(Received for publication, September 25, 1996, and in revised form, December 31, 1996)
From the Department of Physiology and the Department of Internal Medicine, Ohio State University, Columbus, Ohio 43210
The PTC1 chimeric oncogene is
generated by the fusion of the tyrosine kinase domain of the
RET proto-oncogene to the 5-terminal region of another
gene named H4 (D10S170). This oncogene has been detected
only in human papillary thyroid carcinomas. We have previously demonstrated that the putative leucine zipper in the N-terminal region
of H4 can mediate oligomerization of the PTC1 oncoprotein in
vitro. In this study, we further demonstrated that the PTC1 oncoprotein forms a dimer in vivo, and the leucine zipper
is responsible for this dimerization. The H4 leucine zipper-mediated
dimerization is essential for tyrosine hyperphosphorylation and the
transforming activity of the PTC1 oncoprotein. Introducing a
loss-of-function PTC1 mutant into PTC1-transformed NIH3T3
cells suppressed the transforming activity of PTC1 and reversed the
transformed phenotype of these cells, presumably by forming inactive
heterodimers between the two forms of PTC1. Taken together, these data
indicate that constitutive dimerization of the PTC1 oncoprotein is
essential for PTC1 transforming activity and suggest that constitutive
oligomerization acquired by rearrangement or by point mutations may be
a general mechanism for the activation of receptor tyrosine kinase
oncogenes.
The PTC1 chimeric oncogene, formed by intrachromosomal rearrangement between H4 (D10S170) and the RET proto-oncogene, has been detected in 2.5-30% of papillary thyroid carcinomas (1-3). The product of the PTC1 oncogene is a fusion protein containing the N terminus of H4 fused to the tyrosine kinase domain of c-RET (1). The H4 gene shows no significant homology to known genes, and the function of H4 protein is unknown (4). The RET proto-oncogene encodes a receptor-type tyrosine kinase (5, 6), whose ligand has been identified as glial cell line-derived neurotrophic factor (7-9). A variety of mutations involving the RET proto-oncogene have been found to be associated with a number of human neuroendocrine diseases, such as multiple endocrine neoplasia type 2 inherited cancer syndromes (10-12), the congenital developmental defect Hirschsprung's disease (13), and some sporadic medullary thyroid carcinomas (14).
The PTC1 oncoprotein has been demonstrated to be hyperphosphorylated (15) and to exert transforming activity in NIH3T3 cells (16). The causative role of the PTC1 oncogene in human papillary thyroid carcinoma is further supported by the fact that our transgenic mouse model with targeted expression of the PTC1 oncogene in the thyroid gland develops papillary thyroid carcinomas (17). However, the molecular mechanism of PTC1 activation in the development of papillary thyroid tumors has yet to be elucidated. We have previously demonstrated that the PTC1 chimeric oncogene shows unscheduled expression in the thyroid follicular cells and that recombinant proteins containing the putative leucine zipper domain of H4 form oligomeric complexes in vitro (18). As dimerization is considered to be a crucial step for receptor tyrosine kinase activation (19, 20), we hypothesized that both unscheduled expression of RET tyrosine kinase and constitutive oligomerization of PTC1 proteins are responsible for PTC1-transforming activity in the thyroid.
In this study, we further demonstrated that the leucine zipper region of H4 is responsible for the dimerization of the PTC1 oncoprotein in vivo. Our data also indicated that the leucine zipper-mediated dimerization is essential for tyrosine hyperphosphorylation and the transforming activity of PTC1. Furthermore, the transforming activity of PTC1 can be suppressed by introducing a loss-of-function mutant of PTC1 into PTC1-transformed NIH3T3 cells.
The African green monkey kidney cell line, COS-7 cells (ATCC 1651), was maintained in Dulbecco's modified Eagle's medium with 10% fetal bovine serum. Both NIH3T3 and NIH3T3/PTC1 were maintained in Dulbecco's modified Eagle's medium with 10% donor calf serum. All media contain 100 units/ml penicillin and 100 µg/ml streptomycin. The monoclonal antibody MSJ was generated in our laboratory. Its epitope was tentatively mapped to amino acid residues 824-828 of RET.1 The polyclonal antibody, C17, was raised against a synthetic peptide (CKRRDYLDLAASTPSDSL) located at amino acid residues 1011-1027 of the C terminus of the RET tyrosine kinase (Immuno-Dynamics, Inc.). The C17 antibody was purified through an affinity column covalently bound with the synthetic peptide. The monoclonal antibody against phosphotyrosine (4G10) was obtained from Upstate Biotechnology Inc.
Plasmid ConstructsThe PTC1/CMV was constructed by excising
the PTC1 cDNA from the plasmid TPC-1 (15) and inserting it into the
XbaI and ApaI sites of pRc/CMV (Invitrogen). The
PTC1/LNCX was obtained by excising the PTC1 fragment from
PTC1/CMV with HindIII and inserting it into the
HindIII site of pLNCX. To clone PTC1zip/CMV and
PTC1
zip/LNCX, the T7 primer and another primer
(5
-AACGGAACGGCGAGATGA-3
), which is located to a
region just ahead of the leucine zipper, were used to perform
polymerase chain reaction using PTC1/CMV as DNA template. The 250-base
pair polymerase chain reaction product was digested with
BamHI and used to replace the 400-base pair BamHI
fragment of H4 in PTC1/CMV and PTC1/LNCX. To clone PTC1
N/CMV and
PTC1
N/LNCX, a sense primer
(5
-TATGCCATGGCGCCGTTCCGCCTG-3
), which is
located at the beginning of the leucine zipper region, was paired
with an antisense primer (5
-AGTTCTTCCGAGGGAATTCC-3
), which is located
at the beginning of the RET tyrosine kinase domain, to perform
polymerase chain reaction on the PTC1/CMV template. A 200-base pair
BamHI fragment was excised from the polymerase chain
reaction product and used to replace the 400-base pair H4 fragment of PTC1/CMV and PTC1/LNCX. The PTC1
C/CMV was generated by
unidirectional deletion of the PTC1 insert, starting from
the C terminus of PTC1. A mutation (Arg897
Glu), found
in some patients with Hirschsprung's disease (13), was first
introduced into c-RET cDNA by site-directed mutagenesis (Muta-gene M13 in vitro mutagenesis kit, Bio-Rad) and later
subcloned into PTC1/CMV and PTC1
zip/CMV to obtain PTC1HS/CMV and
PTC1
zipHS/CMV.
For transient transfection, 5 × 105 COS-7 cells were transfected with 10-20 µg of various DNA constructs using the calcium phosphate transfection kit (Life Technologies, Inc.). For the focus formation assay, 1.3 × 105 NIH3T3 cells were transfected with 0.3 µg of various DNA constructs. Three weeks later, the foci were stained with Giemsa and counted. For NIH3T3 stable transfectants, 4 × 104 NIH3T3 cells were transfected with 1 µg of various DNA constructs, and the G418-resistant colonies were screened for PTC1 expression by immunoprecipitating the cell lysates with the antibody C17, followed by immunoblot analysis with the antibody MSJ.
Immunoprecipitation and Immunoblot AnalysisCells were lysed in 20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 2 mM EDTA, 1% Nonidet P-40, and 0.5 mM Na3VO4. For immunoprecipitation, cell lysates were incubated at 4 °C overnight with C17 polyclonal antibody that was covalently conjugated to protein A beads (21). For the immunoblot assay, proteins were separated by SDS-PAGE2 and then transferred to a nitrocellulose membrane. The membrane was blocked in TBST buffer (10 mM Tris-HCl, pH 7.5, 100 mM NaCl, and 0.1% Tween 20) containing 5% nonfat dry milk at 4 °C overnight. After incubation with the primary antibody for 1 h and horseradish peroxidase-conjugated anti-mouse IgG (Transduction Laboratories) for an additional hour, the membrane was treated with ECL reagent (Amersham Corp.) and exposed to x-ray film.
Cross-linking was carried out at room temperature with 0.01% glutaraldehyde in 50 mM triethanolamine (pH 8.2) and 100 mM NaCl, using 20-50 µl of lysate of COS-7 cells transfected with various DNA constructs. Aliquots were removed at the times indicated, and the reaction was terminated by boiling the mixture in SDS-PAGE sample buffer. The sample was resolved by SDS-PAGE and detected by immunoblot with the anti-RET antibody MSJ.
Soft Agar AssayThe soft agar assay was carried out as follows. A bottom layer of agar was prepared in 60-mm Petri dishes using 3 ml of 0.5% noble agar (Difco) in normal growth medium. Next, 3 ml of 0.35% noble agar in normal growth medium containing 9 × 103 cells were added on top of the solidified bottom layer. Colonies with anchorage-independent growth were counted 2 weeks later.
To investigate the
oligomerization status of PTC1 oncoprotein in eukaryotic cells, the
PTC1 oncoprotein was expressed in COS-7 cells by transient transfection
with a DNA construct PTC1/CMV. The lysates of COS-7 transfectants
expressing PTC1 were treated with glutaraldehyde to stabilize the
oligomeric protein complexes (Fig. 1b).
Without cross-linking treatment, PTC1 oncoprotein can be detected at
the size of 53 kDa, since the PTC1 used in our experiment is the
alternative spliced isoform with a shorter C terminus (6). After 1 min
of treatment with glutaraldehyde, a 105-kDa protein complex can be
readily detected. By 16 min, the PTC1 monomer can no longer be
detected. We consistently observed that the PTC1 protein complexes had
a stronger signal intensity than the PTC1 monomers in immunoblot
detection. One possible explanation is that the formation of protein
complexes may alter the conformation of the RET tyrosine kinase for
better binding of the MSJ antibody to the epitope site.
To investigate the role of the leucine zipper of H4 in the formation of
PTC1 protein complexes, two forms of PTC1 mutant, PTC1zip and
PTC1
N, were constructed. PTC1
zip had the leucine zipper region of
H4 (amino acid residues 56-102) deleted. PTC1
N had the N terminus
of H4 (amino acid residues 3-52) deleted but retained the leucine
zipper (Fig. 1a). As shown in Fig. 1b, the PTC1
zip (46 kDa) fails to form protein complexes even after a 16-min
treatment with glutaraldehyde. In contrast, PTC1
N (49 kDa) retains
the ability to form protein complexes. These data indicate that the
formation of PTC1 protein complex is mediated by the leucine zipper of
H4.
A co-immunoprecipitation assay was performed to further investigate
whether the PTC1 forms homodimers. A mutated form of PTC1, PTC1C,
had the extreme C terminus of PTC1 (amino acid residues 345-461)
deleted. Therefore, PTC1
C is no longer recognized by the polyclonal
antibody C17 (Fig. 2a). When the PTC1
C was
co-expressed in COS-7 cells with either PTC1, PTC1
zip, or PTC1
N,
the PTC1
C-encoded protein cannot be immunoprecipitated by
C17 unless it forms protein complexes with other forms of PTC1
containing an intact C terminus. The expression of PTC1 and its mutants
in various cell lysates is detected by the MSJ monoclonal antibody,
which can react with all forms of PTC1 protein including PTC1
C (Fig.
2b). The PTC1
C product (38 kDa) cannot be
immunoprecipitated by C17 when it was expressed alone or co-expressed
with PTC1
zip (Fig. 2c, lanes 2 and
4), but can be co-immunoprecipitated when it was co-expressed with
PTC1 or PTC1
N (Fig. 2c, lanes 3 and
5). From the results of the cross-linking experiment and the
co-immunoprecipitation assay we conclude that PTC1 oncoprotein forms a
homodimer in vivo, and the dimerization is mediated by the
leucine zipper domain of H4.
The H4 Leucine Zipper-mediated Dimerization Is Essential for Tyrosine Hyperphosphorylation of the PTC1 Oncoprotein
It has been
shown that dimerization of receptor tyrosine kinases promotes the
transphosphorylation of tyrosine residues on intracellular domains
(20). To investigate whether there is a correlation between
dimerization and transphosphorylation, the tyrosine phosphorylation
levels of PTC1, PTC1zip, and PTC1
N proteins expressed in COS-7
cells were examined (Fig. 3). The equivalent amount of
PTC1, PTC1
zip, and PTC1
N proteins used in the immunoblot analysis
was also shown. The PTC1 and PTC1
N oncoprotein are
hyperphosphorylated on tyrosine residues. However, deletion of the
leucine zipper region of PTC1 dramatically reduced the tyrosine
phosphorylation level of the PTC1
zip-encoded protein.
The H4 Leucine Zipper-mediated Dimerization Is Required for the PTC1 Transforming Activity
To correlate the leucine
zipper-mediated dimerization and tyrosine hyperphosphorylation with the
transforming activity of PTC1, both focus formation assay and soft agar
assay were performed to evaluate the transforming activity of PTC1,
PTC1zip, or PTC1
N. The expression of PTC1, PTC1
zip, or
PTC1
N proteins encoded by the corresponding DNA constructs was
initially confirmed by transient transfection in COS-7 cells, and the
expression levels of these three proteins did not appear to be
different in COS-7 cells. For focus formation assay, NIH3T3 fibroblast
cells were transfected with PTC1/LNCX, PTC1
zip/LNCX, or
PTC1
N/LNCX DNA constructs. The transfected NIH3T3 cells were either
left in the plates for 3 weeks to score for the number of foci formed
(loss of contact inhibition) or were under G418 selection to monitor
the transfection efficiency. As shown in Table I,
although all DNA constructs showed equivalent transfection efficiency,
NIH3T3 cells transfected with PTC1/LNCX or PTC1
N/LNCX, but not with
PTC1
zip/LNCX or vector DNA, lost contact inhibition and displayed
the ability to form foci.
|
For soft agar assay, NIH3T3 stable transfectants expressing PTC1,
PTC1zip, or PTC1
N were established. We consistently observed that
the protein expression levels of PTC1 or PTC1
N were higher than
those of PTC1
zip in corresponding NIH3T3 stable transfectants (Fig.
4). The clones expressing the lowest amount of PTC1
(PTC1-5 and PTC1-10), a clone expressing the highest amount of
PTC1
zip (PTC1
zip-3), and clones expressing both high (PTC1
N-2)
and low (PTC1
N-1) amounts of PTC1
N were selected for soft agar
assay. The result of soft agar assay demonstrated that NIH3T3 cells
expressing PTC1 or PTC1
N, but not PTC1
zip, acquired
anchorage-independent growth in soft agar (Table
II).
|
Although no difference in the ability to form foci was observed between
cells expressing PTC1 and cells expressing PTC1N (Table I), fewer
colonies were formed in soft agar for PTC1
N-expressing cells
compared with PTC1-expressing cells (Table II). Even though two
different clones of both PTC1 and PTC1
N were
tested by soft agar assay, the clonal effect cannot be completely
excluded. Alternatively, the N-terminal region of H4 may contain some
elements contributing to the anchorage-independent growth of PTC1 but
not to the loss of contact inhibition. Taken together, the results of
focus formation assay and soft agar assay indicate that the leucine
zipper-mediated dimerization is required for the PTC1 transforming
activity.
Since intermolecular phosphorylation of dimerized
receptor tyrosine kinases is crucial for the PTC1 activation, it is
possible to suppress the PTC1 transforming activity by introducing a
loss-of-function PTC1 mutant to form inactive heterodimers with PTC1. A
mutation at codon 897 of c-RET, changing arginine to
glutamine, was identified in some patients with Hirschsprung's disease
(13). This mutant has been characterized as a loss-of-function
mutation (22). PTC1HS, a PTC1 mutant containing this
mutation, was transfected into the NIH3T3/PTC1 cells, which were
established by transfecting NIH3T3 cells with human thyroid tumor
genomic DNA that contains the naturally occurring PTC1
transforming gene (1). Since the NIH3T3/PTC1 cells do not have G418
resistance, NIH3T3/PTC1 cells transfected with various DNA constructs
can be selected by G418, and the transformed phenotype of the
G418-resistant colonies was evaluated by their ability to form foci. As
shown in Fig. 5, approximately 68.6% of the mock
transfected (pRc/CMV vector only) NIH3T3/PTC1 cells and 52.5% of the
PTC1zipHS/CMV-transfected NIH3T3/PTC1 cells retained the ability to
form foci. When analyzed by Student's t test, there was no
significant difference between the effects of PTC1
zipHS and mock
transfection. However, only 18.5% of PTC1HS/CMV-transfected NIH3T3/PTC1 cells maintained transformed phenotype, which was significantly different from that of NIH3T3/PTC1 cells transfected with
pRc/CMV vector (p < 0.001) or cells transfected with
PTC1
zipHS/CMV (p < 0.05).
In this study, we demonstrated that the PTC1 oncoprotein forms dimers in vivo and the dimerization of PTC1 is mediated by the leucine zipper in the H4 portion of PTC1. Our data also indicated that the leucine zipper-mediated dimerization was essential for the constitutive phosphorylation and the transforming activity of the PTC1 oncoprotein. Furthermore, a loss-of-function PTC1 mutant can reverse the transformed phenotype of NIH3T3/PTC1 cells, presumably by forming inactive heterodimers.
The PTC1 oncoprotein, with the extracellular and transmembrane domains
of c-RET replaced by H4, is localized in the cytoplasm (15) instead of
the plasma membrane of cells. It was proposed that the translocation of
PTC1 from the cell membrane to the cytoplasm could have provided the
structural basis for its escape from the modulatory effect of the
membrane-associated protein kinase C (23). However, our data indicate
that loss of membrane localization is not enough for the activation of
the RET tyrosine kinase, since PTC1zip, which is also localized in
the cytoplasm, lost both tyrosine hyperphosphorylation and transforming
activity. Our data indicated that the leucine zipper-mediated
dimerization is required for the activation of PTC1 tyrosine
kinase.
In human papillary thyroid carcinomas, three different forms of the PTC oncogenes have been identified in which the RET tyrosine kinase domain becomes fused with the N-terminal sequences of three different genes (1, 24-26). In addition to PTC1, the PTC2 oncoprotein has also been shown to form dimers in vivo (24), and the dimerization is required for the mitogenic activity of PTC2 (27). Although PTC3 has not been shown to form oligomers experimentally, a potential coiled-coil motif was identified in the ELE1 sequence of PTC3.1 In addition to PTC oncogenes, RETI and RETII transforming genes, which were formed by rearrangements during in vitro transfection (28, 29), also form dimers in vivo (30).1 Therefore, constitutive oligomerization of RET appears to be a common mechanism for the activation of rearranged RET oncoproteins. In fact, constitutive oligomerization may serve as a common mechanism for oncogenic activation of other receptor tyrosine kinases in thyroid tumors. The TPR protein, which is involved in a rearrangement with the tyrosine kinase domain of the TRK nerve growth factor receptor in human papillary thyroid carcinoma (31), contains a leucine zipper domain and forms dimers in vivo (32). All of the tyrosine kinase oncogenes formed by rearrangement in human papillary carcinomas seem to be activated by constitutive oligomerization.
Rodrigues and Park (32) have proposed that oligomerization may serve as a general mechanism for oncogenic activation of receptor tyrosine kinases in many types of tumors. They demonstrated that the leucine zipper domain in TPR mediates dimerization of the TPR-MET oncoprotein, and this constitutive dimerization is essential for its transforming activity. Not only can rearrangement cause oligomerization to activate receptor tyrosine kinases, but point mutations have also been shown to promote constitutive dimerization of receptor tyrosine kinases. Point mutations of RET found in MEN2A patients, resulting in the substitution of one of five Cys residues in the extracellular domain, caused constitutive dimerization and activation of RET tyrosine kinase (33, 34). An amino acid substitution of the NEU oncogene has also been shown to induce oncoprotein aggregation, and increase tyrosine kinase activity and transforming activity (35). Therefore, the approach of using a loss-of-function mutant to suppress the transforming activity of the corresponding oncogene provides a new strategy to specifically intervene with the oncogenic activity of various receptor tyrosine kinase oncogenes in a variety of tumors.
We thank Dr. William T. Wong for the NIH3T3 cells, Dr. Massimo Santoro for the NIH3T3/PTC1 cells, and Dr. Benigno C. Valdez for the pLNCX vector. We are grateful to Drs. William T. Wong, Massimo Santoro, and James Lang for valuable suggestions and help.