Download Chromosomal breakpoint positions suggest a direct role for radiation

ã
Oncogene (1999) 18, 6330 ± 6334
1999 Stockton Press All rights reserved 0950 ± 9232/99 $15.00
http://www.stockton-press.co.uk/onc
Chromosomal breakpoint positions suggest a direct role for radiation in
inducing illegitimate recombination between the ELE1 and RET genes in
radiation-induced thyroid carcinomas
YE Nikiforov*,1,2, A Kosho€er1, M Nikiforova2, J Stringer3 and JA Fagin2
1
Department of Pathology, University of Cincinnati College of Medicine, PO Box 670529, Cincinnati, Ohio, OH 45267-0529,
USA, 2Division of Endocrinology, University of Cincinnati College of Medicine, PO Box 670547, Cincinnati, Ohio, OH
45267-0547, USA, 3Department of Molecular Genetics, University of Cincinnati College of Medicine, PO Box 670524, Cincinnati,
Ohio, OH 45267-0524, USA
The RET/PTC3 rearrangement is formed by fusion of
the ELE1 and RET genes, and is highly prevalent in
radiation-induced post-Chernobyl papillary thyroid carcinomas. We characterized the breakpoints in the ELE1
and RET genes in 12 post-Chernobyl pediatric papillary
carcinomas with known RET/PTC3 rearrangement. We
found that the breakpoints within each intron were
distributed in a relatively random fashion, except for
clustering in the Alu regions of ELE1. None of the
breakpoints occurred at the same base or within a similar
sequence. There was also no evidence of preferential
cleavage in AT-rich regions or other target DNA sites
implicated in illegitimate recombination in mammalian
cells. Modi®cation of sequences at the cleavage sites was
minimal, typically involving a 1 ± 3 nucleotide deletion
and/or duplication. Surprisingly, the alignment of ELE1
and RET introns in opposite orientation revealed that in
each tumor the position of the break in one gene
corresponded to the position of the break in the other
gene. This tendency suggests that the two genes may lie
next to each other but point in opposite directions in the
nucleus. Such a structure would facilitate formation of
RET/PTC3 rearrangements because a single radiation
track could produce concerted breaks in both genes,
leading to inversion due to reciprocal exchange via endjoining.
Keywords: RET; gene rearrangement; illegitimate
recombination; radiation-induced; thyroid
Introduction
The Chernobyl nuclear power accident in April 1986
produced one of the most serious environmental
disasters ever recorded and created a rich paradigm
of radiation-induced human neoplasia, o€ering remarkable insights into thyroid carcinogenesis (Kazakov et al., 1992; Nikiforov and Gnepp, 1994; Pacini et
al., 1997). We and others observed that a speci®c type
of the RET gene rearrangement, RET/PTC3, is highly
prevalent (50 ± 58%) in post-Chernobyl pediatric
papillary carcinomas removed 5 ± 7 years after the
accident (Klugbauer et al., 1995; Fugazzola et al., 1995;
*Correspondence: YE Nikiforov
Received 8 April 1999; revised 14 June 1999; accepted 14 June 1999
Nikiforov et al., 1997), but not in tumors from
unexposed children (Bongarzone et al., 1996; Nikiforov et al., 1997). We have also found that this type of
RET rearrangement is associated with a speci®c
histotype of radiation-induced thyroid tumors ± the
solid growth papillary carcinoma, that is rare in
sporadic adult or pediatric populations (Nikiforov et
al., 1997). On the other hand, RAS and p53 mutations,
prevalent in sporadic thyroid cancers, are virtually
absent among post-Chernobyl tumors (Nikiforov et al.,
1996; Suchy et al., 1998). Thus, RET/PTC3 rearrangement is a major genetic event in a well-characterized
population of radiation-induced post-Chernobyl pediatric papillary carcinomas and most likely has direct
association with radiation exposure.
RET/PTC3 results from a paracentric inversion of
the long arm of chromosome 10 and is formed by
fusion of the intracellular domain of RET tyrosine
kinase receptor gene with the ELE1(RFG) gene
(Santoro et al., 1994). Both genes are located at
10q11.2, with a minimum distance of 500 kb between
them (Minoletti et al., 1994). The ELE1 gene is
expressed ubiquitously and drives the expression of
the truncated RET receptor in thyroid follicular cells.
The precise mechanisms of how radiation may
induce oncogenic mutations, such as RET/PTC3
rearrangement, and lead to papillary thyroid carcinoma development are unknown. One possibility is that
radiation directly caused the mutation as a result of
misrepaired DNA damage. It is well known that
radiation produces a random pattern of primary
particle tracks, that (by direct ionization and through
the agency of chemical radicals) cause damage to the
chromatin structure manifesting as single-strand DNA
breaks, double-strand DNA breaks, base damage and
DNA-protein crosslinks (Ward, 1988; Kovacs et al.,
1994; Grosovsky et al., 1988). DNA strand breaks can
be lethal to the cell, and several enzymatic mechanisms
exist to repair them. One such repair mechanism is to
ligate the ends of severed chromosomes (end-joining).
Incorrectly rejoined breaks may result in chromosomal
aberrations such as chromosomal translocations,
deletions, and inversions, through which an oncogene
(such as the RET/PTC3 fusion gene) can form. Indeed,
the induction of the RET gene rearrangement has been
observed in human fetal thyroid tissue transplanted
into SCID mice as early as 2 days following 50 Gy
exposure (Mizuno et al., 1997). The direct action model
or radiation-induced cancer does not explain the long
period of time between radiation exposure and
RET rearrangements in radiation-induced thyroid tumors
YE Nikiforov et al
development of the disease, but this lag period would
be expected if the RET/PTC3 fusion is necessary, but
not sucient to cause tumor growth. The lag period
would be the time required for a second genetic event
to occur. Indeed, it has been shown that constitutive
expression of RET/PTC1 (another form of RET
rearrangement also leading to the activation of the
truncated RET receptor) in well-di€erentiated thyroid
PCCL3 cells was associated with loss of di€erentiated
properties, but was not sucient for transformation
(Santoro et al., 1993). However, malignant phenotype
was obtained by cotransfection with H-RAS or K-RAS
oncogenes.
An alternative to the direct pathway is that radiation
does not act directly to produce the RET/PTC3 fusion,
but instead may cause persistent genomic instability,
which eventually produces one or more mutations that
result in tumor progression (reviewed in Nikiforov and
Fagin, 1998).
Whether the gene fusion occurs early or late in the
process of tumor cell evolution, two general mechanisms for such fusions are possible. One mechanism
would be for concerted breaks to form in the RET and
ELE1 genes. The RET/PTC3 fusion could then be
formed by end-joining. This mechanism would be
expected to be reciprocal. Concerted breaks could
form as the direct result of radiation damage, or due to
the action of an endonuclease. The alternative pathway
to gene fusion would be for breaks to occur
independently and to be joined at random. This
mechanism would not be expected to produce
reciprocal rearrangements.
To explore the potential mechanisms causing the
RET/PTC3 fusion in radiation-induced tumors, we
identi®ed and analysed the precise breakpoint sites in
the RET and ELE1 gene in 12 pediatric postChernobyl papillary thyroid carcinomas. The structures of these breakpoints were most consistent with
the direct mechanism of induction of RET/PTC3
fusion via reciprocal exchange at a pair of double
strand breaks which are in proximity to each other due
to the folding of the chromosome 10.
was obtained. Comparison of the nucleotide composition of the chimeric introns in each tumor with both
contributing germline sequences allowed us to deter-
Results
A series of PCRs with all possible combinations of
primers located sequentially through intervals of 200 ±
250 bp across ELE1 intron 5 (1670 bp) and RET
intron 11 (1843 bp) was performed to amplify genomic
DNA from each tumor (Figure 1). The ampli®cation
product containing ELE1/RET breakpoint was obtained in 12 of 22 tumors. The breakpoint sites within
the two genes were identi®ed by sequencing of PCR
products. Then, the reciprocal sequence of RET/ELE1
Figure 1 Strategy for breakpoint site identi®cation by PCR
ampli®cation of genomic DNA from tumors with RET/PTC3
using all possible combinations of the forward primers (F1 ± F7)
and the reverse primers (R1 ± R7)
Figure 2 Nucleotide sequences of ELE1/RET and RET/ELE1
breakpoint sites in 12 tumors and germ-line sequences of the
ELE1 and RET genes. Arrows indicate positions of the
breakpoints. Nucleotides deleted or inserted (duplicated) are
shown in boxes. Short-sequence homology between germline
RET and ELE1 sequences in the proximity of the breakpoint sites
are indicated in italics and underlined
6331
RET rearrangements in radiation-induced thyroid tumors
YE Nikiforov et al
6332
mine the precise composition of the breakpoints and
modi®cation of sequences at these sites (Figure 2). The
location and distribution of the breakpoint sites in the
RET or ELE1 genes are shown in Figure 3.
Analysis of these data revealed that breakpoints in
the ELE1 gene were asymmetrically distributed, with
nine of 12 breakpoints located in the right half of the
intron. Most of this half of the intron is occupied by
two Alu sequences, and eight breakpoints fell within
one or the other of these two SINES. Breakpoints in
the RET intron 11 were distributed more evenly along
the intron, with four cases (C2, C15, C28, C24)
clustered within an 80 bp region.
None of the RET or ELE1 breakpoints occurred at
exactly the same base or within a similar sequence in
any of the 12 tumors.
There was no long-sequence homology between
regions contributing to these rearrangements, indicating that they are the result of illegitimate rather than
homologous recombination. Short-sequence homology
(3 ± 5 nucleotides) was found at or immediately
adjacent to breakpoints in seven cases.
Each of these inversions was formed via a fairly
precise reciprocal recombination event. Modi®cation of
sequences at the breakpoint sites was minimal, typically
involving a 1 ± 3 nucleotide deletion and/or duplication.
Four cases had su€ered larger deletions of 9 ± 98 bp.
Breakpoints exhibited no particular nucleotide
sequence or composition. In the ELE1 intron, breakpoint sites were located in regions with an average ATcontent of 52%, close to the AT-content of the entire
intron, which is 61%. The same was true for the RET
intron, where breakpoints occurred within regions with
37% AT-content, the same as the AT-content of the
entire intron. Only one of 12 breakpoints in ELE1 gene
was located within a 214 bp AT-rich region (76% AT)
at the 5'-end of the intron. Two of the 24 breakpoint
sites in both genes were found at or close to a poly-A
sequence (C8 and C30 in ELE1), and one breakpoint
was found within a run of six T residues (C11 in
ELE1). Short purine/pyrimidine tracts were seen at two
rearrangement sites (C20, C24). Other direct repeats,
palindromes, chi-like minisatellite octamer sequences,
or telomere-like repeats were not found at or close to
breakpoint sites. No perfect vertebrate topoisomerase
II consensus sequence occurs within either of the
introns. However, if a 1 ± 3 bp mismatch is allowed,
three of 24 breakpoint sites (C27 and C17 in ELE1,
Figure 3 Distribution of breakpoint sites (arrows) in the ELE1
and RET genes constituting the RET/PTC3 rearrangement in 12
post-Chernobyl thyroid carcinomas. Case numbers (in bold) and
nucleotide positions of breakpoints within introns are shown
above arrows. (Alu-Alu repeats; T=topoisomerase II recognition
sequences; Chi-like=Chi-like sequence, poly-A and polyT=polynucleotide sequences)
C24 in RET) were close to topoisomerase II
recognition sites.
Thus, the breakpoint sites in each intron appeared to
be distributed in a relatively random fashion. However,
upon further analysis a certain pattern emerged. Despite
the great range of the chimeric intron lengths (from 304
to 3240 bp), these fusion introns tended to be comprised
of equal lengths of DNA from each gene. To illustrate,
the 304 bp intron in the C20 fusion contained 152 bp
from the ELE1 gene, and 152 bp from the RET gene.
Similarly, the 3240 bp C14 intron contained 1517 bp
from ELE1 and 1723 bp from the RET gene. Moreover,
after alignment of the respective introns in opposite
directions, the position of the breakpoint in one gene
showed a correspondence to the position of the
breakpoint in the other gene in each tumor (Figure 4).
As shown in Figure 4A, such an arrangement aligns ®ve
of the breakpoint pairs directly across from each other.
The other seven pairs of breakpoints can be aligned by
sliding one gene with respect to the other (Figure 4B and
C). Interestingly, in the two other previously reported
cases of Chernobyl-related cancers where fusion
occurred between the ELE1 intron 5 and RET intron
11 (CH10 and CH4) (Bongarzone et al., 1997), the breaks
that were joined to form RET/PTC3 can also be aligned
by this procedure (Figure 4).
Discussion
The analysis of the breakpoint sites in the RET and
ELE1 genes composing RET/PTC3 rearrangement in
these radiation-induced tumors suggests a direct
mechanism of RET/PTC3 induction as a result of
random double-strand DNA breaks produced by
ionizing radiation, rather than a later occurrence of this
mutation due to general destabilization of the genome.
This, if the RET/PTC intrachromosomal inversion arose
as a result of activation of the recombination machinery
or errors in cleavage by enzymes that cut and join DNA
(such as topoisomerase II), one would expect the breaks
to be located within recombinase signal sequences at
both participating loci, to have similarity in sequences at
the breakpoints, or to be clustered at certain speci®c
Figure 4 Alignment of two introns involved in RET/PTC3
demonstrates a spatial arrangement of breakpoints in two genes
in 12 of our cases (solid lines) and two cases (CH4, CH10)
reported by Bongarzone et al. (1997) (dotted lines), as breakpoints in each tumor correspond to one of three stable patterns
(A, B or C)
RET rearrangements in radiation-induced thyroid tumors
YE Nikiforov et al
hypersensitive DNA regions such as AT-rich regions,
Alu repeats, repeated purine/pyrimidine tracts, and other
chromosomal regions with increased lability (fragile)
sites (Ehrlich et al., 1993; Stary and Sarasin, 1992;
Hyrien et al., 1987). By contrast, our results indicate that
in post-Chernobyl tumors the RET/PTC breakpoints
were distributed relatively randomly across the respective introns, except for clustering in the Alu sequences of
one of the two contributing genes, with no breakpoints
occurring at exactly the same base or within an identical
sequence in any of the 12 tumors. Although it is possible
that cleavage within Alu regions favored recombination,
the overall lack of consistency in speci®c DNA sites
implicated in illegitimate recombination in mammalian
cells suggests a direct induction of these genetic events as
a result of random double-strand DNA breaks.
Despite the random distributions of the breakpoint
sites in each intron, the alignment of two introns in
opposite directions disclosed a certain pattern of
correspondence of ELE1 and RET breaks in each
tumor. In addition, breakpoint sites in RET/PTC3 in
two previously reported post-Chernobyl cases (Bongarzone et al., 1997) showed similar patterns of correspondence. Such predictable relative positioning of
chromosomal breaks in two di€erent genes can occur if
the chromosome is folded in a way to space these distant
loci next to each other but pointed in opposite directions
at the time of chromosomal breaks. In such cases,
passage of a single radiation track can produce concerted
double-strand DNA breaks in two chromosomal
fragments closely spaced in the thyroid cell nucleus.
A number of studies indicate that chromosomes and
their domains occupy distinct territories in interphase
nuclei, rather than being randomly dispersed throughout
the nuclear volume (Lichter et al., 1988; Pinkel et al.,
1988). Furthermore, chromosome arms and bands also
form exclusive domains within a given territory (Cremer
et al., 1996). It has also been demonstrated that nuclear
localization of centromeres and other discrete chromosomal regions is non-random, and varies between
di€erent phases of the cell cycle and according to the
cell type (Manuelidis, 1984; Ferguson and Ward, 1992;
Gerdes et al., 1994). Moreover, it has been shown that
for two random chromosomal loci the relationship
between interphase distance and genomic separation
was not always linear, leading to a model of chromatin
arrangement in ¯exible loops of several Mbp arranged
along the chromosomal backbone (Yokota et al., 1995).
Thus, although two chromosomal loci are located at
a considerable linear distance from each other (such as
the ELE1 and RET genes), they may be closely spaced
in the interphase nucleus because of their position at
speci®c areas of chromosomal loop(s).
The position of breakpoint sites in DNA recombination is thought to be determined mostly by sequencespeci®c factors, or secondary DNA structures at the
breakpoint site. Thus, two distant regions consistently
involved in a rearrangement event should possess a
certain degree of nucleotide homology between them or
contain recombinase-speci®c sequences at the cleavage
site. However, the ELE1 and RET breakpoints do not
display such features. Furthermore, our ®ndings suggest
that the preferential occurrence of this rearrangement
may be due in large part to the structural organization of
the chromosome, resulting in the proximity of potentially recombinogenic DNA sequences during a certain
stage of cell cycle. Schematic representation of the most
possible mechanism of RET/PTC3 formation in these
tumors is depicted in Figure 5.
In conclusion, our ®ndings indicate that RET/PTC3
rearrangements in these radiation-induced tumors are a
result of illegitimate reciprocal recombination. The
simplest path to such events is ligation of ends that
were produced by double-strand DNA breaks induced
by radiation at the same time in the same place in a
nucleus.
Materials and methods
Patient population
We studied 22 post-Chernobyl papillary thyroid carcinomas
with RET/PTC3 rearrangements previously identi®ed by RT ±
PCR (Nikiforov et al., 1997). All patients resided in the areas
of Belarus contaminated as a result of the Chernobyl nuclear
accident in April, 1986. They underwent surgery at the
Thyroid Tumor Center in Minsk, Belarus in 1991 and 1992.
There were 15 males and seven females in this group. The age
of patients at the time of surgery ranged from 5 ± 18 years.
Figure 5 Putative mechanism of RET/PTC3 rearrangement
based on the results of the current study. Schematic steps of
RET/PTC3 formation starting from the linear arrangement of the
wild-type genes in the long arm of chromosome 10 (top), to
formation of chromosomal loop dextraposing the two genes
adjacent to each other but pointed in opposite directions, to
simultaneous breaks in both genes at the same sites with
reciprocal exchange via end-joining leading to inversion and
gene rearrangement as demonstrated in a linear fashion (bottom).
Note that as a result of this reciprocal recombination the regions
1 and 2, internal to the target genes, are also inverted
6333
RET rearrangements in radiation-induced thyroid tumors
YE Nikiforov et al
6334
Identi®cation of breakpoints
PCR-based approach was utilized to identify the precise
breakpoint sites of RET/PTC3 rearrangement located within
ELE1 intron 5 (1670 bp) and RET intron 11 (1843 bp). For
these experiments, we determined that the quality of DNA
extracted from paran-embedded tissue (performed as
previously reported in Nikiforov et al., 1996) was sucient
to amplify fragments of 400 ± 500 bp, and that the breakpoints in RET and ELE1 may occur at any site across the
respective introns. Based on the reported sequences of RET
intron 11 (GenBank Acc. # X77860) and ELE1 intron 5
(GenBank Acc. # X77859), we designed primers located
sequentially through intervals of 200 ± 250 bp across both
introns (Figure 1). A series of PCRs in a ®nal volume of
20 ml with all possible combinations of upstream and
downstream primers were performed to amplify the genomic
DNA from each tumor as previously described (Nikiforov et
al., 1996). Ten ml of each PCR product were loaded on
agarose gels, blotted to the nylon membranes and hybridized
with an internal oligonucleotide probe end-labeled with 32P
(Nikiforov et al., 1997). Another 10 ml of the PCR products
were electrophoresed in an agarose gel and stained with
ethidium bromide. The band of interest was excised, puri®ed,
and sequenced with an automated ABI model 377 sequencer
using the ABI PRISM Dye Terminator Cycle Sequencing kit
(Perkin-Elmer, Foster City, CA, USA). If the initially
identi®ed ampli®cation signal was larger than 500 bp or
was visualized only after hybridization, additional internal
primers were designed to narrow the region in order to
facilitate sequencing.
After detecting the breakpoint sites in ELE1/RET we
designed primers to obtain the reciprocal sequence of RET/
ELE1 for each tumor. PCR and sequencing techniques were
identical to those described above. The position and
nucleotide composition of primers for ELE1/RET and
RET/ELE1, as well as of the oligonucleotide probes are
available upon request.
Analysis of breakpoint sites and search for sequence features
The sequences obtained from each tumor where compared
with the available germline sequences of the RET intron 11
and ELE1 intron 5 utilizing the DNAStar program
(DNASTAR, Inc.) to identify the breakpoints sites in both
genes, and to analyse possible modi®cation of nucleotide
sequence at the breakpoint sites.
AT-rich regions, Alu sequences, direct DNA repeats,
palindromes, purine/pyrimidine tracts, telomere-like repeats
(TTAGGG)n, chi-like octamer conserved within minisatellite
repeats (GG(A/T) GG(A/T)CG), as well as vertebrate
topoisomerase
II
consensus
sequences
((A/G)N(T/
C)NNCNNG(T/C)NG(G/T)
TN(T/C)N(T/C))
were
searched for in both introns and in the 100 bp of DNA
surrounding each breakpoint.
Acknowledgments
This work was supported by a TRAC award from Knoll
Pharmaceuticals to Dr Nikiforov and in parts by grants
CA 50706, CA 72597 to Dr Fagin and ES 05652 to Dr
Stringer. We are grateful to Dr Fenoglio-Preiser for
valuable comments.
References
Bongarzone I, Butti MG, Fugazzola L, Pacini F, Pinchera A,
Vorontsova TV, Demidchik EP and Pierotti MA. (1997).
Genomics, 42, 252 ± 259.
Bongarzone I, Fugazzola L, Vigneri P, Mariani L,
Mondellini P, Pacini F, Basolo F, Pinchera A, Pilotti S
and Pierotti MA. (1996). J. Clin. Endocrinol. Metab., 81,
2006 ± 2009.
Cremer C, MuÈnkel C, Granzow M, Jauch A, Dietzel S, Eils
R, Guan X-Y, Meltzer PS, Trent JM, Langowski J and
Cremer T. (1996). Mutat. Res., 366, 97 ± 116.
Ehrlich SD, Bierne H, d'Alencon E, Vilette D, Petranovic M,
Noirot P and Michel B. (1993). Gene, 135, 161 ± 166.
Ferguson M and Ward DC. (1992). Chromosoma, 101, 557 ±
565.
Fugazzola L, Pilotti S, Pinchera A, Vorontsova TV,
Mondellini P, Bongarzone I, Greco A, Astakhova L,
Butti MG, Demidchik EP, Pacini F and Pierotti MA.
(1995). Cancer Res., 55, 5617 ± 5620.
Gerdes MG, Carter KC, Moen PT and Lawrence JB. (1994).
J. Cell Biol., 126, 289 ± 304.
Grosovsky AJ, de Boer JG, deJong PJ, Drobetsky EA and
Glickman BW. (1988). Proc. Natl. Acad. Sci. USA., 85,
185 ± 188.
Hyrien O, Debatisse M, Buttin G and de Saint Vinsent BR.
(1987). EMBO J., 6, 2401 ± 2408.
Kazakov VS, Demidchik EP and Astakhova LN. (1992).
Nature, 359, 21.
Klugbauer S, Lengfelder E, Demidchik EP and Rabes HM.
(1995). Oncogene, 11, 2459 ± 2461.
Kovacs MS, Evans JW, Johnstone IM and Brown JM.
(1994). Radiat. Res., 137, 34 ± 43.
Lichter P, Cremer T, Borden J, Manuelidis L and Ward DC.
(1988). Hum. Genet., 80, 222 ± 224.
Manuelidis L. (1984). Proc. Natl. Acad. Sci. USA, 81, 3123 ±
3127.
Minoletti F, Butti MG, Coronelli S, Miozzo M, Sozzi G,
Pilotti S, Tunnacli€e A, Pierotti MA and Bongarzone I.
(1994). Genes Chrom. Cancer, 11, 51 ± 57.
Mizuno T, Kyoizumi S, Suzuki T, Iwamoto KS and Seyama
T. (1997). Oncogene, 15, 1455 ± 1460.
Nikiforov YE and Gnepp DR. (1994). Cancer, 74, 748 ± 766.
Nikiforov YE, Nikiforova M, Gnepp DR and Fagin JA.
(1996). Oncogene, 13, 687 ± 693.
Nikiforov YE, Rowland JM, Bove KE, Monforte-Munoz H
and Fagin JA. (1997). Cancer Res., 57, 1690 ± 1694.
Nikiforov YE and Fagin JA. (1998). In: Advances in
Molecular and Cellular Endocrinology. Vol 2, LeRoith D.
(ed.). JAI Press Inc: Greenwich, Connecticut, pp. 169 ±
196.
Pacini F, Vorontsova T, Demidchik E, Molinaro E, Agate L,
Romei C, Shavrova E, Cherstvoy E, Ivashkevitch Y,
Kuchinskaya E, Schlumberger M, Rouga G, Felesi M and
Pinchera A. (1997). J. Clin. Endocrinol. Metab., 82, 3563 ±
3569.
Pinkel D, Landegent J, Collins C, Fuscoe J, Seagraves R,
Lucas J, Gray JW. (1998). Proc. Natl. Acad. Sci. USA, 85,
9138 ± 9142.
Santoro M, Dathan NA, Berlingieri MT, Bongarzone I,
Paulin C, Grieco M, Pierotti MA, Vecchio G and Fusco A.
(1994). Oncogene, 9, 509 ± 516.
Santoro M, Melillo RM, Grieco M, Berlingieri MT, Vecchio
G and Fusco A. (1993). Cell Growth Di€eren., 4, 77 ± 84.
Stary A and Sarasin A. (1992). Nucleic Acids Res., 20, 4269 ±
4274.
Suchy B, Waldmann V, Klugbauer S and Rabes HM. (1998).
Br. J. Cancer, 77, 952 ± 955.
Ward JF (1988). Progr. Nucleic Acids Mol. Biol., 35, 95 ± 125.
Yokota H, van den Engh G, Hearst JE, Sachs RK and Trask
BJ. (1995). J. Cell Biol., 130, 1239 ± 1249.