A retrovirus-based protein complementation assay
screen reveals functional AKT1-binding partners
Zhiyong Ding*, Jiyong Liang*, Yiling Lu*, Qinghua Yu*, Zhou Songyang†, Shiaw-Yih Lin*, and Gordon B. Mills*‡
*Department of Molecular Therapeutics, University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Box 950, Houston, TX 77030;
and†Department of Biochemistry and Molecular Biology, Baylor College of Medicine, One Baylor Plaza, Houston, TX 77030
Communicated by Louis Siminovitch, University of Toronto, Toronto, ON, Canada, August 17, 2006 (received for review April 24, 2006)
We developed a retrovirus-based protein-fragment complementa-
tion assay (RePCA) screen to identify protein–protein interactions
in mammalian cells. In RePCA, bait protein is fused to one fragment
of a rationally dissected fluorescent protein, such as GFP, intensely
fluorescent protein, or red fluorescent protein. The second, com-
plementary fragment of the fluorescent protein is fused to an
endogenous protein by in-frame exon traps in the enhanced
retroviral mutagen vector. An interaction between bait and host
protein (prey) places the two parts of the fluorescent molecule in
proximity, resulting in reconstitution of fluorescence. By using
RePCA, we identified a series of 24 potential interaction partners
or substrates of the serine?threonine protein kinase AKT1. We
confirm that ?-actinin 4 (ACTN4) interacts physically and function-
ally with AKT1. siRNA-mediated ACTN4 silencing down-regulates
AKT phosphorylation, blocks AKT translocation to the membrane,
increases p27Kip1levels, and inhibits cell proliferation. Thus, ACTN4
is a critical regulator of AKT1 localization and function.
?-actinin 4 ? AKT1 ? protein–protein interactions ? screen
tation (Co-IP) and mass spectrometry (2), or protein libraries (3).
These approaches, however, do not efficiently identify protein
location of the interaction (Y2H) or difficulties in the Co-IP of
cytoskeletal or membrane proteins. Furthermore, conventional
Y2H approaches yield false-positive signals with transcription fac-
tors precluding screening. The purpose of this study was to develop
a readily applicable, subcellular localization-, library-, and cell-
independent method for the identification of protein–protein in-
teractions, including cytoskeletal and membrane proteins, in mam-
malian cells and apply it to the well characterized AKT
The retrovirus-based protein-fragment complementation assay
(RePCA) utilizes the technical advantages of two emerging tech-
nologies, the enhanced retroviral mutagen (ERM) vector and the
protein-fragment complementation assay (PCA). The ERM vector
functions as an exon trap that efficiently activates and tags endog-
enous genes (4). The ERM vector in all three ORFs can be applied
to any mammalian cell without the effort and bias of cell-specific
libraries. Furthermore, the ERM vector utilizes the endogenous
splicing machinery of the target cell, allowing the evaluation of
as a monomeric enzyme or a fluorescent protein (GFP or a variant
thereof), is rationally dissected into two fragments that will not
reconstitute spontaneously (5–7). When each fragment of the
fluorescent protein is fused to one of a pair of interacting protein
partners, the subsequent protein binding places the fragments in
proximity, creating a functional complex that restores fluorescence
Michnick (9) identified an AKT1 partner with a PCA-based cDNA
library screen using GFP as a reporter. However, the need to
generate cell-specific cDNA libraries and a complex approach to
the identification of candidates renders this approach technically
demanding and labor-intensive. RePCA combines the power of
resently, novel protein–protein interactions are identified by
using yeast two-hybrid (Y2H) systems (1), coimmunoprecipi-
ERM (4) with PCA (8) to provide a facile, sensitive approach for
the identification of context-dependent protein–protein interac-
tions in mammalian cells, allowing native protein folding and
RePCA Screen Design. The intensely fluorescent protein (IFP), also
known as Venus (10), was dissected into two fragments, an IFP
N-terminal portion (IFPN) and IFP C-terminal portion (IFPC), at
residue 158 (11). As shown in Fig. 1A, IFPC was inserted into the
ERM vector, followed by a splice donor, to construct the RePCA
fusion protein. After infection of the target cells, the retrovirus
undergoes reverse transcription and integration into the host ge-
nome. If the integration occurs upstream or inside a host gene, the
IFPC linked in-frame to downstream host exons. The propensity of
the ERM retrovirus to integrate near the start of transcriptional
units increases the chance that a full-length or near-full-length
fusion protein will be created (12).
A host cell line expressing the tetracycline-regulated transac-
tivator tTA or reverse tTA to enable the regulated expression of
the prey from the Tet-responsive promoter is generated by
transfection to express the IFPN-Bait fusion protein (Fig. 1B).
Bait fusion protein-expressing cells are infected with the RePCA
retrovirus in all three frames to generate in-frame IFPC-endog-
enous fusion proteins. Cells in which the retrovirus does not
generate an in-frame fusion protein or wherein IFPC is not fused
to a binding partner of the bait do not fluoresce. Only cells
containing IFPC fused in-frame to an interaction partner of the
bait fluoresce. The fluorescent cells are cloned, and the target
genes are identified by RT-PCR with primers contained in the
ERM vector. The resulting fluorescent clones can be directly
used to characterize the formation, localization, and function-
ality of candidate interactions, providing powerful reagents for
mechanistic exploration without the need for recloning.
Proof-of-Concept Screen for AKT1 Interaction Partners. We chose
AKT1 (also known as PKB-?), which plays a central role in cell
metabolism, survival, growth, and tumorigenesis (13, 14), as a bait
for a proof-of-concept RePCA screen. IFPN-AKT1 HeLa Tet-on
cells with or without transient transfection of IFPC were not
fluorescent, confirming that the fragments do not fluoresce and do
not spontaneously associate (data not shown). Transfection of the
and Q.Y. performed research; Z.S. contributed new reagents?analytic tools; Z.D., J.L., Y.L.,
Q.Y., S.-Y.L., and G.B.M. analyzed data; and Z.D., J.L., Y.L., S.-Y.L., and G.B.M. wrote the
The authors declare no conflict of interest.
Abbreviations: PCA, protein-fragment complementation assay; RePCA, retrovirus-based
PCA; ERM, enhanced retroviral mutagen; Y2H: yeast two-hybrid; IFP, intensely fluorescent
protein; IFPN, N-terminal portion of IFP; IFPC, C-terminal portion of IFP; ACTN4, ?-actinin
4; PDK1, phosphoinositide-dependent kinase 1; PI3K, phosphatidylinositol 3 kinase; Co-IP,
‡To whom correspondence should be addressed. E-mail: email@example.com.
© 2006 by The National Academy of Sciences of the USA
October 10, 2006 ?
vol. 103 ?
no. 41 www.pnas.org?cgi?doi?10.1073?pnas.0606917103
known AKT partner, phosphoinositide-dependent kinase 1
fluorescence, which was enriched in the cell membrane and at
points of cell–cell contact (Fig. 2A). The membrane localization of
the PDK1-IFPC:IFPN-AKT1 complex was completely blocked by
inhibition of phosphatidylinositol 3 kinase (PI3K) with LY294002,
as expected from the known localization of AKT1 and PDK1 (16,
PDK1 reconstitutes IFP fluorescence with appropriate subcellular
IFPN-AKT1 HeLa Tet-on cells were separately infected with
RePCA-IFPC retroviral vectors in all three reading frames. After
selection with puromycin and induction with doxycycline, single
fluorescent cells were isolated by sorting. Expanded clones dis-
played fluorescence with different patterns, i.e., homogenous,
cytoplasmic, nuclear, or membrane, suggesting differential local-
ization of AKT1 and specific binding partners (see Fig. 2 B–D and
genes (10, 30, and 30 clones for reading frame 1, 2, and 3,
respectively), all of which yielded fusion transcripts. In terms of
efficiency, using frame 2 as an example, 2 ? 107cells were sorted,
with the 384 most highly fluorescent cells placed in single wells.
Doxycycline was withdrawn to reduce the expression of proteins
that could have adverse effects on cell survival and growth. Of the
showing fluorescence after doxycycline induction (?14%). We
chose 30 clones based on fluorescence intensity and patterns and
identified 11 independent candidate AKT1 partners (Table 1).
Thus, from 2 ? 107cells, 11 candidates were identified. This pilot
screen was not taken to saturation; therefore, it is likely that
additional AKT1-binding partners would be identified in large-
scale screens. Twenty-four independent candidates were identified
from all three ORFs (Table 1). Some candidates were present as
multiple clones [i.e., BCL2-antagonist of cell death (BAD) was
identified three times (Table 1)] and were likely derived from a
sequencing demonstrated identical fusion transcripts. Limiting cell
propagation before sorting fluorescent cells could increase the
to calculated sizes were detected in all clones (see Fig. 5, which is
published as supporting information on the PNAS web site, for
selected clones) except for AHNAK2, which has an expected
molecular weight of ?600 kDa, precluding detection by Western
blots. The identification of AHNAK2 demonstrates the power of
the ERM vector, which, through the use of endogenous splicing
machinery, is not restricted in terms of size of insert. In 14 of 24
clones, the IFPC fragment was fused to a full-length or near-full-
length protein (Table 1).
ners. (A) AKT1 interaction with PDK1 demon-
strated by PCA. IFPN-AKT1 HeLa Tet-on cells sta-
bly transfected to coexpress PDK1-IFPC yielded
highly fluorescent cells with enhanced fluores-
for 3 hr abrogated membrane localization. (B)
Selected RePCA clones with different fluorescent
patterns. Expanded clones were assessed for flu-
genes activated in each clone are shown under
each image. (C) SSB clone (Table 1, no. 20) show-
6) clone showing membrane fluorescence.
Pilot screen for AKT1-interaction part-
by inserting IFPC into the ERM retroviral vector, followed by a splice donor, in the U3 region of the 3? LTR. A Tet-responsive promoter controls the expression
of IFPC. RePCA retroviruses are generated in packaging cells. After the infection of target cells, integration upstream or inside a host gene may allow the
containing the Tet regulatory complex to enable the expression from a Tet-responsive promoter, is transfected to stably express IFPN-Bait. The cell population
is infected with RePCA viruses to create endogenous protein fusions with IFPC. Fusion of IFPC in-frame to a protein that binds the bait reconstitutes the IFP
molecule, restoring fluorescence. Fluorescent cells are cloned by cell sorting or other approaches and expanded, and target genes are identified by RT-PCR.
Schematic diagram of the RePCA screen. (A) Generation of IFPC fusion with endogenous proteins in mammalian cells. A RePCA vector is constructed
Ding et al.
October 10, 2006 ?
vol. 103 ?
no. 41 ?
Among the targets, BAD is a well characterized AKT substrate
and interaction partner (18). Myosin light chain (atrial?embryonic
alkali) (MYL4) is part of the myosin II complex, a known AKT-
binding partner (19). AHA1, Annexin1, and a Lamin A?C isoform
(LaminB1) have been confirmed as AKT-binding partners by
Co-IP and mass spectroscopy (J. Downward, personal communi-
cation). We also identified two potential AKT substrates because
heme oxygenase 2 (HMOX2) and C14ORF78 (AHNAK2) are
isoforms of HMOX1 and AHNAK1, which are known AKT
substrates (20, 21). Additional candidates have strong links to AKT
(Table 1). For example, CD98 regulates AKT phosphorylation and
activation (22). Both CD98 and AKT interact with integrin-?1 (22,
23). Moesin and AKT interact with the hamartin:tuberin complex
(24–26). The identification of known and likely AKT partners?
substrates validates the screening strategy.
?-Actinin 4 (ACTN4) Is a Functional Partner of AKT1. ?-Actinin
actin-binding proteins, including ACTN4, can bind the p85 regu-
latory subunit of PI3K and translocate out of the membrane after
inhibition of PI3K (27–29). In addition, ACTN4 contributes to the
prognosis of breast cancer (29) and has been implicated in tumor
development, invasion, and metastases (30, 31), making it a can-
doxycycline; therefore, formation of the fluorescent complex was
controlled by Tet-responsive expression of the IFPC fusion protein
(Fig. 3A). The fluorescent IFPN-AKT1::IFPC-ACTN4 complex in
clone IC1-05 was located throughout the cytoplasm in serum-
starved cells (Fig. 3B). Serum induced the translocation of the
AKT1::ACTN4 complex to the membrane, in particular cellular
ruffles, reminiscent of the PDK1::AKT1 complex (Fig. 2A), and to
the nuclear periphery (Fig. 3B). Compatible with the observation
that ACTN4 translocates out of the membrane upon inhibition of
PI3K (29), serum-dependent translocation of the AKT1::ACTN4
complex to the cell membrane was blocked by inhibition of PI3K,
indicating that membrane localization depended on the production
of 3-phosphorylated membrane phosphatidylinositols. Expression
of an exogenous IFPC-ACTN4wt in IFPN-AKT1 HeLa Tet-on
cells resulted in fluorescence, which was enriched at the leading
edge of cells growing in serum (Fig. 3C), confirming that the
apparent interaction between ACTN4 and AKT1 was not due to
clonal variation. Expression of a IFPC-ACTN4?310–665 deletion
construct resulted in homogeneous fluorescence in the cytoplasm
with enhanced fluorescence in the nucleus but completely failed to
Table 1. Interaction partners of AKT1 identified in RePCA screen
No. Protein name
of the fusion
1 BCL2-antagonist of cell
Myosin light chain
Heme oxygenase 2*
BAD2 63 233 HomogenousNP?116784
?12 341 Cytoplasm NP?001002841
5 Activator of heat shock
ATPase homolog 1*
Ribosomal protein L22*
AHA13 28452 NP?036243
Protease inhibitor 6‡
Proteolipid protein 2‡
Zinc finger protein 185‡
factor SUI1 homolog‡
Pyruvate kinase, muscle‡
Sui1iso12 11 223 HomogenousXP?497726
?93 806 CytoplasmNP?872271
Accession nos. and gene symbols are from GenBank.
*Known partners?substrates or potential partners?substrates with known links to AKT.
†RTN4 has five splice variants, A–E, with the same C-terminal sequence. The insertion site for RTN4 was defined according to the RTN4 variant C sequence.
‡Potential AKT partners?substrates without obvious known links to AKT.
www.pnas.org?cgi?doi?10.1073?pnas.0606917103 Ding et al.
mediate localization of the fluorescent complex to the leading edge
of cells growing in serum (Fig. 3C). In contrast, coexpression of
IFPN-AKT1 and IFPC-ACTN4?310–911 did not result in detect-
able fluorescence (Fig. 3C). Thus, residues 665–911 of ACTN4,
which contain two EF motifs, are required for the interaction with
AKT1, whereas residues 310–665 of ACTN4 are critical for the
localization of the AKT1::ACTN4 complex to the leading edge of
cells. Strikingly, the PH domain of AKT1 (32) was not sufficient to
translocate the AKT1::ACTN4 complex to the cell membrane,
suggesting that ACTN4 contributes to AKT1 localization. How-
ever, given that LY294002 blocked translocation (Fig. 3B), the
translocation of the AKT1::ACTN4 complex to the membrane
depends on the production of 3-phosphorylated membrane phos-
phatidylinositols. Therefore, the screening approach not only pro-
vides insight into the binding partners of AKT1, but the resultant
cells can also provide important functional information related to
the localization and formation of the protein complexes.
In clone IC1-05, IFPC-ACTN4 was readily immunoprecipitated
by anti-AKT1 antibodies and detected by immunoblotting with a
polyclonal anti-GFP antibody, which binds both halves of IFP (Fig.
3D). Equal amounts of IFPC-ACTN4 and IFPC-AKT1 were
present in the immunoprecipitated complex, compatible with the
be immunoprecipitated by anti-AKT1 antibodies after in vivo
cross-linking (Fig. 3E; see also Supporting Materials and Methods,
which is published as supporting information on the PNAS web
site), demonstrating an association of endogenous AKT1 with
ACTN4. We were unable to detect a stable association between
endogenous AKT1 and ACTN4 by Co-IP in the absence of
cross-linking (data not shown), potentially because the conditions
required to efficiently release ACTN4 from the cytoskeleton dis-
rupted interactions between AKT1 and ACTN4 or because the
interaction between AKT1 and ACTN4 was transient or indirect.
ACTN4 Silencing Inhibits AKT Translocation, Phosphorylation, Signal-
ing, and Cell Proliferation. ThemembranelocalizationofAKTplays
a pivotal role in AKT activation (13, 14). The distinctive pattern of
ACTN4::AKT1 complex localization suggested that ACTN4 could
contribute to AKT translocation and, thus, activation. In HeLa
Tet-on cells as well as in IOSE80- hTERT (hTERT, human
telomerase reverse transcriptase) cells not expressing exogenous
AKT or ACTN4 IFP fusion proteins (Fig. 4A), an siRNA pool
targeting ACTN4 induced a concordant reduction in ACTN4
protein expression and AKT phosphorylation. Strikingly, ACTN4
silencing increased levels of the cyclin-dependent kinase inhibitor
ACTN4 knockdown and pAKT phosphorylation being concordant
(Fig. 6, which is published as supporting information on the PNAS
web site). Insulin induced translocation of AKT1 fused to a
or ACTN4 fusion proteins) to the cell membrane, which was
blocked by ACTN4 siRNA knockdown, confirming a role for
ACTN4 in the translocation of AKT1 to the membrane (Fig. 4B).
The effects of ACTN4 knockdown on the phosphorylation of
HA-AKT1 were bypassed by membrane-targeted myristylated
AKT1 (Fig. 4C), indicating that ACTN4 knockdown prevents
translocation of AKT1 to the cell membrane. ACTN4 knockdown
did not markedly alter F-actin in HeLa cells, suggesting that the
effects of ACTN4 siRNA were not secondary to the disruption of
the cellular cytoskeleton (Fig. 4D). Finally, knockdown of ACTN4
significantly inhibited HeLa cell proliferation (Fig. 4E), with dif-
ferent siRNAs demonstrating similar activity on ACTN4 knock-
down, decrease in AKT phosphorylation (Fig. 6), and cell prolif-
RePCA to uncover functional protein–protein interactions.
rescence intensity of clone IC1-05 (ACTN4) is Tet-
M1 quadrant. Cells had a mean fluorescence level of
3.2. IC1-05 cells exhibited low-level fluorescence with-
out doxycycline induction, likely because of leaky ex-
pression from Tet-responsive promoters (8.6% of cells
in M1 quadrant). Cells had a mean fluorescence level
of 46.5. After incubation with 2 ?g?ml doxycycline for
48 hr, the mean fluorescence of IC1-05 cells increased
to 112.9 (62% of cells in the M1 quadrant). (B) Trans-
location of IFPN-AKT1::IFPC-ACTN4 complex upon se-
rum stimulation in clone IC1-05. Serum-starved cells
show predominantly cytoplasmic fluorescence. After
serum (10%) stimulation for 60 min, the fluorescent
complex translocated to the leading edge of cells and
the periphery of the nucleus in ?90% of cells. Inhibi-
abrogated serum-induced membrane localization. At
least 100 cells were examined from different fields for
each sample. (C) Confirmation of ACTN4::AKT1 inter-
action by PCA and identification of the residues in
ACTN4 required for interaction with AKT1 and local-
ization of the complex. Coexpression of IFPN-AKT1
with IFPC-ACTN4wt yielded fluorescence enriched at
the leading edge of the cell (ruffles). Coexpression of
IFPN-AKT1 with IFPC-ACTN4?310–665 yielded cyto-
sion of IFPN-AKT1 with IFPC-ACTN4?310–911 did not
spectrin repeats; EF, EF-hand, calcium-binding motif. (D) Association of IFPN-AKT1 with IFPC-ACTN4 in clone IC1-05. Clone IC1-05 cells expressing IFPN-AKT1 and
anti-AKT1; lane 3, total lysate. (E) Association of AKT1 with ACTN4 in parental HeLa Tet-on cells. HeLa Tet-on cells were lysed in RIPA buffer, with in vivo
cross-linking with BASED. Co-IP was performed with anti-AKT1 and Western blotting with anti-ACTN4. Lane 1, Co-IP with normal IgG; lane 2, Co-IP with
anti-AKT1; lane 3, blank; lane 4, total lysate.
AKT1 interaction with ACTN4. (A) The fluo-
Ding et al.
October 10, 2006 ?
vol. 103 ?
no. 41 ?
The identification of known and likely AKT1-binding partners and
substrates (Table 1) validated RePCA as being able to identify
protein–protein interactions and potentially transient interactions
between enzyme and substrate in mammalian cells. A previously
uncharacterized physical and functional interaction was identified
between ACTN4 and AKT1. Functional association was demon-
strated by siRNA-mediated ACTN4 silencing, which inhibits AKT
translocation and phosphorylation, resulting in up-regulation of
p27Kip1protein and a decrease in cell proliferation. An interaction
between ACTN4 and AKT may underlie the pathophysiology of
focal segmental glomerulosclerosis, which is caused by ACTN4
mutations (34), suggesting novel therapeutic approaches. ACTN4
also associates with focal adhesions, tight junctions, and adherens
junctions through interactions with the cytoskeleton and the tight-
junction protein MAGI-1 (35), linking the PI3K?AKT pathway to
motility and invasion.
RePCA has a number of potential advantages. (i) The ability
to perform screens in a homologous mammalian cell environ-
ment allows native protein folding and posttranslational modi-
fications. (ii) RePCA can be used to comparatively analyze
multiple different cell lines or genetic backgrounds, avoiding the
related bias and difficulties in generating cell-specific cDNA
libraries. The host range of the retrovirus can be extended by
pseudotyping with vesicular stomatitis virus G glycoprotein (36).
(iii) RePCA can identify context-dependent interactions under
different activation conditions or genetic manipulations or with
specific drugs. (iv) The derivation of the ERM vector from
Moloney murine leukemia virus (12) results in preferential
integration near the start of transcriptional units, generating a
high frequency of full-length or near-full-length fusion tran-
scripts. Furthermore, because the ERM vector uses native
splicing machinery, the approach can identify splicing-specific
interactions and the identification of binding partners is not
limited by size. (v) The Tet-responsive promoter allows high-
level expression of endogenous targets during screening fol-
lowed by repression of potentially toxic endogenous targets.
Regulated expression of the endogenous target by doxycycline
fluorescence combined with the high level of fluorescence when
the PCA fragments are brought into proximity by interacting
proteins makes the approach applicable to high-throughput
screening. (vii) The resulting clones provide reagents for studies
of the function and localization of protein complexes, suggesting
potential functional consequences of the interactions. These
clones can also be used in high-content drug, genomic, or
chemical genomic screens aimed at preventing the formation of
complexes or blocking the translocation of the complex to
particular subcellular compartments. (viii) The reconstituted
barrel structure of GFP and, by analogy, IFP is relatively stable
(37). Thus, RePCA has the potential to stabilize or trap transient
interactions, such as enzyme–substrate interactions, or to stabi-
lize low-affinity interactions, allowing the identification of com-
ponents of signaling pathways and networks not discoverable by
other approaches. The relative ease and applicability of RePCA
to high-throughput analysis allow sequential indentification of
signaling and cell proliferation. (A) The effects of
ACTN4 siRNA knockdown on AKT signaling. HeLa
Tet-on and human ovarian surface epithelial cells
IOSE80(hTERT) were transfected with ACTN4 siRNA or
a nontargeting siRNA (NT-siRNA) pool (Dharmacon).
Twenty-four hours after transfection, cells were se-
rum-starved for 24 hr and stimulated with 5% FBS for
by 8% SDS?PAGE for the detection of ACTN4 and 12%
for other proteins. ?-actin immunoblotting showed
equivalent loading and specificity of the siRNA. Scan-
independent experiments in HeLa Tet-on cells
(mean ? SE) were obtained with NIH IMAGE 1.63.1
software and are presented as densitometric values of
target siRNA divided by control (nontargeting siRNA).
location to HeLa cell membranes. HeLa cells were sta-
bly transfected to express AKT1-GFP. After serum star-
vation for 18 hr, insulin stimulation (20 ?g?ml) for 10
min induced translocation of AKT1-GFP to the cell
membrane in ?90% of AKT1-GFP HeLa cells in the
presence or absence of nontargeting siRNA. In con-
onstrated insulin-induced translocation of AKT1-GFP
from different fields for each sample. (C) Inhibition of
AKT1 phosphorylation by ACTN4 siRNA is bypassed by
myristylated AKT1. HeLa cells were transfected with a
nontargeting siRNA or ACTN4 siRNA pool. After 24 hr,
the cells were transfected to express Myr-HA-AKT1 or
HA-AKT1. Forty-eight hours after siRNA transfection,
10% FBS and 75 ng?ml IGF-1 for 10 min before lysis.
pAKT1 (473) levels are presented as values relative to
nontargeting siRNA. (D) ACTN4 knockdown does not
alter the actin cytoskeleton. HeLa cells were trans-
effects of ACTN4 knockdown on cell proliferation. Proliferation of HeLa cells transfected with siRNA was assessed by 3,(4,5-dimethylthiazol-2-yl)2,5-
diphenyltetrazolium bromide assay kit (Sigma) 96 hr after transfection.
siRNA-mediated ACTN4 silencing alters AKT
www.pnas.org?cgi?doi?10.1073?pnas.0606917103 Ding et al.
interacting partners and the creation of pathways and networks Download full-text
based on protein–protein interactions. (ix) RePCA can identify
interactions in multiple different intracellular compartments,
particularly cytoskeletal, membrane, and transcription-factor
spectrometry approach. (x) Molecular interactions are detected
directly, not through secondary events, such as transcription
The RePCA approach has a number of potential limitations.
(i) Intron-less genes cannot be captured by the ERM vector.
Fortunately, these genes are relatively uncommon. (ii) The
screen depends on a high-titer virus preparation for targeting a
sufficient number of genes. (iii) Genes not accessible for viral
integration will not be targeted. (iv) The fluorescent protein
fragment may interfere with the formation of protein–protein
interactions. (v) The fluorescent protein fragment may interfere
with membrane insertion of type I integral membrane proteins
or secreted proteins. (vi) The ability to trap enzyme substrate or
weak interactions as well as other sources of binding may result
in false-positive results. However, the use of both N- and
C-terminal orientations for the bait as well as varying the length
of linkers between the engineered GFP fragments and the
bait?prey is likely to alleviate many of these concerns.
Materials and Methods
Cell Lines and Plasmids. HeLa Tet-on cells were from BD Clontech
(Palo Alto, CA). 293T?17 was from American Type Culture
Collection (Manassas, VA). TAg and human telomerase immor-
talized normal ovarian epithelial cells IOSE80(hTERT) (38) were
from N. Auersperg (University of British Columbia, Vancouver,
BC, Canada). Plasmids VYF102 (IFPC vector), 11117-Y101 (ex-
pressing IFPN-AKT1), and 21622-Y108 (expressing PDK1-IFPC)
were from Odyssey Thera, Inc. (San Ramon, CA) (11). Plasmid
PS1941 encoding AKT1-GFP was from Bioimage (Soeborg, Den-
mark). Plasmids pcGP and pVSVG were from Xiao-Feng Qin
(M. D. Anderson Cancer Center). ERM vectors are described in
ref. 4. RePCA vectors for each of the three reading frames were
details, see Supporting Materials and Methods). Full-length ACTN4
cDNA was from Origene Technologies, Inc. (Rockville, MD). The
human AKT1 gene was cloned from OVCAR3 cells by RT-PCR.
Construction of plasmids expressing IFPC-ACTN4wt, IFPC-
ACTN4?310–665, IFPC-ACTN4?310–911, and Myr-HA-AKT1
are described in Supporting Materials and Methods.
RePCA Screen Procedure. HeLa Tet-on cells were stably transfected
with plasmid 11117-Y101 to express IFPN-AKT1. IFPN-AKT1
HeLa Tet-on cells were infected in exponential growth phase.
39, with additional details in Supporting Materials and Methods.
Infected cells were selected with 0.5 ?g?ml puromycin (BD Clon-
tech) for 5 days. In the last 2 days of selection, 2 ?g?ml of
doxycycline (BD Clontech) was added to induce the expression of
IFPC fusions from Tet-responsive promoters. Fluorescent cells
were sorted individually into 96-well plates. Doxycycline was with-
drawn during recovery to reduce the expression of proteins that
could have adverse effects on cell survival and growth (for addi-
tional details, see Supporting Materials and Methods).
Identification of Target Genes. Total RNA was extracted from
expanded clones by using RNeasy Mini kits (Qiagen, Valencia,
CA). Reverse transcription was performed with a random primer
(GC)ACG-3?; n ? AGCT) (4), or a PolyT primer RT-1T (5?-
TTTTTTT-3?) using a SuperScript III kit (Invitrogen, Carlsbad,
CA). The 5? end of the RT-1 primer contains the T7 primer
sequence. The cDNA was PCR-amplified with a specific IFPC
primer (IFPCR, 5?-ACTTCAAGATCCGCCACAACATCGAG-
3?) and the T7 primer (T7-2, 5?-GCAAATACGACTCACTAT-
AGGGATC-3?) by using AccuTaq DNA polymerase (Invitrogen).
Gel-purified PCR products were directly sequenced, with the
resulting sequences used to search GenBank human nonredundant
and expressed-sequence-tag databases by using BLAST.
RNAi and Cell Growth Assay.ApooloffoursiRNAstargetinghuman
ACTN4 and a nontargeting siRNA pool were from Dharmacon
from Ambion (Austin, TX). RNAi silencing was performed ac-
cording to the manufacturer’s protocol. The effects of siRNA
2-yl)2,5-diphenyltetrazolium bromide assay (40).
1. Fields S, Song O (1989) Nature 340:245–246.
2. Lin D, Tabb DL, Yates, JR, III (2003) Biochim Biophys Acta 1646:1–10.
3. Turk BE, Cantley LC (2003) Curr Opin Chem Biol 7:84–90.
4. Liu D, Yang X, Yang D, Songyang Z (2000) Oncogene 19:5964–5972.
5. Johnsson N, Varshavsky A (1994) Proc Natl Acad Sci USA 91:10340–10344.
6. Ghosh I, Hamilton AD, Regan L (2000) J Am Chem Soc 122:5658–5659.
7. Hu CD, Chinenov Y, Kerppola TK (2002) Mol Cell 9:789–798.
8. Michnick SW, Remy I, Campbell-Valois FX, Vallee-Belisle A, Pelletier JN (2000)
Methods Enzymol 328:208–230.
9. Remy I, Michnick SW (2004) Mol Cell Biol 24:1493–1504.
10. Nagai T, Ibata K, Park ES, Kubota M, Mikoshiba K, Miyawaki A (2002) Nat Biotechnol
Drug Dev Technol 1:811–822.
12. Wu X, Li Y, Crise B, Burgess SM (2003) Science 300:1749–1751.
13. Cantley LC (2002) Science 296:1655–1657.
14. Brazil DP, Yang ZZ, Hemmings BA (2004) Trends Biochem Sci 29:233–242.
15. Alessi DR, James SR, Downes CP, Holmes AB, Gaffney PR, Reese CB, Cohen P
(1997) Curr Biol 7:261–269.
16. Andjelkovic M, Alessi DR, Meier R, Fernandez A, Lamb NJ, Frech M, Cron P,
Cohen P, Lucocq JM, Hemmings BA (1997) J Biol Chem 272:31515–31524.
17. Anderson KE, Coadwell J, Stephens LR, Hawkins PT (1998) Curr Biol 8:684–691.
18. Datta SR, Dudek H, Tao X, Masters S, Fu H, Gotoh Y, Greenberg ME (1997) Cell
19. Tanaka M, Konishi H, Touhara K, Sakane F, Hirata M, Ono Y, Kikkawa U (1999)
Biochem Biophys Res Commun 255:169–174.
20. Salinas M, Wang J, Rosa de Sagarra M, Martin D, Rojo AI, Martin-Perez J, Ortiz
de Montellano PR, Cuadrado A (2004) FEBS Lett 578:90–94.
21. Sussman J, Stokoe D, Ossina N, Shtivelman E (2001) J Cell Biol 154:1019–1030.
22. Feral CC, Nishiya N, Fenczik CA, Stuhlmann H, Slepak M, Ginsberg MH (2005) Proc
Natl Acad Sci USA 102:355–360.
23. Wu D, Thakore CU, Wescott GG, McCubrey JA, Terrian DM (2004) Oncogene 23:8659–
24. van Slegtenhorst M, Nellist M, Nagelkerken B, Cheadle J, Snell R, van den Ouweland
A, Reuser A, Sampson J, Halley D, van der Sluijs P (1998) Hum Mol Genet
25. Lamb RF, Roy C, Diefenbach TJ, Vinters HV, Johnson MW, Jay DG, Hall A (2000)
Nat Cell Biol 2:281–287.
26. Dan HC, Sun M, Yang L, Feldman RI, Sui XM, Ou CC, Nellist M, Yeung RS, Halley
DJ, Nicosia SV, Pledger WJ, Cheng JQ (2002) J Biol Chem 277:35364–35370.
27. Fukami K, Furuhashi K, Inagaki M, Endo T, Hatano S, Takenawa T (1992) Nature
28. Shibasaki F, Fukami K, Fukui Y, Takenawa T (1994) Biochem J 302:551–557.
29. Honda K, Yamada T, Endo R, Ino Y, Gotoh M, Tsuda H, Yamada Y, Chiba H,
Hirohashi S (1998) J Cell Biol 140:1383–1393.
30. Menez J, Le Maux Chansac B, Dorothee G, Vergnon I, Jalil A, Carlier MF, Chouaib
S, Mami-Chouaib F (2004) Oncogene 23:2630–2639.
31. Honda K, Yamada T, Hayashida Y, Idogawa M, Sato S, Hasegawa F, Ino Y, Ono M,
Hirohashi S (2005) Gastroenterology 128:51–62.
Gaffney PR, Reese CB, McCormick F, Tempst P, et al. (1998) Science 279:710–714.
33. Liang J, Slingerland JM (2003) Cell Cycle 2:339–345.
34. Kaplan JM, Kim SH, North KN, Rennke H, Correia LA, Tong HQ, Mathis BJ,
Rodriguez-Perez JC, Allen PG, Beggs AH, Pollak MR (2000) Nat Genet 24:251–256.
35. Patrie KM, Drescher AJ, Welihinda A, Mundel P, Margolis B (2002) J Biol Chem
36. Burns JC, Friedmann T, Driever W, Burrascano M, Yee JK (1993) Proc Natl Acad
Sci USA 90:8033–8037.
37. Magliery TJ, Wilson CG, Pan W, Mishler D, Ghosh I, Hamilton AD, Regan L (2005)
J Am Chem Soc 127:146–157.
38. Auersperg N, Maines-Bandiera SL, Dyck HG, Kruk PA (1994) Lab Invest 71:510–
39. Soneoka Y, Cannon PM, Ramsdale EE, Griffiths JC, Romano G, Kingsman SM,
Kingsman AJ (1995) Nucleic Acids Res 23:628–633.
40. Hansen MB, Nielsen SE, Berg K (1989) J Immunol Methods 119:203–210.
Ding et al.
October 10, 2006 ?
vol. 103 ?
no. 41 ?