Download Alternate choice of initiation codon produces a biologically

Oncogene (1999) 18, 1529 ± 1535
ã 1999 Stockton Press All rights reserved 0950 ± 9232/99 $12.00
http://www.stockton-press.co.uk/onc
Alternate choice of initiation codon produces a biologically active product
of the von Hippel Lindau gene with tumor suppressor activity
Catherine Blankenship1,4, Joseph G Naglich2, Jean M Whaley3, Bernd Seizinger1 and
Nikolai Kley*,1
1
Department of Functional Genomics, Genome Therapeutics Corporation, 100 Beaver Street, Waltham, Massachusetts 02154,
USA; 2Department of Oncology, Bristol-Myers Squibb Pharmaceutical Research Institute, Princeton, New Jersey 08540, USA;
3
Department of Metabolic Diseases, Bristol-Myers Squibb Pharmaceutical Research Institute, Princeton, New Jersey 08540 USA
The VHL tumor suppressor gene has previously been
reported to encode a protein of 213 amino acid residues.
Here we report the identi®cation of a second major VHL
gene product with an apparent molecular weight of
18 kD, pVHL18, which appears to arise from alternate
translation initiation at a second AUG codon (codon 54)
within the VHL open reading frame. In vitro and in vivo
studies indicate that the internal codon in the VHL
mRNA is necessary and sucient for production of
pVHL18. pVHL18 can bind to elongin B, elongin C, and
Hs-CUL2. When reintroduced into renal carcinoma cells
that lack a wild-type VHL allele, pVHL18 suppresses
basal levels of VEGF expression, restores hypoxiainducibility of VEGF expression, and inhibits tumor
formation in nude mice. These data strongly support the
existence of two distinct VHL gene products in VHL
tumor suppression.
Keywords: pVHL; VEGF; elongins; Hs-CUL2
Introduction
Inactivation or loss of the VHL gene predisposes
a€ected individuals to various neoplasias associated
with von Hippel Lindau disease (Latif et al., 1993),
characteristically retinal angiomas, hemangioblastomas
of the cerebellum and spinal cord, pheochromocytomas, and renal cell carcinomas (Zbar, 1995,
Maher and Kaelin, 1997). In addition, somatic
inactivation of the VHL gene has been detected in
sporadic renal cell carcinomas and sporadic hemangioblastomas (Zbar, 1995; Gnarra et al., 1994; Shuin et
al., 1994; Whaley et al., 1994; Kanno et al., 1994;
Herman et al., 1994). Most notably, inactivation of the
VHL gene by mutation or hypermethylation has been
detected in approximately 80% of sporadic clear cell
renal carcinomas. Restoration of wild-type VHL
function in VHL mutant renal carcinoma cell lines
inhibits their ability to form tumors in nude mice
(Iliopoulos et al., 1995). These ®ndings attribute the
VHL gene a major role in the development of the most
common form of human kidney cancers.
*Correspondence: N Kley
4
Current address: Department of Molecular Biology, Princeton
University, Princeton, NJ 08540, USA
Received 13 July 1998; revised 25 September 1998; accepted 6
October 1998
Inhibition of hypoxia-inducible mRNAs has been
reported as an important mechanism associated with
pVHL tumor suppression (Ilioupoulos et al., 1996;
Gnarra et al., 1996; Siemeister et al., 1996; Mukhopadhay et al., 1997). Thus, pVHL inhibits expression
of mRNAs encoding vascular endothelial growth factor
(VEGF), mitogenic factors such as PDGF-B and
TGFa (Knebelmann et al., 1998), and the GLUT1
glucose transporter. VHL associated neoplasms are
highly vascular and have been shown to overproduce
VEGF (Wizigmann-Voos et al., 1995). Thus, regulation
of VEGF by pVHL is likely to represent a critical
component of its tumor suppressor activity. Recently, a
role for pVHL in assembly of an extracellular
®bronectin matrix has been proposed, a second
mechanism whereby pVHL could in¯uence tumor
neovascularization (Ohh et al., 1998).
Previous studies have shown that the VHL gene
encodes a protein of 213 residues with an apparent
molecular weight of 30 kD, pVHL30 (Latif et al., 1993;
Iliopoulos et al., 1995). pVHL30 localizes primarily to
the cytoplasm but under certain experimental conditions may shuttle between the cytosol and the nucleus
(Iliopoulos et al., 1995; Los et al., 1996; Lee et al.,
1996; Corless et al., 1997). Although the deduced
amino acid sequence of the predicted VHL protein has
provided no clue to its function, functional studies
have recently provided insights into biochemical
properties of the VHL protein. pVHL has been shown
to physically interact with elongin subunits B and C of
the elongin SIII complex (Duan et al., 1995a; Kibel et
al., 1995), a heterotrimeric protein complex that
regulates transcription elongation in vitro. Whether
elongin B and C regulate transcription elongation in
vivo, however, remains to be determined. Certain, but
not all, tumor-derived mutants of pVHL are defective
in binding to the B/C complex (Duan et al., 1995b;
Kishida et al., 1995; Kibel et al., 1995). In addition,
wild-type pVHL appears to interact with the human
Hs-CUL2 protein (Pause et al., 1997; Lonergan et al.,
1998), a member of a conserved gene family believed to
play a role in ubiquitin-mediated protein degradation
and growth control in lower eukaryotes. Binding of
pVHL to HS-CUL2 appears to be dependent on
binding of pVHL to elongin B/C (Lonergan et al.,
1998), suggesting that the pVHL/B/C/Hs-CUL2 complex may a€ect ubiquitination of target proteins
regulating tumorigenic growth.
Biochemical studies have shown that interaction of
pVHL with elongin B, elongin C and Hs-CUL2 require
an intact C-terminus of pVHL (Kibel et al., 1995;
pVHL isoforms and tumor suppression
C Blankenship et al
1530
Lonergan et al., 1998; Kishida et al., 1995). The role of
the characteristic acidic N-terminus of pVHL, containing eight repeats of the Gly-X-Glu-Glu-X motif, in
tumor suppression is so far unclear. Studies comparing
the mouse and human VHL genes predicted a mouse
pVHL protein that only retains one of eight acidic GlyX-Glu-Glu-X motifs present at the N-terminus of
pVHL30 (Latif et al., 1993; Gao et al., 1995; Duan
et al., 1995b). All known VHL mutations map
downstream of this region and a conserved AUG
codon (codon 54) in the VHL open reading frame
(Zbar, 1995; Maher and Kaelin, 1997). Interestingly,
this conserved AUG codon appears in the context of a
more conserved Kozak consensus sequence. Taken
together with previous observations that anti-VHL
antibodies speci®cally detected multiple proteins in
addition to pVHL30 (Iliopoulos et al., 1995; Kibel et
al., 1995), these observations led us to investigate
whether the VHL gene encodes protein products in
addition to pVHL30, and whether protein products
could arise from internal translation initiation at the
conserved AUG codon within the VHL mRNA.
Here we show that the VHL gene encodes a second
major and biologically active protein, pVHL18, which
arises from translation initiation at an internal AUG
codon (codon 54) and can inhibit tumor formation in
nude mice.
Results
Previous studies have shown that the VHL gene encodes a protein product of 213 amino acid residues,
pVHL, with apparent molecular weight of 30 kD
(pVHL30). A hypothetical VHL protein arising from
internal initiation at codon 54 of the predicted VHL
open reading frame, has a predicted molecular weight
of approximately 17 kD and more hydrophobic characteristics. To test whether detergent-assisted extraction would reveal such a protein by immunoblot and
immunoprecipitation analysis, tTA-786-0 VHL7/7
renal carcinoma cells were transfected with tetOPplasmids encoding pVHL (1 ± 213), pVHL (54 ± 213) or
control plasmid, and selected for stable and tetracycline-inducible expression of VHL transgenes (TG cell
lines), as described (Gossen and Bujard, 1992).
Figure 1a shows a representative immunoblot
analysis of VHL proteins expressed in the presence or
absence of tetracycline, using the previously described
anti-VHL polyclonal antibody pABVHLr1 (Gao et al.,
1995). Analysis of TG1-AR, harboring a VHL
transgene encoding pVHL 1 ± 213 (and thus the acidic
repeat region, AR), revealed at least two major
immunoreactive proteins of approximately 25 kD
(corresponding to the previously reported 30 kD
pVHL protein) and 18 kD. For TG2 and TG3,
encoding pVHL (54 ± 213), a comigrating 18 kD
protein was detected in each case (indicating that the
faster migrating protein detected in TG1-AR lysates
may have arisen by internal translation initiation). As
previously reported (Gao et al., 1995), no immunoreactive proteins were detected with preimmune serum
(data not shown). Importantly, no immunoreactive
proteins were detected in the presence of tetracycline,
which e€ectively inhibited transgene expression by
interfering with the tTA transcription activator
expressed in these cells, or parental 786-0 carcinoma
cells. Thus, NP40 e€ectively facilitates extraction of
both pVHL30 and pVHL18 proteins. When native
VHL products from VHL+/+ 293 human embryonic
kidney were analysed using the same protocol, speci®c
comigrating proteins of 25 and 18 kD (predominant
form) were detected (Figure 1a, lane 1 or 1c, lane 1).
Similar observations were made upon analysis of native
proteins from normal human diploid ®broblasts (FB),
Hela cells, and monkey COS7 cells (COS) (Figure 1b).
Detection of the 25 kD protein was somewhat variable.
The 5' region of the native VHL mRNA has a high GC
content (*80%), predicting an extensive secondary
structure possibly interfering with e€ective use of the
®rst translation start codon. To test the possibility that
the 18 kD protein detected in transgenic TG1-AR cells
or 293 cells arose from internal translation initiation,
stably transfected sublines of TG1-AR were established
with plasmids in which the second ATG in the VHL
open reading frame was mutated to ATT. As shown in
Figure 1c, a 25 kD but no 18 kD pVHL protein was
detected in three distinct cell lines carrying such
mutated transgenes. Similar results were obtained by
analysis of in vitro-translated proteins (not shown).
Thus, pVHL18 is not a breakdown product of
pVHL25.
To further con®rm the identity of the native 18 kD
pVHL protein detected in 293 cells and other human
cell types, 293 cells or Hela cells were metabolically
labeled with 35S-methionine and lysates immunoprecipitated with anti-VHL monoclonal antibody
Figure 1 Identi®cation of cellular pVHL18 protein. (a)
Immunoblot analysis of VHL proteins expressed in parental or
clonal VHL7/7 RCC-786-0 cells carrying VHL transgenes
encoding VHL(1 ± 213) (TG1AR); VHL(54 ± 213) (TG2 and
TG3); or human VHL+/+293 cells. Transgene expression was
regulated by addition of tetracycline (1 mg/ml) to the cell culture
medium as indicated. (b) Immunoblot analysis of VHL proteins
detected in various cells (COS: SV40 T antigen-expressing COS7
monkey kidney cells. FB: normal WI38 diploid human ®broblasts.
Hela: human HPV-transformed cervical carcinoma cells. (c) E€ect
of mutation of the second AUG codon in the VHL open reading
frame on expression of pVHL18 in clonal 786-0 cells (TG*4 ±
6AR) as compared to TG1AR cells. Immunoblot analysis was
performed using polyclonal antibody pAbVHLr1. Mobility of the
MW markers (in kilodaltons) is indicated in the right margin
pVHL isoforms and tumor suppression
C Blankenship et al
IG32 (Kibel et al., 1995). Immune complexes were
resolved by SDS ± polyacrylamide gel electrophoresis
and detected by autoradiography. As shown in Figure
2a, IG32 immunoprecipitates an 18 kD protein from
VHL+/+ 293 cells and Hela cells, as compared to
VHL7/7 786-0 renal carcinoma cells. As expected,
this protein migrated faster than the HA-tagged
VHL(54 ± 213) protein expressed in TG8 clonal cells.
Similar results were obtained by immunoprecipitation
using anti-VHL polyclonal antibody. No immunoreactive 18 kD bands were detected with control antibody.
IP-Western analysis with IG32 and polyclonal antibody con®rmed immunoreactivity of the 18 kD
proteins (not shown). Identity of the 18 kD protein
as pVHL18 was further analysed by comparing peptide
maps of native 18 kD pVHL and in vitro-translated
pVHL18 (54 ± 213). Proteins were metabolically labeled
in vivo or in vitro, immunoprecipitated, comigrating
18 kD proteins excised from the gel and digested with
Staph. aureus V8 protease. The partial proteolytic
peptide maps of the two proteins were identical (Figure
a
2b). Taken together, these results suggest that the
endogenous 18 kD protein identi®ed in human cells
(e.g. 293 cells) is produced by alternate translation
initiation and encodes residues 54 ± 213 of the
conceptual VHL open reading frame.
To study the tumor suppressor function of pVHL18,
we determined its e€ect on renal carcinoma cell
growth. Extensive analysis of the in vitro growth
properties of stably transfected 786-0 cells showed
that expression of pVHL18 has no signi®cant e€ect on
the growth of these cells in monolayers (not shown).
However, pronounced inhibition of their tumorigenic
growth was observed when assayed in nude mice. In
total, 81 mice were injected with either control cell lines
tranfected with control plasmid (cTG lines) or plasmids
expressing pVHL25, pVHL18 or pVHL18 fused to the
HA-tag (hemaglutinin tag). Expression of VHL
proteins in all cell lines was con®rmed by Western
blot analysis (not shown). As previously observed for
pVHL30 (9) or pVHL25 (Table 1), pVHL18 e€ectively
suppressed tumor formation as compared to control
cells (Table 1). Thus, pVHL18 encodes the necessary
domains for tumor suppression.
pVHL30 has been shown previously to inhibit
expression of proteins encoded by hypoxia-inducible
mRNAs (10 ± 12), such as vascular endothelial growth
factor (VEGF). Figure 3 shows that pVHL18 equally
inhibits VEGF mRNA expression. Both untagged and
HA-tagged forms of pVHL18 e€ectively suppressed
basal VEGF expression in those cell lines previously
tested in the nude mouse assay. Although pVHL18
reduces the overall levels of VEGF mRNA, it does not
appear to inhibit induction of VEGF mRNA by the
phorbol ester TPA (Figure 3a). These results are
consistent with previous ®ndings that pVHL30 inhibits
VEGF mRNA at the post-transcriptional level,
whereas TPA induces VEGF mRNA at the transcriptional level. The e€ect of pVHL18 on VEGF
b
Table 1 pVHL18 inhibits tumorigenic growth in nude mice
Clone
Figure 2 Endogenous pVHL18 peptide map. (a) Immunoprecipitation of 35S-methionine-labeled endogenous pVHL18 from 293
and Hela cells, HA-tagged VHL(54 ± 213) from clonal 786-0 cells
(TG8), and associated proteins. Proteins were immunoprecipitated
using monoclonal antibody IG32 or IgG. Arrows in the left
margin indicate the locations of pVHL18 and pVHL25. The
mobility of the molecular weight markers (in kilodaltons) is
indicated in right margins. B and C, elongins B and C. (b)
Endoproteinase Glu-C (V8 protease) digestion of 293 cell
endogenous 35S-labeled pVHL18 and in vitro translated (IVT)
pVHL (54 ± 213). Excised proteins were digested with 0, 0.02 or
0.2 mg of protease and resolved by 15% SDS ± PAGE
T/M
Approximate median tumor weight (mg)
Day 45
Day 60
Control cell lines
c151-B
10/10
cTG-1
6/6
cTG-2
8/8
cTG-3
8/8
cTG-4
6/6
cTG-5
6/6
726
188
198
196
847
663
1614
538
725
1311
NA
NA
VHL-transgenic
TG-1 (AR)
TG-2
TG-3
TG-7 (HA)
TG-8 (HA)
TG-9 (HA)
TG-10 (HA)
35
520
520
0
0
0
0
66
520
520
0
0
0
0
lines
2/5
2/5
4/5
0/6
0/6
0/6
0/6
Typically, 107 RCC-786-0 cells were implanted at a single site on each
mouse and tumors were measured at the indicated times (see
Materials and methods). Cells: 151-B: parental RCC-786-0 cell line
expressing the tTA transgene: cTG1-5: clonal RCC-786-0/151-B cells
transfected with control mutant VHL transgene (see Materials and
methods); TG1(AR): clonal RCC-786-0/151-B cells expressing
pVHL(1 ± 213). TG2, TG3: clonal RCC-786-0/151-B cells expressing
pVHL(54 ± 213). TG7-10: clonal RCC-786-0/151-B cells expressing
HA tagged VHL(54 ± 213). T/M: tumors/mice
1531
pVHL isoforms and tumor suppression
C Blankenship et al
1532
Figure 3 Inhibition of VEGF mRNA expression by wild-type
pVHL18 compared to tumor-derived mutant pVHL18 proteins.
(a) pVHL18 inhibits basal but not phorbol ester TPA-mediated
induction of VEGF mRNA. Total RNA (5 mg/lane) isolated from
each of the indicated transgene-expressing cell lines was analysed
by Northern blot using a human VEGF or GAPDH probes.
Lanes 1 ± 6 contain total RNA from cell lines expressing untagged
or tagged wt VHL proteins (lane 1:TG1AR, lane 2: TG2, lane
3:TG3, lanes 4 ± 6: TG 7 ± 9). Lanes 7 ± 10 contain total RNA
from control cell lines transfected with mutant VHL transgenes
(TG1-4, respectively; see Materials and methods and Table 1).
Cell lines were maintained in the absence of tetracycline and
phorbol ester TPA (upper panel a) or presence of TPA (100 ng/
ml, 8 h) (lower panel b) prior to harvesting RNA. (b) Tetracycline
regulated expression of VEGF mRNA by tumor-derived mutant
of pVHL18 (VHLR167Q). Total RNA (*5 mg/lane) isolated
from each of the indicated transgene-expressing cell lines grown in
the presence (+) or absence (7) of 1 mg/ml tetracycline was
analysed by Northern blot using a human VEGF, GAPDH or
VHL cDNA probes, as indicated. mTG11 and mTG12 are clonal
RCC-786-0/151-B cells expressing a mutant form of HA-VHL
(54 ± 213). Mutation is at codon 167 (R167Q) of the published
VHL ORF. (c) pVHL18 restores hypoxia-inducibility of VEGF
mRNA. Total RNA (*5 mg/lane) was isolated from VHL
transgene-expressing cell lines grown in normoxic (N) or hypoxic
expression appears to be very pronounced. In multiple
experiments we failed to consistently observe tetracycline regulated VEGF expression in response to
pVHL18 (or pVHL25) expression in the presence or
absence of tetracycline, despite the apparent e€ective
regulation of the expression of the VHL transgene
(Figure 3b). Although these ®ndings initially raised
concerns with regard to the physiological role of pVHL
in VEGF mRNA regulation, subsequent analysis
showed that very low level expression of pVHL could
be detected even in the presence of tetracycline (not
shown). Thus, leaky pVHL expression is presumably
sucient to downregulate VEGF mRNA. In contrast,
regulated VEGF mRNA expression was observed with
the naturally occurring mutant pVHL18-R167Q
(Figure 3b). Thus, this mutant pVHL18 protein is
compromised in function as compared to wild-type,
but retains partial activity that can be detected in
overexpression assays. In this context it should be
noted that certain pVHL mutants, including R167W,
retain partial elongin binding activity (Kibel et al.,
1995).
Taken together, these ®ndings indicate that pVHL18
is likely to inhibit tumor formation, at least in part, by
inhibiting VEGF expression. Furthermore, expression
of pVHL18 (and pVHL-R167Q) restored induction of
VEGF mRNA by hypoxia (Figure 3c), indicating that
pVHL18 shares this function with pVHL25 (which
corresponds to the previously reported isoform
pVHL30).
pVHL30 has been shown previously to bind to at
least four cellular proteins, elongin B, elongin C, HsCUL2, and ®bronectin (Duan et al., 1995a; Kibel et
al., 1995; Pause et al., 1997; Lonergan et al., 1998; Ohh
et al., 1998). To test whether pVHL18 retains similar
functions, 786-0 transgenic subclones transfected
expressing pVHL25 or pVHL18 were metabolically
labeled with 35S-methionine, and lysates immunoprecipitated with IG32 monoclonal antibody. Bound
proteins were resolved by SDS ± polyacrylamide gel
electrophoresis and visualized by autoradiography. As
previously reported, pVHL25 speci®cally coimmunoprecipitates with elongin B, elongin C and an 80 kD
protein, p80, presumably Hs-CUL2. Similarly,
pVHL18 bound to all three of these proteins (Figure
4a). A p80 protein was also detected in immunoprecipitates from 293 cells and Hela cells (Figure 2a),
indicating that p80, as previously shown, interacts with
endogenous VHL proteins. To con®rm the identity of
p80, parallel large scale experiments were performed
with pVHL18 (HA) expressing 786-0 cells (TG8) and
vector transfected control cells (cTG2). Lysates were
immunoprecipitated with anti-HA monoclonal antibody, and bound proteins resolved by gradient gel
electrophoresis. As observed with metabolically labeled
proteins (Figure 4a,b), p80 was speci®cally immunoprecipitated from TG8 vs control cells, as visualized by
copper staining. Proteolytic digestion of p80 and mass
spectrometric analysis of peptides revealed four
(H) conditions, as indicated (see Materials and methods).
Northern blots were hybridized with VEGF and GAPDH cDNA
probes
pVHL isoforms and tumor suppression
C Blankenship et al
a
labeled p80 (Figure 4c). These results demonstrate that
pVHL18 indeed binds to cellular Hs-CUL2.
Discussion
Figure 4 Binding of pVHL18 to cellular proteins. (a) Clonal 7860/151-B cells expressing VHL (54 ± 213) (lane 1) or VHL(1 ± 213)
(lane 2) proteins, or control cells lines (cTG2, cTG3), were
metabolically labeled with 35S-methionine, lysed and immunoprecipitated with anti-VHL IG32 monoclonal antibody. Immune
complexes were resolved by 4 ± 20% SDS ± PAGE. The mobility
of the molecular weight markers (in kilodaltons) is indicated in
left margin. Speci®c pVHL co-immunoprecipitating proteins are
indicated: elongin B and C (B/C) and p80. (b) TG8 clonal cells
expressing HA tagged VHL(54 ± 213) protein, or control cTG2
cells, were used for large scale immunoprecipitation experiments
for puri®cation and peptide sequencing of p80. As shown, four
peptides with sequences corresponding to the published Hs-CUL2
sequence were identi®ed. In vitro translated Hs-CUL2 is also
shown to comigrate in SDS ± PAGE with 35S-labeled p80 protein
immunoprecipitated as a HA-pVHL/ p80 complex with anti-HA
monoclonal antibody 12CA5. (c) V8-protease map of p80.
Digestion of 35S-labeled endogenous p80 from TG8 cells and in
vitro translated 35S-labeled Hs-CUL2 with endoproteinase Glu-C
(V8 protease) revealed identical peptide maps. Proteins were
digested with 0, 0.02 or 0.2 mg of protease and resolved by 15%
SDS ± PAGE
peptides with sequences identical to regions of the
predicted Hs-CUL2 protein. In vitro-translated 35SHsCUL2 comigrated with metabolically labeled p80 in
SDS ± PAGE (Figure 4b). Furthermore, identical V8
protease peptide maps were obtained for in vitrotranslated Hs-CUL2 protein and in vivo metabolically
Previous studies have shown that the VHL gene
encodes a protein of 213 amino acid residues with an
acidic N-terminus (residues 1 ± 54) characterized by a
Gly-X-Glu-Glu-X motif repeated eight times (pVHL30
or, as reported here, pVHL25). In this study we
identi®ed a second abundant VHL protein, pVHL18,
which appears to encode residues 54 ± 213 of the
previously characterized pVHL30 product and also
functions as a tumor suppressor.
Two VHL mRNAs have been detected in human
cells (Gnarra et al., 1994; Shuin et al., 1994; Whaley et
al., 1995). Isoform I contains exon 1 (which encodes
both Met1 and Met54), 2 and 3. In isoform II, exon 1
is fused to exon 3. Certain renal carcinoma cell lines
produce only transcript isoform II (Gnarra et al.,
1994), suggesting that even if made (this transcript has
not been shown to encode a protein), it lacks tumor
suppressor activity. As shown in this study, (a)
pVHL18 comigrates with pVHL(54 ± 213) which
retains tumor suppressor activity, (b) the proteolytic
peptide maps of these two proteins are identical, and
(c) coexpression of pVHL18 with pVHL30 in cells (or
in vitro) expressing an extended VHL transgene is
inhibited when the Met54 codon is mutated to encode
isoleucine. Thus, pVHL18 does not appear to arise
from alternative mRNA splicing or proteolytic
degradation of pVHL30 but most likely is produced
by ribosomal scanning and internal translation
initiation at Met54 of the VHL mRNA open reading
frame. Similar observations were made by others
working independently from us (Dr William G
Kaelin, personal communication).
As previously reported for pVHL30 (Iliopoulos et
al., 1995), pVHL18 had no apparent e€ect on renal
carcinoma cell monolayer growth in vitro, indicating
that its ability to inhibit tumor formation in nude mice
is not an unspeci®c growth inhibitory e€ect. Inhibition
of tumor growth correlated with downregulation of
VEGF mRNA, indicating that, at least in part, the
tumor suppressor function of pVHL18 has an antiangiogenic component. It has previously been shown
that inhibition of VEGF expression by pVHL30
correlates with its ability to bind to the cellular
proteins elongin B, elongin C, and Hs-CUL2. Likewise, we show here that pVHL18 binds to these cellular
proteins, indicating that it shares common mechanisms
of action with pVHL30.
As a result of lacking the N-terminal acidic region,
which is part of pVHL30, pVHL18 is predicted to
exhibit a more hydrophobic character, and may not
only share properties with pVHL30 but possibly also
encode distinct biological activities. The production of
protein products as a result of internal translation
initiation has been reported for an increasing number
of viral and cellular mRNAs, and in various cases
these encode proteins with distinct subcellular
localization and biological functions. Examples of
such kind include, for instance, the FGF-2, int-2, the
MOD5 and drosophila antennapeadia genes (Acland et
al., 1990; Gillman et al., 1991; Iizula et al., 1995;
1533
pVHL isoforms and tumor suppression
C Blankenship et al
1534
Vagner et al., 1995). Thus, future studies related to
pVHL18 may reveal novel functions associated with
the VHL gene.
We and others have shown previously that the
region corresponding to residues 54 ± 213 of the VHL
open reading frame is the region of highest homology
between human, mouse and rat VHL genes (Gao et al.,
1993). Furthermore, all known mutations in the VHL
gene invariably map downstream of the internal AUG
codon-54 (Zbar, 1995, Maher and Kaelin, 1997),
indicating that no known mutations would selectively
target pVHL30. Thus, germline and somatic mutations
inactivate both pVHL30 and pVHL18, which, as
shown here, have tumor suppressor activity.
Materials and methods
Cell lines and cDNA cloning
COS-7 (CRL 1651), NIH3T3 (CRL1658), Hela (CCL-2),
293 (CRL 1573), WI-38 (CCL 75), NRK (CRL 6509) and
786-0 (CRL 1932) cells were obtained from the American
Type Culture Collection (Rockville, MD, USA) and
maintained as recommended. VHL7/7 RCC-786-0 renal
carcinoma cells were transfected with pUHD15-1Neo
(tTA-encoding tetracycline activator). Colonies were
isolated, expanded and analysed for tetracycline-mediated
regulation of gene expression by transient transfection with
tetOP-reporter plasmids. One clone was subsequently cotransfected with tTA-responsive expression plasmids,
pUHD10-3 or pUHD10-3HA (containing 5' HA epitope
tag), encoding various VHL proteins: VHL(1 ± 213)
(TG1AR, cDNA spanning nucleotides (nt) 184 ± 1673 of
the ®rst published VHL sequence (Latif et al., 1993);
VHL(54 ± 213) (TG2,TG3: cDNA spanning nt 296 ± 955 of
the ®rst reported human VHL cDNA); mutant VHL(1 ±
213) (TG*4 ± 6: cDNA spanning nt 184 ± 1673 and containing a mutation at codon 54 (ATG to ATT) changing Met54
to Ile54); HA-tagged VHL(54 ± 213) (TG7-10, HA-tag
fused to Met54); HA-tagged mutant VHL(54 ± 213)
containing tumor derived missense mutation R167Q.
Control cell lines contain a mutant VHL transgene
generated by insertion of a cassette containing stop
codons in all three reading frames into the unique NotI
restriction site at nt 384, adjacent to the second ORF ATG
codon (codon 54). Individual colonies were ampli®ed and
assayed for tetracycline (1 mg/ml)-regulated VHL protein
expression by Western blot analysis. Plasmids pUHD151Neo (tTA) and pUHD10-3 (tTA-responsive expression
vector) were obtained from Gossen and Bujard (1992).
Marathon 5' RACE was used to clone the 5' end of HsCUL2 and full length Hs-CUL2, using polyA RNA from
human kidney (Clontech) according to manufacturer's
instructions (Clontech). Hs-CUL2 cDNA was cloned into
pCR3.1 by TA cloning (Invitrogen, Carlsbad, CA, USA).
Recombinant clones were con®rmed by sequencing of both
strands. The cDNA sequence obtained corresponded to the
recently published Hs-CUL2 cDNA sequence (Lonergan et
al., 1998; unpublished data).
Protein analysis
Antibodies and cell-free protein translation assays Anitypuri®ed polyclonal rabbit antiserum to human VHL
protein (pAbVHLr1) was described previously (Gao et
al., 1995). Anti-VHL monoclonal antibody IG32 was
purchased from PharMingen (San Diego, CA, USA),
anti-tubulin antibody from Oncogene Science (Cambridge, MA, USA), and anti-HA monoclonal antibody
(12CA5) from Boehringer Mannheim (Indianapolis, IN,
USA). 35S-methionine-labeled proteins were produced by
coupled in vitro transcription/translation using reticulocyte
lysates, as recommended by the manufacturer (Promega,
Madison, WI, USA). Proteins were resolved by SDS ±
PAGE under denaturing conditions employing either 4 ±
20% or 15% acrylamide gels.
Immunoprecipitation analyses Immunoprecipitation analyses were performed essentially as described by Kibel et
al. (1995). Brie¯y, transfected RCC, HeLa or 293 cells were
labeled metabolically using DMEM minus methionine and
cysteine (Gibco/BRL, Gaithersburg, MD, USA) supplemented with L-[35S]-methionine and L-[35S]-cysteine
[0.5 mCi of Promix (41000 Ci/mmol; Amersham, Arlington Heights, IL, USA) per ml] for 5 h at 378C, 5% CO2.
Cells were lysed in ice-cold EBC bu€er [50 mM Tris-Cl
(pH 8.0), 120 mM NaCl, 0.5% Nonidet P-40] containing
protease inhibitors. The radiolabeled supernatants were
mixed with an equal volume of ice-cold NETN [20 mM
Tris-Cl (pH 8.0), 100 mM NaCl, 1 mM EDTA, 0.5%
Nonidet P-40, protease inhibitors] containing the appropriate antibody and protein A sepharose. The immune
complexes were then washed ®ve times at 48C and resolved
by SDS ± PAGE.
Immunoblot analysis Cells were washed once with PBS
(pH 7.4) and collected by centrifugation. Cell pellets were
resuspended in 26Laemmli sample bu€er and incubated at
958C (5 min). Cellular lysate material was passed ®ve times
through a 25 gauge syringe needle and resolved directly by
SDS ± PAGE. Alternatively, cells were lysed, as described
for immunoprecipitation analyses and resolved by SDS ±
PAGE. Immunoblot reactions were performed as previously described (Gao et al., 1995).
Peptide microsequencing VHL immune complexes were
resolved by 7% SDS ± PAGE. p80 was detected by copper
staining (BioRad, according to manufacturer's instructions), and the corresponding band was excised, destained,
digested with trypsin and microsequenced (William S Lane,
Harvard Microchemistry Facility).
V-8 protease maps Partial proteolytic digests of in vitro
translated pVHL(54 ± 213) and Hs-CUL2 proteins, and
endogenous p18 (VHL) and p80 (Hs-CUL2), were
performed as described by Harlow and Lane (1988).
Hypoxia treatment, TPA treatment and RNA analysis
Cell lines expressing VHL transgenes were plated in
complete medium, and 12 h later cells were either placed
in GasPak pouches (BBL Microbiology Systems) and
incubated for 18 h under hypoxic conditions at 378C, as
indicated by methylene blue decolorization (less than 2%
O2 is achieved under these conditions (BBL)), or incubated
under standard growth conditions (5% CO2, 378C). For
treatments with the phorbol ester TPA, cells were plated in
complete medium, and 24 h later the medium was changed
to complete medium+100 ng/ml TPA. At 24 h following
TPA addition, cells were harvested for RNA analyses.
Total RNA was isolated from cultured cells and analysed
by Northern blot as previously described (Buckbinder et
al., 1995). A human VEGF cDNA corresponding to the
coding region of VEGF isoform 165 was kindly provided
by Dr M Klagsbrun, and a human GAPDH probe was
obtained from Clontech (Palo Alto, CA, USA). Probes
were labeled using 32P-dCTP (New England Nuclear,
Boston, MA, USA).
pVHL isoforms and tumor suppression
C Blankenship et al
Nude mouse xenograft assay
Renal carcinoma cells stably expressing VHL transgenes
were treated with trypsin:EDTA and suspended in basal
culture medium. Cells (107 cells/site) were implanted
subcutaneously into the axillary region of nude mice (nu/
nu; Harland Sprague Dawley). Tumor growth was
measured weekly by calipers and tumor weights were
calculated using the formula for the volume of an elliptical
object,
assuming
volume-weight
equivalence
of
1 mm3=1 mg): Tumor weight (mg)=(1)6(w2)/2 (mm3).
Acknowledgements
We would like to acknowledge Dr William S Lane of the
Harvard Microchemistry Facility for his assistance in
peptide sequencing. In addition we acknowledge Dr Frank
Lee for his help with the tumor suppression experiments in
vivo, and Tim Skinner for expert technical assistance.
References
Acland P, Dixon M, Peters G and Dickson C. (1990). Nature,
343, 662 ± 665.
Buckbinder L, Talbott R, Velasco-Miguel S, Takenaka I,
Faha B, Seizinger BR and Kley N. (1995). Nature, 377,
646 ± 649.
Corless CL, Kibel AS, Iliopoulos O and Kaelin WG. (1997).
Human Pathol., 28, 459 ± 464.
Duan DR, Pause A, Burgess WH, Aso T, Chen DYT, Garret
KP, Conaway RC, Conaway JW, Linehan WM and
Klausner RD. (1995a). Science, 269, 1402 ± 1406.
Duan DR, Humphrey JS, Chen DYT, Weng Y, Sugekawa J,
Lee S, Gnarra JR, Linehan WM and Klausner RD.
(1995b). Proc. Natl. Acad. Sci. USA, 92, 6459 ± 6463.
Gao J, Naglich JG, Laidlaw J, Whaley JM, Seizinger BR and
Kley N. (1995). Cancer Res., 55, 743 ± 747.
Gillman EC, Slusher LB, Martus NC and Hopper AK.
(1991). Mol. Cell. Biol., 11, 2382 ± 2390.
Gnarra JR, Tory K, Weng Y, Schmidt L, Wei MH, Li H,
Latif F, Liu S, Chen F, Duh F-M, Lubensky I, Duan DR,
Florence C, Pozzatti R, Walther MM, Bander NH,
Grossman HB, Brauch H, Pomer S, Brooks JD, Isaacs
WB, Lerman MI, Zbar B and Linehan WM. (1994).
Nature Genet., 7, 85 ± 90.
Gnarra JR, Zhou S, Merrill MJ, Wagner J, Agner J, Krumm
A, Papavassiliou E, Old®eld E, Klausner R and Linehan
WM. (1996). Proc. Natl. Acad. Sci. USA, 93, 10589 ±
10594.
Gossen M and Bujard H. (1992). Proc. Natl. Acad. Sci. USA,
89, 5547 ± 5551.
Harlow E and Lane D. (1988). Antibodies ± a laboratory
manual. Cold Spring Harbor Laboratory Press, Plainview,
NY.
Herman JG, Latif F, Weng Y, Lerman MI, Zbar B, Liu S,
Samid D, Duan DSR, Gnarra JR, Linehan WM and
Baylin SB. (1994). Proc. Natl. Acad. Sci. USA, 91, 9700 ±
9704.
Iizula N, Chen C, Yang G, Johannes G and Sarnov P. (1995).
Curr. Top. Microbio. Immunol., 203, 155 ± 177.
Iliopoulos O, Kibel A, Gray S and Kaelin WG. (1995).
Nature Med., 1, 822 ± 826.
Iliopoulos O, Levy A, Jiang C, Kaelin W and Goldberg MA.
(1996). Proc. Natl. Acad. Sci. USA, 93, 10595 ± 10599.
Kanno H, Kondo K, Ito S, Yamamoto I, Fuji S, Torigoe T,
Sakai N, Hosaka M, Shuin T and Yao M. (1994). Cancer
Res., 54, 4845 ± 4847.
Kibel A, Iliopoulos O, De Caprio JA and Kaelin WG. (1995).
Science, 269, 1444 ± 1446.
Kishida T, Stackhouse TM, Chen F, Lerman MI and Zbar B.
(1995). Cancer Res., 55, 4544 ± 4548.
Knebelmann B, Ananth S, Cohen HT and Sukhatme VP.
(1998). Cancer Res., 58, 226 ± 231.
Latif F, Tory K, Gnarra J, Yao M, Duh FM, Orcutt ML,
Stackhouse T, Kuzmin I, Modi W, Geil L, Schmidt L,
Zhou F, Li H, Wei MH, Chen F, Glenn G, Choyke P,
Walther MM, Weng Y, Duan DSR, Dean M, Glavac D,
Richards FM, Crossey PA, Ferguson-Smith MA, Le
Paslier D, Chumakov I, Cohen D, Chinault AC, Maher
ER, Linehan WM, Zbar B and Lerman MI. (1993).
Science, 260, 1317 ± 1320.
Lee S, Chen DYT, Humphrey JS, Gnarra JR, Linehan WM
and Klausner R. (1996). Proc. Natl. Acad. Sci. USA, 93,
1770 ± 1775.
Lonergan KM, Iliopoulos O, Ohh M, Kamura T, Conaway
RC, Conaway JW and Kaelin WG Jr. (1998). Mol. Cell.
Biol., 18, 732 ± 741.
Los M, Jansen GH, Kaelin WG, Lips CJM, Blijham GH and
Voest EE. (1996). Lab. Invest., 75, 231 ± 238.
Maher E and Kaelin WG. (1997). Medicine, 76, 387 ± 391.
Mukhopadhyay D, Knebelmann B, Cohen HT, Ananth S
and Sukhatme VP. (1997). Mol. Cell. Biol., 17, 5629 ±
5639.
Ohh M, Yauch RL, Lonergan KM, Whaley JM, StemmerRachaminov AD, Louis DN, Gavon BJ, Kley N, Kaelin
WG and Iliopouplos O. (1998). Mol. Cell., 1, 959 ± 968.
Pause A, Lee S, Worrell RA, Chen DY, Burgess WH,
Linehan WM and Klausner RD. (1997). Proc. Natl. Acad.
Sci. USA, 94, 2156 ± 2161.
Shuin T, Kondo K, Torigoe S, Kishida T, Kubota Y, Hosaka
M, Nagashima Y, Kitamura H, Latif F, Zbar B, Lerman
MI and Yao M. (1994). Cancer Res., 54, 2852 ± 2855.
Siemeister G, Weidel K, Mohrs K, Barleon B, MartinyBaron G and Marme D. (1996). Cancer Res., 56, 2299 ±
2301.
Vagner S, Gensac MC, Maret A, Bayard F, Amalric F, Prats
H and Prats AC. (1995). Mol. Cell. Biol., 15, 35 ± 44.
Whaley JM, Naglich J, Gelbert L, Hsia YE, Lamiell JM,
Green JS, Collins D, Neumann HPH, Laidlaw J, Li FP,
Klein-Szanto AJP, Seizinger BR and Kley N. (1994). Am.
J. Human Genet., 55, 1092 ± 1102.
Wizigmann-Voos S, Breier G, Risau W and Plate KH.
(1995). Cancer Res., 55, 1358 ± 1364.
Zbar B. (1995). Cancer Surveys, Genetics and Cancer: a
second look, Ponder BAJ, Cavenee WK and Solomon E.
(eds). Cold Spring Harbor Laboratory Press. pp. 219 ±
232.
1535