Live-cell imaging RNAi screen identifies PP2A-B55alpha and importin-beta1 as key mitotic exit regulators in human cells.

Abstract

When vertebrate cells exit mitosis various cellular structures are re-organized to build functional interphase cells. This depends on Cdk1 (cyclin dependent kinase 1) inactivation and subsequent dephosphorylation of its substrates. Members of the protein phosphatase 1 and 2A (PP1 and PP2A) families can dephosphorylate Cdk1 substrates in biochemical extracts during mitotic exit, but how this relates to postmitotic reassembly of interphase structures in intact cells is not known. Here, we use a live-cell imaging assay and RNAi knockdown to screen a genome-wide library of protein phosphatases for mitotic exit functions in human cells. We identify a trimeric PP2A-B55alpha complex as a key factor in mitotic spindle breakdown and postmitotic reassembly of the nuclear envelope, Golgi apparatus and decondensed chromatin. Using a chemically induced mitotic exit assay, we find that PP2A-B55alpha functions downstream of Cdk1 inactivation. PP2A-B55alpha isolated from mitotic cells had reduced phosphatase activity towards the Cdk1 substrate, histone H1, and was hyper-phosphorylated on all subunits. Mitotic PP2A complexes co-purified with the nuclear transport factor importin-beta1, and RNAi depletion of importin-beta1 delayed mitotic exit synergistically with PP2A-B55alpha. This demonstrates that PP2A-B55alpha and importin-beta1 cooperate in the regulation of postmitotic assembly mechanisms in human cells.

Full text

Page 1

LETTERS
Live-cell imaging RNAi screen identifies PP2A–B55α and
importin-β1 as key mitotic exit regulators in human cells
Michael H. A. Schmitz1,2,3, Michael Held1,2, Veerle Janssens4, James R. A. Hutchins5, Otto Hudecz6, Elitsa Ivanova4,
Jozef Goris4, Laura Trinkle-Mulcahy7, Angus I. Lamond8, Ina Poser9, Anthony A. Hyman9, Karl Mechtler5,6,
Jan-Michael Peters5 and Daniel W. Gerlich1,2,10
When vertebrate cells exit mitosis various cellular structures
are re-organized to build functional interphase cells1. This
depends on Cdk1 (cyclin dependent kinase 1) inactivation
and subsequent dephosphorylation of its substrates2–4.
Members of the protein phosphatase 1 and 2A (PP1 and
PP2A) families can dephosphorylate Cdk1 substrates in
biochemical extracts during mitotic exit5,6, but how this relates
to postmitotic reassembly of interphase structures in intact
cells is not known. Here, we use a live-cell imaging assay and
RNAi knockdown to screen a genome-wide library of protein
phosphatases for mitotic exit functions in human cells. We
identify a trimeric PP2A–B55α complex as a key factor in
mitotic spindle breakdown and postmitotic reassembly of
the nuclear envelope, Golgi apparatus and decondensed
chromatin. Using a chemically induced mitotic exit assay,
we find that PP2A–B55α functions downstream of Cdk1
inactivation. PP2A–B55α isolated from mitotic cells had
reduced phosphatase activity towards the Cdk1 substrate,
histone H1, and was hyper-phosphorylated on all subunits.
Mitotic PP2A complexes co-purified with the nuclear transport
factor importin-β1, and RNAi depletion of importin-β1 delayed
mitotic exit synergistically with PP2A–B55α. This demonstrates
that PP2A–B55α and importin-β1 cooperate in the regulation
of postmitotic assembly mechanisms in human cells.
In the budding yeast, Saccharomyces cervisiae, Cdk1 substrate dephos-
phorylation and mitotic exit depend on the Cdc14 phosphatase7, but this
function does not seem to be conserved in Cdc14 homologues of other
species2,3,8–11. Studies in cycling Xenopus laevis embryonic extracts sug-
gest that phosphatases of both the PP1 (ref. 5) and PP2A (ref. 6) families
can contribute to Cdk1 substrate dephosphorylation during vertebrate
mitotic exit, whereas Ca2+-triggered mitotic exit in cytostatic-factor-
arrested egg extracts depends on calcineurin12,13. Early genetic studies in
Drosophila melanogaster14,15 and Aspergillus nidulans16 reported defects
in late mitosis of PP1 and PP2A mutants. However, the assays used in
these studies were not specific for mitotic exit because they scored pro-
metaphase arrest or anaphase chromosome bridges, which can result
from defects in early mitosis.
Intracellular targeting of Ser/Thr phosphatase complexes to specific
substrates is mediated by a diverse range of regulatory and targeting
subunits that associate with a small group of catalytic subunits3,4,17. It is
possible that mitotic exit in intact cells requires phosphatases that have
not been detected by previous assays using in vitro extracts. In practice,
the short duration of mitotic exit makes it difficult to assay this process,
which explains why previous RNAi screening of cell division regulators18
did not annotate mitotic exit phenotypes.
To assay mitotic exit in live human cells, we measured the timing from
anaphase onset until nuclear reformation. We generated a HeLa cell line
stably expressing a chromatin marker (histone 2B fused to a red fluores-
cent protein; H2B–mCherry19) to visualize the metaphase–anaphase tran-
sition (Fig. 1a–c). To probe for postmitotic nuclear reassembly, we stably
co-expressed a nuclear import substrate (importin-β-binding domain of
importin-α fused to monomeric enhanced green fluorescent protein; IBB–
eGFP20), which is cytoplasmic during mitosis and co-localizes with chroma-
tin regions after reassembly of a functional nuclear envelope (Fig. 1a, b).
To annotate mitotic exit timing automatically in time-lapse micro-
scopy movies, we used computational methods developed in-house
(CellCognition21). Individual cells were detected and tracked over time,
and the mitotic stage of each cell was assigned based on classification of
chromatin morphology (Fig. 1c). Nuclear breakdown and reassembly was
1Institute of Biochemistry, Swiss Federal Institute of Technology Zurich (ETHZ), Schafmattstrasse 18, CH‑8093 Zurich, Switzerland. 2Marine Biological Laboratory,
Woods Hole, MA 02543, USA. 3Current address: Systems and Cell Biology of Neurodegeneration, Division of Psychiatry Research, University of Zurich, August Forel‑
Strasse 1, CH‑8008 Zurich, Switzerland. 4Laboratory of Protein Phosphorylation and Proteomics, Department of Molecular Cell Biology, Faculty of Medicine, KU
Leuven, Gasthuisberg O&N1, Herestraat 49 Box 901, B‑3000 Leuven, Belgium. 5Institute of Molecular Pathology, Dr. Bohr‑Gasse 7, 1030 Vienna, Austria. 6Institute
of Molecular Biotechnology of the Austrian Academy of Sciences, Dr. Bohr‑Gasse 3, 1030 Vienna, Austria. 7Department of Cellular & Molecular Medicine and the
Ottawa Institute of Systems Biology, University of Ottawa, 451 Smyth Road, Ottawa, ON K1H 8M5, Canada. 8Wellcome Trust Centre for Gene Regulation & Expression,
MSI/WTB/JBC Complex, University of Dundee, Dundee, DD1 5EH, UK. 9Max‑Planck‑Institute for Molecular Cell Biology and Genetics, Pfotenhauerstrasse 108,
D‑01307 Dresden, Germany.
10Correspondence should be addressed to D.W.G. (daniel.gerlich@bc.biol.ethz.ch)
Received 27 May 2010; accepted 02 July 2010; published online 15 August 2010; DOI: 10.1038/ncb2092
886
nature cell biology VOLUME 12 | NUMBER 9 | SEPTEMBER 2010
© 20 Macmillan Publishers Limited. All rights reserved10

Page 2

LETTERS
determined by recording changes in the ratio of mean IBB–eGFP fluo-
rescence in chromatin regions versus surrounding cytoplasmic regions
(Fig. 1d). Automated annotation of mitotic exit timing (4.70 ± 0.89 min;
mean ± s.d.) closely matched manual annotation (4.88 ± 0.84 min;
mean ± s.d.; n = 270 cells, Supplementary Information, Fig. S1).
Cells were transfected with 675 different siRNAs targeting a genome-
wide set of 225 annotated human protein phosphatases, including cata-
lytic and associated regulatory and scaffolding subunits (three different
siRNAs per gene, two experimental replicates; for a full list of siRNAs see
Supplementary Information, Table 1). These cells were then imaged over
approximately 24 h with time-intervals of approximately 3.8 min. On
average, this yielded approximately 87 automatically annotated mitotic
events per movie (total: 113,236 mitotic events). The mean mitotic exit
timing was determined for all data sets that contained more than 10
mitotic events (n = 1,278 from 1,350 movies). Five siRNAs reproducibly
scored as ‘hits’ above a significance cut-off at three standard deviations
above the mean of all data points (z-score = 3; Fig. 2a). These five siRNAs
targeted three distinct genes, encoding each of the three subunits of a het-
erotrimeric PP2A complex: PPP2CA (catalytic subunit α, subsequently
labelled as CA), PPP2R1A (scaffold subunit α, subsequently labelled as
R1A), and PPP2R2A (a regulatory B-type subunit, also termed PR55α
or B55α and subsequently labelled as B55α). The remaining four siRNAs
targeting the same PP2A genes also prolonged mitotic exit, albeit below
the cut-off level (Fig. 2a).
0
2
4
6
8
10
a
b
c
d
Time
H2B IBB

H2B
IBB ratio (a.u.)
IBB contour
ProphasePrometaphaseMetaphaseEarly anaphaseLate anaphaseTelophase
0 min1.6
0.0 min
3.35 min
10.05 min
13.40 min
6.70 min
6.44.89.617.628.7–36.7 –33.5–1.6–43.1
–20 –30–40 –1001020 30
Time (min)
IBB H2B
Figure 1 Live‑cell imaging assay of mitotic exit timing. (a) Automated time‑
lapse microscopy imaging of a HeLa cell line stably expressing a chromatin
marker (H2B–mCherry; red) and a nuclear import substrate (IBB–eGFP;
green). The selected images show approximately 13% of a movie field‑
of‑view. For full movie, see Supplementary Information, Movie S1. (b)
Time‑lapse microscopy of a single cell progressing through mitosis; onset
of anaphase is marked as 0 min. (c) Automated detection of chromatin
regions, tracking of cells over time, annotation of mitotic stages and
classification of chromatin morphologies by supervised machine learning.
For full movie, see Supplementary Information, Movie S2. (d) Automated
detection of nuclear breakdown and reassembly. Chromatin regions (red
outline) were defined by automated segmentation of the chromatin‑
associated H2B–mCherry fluorescence (as shown in c). Cytoplasmic regions
(green outline indicates outer boundary, red outline is inner boundary)
were derived by dilation of the chromatin regions. The ratio of mean
IBB–eGFP fluorescence in chromatin versus cytoplasmic regions served
to automatically determine nuclear envelope breakdown (orange bar) and
reformation after anaphase (nuclear import of IBB, blue bar). Anaphase
onset was defined by the classification of chromatin morphology (violet bar,
see c). a.u., arbitrary units. Scale bars, 10 μm.
nature cell biology VOLUME 12 | NUMBER 9 | SEPTEMBER 2010
887
© 20 Macmillan Publishers Limited. All rights reserved10

Page 3

LETTERS
To validate the hits, we tested an extended set of six siRNAs per gene,
which all significantly prolonged mitotic exit (P < 0.001; Supplementary
Information, Fig. S2a–c). The mRNA depletion levels correlated with phe-
notypic penetrance, indicating specificity of the phenotype (Supplementary
a
6420 –2
–2
0
2
4
6
z-score - replicate 1
z-score - replicate 2
Other data points
Control siRNA
CA siRNA
R1A siRNA
B55α siRNA
b
c
Time (min)
0306090
NEBD–anaphase
75
50
25
0
100
Percentage of cells
75
50
25
0
100
Time (min)
0 105 1520
Anaphase–nuclear import
Percentage of cells
B55α siRNA
Control siRNA
CA siRNA
R1A siRNA
B55α siRNA
Control siRNA
CA siRNA
R1A siRNA
def
IBB H2B
IBB
PP2A-B55α siRNA
3.26.49.612.816.022.425.619.20 min–1.6
Control siRNA
B55α (m/h) siRNA
B55α (h) siRNA
Actin
B55α (h) depletion
B55α (h)
B55α−LAP (m)
Percentage of cells
Time (min)
Anaphase–nuclear import
036912
75
50
25
0
100
R1A siRNA
Control siRNA
B55α (h) siRNA
B55α (m/h) siRNA
g
0
2
4
6
–2
–4
–6
z-score
siRNA oligos
Other data
Control siRNA
CA siRNA
R1A siRNA
B55α siRNA
70% 61%
Figure 2 RNAi screen for mitotic exit regulators. (a) RNAi screen of a
genome‑wide library of annotated protein phosphatases using the mitotic
exit assay shown in Figure 1. Individual data points correspond to the
z‑score based on the mean mitotic exit timing in individual movies,
determined in two experimental replicates. Negative controls are black.
siRNA oligos that resulted in phenotypes reproducibly scoring a z‑score
threshold of > 3 (dashed lines) were considered as hits (highlighted by
colours as indicated in legend; siRNAs targeting the same genes as the
hits but with z‑scores below threshold are also highlighted). (b) RNAi
depletion of B55α in a HeLa cell line stably expressing LAP‑tagged
mouse‑B55α from a bacterial artificial chromosome (BAC; for localization
see Supplementary Information, Fig. S6b). Cells were transfected with
siRNA targeting either human B55α alone (B55α h), or both mouse‑ and
human‑B55α (B55α m/h). A quantitative western blot is shown, probed
with an anti‑B55α antibody that recognizes both mouse‑ and human‑
B55α. (c) Rescue of mitotic exit delay phenotype by exogenous B55α.
HeLa cells expressing mouse‑B55α–LAP were RNAi‑depleted for human
(h) or mouse/human (m/h) B55α (as validated in b), or for the scaffolding
subunit of PP2A (R1A), and the timing of mitotic exit was then assessed.
(d) Cumulative histograms for early mitotic progression. Nuclear envelope
breakdown until anaphase onset was timed in control and RNAi‑depleted
HeLa cells stably expressing H2B–mCherry and IBB–eGFP (as in Figure 1;
n ≥ 64, under all conditions). (e) Cumulative histograms of mitotic exit.
Anaphase onset until nuclear import of IBB–eGFP was measured for the
same cells shown in d. (f) Synthetic depletion RNAi screen for mitotic
exit phosphatases. Assay, sample preparation and siRNA library was
identical to the screen shown in a, except that siRNA targeting B55α was
co‑transfected in each experimental condition. Negative controls (black)
were transfected by only non‑targeting siRNA. The plot shows the ranked
z‑scores of a single replicate, calculated as in a. Dashed line indicates
z‑score threshold. (g) Time‑lapse microscopy images of a cell depleted of
all three PP2A–B55α subunits progressing from anaphase through mitotic
exit (as indicated by red and green bars above the images). Full movie
shown in Supplementary Information, Movie S4. For negative control,
see Supplementary Information, Movie S3. Scale bar, 10 μm. Uncropped
image of blot is shown in Supplementary Information, Fig. S9a.
888
nature cell biology VOLUME 12 | NUMBER 9 | SEPTEMBER 2010
© 20 Macmillan Publishers Limited. All rights reserved10

Page 4

LETTERS
Information, Fig. S2a–f). RNAi depletion was also efficient at the protein
level (Supplementary Information, Fig. S2g). Next, we depleted endog-
enous B55α in a HeLa cell line stably expressing eGFP-tagged mouse B55α.
Transfection of siRNA targeting a non-conserved sequence on the human
B55α mRNA efficiently depleted endogenous B55α, but not the exogenous
mouse B55α (Fig. 2b), and these cells showed normal mitotic exit timing
(Fig. 2c). In contrast, transfection of siRNA targeting both human and
mouse B55α mRNA, or R1A, efficiently delayed mitotic exit (Fig. 2c). This
provides validation that on-target depletion of B55α mRNA causes the
observed mitotic exit phenotype.
The PP2A protein phosphatase family is involved in many cellular
processes, including earlier mitotic stages3. It is generally accepted that
the regulatory B-type subunit confers substrate specificity and thereby
regulates diverse functions of PP2A3,17. Depletion of CA or R1A sig-
nificantly prolonged mitotic progression from nuclear envelope break-
down until anaphase (P < 0.001; Fig. 2d; same cells shown in Fig. 2e),
as expected from the known functions of PP2A complexes in spindle
assembly and chromosome cohesion (which involve other B-type subu-
nits3). In contrast, early mitotic progression was unaffected in B55α-
depleted cells. These data indicate that PP2A function is required at
all stages of mitosis, whereas the B55α subunit is rate-limiting only for
post-anaphase progression.
Depletion of PP2A–B55α subunits delayed, but did not arrest, mitotic
exit (Fig. 2e). This may be explained by incomplete RNAi depletion or
the involvement of other unknown factors with redundant function. To
investigate if additional phosphatases become limiting in the absence of
B55α, we screened the phosphatase-targeting siRNA library in a back-
ground of synthetic B55α RNAi depletion. Increased mitotic exit delay
occurred upon co-depletion of R1A or CA subunits with B55α (2–3 oligos
scoring above a z-score threshold of 3; Fig. 2f). However, none of the other
222 phosphatases showed a consistent additive increase in mitotic exit
delay (five other siRNA oligos delayed slightly above the z-score thresh-
old, but could not be confirmed by different siRNAs targeting the same
genes). Co-transfection of siRNAs targeting all three PP2A–B55α subu-
nits (labelled as PP2A–B55α siRNA throughout the manuscript) caused
the most pronounced prolongation of mitotic exit (14.76 ± 6.50 min,
n = 205 versus 4.86 ± 1.07 min in the control, n = 251; mean ± s.d.;
Fig. 2g and Supplementary Information, Fig. S2g). The additive effect
of combinatorial PP2A–B55α subunit depletions suggests that residual
levels of this phosphatase may account for a slow and gradual mitotic exit.
The fact that no other RNAi conditions further delayed mitotic exit in
combination with B55α depletion underlines the particular importance
of PP2A–B55α for mitotic exit.
To address if PP2A–B55α controls postmitotic reassembly of cellular
structures other than the nucleus, we generated a HeLa cell line stably
expressing a fluorescent Golgi marker (galactosyl transferase, GalT–
eGFP22) and H2B–mCherry (Fig. 3a). Time-lapse confocal microscopy
showed that depletion of the PP2A–B55α complex significantly delayed
postmitotic clustering of Golgi fragments (P < 0.001; 25.5 ± 5.86 min,
n = 31 versus 11.17 ± 2.73 min in the control, n = 30; mean ± s.d.; Fig. 3a,
b, e and see Supplementary Information, Fig. S3b, c for single depletion
of B55α). In another mitotic exit assay, we imaged a HeLa cell line stably
expressing H2B–mCherry and eGFP–α-tubulin19 (Fig. 3c). PP2A–B55α-
depleted cells showed significantly delayed disassembly of spindle-pole-
associated microtubules (P < 0.001; Fig. 3c, d, f). Postmitotic chromosome
decondensation was also significantly delayed in PP2A–B55α-depleted
cells (P < 0.001; 18.29 ± 2.29 min, n = 98 versus 8.94 ± 5.89 min in
the control, n = 158; mean ± s.d.; see also Supplementary Information
Fig. S3d). These data show that PP2A–B55α contributes to postmitotic
reassembly of various interphase cell structures.
We next investigated how depletion of PP2A–B55α affects pro-
gression through interphase, using a monoclonal cell line expressing
Control siRNA
Control siRNA
PP2A–B55α siRNA
PP2A–B55α siRNA
a
b
c
d
ef
α-Tub H2B
α-Tub
α-Tub H2B
α-Tub
-4
–40 min 4913 1822263035
–40 min49131722263034
GalT
GalT H2B
GalT H2B
GalT
–20 min 49 13172226 3035
–30 min 581419222730 35
75
50
25
0
100
Time (min)
0 201030
Percentage of cellsPercentage of cells
75
50
25
0
100
Time (min)
020 1030
Anaphase–
Golgi clustering
Anaphase–
spindle disassembly
Control siRNA
PP2A-B55α siRNA
Control siRNA
PP2A-B55α siRNA
Figure 3 PP2A–B55α controls postmitotic Golgi assembly, spindle
breakdown and chromatin decondensation. (a) Images from a confocal
microscopy time‑lapse movie of a control cell expressing H2B–mCherry
and the Golgi marker, GalT–eGFP (for full movie, see Supplementary
Information, Movie S5). Golgi reassembly was scored based on clustering
of the fluorescence into two distinct patches per cell (t = 9 min). (b)
Golgi reassembly in a PP2A–B55α‑depleted cell (for full movie, see
Supplementary Information, Movie S6). (c) Confocal microscopy time‑lapse
images of mitotic spindle disassembly and chromosome decondensation in
a control cell (for full movie, see Supplementary Information, Movie S7).
Spindle disassembly was scored based on the first apparent detachment of
spindle‑pole‑associated microtubules from chromatin masses (t = 9 min).
(d) Mitotic spindle disassembly in a PP2A–B55α‑depleted cell (for full
movie, see Supplementary Information, Movie S8). (e, f) Cumulative
histograms of postmitotic Golgi clustering (e), or spindle disassembly
(f) relative to anaphase onset (t = 0 min). Scale bars, 10 μm.
nature cell biology VOLUME 12 | NUMBER 9 | SEPTEMBER 2010
889
© 20 Macmillan Publishers Limited. All rights reserved10

Page 5

LETTERS
H2B–mCherry and a DNA replication marker (proliferating cell nuclear
antigen) tagged with enhanced GFP (eGFP–PCNA23; Supplementary
Information, Fig. S4a–d). PP2A–B55α-depleted cells had a prolonged
G1 phase (10.49 ± 2.70 h, n = 31 versus 5.77 ± 1.11 h in the control,
n = 39; mean ± s.d.; Supplementary Information, Fig. S4b), as expected
after perturbation of mitotic exit. Conversely, PP2A–B55α-depleted cells
had a significantly shorter G2 phase (P < 0.001; 2.91 ± 0.39 h, n = 42
versus 3.95 ± 0.78 h in the control, n = 37; mean ± s.d.; Supplementary
Information, Fig. S4d). This is consistent with previous observations in
Xenopus embryonic extracts6,24.
The mitotic exit phenotypes observed after PP2A–B55α depletion could
be caused by a failure in Cdk1 inactivation, misregulated Cdk1 substrate
dephosphorylation or both. To discriminate between these possibilities,
we established a mitotic exit assay using chemical inactivation of Cdk1.
Cells were first arrested in metaphase by the proteasome inhibitor MG132
and then forced to exit from mitosis by adding the Cdk inhibitor flavopiri-
dol (still in the presence of MG132) to a final concentration of 20 μM
(ref. 25; Fig. 4a). This treatment promoted changes that are indicative of
mitotic exit, including nuclear reassembly (> 95% of all metaphase cells,
n = 99), Golgi clustering (> 90%, n = 103), chromosome decondensa-
tion and re-attachment to the substratum (Fig. 4b, d). Only chromosome
segregation did not occur, probably owing to the suppression of securin
degradation by MG132. Higher concentrations of flavopiridol did not
further accelerate nuclear reassembly (9.4 ± 1.1 min at 160 μM, n = 30
compared with 9.3 ± 1.3 min at 20 μM, n = 30; mean ± s.d.), indicating
that Cdk1 inhibition was complete. PP2A–B55α depletion significantly
delayed nuclear reassembly in this assay (P < 0.001; 20.67 ± 3.30 min, n
= 41 versus 9.86 ± 0.95 min in the control, n = 29; mean ± s.d.; Fig. 4b,
c, f and see Supplementary Information, Fig. S5a–c for single depletion
of B55α). Golgi reformation was also delayed (36.01 ± 9.38 min, n = 35
versus 16.91 ± 2.27 min in the control, n = 42; mean ± s.d.; Fig. 4d, e, g).
We conclude that a main function of PP2A–B55α in promoting mitotic
exit must be downstream of Cdk1 inactivation.
To test if PP2A–B55α depletion affected dephosphorylation of Cdk1
substrates, extracts were prepared from HeLa cells synchronized to dif-
ferent times after chemical induction of mitotic exit. Cdk1 substrate
phosphorylation, detected on quantitative western blots by an anti-
phosphorylated-Ser antibody that specifically recognizes the Cdk target
sequence K/R-pS-P-X-K/R (where X is any residue and pS is phosphor-
ylated Ser), dropped to approximately 23% within 18 min of flavopiridol
addition in control cells, whereas it remained at approximately 66% in
PP2A–B55α-depleted cells (Fig. 4h and see Supplementary Information,
b
c
d
e
h
fg
Time (min)
Time (min)
Nuclear reassemblyGolgi clustering
PP2A–B55α siRNA
Flavopiridol
Control siRNA
PP2A–B55α siRNA Control siRNA
IBB
–11 min 691114161719
20–1 2 min 691215 1618212324
–21 min 6 111621 2530 3540 4550
–13 min 8131722273237424651
0153045
0 1020
75
50
25
0
100
75
50
25
0
100
IBB H2B
IBB H2B
GalT H2B
GalT H2B
IBB

GalT

GalT

232025
Control siRNA
PP2A-B55α siRNA
ControlsiRNA
PP2A-B55α siRNA
Control siRNA
PP2A-B55α siRNA
1456723811 12 13 14 9 10
Flavopiridol (min)
Anti-phospho(Ser)
Cdk substrates
09 12 15 183609 12 15 1836
a
52 h
Start imaging
1 h
Measure mitotic exit timing
Transfect
siRNAs
MG132Flavopiridol
Percentage of cellsPercentage of cells
Anti-cyclin B1
Anti-nucleolin
Figure 4 PP2A–B55α functions downstream of Cdk1 inactivation. (a)
Experimental protocol for observation of mitotic exit induced by chemical
inactivation of Cdks in absence of proteasome‑mediated degradation.
MG132 is the proteasome inhibitor and flavopiridol is the Cdk inhibitor.
(b, c) Time‑lapse microscopy images of cells expressing H2B–mCherry and
IBB–eGFP. A control cell is shown in b (for full movie, see Supplementary
Information, Movie S9) and a cell transfected with siRNA targeting
PP2A–B55α is shown in c (for full movie, see Supplementary Information,
Movie S10). Dashed red line indicates addition of flavopiridol, green bar
indicates onset of IBB–eGFP nuclear import. (d, e) Golgi reassembly after
chemically induced mitotic exit in cells expressing H2B–mCherry and the
Golgi marker, GalT–eGFP. A control cell is shown in d and a cell transfected
with siRNA targeting PP2A–B55α is shown in e. Dashed red line indicates
addition of flavopiridol, green bar indicates onset of Golgi clustering. (f, g)
Cumulative histograms of nuclear reassembly timing (f) and Golgi reassembly
timing (g) based on the data shown in b–e. (h) Detection of Cdk1 substrate
phosphorylation by an anti‑phosphorylated‑Ser antibody that specifically
recognizes the Cdk target sequence K/R‑pS‑P‑X‑K/R (where X is any residue
and pS is phosphorylated Ser) on a western blot. Samples were prepared
after chemical induction of mitotic exit in synchronized cells in presence of
proteasome inhibitor as in a. In control cells, Cdk substrates dephosphorylate
rapidly (Control siRNA, lanes 1–7). Cells depleted for PP2A–B55α show
delayed dephosphorylation (lanes 8–14). Scale bars, 10 μm. Uncropped
image of blot is shown in Supplementary Information, Fig. S9b.
890
nature cell biology VOLUME 12 | NUMBER 9 | SEPTEMBER 2010
© 20 Macmillan Publishers Limited. All rights reserved10

Page 6

LETTERS
c
f
200
116
97
66
45
31
d
e
Anaphase–Golgi clustering
Time (min)
010203040
Percentage of Cells
75
50
25
0
100
Importin-β1 + R1A siRNA
Importin-β1 + B55α siRNA
Control siRNA
Importin-β1 siRNA
-R1A
-Importin-β1
-Importin-β1
IM
IM
IMIM
R1AB55α
*
**
*
1
2
CA
B55α
R1A
IP: R1A–LAP
Input
Mr(K)
130 -
95 -
72 -
55 -
S212
GISPRG
0.1% (6.3-fold)
Catalytic
pocket
S167
EASPRR
55%
(7.4-fold)
S554
DNSTLQ
0.4%
(12.2-fold)
S8
DDSLYP
T304
RRTPD
0.1%
(8.6-fold)
B55α
CA
R1A
CA unstructured
C-terminus
g
80
60
20
40
0
100
PPase activity
Substrate: H1
MIIM
**
80
60
20
40
0
100
PPase activity
Substrate:
phosphorylase a
ab
IMIMIM
B55α
WT
B55α
S167A
B55α
S167E
Anti-GST
(B55α)
Anti-R1A
Anti-CA
Mr(K)
95-
72-
55-
43-
34-
Cdk1
Cdk1 CycB
R1A
P
B55α
P
CA
R1A
CA
Substr.
Substr.
Substr.
Substr.
Substr.
Substr.
P
P
P
P
P
P
InterphaseMitosis
CycB
Importin β1
h
B55α
Figure 5 Cell‑cycle‑dependent regulation of PP2A–B55α. (a, b) PP2A–B55α,
isolated from nocodazol‑treated mitotic (M) or unsynchronized interphase
cells (I) by pulldown of GST–B55α, was assayed for (a) phosphatase activity
towards Cdk1–cyclin B‑phosphorylated histone H1 (n = 15, values indicate
means ± s.d., asterisks denote P < 0.001; values normalized to interphase
cells) and (b) activity towards its substrate, phosphorylase a (n = 3; values
normalized to interphase cells. (c) Purification of PP2A complexes from
interphase (I) or mitotic (M) HeLa cells stably expressing LAP‑tagged R1A
or B55α baits, resolved by SDS–PAGE and silver staining. Two mitosis‑
specific bands were identified by mass spectrometry: importin‑β1 (1), and
importin‑α1 (2). Expected positions of the bands from endogenous PP2A
subunits are indicated on the right and migration positions of mouse baits are
marked with asterisks. (d) PP2A complexes were purified with R1A–LAP by
immunoprecipitation and resolved on a western blot by probing with anti‑R1A
and anti‑importin‑β1 antibodies. (e) Importin‑β1 function in mitotic exit.
Cells expressing H2B–mCherry and GalT–eGFP were transfected with siRNAs
as indicated. Timing from anaphase (t = 0 min) until Golgi reassembly was
assayed as in Fig. 3 (n ≥ 30 for each condition). (f) Phosphorylation sites on
PP2A–B55α were identified by mass spectrometry and are highlighted in red
on the 3D structure of PP2A–B55α34 and in the associated primary sequences.
The abundance of phosphorylated peptide in the mitotic sample was estimated
by peak area quantification of the elution profiles and is indicated as
percentage of total peptide based on elution profile peak area normalization.
Mitotic phosphorylation increase (indicated in brackets) was estimated
by comparing the normalized peak area quantifications of phosphorylated
peptides in interphase with mitotic samples. (g) Phosphorylation of B55α
Ser 167 affects PP2A complex assembly. GST‑tagged wild‑type‑B55α or GST‑
tagged substitution mutants of B55α (a non‑phosphorylatable S167A mutant
or a phospho‑mimicking S167E mutant) were isolated from unsynchronized
(I) or mitotic (M) cells by GST‑pulldown. PP2A subunits were detected on
western blots by anti‑GST, anti‑R1A and anti‑CA antibodies. (h) Model for
mitotic exit control. Dephosphorylation of a broad range of mitotic Cdk1
substrates promotes reassembly of interphase cells during mitotic exit. A
balance of kinase (Cdk1–cyclin B) and phosphatase (PP2A–B55α) activities
determines the substrate dephosphorylation kinetics during mitotic exit.
Green indicates activated state, red indicates lower activity and P indicates
phosphorylation. Uncropped images of blot are shown in Supplementary
Information, Fig. S9c.
nature cell biology VOLUME 12 | NUMBER 9 | SEPTEMBER 2010
891
© 20 Macmillan Publishers Limited. All rights reserved10

Page 7

LETTERS
Fig. S5e for single depletion of B55α). This supports the conclusion that
PP2A–B55α functions downstream of Cdk inactivation during mitotic
exit and shows that PP2A–B55α is required for timely Cdk1 substrate
dephosphorylation.
Previous studies in Xenopus embryonic extracts5,13 and HeLa cells26
have indicated that there is reduced phosphatase activity towards
Cdk1 targets during early mitosis. Human PP2A–B55α isolated
from mitotic cells had significantly reduced activity towards Cdk1-
phosphorylated histone H1 (P < 0.001; 71 ± 15% versus 100 ± 11% in
interphase; mean ± s.d., n = 15; Fig. 5a). However, we did not detect any
cell cycle-dependent differences in the phosphatase activity towards
another well-characterized model substrate, phosphorylase a (Fig. 5b).
This suggests cell cycle-regulated changes of PP2A–B55α substrate
specificity.
All three PP2A–B55α subunits were expressed at similar levels
in interphase and mitosis (Supplementary Information, Fig. S6a),
indicating that mitotic PP2A–B55α is unlikely to be regulated at the
protein level. We addressed potential changes in PP2A complex com-
position at different cell cycle stages by purification of LAP (localiza-
tion and affinity purification)-tagged R1A or B55α, stably expressed
from endogenous promoters27. R1A co-purified with several mitosis-
specific binding partners (Fig. 5c), two of which were identified by
mass spectrometry as the nuclear transport factors, importin-α1 and
importin-β1. In addition to a function in nuclear transport during
interphase, importin-β1 is part of a mitotic regulatory system involv-
ing the small GTPase Ran, which is known to control mitotic spindle
and nuclear envelope assembly28. After validation of the mitosis-spe-
cific interaction between importin-β1 and R1A by western blotting
(Fig. 5d), we investigated if importin-β1 contributes to mitotic exit pro-
gression. Because the specificity of the nuclear reassembly assay may
be compromised after depletion of a nuclear import factor, we assayed
mitotic exit by monitoring Golgi reformation. Importin-β1 depletion
(Supplementary Information, Fig. S2h, i) significantly delayed post-
mitotic Golgi reassembly (P ≤ 0.001; 14.7 ± 5.0 min, n = 30 versus
10.1 ± 1.4 min in the control, n = 30; mean ± s.d.; Fig. 5e), which
was further increased after co-transfection of siRNAs targeting R1A
or B55α (25.9 ± 6.3 min, n = 30 and 22.5 ± 8.1 min, n = 25, respec-
tively). Importin-β1 depletion also prolonged earlier mitotic stages
(Supplementary Information, Fig. S7), as expected from its known
function in spindle assembly. These data demonstrate that PP2A–B55α
and importin-β1 jointly promote mitotic exit.
By mass spectrometry, we detected five phosphorylation sites on the
PP2A complex purified from mitotic cells (Fig. 5f). The relative quan-
tities of phosphorylated peptides were estimated by a semi-quantitative
approach, using the extracted ion chromatogram for peak area quan-
tification of the peptide elution profiles (Supplementary Information,
Fig. S8). Phosphorylation of B55α at Ser 167 was estimated to be highly
abundant at 55% (mean; n = 2 independent experiments), whereas the
other sites were phosphorylated on < 1% of the eluted peptides (quan-
tifications provided in Fig. 5f; the phosphorylation levels of S8 on R1A
could not be determined because no unmodified peptides containing this
site were detected). PP2A is known to auto-dephosphorylate29, therefore
the absolute phosphorylation levels in cells may be higher. All four of the
quantified phosphorylations were enriched more than fivefold on PP2A
purified from mitotic cells (Fig. 5f; Supplementary Information, Fig. S8).
To test if PP2A complex assembly is regulated by the phosphorylation of
B55α at Ser 167, we isolated PP2A complexes from interphase and mitotic
HeLa cells by expression of a GST (glutathione S-transferase)-tagged B55α
phospho-mimicking S167E mutant. Indeed, the S167E mutant bound less
efficiently to the CA and R1A subunits, compared with wild-type B55α
or a non-phosphorylatable S167A mutant (Fig. 5g and Supplementary
Information, Fig. S9h).
This study provides a comprehensive screen for mitotic exit phos-
phatases in human cells. B55α has been previously shown to increase
PP2A activity towards Cdk1 phosphorylation sites30, consistent with the
possibility that PP2A–B55α promotes mitotic exit by direct dephospho-
rylation of Cdk1 substrates. The in vitro phosphatase activity of mitotic
PP2A–B55α towards Cdk1-phyosphorylated H1 was downregulated.
However, it is not known which of the many putative PP2A substrates
may be affected by this regulation, and therefore we cannot precisely
estimate the extent to which PP2A regulation shapes the kinetics of Cdk1
substrate dephosphorylation.
Our data raise interesting possibilities for PP2A regulatory mecha-
nisms (Fig. 5h). A B55α phospho-mimicking S167E mutation impaired
the binding of R1A and CA subunits, consistent with the possibility
that phosphorylation contributes to a cell cycle-dependent regulation
of PP2A–B55α complex assembly. This may also involve mitotic hyper-
phosphorylation of the CA subunit at Thr 304, as a previous study
showed that a phospho-mimicking T304D mutation also suppresses
assembly of B55α into PP2A complexes31. Importin-β1 may regulate
PP2A by direct binding, by a nuclear-cytoplasmic targeting mechanism
or as a molecular chaperone28. In this context, it is interesting to note
that importin-β1 is structurally related to the R1A subunit and mem-
bers of the Bʹ family of regulatory PP2A subunits32. Even though it is
possible that importin-β1 functions in a mitotic exit pathway independ-
ently of PP2A, the physical and functional interaction observed here
suggests a link between the importin-/Ran and Cdk1-phosphorylation
regulatory systems.
In contrast to the Xenopus embryonic extract system6, depletion
of B55δ in HeLa cells (down to 20% mRNA level; Supplementary
Information, Fig. S2j) did not delay mitotic exit. This may reflect differ-
ent relative expression levels of the B55-subfamily isoforms in the two
systems or technical limitations of the depletion methods. A previous
study proposed that PP1 dominates as a Cdk1-counteracting phos-
phatase in cycling Xenopus embryonic extracts5. The fact that we did
not detect mitotic exit delays after RNAi depletion of any of the PP1
catalytic or regulatory subunits may be related to the cellular context of
our phenotypic assays instead of homogenized extracts or to differences
between embryonic and somatic mitosis. However, potential functional
redundancy between different PP1 catalytic isoforms that were targeted
only individually in our screen may have masked phenotypes; therefore
we cannot rule out the possibility that PP1 also contributes to mitotic
exit in human somatic cells.
In conclusion, our study reveals PP2A–B55α and importin-β1 as
key regulators of cellular reassembly mechanisms during mitotic exit.
Mitotic exit has been recently recognized as a target for improved, next-
generation cancer therapeutics33. B55α and importin-β1 are therefore
good targets for the development of mitotic exit-specific inhibitors.
METHODS
Methods and any associated references are available in the online version
of the paper at http://www.nature.com/naturecellbiology/
892
nature cell biology VOLUME 12 | NUMBER 9 | SEPTEMBER 2010
© 20 Macmillan Publishers Limited. All rights reserved 10

Page 8

LETTERS
Note: Supplementary Information is available on the Nature Cell Biology website
AcKnOWLEDGMEnTS
The authors thank F. Uhlmann and B. Novak for critical comments on the
manuscript. We thank S. Maar, the ETHZ Light Microscopy Centre (LMC), the
ETHZ RNAi Screening Centre (RISC), M. Augsburg (The Max Planck Institute
of Molecular Cell Biology and Genetics; MPI-CBG), M. Leuschner (MPI-CBG)
and A. Ssykor (MPI-CBG) for technical assistance. We thank U. Kutay (ETHZ) for
anti-importin-β1 and anti-nucleolin antibodies, J. Rohrer (University of Zurich)
for providing GalT–eGFP plasmid, J. Ellenberg (EMBL, Heidelberg) for IBB–eGFP
plasmid, M.C. Cardoso (Technical University, Darmstadt) for eGFP–PCNA plasmid
and Sanofi Aventis and the National Cancer Institute for providing flavopiridol.
This work was supported by Swiss National Science Foundation (SNF) research
grant 3100A0-114120, SNF ProDoc grant PDFMP3_124904, a European Young
Investigator (EURYI) award of the European Science Foundation to D.W.G., a MBL
Summer Research Fellowship by the Evelyn and Melvin Spiegel Fund to D.W.G., a
Roche Ph.D. fellowship to M.H.A.S. and a Mueller fellowship of the Molecular Life
Sciences Ph.D. programme Zurich to M.H. M.H. and M.H.A.S. are fellows of the
Zurich Ph.D. programme in Molecular Life Sciences. V.J. and J.G. were supported
by grants of the ‘Geconcerteerde OnderzoeksActies’ of the Flemish government,
the ‘Interuniversitary Attraction Poles’ of the Belgian Science Policy P6/28 and the
‘Fonds voor Wetenschappelijk Onderzoek-Vlaanderen’. A.I.L. is a Wellcome Trust
Principal Research Fellow. A.A.H. acknowledges funding by the Max Planck Society,
MitoCheck (the EU-FP6 integrated project), and a BMBF (Bundesministerium
für Bildung and Forschung) grant, DiGtoP (01GS0859). Work in the groups of
K.M. and J.M.P. was supported by MitoCheck (the EU-FP6 integrated project),
Boehringer Ingelheim, the GEN-AU programme of the Austrian Federal Ministry
of Science and Research (Austrian Proteomics Platform III), MeioSys within the
Seventh Framework Programme of the European Commission and by Chromosome
Dynamics, which is funded by the Austrian Science Foundation (FWF).
AuTHOR cOnTRIbuTIOnS
M.H.A.S. performed all experiments, except the mass spectrometry and in vitro
phosphatase assays, and wrote part of the paper. M.H. implemented software
for automated imaging and data analysis. V.J., E.I. and J.G. performed in vitro
phosphatase assays and B55α phospho-mutant analysis. J.H. and J.M.P. designed
and performed PP2A purification. K.M. and O.H. performed mass spectrometry.
L.T.M. and A.I.L. compiled the phosphatase screening library. I.P. and A.A.H.
generated the cell lines stably expressing LAP-tagged PP2A subunits. D.W.G.
conceived the project, designed the screening strategy and wrote the paper.
cOMPETInG FInAncIAL InTERESTS
The authors declare no competing financial interests.
Published online at http://www.nature.com/naturecellbiology
Reprints and permissions information is available online at http://npg.nature.com/
reprintsandpermissions/
1. Morgan, D. O. The cell cycle: principles of control. (Oxford University Press, 2007).
2. Queralt, E. & Uhlmann, F. Cdk‑counteracting phosphatases unlock mitotic exit. Curr.
Opin. Cell Biol. 20, 661–668 (2008).
3. Bollen, M., Gerlich, D. W. & Lesage, B. Mitotic phosphatases: from entry guards to exit
guides. Trends Cell Biol. 19, 531–541 (2009).
4. Trinkle‑Mulcahy, L. & Lamond, A. I. Mitotic phosphatases: no longer silent partners.
Curr. Opin. Cell Biol. 18, 623–631 (2006).
5. Wu, J. Q. et al. PP1‑mediated dephosphorylation of phosphoproteins at mitotic exit is con‑
trolled by inhibitor‑1 and PP1 phosphorylation. Nat. Cell Biol. 11, 644–651 (2009).
6. Mochida, S., Ikeo, S., Gannon, J. & Hunt, T. Regulated activity of PP2A–B55δ is crucial
for controlling entry into and exit from mitosis in Xenopus egg extracts. EMBO J. 28,
2777–2785 (2009).
7. Stegmeier, F. & Amon, A. Closing mitosis: the functions of the Cdc14 phosphatase and
its regulation. Annu. Rev. Genet. 38, 203–232 (2004).
8. Berdougo, E., Nachury, M. V., Jackson, P. K. & Jallepalli, P. V. The nucleolar phos‑
phatase Cdc14B is dispensable for chromosome segregation and mitotic exit in human
cells. Cell Cycle 7, 1184–1190 (2008).
9. Gruneberg, U., Glotzer, M., Gartner, A. & Nigg, E. A. The CeCDC‑14 phosphatase
is required for cytokinesis in the Caenorhabditis elegans embryo. J. Cell Biol. 158,
901–914 (2002).
10. Mailand, N. et al. Deregulated human Cdc14A phosphatase disrupts centrosome sepa‑
ration and chromosome segregation. Nat. Cell. Biol. 4, 317–322 (2002).
11. Mocciaro, A. et al. Vertebrate cells genetically deficient for Cdc14A or Cdc14B retain
DNA damage checkpoint proficiency but are impaired in DNA repair. J. Cell Biol. 189,
631–639 (2010).
12. Nishiyama, T., Yoshizaki, N., Kishimoto, T. & Ohsumi, K. Transient activation of cal‑
cineurin is essential to initiate embryonic development in Xenopus laevis. Nature 449,
341–345 (2007).
13. Mochida, S. & Hunt, T. Calcineurin is required to release Xenopus egg extracts from
meiotic M phase. Nature 449, 336–340 (2007).
14. Mayer‑Jaekel, R. E. et al. The 55 kd regulatory subunit of Drosophila protein phos‑
phatase 2A is required for anaphase. Cell 72, 621–633 (1993).
15. Axton, J. M., Dombradi, V., Cohen, P. T. & Glover, D. M. One of the protein phosphatase
1 isoenzymes in Drosophila is essential for mitosis. Cell 63, 33–46 (1990).
16. Doonan, J. H. & Morris, N. R. The bimG gene of Aspergillus nidulans, required for com‑
pletion of anaphase, encodes a homolog of mammalian phosphoprotein phosphatase
1. Cell 57, 987–996 (1989).
17. Janssens, V., Longin, S. & Goris, J. PP2A holoenzyme assembly: in cauda venenum
(the sting is in the tail). Trends Biochem. Sci. 33, 113–121 (2008).
18. Neumann, B. et al. Phenotypic profiling of the human genome by time‑lapse microscopy
reveals cell division genes. Nature 464, 721–727 (2010).
19. Steigemann, P. et al. Aurora B‑mediated abscission checkpoint protects against tetra‑
ploidization. Cell 136, 473–484 (2009).
20. Dultz, E. et al. Systematic kinetic analysis of mitotic dis‑ and reassembly of the nuclear
pore in living cells. J. Cell Biol. 180, 857–865 (2008).
21. Held, M. et al. CellCognition: time‑resolved phenotype annotation in high‑throughput
live cell imaging. Nat. Methods doi: 10.1038/nmeth.1486 (in the press).
22. Schaub, B. E., Berger, B., Berger, E. G. & Rohrer, J. Transition of galactosyltransferase
1 from trans‑Golgi cisterna to the trans‑Golgi network is signal mediated. Mol. Biol.
Cell 17, 5153–5162 (2006).
23. Leonhardt, H. et al. Dynamics of DNA replication factories in living cells. J. Cell Biol.
149, 271–280 (2000).
24. Lee, T. H., Turck, C. & Kirschner, M. W. Inhibition of cdc2 activation by INH/PP2A.
Mol. Biol. Cell 5, 323–338 (1994).
25. Potapova, T. A. et al. The reversibility of mitotic exit in vertebrate cells. Nature 440,
954–958 (2006).
26. Skoufias, D. A., Indorato, R. L., Lacroix, F., Panopoulos, A. & Margolis, R. L. Mitosis
persists in the absence of Cdk1 activity when proteolysis or protein phosphatase activity
is suppressed. J. Cell Biol. 179, 671–685 (2007).
27. Poser, I. et al. BAC TransgeneOmics: a high‑throughput method for exploration of
protein function in mammals. Nat. Methods 5, 409–415 (2008).
28. Harel, A. & Forbes, D. J. Importin‑β: conducting a much larger cellular symphony. Mol.
Cell 16, 319–330 (2004).
29. Guo, H. & Damuni, Z. Autophosphorylation‑activated protein kinase phosphorylates
and inactivates protein phosphatase 2A. Proc. Natl Acad. Sci. USA 90, 2500–2504
(1993).
30. Mayer‑Jaekel, R. E. et al. Drosophila mutants in the 55 kDa regulatory subunit of
protein phosphatase 2A show strongly reduced ability to dephosphorylate substrates
of p34cdc2. J. Cell Sci. 107, 2609–2616 (1994).
31. Longin, S. et al. Selection of protein phosphatase 2A regulatory subunits is mediated by
the C terminus of the catalytic Subunit. J. Biol. Chem. 282, 26971–26980 (2007).
32. Shi, Y. Serine/threonine phosphatases: mechanism through structure. Cell 139, 468–
484 (2009).
33. Huang, H. C., Shi, J., Orth, J. D. & Mitchison, T. J. Evidence that mitotic exit is a better
cancer therapeutic target than spindle assembly. Cancer Cell 16, 347–358 (2009).
34. Xu, Y., Chen, Y., Zhang, P., Jeffrey, P. D. & Shi, Y. Structure of a protein phosphatase
2A holoenzyme: insights into B55‑mediated Tau dephosphorylation. Mol. Cell 31,
873–885 (2008).
nature cell biology VOLUME 12 | NUMBER 9 | SEPTEMBER 2010
893
© 20 Macmillan Publishers Limited. All rights reserved 10

Page 9

METHODS
DOI: 10.1038/ncb2092
METHODS
Cell lines and plasmids. The HeLa Kyoto cell line was obtained from S. Narumiya
(Kyoto University, Japan) and cultured in Dulbecco’s modified eagle medium
(DMEM; GIBCO) supplemented with 10% (v/v) foetal bovine serum (PAA
Laboratories) and 1% (v/v) penicillin–streptomycin (Invitrogen). All live-cell
imaging experiments were performed using monoclonal reporter cell lines that
were generated as previously described35. For a complete list of plasmids and cell
lines see Supplementary Information, Tables S3 and S4. Cells were grown either
in 96-well microtitre plates (Greiner) or on LabTek chambered coverslips (Nunc)
for live-cell microscopy. Live-cell imaging was performed in DMEM containing
10% (v/v) foetal calf serum and 1% (v/v) penicillin–streptomycin, but without
phenol red and riboflavin to reduce autofluorescence of the medium35.
The bacterial artificial chromosomes (BACs), RP24-103C16, harbouring
mouse PP2A–B55α (PPP2R2A), and RP24-255O20, harbouring mouse PPP2CA,
were obtained from the BACPAC Resources Center (http://bacpac.chori.org). The
LAP (eGFP–IRES (internal ribosome entry site)–neomycin) cassette was PCR
amplified using primers that contain 50 nucleotides homologous to the carboxy
terminus of each of the target genes. Recombineering and stable transfection of
the modified BAC was performed as previously described27.
Live-cell imaging. Automated microscopy with reflection-based laser autofocus
was performed on a Molecular Devices ImageXpress Micro screening micro-
scope equipped with a ×10, 0.5 N.A. S Fluor dry objective (Nikon), controlled by
Metamorph macros developed in-house21. Cells were maintained in a microscope
stage incubator at 37 °C in a humidified atmosphere of 5% CO2 throughout the
experiment. Illumination was adjusted so that the cell death rate was below 5%
in untreated control cells. Confocal microscopy was performed on a customized
Zeiss LSM 510 Axiovert microscope using a ×20, 0.8 N.A. Plan-Apochromat dry
objective, a ×40, 1.3 N.A. oil DIC EC Plan-Neofluar objective, or a ×63, 1.4 N.A.
oil Plan-Apochromat objective (Zeiss). The microscope was equipped with piezo
focus drives (piezosystem jena), custom-designed filters (Chroma) and an EMBL
incubation chamber (European Molecular Biology Laboratory), which provided
a humidified atmosphere at 37 °C with 5% CO2.
For imaging chemically induced mitotic exit, cells were seeded in chambered
LabTek coverslips overnight and then transfected with the indicated siRNAs.
After transfection (52 h), the medium was replaced with imaging medium con-
taining 30 μM MG132 (proteasome inhibitor; Sigma). The chambered coverslips
were placed into the microscope stage incubator, which was maintained at 37 °C
in a humidified atmosphere of 5% CO2, for 45 min, when imaging locations with
metaphase-arrested cells were selected. These cells were imaged for 10–15 min,
before mitotic exit was induced with 20 μM flavopiridol (Cdk inhibitor)25.
Image analysis and statistical analysis. Automated image analysis was per-
formed by CellCognition software, which was developed in-house21 (http://www.
cellcognition.org). Cell nuclei and mitotic chromosome masses were detected
by local adaptive thresholding. Cytoplasmic regions were derived by region-
growing the chromatin segmentation to a fixed size. Next, texture and shape
features were calculated for each object and samples for mitotic classes were
manually annotated for supervised classification. Support vector machine clas-
sification was performed by radial-based kernel and probability estimates. Cells
were tracked over time using a constrained nearest-neighbour approach, with an
algorithm that supported trajectory splitting and merging. Mitotic events were
detected in the graph structure on the basis of the transition from prophase to
prometaphase. Nuclear envelope reassembly was defined as an increase in the
ratio of the mean nuclear versus cytoplasmic IBB–eGFP fluorescence > 1.5-fold
above the ratio at the time of chromosome segregation. In the RNAi screen,
mean mitotic exit timing was normalized per 96-well plate to compensate for
slight differences in the temporal sampling rate. z-scores were calculated based
on the mean and standard deviation of all data points. All statistical testing was
by a two-tailed Student’s t-test.
RNAi. The human siRNA phosphatase library was based on version V3.0 from
Qiagen and complemented with custom siRNAs targeting missing phosphatases.
For a complete list of siRNA oligos see Supplementary Information, Tables S1 and
S2. RNAi duplexes were transfected in liquid phase with either Oligofectamine
(Invitrogen) or HiPerfect (Qiagen) according to the manufacturer’s instruc-
tions. Final siRNA concentrations were 50 nM for Oligofectamine or 10 nM for
HiPerfect. Cells were reverse transfected in 96-well microtitre plates and incubated
for approximately 40 h before imaging.
Quantitative real-time PCR. mRNA was extracted from cells 40 h after
transfection of siRNAs using the TurboCapture 8 mRNA Kit (Qiagen) and
cDNA was prepared using random hexamer primers (Microsynth) and
Ready-To-Go You-Prime First-Strand beads (GE Healthcare). Real-time
PCR was performed using the LightCycler 480 SYBR Green I Master sys-
tem (Roche Diagnostics). Primers were designed using AutoPrime soft-
ware (www.autoprime.de) or Clone Manager. Primer pairs for the indicated
genes were: CA (5ʹ-GGAGCTGGTTACACCTTTG-3ʹ and 5ʹ-CCAGTTA-
TATCCCTCCATCAC-3ʹ), R1A (5ʹ-CTTCAATGTGGCCAAGTCTC-3ʹ
and 5ʹ-TCTAGGATGGGCTTGACTTC-3ʹ) B55α (5ʹ-ATTCGGCTA TGTG-
ACATGAG-3ʹ and 5ʹ-GACCTGTTACT GGGATCTTC-3ʹ), B55δ (5ʹ-CTG
AAAGACGAAGATG GAAG-3ʹ and 5ʹ-AATATTGGG ACCCGTAGC-3ʹ),
importin-β1 (5ʹ- CAGATACG AGGGTACGAGTG-3ʹ and 5ʹ-TTTCAT TGCTTC-
GATTGTG-3ʹ) and GAPDH (5ʹ-CGTGTCAG TGGTGGACCTGACC-3ʹ and
5ʹ-CTGCTTCACCACCT TCTTGATGTCA-3ʹ).
Protein blotting. Pelleted cells were washed with PBS and total protein lysates
prepared in lysis buffer (50 mM Tris at pH 7.5, 150 mM NaCl, 1% (v/v) Nonidet
P-40, 10% (v/v) glycerol and 2 mM EDTA), supplemented with Mini-Complete
protease inhibitor tablet (Roche), and 20 mM β-glycerophosphate. Protein con-
centrations were determined using a BCA kit (Pierce). The following primary
antibodies were used: polyclonal rabbit-anti-PPP2CA (1:2,000, Cell Signaling),
polyclonal rabbit-anti-PPP2R1A (1:2,000, Cell Signaling), monoclonal mouse-
anti-PP2A–B55α (1:500, Santa Cruz), polyclonal rabbit-anti-phosphorylated-
Ser Cdks substrate (1:1,000, Cell Signaling), monoclonal mouse-anti-cyclin B1
(1:5,000, Santa Cruz) and monoclonal mouse-anti-actin (1:50,000, Millipore).
For the chemically induced mitotic exit assay, nocodazole-arrested cells
(100 ng ml–1 for 17 h; Sigma) were incubated for 30 min in 30 μM MG132 (Sigma),
collected by shake-off, washed in PBS containing MG132 and resuspended in
800 μl of PBS (containing MG132). This suspension was divided into 100 μl
aliquots in 1.5 ml centrifugation tubes at 37 °C. Mitotic exit was induced by
adding flavopiridol to a final concentration of 20 μM (provided by the National
Cancer Institute with permission by Sanofi-aventis). Cell aliquots were lysed at
3 min intervals over 18 min by adding 5× concentrated SDS (sodium dodecyl
sulphate) cracking buffer to the respective aliquot, and boiling it for 5 min at
95 °C. Quantitative western blotting was performed using the Odyssey system
(LICOR) or the FluorChem system (Alpha Innotec).
Immunoprecipitation and GST pulldowns. Subconfluent HeLa Kyoto cells
expressing GST–B55α were untreated or arrested in mitosis with nocodazole
(100 ng ml–1 for 16 h; M). Following collection by scraping (untreated cells) or
shake off (cells arrested in mitosis), cells were washed in PBS, pooled and lysed in
1 ml NET buffer (50 mM Tris at pH 7.4, 150 mM NaCl, 1% (w/v) Nonidet P-40 and
15 mM EDTA), supplemented with Mini-Complete protease inhibitors (Roche)
and, when appropriate, with phosphatase-inhibitor tablet (Roche). Equal amounts
of the cleared lysates (as measured using the BCA kit; Pierce) were added to 25 μl
of glutathione–Sepharose (GE Healthcare) and allowed to bind for 1 h at 4 °C. As
previously shown31, this method allows isolation of catalytically competent PP2A–
B55α trimers. After four washes in NENT-100 (ref. 31) and one wash in PP2A assay
buffer (20 mM Tris at pH 7.4 and 5 mM DTT; dithiothreitol), PP2A assay buffer
was added to the beads for phosphatase assays on two different substrates (histone
H1 and phosphorylase a) for at least two times and with three replicates for each
measurement (final volume 100–120 μl). A small aliquot was used for western
blotting with anti-C and anti-A (monoclonals, 1:2,000 dilution, S. Dilworth) and
anti-GST antibodies (1:10,000 dilution, Sigma) to check for equal input.
Phosphatase assays. Histone H1 (6 μg; Roche) was phosphorylated by Cdk1–
cyclin B1 (Biaffin) or Cdk2–cyclin A36 in the presence of 32P-labelled ATP (GE
Healthcare) to a level of 5–7 moles per mole of histone. 32P-radiolabelled histone
H1 was precipitated in 25% (v/v) trichloroacetic acid (TCA) and washed; twice
in 25% (v/v) TCA, once in a solution of acetone containing HCl (200:1) and once
in acetone. This pellet was solubilized in PP2A assay buffer and used at a concen-
tration of 0.3 μM. Following different incubation times with PP2A–B55α at 30 °C
(2, 5, 10 or 20 min), release of inorganic 32P-phosphate was measured through

nature cell biology
© 20 Macmillan Publishers Limited. All rights reserved10

Page 10

DOI: 10.1038/ncb2092
METHODS
extraction of a phosphate–molybdate complex and scintillation counting37.
Under these conditions, phosphate release was < 10% of the phosphorylated
substrate and thus remained in the linear range. Phosphorylase a phosphatase
assays were performed as previously described38, by measuring the liberated
TCA-soluble 32P-labelled phosphate following a 10- or 20-min incubation with
PP2A–B55α. Each measurement of phosphatase activity was performed in trip-
licate. PP2A–B55α activities from mitotic cells were normalized to PP2A–B55α
activities from interphase cells.
Affinity purification of protein complexes. HeLa cell pools expressing LAP-
tagged mouse PP2A subunits from bacterial artificial chromosomes (BACs)
were cultured in ten 25-cm square tissue culture trays. Cells were harvested
from two culture conditions: prometaphase arrest induced by incubation in
0.1 μg ml–1 nocodazole for 18 h and during exponential growth, yielding cells
typically > 90% in interphase. PP2A complexes were isolated from concentrated
extracts of these cells using the two-step purification procedure previously
described27, except that okadaic acid was included in the extract buffer and
all subsequent solutions to inhibit auto-dephosphorylation. Purified protein
complex (20%) was analysed by SDS–PAGE and silver staining. The remaining
sample was processed as previously described39, then divided into two aliquots
for parallel in-solution digestion by trypsin and subtilisin and analysis by liquid
chromatography–mass spectrometry for protein identification and phospho-
rylation site mapping.
Trypsin digestion of SDS–PAGE gel slices. Selected silver-stained protein bands
were excised from an SDS–PAGE gel, cut into smaller pieces and washed by
incubating in a shaking incubator once with 200 μl of 50 mM triethyl ammonium
bicarbonate (TEAB), then once with 100 μl acetonitrile (ACN) plus 100 μl of
50 mM TEAB. Each wash step was performed in a shaking incubator at room
temperature. This two-step wash procedure was repeated, and then excess liquid
removed. To shrink the gel pieces, 100 μl of ACN was added.
Proteins were reduced by incubating the gel pieces in 100 μl of 1 mg ml–1 DTT
in 50 mM TEAB at 57 °C for 30 min, and removing the excess liquid. Proteins
were alkylated by incubating the gel pieces with 100 μl of 5 mg ml–1 iodoaceta-
mide in 50 mM TEAB at room temperature, in the dark, for 35 min. Gel pieces
were subjected to the two-step wash and shrinking procedures, as previously
described, and were then centrifuged in a vacuum concentrator (SpeedVac,
Thermo Scientific) until dry (5–7 min).
Trypsin Gold (Promega) was first dissolved to a concentration of 100 ng μl–1 in
50 mM acetic acid, then diluted in 50 mM TEAB to a concentration of 12 ng μl–1.
Gel pieces were incubated with 20 μl trypsin–TEAB solution at 4 °C for 5 min.
Excess liquid was removed, and 20 μl of 50 mM TEAB was added, followed by
incubation at 37 °C overnight. After centrifugation, the supernatant was trans-
ferred into a fresh tube and stored at 5 °C. Formic acid (20 μl of a 5% (v/v) solu-
tion) was added to the gel pieces, followed by sonication for 10 min in a cooled
ultrasonic waterbath. The sample was spun and the supernatant transferred to
the tube. This formic acid/sonication step was repeated once, and the superna-
tant pooled with the first, generating a sample of volume 60 μl, ready for analysis
by mass spectrometry.
Nano-liquid chromatography–mass spectrometry. The nano-HPLC (high-per-
formance liquid chromatography) system used in all experiments was an UltiMate
3000 Dual Gradient HPLC system (Dionex), equipped with a nanospray source
(Proxeon), coupled to an LTQ FT mass spectrometer (Thermo Fisher Scientific)
in the first analysis and coupled to an LTQ Velos Orbitrap mass spectrometer
(Thermo Fisher Scientific) in the second analysis. The LTQ FT was operated
in data-dependent mode using a full scan in the ICR (ion cyclotron resonance)
cell followed by tandem mass spectrometry scans of the five most abundant ions
in the linear ion trap. Tandem mass spectrometry spectra were acquired in the
multistage activation mode, where subsequent activation was performed on frag-
ment ions resulting from the neutral loss of –98, –49 or –32.6 m/z. Precursor ions
selected for fragmentation were put on a dynamic exclusion list for 180 s and
monoisotopic precursor selection was enabled. Phosphorylated peptides identi-
fied by database search and validated by manual inspection were put onto an
inclusion list and replicate analyses were carried out on an LTQ Velos Orbitrap
mass spectrometer.
Analysis of mass spectrometry data. For peptide identification, all tandem
mass spectrometry spectra were analysed using Mascot 2.2.0 (Matrix Science)
against a customized protein sequence database comprising the complete human
sequences from Swiss-Prot, TrEMBL, PIR, GenBank, EMBL, DDBJ, RefSeq and
Celera (hKBMS), plus the Swiss-Prot database entries corresponding to the mouse
‘bait’ proteins. The following search parameters were used: carbamidomethylation
on cysteine was set as a fixed modification, and oxidation on methionine, and
phosphorylation on serine, threonine and tyrosine were set as variable modifica-
tions. Monoisotopic masses were analysed using an unrestricted range of protein
masses for tryptic peptides. The peptide mass tolerance was set to ± 5 p.p.m. and
the fragment mass tolerance to ± 0.5. The maximal number of missed cleavages
was set to two. Each tandem mass spectrometry spectrum corresponding to a
predicted phosphorylated peptide was validated by visual inspection. A label-
free quantification approach was used to determine abundance change from
interphase to mitosis of the different peptides by peak area integration. For peak
area quantification, we extracted the ion chromatograms of the peptides with a
mass tolerance of 3 p.p.m. using the Qual Browser application from the Xcalibur
software package. The mass traces of peak area integration for the phospho-
rylated peptides shown in Fig. 5f were extracted from the base peak chroma-
togram with ± 3p.p.m. mass tolerance, manually reviewed and are shown in
Supplementary Information, Fig. S8.
35. Schmitz, M. H. & Gerlich, D. W. Automated live microscopy to study mitotic gene func‑
tion in fluorescent reporter cell lines. Methods Mol. Biol. 545, 113–134 (2009).
36. Amniai, L. et al. Alzheimer disease specific phosphoepitopes of Tau interfere with
assembly of tubulin but not binding to microtubules. FASEB J. 23, 1146–1152
(2009).
37. Shacter‑Noiman, E. & Chock, P. B. Properties of a Mr = 38, 000 phosphoprotein
phosphatase. Modulation by divalent cations, ATP, and fluoride. J. Biol. Chem. 258,
4214–4219 (1983).
38. Waelkens, E., Goris, J. & Merlevede, W. Purification and properties of polycation‑
stimulated phosphorylase phosphatases from rabbit skeletal muscle. J. Biol. Chem.
262, 1049–1059 (1987).
39. Gregan, J. et al. Tandem affinity purification of functional TAP‑tagged proteins from
human cells. Nat. Protoc. 2, 1145–1151 (2007).
nature cell biology
© 20 Macmillan Publishers Limited. All rights reserved 10

Page 11

supplementary information
www.nature.com/naturecellbiology
1
DOI: 10.1038/ncb2092
Figure S1 Validation of automated image analysis. (a) Histograms of mitotic exit timing based on IBB-import after anaphase onset (t = 0) of the same dataset
analysed manually (left, white bars), or automatically (right, black bars). (b) Automated annotation is highly reproducibly between movies (three independent
movies shown).
Supplementary Information, Figure S1
a
1.6
Time [min]
3.24.86.4 8.01.6 3.2
Time [min]
4.8 6.4
8.0
1.6
Time [min]
3.24.86.48.01.63.2
Time [min]
4.8 6.48.01.6
3.2
Time [min]
4.86.48.0
b
60
40
20
0
80
60
40
20
0
80
60
40
20
0
80
60
40
20
0
80
% of Cells
60
40
20
0
80
% of Cells
Manual annotationAutomated annotation
Movie 1Movie 2 Movie 3
© 2010 Macmillan Publishers Limited. All rights reserved.

Page 12

supplementary information
Supplementary Information, Figure S2
2
www.nature.com/naturecellbiology
Figure S2 Validation of RNAi efficiency. (a-c) Mitotic exit timing measured
as in Figure 1, for six different siRNAs targeting CA (n ≥ 382 for each siRNA
condition; mean±s.d) (a), R1A (n ≥ 175 for each siRNA condition) (b), or
B55α (n ≥ 108 for each siRNA condition) (c). (d-f) Quantification of mRNA
knockdown 40 h post transfection by real-time PCR for the same siRNAs as in
(a-c), targeting CA (d), R1A (e), or B55α (f), normalized against GAPDH (n =
3 for each condition; mean±s.d). See Supplementary Information, Table 2 for
siRNA sequences. (g) Protein depletion levels of CA, R1A, and B55α, detected
by Western blotting in cells depleted for the indicated siRNAs 60 h post-
transfection. Note that depletion of CA or R1A co-depletes other subunits of the
PP2A-B55α complex, consistent with previous reports39, 40. (h) Quantification
of Importin b1 mRNA knockdown 48 h post transfection. (i) Importin b1
protein levels, detected by Western blotting 64 h after siRNA transfection.
(j) Quantification of B55d mRNA knockdown 48 h post transfection.
ca
g
b
2
4
6
8
10
Time [min]
2
4
6
8
10
Time [min]
2
4
6
8
10
Time [min]
siCtrl
siB55α_1
siB55α_3
siB55α_5
siB55α_6
siB55α_7
siB55α_8
siCtrl
siCA_3 siCA_5siCA_6siCA_7
siCA_8siCA_9
siCtrl
siR1A_1
siR1A_5siR1A_6siR1A_7
siR1A_8siR1A_9
efd
siCtrl
siB55α_1
siB55α_3
siB55α_5
siB55α_6
siB55α_7
siB55α_8
siCtrl
siCA_3siCA_5siCA_6siCA_7siCA_8siCA_9
0.25
0.50
0.75
1.00
relative B55α
mRNA levels
0.25
0.50
0.75
1.00
relative CA
mRNA levels
siCtrl
siR1A_1
siR1A_5
siR1A_6siR1A_7
siR1A_8
siR1A_9
0.25
0.50
0.75
1.00
relative R1A
mRNA levels
α-R1A
α-B55α
α-CA
loading ctrl.
siCtrl
siCAsiR1AsiB55α
siPP2A-B55α
α-actin
α-Importin β1
Importin β1 depletion:
81% 78%
77%
siImportin β1_1
siImportin β1_1
siCtrl
siCtrl
siImportin β1_2
siImportin β1_2
siImportin β1_3
siImportin β1_3
0.25
0.50
0.75
1.00
relative Importin β1
mRNA levels
i
h
siB55δ_1
siCtrl
siB55δ_2siB55δ_5
0.25
0.50
0.75
1.00
relative B55δ
mRNA levels
j
© 2010 Macmillan Publishers Limited. All rights reserved.

Page 13

supplementary information
www.nature.com/naturecellbiology
3
Figure S3 Phenotypes of B55α single depletions. (a) Nuclear reassembly
timing in a B55α-depleted cell, see also Supplementary Information,
Movie S11. (b) Golgi reassembly timing in a B55α-depleted cell, see also
Supplementary Information, Movie S12. (c-e) Cumulative histograms of
postmitotic Golgi clustering (c), chromosome decondensation (d), and spindle
disassembly (e) relative to anaphase onset (0 min). Scale bars: 10 mm.
Supplementary Information, Figure S3
c
e
d
Anaphase - Golgi ClusteringAnaphase - Chromosome
Decondensation
75
50
25
0
100
% of Cells
75
50
25
0
100
% of Cells
siB55α
siControl
siB55α
siControl
a
IBB H2B
IBB
siB55α
3.26.4 12.89.60 min-1.6
b
siB55α
GalT H2B
GalT
13
-2
0 min491622273136
16.022.425.619.2
Time [min]
Time [min]
0
0
10
10
51520
20
30
25
Time [min]
010 5 152025
75
50
25
0
100
% of Cells
Anaphase - Spindle disassembly
siB55α
siPP2A-B55α
siControl
© 2010 Macmillan Publishers Limited. All rights reserved.

Page 14

supplementary information
4
www.nature.com/naturecellbiology
Figure S4 Effect of PP2A-B55α-depletion on cell cycle progression. (a)
Cell cycle staging by DNA replication factor pattern of EGFP-PCNA. (b, c,
d) Cumulative histograms of G1-phase (b), S-phase (c), and G2-phase (d)
duration. Scale bars: 10 mm.
Supplementary Information, Figure S4
bcd
G1-phase G2-phaseS-phase
Time [h]Time [h] Time [h]
0369121501
2
3
4
502
4
6
8
75
50
25
0
100
75
50
25
0
100
75
50
25
0
100
% of Cells
siB55α
siPP2A-B55α
siControl
a
G1mitosisG2
early-SNEBDmid-Slate-S
PCNA H2B
PCNA
0 min20290570660680870
© 2010 Macmillan Publishers Limited. All rights reserved.

Page 15

supplementary information
www.nature.com/naturecellbiology
5
Figure S5 B55α functions downstream of Cdk1-cyclin B inactivation.
(a) Movie of a cell transfected with non-silencing control siRNA, following
the protocol indicated in Figure 4a. Arrowhead indicates onset of nuclear
import of IBB-EGFP. (b) Time-lapse images of a cell transfected with
siRNA targeting B55α, see also Supplementary Information, Movie S13.
(c, d) Cumulative histograms of nuclear envelope assembly
(c) and Golgi clustering (d) timing based on the data shown in (a-b, and
data not shown). (e) Chemical induction of mitotic exit in presence of
proteasome inhibitor in nocodazole arrested mitotic cells results in rapid
dephosphorylation of Cdk substrates (siControl, lanes 1-7). Cells depleted
for B55α show delayed dephosphorylation (siB55α, lanes 8-14).Scale
bars: 10 mm.
Supplementary Information, Figure S5
a
b
cd
e
siB55α
siControl
-2
-2
-1
Flavo.
-11 min5
8
91112141617
1 min4791112141617
IBB
IBB H2B
IBB
IBB H2B
α-Cyclin B1
siControlsiB55α
091215183609 12151836
Flavopiridol [min]
α-phospho(Ser)
Cdk substrates
1456723811121314910
75
50
25
0
100
% of Cells
75
50
25
0
100
% of Cells
Time [min]
01051520
Time [min]
NE ReassemblyGolgi Clustering
0201030
siControl
siB55α
siControl
siB55α
α-Nucleolin
© 2010 Macmillan Publishers Limited. All rights reserved.