Research Article
3429
Melanoma-associated mutations in protein
phosphatase 6 cause chromosome instability and DNA
damage owing to dysregulated Aurora-A
Dean Hammond1,`,*, Kang Zeng1,`,§, Antonio Espert2, Ricardo Nunes Bastos1, Ryan D. Baron1,
Ulrike Gruneberg2," and Francis A. Barr1,"
1
University of Oxford, Department of Biochemistry, South Parks Road, Oxford OX1 3QU, UK
University of Oxford, Sir William Dunn School of Pathology, South Parks Road, Oxford OX1 3RE, UK
2
*Present address: University of Liverpool, Institute of Integrative Biology, Liverpool L69 7ZB, UK
§
Present address: Paterson Institute for Cancer Research, Wilmslow Road, Manchester M20 4BX, UK
`
These authors contributed equally to this work
"
Authors for correspondence ([email protected]; [email protected])
Journal of Cell Science
Accepted 7 May 2013
Journal of Cell Science 126, 3429–3440
2013. Published by The Company of Biologists Ltd
doi: 10.1242/jcs.128397
Summary
Mutations in the PPP6C catalytic subunit of protein phosphatase 6 (PP6) are drivers for the development of melanoma. Here, we analyse
a panel of melanoma-associated mutations in PPP6C and find that these generally compromise assembly of the PP6 holoenzyme and
catalytic activity towards a model substrate. Detailed analysis of one mutant, PPP6C-H114Y, in both primary melanoma and engineered
cell lines reveals it is destabilized and undergoes increased proteasome-mediated turnover. Global analysis of phosphatase substrates by
mass spectrometry identifies the oncogenic kinase Aurora-A as the major PP6 substrate that is dysregulated under these conditions.
Accordingly, cells lacking PPP6C or carrying the PPP6C-H114Y allele have elevated Aurora-A kinase activity and display chromosome
instability with associated Aurora-A-dependent micronucleation. Chromosomes mis-segregated to these micronuclei are preferentially
stained by the DNA damage marker c-H2AX, suggesting that loss of PPP6C promotes both chromosome instability and DNA damage.
These findings support the view that formation of micronuclei rather than chromosome instability alone explains how loss of PPP6C,
and more generally mitotic spindle and centrosome defects, can act as drivers for genome instability in melanoma and other cancers.
Key words: Aurora-A, Protein phosphatase 6, Chromothripsis
Introduction
Genomic instability is considered a driving force for a number of
human malignancies including cancers (Gordon et al., 2012;
Kops et al., 2005; Thompson et al., 2010). Because of this there is
an intense research focus aimed at understanding both how a
stable genome is maintained and how increased genome
instability can be exploited by targeting pathways that
differentiate normal and cancer cells. Alterations in multiple
pathways can promote genomic instability and these can be
grouped into S-phase pathways important for DNA replication
and repair of DNA damage, and M-phase pathways important for
accurate genome segregation. In the case of M-phase pathways,
mutations or overexpression of proteins involved in centrosome
function, sister chromatid cohesion, and mutations in spindle
checkpoint associated proteins and other components of the
mitotic machinery all lead to increased genome instability (Cahill
et al., 1998; Kops et al., 2005; Thompson et al., 2010; Zhou et al.,
1998). Of the mitotic regulators, the Aurora-A mitotic kinase is
the best understood in terms of its impact in human cancers.
Aurora-A is amplified or upregulated in a wide range of
malignancies including cancers of the pancreas, colon, breast,
liver and stomach, and it has been proposed that this may
constitute a major driver in the pathway to chromosomal
instability (Bischoff et al., 1998; Karthigeyan et al., 2010; Lens
et al., 2010; Lentini et al., 2007; Sen et al., 1997; Zhou et al.,
1998). Aurora-A has therefore been formally classified as an
oncogene (Bischoff et al., 1998). This has suggested that
inhibition of Aurora kinase activity may be useful in the
treatment of human cancers and led to the development of
specific drugs targeting Aurora-A and B (Lens et al., 2010). Early
pre-clinical investigations of Aurora-A and Aurora B kinase
inhibitors have demonstrated promising results and phase I/II
trials investigating a number of these compounds are underway
(Lens et al., 2010).
Recent evidence suggests that mutations in the catalytic
subunit of protein phosphatase 6 are found in 9 to 12.4% of
melanomas surveyed and may act as drivers for the development
of melanoma (Hodis et al., 2012; Krauthammer et al., 2012). At
present the basis for this driver effect remains unclear due to the
lack of data on the consequence these mutations have for PP6
activity and the limited information about PP6 substrates. PP6 is
reported to act in the regulation of NF-kappaB, DNA-dependant
protein kinase (DNA-PK), histone c-H2AX and Aurora-A
(Douglas et al., 2010; Mi et al., 2009; Stefansson and
Brautigan, 2006; Zeng et al., 2010). However, it is not known
if the action of PP6 on any of these substrates explains how the
reported melanoma-associated changes in PPP6C promote
tumour development. Here we set out to explain how mutations
in PPP6C result in altered regulation of PP6 substrates. To do this
we first define the consequences of mutations in PPP6C for PP6
3430
Journal of Cell Science 126 (15)
enzyme stability and activity, and then develop an unbiased
proteomic screen that identifies Aurora-A as the key substrate of
PP6 in dividing cells. Using engineered stable cell lines and
PPP6C mutant primary melanoma cells we show that Aurora-A is
dysregulated when PP6 activity is perturbed and that this results
in chromosomal instability and DNA damage.
Results
Journal of Cell Science
Melanoma-associated loss-of-function mutations in PPP6C
PPP6C is mutated at multiple sites in 9 to 12.4% of melanoma
samples tested (Hodis et al., 2012; Krauthammer et al., 2012; Stark
et al., 2012). It has been argued by one study that the observed
pattern of change is characteristic of gain-of-function mutations,
although this is speculative since in all cases the functional
consequences of these alterations are currently unknown (Hodis
et al., 2012). To better understand the potential effect of these
mutations, a homology model for PP6 was created using a
sequence-structure comparison strategy (Shi et al., 2001; Yang
et al., 2012) (Fig. 1A). This analysis of the subunits comprising
PP6 hinted at their likely roles and structures. As expected the PP6
catalytic subunit PPP6C showed a high degree of sequence
homology to PP2A and PP1 catalytic subunits (Shi, 2009).
Intriguingly, the SAPS subunits are likely to be equivalent to the
B56 subunit of PP2A involved in substrate recognition (Fig. 1A;
supplementary material Fig. S1A). Finally, the ANKRD subunits
of PP6 are predicted to contain .17 ankyrin repeats and are
therefore expected to form an extended curved surface
(supplementary material Fig. S1B). This is reminiscent of the
PPP1R1 scaffolding subunits of PP2A that have HEAT
(huntingtin/elongation factor 3/protein phosphatase 2A/TOR1)
repeat motifs (Groves et al., 1999; Xing et al., 2006). We therefore
propose that the PP6 ANKRD subunit forms a platform upon
which the catalytic and SAPS regulatory subunits sit, shown
schematically in the model as a grey outline (Fig. 1A).
Most important for this work, a homozygous mutation at
position 340 (C to T) of the PP6 catalytic subunit (PPP6C)
changing histidine 114 into a tyrosine has been observed in
multiple independently isolated malignant melanoma samples
(Hodis et al., 2012; Stephens et al., 2004; van Haaften et al.,
2009). Although widely conserved in PP2A and PP1 catalytic
subunits and in close proximity to the catalytic centre, histidine
114 does not form part of the network of residues interacting
directly with the bound manganese ions Fig. 1A), and a direct
role in catalysis or metal binding is therefore unlikely (Shi,
2009). However, the high degree of conservation suggests it is
essential for phosphatase function and therefore may be
important for the assembly of the active holoenzyme.
To test these possibilities, recombinant PP6 holoenzymes were
purified containing either wild type PPP6C, the PPP6C tumour
mutant H114Y, a wide panel of PPP6C tumour mutations (Hodis
et al., 2012; Krauthammer et al., 2012), or a phosphatase dead
version of PPP6C containing two mutations abrogating binding
of the metal ions required for catalysis. Western blot analysis of
the phosphatase complexes isolated using the different versions
of the catalytic subunit showed that in contrast to the wild-type
PPP6C, PPP6C-H114Y did not efficiently co-precipitate the PP6
SAPS2 or ANKRD28 subunits (Stefansson and Brautigan, 2006)
(Fig. 1B). Previous studies have shown that PP2A, PP4 and PP6
catalytic subunits interact with alpha4, a chaperone transiently
required to protect the PP2A family catalytic subunits from
degradation until the holoenzyme complex has been assembled
(Kong et al., 2009; Prickett and Brautigan, 2006). Catalytic
subunits unable to associate with the regulatory subunits are
expected to show stronger association with alpha4 but to be
eventually degraded by the proteasome. As expected, alpha4 was
not present in wild-type PP6 holoenzymes, but was found to
stably precipitate with PPP6C-H114Y (Fig. 1B), supporting the
idea that regulatory subunit binding and alpha4-binding are
mutually exclusive (Kong et al., 2009). Consistent with increased
degradation of PPP6C-H114Y, the levels of this protein were
stabilised by the proteasome inhibitor MG132 (Fig. 1C). Low
levels of alpha4 were also present in the PPP6C-PD complexes
(Fig. 1B) and PPP6C-PD was also stabilised by the addition of
MG132 (Fig. 1C), suggesting PPP6C-PD has a partial assembly
defect. Consistent with this idea PPP6C-PD complexes contain
normal levels of SAPS2 but lack ANKRD28 (Fig. 1B). In
contrast, PPP6C-H114Y lacks both the SAPS2 and ANKRD28
subunits (Fig. 1B). Furthermore, only the wild-type PPP6C
efficiently dephosphorylated the generic phosphatase substrate
p-nitrophenylphosphate (Fig. 1D). PPP6C-H114Y showed little
activity above the background value observed with a buffer
control or the catalytically inactive PPP6C-PD (Fig. 1D).
Extending these results, a similar loss of interaction with the
SAPS2 regulatory subunit (Fig. 1E) and reduction in PP6 catalytic
activity (Fig. 1F) was observed for a wider panel of melanomaassociated mutations in PPP6C. Compromised PP6 holoenzyme
assembly and catalytic activity are therefore features generally
associated with the melanoma-associated mutations in PPP6C, and
in at least one case, PPP6C-H114Y, this also destabilises the
catalytic subunit and results in its turnover by the proteasome. This
agrees with the proposal that PPP6C mutations are loss- rather than
gain-of-function (Krauthammer et al., 2012). Because of these
effects, tumour cells carrying PPP6C mutations will therefore lack
PP6 catalytic activity leading to dysregulation of PP6 substrates.
Testing whether or not this is of significance for the development
of the tumour requires a definitive identification of the cellular
pathways in which PP6 functions and the target proteins it acts on.
Unbiased identification of cellular targets for PP6 in mitosis
Current procedures for identifying the substrates of many
enzymes rely on candidate approaches and other inherently
biased strategies. Using such candidate approaches PP6 has
previously been reported to act in the regulation of NF-kappaB,
DNA-PK, histone c-H2AX and Aurora-A (Douglas et al., 2010;
Mi et al., 2009; Stefansson and Brautigan, 2006; Zeng et al.,
2010). However, because of the limitations already mentioned it
remains unclear if these reflect the major functions for PP6 that
could promote tumorigenesis. Here we describe a procedure for
the identification protein phosphatase substrates that overcomes
these limitations and apply it to the identification of PP6
substrates in dividing cells. The principle is based on the
observation that phosphorylation is rapidly lost in samples
incubated on ice for 2 hours after lysis if relevant inhibitors are
absent, but is retained if the enzyme responsible for
dephosphorylation is inactivated by mutation or depleted from
the cells used to make the lysate (Zeng et al., 2010). This can be
exploited in the form of a general unbiased method to identify the
substrates of phosphatases. Whole cell extracts from phosphatase
active and inactive states were subjected to in-solution digestion
followed by strong-cation exchange chromatography and
titanium dioxide phosphopeptide enrichment steps prior to
reverse phase online-LC mass spectrometry (Fig. 2A).
3431
Journal of Cell Science
PP6 loss is a driver for chromosome instability
Fig. 1. Melanoma-associated mutations in PPP6C compromise holoenzyme assembly and phosphatase activity. (A) A homology model of PP6 showing the
PPP6C and SAPS1 subunits, and an outline (grey) to indicate the presumed position of the ANKRD subunit. Human cancer mutations are highlighted in magenta
and also shown on the schematic of the PPP6C subunit (Hodis et al., 2012; Krauthammer et al., 2012). The enlarged region of the catalytic centre shows the
interaction (dotted grey lines) of the side chains of Asn113, His163 and His237, and Asp53, His55 and Asp81 with the two manganese ions. One of these residues,
His55 (magenta) is mutated in tumours (Krauthammer et al., 2012). Alignment of the PPP family catalytic subunits shows that His114 is invariant in all family
members, but the model suggests that is not directly involved in metal ion binding. It is predicted to form a salt bridge (dotted yellow line) to Asp84, and might
therefore be structurally important for positioning two of the catalytic residues, Asp81 and Asn113, and the a5–a6 helices. (B) PP6 holoenzymes were prepared
from HeLa Flp-In cells stably expressing inducible FLAG-tagged wild-type PPP6C, phosphatase dead (PD) or PPP6C H114Y. Purified holoenzyme complexes
were blotted for PPP6C, the PP6 SAPS2 and ANKRD28 subunits, and the alpha4 PPP catalytic subunit chaperone. (C) HeLa cells stably expressing inducible
FLAG-tagged PPP6C constructs were induced with doxycycline (Dox) for 12 hours, and then treated with MG132, as indicated, prior to cell lysis. Cell lysates
were blotted for PPP6C or actin as a loading control. (D) PP6 holoenzymes were tested for activity towards the generic phosphatase substrate p-nitrophenol
(pNPP). Absorbance at 405 nm is plotted in the graph, error bars indicate the standard deviation (n53). (E) PP6 holoenzymes were prepared from HEK293 cells
transfected with FLAG-tagged wild-type PPP6C, phosphatase dead (PD), PPP6C H114Y, S167F, R264C, S270L, K132N, P186S, P259S, G52K, H55Y, L110F,
G112E, Q189R and L305F. The purified holoenzyme complexes were western blotted for PPP6C, and the PP6 SAPS2 and ANKRD28 subunits. (F) PP6
holoenzymes were tested for activity towards the generic phosphatase substrate p-nitrophenol (pNPP). Activity relative to the control (set to 1.0) is plotted in the
graph, error bars indicate the standard deviation from the mean (n52).
Phosphopeptides in the different conditions were identified and
filtered using an analysis pipeline created with the OpenMS
software (Fig. 2A; supplementary material Fig. S2) (Kohlbacher
et al., 2007). Candidate PP6 substrates were defined as those
phosphopeptides lost in the control and untreated samples but
protected in samples depleted of the PP6 catalytic (PPP6C) and
regulatory (SAPS) subunits and by the PPP inhibitor okadaic
acid.
Journal of Cell Science 126 (15)
Journal of Cell Science
3432
Fig. 2. Identification of PP6 targets, using a global phosphopeptide substrate screen. (A) A schematic for the Global Analysis of Phosphatase Substrates
(GAPS). Whole cell extracts were prepared as described in the methods from phosphatase active (siControl and untreated) and inactive states (okadaic acid
treated, siPPP6C, siSAPS/PPP6R). These lysates were then subjected to in-solution digestion followed by strong-cation exchange chromatography (SCX) and
titanium dioxide phosphopeptide enrichment steps. This generated 58 fractions for analysis by reverse phase online-LC collision-induced dissociation (CID) mass
spectrometry. The resulting data files were analysed using a combined Mascot and OpenMS pipeline (supplementary material Fig. S1). (B) This analysis identified
31,884 phosphorylation sites. Venn diagrams show the number of phosphorylation sites identified in the different conditions. Note: some peptides contain more
than one phosphorylation site. (C) Mass spectra for the candidate Aurora-A T-loop phosphopeptide identified using GAPS. The regions for the double- and
triple-charged peptide envelopes at m/z 1152.04 and 768.36 are shown. (D) MS2 spectra from the siPPP6C and siSAPS/PPP6R samples, confirming the peptide
sequence and position of the phosphorylation site at T288.
This analysis identified 198 phosphorylation sites protected in
both PPP6C-depleted cells and by okadaic acid treatment, and
450 phosphorylation sites protected in both SAPS depleted cells
and by okadaic acid treatment (Fig. 2B). These sites mapped to
122 and 258 peptides, respectively, of which two were common
to both datasets (supplementary material Table S1). Filtering for
Journal of Cell Science
PP6 loss is a driver for chromosome instability
high quality Mascot identification and feature assignments left a
single phosphopeptide species identified in two different charge
states, the Aurora-A T-loop peptide (supplementary material
Table S1). The other lower confidence assignment excluded due
to low feature quality and Mascot scores was the DNA damage
repair protein XRCC1 (supplementary material Table S1).
Comparison of the original mass spectra for the double-charged
Aurora-A peptide m/z51152.04 shows it is protected only in the
samples where PP6 is inactive (Fig. 2C) and that the
phosphorylation maps to T288 (Fig. 2D). Similar results were
obtained for the triple-charged Aurora T-loop peptide m/z
5768.36 (Fig. 2C). Blotting confirmed that the Aurora-A T288
residue is dephosphorylated in the absence of phosphatase
inhibitors, as are other phosphoproteins PRC1 and MPS1
(Fig. 3A). However, only Aurora-A T288 phosphorylation was
protected by both okadaic acid treatment and depletion of PPP6C
or the SAPS regulatory subunits (Fig. 3B). By contrast, PRC1
dephosphorylation at T481 (Mollinari et al., 2002) was prevented
by okadaic acid treatment, but still occurred in extracts lacking
PP6 activity (Fig. 3B). The PRC1 T481 phosphopeptide was also
identified by mass spectrometry in the list of peptides protected
only in okadaic acid treated samples.
These findings show that dephosphorylation of Aurora-A at
T288 requires PP6. To prove that this is a direct consequence of
PP6 activity, recombinant PP6 holoenzyme complexes containing
either wild-type or mutant catalytic subunits were incubated with
Aurora-A–TPX2 complex purified from mitotic cells. In this assay,
wild-type PP6 but not a non-specific phosphatase (CIP) readily
dephosphorylated T288-phosphorylated Aurora-A (Fig. 3C)
(Bayliss et al., 2003; Zeng et al., 2010). No dephosphorylation
of Aurora-A T288 was observed with PP6-PD or PP6-H114Y
(Fig. 3C), and similar results were seen with another tumour
3433
mutation S167F (supplementary material Fig. S3A). PP6 activity is
therefore necessary and sufficient for the dephosphorylation of
activated Aurora-A. Furthermore, these results indicate that the
PPP6C-H114Y and other tumour-associated loss-of-function
mutations will lead to a specific amplification of Aurora-A
activity.
Aurora-A activity is amplified in cells expressing PP6
H114Y
The results presented thus far indicate that the tumour-associated
PPP6C-H114Y mutation renders the PP6 catalytic subunit unable
to interact with the regulatory subunits and therefore leads to a
PP6 null state in which Aurora-A activity should be amplified. To
test the effect of the PPP6C-H114Y mutation on Aurora-A
localisation and activity, matched cell lines were created
containing a single integrated copy of a doxycycline-inducible
FLAG-tagged PPP6C wild-type, PPP6C-PD or PPP6C-H114Y.
These transgenes carry silent mutations rendering them resistant
to the PPP6C siRNA duplex 08. When induced with doxycycline
and treated with siPPP6C08 these cells replace the endogenous
PPP6C with the form encoded by the transgene. In control cells,
activated Aurora-A detected by the phosphorylated T288
antibody was tightly concentrated around the mitotic spindle
poles (Fig. 4A, arrowheads), and the Aurora-A regulated spindle
assembly factor NuMA was recruited to the mitotic spindle
(Fig. 4B) (Kettenbach et al., 2011). By contrast, in cells depleted
of PPP6C the staining of both Aurora-A and T288
phosphorylated Aurora-A was elevated and spread along the
spindle microtubules (Fig. 4A, arrows), and NuMA was lost from
the spindles as a consequence (Fig. 4B). Under these conditions
the organisation of the chromatin and mitotic spindle was also
altered (Fig. 4A). Replacement of wild-type PPP6C by H114Y or
Fig. 3. PP6-H114Y lacks Aurora-A phosphatase activity. (A) HeLa
cell lysates were prepared in lysis buffer lacking okadaic acid (OA).
Samples were incubated on ice for 2 hours and okadaic acid was added at
the times shown. All samples were blotted with phosphospecific
antibodies against Aurora-A (pT288), PRC1 (pT481), and antibodies
against Aurora-A, PRC1 and MPS1. MPS1 is phosphorylated at multiple
sites during mitosis and this reduces its mobility (Dou et al., 2011).
Increased mobility, shown by a downshift on SDS-PAGE can therefore be
used as a marker for MPS1 dephosphorylation. Tubulin and cyclin B were
used as loading controls. (B) HeLa cells were transfected for 48 hours
with control, PPP6C si08 or SAPS1/2/3 siRNA (siPPP6R) duplexes.
Lysates were then prepared from these cells in lysis buffer containing (+)
or lacking (2) okadaic acid in the combinations shown in the figure.
After 2 hours incubation on ice the samples were blotted using the
antibodies shown. (C) Purified Aurora-A–TPX2 complexes were
incubated with wild-type or catalytically inactive recombinant PP6
holoenzymes, PP6-WT, and PP6-PD and PP6-H114Y, respectively,
buffer, mock purified enzyme, or 2.5 units calf intestinal phosphatase
(CIP) for 2 hours then blotted as shown in the figure.
Journal of Cell Science
3434
Journal of Cell Science 126 (15)
Fig. 4. Loss of PP6 results in
amplification of Aurora-A kinase activity.
Flp-In cell lines containing a single
integrated copy of a doxycycline-inducible
FLAG-tagged PPP6C wild-type or H114Y
were transfected with control or PPP6C08
siRNA duplexes for 24 hours, then induced
with doxycycline for 24 hours. The cells
were fixed, and then stained with antibodies
for (A) Aurora-A, T288 phosphorylated
Aurora-A (pT288) and tubulin, or
(B) NuMA and tubulin. DNA was stained
with DAPI. Scale bars: 10 mm. (C) Histone
H3 kinase assays were performed using
Aurora-A–TPX2 complexes purified from
cells expressing only PPP6C wild-type,
phosphatase dead (PD) or PPP6C-H114Y.
In addition, incubations using a mock
isolation (control) and a buffer control were
performed. Radioactive incorporation into
histone H3 ([32P]H3) was measured and is
plotted in the graph; error bars show the
standard deviation (n53). A paired Student’s
t-test indicated a significant (**P50.011)
difference in kinase activity between wildtype control and PPP6C-H114Y conditions.
A Coomassie-Blue-stained gel shows equal
loading of histone H3. Samples were also
blotted for Aurora-A and T288
phosphorylated Aurora-A (pT288).
S167F mutant forms caused the same spread of Aurora-A and the
active form of Aurora-A along the mitotic spindle (Fig. 4A,
arrows and S3B), and NuMA was also lost from the mitotic
spindle (Fig. 4B and supplementary material Fig. S3C). These
results support the notion that Aurora-A activity is increased in
cells expressing PPP6-H114Y or S167F. To confirm this idea
Aurora-A–TPX2 complexes were isolated from cells expressing
wild-type and mutant forms of PPP6C and their activity tested
using in vitro kinase assays. This approach revealed that there is a
200–250% increase in Aurora-A activity in cells expressing
PPP6C-H114Y or the PPP6C-PD catalytic mutant when
compared to control cells (Fig. 4C). The overall level of
activated T288 phosphorylated Aurora-A was also increased
under these conditions (Fig. 4C). Together, these results show
that when PP6 function is compromised by the PPP6C-H114Y
mutation Aurora-A targeting to the mitotic spindle is increased
and Aurora-A activity is amplified.
To extend key aspects of these findings LB373-MEL primary
melanoma cells homozygous for the PPP6C H114Y mutation
were then compared to a control melanoma cell line LB2518MEL carrying wild-type PPP6C (Stephens et al., 2004; van
Haaften et al., 2009). In agreement with biochemical analysis of
PPP6C-H114Y mutant stability (Fig. 1C), LB373-MEL tumour
cells showed reduced levels of PPP6C compared to the LB2518MEL control cells and an upshifted T288 phosphorylated form
of Aurora-A (Fig. 5A). When lysates were prepared in the
absence of phosphatase inhibitors, Aurora-A remained in a
phosphorylated state in LB373-MEL cells but was rapidly
dephosphorylated in control LB2518-MEL melanoma cells
(Fig. 5B), demonstrating that the absence of PPP6C protein in
LB373-MEL also leads to functional loss of the Aurora-A T-loop
phosphatase. These cells also showed greatly increased
micronucleation relative to LB2518-MEL control cells (Fig. 5C),
fitting with the idea that they have mitotic defects and are
genomically unstable as a consequence. The LB373-MEL cells
also showed increased levels of total and pT288 activated AuroraA at the mitotic spindle relative to the LB2518-MEL cells
(Fig. 5D). Confirming the idea that LB373-MEL cells display an
Aurora-A amplification phenotype, limiting inhibition of AuroraA with the specific drug MLN8537 significantly (P,0.001)
reduced the extent of micronucleation by nearly 25% (Fig. 6A,B).
Interestingly, the same treatment caused a slight but significant
(P,0.001) 7.3% increase in nucleation in the LB2518-MEL cells,
suggesting that reduced Aurora-A activity also leads to increased
genome instability (Fig. 6A,B).
Together, these results from engineered model cell lines and
primary tumour cells provide compelling evidence that mutations
in PPP6C can result in altered Aurora-A regulation and genome
instability. One caveat is the difficulty of proving that the
micronucleation phenotype observed in LB373-MEL cells is a
direct consequence of the loss of PP6 activity due to the
acquisition of inactivating mutations in PPP6C. To investigate
whether reduction of PP6 activity directly induces
micronucleation and chromosomal instability, engineered cell
lines where the PP6 catalytic subunit can be rapidly replaced by
wild-type or mutant forms were therefore used.
Loss of PP6 function results in chromosome instability
Replacement of endogenous PPP6C with a wild-type
si08ResPPP6CWT transgene did not result in any alterations in
PP6 loss is a driver for chromosome instability
3435
nuclear morphology (Fig. 7A) or increased micronucleation
(Fig. 7B). When the transgene was not induced, or both
endogenous and transgene expressed PPP6C was depleted using
siPPP6C07 nuclear morphology was altered and the cells became
micronucleated (Fig. 7A,B). Replacement of endogenous PPP6C
with the si08ResPPP6CH114Y resulted in the formation of multilobed nuclei (Fig. 7A) and caused micronucleation in 50% of
cells (Fig. 7B). Similar effects were seen when a catalytically
inactive si08ResPPP6CPD transgene was used (Fig. 7A,B). The
formation of micronuclei is consistent with the idea that
chromosome segregation is perturbed in these cells and that
PPP6-H114Y causes chromosome instability. To directly test the
idea that reduction of PP6 activity promotes chromosomal
instability, fluorescence in-situ hybridisation (FISH) was
performed using probes directed towards chromosomes 2 and
8. The parental cell line is known to be triploid for these
chromosomal loci (Macville et al., 1999) and this was confirmed
by the analysis (Fig. 7C). By contrast, chromosomes 2 and 8
showed altered segregation in the absence of PP6 function, with
,5% single chromosome mis-segregation events for either one of
these loci resulting in 10.6–10.7% cells with altered chromosome
number (Fig. 7C).
Journal of Cell Science
Elevated chromosome instability in PPP6C mutant primary
tumour cells
Fig. 5. PPP6C-H114Y melanoma cells show loss of PP6 function and
amplified Aurora-A phosphorylation. (A) LB2518-MEL and LB373-MEL
cells were arrested in mitosis with nocodazole and cell lysates were blotted
for pT288 Aurora-A, Aurora-A, cyclin B, PPP6C and actin, as a loading
control. (B) LB2518-MEL and LB373-MEL cells were arrested in mitosis
with nocodazole, harvested and lysed in the presence or absence of
phosphatase inhibitors (PP Inh). Lysates were blotted for Aurora-A pT288
and total Aurora-A. The change in Aurora-A pT288 was measured,
normalized to total Aurora-A, and plotted on the graph; error bars show the
standard deviation (n53). An independent sample Student’s t-test indicated a
significant (**P50.018) difference in Aurora-A phosphorylation between
LB2518-MEL control and LB373-MEL PPP6C-H114Y cells in the absence of
phosphatase inhibitors. Compared with the control LB2518-MEL cells, no
significant reduction in Aurora-A phosphorylation occurred in the absence of
phosphatase inhibitors in LB373-MEL (P50.513). (C) LB373-MEL and
matched control LB2518-MEL cells were fixed and stained with CREST
serum to detect centromeres and kinetochores, and DAPI to detect DNA.
Cells with CREST-positive micronuclei, indicating chromosome segregation
defects in mitosis, were counted. This is expressed as a percentage, with
standard deviation for three independent experiments and shown on the
images in the figure. (D) LB373-MEL and matched control LB2518-MEL
cells in mitosis were stained for Aurora-A and pT288 Aurora-A. DAPI was
used to detect DNA. Scale bars: 10 mm.
To further support the idea that PPP6 mutation results in
chromosome instability, primary melanoma cell lines were
analysed. FISH was performed using probes for chromosomes
2 and 8, or 7 and 15 in the PPP6C mutant melanoma cell line
LB373-MEL, the PPP6C wild-type melanoma cell line LB2518MEL and control diploid RPE1 cells. As expected, the control
RPE1 cells were diploid and analysis of both chromosomes 2 and
8 or 7 and 15 showed fewer than 5% of cells had a non-modal
chromosome number (Fig. 8A and supplementary material Fig.
S4). The PPP6C wild type melanoma cells line had altered copy
number for all chromosomes tested, and 20–25% of cells
displayed a non-modal chromosome number (Fig. 8A and
supplementary material Fig. S4). Strikingly, up to 80% of the
LB373-MEL PPP6C-H114Y mutant melanoma cells had a nonmodal chromosome number (Fig. 8A and S4). This high level of
cells with a non-modal chromosome number reflected a
statistically significant (P,0.001) increase in chromosome
instability, when compared to an equivalent PPP6C wild-type
melanoma (Fig. 8A). This pattern of chromosome instability is
consistent with the high frequency of micronucleation observed
in the LB373-MEL PPP6C-H114Y cells and the low level seen in
the LB2518-MEL PPP6C wild-type cells (Fig. 5C). These data
therefore support the hypothesis that mutation of PPP6C and
consequent loss of PP6 function correlate with increased
chromosomal instability and therefore directly promote the
formation of micronuclei.
Micronucleation and DNA damage as a consequence of
reduced PP6 activity
Although there is strong evidence to suggest that chromosomal
instability promotes tumorigenesis, the precise molecular
mechanisms are unclear. An influential study on the origins of
chromothripsis, the catastrophic re-arrangement of single
chromosomes or chromosome arms in cancer cells and
considered a catalyst or driver for tumour evolution, has
suggested that micronucleation induced by chromosomal
3436
Journal of Cell Science 126 (15)
Journal of Cell Science
Fig. 6. Reduced genome instability in LB373-MEL
cells following partial Aurora-A inhibition.
(A) LB2518-MEL and LB373-MEL cells were grown
for 1 week in the presence or absence of 20 nM
MLN8357, a specific Aurora-A inhibitor. The cells
were fixed and then stained with CREST antiserum to
mark centromeres, and DAPI to detect DNA. (B) The
number of cells with fragmented nuclei or micronuclei
was counted, and is plotted in the graph as a percentage
(n53, for 300 hundred cells per experiment, error bars
indicate the standard deviation). A Chi-squared test
demonstrated a statistically significant difference in
micronucleation following MLN8537 treatment in
both cell lines (**P,0.001). Scale bars: 10 mm.
instability may be one such trigger (Crasta et al., 2012; Forment
et al., 2012; Stephens et al., 2011). Such micronuclei are
defective and delayed in DNA replication, resulting in DNA
damage and premature DNA condensation (Crasta et al., 2012).
However, how this would be initiated or what the drivers for
these events could be was not investigated.
To test whether loss of PP6 function induces micronuclei
exhibiting DNA damage, PPP6C-depleted HeLa cells were stained
with antibodies against the Ser139-phosphorylated form of histone
H2A.X (c-H2AX), an early marker for DNA damage (Rogakou et al.,
1998) (Fig. 8B). In these cells, loss of PPP6C resulted in a significant
increase in c-H2AX-positive micronuclei (Fig. 8B). Similar results
were obtained in the primary tumour cells lines, LB373-MEL PPP6CH114Y mutant melanoma cells showed a significant increase in cH2AX-positive micronuclei relative to the matched control (Fig. 8C),
consistent with the view that chromosomal instability combined
with micronucleation results in DNA-damage. Together, these
observations reveal that loss of PP6 results in a significant increase
in chromosomal instability, micronucleation and DNA damage, and
suggest an attractive explanation of why mutations in PPP6C may act
as driver mutations in UV-induced skin cancer (Hodis et al., 2012;
Krauthammer et al., 2012).
Fig. 7. Chromosome instability is a hallmark of
PP6 inactivation. (A) HeLa Flp-In cells stably
expressing inducible PPP6C si08-resistant wild-type
(si08ResPPP6CWT) phosphatase dead mutant
(si08ResPPP6CPD), H114Y (si08ResPPP6CH114Y)
forms of PPP6C tagged at the N-terminus with a
FLAG epitope were transfected with control or si07
and si08 duplexes targeting PPP6C for 48 hours.
Doxycycline (1 mg/ml) was included in the growth
medium for the last 24 hours to induce expression of
the transgenes. The cells were fixed and then stained
with DAPI, or blotted for PPP6C and actin as a
loading control. (B) Micronucleation and abnormal
nuclear morphology was scored for each condition
and is plotted in the graph. Error bars indicated the
standard deviation from the mean (n53). (C) Control
and PPP6C-depleted cells were analyzed using FISH
with probes for loci on chromosomes 2 and 8.
Synchronized cells were used to allow the analysis of
the products of a single round of mitosis. Examples
images are shown, with chromosome 2 marked in red
and chromosome 8 marked in green. The frequency
of chromosome mis-segregation was counted for the
progeny of 1000 mitotic events in all conditions and
plotted on the graphs. Because the cells are stably
triploid at the outset this category is not shown. Scale
bars: 10 mm.
PP6 loss is a driver for chromosome instability
3437
Journal of Cell Science
Fig. 8. Micronucleus-associated DNA damage in the absence
of PP6 activity. (A) LB2518-MEL, LB373-MEL and RPE1 cells
were grown on 12-well micro-chambers and processed for FISH
analysis for chromosomes 2 and 8, or 7 and 15. Representative
images for chromosome 2 and 8 are shown on the left-hand side,
with colour-coded numbers to indicate how many copies of each
chromosome were detected. For chromosomes 2 and 8, or 7 and
15, 660 or 672 cells per cell line, respectively, were examined.
Modal chromosomal number was calculated for each cell line,
and individual cells were then classified as having the modal or a
non-modal chromosomal number (error bars show the standard
deviation, n52). Using either the chromosome 2 and 8, or the
chromosome 7 and 15 datasets, cell lines were compared in
pairwise fashion using a Chi-squared test. Bonferroni correction
was applied to account for pairwise testing, with P,0.0166
representing significance at the 5% level. (B) HeLa cells were
depleted of PPP6C by siRNA for 60 hours were stained with
antibodies to c-H2A.X and CREST serum. DNA was visualized
with DAPI. (C) LB2518-MEL and LB373-MEL cells were
stained with antibodies to c-H2AX and DNA was visualized with
DAPI. In both B and C the number of c-H2AX-positive
micronuclei were counted (600 cells from three different
experiments, error bars indicate standard deviations). (D) In this
model PPP6C loss-of-function mutations result in chromosome
instability combined with micronucleation due to a failure to
properly segregate chromosomes to a single defined nucleus.
Although chromosome instability alone promotes aneuploidy,
chromosome instability combined with micronucleation results
in DNA damage to those chromosomes present in the
micronucleus, which is reminiscent of the process of
chromothripsis. Scale bars: 10 mm.
Discussion
If Aurora-A is an oncogene, is PPP6C a tumour
suppressor?
All of the mutations tested resulted in defective assembly of PP6
holoenzyme complexes and are therefore expected to result in
reduced enzyme activity. For the PPP6C-H114Y mutation, where
a primary tumour cell line exists, it was confirmed that this change
leads to activation of the major PP6 substrate, Aurora-A. Based on
these findings, PPP6C could therefore be considered a candidate
tumour-suppressor gene, opposing Aurora-A already classified as
an oncogene (Bischoff et al., 1998). Corroborating these findings,
PPP6C has recently been identified as ‘driver gene’ for melanoma
(Hodis et al., 2012; Krauthammer et al., 2012). Mutations in a
‘driver gene’ are thought to confer a selective advantage to an
emerging tumour and are thus distinct from ‘passenger’ mutations
that do not confer a selective advantage. Our work provides a
biological basis explaining why mutations in PPP6C have ‘driver’
properties: they destabilize the PP6 holoenzyme and result in
amplified Aurora-A activity, which in turn perturbs efficient
chromosome congression and segregation.
Although chromosome instability has long been recognized as
a hallmark and potential driving force of human cancers (Boveri,
2008; Holland and Cleveland, 2012; Lengauer et al., 1998;
McGranahan et al., 2012), the mechanisms behind chromosome
instability in tumours continue to be largely unexplained.
The analysis of tumour-associated PPP6C mutations and
their biological consequences demonstrates how chromosome
instability may arise. Importantly, loss of PP6 function not only
Journal of Cell Science
3438
Journal of Cell Science 126 (15)
results in chromosome instability, it also promotes
micronucleation due to defective chromosome segregation
(Zeng et al., 2010), which emerging evidence shows is the
origin of a phenomenon termed chromothripsis or chromosome
shattering (Crasta et al., 2012; Forment et al., 2012). This type of
genomic re-arrangements is thought to contribute significantly to
tumour evolution however how it is initiated remains mysterious.
The findings presented here support the view that it is the
combination of chromosome instability and micronucleation,
with resulting DNA damage that may explain why PPP6C
mutations act as drivers in melanoma. These ideas can be
summarised in the following scenario for PPP6C mutations in
UV-induced skin cancer progression (Fig. 8D). Exposure of skin
cells to UV results in undirected DNA damage at a variety of
loci. Specific mutations in the coding region for PPP6C result in
loss of PP6 catalytic activity as well as reduced protein stability.
These cells will enter mitosis with amplified Aurora-A activity,
resulting in impaired spindle formation, chromosome segregation
errors and thus micronucleation as we have shown previously
(Zeng et al., 2010). Micronuclei are impaired in their DNA
replication capacity due to limiting number of nuclear pores and
structural defects in the nuclear envelope (Géraud et al., 1989;
Hoffelder et al., 2004). These cells will therefore undergo
successive rounds of mitosis with under-replicated micronuclear
DNA driving chromosome fragmentation as a consequence
(Crasta et al., 2012). The resulting DNA damage provides a
mechanism to promote tumorigenesis, since within a population
of such cells there are likely to be a number with one or more
additional driver mutations. We therefore propose that loss of PP6
activity predisposes cells to chromothripsis and tumorigenesis
through the pronounced accumulation of micronuclei in an
Aurora-A-dependent fashion.
Identification and management of PPP6C mutant tumours
As already discussed, Aurora-A is amplified in many cancers and
is currently the focus of much interest as a drug target for cancer
therapy (Bischoff et al., 1998; Karthigeyan et al., 2010; Sen et al.,
1997; Zhou et al., 1998). As we show, loss of PP6 function leads
to increased Aurora-A activity and creates a state similar to direct
amplification of the Aurora-A locus. Aurora-A inhibitors may
therefore prove valuable to treat cancers with dysregulated
PPP6C, with the caveat that they increase genome instability in
normal cells. An intriguing immediate possibility is that AuroraA inhibitors already in clinical trials could be used at low doses
to help manage amplified Aurora-A activity in PPP6C
mutant tumours. This would require the development of
simple screens for changes in PPP6C, but as we have shown
here, these mutations appear to destabilize the protein and
immunohistochemistry could therefore be used to detect
reduction or loss of PPP6C in tumour biopsies.
Materials and Methods
Reagents and antibodies
Laboratory reagents were obtained from Thermo Fisher Scientific and SigmaAldrich. Commercial antibodies were used to detect a-tubulin (mouse, DM1A;
Sigma-Aldrich), Aurora-A (rabbit, AB12875; Abcam), phospho-Aurora-A pT288
(rabbit 3079; Cell Signaling Technology), pan phospho-Aurora-A/B/C pT288/232/
198 (rabbit, 2914; Cell Signaling), phospho-PRC1 T481 (goat, sc-11768, Santa
Cruz), PPP6C (rabbit, A300-844A; Bethyl Laboratories, Inc.), SAPS1–3 (rabbit,
A300-968A, A300-969A and A300-972A; Bethyl), MPS1/TTK (mouse, [N1],
ab11108; Abcam), TPX2 (mouse, ab32795; Abcam) and cyclin B1 (mouse, GNS3;
Millipore). Affinity purified primary and secondary antibodies were used at 1 mg/
ml final concentration and sera were used at 1:1000 dilution. Secondary antibodies,
raised in donkey, to mouse, rabbit and sheep/goat conjugated to HRP, Alexa Fluor
488, Alexa Fluor 555, Alexa Fluor 568 and Alexa Fluor 647 were obtained from
Molecular Probes.
Cell culture and microscopy
LB2518-MEL and LB373-MEL melanoma primary cell lines were cultured in
Iscove’s medium supplemented with 10% fetal calf serum, AAG (0.24 mmol/L of
L-asparagine, 0.55 mmol/L of L-arginine, 1.5 mmol/L of L-glutamine) and HITES
(10 nM hydrocortisone, 5 mg/ml insulin, 100 mg/ml transferrin, 10 nM 17-bestradiol, 30 nM sodium selenite). Stable HeLa cells were created as described
previously and cultured in growth medium (DMEM containing 10% foetal bovine
serum) at 37 ˚C and 5% CO2 (Bastos et al., 2012). For plasmid transfection and
siRNA transfection, Mirus LT1 (Mirus Bio LLC) and Oligofectamine (Invitrogen),
respectively, were used according to the manufacturer’s instructions. Microscopy
was performed exactly as described elsewhere (Dunsch et al., 2012).
Homology modelling
Homology models for the PPP6C and SAPS1 subunits of PP6 were built using
FUGUE to carry out sequence-structure comparison (Shi et al., 2001). For PPP6C
the best match was seen to the PP1 and PP2A catalytic subunits structures, and the
PDB structure 1tcoa was used for model building. For SAPS1 a good match was
seen with the PP2A-B56 regulatory subunit in PDB structure 3fgab. Homology
models were refined with Modeller and Chimera (http://www.cgl.ucsf.edu/
chimera) (Pettersen et al., 2004; Yang et al., 2012) then exported to Pymol
1.5.0.5 to create figures (The PyMOL Molecular Graphics System, Version 1.5.0.5
Schrödinger, LLC.).
Isolation of peptides from whole mitotic cell extracts
Proteins were isolated from cells arrested in mitosis with 100 ng/ml nocodazole for
16 h. Mitotic cells were harvested by shake off, washed three times with warm
PBS and then re-suspended in warm growth medium. Cells were incubated in a 5%
CO2 incubator at 37 ˚C for 20 min and then washed twice with cold PBS. The cells
were lysed in a buffer containing 20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1%
NP-40, 0.1% sodium deoxycholate and 1:250 protease inhibitor cocktail (Sigma),
left on ice for 75–100 min, then centrifuged at 21,000 g for 15 min at 4 ˚C to
remove insoluble debris. Lysates containing 25 mg of mitotic cell protein extract
were mixed with five volumes 100% trichloroacetic acid (TCA) and incubated
overnight on ice. Precipitates were centrifuged for 8 min at low speed (1006g)
then washed twice with ice-cold acetone. The pellets were dried at 37 ˚C to remove
acetone and re-suspended in 10 ml UA (8M urea, 100 mM Tris-HCl [pH 8.5])
with 100 mM dithiothreitol (DTT) and loaded into Amicon Ultra 15 30 kDa spinconcentrators (Millipore) for tryptic digestion by filter-aided sample preparation
(Wiśniewski et al., 2009). After 20-fold concentration at 47006g, 10 ml UA was
added and the sample concentrated again. Concentrates were then mixed with 2 ml
of 50 mM iodoacetamide (IAA) in UA buffer and incubated at room temperature
(RT) for 30 min then centrifuged for 45 min. Concentrates were resuspended in
10 ml of 50 mM ammonium bicarbonate (Ambic) and the concentration process
repeated twice to remove urea. Trypsin, 4 ng/ml final concentration, was added to
the final concentrate and digestion performed at 37 ˚C overnight. Tryptic peptides
were collected by centrifugation for 10 min at 36006g. Two further elutions with
2 ml of 50 mM Ambic were performed and the eluates pooled. Pooled eluates
were desalted using 1 g (20 cc) Oasis HLB plus cartridges (part number
186000117, Waters Corporation) according to the manufacturer’s instructions.
Peptides were eluted with 90% acetonitrile (Acn)/0.5% formic acid, pooled
together and vacuum dried.
Phosphopeptide enrichment strategy and mass spectrometry
To enrich phosphopeptides a sequential strategy of strong cation exchange (SCX)
chromatography followed by titanium dioxide (TiO2) binding was employed
(Olsen and Macek, 2009). Dried peptides were dissolved in buffer A (5 mM
ammonium formate, 20% Acn, acidified with trifluoroacetic acid (TFA) to pH 2.7
and centrifuged for 10 min at 21,0006g at 4 ˚C prior to loading onto a 1 ml
Resource S column (GE Healthcare) via a 5 ml injection loop. Peptides were
eluted at 1 ml/min collecting 1 ml fractions by a 20 min linear gradient from
100% buffer A to 50% buffer B (200 mM ammonium formate, 20% Acn, acidified
with TFA to pH 2.7) followed by a 5 min linear gradient ramp from 50% to 100%
buffer B and a final 5 min with 100% buffer B. Half of each fraction was vacuum
dried and reconstituted in 40 ml GA solution (80 mg/ml glycolic acid, 80% Acn
and 2% TFA) for TiO2-based enrichment of phosphopeptides (Rappsilber et al.,
2003; Rappsilber et al., 2007). A 200 ml pipette tip was plugged with C8 matrix,
4 mg of TiO2 bead slurry added, equilibrated with 40 ml of 0.6% NH4OH and then
40 ml of GA. Peptides were applied to the tips at low flow rate (2000 rpm in a
bench-top micro-centrifuge for ,5 min). The tips were washed with 40 ml of GA,
40 ml of a mixture of 80% Acn and 0.2% TFA and finally 40 ml of 20% Acn.
Phosphopeptides were eluted first with 40 ml of 0.6% NH4OH and 40 ml of 60%
Acn and the eluates pooled (80 ml). A second phase of elution, in 40 ml of 1%
pyrrolidine, followed by 40 ml of 60% Acn was also carried out. All eluates were
PP6 loss is a driver for chromosome instability
dried, re-dissolved in 0.5% formic acid, then analyzed by online LC-MS2 with a
nanoAcquity UPLC system (Waters Corporation) and a hybrid mass spectrometer
(LTQ-Orbitrap XL ETD; Thermo Fisher Scientific) fitted with a nanoelectrospray
source (Proxeon) as described previously (Zeng et al., 2010). All spectra were
acquired using Xcalibur software (version 2.0.7; Thermo Fisher Scientific).
Journal of Cell Science
Data processing using OpenMS
MSConvert (part of MSFileReader version-13_04.14.2009; Thermo Fisher
Scientific) was used to convert data files acquired by Xcalibur into mzML
format. OpenMS (version 1.8.0) proteomics pipeline (Kohlbacher et al., 2007;
Sturm et al., 2008) was used to create a data filtering/analysis procedure to directly
compare matched SCX/TiO2 samples. Data acquired in profile mode were
centroided, then MS1 peaks picked using the high-resolution PeakPicker peakpicking algorithm. Peptide ions with charge states 2+ to 5+ within an intensity
range of 56104 and 16109 were then selected. MS features were then identified
with the FeatureFinder centroided data algorithm. Features in the equivalent
fractions from different samples were matched together using the FeatureLinker
‘unlabelled’ algorithm with a maximum allowed RT or m/z difference of
10 seconds or 10 ppm, respectively. The outputs, either siPPP6C, plus OA,
siControl, minus OA, or siPPP6R, plus OA, siControl, minus OA were merged into
one consensus XML file. Each sample is linked to a numbered consensus ‘element
map’ within this file.
In parallel to the batch processing of data files described already, Mascot
(version 2.2.0.3; Matrix Science) was used to search each input mzML data file
using the MascotAdapterOnline tool, against the human international protein index
database (version 3.60; 79,371 protein entries), producing a corresponding idXML
search result file for each SCX fraction/TiO2 enrichment. Search parameters were:
precursor mass tolerance 6 1 Da, and 6 0.8 Da for MS2 fragmentation spectra,
enzyme specificity was set to trypsin with two missed-cleavages, Cyscarbamidomethylation as a fixed modification, and Met-oxidation and
phosphorylation of Ser/Thr as variable modifications. The idXML files from
parallel SCX/TiO2 samples were then concatenated using IDMerger. To match
features with peptide sequences merged consensusXML files were annotated with
the peptide sequence information from the corresponding merged idXML file
using TOPPView. Perl scripts were used to extract only PPP6-sensitive
phosphorylation events from the data, inserting a ‘user parameter metadata tag’
(‘‘PP6 substrate’’) into the XML file structure if a phosphopeptide is present in the
siPPP6C or siSAPS/PPP6R and plusOA, but not in control samples (siGL2control
and untreated/minusOA); that is, only in consensus element maps 0 and 1 but not 2
and 3. A pairwise comparison of candidate phosphopeptides identified in both the
siPPP6C and siSAPS/PPP6R (supplementary material Table S1) gives a high
confidence lists of PP6 substrates, complete with Mascot scores, feature quality,
mass, charge state, LC retention time and intensity values.
3439
In situ hybridisation on synchronized cells
For fluorescence in-situ hybridisation, 106 cells were seeded in 10 cm dishes then
transfected with control or PPP6C siRNA duplexes after 24 hours. After 48 hrs
siRNA, the cells were arrested with 2 mM thymidine for 20 hrs, then washed three
times in warm PBS and once with fresh growth medium. Mitotic cells were shaken
from the dishes after 9–10 hrs, re-seeded on glass slides, then incubated at 37 ˚C in
a 5% CO2 incubator. After 5–6 hrs, when the cells had attached to the glass and
completed mitosis, the slides were treated in hypotonic solution (0.56% KCl) for
30 min at 37 ˚C, then fixed by dipping in ice cold 3:1 methanol:acetic acid for
30 min changing the fix every 10 min. The slides were air dried overnight then
washed with 26SSC (206SSC stock: 17.5 g NaCl and 8.8 g sodium citrate
dissolved in 80 ml deionised water, adjusted to pH 7.0 with HCl, and made to a
final volume of 100 ml) for 5 min at room temperature. Dehydration was carried
out by a series of 2 min washes in 70%, 85% and 100% EtOH. Slides were placed
at 37 ˚C then 2 ml of the FISH satellite enumeration probes for chromosomes 2 and
8 labelled with FITC and Texas-Red (Cytocell Cambridge, UK), respectively, were
added in 10 ml hybridization buffer. The liquid was covered with 22622 mm
coverslips and the edges sealed with rubber cement. The slides were heated at 76 ˚C
for 3 min using a metal heating block, then kept at 37 ˚C in a water bath overnight.
After removal of the rubber cement and coverslips the slides were washed in
0.256SSC for 2 min at 72 ˚C, then again in 26SSC with 0.05% [vol/vol]
TWEEN20 for 30 sec at room temperature. Finally, the slides were dried and a
coverslip mounted over the hybridized region using Moviol 4–88 containing 1 mg/
ml DAPI. Unsynchronised melanoma cells were processed for FISH analysis in the
same way but were grown in 12-well micro chambers (ibidi, Martinsried,
Germany).
Online supplemental material
Supplementary material Table S1 contains mass spectrometry data for the PPP6C
and PPP6R/SAPS activity sensitive phosphopeptides. Supplementary material Figs
S1–S4 describe modelling of PP6 scaffold and regulatory subunits, the GAPS
processing pipeline and provide additional evidence that melanoma-associated
mutations reduces PP6 activity towards Aurora-A. Example batch files for GAPS
are provided (supplementary material Fig. S5).
Acknowledgements
LB2518-MEL and LB373-MEL-D cells were kindly provided by Drs
Francis Brasseur and Nicolas van Baren (Ludwig Institute for Cancer
Research, Brussels). We thank Dr John Brognard (Paterson Institute
for Cancer Research, Manchester) for highlighting the S167F
mutation, and Dr Victoria Furio for advice on analysis of
chromosome distributions.
Time-course of de-phosphorylation assay
Mitotic cell suspensions prepared as described above were pelleted by
centrifugation for 5 min at 300 g and equally divided into ten 1.5 ml
microcentrifuge tubes on ice. One cell pellet was immediately lysed in a buffer
containing 20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% NP-40, 0.1% sodium
deoxycholate, 40 mM b-glycerol phosphate, 10 mM NaF, 1 mM okadaic acid,
1:100 Sigma phosphatase inhibitor cocktail III and 1:250 Sigma protease inhibitor
cocktail. The remaining cell pellets were immediately lysed in the same lysis
buffer without b-glycerophosphate, NaF, okadaic acid and phosphatase inhibitors.
Instead, 1 mM okadaic acid was subsequently added at specific time-points up to
130 min. Cell lysates were then centrifuged for 15 min at 21,0006g to pellet
insoluble debris. Cell lysates were kept at 4 ˚C throughout the whole experimental
period. The phosphorylation status of various mitotic proteins was analysed by
blotting on normal gels or gels containing 50 mM Phos-tag reagent (Wako
Chemicals GmbH, Neuss, Germany) charged with Mn2+ to better resolve
phosphorylated from non-phosphorylated proteins.
Phosphatase and kinase assays
Isolation of recombinant PP6 holoenzyme complexes and Aurora-A–TPX2
complexes were performed exactly as described previously (Zeng et al., 2010).
For p-nitrophenol phosphate (pNPP) phosphatase assays, purified PP6 complexes
were incubated with 5 mM pNPP in 50 mM Tris-HCl, pH 8.0, 0.5 mM MnCl2,
2 mM DTT for 60 mins at 30 ˚C. 50 ml reactions were stopped with 200 ml 0.5%
SDS and the absorption was read at 405 nm. Phosphatase assays using Aurora-A–
TPX2 complexes were performed exactly as described previously (Zeng et al.,
2010). For in vitro kinase assays a modification of a published method was used
(Preisinger et al., 2005). Control or Aurora-A complexes were isolated from
3615 cm dishes of HeLa Flp-in cells each, depleted for endogenous PPP6C using
PPP6C siRNA oligo 08 and expressing si08PPP6Cwt, si08PPP6CPD or
si08PPP6CH114Y. Immunoprecipitates were incubated in 50 mM Tris-HCl
pH 7.3, 50 mM KCl, 10 mM MgCl2, 20 mM b-glycerophosphate, 15 mM
EGTA, 100 mM ATP, 0.5 ml [c-32P]-ATP, 1 mg histone H3 substrate in 20 ml
final volume for 30 minutes at 30 ˚C. Reactions were analysed by SDS-PAGE and
autoradiography for histone H3 phosphorylation and blotting for Aurora-A levels.
Author contributions
U.G. and F.A.B. conceived and directed the study, and wrote the
manuscript. F.A.B. built homology models of PP6. D.H., R.N.B. and
F.A.B. developed and performed the phosphatase substrate screening
and mass spectrometry. K.Z., A.S. and U.G. carried out the cell
biological and biochemical analysis of PP6 function. A.S. and R.D.B.
performed the statistical analysis of genome instability.
Funding
This work was supported by Cancer Research UK [grant numbers
C20079/A9473to F.A.B., C24085/A8296 to U.G.].
Supplementary material available online at
http://jcs.biologists.org/lookup/suppl/doi:10.1242/jcs.128397/-/DC1
References
Bastos, R. N., Penate, X., Bates, M., Hammond, D. and Barr, F. A. (2012). CYK4
inhibits Rac1-dependent PAK1 and ARHGEF7 effector pathways during cytokinesis.
J. Cell Biol. 198, 865-880.
Bayliss, R., Sardon, T., Vernos, I. and Conti, E. (2003). Structural basis of Aurora-A
activation by TPX2 at the mitotic spindle. Mol. Cell 12, 851-862.
Bischoff, J. R., Anderson, L., Zhu, Y., Mossie, K., Ng, L., Souza, B., Schryver, B.,
Flanagan, P., Clairvoyant, F., Ginther, C. et al. (1998). A homologue of Drosophila
aurora kinase is oncogenic and amplified in human colorectal cancers. EMBO J. 17,
3052-3065.
Boveri, T. (2008). Concerning the origin of malignant tumours by Theodor Boveri.
Translated and annotated by Henry Harris. J. Cell Sci. 121 Suppl. 1, 1-84.
Cahill, D. P., Lengauer, C., Yu, J., Riggins, G. J., Willson, J. K., Markowitz, S. D.,
Kinzler, K. W. and Vogelstein, B. (1998). Mutations of mitotic checkpoint genes in
human cancers. Nature 392, 300-303.
Journal of Cell Science
3440
Journal of Cell Science 126 (15)
Crasta, K., Ganem, N. J., Dagher, R., Lantermann, A. B., Ivanova, E. V., Pan, Y.,
Nezi, L., Protopopov, A., Chowdhury, D. and Pellman, D. (2012). DNA breaks and
chromosome pulverization from errors in mitosis. Nature 482, 53-58.
Dou, Z., von Schubert, C., Körner, R., Santamaria, A., Elowe, S. and Nigg, E. A.
(2011). Quantitative mass spectrometry analysis reveals similar substrate consensus
motif for human Mps1 kinase and Plk1. PLoS ONE 6, e18793.
Douglas, P., Zhong, J., Ye, R., Moorhead, G. B., Xu, X. and Lees-Miller, S. P.
(2010). Protein phosphatase 6 interacts with the DNA-dependent protein kinase
catalytic subunit and dephosphorylates gamma-H2AX. Mol. Cell. Biol. 30, 13681381.
Dunsch, A. K., Hammond, D., Lloyd, J., Schermelleh, L., Gruneberg, U. and Barr,
F. A. (2012). Dynein light chain 1 and a spindle-associated adaptor promote dynein
asymmetry and spindle orientation. J. Cell Biol. 198, 1039-1054.
Forment, J. V., Kaidi, A. and Jackson, S. P. (2012). Chromothripsis and cancer:
causes and consequences of chromosome shattering. Nat. Rev. Cancer 12, 663-670.
Géraud, G., Laquerrière, F., Masson, C., Arnoult, J., Labidi, B. and HernandezVerdun, D. (1989). Three-dimensional organization of micronuclei induced by
colchicine in PtK1 cells. Exp. Cell Res. 181, 27-39.
Gordon, D. J., Resio, B. and Pellman, D. (2012). Causes and consequences of
aneuploidy in cancer. Nat. Rev. Genet. 13, 189-203.
Groves, M. R., Hanlon, N., Turowski, P., Hemmings, B. A. and Barford, D. (1999).
The structure of the protein phosphatase 2A PR65/A subunit reveals the conformation
of its 15 tandemly repeated HEAT motifs. Cell 96, 99-110.
Hodis, E., Watson, I. R., Kryukov, G. V., Arold, S. T., Imielinski, M., Theurillat,
J.-P., Nickerson, E., Auclair, D., Li, L., Place, C. et al. (2012). A landscape of
driver mutations in melanoma. Cell 150, 251-263.
Hoffelder, D. R., Luo, L., Burke, N. A., Watkins, S. C., Gollin, S. M. and Saunders,
W. S. (2004). Resolution of anaphase bridges in cancer cells. Chromosoma 112, 389397.
Holland, A. J. and Cleveland, D. W. (2012). Losing balance: the origin and impact of
aneuploidy in cancer. EMBO Rep. 13, 501-514.
Karthigeyan, D., Prasad, S. B., Shandilya, J., Agrawal, S. and Kundu, T. K. (2010).
Biology of Aurora A kinase: Implications in cancer manifestation and therapy. Med.
Res. Rev. 31, 757–793.
Kettenbach, A. N., Schweppe, D. K., Faherty, B. K., Pechenick, D., Pletnev, A. A.
and Gerber, S. A. (2011). Quantitative phosphoproteomics identifies substrates and
functional modules of Aurora and Polo-like kinase activities in mitotic cells. Sci.
Signal. 4, rs5.
Kohlbacher, O., Reinert, K., Gröpl, C., Lange, E., Pfeifer, N., Schulz-Trieglaff, O.
and Sturm, M. (2007). TOPP – the OpenMS proteomics pipeline. Bioinformatics 23,
e191-e197.
Kong, M., Ditsworth, D., Lindsten, T. and Thompson, C. B. (2009). Alpha4 is an
essential regulator of PP2A phosphatase activity. Mol. Cell 36, 51-60.
Kops, G. J., Weaver, B. A. and Cleveland, D. W. (2005). On the road to cancer:
aneuploidy and the mitotic checkpoint. Nat. Rev. Cancer 5, 773-785.
Krauthammer, M., Kong, Y., Ha, B. H., Evans, P., Bacchiocchi, A., McCusker,
J. P., Cheng, E., Davis, M. J., Goh, G., Choi, M. et al. (2012). Exome sequencing
identifies recurrent somatic RAC1 mutations in melanoma. Nat. Genet. 44, 10061014.
Lengauer, C., Kinzler, K. W. and Vogelstein, B. (1998). Genetic instabilities in human
cancers. Nature 396, 643-649.
Lens, S. M., Voest, E. E. and Medema, R. H. (2010). Shared and separate functions of
polo-like kinases and aurora kinases in cancer. Nat. Rev. Cancer 10, 825-841.
Lentini, L., Amato, A., Schillaci, T. and Di Leonardo, A. (2007). Simultaneous
Aurora-A/STK15 overexpression and centrosome amplification induce chromosomal
instability in tumour cells with a MIN phenotype. BMC Cancer 7, 212.
Macville, M., Schröck, E., Padilla-Nash, H., Keck, C., Ghadimi, B. M., Zimonjic, D.,
Popescu, N. and Ried, T. (1999). Comprehensive and definitive molecular
cytogenetic characterization of HeLa cells by spectral karyotyping. Cancer Res. 59,
141-150.
McGranahan, N., Burrell, R. A., Endesfelder, D., Novelli, M. R. and Swanton,
C. (2012). Cancer chromosomal instability: therapeutic and diagnostic challenges.
EMBO Rep. 13, 528-538.
Mi, J., Dziegielewski, J., Bolesta, E., Brautigan, D. L. and Larner, J. M. (2009).
Activation of DNA-PK by ionizing radiation is mediated by protein phosphatase 6.
PLoS ONE 4, e4395.
Mollinari, C., Kleman, J. P., Jiang, W., Schoehn, G., Hunter, T. and Margolis, R. L.
(2002). PRC1 is a microtubule binding and bundling protein essential to maintain the
mitotic spindle midzone. J. Cell Biol. 157, 1175-1186.
Olsen, J. V. and Macek, B. (2009). High accuracy mass spectrometry in large-scale
analysis of protein phosphorylation. Methods Mol. Biol. 492, 131-142.
Pettersen, E. F., Goddard, T. D., Huang, C. C., Couch, G. S., Greenblatt, D. M.,
Meng, E. C. and Ferrin, T. E. (2004). UCSF Chimera – a visualization system for
exploratory research and analysis. J. Comput. Chem. 25, 1605-1612.
Preisinger, C., Körner, R., Wind, M., Lehmann, W. D., Kopajtich, R. and Barr,
F. A. (2005). Plk1 docking to GRASP65 phosphorylated by Cdk1 suggests a
mechanism for Golgi checkpoint signalling. EMBO J. 24, 753-765.
Prickett, T. D. and Brautigan, D. L. (2006). The alpha4 regulatory subunit exerts
opposing allosteric effects on protein phosphatases PP6 and PP2A. J. Biol. Chem.
281, 30503-30511.
Rappsilber, J., Ishihama, Y. and Mann, M. (2003). Stop and go extraction tips for
matrix-assisted laser desorption/ionization, nanoelectrospray, and LC/MS sample
pretreatment in proteomics. Anal. Chem. 75, 663-670.
Rappsilber, J., Mann, M. and Ishihama, Y. (2007). Protocol for micro-purification,
enrichment, pre-fractionation and storage of peptides for proteomics using StageTips.
Nat. Protoc. 2, 1896-1906.
Rogakou, E. P., Pilch, D. R., Orr, A. H., Ivanova, V. S. and Bonner, W. M. (1998).
DNA double-stranded breaks induce histone H2AX phosphorylation on serine 139.
J. Biol. Chem. 273, 5858-5868.
Sen, S., Zhou, H. and White, R. A. (1997). A putative serine/threonine kinase encoding
gene BTAK on chromosome 20q13 is amplified and overexpressed in human breast
cancer cell lines. Oncogene 14, 2195-2200.
Shi, Y. (2009). Serine/threonine phosphatases: mechanism through structure. Cell 139,
468-484.
Shi, J., Blundell, T. L. and Mizuguchi, K. (2001). FUGUE: sequence-structure
homology recognition using environment-specific substitution tables and structuredependent gap penalties. J. Mol. Biol. 310, 243-257.
Stark, M. S., Woods, S. L., Gartside, M. G., Bonazzi, V. F., Dutton-Regester, K.,
Aoude, L. G., Chow, D., Sereduk, C., Niemi, N. M., Tang, N. et al. (2012).
Frequent somatic mutations in MAP3K5 and MAP3K9 in metastatic melanoma
identified by exome sequencing. Nat. Genet. 44, 165-169.
Stefansson, B. and Brautigan, D. L. (2006). Protein phosphatase 6 subunit with
conserved Sit4-associated protein domain targets IkappaBepsilon. J. Biol. Chem. 281,
22624-22634.
Stephens, P., Hunter, C., Bignell, G., Edkins, S., Davies, H., Teague, J., Stevens, C.,
O’Meara, S., Smith, R., Parker, A. et al. (2004). Lung cancer: intragenic ERBB2
kinase mutations in tumours. Nature 431, 525-526.
Stephens, P. J., Greenman, C. D., Fu, B., Yang, F., Bignell, G. R., Mudie, L. J.,
Pleasance, E. D., Lau, K. W., Beare, D., Stebbings, L. A. et al. (2011). Massive
genomic rearrangement acquired in a single catastrophic event during cancer
development. Cell 144, 27-40.
Sturm, M., Bertsch, A., Gröpl, C., Hildebrandt, A., Hussong, R., Lange, E., Pfeifer,
N., Schulz-Trieglaff, O., Zerck, A., Reinert, K. et al. (2008). OpenMS – an opensource software framework for mass spectrometry. BMC Bioinformatics 9, 163.
Thompson, S. L., Bakhoum, S. F. and Compton, D. A. (2010). Mechanisms of
chromosomal instability. Curr. Biol. 20, R285-R295.
van Haaften, G., Dalgliesh, G. L., Davies, H., Chen, L., Bignell, G., Greenman, C.,
Edkins, S., Hardy, C., O’Meara, S., Teague, J. et al. (2009). Somatic mutations of
the histone H3K27 demethylase gene UTX in human cancer. Nat. Genet. 41, 521-523.
Wiśniewski, J. R., Zougman, A., Nagaraj, N. and Mann, M. (2009). Universal sample
preparation method for proteome analysis. Nat. Methods 6, 359-362.
Xing, Y., Xu, Y., Chen, Y., Jeffrey, P. D., Chao, Y., Lin, Z., Li, Z., Strack, S., Stock,
J. B. and Shi, Y. (2006). Structure of protein phosphatase 2A core enzyme bound to
tumor-inducing toxins. Cell 127, 341-353.
Yang, Z., Lasker, K., Schneidman-Duhovny, D., Webb, B., Huang, C. C., Pettersen,
E. F., Goddard, T. D., Meng, E. C., Sali, A. and Ferrin, T. E. (2012). UCSF
Chimera, MODELLER, and IMP: an integrated modeling system. J. Struct. Biol. 179,
269-278.
Zeng, K., Bastos, R. N., Barr, F. A. and Gruneberg, U. (2010). Protein phosphatase 6
regulates mitotic spindle formation by controlling the T-loop phosphorylation state of
Aurora A bound to its activator TPX2. J. Cell Biol. 191, 1315-1332.
Zhou, H., Kuang, J., Zhong, L., Kuo, W. L., Gray, J. W., Sahin, A., Brinkley, B. R.
and Sen, S. (1998). Tumour amplified kinase STK15/BTAK induces centrosome
amplification, aneuploidy and transformation. Nat. Genet. 20, 189-193.