Download A novel exon within the mdm2 gene modulates translation

ã
Oncogene (1999) 18, 7026 ± 7033
1999 Stockton Press All rights reserved 0950 ± 9232/99 $15.00
http://www.stockton-press.co.uk/onc
A novel exon within the mdm2 gene modulates translation initiation in vitro
and disrupts the p53-binding domain of mdm2 protein
Nik Veldhoen1, Su Metcalfe2 and Jo Milner*,1
1
YCR P53 Research Group, Department of Biology, University of York, York, YO10 5DD, UK; 2Department of Surgery,
Addenbrooke's Hospital, University of Cambridge, Cambridge, CB2 2QQ, UK
The mdm2 protein interacts with a number of proteins
involved in cell growth control. Such interactions favour
cell proliferation and may explain the oncogenic
potential of mdm2 when over-expressed in cells.
Interaction with the tumour suppressor p53 involves the
N-terminus of mdm2 and targets p53 for rapid
degradation by the ubiquitin pathway. We now describe
a novel, highly conserved exon of mdm2 (exon a) which
includes an in-frame UGA stop codon. Expression of
exon a disrupts in vitro translation of the p53 binding
domain of mdm2. We propose that exon a induces
translation re-initiation at an internal AUG codon within
the mdm2a mRNA isoform. The putative mdm2a protein
lacks the N-terminus of mdm2 and shows little, if any,
binding capacity for p53. Mdm2a mRNA is expressed in
a tissue-speci®c manner and is observed predominantly in
testis and peripheral blood lymphocytes. We propose that
mdm2a expression may provide a mechanism for
uncoupling mdm2-p53 interaction in certain cell types
and/or under speci®c conditions of cell growth.
Keywords: mdm2; p53; regulatory exon; alternative
splicing
Introduction
In normal cells the mdm2 protein can bind proteins
involved in cell growth suppression and in cell
proliferation. The e€ects of such interactions favours
cell proliferation and this may account for the
oncogenic nature of mdm2 when over-expressed
(Haines, 1997; Momand and Zambetti, 1997; Piette et
al., 1997). Two major tumour suppressors, p53 and
pRB, are bound by mdm2 with subsequent inhibition
of their cell growth suppressor properties (Momand et
al., 1992; Oliner et al., 1993; Xiao et al., 1995).
Conversely, transcriptional activation by E2F-1/DP1 is
stimulated when complexed with mdm2 protein
(Martin et al., 1995). The E2F-1/DP1 heterodimer
facilitates the progression of cells through S phase and
mdm2 is predicted to enhance this function and
promote cell division. Indeed, targetted over-expression of an mdm2 minigene in the mammary gland
results in uncontrolled entry into S phase, polyploidy
*Correspondence: J Milner
Received 17 March 1999; revised 9 August 1999; accepted 23 August
1999
and tumour formation in transgenic mice (Lundgren et
al., 1997). The e€ect is independent of p53 since both
wild type and p53 null mice were similarly a€ected.
The mdm2 protein can be divided into sub-domains
(reviewed by Piette et al., 1997). Human mdm2 is
composed of 491 residues (see Figure 1) and
interactions with p53 and E2F-1/DP1 occur within
the N-terminal half of the protein (Chen et al., 1993;
Leng et al., 1995; Martin et al., 1995; Freedman et al.,
1997). Other interactions occur at distinct sites within
mdm2, including L5 protein binding within the central
acidic domain (residues 237 ± 260) and RNA binding at
the C-terminal RING ®nger domain (residues 438 ±
477) (MareÂchal et al., 1994; Elenbaas et al., 1996).
Interaction between mdm2 and Numb, a cell fate
regulator, has recently been reported and raises the
possibility that this pluripotential protein may also
in¯uence cell di€erentiation and survival (JuvenGershon et al., 1998).
The interaction between p53 and mdm2 has
attracted a lot of attention and the molecular structure
of the binding domain of p53 in complex with the Nterminus of mdm2 has now been resolved (Kussie et
al., 1996). The p53 binding domain of mdm2 (residues
15 ± 125 for the human form) contains a deep
hydrophobic
cleft
into
which
p53
residues
TFSDLWKLL (amino acids 18 ± 26) bind as an
amphipathic alpha helix. Binding e€ectively blocks
the transactivation domain of p53 and targets p53 for
rapid degradation by the ubiquitin-mediated proteolytic pathway (Haupt et al., 1997; reviewed by Kubbutat
and Vousden, 1998). Indeed, there is evidence that
mdm2 directly functions as a ubiquitin ligase for p53
(Honda et al., 1997; Honda and Yasuda, 1999).
Nuclear-cytoplasmic shuttling by mdm2 may also be
important in accelerating the degradation of p53 (Roth
et al., 1998). These various properties of mdm2 serve to
down-regulate p53 activity and this is essential for
normal embryonic growth and development (Montes
de Oca Luna et al., 1995).
The oncogenic potential of mdm2 was ®rst
demonstrated by Fakharzadeh et al. (1991) and is
now recognized as an important factor in human
cancer. Gene ampli®cation and overexpression of
mdm2 protein is common in soft tissue sarcomas
(20%), osteosarcomas (16%) and oesophageal cancer
(13%) (see Momand et al., 1998). One consequence of
mdm2 overexpression is predicted to be the inhibition
of p53 function, and this would contribute towards
genomic instability and oncogenic transformation.
Abnormal transcripts of mdm2 in tumours have also
been reported (Haines et al., 1994; Maxwell, 1994;
A novel mdm2 exon modulates translation
N Veldhoen et al
Sigalas et al., 1996; Matsumoto et al., 1998; Kraus et
al., 1999) but are not evident in normal cells (Montes
de Oca Luna et al., 1996).
In the present study we have identi®ed a novel exon
within the mdm2 gene which is expressed in normal
tissue and is 99% conserved between human and
canine mdm2. It is located between the previously
identi®ed exons 4 and 5 of mdm2 and is ¯anked by
conventional intron/exon splicing boundaries. This
unique exon (termed exon a) disrupts translation of
the p53 binding domain of mdm2 and may represent a
mechanism for uncoupling mdm2-p53 interaction in
certain normal cell types and/or under speci®c
conditions of cell growth.
approximate intronic distances of exon a from exons 4
and 5 are shown in Figure 2b. DNA sequencing within
this region of the human and canine mdm2 gene also
indicated that the exon-intron boundaries ¯anking the
a exon conform to the GT ± AG rules required for
accurate mRNA splicing (Table 1) (Padgett et al.,
1986).
These results demonstrate that both mdm2 and the
mdm2a splice variant are highly conserved between
human and canine species. For simplicity, in the text
these two cDNAs and their related proteins are
Results
Isolation of canine mdm2 cDNA reveals a novel exon
This study forms part of a systematic analysis of cell
growth control and spontaneous tumour development
in the dog. Having cloned and sequenced canine p53
(Veldhoen and Milner, 1998) we next sought to
characterize canine mdm2 and its interaction with
p53. A canine homologue of mdm2 cDNA was isolated
by reverse transcriptase-PCR (RT ± PCR) using mRNA
derived from canine mammary tumour tissue (Materials and methods). The predicted canine mdm2
(cMDM2) protein shares 93% amino acid identity
with human mdm2 (hMDM2) and 81% identity with
murine mdm2 (Figure 1). A second mdm2 cDNA,
called c-mdm2a, was isolated from total RNA derived
from canine peripheral blood leukocytes (PBLs). The cmdm2a cDNA is unusual in that it contains an
additional 87 base pair sequence which encodes 29
in-frame codons, including a UGA codon (Figures 1
and 2a). This additional sequence, termed the a exon,
lies within the sequence encoding the p53 binding
domain of mdm2.
The a exon is highly conserved
We next asked if human cells also express a mdm2a
transcript and screened for the presence of mdm2a
mRNA in human lymphocytes. cDNA ampli®cation
using total RNA isolated from human blood
lymphocytes lead to the identi®cation of a h-mdm2a
splice variant (Figure 1). The a exon sequence is highly
conserved between canine and human mdm2 genes,
with only a single base pair di€erence between the two
species (see Figure 2a). Such a high level of
evolutionary conservation suggests that the a exon
may represent a functionally important region of the
mdm2 gene.
The a exon of both canine and human mdm2 are
located at identical sites within their respective cDNAs
(see Figure 1). This site corresponds to an exon-intron
boundary in the murine mdm2 gene (between exons 4
and 5; Montes de Oca Luna et al., 1996; Jones et al.,
1996). To characterize the position of the a exon in
more detail we performed genomic PCR analysis using
oligonucleotide primers that anneal to the a exon and
to exons 4 and 5 (Materials and methods). The
ampli®ed DNA products served to localize the a exon
within the human and canine mdm2 genes and the
Figure 1 Alignment of mdm2-related protein sequences. The
coding potential of canine mdm2 and mdm2a, and human mdm2a
is depicted along with previously published sequences for murine
mdm2 (Cahilly-Snyder et al., 1987), human mdm2 (Oliner et al.,
1992), and Xenopus mdm2 (MareÂchal et al., 1997). The presence
of canine and human mdm2a transcripts were con®rmed in PBLs
isolated from ®ve di€erent dogs and seven separate human
individuals. Highly conserved regions are shaded while proposed
nuclear localization signals (NLS) and a nuclear export signal
(NES) are boxed (Piette et al., 1997; Roth et al., 1998). Amino
acid residues that contribute to p53 protein binding are identi®ed
above the sequence alignment by black circles (Kussie et al., 1996;
BoÈttger et al., 1997; Freedman et al., 1997). The positions within
mdm2a of the UGA codon and the downstream AUG
(corresponding to AUG62 in mdm2) are shown by an asterisk
and an open triangle, respectively
7027
A novel mdm2 exon modulates translation
N Veldhoen et al
7028
referred to as mdm2 and mdm2a, with the species of
origin given in the relevant ®gures.
Translation of mdm2a protein initiates from an internal
AUG codon
The a exon contains an in-frame UGA codon, the
context of which indicates an ecient stop signal for
translation (UGAG, Figure 2a; McCaughan et al.,
1995). We were therefore surprised when in vitro
translation of mdm2a cRNA yielded detectable
protein (Figure 3a, lane 3). Translational readthrough seemed unlikely since this would generate
an mdm2a protein product containing 29 additional
amino acids, whereas the in vitro translated mdm2a
protein had a lower apparent molecular weight than
mdm2 (Figure 3a, lane 3 compared with lanes 1 and
2).
The above results might be explained if translation
of mdm2a cRNA begins at the proximal AUG1, stops
at the UGA codon within exon a, and subsequently reinitiates at a downstream AUG. Alternatively, translation of mdm2a RNA may be accomplished by internal
ribosome entry to the transcript and translation
initiation at an AUG codon located downstream of
the UGA stop codon. These alternative mechanisms
are represented schematically in Figure 3b. Both
alternatives would generate an mdm2 protein product
with a lower molecular weight than full length mdm2,
consistent with the observed results (Figure 3a). AUG62
(codon 62 of mdm2) is the ®rst start codon downstream of the UGA stop signal and we constructed an
N-terminal deletion mutant of mdm2, mdm2DN, which
is predicted to initiate translation at AUG62 (Figure
3c). When translated in vitro, the truncated mutant comigrated with mdm2a suggesting that the N-terminus
of the putative mdm2a protein is equivalent to AUG62
of mdm2 (Figure 3a, lanes 3 and 4).
Mdm2a translation in vitro occurs through a
re-initiation mechanism
We next sought to discriminate between the two
models presented in Figure 3b. Does translation of
mdm2a start at AUG62 following internal ribosome
entry? Alternatively, does translation initiation begin at
AUG1 and subsequently re-initiate at AUG62 due to the
presence of the stop codon in exon a? If the latter
alternative applies we reasoned that removal of the
stop signal would permit translational read-through of
the a exon open reading frame. The resulting protein
would contain 29 additional residues compared with
mdm2 (since this represents the in-frame coding
capacity of the a exon) and migrate slower on a
polyacrylamide gel. When the in-frame UGA stop
codon was mutated to CGA (alanine) the mutant
mdm2a protein displayed a slightly larger apparent
molecular weight than full-length mdm2 (Figure 3a,
lane 5). This is consistent with translation initiation at
AUG1 and read through of the a exon coding
sequence. Overall, these results suggest that primary
initiation of translation of mdm2a mRNA occurs at
AUG1 and subsequently re-initiates at an internal
AUG codon (corresponding to position 62 in the
mdm2 sequence) due to the presence of the UGA stop
codon in exon a. This re-initiation model is represented
in Figure 3b, and the various mdm2 constructs used in
the experiment are shown in Figure 3c.
Exon a regulates mdm2a translation eciency
Figure 2 Characterization of the a exon sequence within human
and canine mdm2 genes. (a) The RNA sequence of the a exon
from canine and human mdm2a transcripts. The potential amino
acid sequence encoded by each a exon is shown above (canine) or
below (human) the RNA. The single nucleotide di€erence
between the two exons is depicted by an open circle. The UGA
codon within each a exon is identi®ed by an asterisk. (b)
Schematic diagram showing the location of the a exon within
the human and canine mdm2 genes. Approximate base pair (bp)
lengths are indicated
Translation eciency of an open reading frame can be
down-regulated by the presence of upstream stop
codons within bicistronic mRNA (for examples see
Kaufman et al., 1987; Luukkonen et al., 1995). The
organization of mdm2a mRNA suggests that translation eciency may be similarly a€ected by the UGA
stop codon in exon a. This would be consistent with
the lower amount of in vitro translated mdm2a protein
which was routinely observed compared with mdm2
(see Figure 3a). In detailed time course studies we
compared the eciency of translation of mdm2 and
mdm2a mRNAs using rabbit reticulocyte lysate for
Table 1 Exon-intron boundaries within the region of the human and canine mdm2 genes containing the a exon
Species
Exon
DNA Sequence of exon-intron boundaries
Exon
Canine
4
canine a
4
human a
AAAGAGgtaagctaaa ± ND
ATAGAGgtgagctgct ± tctatttcagGTGATA
AAAGAGgtaagctgaa ± tctatccaagGACTTC
ATAGAGgtgagctgat ± tgtatttcagGTTCTT
canine a
5
human a
5
Human
Exon DNA sequence is shown in uppercase while intron DNA sequence is depicted in lowercase. Invariant dinucleotides used in RNA splicing
are shaded. ND, not determined
A novel mdm2 exon modulates translation
N Veldhoen et al
translation (Materials and methods). The rate of
accumulation of radiolabelled mdm2 protein was
rapid over the ®rst 10 min of translation, and began
to plateau after 20 ± 30 min (Figure 4). No further
increase in the rate of protein accumulation was
observed after 40 min (data not shown). In contrast,
radiolabelled mdm2a protein showed a threefold slower
rate of accumulation within the ®rst 10 min of
translation (Figure 4), with a much lower overall
yield. Mutation of the UGA stop codon to GCA
restored the eciency of translation of mdm2a mRNA
Figure 3 Translation of canine mdm2-related proteins. (a)
Migration of human and canine mdm2-related proteins on a
12% SDS-polyacrylamide gel. Protein of human origin is shown
by `h' and canine proteins are identi®ed by `c'. (b) Two alternative
models for the translation of mdm2a depicting a re-initiation
process and internal ribosome entry to the RNA message. (c)
Schematic representation of the canine cDNA constructs
encoding mdm2, mdm2a, mdm2DN, and mdm2RT. Regions
encoding functional protein domains are shaded. The location of
translation initiation and termination codons described in the text
are identi®ed
to that observed for mdm2 (data not shown). We
conclude that the stop codon in exon a reduces the rate
of translation of mdm2a protein in vitro.
Mdm2a lacks an intact p53 binding domain and fails to
bind p53
Re-initiation of mdm2a translation at AUG62 would
result in disruption of the p53 binding site of mdm2
(residues 15 ± 125; Chen et al., 1993; Kussie et al., 1996;
BoÈttger et al., 1997; Freedman et al., 1997). To
determine if this crucial binding domain is lost, we
characterized the immunoprecipitation pro®le of the
mdm2 and mdm2a proteins. Controls showed that
human and canine mdm2 share six epitopes (Table 2),
including the epitopes for antibodies 4B2 (amino acids
19 ± 50; Chen et al., 1993) and 3G5 (amino acids 66 ±
69; BoÈttger et al., 1997). These two epitopes are located
within the p53 binding domain of mdm2. The mdm2a
and mdm2DN proteins were recognized by all
antibodies except 4B2 and 3G5 (Table 2) thus
con®rming that these proteins lack an intact Nterminus and an intact p53 binding domain.
To assess the interaction between mdm2a and p53
the two proteins were co-translated in vitro and
analysed by co-immunoprecipitation, using PAb421
(directed towards the C-terminus of p53) and 4B11
(directed towards the C-terminus of mdm2). Canine
mdm2 and the mdm2DN mutant, which lacks the ®rst
61 residues of mdm2, were included as controls. The
results of the co-immunoprecipitation experiments are
summarized in Table 2. Complexes between canine
mdm2 and canine p53 were observed following
reciprocal immunoprecipitations with PAb421 and
4B11 (Table 2). No complex formation was observed
between canine p53 and mdm2DN protein (Table 2).
Figure 4 Translation kinetics of mdm2-related proteins. The rate
of protein accumulation from human mdm2 (closed circle), canine
mdm2 (open circle), canine mdm2a (closed square), and mdm2DN
(open square) are shown. Radioactive label incorporation (c.p.m.)
for each protein sample was adjusted for methionine content. The
rate of protein accumulation was observed to plateau after 40 min
(data not shown)
7029
A novel mdm2 exon modulates translation
N Veldhoen et al
7030
Table 2
Protein
h
mdm2
mdm2c
mdm2ac
mdm2DNc
mdm2RTc
Immunoprecipitation pro®les and p53 binding activities of mdm2-related proteins
p53 protein
binding
4B2
3G5
+
+
7
7
ND
+
+
7
7
+
+
+
7
7
+
Antibody binding
2A9
SMP14
+
+
+
+
+
+
+
+
+
+
2A10
4B11
+
+
+
+
+
+
+
+
+
+
Comparison of antibody reactivity and p53 protein binding of mdm2-related proteins. Strong antibody binding is shown by`+',while reduced or
no antibody binding is depicted by `+'and `7' respectively. Association with p53 protein is shown by `+' and no p53 binding by `7'. Human
mdm2 protein labelled with an `h' and canine mdm2 proteins with a `c'. ND, not determined
Co-translation of p53 with mdm2a also yielded little, if
any detectable p53-mdm2a complexes (Table 2)
demonstrating that incorporation of the a exon of
mdm2 disrupts protein ± protein interaction between
p53 and the mdm2a isoform.
Mdm2a is highly expressed in leukocytes and testicular
tissue
Given the potential functional signi®cance of the a
exon (see Discussion) we next asked if mdm2a mRNA
is similarly expressed in all tissues, or is expressed in a
tissue-speci®c manner. For this purpose, we investigated the expression pattern of mdm2a mRNA in a
number of di€erent tissue types using a nonquantitative nested RT ± PCR (Materials and methods). Controls showed that mdm2 mRNA was
detectable in all tissues examined (Figure 5, upper
panel).
An oligonucleotide primer speci®c for the a exon
was used in the second round of DNA ampli®cation to
detect the presence of mdm2a mRNA. Most tissues
showed little, if any mdm2a expression (Figure 5, lower
panel). The exceptions were lymphoid and testicular
tissues in which mdm2a transcripts were clearly
detectable. These observations were extended to
include normal lymphocytes from a total of seven
human individuals and normal leukocytes from ®ve
individual dogs: all gave results similar to those
presented for canine peripheral blood leukocytes (see
Figure 5, lower panel for canine leukocytes). These
results indicate that mdm2a mRNA can be expressed
in a tissue-speci®c manner and is present in lymphoid
and testicular tissues.
Discussion
In the process of studying malignant transformation in
the dog we have now cloned and sequenced canine p53
and mdm2 cDNAs (Veldhoen and Milner, 1998 and
this paper). This systematic approach has revealed the
presence of a novel `a exon' in the canine mdm2 gene
and an almost identical a exon in human mdm2. The a
exon exhibits a number of remarkable properties and
represents a novel mechanism for the determination of
protein structure by causing re-initiation of translation
with consequent loss of the N-terminus of mdm2
protein.
Modi®cation of protein structure and function by
alternative mRNA splicing is not unusual. For
Figure 5 Expression of mdm2a mRNA in di€erent tissues of the
dog. Ampli®ed mdm2 control and mdm2a cDNA derived from
each tissue type were run on a 1% agarose gel
example, expression of the alternatively spliced exon
5 of the Wilms tumour suppressor gene leads to the
incorporation of an additional 17 amino acid sequence
that disrupts the regulatory domain of the WT1
protein (Bruening and Pelletier, 1996). Mechanistically, however, this modi®cation of protein structure is
distinct from the system we describe for mdm2a since
exon 5 of WT1 (i) does not contain an in-frame stop
codon and (ii) does not a€ect translation of the Nterminus of the protein product.
A mechanism much closer to the one we describe for
mdm2a is indicated for the expression of the cAMP
response modulator (CREM). Two CREM transcripts,
CREM 23 and CREM 24, contain a 100 bp exon (exon
c) with an in-frame stop codon (Gellerson et al., 1997).
The mRNA is nevertheless translated into functional
CREM isoforms and, signi®cantly, both protein
isoforms lack the N-terminus and appear to arise by
initiation at an internal AUG. However, further studies
of CREM 23 and CREM 24 are required to support a
re-initiation model for translation and to determine the
signi®cance of the stop codon in exon c. Thus, we
believe that the mdm2a exon is the ®rst example of a
mammalian exon with the capacity to `reprogram'
translation initiation of protein synthesis.
The remarkable conservation of the a exon suggests
that it comprises a functionally important component
of the mdm2 gene, and our results suggest that the a
exon may regulate mdm2 expression at the level of
translation. Thus, in vitro, exon a causes translation of
the mdm2a transcript to stop, due to the presence of an
in-frame stop codon, and then re-initiate at a down-
A novel mdm2 exon modulates translation
N Veldhoen et al
stream AUG codon (which corresponds to position 62
in mdm2) (see Figure 3). The resulting protein product
lacks the N-terminus of mdm2 and the p53 binding
domain is disrupted (Table 2). The biological consequences of this e€ect are likely to be profound. It is
well established that interaction with p53 is a vital
function of mdm2, and that mdm27/7 mouse embryos
are non-viable (Montes de Oca Luna et al., 1995).
Therefore any mechanism which disrupts mdm2-p53
interaction must be under stringent control. This would
be consistent with the observed reduced rate of
translation and cell type speci®c expression of mdm2a
mRNA (see Figures 3a and 5). Studies are in progress
to evaluate the translation potential of mdm2a in vivo,
using cells of di€erent tissue types in case translational
control is also cell type speci®c.
The expression of the mdm2 gene has previously
been shown to involve a combination of di€erential
mRNA splicing and internal translation initiation
(Olson et al., 1993; Barak et al., 1994; Gudas et al.,
1995; Kraus et al., 1999). Codons AUG50 and AUG62
within mdm2 mRNA can be used as sites of translation
initiation (Barak et al., 1994; Saucedo et al., 1999).
However, the precise factors that modulate selection of
these internal translation start sites remains poorly
de®ned. In the case of the alternatively spliced mdm2a
transcript, the presence of the a exon may serve to
redirect translation initiation to AUG62 (see Figure 3).
Mdm2 proteins derived from these internal initiation
events, such as the mdm2a isoform, lack a functional
p53 binding domain and may be involved in p53independent activities within the cell (Saucedo et al.,
1999). For example, N-terminally truncated mdm2
protein can rescue TGFb-induced growth arrest by
a€ecting the Rb/E2F pathway (Sun et al., 1998). Thus,
expression of the mdm2a protein isoform could serve
as a proliferative signal in certain normal cell types.
Preliminary results using speci®c primers to compare
mdm2 and mdm2a mRNA levels in normal canine
testicular tissue indicate that mdm2a mRNA is
approximately an order of magnitude lower than
mdm2. Interestingly, in canine testicular tumour
mdm2 expression was two orders of magnitude higher
than mdm2a (Watterson and Milner, unpublished
observations). This would be consistent with abnormal
down-regulation of p53 (by mdm2) in the tumour cells
whilst retaining putative proliferative functions of
mdm2a.
Tissue-speci®c expression of mdm2a mRNA in
normal testicular tissue and peripheral blood lymphocytes is intriguing. Several lines of evidence indicate
that p53 may function in the maturation of
spermatocytes and of lymphoid cells (Rotter et al.,
1993, 1994; Shick et al., 1997). Moreover there is
evidence that spatial and cyclical expression of p53 in
the testis plays a role in the meiotic process of
spermatogenesis (Almon et al., 1993). In order to
allow cell-type speci®c functions it may be necessary
to uncouple p53 from mdm2-mediated nuclear export
and degradation. This could be achieved by expression
of mdm2a mRNA since it contains the a exon and
would encode mdm2 protein products unable to
associate with p53. Such a model might explain the
high levels of mdm2a mRNA observed in testicular
and lymphoid cells compared with other cell types. In
addition, mdm2a protein may have intrinsic functions
speci®c to these cell types (see above). These
considerations lead us to suggest that the a exon of
mdm2 may represent an important determinant of
mdm2 function in normal testicular and lymphoid
cells.
Materials and methods
RNA isolation and RT ± PCR ampli®cation of mdm2 cDNA
sequences
Total RNA was prepared from human blood lymphocytes
and various canine tissues using the RNeasy kit as per the
manufacturer's protocol (Qiagen). In some cases, messenger
RNA was further puri®ed using the Oligotex mRNA kit as
per the manufacturer's protocol (Qiagen). Ampli®cation of
mdm2 cDNA by reverse transcriptase-PCR (RT ± PCR) from
human and canine total RNA was carried out using the
degenerate mdm2-speci®c primers 5'MDM (5'-GCGGTACCAGGCM AATGTGCAATACCAAC-3') and 3'MDM (5'GCGAATTCAGGTCARCTAGKK GAARTAASTTAG-3')
and the Access RT ± PCR kit as per the manufacturer's
protocol (Promega).
Cloning of mdm2 cDNA sequences from human and canine
tissues
Human peripheral blood lymphocytes were obtained from
blood samples from healthy volunteers. Samples were diluted
1 : 1 in sterile PBS and lymphocytes puri®ed on Ficoll
according to standard methods. Canine leukocytes were
obtained from pre-operative blood samples and prepared as
above. The fractionated lymphoid cells were cryo-preserved
in 10% DMSO, 50% FCS and 40% RPMI until required.
Solid tissues of canine origin were obtained during routine
surgical resection of tumour mass and normal tissue
surrounding the tumour material. Canine tissue was stored
at 7808C until required for RNA extraction. Mdm2 cDNA
ampli®ed from human or canine mRNA was cut with KpnI
and EcoRI and ligated into the same sites in pBluescript
SK+. The plasmid pHDM2 contained human mdm2 cDNA
isolated from blood lymphocytes, the plasmid pCDM2
contained canine mdm2 cDNA derived from a benign
mammary tumour, and the plasmid pCDM2a contained the
mdm2a cDNA isolated from canine peripheral blood
leukocytes. Two canine mdm2a mutants were generated by
DNA ampli®cation and cloning. A deletion mutant lacking
the ®rst 90 codons, called mdm2DN (plasmid pCDM2DN),
was constructed using the primers DH/Cup (5'GGCGGTACCAG TATATTATGACTAAACG-3') and
3'CDM (5'-CGCGGAATTCAGGTCAAC TAGGGGAAATAAGTTAG-3') and cloned into the KpnI and EcoRI sites of
pBluescript SK+. A `read-through' mutant, called mdm2RT
(plasmid pCDM2RT), was constructed using the primers
CDM2RT-(5'-GGCGGATCCCAGGTTAGAACTTCT-CACTAGAGATACAGCAGTACAGTATAC-3') and 3'CDM
which contained an alanine codon substitution at UGA71 in
mdm2a. Plasmid pK9 contains canine p53 cDNA cloned into
pLitmus 29 (Veldhoen and Milner, 1998). All cDNA
constructs were con®rmed by manual DNA sequencing
using a T7 DNA polymerase-based sequencing kit as per
the manufacturer's protocol (Pharmacia). Plasmids were
maintained in E. coli XL-1 Blue MRF'.
Identi®cation of exon-intron boundaries ¯anking the a exon
within mdm2
DNA ampli®cation was performed using genomic DNA
isolated from human and canine blood. The canine primer
pairs included 98.14 (5'-GGCGGTACCAAAAAG ACACT-
7031
A novel mdm2 exon modulates translation
N Veldhoen et al
7032
TATACTATG-3')/cMDM2ad n (5'-GGGATCCATGGATGCCCAAGAAGTC-3') and cMDM2a u p (5'-GGGCATCCATGGATCCCAGGTTAAG-3')/98.10 (5'-GGCAGAATTCGTTTAGTCATAATATACTGG-3'). The human primer
pairs included 98.14/hMDM2adn (5'-TTCCTGGGATCCAGGGATGCCCAAG-3') and hMDM2aup (5'-GGGCATCCCTGGATCCCAGGTTAAG-3')/98.10. The 30 ml amplification reactions contained 16PC2 bu€er (50 mM Tris-HCl
(pH 9.1), 16 mM ammonium sulphate, 3.5 mM MgCl2, and
150 mg/ml BSA, 0.25 mM of each primer, 166 mM dNTPs
(dATP, dCTP, dGTP, dTTP), 250 ng genomic DNA, and
12.5 units Taq Supreme (Helena Biosciences). The thermocycle program included a denaturation step at 948C (7 min),
40 cycles of 948C (60 s), 588C (60 s), and 688C (3 min) and a
®nal elongation step at 688C (7 min). Ampli®ed DNA
products were cloned into pBluescript SK+ and manually
sequenced using a T7 DNA polymerase-based sequencing kit
(Pharmacia).
Detection of mdm2a mRNA expression in di€erent tissues
Mdm2a expression within various canine tissues was assessed
by performing mdm2-speci®c RT ± PCR on total RNA or
puri®ed mRNA, as detailed above, followed by secondary
nested PCR using the primer 98.4 (5'-GGCGGTACCAATGA AAGAGGACTTCTTGGG-3') and the degenerate
primer IIDN (5'-CCAGGCCTYACG AAGGGYCCARCATCTNTTRCA-3'). Mdm2 control reactions were performed using primers 98.5 (5'-TATTATGACTAAACGATTG-3') and IIDN. The 20 ml reactions contained 16PC2
bu€er, 0.25 mM of each primer, 200 mM dNTPs (dATP,
dCTP, dGTP, dTTP), 1 ml initial RT ± PCR sample, and 0.05
units Taq Supreme (Helena Biosciences). The thermocycle
program included a denaturation step at 948C (7 min), 35
cycles of 948C (30 s), 588C (30 s), and 688C (30 s) and a ®nal
elongation step at 688C (7 min).
Expression of mdm2 proteins in vitro
Plasmids pHDM2, pCDM2, pCDM2a, pCDM2DN,
pCDM2RT, and pK9 were linearized with EcoRI. Each
50 ml transcription reaction contained 40 mM Tris-HCl pH
7.5, 6 mM MgCl2, 2 mM spermidine, 10 mM dithiothreitol, 20
units RNasin (Promega), 0.5 mM each ATP, CTP and UTP,
0.025 mM GTP, 2 mg template DNA, 0.5 mM m7GpppG, and
40 units T7 RNA polymerase (Promega). Transcription was
allowed to proceed at 378C for 30 min and GTP was added
to a ®nal concentration of 1 mM. A further incubation of
60 min at 378C was performed followed by phenol:chloroform extraction, ethanol precipitation, and resuspension in 40 ml of water. Further puri®cation of mRNA
transcripts was performed using the RNeasy kit as per the
manufacturer's protocol (Qiagen). Human and canine mdm2
proteins and canine p53 protein were translated in a rabbit
reticulocyte lysate system (Promega) as described previously
by Gamble and Milner (1998) and Veldhoen and Milner
(1998). The rate and eciency of translation was determined
by TCA precipitation of 2 ml of each reaction on glass ®lters
followed by scintillation counting. Translated protein
products were separated on a 12% SDS-polyacrylamide gel
and visualized by autoradiography with Fuji RX ®lm at
room temperature.
Immunoprecipitation
Protein conformation was determined using an immunoprecipitation method as described by Cook and Milner (1990).
The following anti-mdm2 antibodies were used; 10 ml
hybridoma supernatants containing antibodies 4B2, 3G5,
2A9, 2A10, and 4B11 and 1 ml puri®ed SMP14 monoclonal
antibody (Oncogene Science). Antibody PAb416 (30 ml),
directed towards the large T-antigen of SV40, was used as
a negative control. For co-immunoprecipitation experiments,
canine p53 protein and canine and human mdm2 proteins
were co-expressed in rabbit reticulocyte lysate. Complexes
were immunoprecipitated using antibody 4B11 directed to
mdm2-related proteins or antibody 421 directed towards
canine p53 protein. Immunoprecipitated proteins were
resolved by 15% SDS-polyacrylamide gel electrophoresis
and visualized by autoradiography with Fuji RX ®lm at
room temperature.
Acknowledgements
We thank Arnold Levine and Stephen Picksley for making
available the anti-mdm2 antibodies 4B2, 3G5, 2A9, 2A10,
and 4B11. We are also indebted to Andrei Okorokov and
Ming Jiang, YCR p53 Research Group, for many helpful
discussions. Photographic work was prepared by Meg
Stark. This work was supported by a grant from
Yorkshire Cancer Research to J Milner.
References
Almon E, Gold®nger N, Kapon A, Schwartz D, Levine AJ
and Rotter V. (1993). Dev. Biol., 156, 107 ± 116.
Barak Y, Gottlieb E, Juven-Gershon T and Oren M. (1994).
Genes Dev., 8, 1739 ± 1749.
BoÈttger A, BoÈttger V, Garcia-Echeverria C, CheÁne P,
Hochkeppel H-K, Sampson W, Ang K, Howard SF,
Picksley SM and Lane DP. (1997). J. Mol. Biol., 269, 744 ±
756.
Bruening W and Pelletier J. (1996). J. Biol. Chem., 271,
8646 ± 8654.
Cahilly-Snyder L, Yang-Feng T, Francke U and George DL.
(1987). Somat. Cell Mol. Genet., 13, 235 ± 244.
Chen J, MareÂchal V and Levine AJ. (1993). Mol. Cell. Biol.,
13, 4107 ± 4114.
Cook A and Milner J. (1990). Br. J. Cancer, 61, 548 ± 552.
Elenbaas B, Dobbelstein M, Roth J, Shenk T and Levine AJ.
(1996). Mol. Med., 2, 439 ± 451.
Fakharzadeh SS, Trusko SP and George DL. (1991). EMBO
J., 10, 1565 ± 1569.
Freedman DA, Epstein CB, Roth JC and Levine AJ. (1997).
Mol. Med., 3, 248 ± 259.
Gamble J and Milner J. (1988). Virology, 162, 452 ± 458.
Gellersen B, Kempf R and Telgmann R. (1997). Mol.
Endocrin., 11, 97 ± 113.
Gudas JM, Nguyen H, Klein RC, Katayose D, Seth P and
Cowan KH. (1995). Clin. Cancer Res., 1, 71 ± 80.
Haines DS. (1997). Leuk. Lymph., 26, 227 ± 238.
Haines DS, Landers JE, Engle LJ and George DL. (1994).
Mol. Cell. Biol., 41, 1171 ± 1178.
Haupt Y, Maya R, Kazaz A and Oren M. (1997). Nature,
387, 296 ± 299.
Honda R, Tanaka H and Yasuda H. (1997). FEBS Lett., 420,
25 ± 27.
Honda R and Yasuda H. (1999). EMBO J., 18, 22 ± 27.
Jones SN, Ansari-Lari MA, Hancock AR, Jones WJ, Gibbs
RA, Donehower LA and Bradley A. (1996). Gene, 175,
209 ± 213.
Juven-Gershon T, Shifman O, Unger T, Elkeles A, Haupt Y
and Oren M. (1998). Mol. Cell. Biol., 18, 3974 ± 3982.
Kaufman RJ, Murtha P and Davies MV. (1987). EMBO J.,
6, 187 ± 193.
A novel mdm2 exon modulates translation
N Veldhoen et al
Kraus A, Ne€ F, Behn M, Schuermann M, Muenkel K and
Schlegel J. (1999). Int. J. Cancer, 80, 930 ± 934.
Kubbatat MHG and Vousden KH. (1998). Mol. Med., 4,
250 ± 256.
Kussie PH, Gorina S, MareÂchal V, Elenbaas B, Moreau J,
Levine AJ and Pavletich NP. (1996). Science, 274, 948 ±
953.
Leng P, Brown DR, Shivakumar CV, Deb S and Deb SP.
(1995). Oncogene, 10, 1275 ± 1282.
Lundgren K, Montes de Oca Luna R, McNeill YB, Emerick
EP, Spencer B, Bar®eld CR, Lozano G, Rosenberg MP
and Finlay CA. (1997). Genes Dev., 11, 714 ± 725.
Luukkonen BG, Tan W and Schwartz S. (1995). J. Virol., 69,
4086 ± 4094.
MareÂchal V, Elenbaas B, Piette J, Nicolas JC and Levine AJ.
(1994). Mol. Cell. Biol., 14, 7414 ± 7420.
MareÂchal V, Elenbaas B, Taneyhill L, Piette J, Mechali M,
Nicolas J-C, Levine AJ and Moreau J. (1997). Oncogene,
14, 1427 ± 1433.
Martin K, Trouche D, Hagemeier C, Sùrenson TS, La
Thangue NB and Kouzarides T. (1995). Nature, 375, 691 ±
694.
Matsumoto R, Tada M, Nozaki M, Zhang CL, Sawamura Y
and Abe H. (1998). Cancer Res., 58, 609 ± 613.
Maxwell SA. (1994). Anticancer Res., 14, 2541 ± 2548.
McCaughan KK, Brown CM, Dalphin ME, Berry MJ and
Tate WP. (1995). Proc. Natl. Acad. Sci. USA, 92, 5431 ±
5435.
Momand J, Jung D, Wilczynski S and Niland J. (1998). Nucl.
Acids Res., 26, 3453 ± 3459.
Momand J and Zambetti GP. (1997). J. Cell. Biochem., 64,
343 ± 352.
Momand J, Zambetti GP, Olson DC, George D and Levine
AJ. (1992). Cell, 69, 1237 ± 1245.
Montes de Oca Luna R, Tabor AD, Eberspaecher H, Hulboy
DL, Worth LL, Colman MS, Finlay CA and Lozano G.
(1996). Genomics, 33, 352 ± 357.
Montes de Oca Luna R, Wagner DS and Lozano G. (1995).
Nature, 378, 203 ± 206.
Oliner JD, Kinzler KW, Meltzer PS, George DL and
Vogelstein B. (1992). Nature, 358, 80 ± 83.
Oliner JD, Pietenpol JA, Thiagalingam S, Gyuris J, Kinzler
KW and Vogelstein B. (1993). Nature, 362, 857 ± 860.
Olson DC, Marechal V, Momand J, Chen J, Romocki C and
Levine AJ. (1993). Oncogene, 8, 2353 ± 2360.
Padgett RA, Grabowski PJ, Konarska MM, Seiler S and
Sharp PA. (1986). Annu. Rev. Biochem., 55, 1119 ± 1150.
Piette J, Neel H and MareÂchal V. (1997). Oncogene, 15,
1001 ± 1010.
Roth R, Dobbelstein M, Freedman DA, Shenk T and Levine
AJ. (1998). EMBO J., 17, 554 ± 564.
Rotter V, Aloni-Grinstein R, Schwartz D, Elkind NB,
Simons A, Wolkowicz R, Lavigne M, Beserman P,
Kapon A and Gold®nger N. (1994). Semin. Cancer Biol.,
5, 229 ± 236.
Rotter V, Schwartz D, Almon E, Gold®nger N, Kapon A,
Meshorer A, Donehower LA and Levine LA. (1993). Proc.
Natl. Acad. Sci. USA, 90, 9075 ± 9079.
Saucedo LJ, Myers CD and Perry ME. (1999). J. Biol. Chem.,
274, 8161 ± 8168.
Sigalas I, Calvert AH, Anderson JJ, Neal DE and Lunec J.
(1996). Nat. Med., 2, 912 ± 917.
Shick L, Carman JH, Choi JK, Somasundaram K, Burrell M,
Hill DE, Zeng Y-X, Wang Y, Wiman KG, SalhanyK,
Kadesch TR, Monroe JG, Donehower LA and El-Deiry
WS. (1997). Cell Growth Di€., 8, 121 ± 131.
Sun P, Dong P, Dai K, Hannon GJ and Beach D. (1998).
Science, 282, 2270 ± 2272.
Veldhoen N and Milner J. (1998). Oncogene, 16, 1077 ± 1084.
Xiao Z-X, Chen J, Levine AJ, Modjtahedi N, Xing J, Sellers
WR and Livingston DM. (1995). Nature, 375, 694 ± 698.
7033