcalf,R.A., Welsh,J.A., Bennett,W.P.,
Lane,D.P. and Harris,C.C. (1992) Mutations of p53 and ras genes in radon
associated lung cancer from uranium miners. Lancet, 399, 576–580.
11. Taylor,J.A., Watson,M.A., Devereux,T.R., Michels,R.Y., Saccomanno,G.
and Anderson,M. (1994) p53 mutation hotspot in radon-associated lung
cancer. Lancet, 343, 86–87.
12. Bartsch,H., Hollstein,M., Mustonen,R., Schmidt,J., Spiethoff,A., Wesch,H.,
Wiethege,T. and Muller,K.M. (1995) Screening for putative radon-specific
p53 mutation hotspot in German uranium miners. Lancet, 346, 121.
13. McDonald,J.W., Taylor,J.A., Watson,M.A., Saccomanno,G. and
Devereux,T.R. (1995) p53 and K-ras in radon-associated lung
adenocarcinoma. Cancer Epidemiol. Biomarkers Prevent., 4, 791–793.
14. Samet,J.M. (1989) Radon and lung cancer. J. Natl Cancer Inst., 81,
745–757.
15. Moolgavkar,S.H., Luebeck,E.G., Krewski,D. and Zielinski,J.M. (1993)
Radon, cigarette smoke and lung cancer: a reanalysis of the Colorado
plateau uranium miners’ data. Epidemiology, 4, 204–217.
16. Sankaranarayanan,K. and Chakraborty,R. (1995) Cancer predisposition,
radiosensitivity and the risk of radiation-induced cancers. Radiat. Res.,
143, 121–143.
17. Zhu,L.X., Waldren,C.A., Vannais,D. and Hei,T.K. (1996) Cellular and
molecular analysis of mutagenesis induced by charged particles of defined
linear energy transfer. Radiat. Res., 145, 251–259.
18. Evans,H.H. (1994) The prevalence of multilocus lesions in radiationinduced mutants. Radiat. Res., 137, 131–144.
19. Aguilar,F., Hussain,S.P. and Cerutti,P. (1993) Aflatoxin B1 induces the
transversion of G to T in codon 249 of the p53 tumor suppressor gene in
human hepatocytes. Proc. Natl Acad. Sci. USA, 90, 8586–8590.
20. Hussain,S.P., Aguilar,F., Amstad,P. and Cerutti,P. (1994) Oxy-radical
induced mutagenesis of hotspot codons 248 and 249 of the human p53
gene. Oncogene, 9, 2277–2281.
21. Kennedy,C.H., Mitchell,C.E., Fukushima,N.H., Neft,R.E. and Lechner,J.F.
(1996) Induction of genomic instability in normal human bronchial
epithelial cells by 238Pu α-particles. Carcinogenesis, 17, 1671–1676.
22. Elliget,K.A. and Lechner,J.F. (1992) Normal human bronchial epithelial
cell cultures. In Freshney,R.I. (ed.), Culture of Epithelial Cells. Wiley2Liss,
New York, NY, pp. 181–196.
23. Venitt,S. and Biggs,P.J. (1994) Radon, mycotoxins and uranium mining.
Lancet, 343, 795.
24. Jin,Y., Yie,T.A. and Carothers,M.A. (1995) Non-random deletions at the
dihydrofolate reductase locus of chinese hamster ovary cells induced by
alpha-particles stimulating radon. Carcinogenesis, 16, 1981–1991.
25. Jaberaboansari,A., Dunn,W.C., Preston,R.J., Mitra,S. and Waters,L.C.
(1991) Mutations induced by ionizing radiation in a plasmid replicated in
human cells. II. Sequence analysis of alpha particle-induced point
mutations. Radiat. Res., 127, 202–210.
26. Stegelmeier,B.L., Gillett,N.A., Rebar,A.H. and Kelly,G. (1991) The
molecular progression of plutonium-239-induced rat lung carcinogenesis
Ki-ras expression and activation. Mol. Carcinogen., 4, 43–51.
27. Boring,C.C., Squires,T.S. and Tong,T. (1993) Cancer statistics. J. Clin.
Cancer, 43, 7–26.
p53 mutability with
238Pu
28. Kadhim,M.A., Lorimore,S.A., Hepburn,M.D., Goodhead,D.T., Buckle,V.J.
and Wright,E.G. (1994) Alpha-particle-induced chromosomal instability
in human bone marrow cells. Lancet, 344, 987–988.
29. Tindall,K.R., Stein,J. and Hutchinson,F. (1988) Changes in DNA base
sequence induced by gamma-ray mutagenesis of lambda phage and
prophage. Genetics, 118, 551–569.
30. Strauss,B.S. (1991) The ‘A Rule’ of mutagen specificity: a consequence
of DNA polymerase bypass of non-instructional lesions? BioEssays, 13,
79–84.
31. Loeb,L.A. and Preston,B.D. (1986) Mutagenesis by apurinic/apyrimidinic
sites. Annu. Rev. Genet., 20, 201–230.
32. Amstad,P., Hussain,S.P. and Cerutti,P. (1994) Ultraviolet B light induced
mutagenesis of p53 codons 248 and 249 in human skin fibroblasts. Mol.
Carcinogen., 10, 181–188.
33. Moraes,E.C., Keyse,S.M., Pidoux,M. and Tyrrell,R. (1990) Mutagenesis
by hydrogen peroxide treatment of mammalian cells: a molecular analysis.
Carcinogenesis, 11, 282–293.
34. Hollstein,H., Shomer,B., Greenblatt,M., Soussi,T., Hovig,E., Montesano,R.
and Harris,C.C. (1996) Somatic point mutations in the p53 gene of human
tumors and cell lines: updated compilation. Nucl. Acids Res., 24, 141–146.
Received on July 10, 1996; revised on September 20, 1996; accepted on
September 23, 1996
125