Download Cooperation between the RING+B1-B2 and coiled-coil

Oncogene (1998) 16, 2905 ± 2913
 1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00
http://www.stockton-press.co.uk/onc
Cooperation between the RING+B1-B2 and coiled-coil domains of PML is
necessary for its e€ects on cell survival
Marta Fagioli1, Myriam Alcalay2, Lucia Tomassoni2, Pier Francesco Ferrucci2 ,
Amedea Mencarelli1, Daniela Riganelli1, Francesco Grignani1, Tullio Pozzan3, Ildo Nicoletti1,
Fausto Grignani1 and Pier Giuseppe Pelicci1,2
1
Istituto di Medicina Intema e Scienze Oncologiche, UniversitaÁ degli Studi di Perugia, Policlinico Monteluce, 06100, Perugia;
European Institute of Oncology, Department of Experimental Oncology, 20141, Milan; 3Department of Biomedical Sciences and
CNR Center for the Study of Mitochondrial Physiology, University of Padova, Via Trieste 75, 35121 Padova, Italy
2
PML/RARa is the abnormal protein product of the Acute
Promyelocytic Leukemia-speci®c 15;17 translocation.
Both the PML and RARa components are required for
the PML/RARa biological activities, namely its capacity
to block di€erentiation and to increase survival of
haematopoietic precursors. The physiological role of
PML and its contribution to the function of the fusion
protein are unknown. PML localizes to the cytoplasm and
within speci®c nuclear bodies (NBs). In vitro, overexpression of PML correlates with suppression of cell
transformation. The PML aminoterminal portion retained
within the PML/RARa protein contains the RING ®nger,
two newly de®ned cystein/histidine-rich motifs called Bboxes (B1 and B2) and a coiled-coil region. We report here
that PML has a growth suppressive activity in all the cell
lines tested, regardless of their transformed phenotype, and
that the cellular basis for the PML growth suppression is
induction of apoptotic cell death. Analysis of various
nuclear and cytoplasmic PML isoforms showed that the
PML growth suppressive activity correlates with its
nuclear localization. Analysis of the localization and
growth suppressive activity demonstrated that: (i) the
Ring+B1-B2 and coiled-coil regions are both indispensable and sucient to target PML to the NBs; (ii)
individual deletions of the various PML domains have no
e€ect on its growth suppressor activity; (iii) the Ring+B1B2 region exerts a partial growth suppressor activity but
its fusion with the coiled-coil region is sucient to
recapitulate the suppressive function of wild type PML.
These results indicate that PML is involved in cell survival
regulation and that the PML component of the fusion
protein (Ring+B1-B2 and coiled-coil regions) retains
intact biological activity, thereby suggesting that the
e€ects of PML/RARa on survival derive from the
activation of the incorporated PML sequence.
Keywords: PML; Ring; B-boxes; coiled-coil; apoptotic
cell death
Introduction
The PML gene was identi®ed by cloning the
breakpoint sites of the 15;17 chromosome translocaCorrespondence: PG Pelicci, Department of Experimental Oncology,
European Institute of Oncology Via Ripamonti 435, 20141 Milan,
Italy
Received 17 July 1997; revised 29 December 1997; accepted 5 January
1998
tion associated with acute promyelocytic leukaemia
(Alcalay et al., 1991; Borrow et al., 1990; de The et
al., 1990). It is localized on chromosome 15 and as
a consequence of the translocation is fused with the
retinoic acid receptor a gene (RARa) on 17. The
PML/RARa fusion gene is transcriptionally active in
all APL cases and encodes for a PML/RARa fusion
protein (de The et al., 1991; Kakizuka et al., 1991;
Pandol® et al., 1991). PML/RARa transgenic mice
develop haematopoietic alterations with typical APL
phenotypic features (Grisolano et al., 1996; Brown et
al., 1997) and expression of PML/RARa in chicken
haematopoietic progenitors induces acute leukaemia
(Altabef et al., 1996). Ectopic expression of PML/
RARa into human haematopoietic cell lines blocks
di€erentiation induced by various stimuli and prolongs cell survival by inhibiting apoptosis (Grignani
et al., 1993; Rogaia et al., 1995), suggesting that
these two biological activities underlie the leukaemogenic potential of PML/RARa in vivo. Extensive
structure-function analysis has revealed that the
integrity of both PML and RARa domains is
crucial for the di€erentiation block (Grignani et
al., 1996) and the e€ect on cell survival (Ferrucci et
al., 1997).
The molecular mechanisms through which PML/
RARa exerts its biological activities are poorly
understood. It has been proposed that the fusion
protein has a dominant negative e€ect on PML and
RARa endogenous pathways, since, in APL cells, it
localizes in nuclear microspeckled structures and
provokes delocalization of the endogenous PML and
RXR, a RARa co-factor (Dyck et al., 1994; Flenghi et
al., 1995; Koken et al., 1994; Weis et al., 1994).
Alternatively, PML/RARa biological activities might
be the consequence of activation of both PML and
RARa functional domains within the context of the
fusion protein (Grignani et al., 1996).
The physiological function and biochemical activities
of the PML protein are unknown. PML belongs to a
family of proteins which share an aminoterminal
tripartite domain characterized by the C3HC4 zinc
®nger motif, named Ring ®nger, one or two additional
cysteine-rich regions (B-boxes) and a coiled-coil region
(Freemont et al., 1991; Lovering et al., 1993; Reddy et
al., 1992). Many members of this family are
transcriptional regulators, some display DNA binding
function and two (RFP, T18/TIF1) are, like PML,
involved in genetic recombinations that lead to the
formation of transforming fusion proteins (RFP/ret
and T18/B-raf, respectively) (Gandini and Pandol®,
PML growth suppressive activity
M Fagioli et al
2906
1994; Le Douarin et al., 1995; Tkahashi et al., 1988).
The RING ®nger motif, the B-boxes and the coiled-coil
regions are retained in the PML/RARa protein.
PML is a nuclear phosphoprotein (Chang et al.,
1995) and is localized to speci®c matrix-associated
subdomains called nuclear bodies (PML-NB), also
known as POD (PML oncogenic domains), ND10
(Nuclear Domain 10) or Kr bodies (Dyck et al., 1994;
Everett and Maul, 1994; Weis et al., 1994). Several
proteins have been identi®ed that colocalize with PML
within NBs: the SP100 protein, originally identi®ed as
an autoantigen in patients with primary biliary
cirrhosis (Szostecki et al., 1990); two proteins of
unknown function of 65 kDa (Epstein, 1984) and
55 kDa (NDP55) (Ascoli and Maul, 1991); NP52
(Korioth et al., 1995); the PIC-1 PML interacting
protein (Boddy et al., 1996); the retinoblastoma protein
(Alcalay et al., 1998) and Int-6 (Desbois et al., 1996).
The PML-NBs are thought to be critical targets during
viral infection. In fact, a drastic redistribution of the
PML-NBs into ®brous structures has been shown to
occur during infection by DNA viruses such as Herpes
Simplex and Adenoviruses (Carvalho et al., 1995;
Doucas et al., 1996; Everett and Maul, 1994; PuvionDutilleul et al., 1995) together with an increase in size
and number after treatment of cells with interferon
(Chelbi-Alix et al., 1995; Lavau et al., 1995; Stadler et
al., 1995).
PML overexpression suppresses the transformed
phenotype in some experimental systems. It decreases
the capacity of activated (neu) or cooperative
oncogenes (Ha-Ras plus myc or mutated p53) to
transform NIH3T3 cells and rat embryo ®broblasts,
respectively (Liu et al., 1995; Mu et al., 1994), but has
no e€ect on Ras transformation in the same cell types.
As the localization of NBs is lost in various PML-non
functional mutants, the anti-tumor e€ect seems to
depend on the integrity of the NBs (Le et al., 1996).
Based on these data, it has been suggested that, in APL
cells, the PML growth suppressive function is lost as a
consequence of the fusion with RARa and that the
resulting PML/RARa protein dominates over the
endogenous PML by provoking its delocalization
from the NBs (Le et al., 1996). However, whether the
ability of PML to suppress transformation in certain
experimental systems re¯ects a more general function
of growth suppression and whether and how this
function is lost in APL remain to be established.
We report here that PML suppresses cell growth in
all cell lines tested regardless of their transformed
phenotype; that the mechanism of action is due to an
induction of cellular death and that it depends on the
cooperation of the PML domains retained by the
PML/RARa fusion protein. These results have
important implications for the physiological role of
PML and the leukaemic activity of PML/RARa.
Results
PML overexpression suppresses growth in a variety of
cell lines
To investigate the potential of PML to suppress cell
growth we tested the capacity of several cell lines to
form G418-resistant colonies in the presence of PML
overexpression (colony formation assay). For this
purpose both transformed cell lines (C33a and HeLa
cervix carcinoma; SAOS-2 and U20S osteosarcoma)
and normal ®broblasts (NIH3T3 and primary mouse
®broblasts) were used. All cell lines were transfected
with a PML expression vector (PML3-pCDNA3; see
methods) or the empty pCDNA3 expression vector
and, after 15 days of G418 selection, stained with
crystal violet and evaluated for the number of
resistant colonies. There was a dramatic drop in the
capacity of all the cell lines to form colonies in the
presence of PML overexpression (Figure 1). The level
of PML expression 48 h after transfection was high
(not shown) but became very low or absent after
selection, as evaluated by Western blotting of the rare
G418-resistant colonies (not shown). These results
demonstrates that PML has growth suppressive
activity in all the cell lines tested regardless of their
transformed phenotype. The antiproliferative e€ect
appeared to be both Rb and p53 independent, since
the C33a and HeLa cell lines, which both carry
genetic and/or functional inactivation of p53 and Rb,
were sensitive to the growth-suppressive action of
PML3.
Inducible expression of PML3 in U937 cells induces cell
death
To determine the mechanism through which PML
exerts its growth suppressive function, PML was
expressed under the control of the metallothionine
(MT) zinc (Zn) inducible promoter (pMTPML3) in a
variety of cell lines. However, inducible PML
expression was obtained only in the U937 cell line.
Several clones were derived from electroporated U937
cells and analysed by Western blotting for PML
expression. Clone MTP#F1 was selected because the
exogenous PML protein was not detectable prior to
Zn induction, while high levels were documented after
Zn (see panel inside Figure 2a). The e€ect of PML
overexpression on cell proliferation was evaluated by
comparing the growth potential of one U937 cell clone
carrying the MT empty expression vector (MT#8) and
MTP#F1 cells. The growth rate of Zn-untreated
Figure 1 E€ects of PML expression on growth of a variety of
human cell lines. Triplicates of cells from the indicated cell lines
were transfected by calcium phosphate precipitation procedure
with the empty pCDNA3 expression vector (C: control) or with
the same plasmid carrying the PML3 isoform (PML: PML3
isoform). After 48 h from the transfection cells were splitted,
diluted 1 : 25, 1 : 50 and 1 : 100 and re-plated in medium containing
500 mg/ml of G418. After 15 ± 21 days of antibiotic selection a
crystal violet staining was perform to detect the surviving
colonies. The results are expressed as percentage of cloning
eciency with respect to the control empty expression vector
PML growth suppressive activity
M Fagioli et al
during the 4 days of culture (approximately 20 ±
25%).
Taken together these data suggest that PML
overexpression induces apoptotic cell death in U937
cells and that this e€ect depends on the levels of PML
protein expression.
a
A PML isoform with cytoplasmic localization has no
e€ects on cell growth
b
+Zn
MTP#F1
MT#8
–Zn
Figure 2 Growth inibition and cell death induction by inducible
expression of PNL3 in U937 cells. (a) Growth curves of
U937MTP#F1 and U937MT#8. Triplicate cultures of each of
the indicated clones were cultured in the presence or absence of
100 mM ZnSO4 and counted daily. The PML3 expression levels
were determined by Western Blotting and are shown in the panel
inside. (b) Representative TdT staining (day 4) of control (MT#8)
and PML overexpressing (MTP#Fl) U937 cells, in absence or in
presence of Zn treatment
MTP#F1 cells was comparable to those of Znuntreated and Zn-treated MT#8 control cells. It,
therefore, appears that MTP#F1 cells grow as the
control in the absence of Zn induction and that Zn
treatment has not signi®cant e€ects on growth of
U937 cells. However, Zn treatment of MTP#F1 cells
almost completely blocked the expansion of the cell
population, suggesting that PML overexpression
prevents growth of U937 cells.
To determine the mechanism through which PML
overexpression produces growth suppression, the cell
cycle distribution and percentage of apoptotic cells
was evaluated every 24 h in both the MT#8 control
and MTP#F1 clones by ¯ow cytometry of propidium iodide-stained nuclei and the TdT assay,
respectively. FACS analysis revealed no marked
variations in the proportion of cycling cells in
control and PML populations (not shown). Instead,
the TdT assay revealed that the percentage of
positive cells was low in Zn-untreated and Zntreated MT#8 cells and in Zn-untreated MTP#Fl
cells (approximately 2 ± 3%), during the 4 days of
the culture. Instead, in the MTP#F1 cells, the
number of TdT positive cells was high and constant
PML is a nuclear protein with a minor component of
cytoplasmic localization, as revealed by cell fractionation and immuno¯uorescence studies (Koken et al.,
1994; Flenghi et al., 1995 and our unpublished
results). Several PML isoforms have been identi®ed
which di€er for their cellular localization. The PML 34-7 isoform lacks the putative nuclear localization
signal (NLS) and is localized to the cytoplasm (Fagioli
et al., 1992; Flenghi et al., 1995). It contains the
Ring+BI-B2 boxes, the coiled-coil region and a short
tract of the predicted a-helix region (Figure 3a). All
other PML isoforms (PML1, PML2, PML3 and
PML4) are nuclear and di€er in their carboxyterminal region (Fagioli et al., 1992; Flenghi et al.,
1995). Transient expression of the PML 3-4-7 cDNA
into U20S yielded the expected 47 kDa a-PML
immunoreactive polypeptide (Figure 3b). However, it
lacked growth suppressive activity on the same cells,
as revealed by the colony assay (Figure 3c). The
PML3 (Figure 3c), PML1 and PML2 (data not
shown) isoforms gave comparable levels of growth
inhibition. These results suggest that nuclear localization is a prerequisite of the PML growth suppressive
activity.
Single deletions of the cysteine-histidine, coiled-coil or
carboxy-terminal regions exert no e€ect on the PML
growth suppressive activity
To de®ne the regions responsible for the PML growth
suppressive activity, we constructed PML mutants
lacking the aminoterminal cysteine-histidine rich
region containing the Ring+BI-B2 boxes (DCPML3),
the coiled-coil region (DHPML3), or the carboxyterminal region (PML-Pr). This last mutant was
engineered to contain the PML sequences which are
retained within the PML/RARa fusion protein (Figure
3a). The DCPML3, DHPML3 and PML-PR mutants
all localize to the nucleus. Their corresponding
expression vector were then transfected into U20S
cells to perform a colony formation assay. The levels
of expression, as determined by Western blot analysis
48 h after transfection, were high and roughly the
same for all mutants (see Figure 3b for representative
results) but became low or absent after selection, as
adjudged by Western blotting of the rare G418
resistant colonies (not shown). The results of the
colony assays (Figure 3c) showed that the PML
growth suppressive activity was retained in the
DCPML3 and PML-Pr mutants and partially reduced
in the DHPML3 mutant. As the biological activity of
PML was not signi®cantly a€ected by the deletion of
the Ring plus B1-B2 boxes, coiled-coil or carboxyterminal region, these domains do not appear to be
indispensable for the growth suppressor function of
PML.
2907
PML growth suppressive activity
M Fagioli et al
2908
Additive e€ects of di€erent PML regions on growth
suppression
Since none of the relevant PML domains appeared to
be crucial for the PML growth suppressive activity, we
investigated whether they might mediate independent
e€ects on cell growth. The answer was sought by
independently expressing the PML aminoterminal
cysteine-histidine rich region or the cysteine-histidine
rich plus coiled-coil regions in the U20S cells. These
a
R
B1
B2 1 2 3 4
α-H
s/p
1 23 4
α-H
s/p
B2
R
B1
B2
R
B1
B2 1 2 3 4
R
B1
B2 1 2 3 4
α-H
α-H
s/p
s/p
b
PML3
∆CPML3 ∆HPML3 PML-PR PML 3-4-7
C
96 —
66 —
42 —
c
Figure 3 E€ects of PML and mutant expression on growth of
U20S cells. (a) Modular organization of PML3 and PML
mutants. P: PML prolin-rich region; R: RING ®nger domain;
B1, B2: B-boxes; 1,2,3,4: the four clusters of hydrophobic
aininoacids heptads that constitute the coiled-coil region. a-H:
a-helix region; SP: serine proline-rich region. The bar above each
mutant represents the putative Nuclear Localization Signal
(NLS). The triangle indicates the breackpoint site (BP) of
t(15;17). (b) Western blot analysis of PML isoforms and mutant
protein expression. U20S were transfected with the indicated
expression vectors and analysed 48 h after. The PG antibody,
directed against a portion of a-helix common to all the mutants,
was used. Molecular weight markers are given on the left of the
blot. A control lane (C) of cells transfected with the empty
pCDNA3 expression vector was included. (c) Colony formation
assay. The assay was performed as described in Figure 1. The
results are expressed as percentage of cloning eciency with
respect to the control empty expression vector
two PML mutants were, however, not able to enter the
nucleus (data not shown), probably because they both
lacked the PML putative NLS, which has been mapped
between AA 448 and 494 (Flenghi et al., 1995; Kastner
et al., 1992). To target these two mutants to the
nucleus we ®rst demonstrated that the 448 ± 494 region
is involved in nuclear targeting of PML by adding this
sequence to a di€usible protein, the green ¯uorescence
protein (GFP) and analysing its cellular distribution.
The GFP is a highly ¯uorescent protein of the
bioluminescent jelly®sh Aequora victoria which emits
green light upon excitation with ultraviolet or blue
light (Chal¯e et al., 1994). Here we used a mutant of
GFP (Ser65→Thr) which emits a brighter green light
(the emission spectra is seven time higher than the
wild-type GFP) upon excitation with only blue light
(Heim et al., 1995). Its expression into mammalian cells
results in a di€use cytoplasmic and nuclear ¯uorescent
pattern (Rizzuto et al., 1995 and Figure 5) that can be
seen both in ®xed and living transfected cells. Addition
of the PML NLS to the GFP protein (GT1 construct,
Figure 4a) gave an exclusively nuclear ¯uorescent
pattern (Figure 5). Next the GT1 polypeptide was
fused to the PML aminoterminal cysteine-histidine rich
region (GT2 construct), to the PML cysteine-histidine
rich plus coiled-coil regions (GT8 construct) and to the
full-length PML3 (PMLB1 construct) (Figure 4a). The
resulting GFP-fusion proteins were transfected into
U2OS cells and analysed for their localization (Figure
5). The GT2 construct gave a nuclear di€use pattern of
staining, with the exclusion of nucleoli, while the GT8
stained within NBs, morphologically indistinguishable
from those of the PMLB1 protein (Figure 5). The GT1,
GT2, GT8 and PMLB1 plasmids were then used to
perform a colony formation assay, using U2OS cells, as
previously described. The levels of expression, as
determined by Western blot analysis 48 h after
transfection, were roughly the same for all GFPfusion proteins (Figure 4b). The results, expressed as
percentage of cloning eciency with respect to the
control GT1, demonstrated that the cysteine-histidine
region alone (GT2) had a slight but consistent growth
suppressive activity while the cysteine-histidine plus the
coiled coil (GT8) had a near wild-type e€ect (Figure
4c). It, therefore, appears that the cysteine-histidine
rich region exerts a weak, yet independent, activity on
growth regulation and that its fusion with the coiledcoil region is sucient to recapitulate the suppressive
function of wild type PML. In summary, these data,
together with those described in the previous paragraph showing that individual deletions of the various
PML domains have no e€ect on its growth suppressor
activity, suggest that the three PML domains have
independent and additive signalling potential toward
growth regulation.
Cellular localization of PML mutants
To determine whether the growth suppressive e€ect of
PML requires integrity of the NBs, we analysed the
cellular localization of the various PML mutants after
transient transfection in mouse NIH3T3 ®broblasts. As
our anti-PML antibodies do not recognize mouse
PML, they allow selective monitoring of exogenous
PML expression (Flenghi et al., 1995). The DCPML3
and DHPML3 mutants produced a di€use nuclear
PML growth suppressive activity
M Fagioli et al
pattern with the exclusion of nucleoli (Figure 6b and c,
respectively), while the PML-PR yielded a speckled
pattern with larger and less numerous NBs (Figure 6d)
a
R
B1 B2
R
B1 B2 1 2 3 4
R
B1 B2 1 2 3 4 α-H
s/p
b
PMLB1 GT2
GT8
C
96 —
66 —
42 —
c
Figure 4 E€ects of PML-GFP fusion proteins on growth of
U20S cells. (a) Modular organization of GFP- fusion protein.
Green Fluorescence Protein (GFP) is represented as an open
circle while the NLS of PML (not in scale) as an open rectangle.
P: PML prolin-rich region; R: RING ®nger domain; B1, B2: Bboxes; 1,2,3,4: clusters of hydrophobic aminoacids heptads. a-H:
a-helix region; SP: serine proline-rich region. (b) Western blot
analysis of the GFP-fusion proteins expression. Proteins are
indicated above the blot. A control lane of cells transfected with
the empty pCDNA3 expression vector was included. Blots were
decorated with the PG-M3 antibody, directed against the
aminoterminal portion of PML. (c) Colony formation assay.
The assays were performed as described in Figure 1. The results
were represented as percentage of cloning eciency normalized
with the transfection ecency. To establish the transfection
percentage of each construct, spontaneous ¯uorescence of GFP
was assessed on an aliquot of cells (approximately 1000 cells per
transfection). Transfection eciency ranged from 15 ± 30% among
di€erent constructs and experiments. Results shown in this panel
are comparable with those of Figure 3c. In fact, in a set of
separate experiments, we have documented that the GT1 and
pCDNA3 constructs and the PMLB1 and PML3 constructs
scored similarly in colony formation assays using U20S cells (data
not shown)
than in the transfected wt PML (Figure 6a). The
localization of these mutants is consistent with that
observed with the various GFP-PML fusion proteins
(Figure 5). In particular, the GT8 and the PML-PR
mutants, both lacking the PML C-terminal region,
gave a similar speckled pattern of nuclear localization.
In summary, the results indicate that the RING+B1B2 and the coiled-coil are required and sucient for
PML NB targeting since DCPML3, DHPML3 and the
GT2 mutants are nuclear di€use and the GT8 mutant
is speckled. Furthermore, targeting to NBs is not
necessary for the PML growth suppressive e€ect since
two growth suppressor PML mutants (DCPML3 and
DHPML3) had a di€use and two others (GT8 and
PML-PR) a speckled patterns of localization.
Discussion
We provided evidence that overexpression of PML
potently suppresses growth in various cell lines and
that the mechanism underlying growth suppression is
induction of apoptotic cell death. Other groups have
reported that PML overexpression exerts a negative
e€ect on the cellular transformation triggered by
certain oncogenes (neu) and cooperative oncogenes
(Ha-Ras plus myc, Ha-Ras plus mutated p53) (Liu et
al., 1995; Mu et al., 1994). Since we demonstrated that
PML overexpression a€ects growth rather than
transformation, and that this e€ect is also evident in
non-transformed cells, the previously described antitransformation potential of PML might be the
consequence of a more general e€ect on cell growth.
The relevance of these data with regard to the
physiological function of PML is not clear, as the
PML-antiproliferative e€ect was established in nonphysiological conditions, e.g. upon expression of the
PML gene under the control of a heterologous
promoter. Analysis of PML expression in normal
tissues revealed that it is mainly expressed in postmitotic cells (Flenghi et al., 1995; Gambacorta et al.,
1996). In addition, PML-NBs have been shown to be
target of DNA viruses (Carvalho et al., 1995; Doucas
et al., 1996; Everett and Maul, 1994; Puvion-Dutilleul
et al., 1995) and PML itself is up-regulated by the
antiproliferative and anti-viral agent interferon (ChelbiAlix et al., 1995; Lavau et al., 1995; Stadler et al.,
1995) and forms stable complexes with the adenovirus
E1A transforming protein (Alcalay et al., 1998). All
together these ®ndings suggest that PML is physiologically involved in negative growth regulation, possibly
by modulating cell survival.
To investigate the molecular mechanisms through
which PML induces growth suppression, we have
analysed the biological activity of di€erent PML
isoforms and mutants. Several PML isoforms have
been identi®ed that di€er for their carboxy-terminal
region (Fagioli et al., 1992). They all contain a putative
NLS and localize to the nucleus, with the exception of
the PML 3-4-7 isoform that lacks the NLS and
localizes to the cytoplasm (Flenghi et al., 1995). The
nuclear PML1, PML2 and PML3 isoforms showed a
potent growth suppressive activity, while the cytoplasmic PML 3-4-7 isoform had no e€ect on cell growth,
thereby suggesting that PML exerts its biological
activity within the nucleus. We have then investigated
2909
PML growth suppressive activity
M Fagioli et al
2910
the biological activity of various nuclear PML mutants
which were designed to explore the relevance of three
regions of the PML protein: the Ring ®nger and the
B1-B2 boxes; the coiled-coil region, which is a proteinprotein dimerization interface; the carboxy-terminal
region, that corresponds to the region of PML lost
during the recombination with RARa in the t(15;17)
translocation. None of these regions appears to be
indispensable for the growth suppressive e€ect of
PML, since individual deletions had no e€ect on the
biological activity of PML. However, since growth
inhibitory activity was weak, though consistent, with
GFP
GT1
GT2
the RING+B1-B2 region alone, while it was fully
expressed by the RING+B1-B2 plus coiled-coil
regions, they may exert both independent and
cooperative e€ects on growth.
The mechanisms through which PML overexpression a€ects cell growth are not clear. Although the
PML RING ®nger has no assigned function, RING
domains of two other proteins (TIF1 and BRCA1) are
thought to act as protein-protein interaction domains
(Friedman et al., 1996; Wu et al., 1996). The coiled-coil
region of PML has been shown to be a protein-protein
interaction domain (Kastner et al., 1992). The B1 and
GT8
PMLB1
Figure 5 Cellular localization of PML-GFP fusion proteins in NIH3T3 cell line. Cells were transiently transfected with the
indicated cDNAs using calcium phosphate precipitation procedure. The GFP ¯uorescence was detected after ®xation of cells with
3.7% paraphormaldeide for 10 min and methanol for 5 min at room temperature (upper panels); the lower panels show the
overlapped images, by software, of 4',6-diamidino-2-phenylindole (DAPI) staining (for identi®cation of nuclear morphology) and
GFP
Figure 6 Immuno¯uorescence analysis of the wt and mutant PML localization in NIH3T3 cells. Cells were transiently transfected
using the calcium phosphate precipitation procedure and then stained with two di€erent antibodies: PG-M3 to stain PML3 (a),
DHPML3 (c) and PML-PR (d) mutant proteins and PG to stain the DCPML3 mutant protein (b)
PML growth suppressive activity
M Fagioli et al
B2 boxes and the carboxy-terminal regions also have
no assigned functions. However, we have recently
proved that they are involved in the formation of
PML-Rb complex (Alcalay et al., 1998). It may,
therefore, be that the three PML domains direct
multiple protein-protein interactions which are involved in the growth suppressive function of PML.
The PML/Rb interaction, however, does not appear to
be involved since PML overexpression induces growth
suppression in Rb7/7 cells, as well (Alcalay et al.,
1998).
It has previously been reported that the capacity of
PML to suppress neoplastic transformation depends on
its localization to the NBs (Le et al., 1996). We have
demonstrated that the RING, B1+B2, the coiled-coil
of PML are both indispensable and sucient to target
PML to the NBs. Fusion of these regions to a
di€usible protein, such as the GFP, target the GFP
to the NBs and individual deletions of these regions
abrogate the NB localization of PML. However,
although some of the PML mutants that are targeted
to the NBs are biologically active, others exhibit
similar biological activity and a nuclear di€use
localization. In summary, the PML regions involved
in NB targeting and growth suppression do not
completely overlap.
The lack of correlation between NB localization and
the growth suppressive e€ect suggests that the NBs are
not the only active sites of PML. Biochemical analysis
of PML cell distribution by fractionation experiments
(unpublished results) have revealed that the wild-type
and carboxy-terminal deleted PML molecules are
predominantly localized in the nuclear matrix and
less in the soluble nuclear fraction. Since both proteins
localize to NBs, this correlates with previously reported
®ndings that PML-NBs are matrix-associated (Chang
et al., 1995). On the other hand, the PML mutants with
deletions of the cysteine-histidine or coiled-coil regions
are nuclear di€use and mainly localized to the soluble
nuclear fraction, with only a minor component
associated with the nuclear matrix. Overall these data
suggest that the PML that is localized to the matrixassociated NBs and the nuclear di€use and soluble
PML are both involved in transducing negative signals
to cell growth.
The data presented here have important implications for understanding the mechanisms through
which the PML/RARa protein exerts its biological
activities. The current hypothesis is that fusion with
RARa results in the loss of the PML growth
suppressive function and that PML/RARa inactivates the endogenous PML growth suppressive
activity by destroying the PML-NBs (Le et al.,
1996). A number of ®ndings argue against this
hypothesis: (i) the PML portion of the PML/RARa
fusion protein (PML-PR mutant) retains growthsuppressive activity; (ii) the NBs are not the only
active sites of the PML protein; (iii) PML/RARa
increases survival of a subset of haematopoietic cells
(Grignani et al., 1993; Rogaia et al., 1995), but
suppresses growth of all other cell types tested
(Ferrucci et al., 1997). An alternative hypothesis is
that PML/RARa incorporates the biological activity
of PML. This would result in the capacity of PML/
RARa to induce cell death in every cell type, except
in a subset of haemopoietic cells, where it, instead,
promotes survival. In the latter, the PML portion of
the fusion protein might acquire survival-stimulating
functions as a consequence of its fusion with RARa
and physical or functional interactions with tissue
speci®c factors. The identi®cation of haematopoieticspeci®c PML/RARa interacting proteins would help
clarify this issue.
Materials and methods
PML mutants and expression vectors
The various PML2 and PML3 mutants were obtained from
the PML M51 and M58 clones (Fagioli et al., 1992),
respectively by PCR base technologies and controlled by
DNA sequence. For the DH mutants construction the two
PML sequences ¯anking the coiled-coil region were PCR
ampli®ed with the oligonucleotides 5'-GGATGCCTGGAGGCGTCGGGC-3' and 5'-GTCGCACTCGAGCTCACTGT-3'; 5'-CTGCCGCCTGCGCCTCGAGGA-3' and
5'-TCGGGCAGGTCAACGTCAAT-3'. The two fragments were ligated through the XhoI sites introduced by
PCR (underlined). The resulting PML fragment, which
lacked aminoacids 226 ± 361, was introduced in the PML2
M51 and PML3 M58 cDNAs between SphI and HincII. To
obtain the DC mutant the PML region from bp 721 ± 881
(primers: 5'-CTTGACAGCGCCACCATGGAGCTCA-3'
and 5'-GCGCGGCCCAGCTGGCCGACG-3') was ampli®ed and substituted to the 5' PML3 M58 sequences at the
S®I site. The 5' oligonucleotides contained a start codon
(underlined) in the context of a Kozak consensus. The
PML-PR cDNA, encoding a PML protein truncated at the
aminoacid 553, was constructed by substituting, at the
HincII site, the 3' portion of a PCR-modi®ed PML cDNA
carrying a TAG stop codon (primers: 5'-AGAGCCTGCAAGCTGCCGT-3' and 5'-ACAACGGGATCCTATGCCT-3'; the stop codon is underlined). PML 3-4-7 clone
has just been described (Fagioli et al., 1992). Using oligo
NLS1 (from 1403 to 1421 bp: 5'-AGTCAGTGCCCGGGGCACA-3') and NLS2 (from 1556 to 1537: 5'-TCCTCAGATCACATCTTGAT-3'), the region containing the
putative PML nuclear localization signal was engineered
to contain a 5' SmaI site (underlined in oligo NLS 1) and a
3' stop codon (bold typed in oligo NLS2) and then ligated
to the 3' extremity of GFP cDNA (Chal¯e et al. 1994)) by
blunt-end ligation into aEcoRV site that mutagenized the
GFP TGA stop codon. The resulting GFP-NLS construct
(GT1) was modi®ed to introduce a SacI 5' site at the ATG
start codon and then ligated to the EcoRI ± SacI 740 bp
and to the EcoRI ± SacI 1162 bp fragement of PML3 M58
clone to produce the GT2 and the GT8 cDNA clones,
respectively. The PMLB1 cDNA clone was obtained by
introducing a BamHI site at the TGA site of PML3 and at
the ATG of GFP. All the described mutants and the
PML1, PML2 and PML3 (Fagioli et al., 1992) cDNAs
were subcloned into the pCDNA3 (Invitrogen, De Schelp
12, 9351 NV Leek, The Netherlands) or pMT (Grignani et
al., 1993) expression vectors.
Cell culture and transfection
Cell lines that grown in suspension were maintained in
RPMI 1640 medium supplemented with 10% FCS plus
appropriated growth factors and then transfected by
electroporation. Cells were selected in G418 and subcloned under limiting dilution conditions. All adherent cell
lines were cultured in Dulbecco modi®ed Eagle's medium
supplemented with 10% FCS, except NIH3T3 which was
cultured in 10% BCS. The cells were then transfected by
the calcium phosphate precipitation procedure.
2911
PML growth suppressive activity
M Fagioli et al
2912
Western blotting
Western blot analysis was performed as described (Pandol®
et al., 1991) on di€erent cell lysates using the PG-M3
(Flenghi et al., 1995) or PG anti-PML antibodies and
revealed by the ECL method (Amersham, Little Chalfont,
UK).
Colony formation assay
U20S, C33a, SAOS, HeLa, NIH3T3 and PEF cells were
plated at a concentration of 56105 in a 10 cm plate and
transfected with 10 mg of cloned or control DNAs. After
48 h cells were splitted 1 : 25, 1 : 50 and 1 : 100 and
subcoltured into medium containing 500 mg/ml of G418.
A small amount was plated in chamber slides (Nunc Inc.,
Naperville, Illinois, USA) to perform anti-PML immuno¯uorescence staining to verify transfection ecency. Drug
resistant colonies that appeared 14 ± 21 days later were
either isolated for further analysis or quanti®ed after being
stained with crystal violet.
Immuno¯uorescence staining
U937 cells were collected before or after 100 mM ZnSO4
treatment, cytocentrifuged and ®xed in methanol at room
temperature for 5 min followed by acetone at 7208C for
2 min. Adherent cells were plated in chamber slides and
®xed in paraformaldehyde 3.7% for 10 min followed by
methanol for 5 min at room temperature. PML staining
was performed with three di€erent antibodies: the PG-M3
monoclonal antibody, directed against the PML Nterminal region (Flenghi et al., 1995); the PG serum,
against aminoacids 228 ± 388 of the the a-helix (a gift from
P Greer); the anti PML-3 serum against the PML3 unique
C-terminal region (aminoacids 592 ± 607) (Fagioli et al.,
1992). Cells were then stained with FITC- or rhodamineconjugated anti-mouse or anti-rabbit Ig antibodies (Southern Biotechnology Associates, Birmingham, AL, USA).
Preparations were examined on a Olympus BX-60
¯ourescence microscope equipped with a chilled 3CCD
digital colour camera (C5810 Hamamatsu Photonics,
Japan). Images were captured with a 24-bit board (Image
grabber 24, Neotech, London UK) on a 8100/80 Power
Macintosh computer (Apple, Cupertino, USA). For
colocalization experiments, the same microscopic ®elds
were examined with distinct cubes for ¯uorescein
(excitation ®lter 470 ± 490, dichroic mirror 505, barrier
®lter 515 ± 550 nm) and rhodamine (excitation ®lter 510 ±
550, dichroic mirror 570, barrier ®lter 590 nm). The images
were directly superimposed by the C5810 3CCD control
unit.
Cell growth, cell cycle phase distribution and apoptosis
U937 cells were plated at 26105 per ml in RPMI with 10%
FCS in the presence and absence of 100 mM ZnSO4.
Proliferation was evaluated by direct cell counting
(trypan blue exclusion method). At various time points,
the cells were analysed for cell cycle distribution by ¯ow
cytometric analysis of propidium iodide-stained nuclei, as
described (Pagliacci et al., 1993). To identify apoptosis
cells were collected, cytocentrifuged, ®xed in 1% paraformaldehyde and subjected to TdT (terminal-deoxynucleotidyl transferase) assay (Gorczyca et al., 1993).
Acknowledgements
We thank Dr Catia Scassellati for technical assistance.
Marta Fagioli was supported by `A. Castelnuovo Foundation' fellowship, Pier Francesco Ferrucci by an Italian
Association of Internal Medicine fellowship and the Italian
Cancer Research Federation (FIRC). The investigation was
supported by grants from AIRC, CNR-ACRO and EC
(Biomed and Biotech programs).
References
Alcalay M, Tomassoni L, Colombo E, Stlodt S, Grignani Fr,
Fagioli M, Szekely L, Helin K and Pelicci PG. (1998). Mol.
Cell. Biol., in press.
Alcalay M, Zangrilli D, Pandol® PP, Longo L, Mencarelli A,
Giacomucci A, Rocchi M, Biondi A, Rambaldi A, Lo
Coco F, Diverio D, Donti E, Grignani F and Pelicci PG.
(1991). Proc. Natl. Acad. Sci. USA, 88, 1977 ± 1986.
Altabef M, Garcia M, Lavau C, Bae S, Dejean A and
Samaraut J. (1996). EMBO J., 15, 2707 ± 2716.
Ascoli C and Maul GJ. (1991). J. Cell. Biol., 112, 785 ± 795.
Boddy MN, Howe K, Etkin LD, Solomon E and Freemont
PS. (1996). Oncogene, 13, 971 ± 982.
Borrow J, Goddard AD, Sheer D and Solomon E. (1990).
Science, 249, 1577 ± 1581.
Brown D, Kogan S, Lagasse E, Weissman I, Alcalay M,
Pelicci PG, Atwater S, and Bishop M. (1997). Proc. Natl.
Acad. Sci. USA, 94, 2551 ± 2556.
Chal¯e M, Tu Y, Euskirchen G, Ward WW and Prasher DC.
(1994). Science, 263, 802 ± 805.
Carvalho T, Seeler JS, Ohman K, Jordan P, Petterson U,
Akusjarvi G, Carino-Fonseca M and Dejean A. (1995). J.
Cell. Biol., 131, 45 ± 56.
Chang KS, Fan YH, Andreef M, Liu J and Mu ZM. (1995).
Blood, 85, 3646 ± 3653.
Chelbi-Alix MK, Pelicano L, Quignon F, Koken MHM,
Venturini L, Stadler M, Pavlovic J, Degos L and de The H.
(1995). Leukemia, 9, 2027 ± 2033.
Desbois C, Rousset R, Bantignies F and Jalinot P. (1996).
Science, 273, 951 ± 953.
de The H, Chomienne C, Lanotte M, Degos L and Dejean A.
(1990). Nature, 347, 558 ± 561.
de The H, Lavau C, Marchio A, Chomienne C, Degos L and
Dejean A. (1991). Cell, 66, 675 ± 684.
Doucas V, Ishov AM, Romo A, Juguilon H, Weitzman MD,
Evans RM and Maul GG. (1996). Genes Dev., 10, 196 ±
207.
Dyck JA, Maul GG, Miller Jr WH, Chen JD, Kakizuka A
and Evans R. (1994). Cell, 76, 333 ± 343.
Epstein AL. (1984). J. Virol., 50, 372 ± 379.
Everett RD and Maul GG. (1994). EMBO J., 13, 5062 ±
5069.
Fagioli M, Alcalay M, Pandol® PP, Venturini L, Mencarelli
A, Simeone A, Acampora D, Grignani F and Pelicci PG.
(1992). Oncogene, 7, 1083 ± 1091.
Ferrucci PF, Grignani Fr, Pearson M, Fagioli M, Grignani
F, Nicoletti I, Pelicci PG. (1997). Proc. Natl. Acad. Sci.
USA, 94, 10901 ± 10906.
Flenghi L, Fagioli M, Tomassoni L, Pileri S, Gambacorta M,
Pacini R, Grignani Fr, Casini T, Ferrucci PF, Martelli
MF, Pelicci PG, Falini B. (1995). Blood, 85, 1871 ± 1880.
Freemont PS, Hanson IM and Trowsdale J. (1991). Cell, 64,
483 ± 484.
Friedman JR, Fredericks WJ, Jensen DE, Speicher DW,
Huang X, Neilson EG and Rauscher III FG. (1996). Genes
Dev., 10, 2067 ± 2078.
PML growth suppressive activity
M Fagioli et al
Gambacorta M, Flenghi L, Fagioli M, Pileri S, Leoncini L,
Bigema B, Pacini R, Natali Tanci L, Pasqualucci L, Ascani
S, Mencarelli A, Liso A, Pelicci PG and Falini B. (1996).
Am. J. Path., 149, 2023 ± 2035.
Gandini D and Pandol® PP. (1994). Blood, 84 (Supp. 1),
439a.
Gorczyca W, Gong J and Darzynkiewicz Z. (1993). Cancer
Research, 53, 945.
Grignani Fr, Ferrucci PF, Testa U, Talamo G, Fagioli M,
Alcalay M, Mencarelli A, Grignani F, Peschle C, Nicoletti
I and Pelicci PG. (1993). Cell, 74, 423 ± 431.
Grignani Fr, Testa U, Rogaia D, Ferrucci PF, Pinto A,
Aldinucci D, Gelmetti V, Fagioli M, Alcalay M, Seeler J,
Grignani F, Nicoletti I, Peschle C and Pelicci PG. (1996).
EMBO J., 15, 4949 ± 4958.
Grisolano JL, Wesselschmidt RL, Pelicci PG and Ley TJ.
(1996). Blood, 89, 376 ± 387.
Heim R, Cubitt AB and Tsien RY. (1995). Nature, 373, 663 ±
664.
Kakizuka A, Miller Jr WH, Umesono K, Warrel Jr RP,
Frankel SR, Murty VVS, Dmitrovsky E and Evans RM.
(1991). Cell, 66, 663 ± 674.
Kastner P, Perez A, Lutz Y, Rochette-Egly C, Gaub MP,
Durand B, Lanott M, Berger R and Chambon P. (1992).
EMBO J., 11, 629 ± 642.
Koken MHM, Puvion-Dutilleul F, Guillemin MC, Viron A,
Cruz-Linares G, Stuurman N, de Jong L, Szostecki C,
Calvo F, Chomienne C, Degos L, Puvion E and de The H.
(1994). EMBO J., 13, 1073 ± 1083.
Korioth F, Gie€ers C, Maul GG and Frey J. (1995). J. Cell
Biol., 130, 1 ± 13.
Lavau C, Marchio A, Fagioli M, Jansen J, Falini B, Lebon P,
Grosveld F, Pandol® PP, Pelicci PG and Dejean A. (1995).
Oncogene, 11, 871 ± 876.
Le XF, Yang P and Chang KS. (1996). J. Biol. Chem., 1,
130 ± 135.
Le Douarin B, Zechel C, Garnier JM, Lutz Y, Tora L, Pierrat
B, Heery D, Gronemeyer H, Chambon P and Losson R.
(1995). EMBO J., 14, 2020 ± 2033.
Liu JH, Mu ZM and Chang KS. (1995). J. Exp. Med., 181,
1965 ± 1973.
Lovering R, Hanson IM, Borden KLB, Martin S, O'Reilly
NJ, Evan GI, Rahman D, Pappin DJC, Trowsdale J and
Freemont P. (1993). Proc. Natl. Acad. Sci. USA, 90,
2112 ± 2116.
Mu ZM, Chin KV, Liu JH, Lozano G and Chang KS. (1994).
Mol. Cell. Biol., 14, 6858 ± 6867.
Pagliacci MC, Fumi G, Migliorati G, Grignani F, Riccardi C
and Nicoletti I. (1993). Lymphokine Cytokine Res., 12,
439 ± 447.
Pandol® PP., Grignani Fr, Alcalay M, Mencarelli A, Biondi
A, Lo Coco F, Grignani F and Pelicci PG. (1991).
Oncogene, 6, 1285 ± 1292.
Puvion-Dutilleul F, Chelbi-Alix MK, Koken K, Quignon F,
Puvion E and de The H. (1995). Exp. Cell Res., 218, 9 ± 16.
Reddy BA, Etkin LD and Freemont PS. (1992). TIBS, 17,
344 ± 345.
Rizzuto R, Brini M, Pizzo P, Murgia M and Pozzan T.
(1995). Curr. Biol., 5, 635 ± 642.
Rogaia D, Grignani Fr, Nicoletti I and Pelieci PG. (1995).
Leukemia, 9, 1467 ± 1472.
Stadler M, Chelbi-Alix MK, Koken MHM, Venturini L, Lee
C, Saib A, Quignon F, Pelicano L, Guillemin MC,
Schindler C and de The H. (1995). Oncogene, 11, 2565 ±
2573.
Szostecki C, Guldner HH, Netter HJ and Will H. (1990). J.
Immunol., 145, 4338 ± 4347.
Tkahashi M, Inaguma Y, Hiai H and Hirose F. (1988). Mol.
Cell. Biol., 8, 1853 ± 1856.
Weis K, Rainbaud S, Lavau C, Jansen J, Carvalho T,
Carmo-Fonseca M, Lamond A and Dejean A. (1994).
Cell, 76, 345 ± 358.
Wu LC, Wang ZW, Tsan JT, Spillman MA, Phung A, Xu
XL, Yang, Hwang L, Bowcock AM and Baer R. (1996).
Nature Gen., 14, 430 ± 440.
2913