 |
INTRODUCTION |
Classical cyclins promote cellular proliferation. They form
complexes with specific cyclin-dependent kinases
(CDKs)1 (1, 2), thereby
enabling CDK activation. Activated CDKs trigger cell cycle transitions
through phosphorylation of specific targets such as the tumor
suppressor Rb and cytoskeletal proteins (3). Some recently identified
cyclins and CDKs do not promote cell cycle progression per
se (4). For example, the mammalian p35-CDK5 complex is essential
for the regulation of neurite outgrowth (5, 6). Mammalian cyclin H-CDK7
and cyclin C-CDK8 regulate transcription by phosphorylating the
carboxyl terminus of RNA polymerase II (7).
Cyclin G2 is an unconventional cyclin highly expressed in cells
undergoing apoptosis (8, 9). In contrast to conventional cyclins,
cyclin G2 expression is up-regulated in B cells responding to
growth-inhibitory stimuli and during antigen-induced cell cycle arrest
and apoptosis (9), but is low in proliferating B cells (9). Cyclin G2
is also strongly expressed in adult brain cerebellum, a terminally
differentiated tissue (8, 9). Cyclin G2 shows 26% amino acid identity
with cyclin A, but its closest homolog is cyclin G1 (53% amino acid
identity). Initial cloning and sequencing of cyclin G1, then called
cyclin G, predicted a protein starting at the cyclin box and lacking an
NH2-terminal region typically present in other cyclins (10,
11). Cyclin G was later found to exist in a 45-amino acid
NH2-terminally extended form (8, 12) and renamed cyclin G1.
The cyclin G1 gene was identified as a transcriptional target of the
tumor suppressor and cell cycle checkpoint regulator p53 (11, 12). As
DNA damage induces cyclin G1 mRNA expression, it may function in
cell cycle checkpoint control (11-13). The third member of this novel
family of cyclins, cyclin I, exhibits 30% amino acid identity with
cyclin G2 and G1. Cyclin G1 and I mRNAs are constitutively
expressed throughout the cell cycle (8-10, 14). All three cyclins are
expressed in tissues rich in terminally differentiated cells,
including cardiac and skeletal muscle and various brain regions (8-10,
14). The overall expression profile of the G cyclins is atypical and
not associated with promotion of cellular proliferation but rather
suggests that they act as inhibitory coordinators of the cell cycle in
some cell types and may assist in maintaining the quiescent state of differentiated cells.
Although cyclin G2 and G1 are sequence homologs, their expression is
differentially regulated during development (9), and they probably
serve unique physiological functions. In contrast to cyclin G1, the
cyclin G2 gene is not a transcriptional target of p53, and its
expression is independent of p53 (8, 9, 13, 15). The subcellular
distribution of these two homologs is also distinctive; the primarily
nuclear localization of cyclin G1
(15)2 contrasts with the
mainly cytoplasmic and less frequent nuclear expression of cyclin G2
(this report and Ref. 15). Upon receptor-activated negative signaling
in B lymphocytes, cyclin G2 transcript levels are dramatically elevated
over an extended period, whereas cyclin G1 expression is unaffected (8,
9). In lymphocytes, cyclin G1 expression is constitutive throughout the
cell cycle, whereas cyclin G2 expression oscillates, peaking in late
S/early G2 phase of the cell cycle (8, 9). Among
differentiated tissues, cyclin G2 expression is highest in cerebellum,
but cyclin G1 is most abundant in muscle, where cyclin G2 is low
(8-10).
Controlled dephosphorylation of distinct substrates is critical for
regulation of cellular division and differentiation. Protein phosphatase 2A (PP2A) is a highly conserved serine/threonine
phosphatase essential for a plethora of cellular functions including
signal transduction, translational control, endosome trafficking, and cell cycle regulation (16-20). It is the chief target of various natural toxins and viral oncogenes (16, 21-23). PP2A dephosphorylates physiological substrates of CDKs in vitro (24) and the
Rb-related protein p107 in vivo (25). Because of its effects
on the activity of the dual specificity phosphatase Cdc25, a key
regulator of the mitotic CDK p34cdc2, PP2A activity is thought
to prevent premature entry into mitosis (26, 27). The diversity of PP2A
function is paralleled by its requisite localization to distinct
subcellular sites and specific targeting to a variety of cellular
substrates, a crucial regulatory process for the proper function of
many protein kinases (28). The 36-kDa catalytic C subunit of PP2A
(PP2A/C) exists in cells in either a dimeric core complex with a 65-kDa
"scaffolding" A subunit (PP2A/A, PR65) or, more often, in a
trimeric complex containing variable regulatory B subunits bound to the
A/C dimer (16, 29, 30). The abundance of known and possible trimeric
PP2A complexes that could be formed from two C, two A, and at least 16 B subunits (encompassed by the sequence-unrelated B (PR55), B' (PR56),
and B" (PR72) families) reflects the medley of functions for this phosphatase (16, 31-35). The various regulatory B subunits direct PP2A
to distinct subcellular localizations and modulate its substrate specificity (16, 36-41). Unlike other B subunits, some B' subunits contain a classic bipartite nuclear localization signal (31, 38, 42).
Studies in yeast showed that rabbit B' isoforms can complement
mutations in the yeast B' subunit Rts1p, but Rts1p cannot complement
mutations in the B subunit Cdc55p, indicating that different regulatory
B subunits confer enzyme specificity (43). Thus, the discovery that
cyclin G1 interacts with PP2A B' subunits (44) implicates cyclin G1 in
the regulation of PP2A and contrasts it with the classical cell
cycle-promoting cyclins.
For a better understanding of cyclin G2 function, we investigated
potential binding partners of this novel cyclin. We show that cyclin G2
interacts with various PP2A B' subunits in vitro and in
living cells, where they colocalize at distinct sites. Okamoto et al. observed that cyclin G1 interacts
with two distinct B' subunits; however, the composition of cyclin
G1-PP2A/B' complexes was not further resolved (44). To understand the
function of the cyclin G2 complex, we examined whether the cyclin
G2-PP2A/B' complex contains the other two major subunits of the
prototypical PP2A trimeric A-B'-C complex. It is possible that members
of the G cyclin family sequester B' subunits away from the PP2A/A-C
dimer and thereby change the pool of distinctly targeted PP2A
holoenzyme. Alternatively, cyclin G1 or G2 may directly regulate either
the activity or localization of PP2A by forming multimeric complexes. We found that cyclin G2 and G1 cannot only associate with a variety of
B' subunits but can also directly interact with the catalytic C subunit
of PP2A. Our data indicate that this cyclin G2-C subunit association
can also occur in vivo and that association of cyclin G2
with PP2A/B' and C excludes the A subunit. Accordingly, cyclin G2 can
form catalytically active trimeric complexes with PP2A/B' and C
subunits and thus may alter PP2A targeting or substrate specificity. A
functional consequence of high cyclin G2 expression may be a change in
the balance or local concentration of distinct PP2A complexes. This
could lead to alterations in signal transduction, cell cycle
regulation, cytoskeletal networks, or mitosis. Indeed, we observe an
induction of cell cycle arrest and unusual nuclear DNA structures in
cells overexpressing cyclin G2, indicative of an aberrant
mitosis/cytokinesis. Together, our studies point to a role for cyclin
G2 in PP2A regulation and provide a mechanism through which cyclin G2
up-regulation could contribute to cell cycle inhibition.
| |
MATERIALS AND METHODS |
Reagents and Antibodies--
Glutathione- and protein
G-Sepharose, horseradish peroxidase (HRP)-conjugated protein A, and ECL
detection kits were purchased from Amersham Biosciences. The mouse
anti-HA antibody was obtained from Babco (Richmond, CA), the mouse
anti-PP2A/C from Transduction Laboratories (Lexington, KY), mouse
anti-V5 from Invitrogen (Carlsbad, CA), and goat anti-lamin B from
Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Rat monoclonal
anti-PP2A/A antibodies 6G3 and 6F9 were kindly provided by Dr. Gernot
Walter (University of California, San Diego, CA) (30), and sheep
anti-PP2A/C and rabbit anti-pan-PP2A/B' polyclonal antibodies were from
Dr. Brian E. Wadzinski (Vanderbilt University, Nashville, TN) (40, 45).
Rabbit anti-green fluorescent protein (GFP) antiserum and the DNA
binding fluorescent dye TOTO-3 were purchased from Molecular Probes,
Inc. (Eugene, OR). HRP-conjugated rabbit anti-rat and donkey anti-sheep
IgGl; fluorescein isothiocyanate (FITC)-conjugated donkey anti-rabbit
and anti-mouse IgG; lissamine rhodamine sulfonyl chloride
(LRSC)-conjugated donkey anti-rabbit, anti-mouse, and anti-goat IgG
antibodies; and Cy5-conjugated donkey anti-goat IgG were from Jackson
ImmunoResearch (West Grove, PA). HRP-conjugated goat anti-mouse IgG was
from Bio-Rad and HRP-conjugated rat anti-mouse Ig 
light chain was
from Zymed Laboratories Inc. (South San Francisco,
CA). All other reagents were of standard quality from established suppliers.
Production of Cyclin G1 and G2 Fusion
Proteins--
Complementary DNA templates encoding murine cyclin G1
and G2 (8, 9) and human PP2A/C (with and without an HA tag (46)) were
amplified by PCR using oligonucleotides containing engineered endonuclease restriction sites for subcloning in frame with a 5'
nucleotide sequence encoding glutathione S-transferase (GST) or a polyhistidine tag and T7 epitope with standard methods as described (8, 9, 47, 48). All PCR fragments were cloned into the TA
cloning vector PCR.1 (Invitrogen) and confirmed by PCR sequencing with
the AmpliTaq system (PerkinElmer Life Sciences). Cyclins G1
and G2 and PP2A/C sequences were then subcloned into pGEX 4T-1
(Amersham Biosciences), and pTrcHisA (Invitrogen) and DNA sequences
were verified as above.
GST and polyhistidine (His) fusion proteins were expressed in
Escherichia coli (NovaBlue; Novagen, Madison, WI) and
purified following the manufacturer's protocols with modifications.
Induced bacterial pellets were resuspended in 50 ml of TBS (150 mM NaCl, 15 mM Tris-Cl, pH 7.4) containing 0.1 mg/ml lysozyme. After 10 min on ice, protease inhibitors (1 µg/ml
pepstatin A, 1 µg/ml leupeptin, 1 µg/ml aprotinin, 200 nM phenylmethanesulfonyl fluoride), 15 mM
dithiothreitol (GST) or 10 mM 
-mercaptoethanol (His), 10 mM EDTA (GST proteins only), and 1.5% sarkosyl were added.
Insoluble material was removed by centrifugation (30,000 rpm, 60 min,
4 °C, 45Ti rotor), and supernatants were frozen and stored at
80 °C. For affinity purification, bacterial lysates were thawed
and diluted 1:5 in TBS containing 2-3% Triton X-100 to neutralize sarkosyl. Lysates were incubated with 1 ml of glutathione-Sepharose (GST proteins) or Ni2+-nitrilotriacetic acid resin (Qiagen;
His) at 4 °C for 2 h. The resins were washed three times with
TBS containing 0.05% Tween 20 and once with TBS only. The purity of
all immobilized proteins was confirmed by SDS-PAGE and Coomassie
staining, revealing a single major band at the predicted molecular
mass. The identity of each fusion protein was further confirmed by
immunoblot analysis with anti-GST and anti-T7 antibodies (47), with
antibodies produced against multiple antigen peptides derived from
NH2-terminal sequences of cyclin G1 and G2, respectively
(see below),3 and with
anti-PP2A/C antibodies. For elution of cyclin G1- and G2-TrcHis,
200-500 mM imidazole was added to the washing buffer.
Cyclin G2 Antibodies--
New Zealand White rabbits were
immunized with affinity-purified full-length cyclin G2-GST fusion
protein eluted from glutathione-Sepharose with 15 mM
glutathione in 50 mM triethanolamine, pH 11.5, and dialyzed
against TBS. Antibodies were also raised against a
multiple-antigen peptide (49) corresponding to the first 16 amino acids
of the NH2-terminal cyclin G2 sequence
(MKDLGAKHLAGGEGVQ-K-8-MAP; anti-cyclin G2-NT). Rabbits were immunized
every 3-4 weeks with 100-250 µg of the cyclin G2-GST fusion protein
or 1-2 mg of the multiple antigen peptide solubilized in PBS and
emulsified in an equal volume of Freund's complete (initial injection)
or incomplete adjuvant in multiple subcutaneous sites. Antiserum was
preabsorbed for 90 min at 4 °C with 10 mg/ml rat liver acetone
powder (Sigma; cyclin G2 mRNA is not detectable in liver), cleared
by centrifugation, incubated for 2-4 h with cyclin G1-GST cross-linked
with dimethyl pimelimidate to glutathione-Sepharose (50), and
centrifuged to remove antibodies against the GST moiety of the fusion
protein or those cross-reactive with cyclin G1. Supernatants as well as the antisera produced against the multiple-antigen peptide preadsorbed on liver extract were then affinity-purified by incubation for 4-10 h
at 4 °C with cyclin G2-GST cross-linked with dimethyl pimelimidate to glutathione-Sepharose. Antibodies were eluted with 10 ml of 100 mM glycine, pH 2.5, and the eluates were neutralized with 1 ml of 1 M Tris-Cl, pH 8, and concentrated.
Cyclin G2 Expression Constructs and Transient Transfection of
HEK293 Cells--
HEK293 fibroblasts were cultured in Dulbecco's
modified Eagle's medium (Invitrogen), supplemented with 10%
heat-inactivated fetal bovine serum (NovaTech), 2 mM
L-glutamine, 1 mM sodium pyruvate at 37 °C
in a humidified chamber with 5% CO2. Cultures were plated at 20-30% and maintained at 50-90% confluence. DNA constructs for
expression of V5 epitope-tagged cyclin G2 or cyclin G2-GFP fusion
proteins in mammalian cells were prepared by PCR amplification of
murine cyclin G2 cDNAs with modified oligonucleotide primers. The
forward and reverse primers had engineered endonuclease restriction sites (BamHI and BglII in forward primer,
XhoI and Eco47III in reverse primer). The forward
primer contains an optimal Kozak translation initiation sequence; the
reverse primer lacked the antisense sequence for an in frame stop
codon. PCR fragments were cloned into pT-Adv
(CLONTECH, Palo Alto, CA) and sequenced. A 1.1-kbp
BamHI/XhoI restriction fragment encompassing
cyclin G2 was first subcloned into pIND-V5HisC and then a larger
BamHI/PmeI fragment encompassing cyclin G2 was
shuttled into BamHI/EcoRV-digested pcDNA3
(Invitrogen) to yield pcDNA3-cycG2-V5His for the constitutive expression of V5 epitope-tagged cyclin G2 in transient transfection. To
construct a constitutive expression vector for GFP-tagged cyclin G2, a
1.1-kbp NheI/Eco47III fragment carrying cyclin G2
was isolated from the above pIND-cycG2-V5His clone and shuttled into an
NheI/Ecl136II-cut derivative of pIND
(Invitrogen), pIND-GFP, that contains the 700-bp NheI/NotI fragment encompassing the GFP coding
sequence and a 5' polylinker sequence. A 1.9-kbp
BglII/NotI fragment from this pIND-cycG2-GFP
clone containing cyclin G2 fused in frame with GFP coding sequences was
subcloned into BamHI/NotI-digested pcDNA3 (now pcDNA3-cycG2-GFP) for constitutive expression of GFP-tagged cyclin G2.
HEK293 cells were transiently transfected using a modified calcium
phosphate precipitation protocol (51) with one of the above
pcDNA3-cyclin G2 expression constructs, expression constructs for
the HA-tagged PP2A B' subunits (pCGN B'2, B'3, or B'4 or pCMVneoHA B'
(31)), for NH2-terminally HA-tagged PP2A/C
subunit (HA-PP2A/C in pcDNA3+ kindly provided by Dr. Brian E. Wadzinski, Vanderbilt University, Nashville, TN (46)) or rabbit
skeletal muscle PP2A/A in pcDNA3 or a combination of these
constructs. An equal molar amount of pAdVantage (Promega) was included
in each transfection mix to enhance transfection and translation efficiency (52). Cells were harvested 24-36 h post-transfection.
Immunoprecipitation and Affinity Chromatography of Cyclin G2 and
PP2A from Cell Lysates--
Cells were lysed with radioimmune
precipitation buffer (10% glycerol, 1% Nonidet P-40, 0.4%
deoxycholate, 150 mM NaCl, 5 mM EGTA, 5 mM EDTA in 50 mM Tris, pH 7.4) containing
pepstatin A (1 µg/ml), leupeptin (10 µg/ml), aprotinin (20 µg/ml), and phenylmethanesulfonyl fluoride (200 nM).
Lysates were cleared of insoluble material by centrifugation. 0.05%
SDS was added before precipitation with either 5-10 µg of
affinity-purified anti-cyclin G2, 10 µg of anti-HA, 6-10 µl of
anti-PP2A/A (6F9 on protein G-Sepharose (30)), or 10 µl (~10 µg)
of sheep anti-PP2A/C (19) and 15 µl of protein A-Sepharose (Repligen,
Needham, MA) or with 15 µl of microcystin-Sepharose (Upstate
Biotechnology, Inc., Lake Placid, NY). For in vitro
interaction studies with cyclin G1- and G2-GST fusion proteins
immobilized on glutathione-Sepharose, cell lysates were added to 20 µl of the resins and mixed for 1-4 h at 4 °C. Samples were washed
with 0.1% Tween 20 and 0.5% Nonidet P-40 in TBS before extraction
with denaturing buffer, SDS-PAGE in 10-20% gradient acrylamide gels, and immunoblotting as described (53). Anti-HA immunoprecipitates of
HA-tagged PP2A/C were also extracted at 60 °C for 20 min in a
thermomixer with 40 µl of 1.5% SDS, 50 mM Tris-Cl, pH
8.0, containing 5 mM dithiothreitol and protease inhibitors
(as above). The extracts were diluted with 360 µl of TBS, protease
inhibitors, and 1% Triton X-100 and added to cyclin G1- and G2-GST
fusion proteins immobilized on glutathione-Sepharose for further
interaction studies.
Immunofluorescence Confocal Microscopy--
HEK293 and CHO cells
were seeded at 1.5 × 105 cells/35-mm well onto
a glass coverslip coated with 10 µg/ml collagen and 1 µg/ml poly-L-lysine. Cells were allowed to adhere and were grown
for 14-18 h before transfection with a ratio of 1 µg of DNA per 3 µl of the LipofectAMINE (Invitrogen) transfection reagent in 1 ml of
serum-free medium (OptiMEM; Invitrogen) according to protocols of the
manufacturer with the following modifications. The serum-free medium
was removed 3 h after transfection and replaced with the complete
medium described above; 2 h later, the medium was exchanged again.
Coverslips were removed 24-36 h after transfection and rinsed with PBS
followed immediately by fixation with 4% paraformaldehyde for 15 min
at room temperature. For removal of soluble proteins and cellular
lipids by preextraction with Triton X-100, coverslips were briefly
rinsed in the microtubule-stabilizing buffer PHEM (45 mM
PIPES, 45 mM HEPES, 10 mM EGTA, 5 mM MgCl2, pH 6.9), followed by extraction of
the cells for 90-95 s in PHEM containing 0.5% Triton X-100 and by
another rapid rinse in PHEM immediately prior to fixation essentially
as described (54). After fixation, the coverslips were rinsed in PBS
and stored at 4 °C in PBS plus 0.02% sodium azide.
Specimens were permeabilized with 0.3% Triton X-100 in PBS for 20 min
(unless preextracted; see above), incubated with blocking solution (PBS
containing 2% glycerol, 25 mM NH4Cl, 2.5%
fetal bovine serum, and 0.5% donkey serum) for 2 h at 20 °C or
overnight at 4 °C and then with primary antibodies in blocking
solution for 1.5 h at 20 °C, and then washed three times with
PBS and four times with TBS. Coverslips were incubated for 30 min at
20 °C with secondary antibodies (and 6 µM TOTO-3 when
staining for DNA) in blocking solution containing 20 mM
Tris-Cl followed by washing five times with TBS and twice with
H2O before mounting with the ProLong Antifade Kit from
Molecular Probes. Of note, control experiments were performed to
exclude the possibility of nonspecific labeling or, in the case of
double-labeling, cross-reactivity of secondary antibodies with primary
antibodies as well as bleed-through from one channel into another.
Cell Cycle Analysis by Fluorescence-activated Cell Sorting (FACS)
and Flow Cytometry--
The DNA distribution profiles of the cyclin
G2, GFP-expressing, and null cells were determined through a
combination of cell sorting and flow cytometry using a
FacstarPLUS dual laser cytometer (Becton Dickinson, San
Jose, CA). Cultures transfected with either a cyclin G2-GFP or control
GFP expression vector were harvested, washed, and resuspended at
1.5 × 106 cells/ml in media supplemented with
20 µg/ml of the cell-permeable DNA dye Hoechst 33342 and incubated
for ~45 min at 37 °C. Cell suspensions were then filtered through
a 70-µm Nitex nylon mesh (Sefar America, Inc.) and further incubated
at room temperature with 5 µg/ml of the cell-impermeable DNA dye
propidium iodide (PI) for 5 min. The cells were sorted with the
cytometer's argon I-90 laser set at 488 nm in the primary position
(GFP, G2-GFP) and the krypton laser 300 series tuned to a multiline
ultraviolet spectrum in the secondary position for Hoechst 33342 (DNA).
The GFP signal was collected through a 450/50 band pass filter. The PI
signal was excited with the argon laser and collected through a 660/22
band pass filter. Both lasers were set with 100 milliwatts of light.
Each sample was sorted, and data were collected (50,000 events) using
CellQuest acquisition and analysis software (Becton Dickinson). Final
analysis of the collected data was done using FlowJo 3.2 software (Tree
Star, Inc., San Carlos, CA). PI-positive cells and doublets were
excluded to ensure that only single viable cells were used for analysis
of DNA content and GFP expression.
BrdUrd incorporation was determined to measure relative amounts of DNA
synthesis as described (8, 9) with the following modifications. The
BrdUrd (20 µM) pulse (15 min)-labeled cells were fixed in
a PBS buffer containing 10 mM EDTA and 0.5%
paraformaldehyde for 5-10 min to retain the GFP signal and washed
again prior to fixation and permeabilization in ice-cold 70% methanol.
The incorporated BrdUrd present in the denatured DNA was detected with
an unlabeled mouse anti-BrdUrd monoclonal primary antibody (40 ng/µl;
Caltag) in conjunction with an Alexa-660-labeled goat anti-mouse
secondary antibody (1:500; Molecular Probes). The cell populations in
each transfection were measured according to the amount of green
fluorescence (G2-GFP, GFP) or no fluorescence (nonexpressing) DNA
content (PI) and BrdUrd incorporation (Alexa 660 fluorescence) by flow
cytometry using a FACSCalibur cytometer equipped with two air-cooled
lasers. 50,000 events were acquired using CellQuest software (Becton
Dickinson). The individual cells of each of these segregated
populations were then analyzed for their DNA content profile and degree
of DNA synthesis (BrdUrd incorporation) using FlowJo 3.4 analysis software.
Kinase and Phosphatase Assays--
HEK 293 and CHO cells were
transfected with pcDNA3-G2GFP, pcDNA3-G2V5, or pcDNA3-GFP
and for phosphatase assays cotransfected with pcDNA3-HAPP2A/C. For
CDK2 kinase activity assays, the G2GFP- or GFP-transfected cell
population was collected ~27-30 h post-transfection (~60%
confluence) and sorted live via an SE Vantage high speed FACS cytometer
(Becton Dickinson) into GFP-positive and negative cell populations
using an argon laser tuned to 488 nm. The cytometer was set at 35 p.s.i. with a droplet frequency of 5900 kHz. Doublets were excluded
using "time of flight" measurements. The expressing population
collection gate included cells in the upper two-thirds of GFP signal
intensity, and nonexpressor gate was limited to those under the first
decalog with the lowest GFP signal. The sorted cell populations
(~5 × 106) were kept at 2-4 °C and concentrated
by centrifugation. Cell pellets were lysed in radioimmune precipitation
buffer (supplemented with 25 mM NaF, 20 mM
-glycerol phosphate, 10 mM EGTA, 20 mM NaPPi, and 2 µM microcystin), and the lysates
were cleared of insoluble material by centrifugation at 21,000 × g for 10 min. Immunoprecipitations of CDK2 were performed in duplicate
from cleared supernatants containing equivalent amounts (80 or 100 µg
each) of protein using 10 µg of rabbit anti-CDK2 antibodies (sc-54;
Santa Cruz Biotechnology). The anti-CDK2 and control immunocomplexes were washed three times with buffer (50 mM Hepes, pH 7.4, 10 mM MgCl2, and 5 mM
MnCl2), incubated for 30 min at 37 °C in 40 µl of
kinase buffer (50 mM Hepes, pH 7.4, 10 mM
MgCl2, and 5 mM MnCl2, 1 mM dithiothreitol, 10 µM ATP, 2 µg of
histone H1 (Calbiochem) supplemented with 10 µCi of
[-32P]ATP). Reactions were analyzed by SDS-PAGE and
autoradiography. In parallel, duplicate CDK2 and control
immunocomplexes and total lysate (80 µg of protein) were
immunoblotted with anti-CDK2.
For phosphatase assays, cell pellets harvested from the various
transfections or rat brain cerebellum were solubilized in modified
radioimmune precipitation buffer (without phosphatase inhibitors), and
cleared lysates containing 1.5 or 3.0 mg of protein, respectively, were
subjected to immunoprecipitation in triplicate with either rabbit
anti-cyclin G2 or control IgG. The immunocomplexes were collected by
centrifugation, washed three times with phosphatase wash buffer (150 mM NaCl, 15 mM Tris, pH 7.4, 1% Nonidet P-40, 0.4% deoxycholate, and 0.05% SDS), and washed one time with
phosphate-free double-distilled H2O, and then bound
phosphatase activity was determined by measuring the phosphate released
following incubation with a target phosphopeptide (RRApTVA; where pT
represents phosphothreonine). Amounts of released phosphate were
determined using a malachite green/molybdate-based assay following
protocols prescribed by the manufacturer (serine/threonine phosphatase
assay system; Promega) with modifications. Immunocomplexes were mixed
at room temperature in 50 µl of assay buffer (330 µM
phosphopeptide, 165 µM imidazole, pH 7.2, 660 µM EDTA, 0.066% -mercaptoethanol, 0.066% bovine
serum albumin) for 2 h. 45 µl of the reaction eluate was
collected after centrifugation and added to a 96-well microtiter plate
followed by the addition of an equal volume of the molybdate/malachite green dye. After 15 min at room temperature, absorbance was measured at
600 nm. The amount of phosphate released in the immunocomplexes was
calculated from a curve of phosphate standards run in parallel.
| |
RESULTS |
Cellular Effects of Ectopic Cyclin G2 Expression and Its
Subcellular Distribution--
Several trials to establish stable cell
clones constitutively or inducibly expressing cyclin G2 yielded only
weak expressors rapidly lost upon expansion. Therefore, transient
transfection assays were utilized to characterize cyclin G2. All
figures shown were obtained with an anti-cyclin G2 antibody produced
against a full-length cyclin G2-GST fusion protein. The specificity of its immunoreactivity was confirmed on immunoblots of cyclin G1-GST, G2-GST, and GST alone (cyclin G2 antibodies do not cross-react with
cyclin G1; Fig. 1A,
lanes 4-6) and in parallel assays with anti-cyclin G2-NT antibodies raised against a peptide derived from
NH2-terminal sequences of cyclin G2 (data not shown).
Anti-cyclin G2 antibodies recognized a single protein band of about 48 kDa and a group of bands just below 70 kDa in total cell lysates of HEK293 cells expressing cyclin G2-V5His and G2-GFP fusion proteins, respectively (Fig. 1A, lanes 1 and
2). Apparent Mr values of the immunoreactive bands are as expected for the fusion proteins. The
detection of multiple bands migrating within the same
Mr range in lysates from cultures expressing
G2-GFP probably reflects limited proteolytic breakdown of the cyclin
G2-GFP protein. No immunoreactive bands were detectable in cell lysates
from HEK293 cells expressing GFP alone (Fig. 1, lane
3).
View larger version (40K):
[in this window]
[in a new window]
|
Fig. 1.
Characterization of anti-cyclin G2 antibodies
and the effect of cyclin G2 overexpression. HEK293 or CHO cells
were transfected with pcDNA3 expression constructs encoding
either G2-GFP, G2-V5His, or GFP alone. A, for
immunoblotting, 50 µg of protein from HEK293 lysates cleared by
centrifugation (lanes 1-3) and 5-10 ng of
affinity-purified GST fusion proteins (lanes
4-6) were loaded as indicated. Blots were probed with
affinity-purified anti-cyclin G2 antibody, followed by HRP-conjugated
protein A. The migrations of the immunoreactive fusion proteins present
in each lane are indicated by the arrows, and the apparent
molecular masses of the protein standards are depicted at the
left side of the blot. B-D,
immunofluorescence confocal microscopy of cyclin G2-expressing HEK293
(B and C) or CHO (D) cell cultures.
Cultures expressing cyclin G2-V5His (B (left
side of panels) and D) or
cyclin G2-GFP (B (right sides) and
C) were fixed either before permeabilization with Triton
X-100 (B and C) or after preextraction with
Triton X-100 (D) and stained with affinity-purified
anti-cyclin G2 antibody (red), followed by LRSC-conjugated
donkey anti-rabbit IgG (B (middle) and
D (left)), with anti-V5His, followed by
FITC-conjugated donkey anti-mouse IgG (B (left
side of left panel) and D
(middle)), with anti-lamin B antibodies in conjunction with
LRSC-conjugated donkey anti-goat IgG (C (red in
left panel); anti-lamin B immunosignals are also
separately shown in the middle panel in
black and white for better contrast),
and with TOTO-3 for DNA detection in transfected and untransfected
cells (pseudocolored blue in B (right)
and C (left) and shown in black
and white in D (right)).
The percentages of cells with abnormal nuclei upon cyclin G2 expression
in experiments represented by the images were 50%
(B, left) and 64% (B,
right), 55% (C), and 54% (D).
|
|
HEK293 and CHO cultures expressing cyclin G2-GFP and G2-V5His were
labeled with anti-cyclin G2 (red) and in the latter case anti-V5 (green) antibodies (Fig. 1). No anti-cyclin
G2 immunoreactivity was observed in cells not expressing cyclin G2
fusion proteins (see no GFP or anti-V5 signal in Fig. 1B;
nuclei detected with the DNA dye TOTO-3 pseudocolorized in
blue). Accordingly, the anti-cyclin G2 antibodies
specifically recognize the ectopically expressed cyclin G2 fusion
proteins and do not cross-react with other proteins in these cells.
Lipid-mediated transfection of cytomegalovirus-driven expression
vectors for cyclin G2-GFP and G2-V5His resulted in an expression
frequency of about 15-20% of the HEK293 and 30-60% of the CHO cell
populations 20-24 h post-transfection. In contrast to control GFP
transfectants, the number of cyclin G2-expressing cells decreased
substantially in both cell lines by 48-56 h, suggesting inhibitory
effects of ectopic cyclin G2 expression on cells. Furthermore, DNA
staining revealed an aberrant, fragmented, or multilobed appearance of
the nuclei; at 29-34 h post-transfection, ~60% of the cyclin
G2-positive cells had abnormal nuclei (see Fig. 1B). Less
than 10% of mock or empty GFP vector-transfected cells showed nuclear
abnormalities. To further characterize these nuclear abnormalities,
HEK293 cultures expressing cyclin G2 fusion proteins were immunostained
with antibodies against lamin B, a structural protein of the nuclear
envelope that is degraded by caspases in apoptotic cells. Cells
expressing cyclin G2-GFP (Fig. 1C, left
panels, green) contained nuclei with multiple
DNA-stained lobes (Fig. 1C, right
panels), each surrounded by a lamin B (Fig. 1C,
middle panels)-positive membrane, indicating an
intact nuclear envelope. This observation, together with the absence of
DNA overcondensation in the majority of these cyclin G2-overexpressing
cells (see Fig. 1, B-D), argues that the principal nuclear
aberration observed does not reflect apoptotic nuclei but rather
results from defect in a mitotic or cytokinetic process. Furthermore,
TUNEL staining did not reveal an obvious correlation of the cyclin
G2-expressing cells displaying nuclear aberrations with TUNEL-positive
apoptotic nuclei (data not shown).
Cyclin G2-GFP as well as G2-V5His is detectable throughout the
cytoplasmic region and often but not always in the nucleus. In many CHO
cells (e.g. Figs. 1D and 4D) and also
some HEK293 cells, this distribution appears rather punctate (Fig.
1C and data not shown). In addition, spots with strong
cyclin G2-GFP fluorescence or cyclin G2 immunosignals are visible in
the perinuclear region of HEK293 cells. An overlay of the red and green
channels results in a yellow color throughout most of the cyclin
G2-expressing cells. Because of the often quite pronounced punctate
distribution pattern for cyclin G2, we examined whether cyclin G2 is
associated with a Triton X-100-insoluble fraction of transfected cells.
In fact, preextraction of HEK293 cells (data not shown) and CHO cells for 90 s with Triton X-100 in the cytoskeleton-stabilizing PHEM buffer prior to fixation revealed that a large portion of cyclin G2 was
not extractable (Figs. 1D, 4, and 6E). In
contrast, other proteins such as BiP, a marker protein for the
endoplasmic reticulum and ectopically expressed GFP, which behaves as a
soluble cytosolic marker, were efficiently removed by this Triton X-100
preextraction method (data not shown). Our data indicate that cyclin G2
is to a significant extent associated with Triton X-100-insoluble
subcellular structures (e.g. nuclei and the cytoskeleton).
Cyclin G2 Expression Promotes Cell Cycle Arrest--
Our earlier
observations of up-regulation of endogenous cyclin G2 expression during
cell cycle arrest and apoptosis prompted us to examine the effect of
ectopic cyclin G2 expression on cell cycle progression in CHO and
HEK293 cells. Live cell cultures ectopically expressing cyclin G2GFP or
GFP were harvested and stained with the cell-permeant DNA dye Hoechst
33342 and the cell-impermeant DNA dye PI. Flow cytometric
analysis revealed the DNA content of the GFP-positive and PI-negative
cells (Fig. 2). DNA content profiles of
the nonexpressing (GFP-negative) fractions of both populations appeared
normal, exhibiting no appreciable difference between themselves and
nontransfected cycling cell populations (Fig. 2A and data
not shown). However, populations expressing moderate and high levels of
cyclin G2GFP revealed a strong G1 phase cell cycle arrest
by 21 h post-transfection. Each cyclin G2-positive population
showed a considerable accumulation of cells with a G1 phase
DNA content concomitant with a decrease of cells in both the S and
G2 phases. In contrast, control cells expressing GFP alone
continued to grow asynchronously with cell cycle distributions comparable with those of parental cells lacking GFP expression (Fig 2A). Similar results were observed with HEK cyclin G2
and GFP transfectants, the percentage of cyclin G2-positive
cells accumulated in the G1 phase (73%) relative to the
normal cell cycle distribution of GFP alone and parental control cells.
A similar accumulation of G1 phase cells was observed with
cells expressing untagged cyclin G2 (coexpressed with GFP) but never in
the control GFP-alone populations (see below). Comparison of the DNA
profiles of CHO and HEK cyclin G2GFP expressors with nonexpressors and
parallel GFP control populations harvested after longer times in
culture showed a clear accumulation of cells with a broad
G1/S phase distribution and an alteration of the
G2/M phase profile suggestive of aneuploidy (data not
shown).
View larger version (56K):
[in this window]
[in a new window]
|
Fig. 2.
Flow cytometry and cell cycle analysis of
cyclin G2-expressing cells. A, analysis of DNA content
and GFP signal intensity by flow cytometry of intact (propidium
iodide-negative) live CHO cells transfected with either cyclin G2GFP or
control GFP expression plasmids, stained with the cell-permeant DNA dye
Hoechst 33342, 21 h post-transfection. The amounts of DNA
(Hoechst) relative to the expressed GFP signal in a total population of
transfected cells are shown as dot plots (upper
panels) and DNA profiles of the expressing versus
nonexpressing gated populations in the labeled histograms
below each corresponding dot plot. We determined the
percentage of cells in the G1, S, and G2 + M
phases (given at the right for each DNA histogram with the
FlowJo Watson Pragmatic cell cycle analysis program; similar results
were obtained applying the Dean-Jett-Fox cell cycle modeling
algorithm). B, bivariate flow cytometric analysis of the DNA
content (propidium iodide) versus GFP signal and new DNA
synthesis (BrdUrd incorporation detected with an anti-BrdUrd primary
and Alexa 660 fluorophore-labeled secondary antibody) of CHO cells
transfected with cyclin G2GFP DNA (27 h) in dot plot format. The
left panel shows the DNA profile of the total
transfected culture and the population gates set for analysis of BrdUrd
incorporation shown in corresponding labeled dot plots in the
right two panels. The percentage of
cells incorporating the BrdUrd label into DNA during the 15-min pulse
of each population is shown to the upper right of
the correspondingly shaded gate (identical for both populations), and
the calculated cell cycle distribution derived from the total DNA dot
plot is shown on the side. C, CDK2 and CDK4
kinase activity of FACS-sorted live populations of CHO cells expressing
or not expressing either cyclin G2GFP or GFP. The upper and
middle panels on the left and
lowest panel on the right are
autoradiograms of 32P-labeled histone H1 phosphorylated by
CDK2 isolated by immunoprecipitation from identical amounts (80 µg)
of lysates prepared from the nonexpressors (no), expressors
(high), and total population of cells, as indicated. No
kinase activity was detectable after immunoprecipitation with control
nonimmune IgG (NRS IgG; middle panel
on the left). Bottom panel on
left, anti-CDK2 immunoblot of lysate (right) and
CDK2 immunoprecipitated in parallel from identical amounts of the same
extract used in the kinase assay shown in the middle
panel. Quantification of the above signals indicated an
approximate 7-13-fold reduction of CDK2 kinase activity in cyclin
G2-expressing cells. The top panel on the
right shows the specificity of CDK4 immune complexes
compared with control IgG for RbGST phosphorylation (autoradiogram of
32P-labeled Rb-GST). The middle two
panels on the right are the autoradiogram of the
CDK4 activity toward Rb-GST in anti-CDK4-immunoprecipitates from sorted
and total population lysates (indicated at the top of the
right panels; 200 µg of protein) and the
respective CDK4 immunoblot signals. In parallel, a determination of
histone H1 activity in anti-CDK2 immunoprecipitates (lower
right) was obtained from the same lysates (100 µg of
protein) of the respective sorter populations used for the CDK4
assay.
|
|
To further dissect the cell cycle status of cyclin G2-expressing
populations, cyclin G2GFP and control GFP transfectant cultures were
pulse-labeled with BrdUrd for 15 min immediately prior to harvest (28 h
post-transfection) and analyzed for the amount of newly synthesized DNA
via measurement of BrdUrd label incorporation. GFP expression in the
control did not alter BrdUrd incorporation profiles relative to the
nonexpressing population of the same transfection (Fig. 2B)
and the nontransfected parental control (data not shown). In contrast,
the comparable population of cyclin G2GFP-positive cells exhibited very
little BrdUrd incorporation (5%), with the majority of the cells
arrested in G1 phase (Fig. 2B). Longer pulses of
45 min revealed a similar inhibition of DNA synthesis in cyclin
G2GFP-expressing cells (9% BrdUrd incorporation) compared with 31% of
the nonexpressing cells and control GFP-expressing cells (data not shown).
CDK2 kinase activity is crucial for normal transition of a cell through
the G1/S restriction point and passage through S phase. To
determine whether CDK2 activity was reduced in cyclin G2GFP-expressing CHO and HEK cells compared with GFP control cells, moderate to high
expressing cells and nonexpressing cells of each culture were collected
live from the same transfected culture via cell sorting.
Immunoprecipitations were performed in duplicate with anti-CDK2 and
control IgG from identical protein amounts (80 µg) of lysate of each
sorted population. Immunocomplexes were incubated with histone H1
substrate and [-32P]ATP followed by SDS-PAGE and
autoradiography to test for CDK2 activity (Fig. 2C,
upper panel). CDK2 activity immunoprecipitated from cyclin G2GFP-expressing CHO cells was significantly reduced in
comparison with the corresponding population of control GFP expressors
and nonexpressing cell populations of each transfection (Fig.
2C, upper left). Similar results were
obtained in HEK293 cells (data not shown). Next, the
CDK2-dependent histone H1 kinase activity of CHO cell
cyclin G2GFP-expressing and nonexpressing populations was compared with
the amount of CDK2 immunoprecipitated and present in each lysate (Fig.
2C, lower left panels). No
kinase activity was precipitated by preimmune antisera
(NRS), and again the cyclin G2GFP-positive population
contained a dramatically reduced CDK2 activity. Quantification of the
kinase activity relative to the amount of CDK2 immunoprecipitated or
present in total lysate indicated about a 10-fold reduction in
CDK2-specific kinase activity in the cyclin G2-expressing population
compared with the nonexpressors (Fig. 2C, lower
left two panels). To determine whether
the cell cycle block in cyclin G2-expressing cells was in early
G1 phase or at the G1/S phase boundary, a
similar analysis of CDK4 activity was performed using Rb-GST as a
kinase substrate. CDK4 activity is critical for entry into and
transition through the early G1 phase. Interestingly,
cyclin G2 expressors contained increased levels of CDK4 activity
relative to that in nonexpressors and the total population (Fig.
2C, right panels), probably due to the
accumulation of cells in the early G1 phase. These data
indicate that cyclin G2-expressing CHO and HEK293 cells can enter and
progress through early G1 phase but cannot effectively move
beyond the G1/S phase of the cell cycle.
Cyclin G2 Is Associated with PP2A B'3 and B' Subunits in
Living Cells--
We were unable to find any typical CDK activity or a
cell cycle-related CDK associated with cyclin G2. Because cyclin G1
associates with B' regulatory subunits of PP2A, an enzyme that also
plays a critical role in cell cycle progression, we tested for cyclin G2-PP2A interactions. PP2A B' subunits are encoded by six different genes that give rise to at least nine distinct proteins, including splice variants. Varying nomenclatures categorizing these proteins have
been instituted (31, 34, 35, 38, 42). Csortos et al. (31)
cloned and described four different rabbit B' subunits (B'
-
),
including B' and B' subunits. Okamoto et al.
determined that two murine B' subunits interact with cyclin G1, which
they termed B' and B' (44). These are the respective murine
homologs of the rabbit B' and B' subunits (31).
We investigated the potential interaction of cyclin G2 and cyclin G1
with the rabbit PP2A B' and B' subunits (nomenclature established
in Ref. 31). Initially purified and immobilized GST, cyclin G2-GST,
cyclin G1-GST, and polyhistidine-tagged cyclin G1 and cyclin G2 fusion
proteins were tested in vitro for interactions with
HA-tagged B' or B' splice variant B'2, B'3, or B'4 (31) present in transfected cell lysates. All of the HA-tagged subunits bound to cyclin G2 as well as cyclin G1-GST and polyhistidine-tagged fusion proteins but not to control GST (data not shown). These results
indicated that in vitro, cyclin G2 interacts with splice variants of PP2A/B' and PP2A/B' and that cyclin G1 binds not only
B' and splice forms of B' but also B' subunits.
To investigate the possible association of cyclin G2 with PP2A B'
subunits in mammalian cells, lysates were obtained from HEK293 cultures
transiently expressing cyclin G2 with and without HA-tagged B'
subunits. Anti-cyclin G2 immunocomplexes isolated from lysates of these
transfected cultures were immunoblotted with anti-HA antibodies.
Immunoreactive proteins migrating with an apparent
Mr of 62,000 and 69,000, corresponding to
HA-tagged B'3 and B', respectively, were present in cyclin G2
immunocomplexes isolated from the corresponding cyclin G2-PP2A/B'
cotransfectants (Fig. 3A,
top, lanes 5, 6,
8, and 9) but not in those from control GFP-PP2A/B' cotransfectants (Fig. 3A, top,
lanes 2 and 3) or singly transfected
controls (Fig. 3A, top, lanes
1, 4, and 7). Immunoprecipitates isolated with GFP-specific antibodies from control cultures transfected with an expression vector for GFP with or without vectors for the
HA-tagged B' subunits did not contain any HA-immunoreactive proteins
(Fig. 3A, top, lanes
10-12). This experiment also suggests that B'
coprecipitated with cyclin G2 more efficiently than B'3; significantly more PP2A/B' was present in cyclin G2 immunocomplexes isolated from G2-GFP-transfectant lysates (Fig. 3A,
top, lanes 5 and 6) despite
the presence of similar amounts of GFP-tagged G2 in these samples (Fig.
3A, bottom, lanes 5 and
6). Although much smaller amounts of cyclin G2
immunoprecipitated from lysates of the cyclin G2-V5His plus B'
cotransfectants (relative to the B'3 cotransfectants), similar
amounts of these two HA-tagged B' subunits copurified with the two
cyclin G2 immunocomplexes (Fig. 3A, bottom,
lanes 8 and 9).
View larger version (87K):
[in this window]
[in a new window]
|
Fig. 3.
Co-immunoprecipitation of cyclin G2 with PP2A
B' subunits. HEK293 cells were co-transfected with DNA constructs
encoding GFP, G2-GFP, or G2-V5His and either an irrelevant protein
(Cont.) or various HA-tagged PP2A B' subunits as indicated
at the top of each panel. A, cleared
cell lysates containing 400 µg (lanes 1-12),
500 µg (lanes 13-15), or 1 mg of total protein
(lanes 19-21) were used for immunoprecipitations
with anti-G2 or anti-GFP antibodies as depicted at the
bottom (IP). Lysates containing 200 µg
(lanes 16-18) or 50 µg of protein
(lanes 22 and 23) were also directly
loaded. Immunoblots were incubated with anti-HA antibodies
(top panels) followed by either HRP-conjugated
goat anti-mouse IgG (lanes 1-18) or rat
anti-mouse Ig light chain (lanes 19-23). The
latter was used to probe for PP2A/B'4, which migrates close to the
IgG heavy chain, to avoid detection of the heavy chain of the
immunoprecipitating mouse anti-HA antibody (as observed in some
exposures of blots probed with the goat anti-mouse IgG; e.g.
the band at 50 kDa in lanes 1-12). Blots were
stripped with SDS (53) and reprobed with anti-cyclin G2 antibodies,
followed by HRP-labeled protein A (bottom
panels). B, cleared cell lysates containing 1 mg
(lanes 1-4 and 18-20) or 800 µg of
total protein (lanes 5-11) were used for
immunoprecipitations with anti-G2, anti-HA, or control antibodies as
given at the bottom (IP). Lysates containing 100 µg (lanes 12-14) or 50 µg of protein
(lanes 15-17) were also directly loaded. Blots
were incubated with anti-cyclin G2 antibodies, followed by HRP-labeled
protein A (top panels). Blots were stripped with
SDS (53) and reprobed with anti-HA (bottom
panels) followed by either HRP-conjugated goat anti-mouse
IgG (lanes 1-14) or rat anti-mouse Ig light
chain (lanes 15-20) to avoid detection of the
heavy chain of the immunoprecipitating antibody when probing for
PP2A/B'4, which migrates close to the IgG heavy chain. Positions of
marker proteins are depicted at the left or right
of each blot (in kDa).
|
|
For a more quantitative analysis, analogous immunoprecipitation
experiments using anti-cyclin G2 antibodies were performed from lysates
containing 500 µg of protein followed by immunoblotting and
immunosignals compared with those obtained from direct loading of 100 µg of the respective lysates. Under these conditions,
immunoprecipitation with anti-cyclin G2 yielded amounts similar to or
slightly higher than those present in the directly loaded samples (Fig.
3A, bottom; compare lanes
13-15 with lanes 16-18). The
immunosignals for HA-tagged B'3 and B' subunits coprecipitated
with cyclin G2 are roughly 10-20% and above 50%, respectively, of
those in the corresponding whole lysates (Fig. 3A,
top; compare lanes 14 and
15 with lanes 17 and 18).
Collectively, these results suggest that 10-20% of HA-tagged
PP2A/B'3 and 50% or more of HA-tagged PP2A/B' in these lysates
are associated with cyclin G2. HA-tagged PP2A/B'4
coimmunoprecipitated with cyclin G2-V5His and G2-GFP, and the extent of
these interactions is similar to that between cyclin G2 and PP2A/B'3
(Fig. 3A, lanes 19-23).
Complementary results were achieved by immunoprecipitation of HA-tagged
B' subunits with anti-HA antibodies and immunoblotting with anti-cyclin
G2 antibodies. Cyclin G2-GFP and G2-V5His fusion proteins were
coimmunoprecipitated with B' subunits from cell lysates of the
B'-cyclin G2 cotransfectants (Fig. 3B, top,
lanes 9, 11, 19, and
20) but not from singly transfected cyclin G2 or B' subunit
expression only controls (Fig. 3B, top,
lanes 6, 7, and 15).
Reprobing this blot with anti-HA antibodies demonstrated again that
relatively more cyclin G2 is associated with B' than with B'3 in
transfected cell extracts; the cyclin G2 signal is several times
stronger after immunoprecipitation from B' than from B'3
cotransfectants (Fig. 3B, top, lanes
9 and 11), although the anti-HA signal is
comparatively less in the former than in the latter sample (Fig.
3B, bottom, lanes 9 and
11).
Colocalization of Cyclin G2 with PP2A B' Subunits in Mammalian
Cells--
To determine whether cyclin G2 and B' subunits are
colocalized in vivo, HEK293 and CHO cells were singly or
doubly transfected with expression constructs for cyclin G2-GFP or
G2-V5His and for HA-tagged B', B'4 or B'3 (Fig.
4). Similar to cyclin G2 (see Fig. 1,
B-D), immunoreactivity for HA-tagged B' subunits is present throughout the cytoplasmic region and often in the nucleus, whether expressed with (Fig. 4A, upper panels)
or without (Fig. 4, B and C) cyclin G2. Both
B'3 and B' contain classic bipartite nuclear localization signals
in their COOH termini, and the human B' homolog has been shown to be
both nuclear and cytoplasmic (31, 38, 40, 42). Although B'4 lacks
this nuclear targeting sequence, it still efficiently localizes to the
nucleus (31, 38). Clear examples of nuclear colocalization are seen in
the cyclin G2-PP2A/B'- and B'3-cotransfected CHO cells shown in Fig 4, D-F. Although nuclear colocalization was more
pronounced in CHO cells, it was also observed in several HEK293
cotransfectants (data not shown). B' and B' partially rescue the
growth defect of the mutant Rts1p, a B' subunit in yeast (43). Rts1p is
a cytoplasmic protein, whereas B' is both nuclear and cytoplasmic. B' complementation of the Rts1p mutant results in abnormal
morphology (43). Of note, in some cases (Fig. 4 and data not shown)
cells that overexpress PP2A/B' or - contained multilobed or
fragmented nuclei similar to cyclin G2-GFP- or G2-V5His-expressing
cells (Fig. 1, B-D). The observations that overexpression
of either cyclin G2 or B' subunits often results in comparable
nuclear aberrations suggest that they are involved in coordinating
similar processes.
View larger version (60K):
[in this window]
[in a new window]
|
Fig. 4.
Colocalization of PP2A B' subunits with
cyclin G2 by immunofluorescence confocal microscopy. CHO cells
were transfected with plasmids for expression of HA-tagged
(A, red) B'3 or B' (B and
C, red) and cyclin G2-GFP (A and
B, green) or G2-V5His (C,
green). After 24-29 h, cells were fixed either before
permeabilization with Triton X-100 (A, upper
panel) or after preextraction with Triton X-100
(A (lower panel), B, and
C). Cultures were stained with anti-HA antibodies followed
by LRSC-conjugated donkey anti-mouse IgG to visualize HA-tagged
PP2A/B'4 and B' and with TOTO-3 for labeling of DNA
(blue channel in right
panels). Those cultures co-expressing V5His-tagged cyclin G2
were also labeled with anti-cyclin G2 antibodies, followed by
FITC-conjugated donkey anti-rabbit IgG (C).
|
|
Distribution in the cytoplasmic region for cyclin G2 fusion proteins
expressed alone (Fig. 1, B-D) or together with exogenous B'
subunits (Fig. 4) varied between a smooth and a quite punctate pattern.
Similar variations were seen in the staining patterns for HA-tagged
PP2A B' subunits whether expressed alone or in combination with cyclin
G2 (Fig. 4 and data not shown). In CHO cells, a punctate distribution
prevailed for cyclin G2 and B' subunits (Fig. 4), with occasional
smooth distributions (data not shown). Preextraction of HEK293 and CHO
cells with Triton X-100 does not abolish cyclin G2 signals in singly
transfected cells (Fig. 1D and data not shown) or in cells
co-expressing B' subunits (Fig. 4A, lower
panels, B and C), indicating that
cyclin G2 is associated with detergent-resistant subcellular
structures. The same is true for B' subunits coexpressed with GFP
rather than cyclin G2; after preextraction with Triton X-100, GFP
signals were completely gone, indicating that GFP is efficiently
solubilized by the detergent (see Fig. 6F and data not
shown). In contrast to GFP but similar to cyclin G2, PP2A/B' immunoreactivity was still present after preextraction (data not shown). The same was true for PP2A/B'3 expressed with
(e.g. Fig. 4A, lower panel)
or without cyclin G2 (data not shown). G2-GFP- or G2-V5His-positive
dots (Fig. 4) often, though not always, matched those for B'3 (Fig.
4A) or (Fig. 4, B and C), whether
cells were preextracted with Triton X-100 (Fig. 4, A,
lower panel, B, and C) or
not (Fig. 4A, upper panel). In
general, preextraction did not change the staining pattern for either
cyclin G2 or B' subunits (e.g. compare the unextracted cell
in Fig. 4A, upper panel, with the
preextracted cells in Fig. 4B). Untransfected cells
(visualized with the TOTO-3 DNA stain; blue
channel in right panels in Fig. 4) did
not show any immunofluorescence for anti-HA or anti-cyclin G2
antibodies, confirming the antibody specificities. Collectively, our
data show that both cyclin G2 and B' subunits are to a significant
degree colocalized with each other before and after Triton X-100
extraction. Accordingly, cyclin G2 and B' subunits are not just
codistributed throughout the cytoplasm like soluble cytosolic proteins
such as GFP. Rather, cyclin G2 and PP2A/B' and PP2A/B' are part
of subcellular structures resistant to Triton X-100. Our findings
indicate that cyclin G2 and PP2A/B' and - are present at the same
subcellular compartments, an important criterion for establishing their
interaction in vivo.
Cyclin G2 Associates with B' but Not A Subunits of
PP2A--
Association of cyclin G2 with B' subunits could reflect the
existence of a cyclin G2 multimeric complex containing the A and C
subunits of PP2A, or it could sequester B' subunits away from a PP2A
core complex. Because B' subunits are thought to target and modulate
the catalytic activity of PP2A/C, it is important to define which
subunits can form complexes with cyclin G2. Since PP2A/A is a
scaffolding protein for B and C subunits, we first tested if endogenous
or overexpressed PP2A/A present in HEK293 lysates interacts with
immobilized cyclin G1- and cyclin G2-GST as found for PP2A/B' subunits.
Several attempts to pull down either endogenous or overexpressed PP2A/A
with immobilized cyclin-GST fusion proteins yielded negative results;
the A subunit did not bind whether ectopically expressed HA tagged B'
subunits were present or not in the lysates (data not shown; see Fig.
6A, upper panels). Next, PP2A/B'3
or B' were co-expressed with cyclin G2-GFP or G2-V5His for
coimmunoisolation with anti-cyclin G2 antibodies (see Fig. 3) followed
by immunoblotting with A subunit-specific antibodies. Despite the
presence of HA-tagged B'3 and B' (Fig. 5A, left
top) associated with cyclin G2 (Fig. 5A,
left bottom), we could not detect the A subunit
in these immunocomplexes (Fig. 5A, left
middle). Reciprocal immunoprecipitations with antibodies against the A subunit from corresponding cell lysates and subsequent immunoblotting with anti-cyclin G2 antibodies did not indicate any
association of cyclin G2-immunoreactive proteins with the A subunit
(Fig. 5A, right bottom). In contrast,
immunosignals for HA-tagged B' subunits present in the anti-PP2A/A
immunocomplexes were at least as strong as those for B' subunits
present in cyclin G2 precipitates (Fig. 5A, compare
right and left panels on
top), indicating that an amount of B' subunit-containing
protein complexes were immunoprecipitated that was similar if not
greater than that precipitated with anti-cyclin G2. As expected,
reprobing of the blots of anti-PP2A/A immunoprecipitates with
anti-PP2A/A antibodies yielded strong immunoreactive bands of ~62 kDa
(Fig. 5A, right middle). Although it
is impossible to rule out the presence of minor amounts of PP2A/A in
cyclin G2 complexes, these results show that a large portion of cyclin
G2-PP2A/B' complexes do not contain PP2A/A and vice versa.
View larger version (74K):
[in this window]
[in a new window]
|
Fig. 5.
Cyclin G2 association is detectable with B'
but not A subunits. Lysates were prepared from HEK293 cells
expressing of GFP, G2-GFP, or G2-V5His and HA-tagged PP2A B' or A
subunits as indicated at the top (A) or
bottom (B) of each panel
(Transfect). A, lysates containing 500 µg of
protein were used for immunoprecipitations with anti-cyclin G2
(left panels) or rat anti-PP2A/A antibodies
(clone 6F9; right panels) and immunoblotting.
Blots were initially probed with anti-cyclin G2 (bottom
left) and/or rat anti-PP2A/A (clone 6G3; middle
right). Blots were then stripped with SDS and reprobed
in reverse order before a final stripping and reprobing with the
anti-HA antibody. HRP-coupled secondary antibodies were rabbit anti-rat
IgG and goat anti-mouse IgG for probings with anti-PP2A/A and anti-HA
antibodies, respectively. Anti-cyclin G2 antibodies were detected with
HRP-labeled protein A. B, cleared cell lysates containing 1 mg of protein were used for immunoprecipitations with anti-cyclin G2
(left panels) or rat anti-PP2A/A antibodies
(clone 6F9; right panels). Lysates containing 100 µg of protein (lanes 7-9) were also directly
loaded. Blots were probed with rat anti-PP2A/A (clone 6G3), followed by
HRP-coupled rabbit anti-rat IgG, and stripped and reprobed with
anti-cyclin G2, followed by HRP-labeled protein A.
|
|
To test whether overexpression of PP2A/A with cyclin G2 could force
their interaction, both proteins were ectopically expressed in HEK293
cells and immunoisolated with anti-cyclin G2 and anti-PP2A antibodies.
Anti-cyclin G2 immunoprecipitates did not contain any detectable PP2A/A
(Fig. 5B, top left panel,
lanes 1, 3, and 5; data not
shown) in contrast to that seen in total cell lysates of both
overexpressed and endogenous PP2A/A (Fig. 5B, top
left, lanes 7-9) and the presence of
abundant amounts of cyclin G2 in the anti-cyclin G2 immunocomplexes
(Fig. 5B, bottom left). Likewise, cyclin G2 was not detectable in anti-PP2A/A immunoprecipitates (Fig.
5B, lower right; data not shown), but
reprobing of these immunoblots with anti-PP2A antibodies demonstrated
the efficient immunoprecipitation of overexpressed and endogenous
PP2A/A (Fig. 5B, top right,
lanes 11, 13, and 15).
Accordingly, PP2A/A appears to be absent from complexes containing
cyclin G2 and vice versa.
Cyclin G2 Exists in a Complex with the C Subunit of PP2A--
To
test whether PP2A/C can interact with cyclin G1 or G2, HEK293 extracts
were incubated with immobilized cyclin G1-GST and G2-GST (Fig.
6A).
Endogenous PP2A/C specifically bound to immobilized cyclin G1-GST and
G2-GST fusion proteins but not control GST. The amount of PP2A/C that
associated with cyclin G1-GST and G2-GST was not significantly altered
by the presence or absence of PP2A/B' subunits (Fig. 6A,
compare left and right bottom
panels). After probing with PP2A/A antibodies, we could not
detect the A subunit in any of the pull-down pellets in contrast to
that in the applied lysates. This observation suggests that endogenous
PP2A/C binding to cyclin G1 or G2 cannot be improved by co-expressing
B' subunits and does not require A subunits. However, it is possible
that unknown endogenous B' subunits are present in saturating amounts and that additional B' subunits cannot increase the binding.
View larger version (109K):
[in this window]
[in a new window]
|
Fig. 6.
Cyclin G2 associates with endogenous PP2A/C
and colocalizes with PP2A/C in transfected
cells. A, cyclin G1-GST and G2-GST pulled down
endogenous PP2A/C but not PP2A/A. Lysates (500 µg of protein) were
incubated with immobilized GST, cyclin G2-GST, or G1-GST fusion
proteins for pull-down assays or were directly loaded (20 µg of
protein), as indicated. The bottom parts of each
blot were probed with mouse anti-PP2A/C followed by HRP-labeled goat
anti-mouse IgG, and top parts were probed with
rat anti-PP2A/A (clone 6G3) followed by HRP-conjugated rabbit anti-rat
IgG. B and C, cyclin G2 is associated with
endogenous PP2A/C. HEK293 cells expressing GFP, G2-GFP, and G2-V5His
with or without HA-tagged PP2A B'3 or - as indicated were lysed.
Lysates (1 mg of protein) were used for pull-down with
microcystin-agarose (Mcys) or control resin (Sepharose
4B-CL; Seph.) (B) or for immunoprecipitations
with sheep anti-PP2A/C (C) before SDS-PAGE. Lysates (50 µg
of protein; B, lanes 1, 5,
6, 9, 10, and 13) were also
directly applied. The top half of each blot was
probed with anti-cyclin G2 followed by HRP-conjugated protein A, and
the bottom was probed with mouse anti-PP2A/C followed by HRP-labeled
goat anti-mouse IgG. The arrows indicate the position of
cyclin G2 fusion proteins and PP2A/C. D-F,
immunofluorescence confocal microscopy colocalization of PP2A/C with
cyclin G2. CHO cells were co-transfected with plasmids
encoding HA-tagged PP2A/C (D-F, left
panels; red) plus cyclin G2-V5His (D
and E, middle left
subpanels; green), G2-GFP (D and
E, middle right subpanels;
green), or GFP (F; green). Cells were
fixed either before permeabilization with Triton X-100 (D
and upper right subpanel in
F) or after pre-extraction with Triton X-100 (E
and left and lower right
subpanel of F). Cultures were stained with
anti-HA primary followed by LRSC-conjugated secondary antibodies to
visualize HA-tagged PP2A/C and TOTO-3 to label DNA (pseudocolored
blue in channel merge shown at right). Cultures
co-expressing cyclin G2-V5His were labeled with anti-cyclin G2
antibodies followed by FITC-conjugated anti-rabbit IgG.
|
|
Microcystin is a toxin that specifically and tightly binds to the
catalytic site of both protein phosphatase 1 and PP2A. We used a
microcystin resin to precipitate endogenous PP2A/C from lysates of
cells singly or doubly transfected with expression vectors for cyclin
G2 and HA-tagged B' subunits. Precipitated protein complexes contained
endogenous PP2A/C and cyclin G2 (Fig. 6B). The fraction of
both proteins isolated on microcystin-Sepharose can roughly be
estimated by comparing immunosignals from 50 µg of protein lysate and
the microcystin precipitates from corresponding lysates containing 1 mg
of protein (Fig. 6B, compare lanes 1 to 3 and 4 to 5, etc.). The percentage
of cyclin G2 isolated with endogenous PP2A/C on microcystin-Sepharose
varied between different experiments from ~1% (Fig. 6B,
upper panels, compare lanes
1 and 3) to 5% (Fig. 6B, compare
lane 6 with lane 7 and
lane 10 with lane 11).
Microcystin-Sepharose precipitation isolated about 15-30% of the
total PP2A/C from the lysate regardless of the presence of cyclin G2,
HA-tagged B' subunits singly or in combination (Fig. 6B,
lower panels, compare lanes
1 and 3, etc.). Accordingly, 2-20% of cyclin G2
is associated with endogenous PP2A/C. This amount might be greater if a
substantial amount of cyclin G2 dissociates from the complex during
isolation or is associated with PP2A/C at detergent-resistant sites.
The formation of a complex that contains both endogenous PP2A/C and
cyclin G2 was further confirmed by coimmunoprecipitation of cyclin G2
with anti-PP2A/C antibodies (Fig. 6C). To further examine
whether PP2A/C interacts with cyclin G2 in intact mammalian cells, we
tested whether cyclin G2 colocalized with PP2A/C at distinct sites.
Detection of endogenous PP2A/C in either HEK293 or CHO cells by
immunofluorescence with any of the anti-PP2A/C-specific antibodies
available to us was unsatisfactory. Therefore, we tested whether
HA-tagged PP2A/C can associate with coexpressed cyclin G2-GFP or
G2-V5His in HEK293 cells. HA-PP2A/C coprecipitated with cyclin G2
immunocomplexes isolated from the corresponding cell lysates, and no
signal was detected in immunoprecipitates from control lysates (data
not shown), corroborating that HA-PP2A/C specifically coprecipitated
with cyclin G2. Complementary immunoprecipitation experiments of tagged
PP2A/C with anti-HA antibodies followed by reciproc
,
TOP
ABSTRACT
INTRODUCTION
