Download Important roles for novel protein phosphatases dephosphorylating

Biochemical Society Transactions
~
~~
~~~
Important roles for novel protein phosphatases dephosphorylating serine and
threonine residues
884
Patricia T. W. Cohen
Medical Research Council Protein Phosphorylation Unit, Department of Biochemistry, The University of Dundee,
Dundee DDI 4HN, U.K.
Introduction
Phosphorylation of proteins on serine and
threonine residues is a key mechanism in the
regulation of a wide variety of cellular functions,
including cell division, hormonal action and signalling from the plasma membrane to the nucleus
[ 1-01. The revsibility of protein phosphorylation
requires that dephosphorylation as well as phosphorylation processes must be elucidated to understand the regulation of cellular functions by this
mechanism fully. Four major protein phosphatase
(PP) catalytic subunits were identified originally by
enzymic criteria in mammalian cells, and termed
PP1, PP2A, PP2H and PP2C [4] and isoforms of
each of these phosphatases were identified subsequently by molecular cloning. These studies also
showed that PP1, PP2A and PP2H are related in
structure, while PP2C does not contain the regions
predicted to be essential for the catalytic activity of
the PPl/PP2A/PP2H family [ S , 61.
Size of the PP I /PPZA/PPZB family
Over the last few years, several novel protein phosphatases belonging to the PP 1/PP2A/PP2H family
have been identified from their cDNAs. These
include PPX [7] in mammals, PPV [6] and PPY [S]
in Drosophila melanogaster, and PPZl [6, 9, 101 and
PPZ2 191 in Saccharomyces cerevisiae. Others have
been identified by the analysis of a mutant of the
protein SIT4 in S. cerevkiae [ 111 and a mutant of the
protein rdgC in DrosophiZa [ 121. PCR studies were
initiated [ 131 to estimate the number of different
catalytic subunits in this family of protein phosphatases. Oligonucleotides constructed to sequences
conserved between mammalian and bacteriophage
I protein phosphatases [ S ] were used with genomic
DNA from S. cerevisiae, D. melanogaster and Homo
sapiens. Sequence determination of the PCR fragments identified both known and novel protein
phosphatases in S cerevisiae and Drosophila and
novel protein phosphatase genes (or pseudogenes)
in man. In Drosophila, seven novel and two protein
phosphatase genes that were already known were
identified by this procedure. The results suggest
Abbreviation used: I T , protein phosphatase; MAP
kinase, mitogen-activated protein kinase.
Volume 21
that, if we detected all eight protein phosphatase
genes that were known in Drosophila at that time,
then we should have identified 28 novel protein
phosphatase genes, bringing the total number of
protein phosphatase genes in this species to 36.
Since the human genome may encode tenfold the
number of proteins encoded by the Drosophila
genome and the complexity of protein phosphorylation undoubtedly increases in higher eukaryotes,
there may be more than 300 genes for the catalytic
subunits of this family of protein phosphatases in
mammals. PP1, PP2A and PP2H catalytic subunits
all bind at least two different regulatory subunits,
indicating that >900 genes or approx. 1% of
human genes may encode this family of protein
phosphatases.
Protein phosphatases in the P P l and PP2A
subfamilies identified from full-length cI )NAs or
genes are shown in Table 1 for the three species:
man, Drosophila and S. cerevisiae. T o determine
whether the novel protein phosphatases play distinct roles from those of PP1 and PP2A, we have
used several approaches to investigate their functions. The results point to specific and key roles for
mammalian PP4 (PPX), Drosophila PPV and S
cerevisiae PPZ 1 and PPZ2.
PP4 (initially termed PPX) [ 141
Rabbit PP4 was expressed from its cDNA in the
baculovirus/insect cell system. Although mostly
insoluble, the 10% that was soluble and active was
purified partially to remove contaminating insect
cell protein phosphatases. Examination of its substrate specificity showed that PP4 dephosphorylates
serine and threonine residues. As with PP2A, PP4
preferentially dephosphorylates the a subunit
rather than the p subunit of phosphorylase kinase,
and is unaffected by inhibitor-1 and by inhibitor-2.
PP4 and PP2A were also inhibited similarly by the
tumour promoter okadaic acid and the liver toxin
microcystin (the ICirrvalues for okadaic acid were
0.07 nM for PP2A and 0.2 nM for PP4, while for
microcystin they were 0.2 nM and 0.8 nM respectively). However, when PP4 and PP2A catalytic subunits were matched for phosphorylase phosphatase
activity, PP4 had a similar activity towards the
synthetic peptide RRATj’P-VA, but was less active
Signalling from the Plasma M e m b r a n e to the Nucleus
Table I
Protein
phosphatases in the
subfamilies
PPI
and
PPZA
The novel enzymes in each family are indicated. References for
the sequences of PPI and PPZA isoforms are given in [ 131. References for novel phosphatase sequences are PPY [8]. PPZ I [6,
10, 241 PPZ2 [9. 291, PPQ 1291, mammalian PP4 [ 141, Drosophria
PP4 [30],PPV [ 181, SIT4 [ I I ] and PPG [3 I ]
Mammals
Drosophilo
S
PP I a
PPlg
PPI y
PPI 878
PPI 96A
PPI 13C
PPI 9c
PPY
PP I (DIS2)
cerevisioe
PP I subfamily
Established
Novel
PPZ I
PPZ2
PPQ
PPZA subfamily
Established
PP2Aa
PPZAB
PPZA
Novel
PP4 (PPX)
PP4 (PPX)
PPV
PPH2 I
PPH22
PPH3
SIT4
PPG
than ”2.4
towards all other substrates tested
(casein, histone EI 1, caldesmon. €IMG-I). The
I’P2A:PP4 activity ratio varied from 1 : l to 13:l
with different substrates, indicating that the specificities of the two enzymes are distinct. Furthermore,
despite 65% amino acid sequence identity to PP2A,
I’P4 did not bind the 65 kDa regulatory subunit of
I’P2A.
Evidence pointing towards a specific function
for PP4 came from immunofluorescent localization
studies using biotinylated, affinity-purified antibodies against the PP4 protein, coupled to an
avidin-fluorescein
stain. These experiments
demonstrated that although PP4 was present in the
cytoplasm and more strongly in the nucleus of
human cells, it localized intensely to the centrosomes. The localization was judged to be specific
since it could be blocked by pre-incubation of the
antibody with excess PP4. An identical localization
pattern was obtained using an antibody raised
against a synthetic peptide corresponding to amino
acids 287-305 of PP4.
The centrosome is one of the microtubuleorganizing centres of the interphase cell, and at
mitosis it duplicates to form the spindle-pole bodies.
Localization of PP4 was observed at the centrosomes of interphase cells and at all stages of mitosis
except telophase. Therefore, at telophase, either the
PP4 epitope must have an altered conformation or
PP4 must be released from the centrosomes. Since
telophase is the point at which the mitotic spindle
disappears and the interphase network of microtubules reforms, PP4 may be involved in this
process. Examination of the localization of PP4 at
higher magnification revealed that it was absent
from the middle of the centrosome, but it co-localized with antibodies that are known to detect the
pericentriolar material, the region thought to be
responsible for initiating the growth of microtubules. Several studies have shown that a protein
phosphatase sensitive to nanomolar concentrations
of okadaic acid is required for microtubule nucleation, and it has been assumed to be PP2A [ 15, 161.
However, affinity-purified antibodies to PPZA localized predominantly to the cytoplasm, and were not
evident at the centrosomes in interphase or in dividing cells [ 141. It can be inferred therefore that PP4,
rather than PP2A, may regulate microtubule
nucleation. The signal(s) that regulate PP4 activity
is not known, but one possibility is the cdc2-cyclin
A complex, which has recently been shown to
initiate microtubule growth in vitro [ 171.
Drosophilu PPV
PPV [18] was identified from a cDNA in a Drosophila head library, but subsequent studies showed
that PPV mRNA was only expressed at very low
levels in adult tissues, while it was abundantly
expressed in the early embryo. In sztu hybridization studies on Drosophih embryos around
the time of cell formation (nuclear-division cycle
14) showed that PPV mRNA was localized predominantly at the periphery of the embryo, in
the region undergoing cellularization. The PPV
protein, detected with affinity-purified antibodies
against a PPV-specific peptide also localized to the
cytoplasm of cells at the cortex. No PPV could be
detected in the nucleus. The substantial rise in PPV
occurring transiently over the cellularization period
is the time at which zygotic transcription rises and
the nuclear-division cycle changes from one in
which mitosis rapidly follows DNA synthesis, to the
more characteristic cell cycle of eukaryotes that
includes G1 and G2 phases, suggesting that PPV
may be involved in one or more of these processes.
I993
Biochemical Society Transactions
Since the amino acid sequence of PPV is most
similar to that of SIT4 in S. cermzi-iae (65% identity),
it was pertinent to test whether SIT4 and PPV
might be homologues. SIT4 regulates the transcription of a number of genes, including G1 cyclins
[ 191, and, in certain genetic backgrounds, an allele
of SIT$ sit4-1 02, causes a temperature-sensitive
cell-cycle arrest in late G I [20]. Transformation of
this mutant with PPV cDNA, placed under the
control of the alcohol dehydrogenase promoter,
allowed growth at the restrictive temperature. In
contrast, cDNA for other protein phosphatases
under the same promoter did not rescue sit4-102.
Proof that PPV could perform all the in vivo functions of SIT4 came from the demonstration that
PPV cDNA, when under the SIT4 promoter, could
completely replace the S I T 4 gene in S. cerevisiae.
The large increase in PPV occurring at the
time when zygotic transcription increases, and the
demonstration that PPV is the functional homologue of SIT4, indicate a role for PPV in the regulation of transcription. However, there is, as yet, no
evidence for a rise in the transcription of any G1
cyclin in the early Drosophila embryo. SIT4 has also
been shown to be essential for bud emergence in S.
cerevisiae, and a parallel role for PPV in Drosophila
could be in the formation of cells in the DrosophiZu
embryo at nuclear-division cycle 14. The mRNA
for the genes serendipi& and nullo, which are
required for cellularization, rise transiently [2 11, and
it may be that their transcription is under the
control of PPV.
Recently, ppe 1, the probable Schkosuccharomyces pombe homologue of PPV and SIT4, has been
identified [22]. Deletion of the ppel gene causes
arrest of cells at G2, rather than G1 as seen in S.
cerevkzize. Other workers identified ppel as a
suppressor of the mutant piml, in which mitosis is
uncoupled from DNA synthesis [23]. These
analyses implicate ppel in a process in the G2-M
transition that may couple DNA synthesis and
mitosis in S. pombe. Therefore, it may be that the
PPV/SIT4/ppe 1 phosphatase is involved both in
the G1-S and G2-M check-points.
A role in the cell cycle suggests that the
activity of PPV may be tightly regulated. Comparison of the PPV and SIT4 sequences revealed amino
acids that were identical in these two protein phosphatases, but were different from those in other
known protein phosphatases. The only section
where these were more than just isolated amino
acids was in the N-terminal region. Therefore, a
chimeric construct (designated pADH V: 13V) was
made, in which the N-terminal 5 5 amino acids of
Volume 21
PPV were attached to the catalytic domain of a
Drosophila isoform of PP1 (PP1-13C) and placed
under the control of the alcohol dehydrogenase
promoter. Rescue of sit4-102 was achieved by
transformation with pADH V:13C but not by
pADH 13C:V in which the N-terminus of PP1 was
attached to the catalytic domain of PPV [8]. It is
remarkable that the N-terminal region of PPV (a
PP2A-like phosphatase) could effect the rescue of
the SIT4 mutant when attached to a PP1 catalytic
region. These results identify the N-terminal 5 5
amino acids of PPV as being essential for PPV
function. This domain is unlikely to target PPV to a
particular location within the cell, since the
immunofluorescent localization studies described
above showed a diffuse cytoplasmic staining for
PPV. A more plausible explanation is that the
N-terminal domain is involved in binding to regulatory subunits, which may influence the specificity of
the catalytic domain and/or control its activity in
response to incoming signals.
S. cerevisiae P P Z l and PPZ2 [24]
Although PPZl and PPZ2 were originally identified
in a commercial rabbit brain cDNA library
(Clontech), subsequent analyses demonstrated that
they did not encode brain phosphatases but novel S.
cereviszize enzymes [9]! Both phosphatases contain a
catalytic domain that is preceded by a long Nterminal domain, rich in serine and asparagine in
PPZ1, and serine and arginine in PPZ2. Although
the N-terminal domains are only 43% identical to
each other, the catalytic domains are 93% identical,
indicating that these two phosphatases are likely to
have similar or overlapping functions. To determine
their cellular roles, deletions were made in the PPZl
and PPZ2 genes in regions encoding the conserved
motifs predicted to be essential for catalytic activity.
Mutant strains disrupted in either or in both genes,
showed an enlarged cell size when examined under
the microscope in stationary phase. Analysis by
light scatter measurements in a fluorescence-activated cell sorter confirmed this size difference and
showed that the mutant ppzlppz2 cells could be
restored to normal or near normal size by the inclusion of 1 M sorbitol in the growth medium. The
integrity of the cell wall and plasma membrane
were assessed by measuring the release of RNA
from cells labelled with ['Hluridine. Although little
difference between wild-type and mutant cells was
observed at 28"C, cell lysis was increased in ppzl
and ppzlppz2 cells at 37°C compared with wildtype cells. In addition, lysis was slightly increased in
ppz2 cells and markedly elevated in ppzl and
Signalling from the Plasma Membrane t o the Nucleus
ppzlppz2 strains in the presence of caffeine at 28°C.
1 M sorbital partially suppressed the lytic phenotY Pe.
Several protein kinase mutants give rise to a
lytic phenotype. Conditional mutants of the protein
PKCl (protein kinase C1) display a cell-cyclespecific lysis defect [25,261 that can be suppressed
by four different protein kinases. Genetic analyses
demonstrated their order of action to be
PKCl +BCK1 -MKKl/MKKZ-MPKl
where
MKKl and MKK2 are related to mammalian
mitogen-activated protein (MAP) kinase kinases
and MPKl to mammalian MAP kinase. Strains
carrying a deletion of RCK1, a double deletion of
MKKl and MKK2 or a deletion of MPKl all show
a temperature-sensitive cell lysis defect that can be
suppressed by osmotic stabilizers [27]. Mutants of
two other protein kinases, PRS2 and HOG1, which
are involved in regulating the accumulation of intracellular glycerol in response to extracellular
osmolarity changes, also show cell lysis defects, but,
in contrast to PPZ disruptants, their lytic phenotypes are not suppressed by sorbitol [28].PPZl and
PPZ2 are therefore more likely to function in the
PKCl pathway than in the PHS2 pathway. Since the
phenotype of increased susceptibility to cell lysis of
the PPZ mutants is similar to that of BCKl (bypass
of protein kinase C), MKKlIMKK2 and MPKl
deletion mutants, it appears more likely that PPZ
might act in concert with this kinase cascade rather
than reverse it. Hy analogy to the regulation of
glycogen metabolism by insulin in mammalian
skeletal muscle, where PP1 is activated by a protein
kinase that is linked to the MAP kinase cascade,
PPZl/PPZ2 might be phosphorylated and
activated by a component of the yeast PKCl /MAP
kinase cascade. T h e serine rich N-terminal domains
of PPZl and PPZ2 contain many phosphorylation
sites for MAP kinase, protein kinase C and other
protein kinases [24].
The exact mechanism by which PPZl and
PPZ2 maintain cell integrity is not known, but it
could involve cell wall construction, organization of
the cytoskeleton and/or osmosensing. Several lines
of evidence suggest that PPZl and PPZ2 may be
membrane-bound. Both N-termini possess good
consensus sequences for myristoylation, which
would favour a membrane location. Two short
sequences present in the N-terminal domains of
both PPZl and PPZ2 are similar to short sections
of a yeast osmotic growth protein and to
desmoplakin respectively, both of which are located
at or near the cell membrane. It is therefore
plausible that PPZl and PPZ2 may dephosphory-
late proteins at or near the surface of the cell membrane [24]. However, disruption of PPZl and PPZ2
in a certain genetic background can affect cell
growth (M. X. Chen and P. T. W. Cohen,
unpublished work) indicating that PPZl and PPZ2
could carry out other functions, and perhaps also
prevent cell lysis by regulating the transcription of
the enzymes involved in maintenance of cell
integrity.
In conclusion, PCR studies indicate that the
PPl/PPZA/PPZB family is much larger than previously supposed, and analysis of phosphatases
related to PP1 and PPZA have revealed that four of
these novel enzymes PP4, PPV, PPZl and PPZ2
play key roles in cellular regulation that are distinct
from those of PP1 and PP2A.
This work was supported by the Medical Research
Council, London and by the Cancer Research Campaign.
1. Cohen, P. (1992) Trends Biochem. Sci. 17,408-413
2. Meek, D. W. and Street, A. J. (1992) Biochem. J. 287,
1-15
3. Yanagida, M., Kinoshita, N., Stone, E. M. and
Yamano, H. (1992) in Regulation of the Eukaryotic
Cell Cycle (J. Marsh, ed.), pp. 130- 146 John Wiley &
Sons, Chichester
4. Cohen, P. (1989) Annu. Rev. Bichem. 58,453-508
5. Cohen, P. T. W. and Cohen, P. (1989) Riochem. J.
260.93 1-934
6. Cohen, P. T. W., Hrewis, N. D., Hughes, V. and
Mann, D. J. (1990) FEBS Lett. 268, 355-359
7. da Cruz e Silva, 0. H., da Cruz e Silva, E. F. and
Cohen, P. T. W. ( 1 988) FEBS Lett. 242, 106- 110
8. Dombradi, V., Axton, J. M., Glover, Ll. M. and Cohen,
P. T. W. (1989) FEBS Lett. 247, 391-395
9. da Cruz e Silva, E. F., Hughes, V., McDonald, P.,
Stark, M. J. and Cohen, P. T. W. (1991) Hiochim.
Biophys. Acta 1089,269-272
10. Posas, F., Casamayor, A., Morral, N. and Ariiio, J.
(1992) J. Biol. Chem. 267, 1 1734- 11740
11. Arndt, K. T., Styles, C. A. and Fink, G. R. (1989) Cell
56,527-537
12. Steele, F. R., Washburn, T., Kieger, R. and OTousa,
J. E. (1992) Cell 69,669-676
13. Chen, M. X., Chen, Y. H. and Cohen, P. T. W. (1992)
FEBS Lett. 306, 54-58
14. Rrewis, N. D., Street, A. J., Prescott, A. R. and Cohen,
P. T. W. (1993) EMBO J. 12,987-996
5. I’icard, A,, Capony, J. P., Brautigan, D. I,. and Dorke,
M. (1989)J. Cell Biol. 109, 3347-3354
6. Rime, H., Huchon, D.. Jessus, C., Goris, J., Merlevede,
W. and Ozon, K. (1990) Cell Differ. Dev. 29, 47-58
7. Buendia, B., Draetta, G. and Karsenti, E. (1 992) J. Cell
Riol. 116, 1431-1442
8. Mann, D.
Dombradi, V. and Cohen. P. T. W.
(1993) EMBO J., in the press
J.?
I993
887
Biochemical Society Transactions
888
19. Fernandez-Sarahia, M. J., Sutton, A., Zhong, T. and
Arndt, K. T. ( 1992) Genes Dev. 6 , 2 4 17-2428
20. Sutton, A., Immanuel, I>. and Arndt. K. T. (1991)
Mol. Cell. Hiol. 11,2133-2148
21. Kose, I,. S. and Wiesehaus. E. (1992) Genes Dev. 6,
1255-1268
22. Shimanuki, M., Kinshita, N., Ohkura, H., Yoshida, T.,
Toda, T. and Yanagida, M. (1993) Mol. Biol. Cell 4,
303-3 13
23. Matsumoto, T. and Beach, I). (1993) Mol. Cell. Hiol.
4,337-345
24. Hughes, V., Muller, A,, Stark, M. J. R. and Cohen.
1'. T. W. (1993) Eur. J. Hiochem. 216,2h0-270
25. Levin, 1). E. and Hartlett-Heubusch, E. (1992) J. Cell
Hiol. 1 16, 122 1- 1229
20. I'aravicini, G., Cooper, M.. Friedeli, I+ Smith, 1). J.,
Carpenter, J.-I,., Klig, 1,. S.and I'ayton, M. A. (1002)
Mol. Cell Hiol. 12, 4896-4005
27. Errede, H. and Levin, L). E. (1 993) Curr. Opin. Cell
Hiol. 5, 254-260
28. Hrewster, J. I,., de Valoir, T., Dwyer, N. l).,Winter, E.
and Gustin, M. C. ( I 993) Science 259, 1760- 1762
29. Chen, M. X., Chen, Y. H. and Cohen, 1'. T. W . (1003)
Eur. J. Hiochem., in the press
30. Hrewis, N. 1). (1992) 1'h.D. thesis, llundee llniversity
31. I'osas, F.?Clotet, J., Muns, M. T., Corominas, J. A.,
Casamayor, A. and Ariiio, J. (1003) J. Hiol. Cheni.
268, 1349-1 354
Received 4 August 1993
Role of p2 I ras in growth factor signal transduction
Boudewijn M.Th. Burgering, Gijsbertus J. Pronk, Jan Paul Mederna, Loesje van der Voorn. Alida M. M. de
Vries Smits, Pascale C. van Weeren and Johannes L. Bos
Laboratory of Physiological Chemistry, Utrecht University, 352 I GG Utrecht, The Netherlands
Introduction
r .
I he family of human rus proto-oncogenes consists
of three closely related genes: 11-rus, K-rus and
N-rus [ 1 I. They encode homologous proteins of
molecular mass 21 kI)a (p21r"')9located at the inner
side of the plasma membrane. The rus proto-oncogenes acquire their oncogenic potential when point
mutations occur at amino acid positions 12, 13 or
61, resulting in a single amino acid substitution in
the protein [ 21. T o understand the consequences of
these oncogenic mutations, it is crucial to understand the role that normal ~ 2 1 " ' fulfils in cell
growth and differentiation.
Regulation of p2 I ras activity
The notion that p21"' can bind the guanine nucleotides G T P and CLIP is central to our understanding
of the function of p2lras in growth factor-induced
signal transduction. Through sequential binding of
(;I>P and GTP, p21"' can function as a molecular
switch in signal transduction [ 3, 41. Microinjection
of p21"' proteins bound to CLIP or G T P has
demonstrated that p2 1'"'-GI>P is biologically
inactive. and hence represents the 'off state,
Abbreviations used: EGF. epidermal growth factor;
EKK2, extracellular-signal-regulated kinase 2; GAP,
WI'ase-activating protein, GKF, guanine-nucleotide
releasing factor; MAP kinase. mitogen-activated protein
kinase; I'DGF, platelet-derived growth factor; I'KC,
protein kinase C; I'MA, phorbol 12-myristate 13-acetate.
Volume 2 I
whereas p2 1 "'--GTP is biologically active, and thus
represents the 'on' state of this switch. Two classes
of accessory proteins are involved in the regulation
of GDP/GTP binding to p21"". Firstly, the dissociation of GDP from p21"' occurs very slowly,
and is catalysed by the interaction of p21"' with
guanine-nucleotide releasing factors ((;KFs). Some
of these proteins showing specificity towards p2 1
have been identified. Of these, p 140';"'" appears t o
be exclusively expressed in brain tissue [ 51,
whereas two others (mSos1 and mSos2) appear t o
be ubiquitously expressed [6]. Once GTP is bound
to p21"', a transient interaction is thought to occur
between p2 1 ""-GTP and its effector molecule. The
effector-binding region of p2 1"" has been determined by genetic methods and spans roughly the
region amino acid 30-40 [7]. Although the identity
of the p2 1"I' effector molecule is still elusive, recent
results indicate that this may be the raf-1 kinase.
In vitro association of raf-1 kinase with p21""GTP has been demonstrated [XI, and by using the
yeast two-hybrid system it has been shown that this
interaction may also occur in azvo 101. Furthermore,
genetic evidence suggests that raf- 1 kinase acts
downstream of p21"" [lo].
Eventually, after activation of downstream signalling, GTP on p2 1rr'' will be hydrolysed. The
protein p2 1'",' itself displays a low GTPase activity,
which is enhanced upon interaction with GI'Paseactivating proteins (GAPS), such as p120';'"' and
neurofibromin. It is still not clear whether GtZI's, as
"I'