tethering by lamin A stabilizes and targets the ING1 tumour suppressorXijing Han1,6, Xiaolan Feng1,6, Jerome B. Rattner2, Heather Smith1, Pinaki Bose1, Keiko Suzuki1, Mohamed A. Soliman1,3, Michelle S. Scott4, Brian E. Burke5 and Karl Riabowol1,7
ING proteins interact with core histones through their plant homeodomains (PHDs)1–4 and with histone acetyltransferase (HAT) and histone deacetylase (HDAC) complexes to alter chromatin structure5–7. Here we identify a lamin interaction domain (LID) found only in ING proteins, through which they bind to and colocalize with lamin A. Lamin knockout (LMNA–/–) cells show reduced levels of ING1 that mislocalize. Ectopic lamin A expression increases ING1 levels and re-targets it to the nucleus to act as an epigenetic regulator6,8. ING1 lacking the LID does not interact with lamin A or affect apoptosis. In LMNA–/– cells, apoptosis is not affected by ING1. Mutation of lamin A results in several laminopathies, including Hutchinson-Gilford progeria syndrome (HGPS), a severe premature ageing disorder9. HGPS cells have reduced ING1 levels that mislocalize. Expression of LID peptides to block lamin A–ING1 interaction induces phenotypes reminiscent of laminopathies including HGPS9,10. These data show that targeting of ING1 to the nucleus by lamin A maintains ING1 levels and biological function. Known roles for ING proteins in regulating apoptosis and chromatin structure indicate that loss of lamin A–ING interaction may be an effector of lamin A loss, contributing to the HGPS phenotype.
The five mammalian ING genes (ing1–5) encode evolutionarily con-served proteins11 that are frequently inactivated in cancers and have roles in apoptosis and cell senescence12,13 through binding histone H3 in a methylation-sensitive (ING1 and ING2) or -insensitive (ING4) man-ner1–4. Binding targets acetylation and deacetylation complexes to alter the epigenetic code, as ING proteins are stoichiometric components of HAT and HDAC complexes7,8. ING proteins also affect DNA damage recognition and repair by inducibly binding proliferating cell nuclear antigen (PCNA) after UV-induced DNA damage14, and by making chro-matin accessible to DNA repair complexes15.
Lamins are type V intermediate filament proteins that bind diverse proteins16 and are essential elements of the nuclear lamina and inner nuclear membrane. The three human lamin genes (LMNA, LMNB1, LMNB2) encode at least seven different proteins. Lamins A and C are alternatively spliced isoforms of the LMNA gene, and both bind a mix-ture of core histones17. Located on the peripheral lamina, lamins help to maintain the shape and stability of the nuclear envelope. Lamins are also involved in the regulation of DNA replication and transcription, and this function is partially attributed to their associations with PCNA and the retinoblastoma (Rb) protein16,18, where subnuclear localization and proteosomal degradation of Rb are regulated by lamin A binding19.
ING proteins contain a conserved sequence of about 50 amino acids, which is unique in the human proteome11. It is located in a region distinct from other domains, including the PHD that binds trimethylated histone H3 at Lys 4 (H3K4Me3) in a methylation-sensitive manner to read the histone code (Fig. 1a, b). To identify proteins that could interact with this region, we transfected HEK293 cells with a His-tagged construct encoding a triple-repeat of this region attached to a single nuclear locali-zation signal (Fig. 1c) to properly target it to the nucleus where ING1 is predominantly localized8. We used Ni-NTA columns to pull down associated proteins. After digestion with trypsin, binding partners were identified by liquid chromatography-mass spectrometry (LC/MS/MS). Using a significance threshold of P < 0.05 (ref. 20), we found nine inter-actions that were specific when compared with a non-specific His–GFP negative control run in parallel. Two of the top five scores were isoforms of lamin A/C (Fig. 1d). We subsequently designated this region the lamin interaction domain (LID).
To verify MS data, we performed reciprocal co-immunoprecipita-tions using HEK293 cells, Hs68 human diploid fibroblasts and fibrob-lasts from HGPS patients. In immunoprecipitation-western blots of HEK293 cells transfected with the ING1 constructs (Fig. 1c), we found that wild-type p33ING1b (hereafter called ING1) and the LID+NLS
1Departments of Biochemistry & Molecular Biology and Oncology, Faculty of Medicine, University of Calgary, 3330 Hospital Dr. NW, Calgary, Alberta, Canada, T2N 4N1. 2Departments of Cell Biology and Anatomy, Biochemistry & Molecular Biology and Oncology, Faculty of Medicine, University of Calgary, 3330 Hospital Dr. NW, Calgary, Alberta Canada T2N 4N1. 3Department of Biochemistry, Faculty of Pharmacy, Cairo University, Cairo, Egypt. 4Current address: School of Life Sciences Research, College of Life Sciences, University of Dundee, Scotland. 5Department of Anatomy and Cell Biology, University of Florida College of Medicine, 1600 SW Archer Road, Gainesville, FL 32606, USA.6These authors contributed equally to the manuscript7Correspondence should be addressed to K.R. (e-mail: [email protected])
Received 2 June 2008; accepted 26 August 2008; published online 5 October 2008; DOI: 10.1038/ncb1792
nature cell biology volume 10 | number 11 | november 2008 1333
proteins avidly bound endogenous lamin A, whereas the LID deletion mutant and the NLS fragment did not (Fig. 2a), despite similar expres-sion levels (lower panels of Fig. 2a). Reciprocal immunoprecipitation of endogenous lamin A confirmed that full-length ING1 and the LID+NLS were recovered in lamin A immunoprecipitations, whereas the LID deletion and NLS control proteins were not (Fig. 2b). Analyses of cells expressing ING1–ING5 suggest that ING1–4 also co-precipitates with lamin A (Supplementary Information, Fig. S1). Immunoprecipitation-westerns of cells transfected with different regions of lamin A showed that ING1 and the LID+NLS both bound the amino end corresponding to the rod domain, which is important for lamin dimerization (Fig. 2c), whereas the carboxyl region that binds other proteins, including actin, histones, emerin and others16–18, was not bound. To determine whether endogenous wild-type and mutant forms of lamins bound to endog-enous ING1, we performed immunoprecipitation-western assays using normal human Hs68 fibroblasts (wild-type lamin A/C), Prog1 HGPS and Prog 2 HGPS fibroblasts. Prog1 HGPS fibroblasts have a Gly 608→Gly 608 silent mutation in one lamin A allele that affects splic-ing and produces mutant LAΔ50 (progerin). Prog2 HGPS fibroblasts have an Arg 644→Cys 644 mis-sense mutation in one allele of lamin A,
resulting in defects in the cleavage recognition sequence by Zmpste24, an endoprotease essential for pre-lamin A processing10,21. Both lamins A and C were recovered in ING1 immunoprecipitates (Fig. 2d), whereas progerin was not. Interaction was noted when ING1 and progerin were overexpressed (data not shown), suggesting that although interaction can occur, it does so with reduced avidity between ING1 and progerin. To confirm that ING1 interacts directly with the lamins, lysates from normal human Hs68 fibroblasts, Prog1 and Prog2 HGPS fibroblasts, murine fibroblasts (LMNA+/+) and syngeneic murine lamin knockout fibroblasts (LMNA–/–) were electrophoresed and probed with purified HIS–LID+NLS in far western protein overlay experiments. Purified HIS–LID+NLS interacts with lamin A, but interactions are not seen in cells that do not express lamin A, and progerin does not seem to interact directly (Fig. 2e). This indicates that ING1–lamin A binding is indeed direct and that partial renaturation in this assay restored protein structure sufficiently to recapitulate differential binding between ING1 and the wild-type and progerin forms of lamin A.
As ING function as an epigenetic regulator is affected by subcel-lular localization8, and localization of the tumour suppressor Rb was reported to be affected by lamin A19, we examined ING1 levels in nuclear
PCNA
14-3-3
Sap30 PIsH3K4me3
ARF
a
karyopherins
b
c1
1 73
73 196
127
dRANK DESCRIPTION MASS SCORE PEPTI DES MATCHED ION SCORE RANGE
1 LAMIN A/C isoform 2 65,096 187 8 15–94
2 TCP-1-β 22,929 150 2 96–108
3 SAP 145 97,596 129 6 14–75
4 LAMIN A/C isoform 3 70,618 128 8 6–81
5 SWI/SNF 170K subunit 132,650 91 3 18–67
PIP LID NLS PHD PBRPB
ING1-WT
ING1-∆LID
ING1-LID+NLS
NLS
ING1-TriLID+NLS
138 196
279
279
73 126 73 126 73 196
Figure 1 Structural features of ING1 and potential binding partners of the LID. (a) p33ING1b (ING1) consists of 279 amino acids and possesses several functional domains. ING1 interacts with PCNA, SAP30, ARF, 14-3-3, karyopherins, methylated histone H3 (H3K4me3) and specific phosphoinositides (PIs) in the binding regions indicated8. Here we map a domain conserved within and unique to ING proteins (amino acids 74–126 of ING1) that binds lamin A. This area is contained within the SAP30 interaction domain (SAID) previously mapped to the first 124 amino acids of ING1 (ref. 6). PIP, PCNA-interacting protein motif; PB, partial bromodomain;
LID, lamin interaction domain; NLS, nuclear localization signal; PHD, plant homeodomain; PBR, polybasic region. (b) Multiple sequence alignments of the LID of ING1-5. The sequences of the LID were analysed by T-Coffee and visualized by the GENEDOC programs. Residues shaded in black indicate high conservation in the whole ING family and those in grey are less conserved ones. (c) Schematic representation of ING1b constructs used in this study. (d) Top five protein hits of liquid-chromotography mass spectrometry (LC/MS/MS) analyses (determined after subtracting negative-control hits) that may interact with the newly identified LID.
1334 nature cell biology volume 10 | number 11 | november 2008
and cytoplasmic fractions of wild-type (LMNA+/+) and LMNA–/– cells. Lamin A loss reduced the amount of ING1 found in the nucleus (Fig. 2f). Indirect immunofluorescence microscopy of LMNA–/– cells showed strong nuclear ING1 staining in the presence, but not the absence of lamin A (Fig. 3a), perhaps contributing to the altered chromatin dis-tribution seen in lamin A knockout22 and HGPS23 cells, and with DAPI staining of LMNA–/– cells (Fig. 3a). Reduced levels and/or altered forms of lamin A also seem to affect ING1 subcellular distribution as both the Prog1 and Prog2 HGPS strains showed significantly more cytoplasmic ING1 staining, and significantly less nuclear staining than normal Hs68 fibroblasts (Fig. 3b). Furthermore, in murine embryonic stem (ES) cells, which do not express lamin A in the undifferentiated state, ING1 protein localized in both the cytoplasm and nucleus of undifferentiated clus-ters of ES cells (Fig. 3c), whereas ES cells that began to differentiate, as indicated by lamin A expression (arrow in Fig. 3c), also localized ING1 to the nucleus. To further test whether mislocalization was due to the absence of lamin A and not to possible secondary effects or mutations
in knockout cells, LMNA–/– cells were transfected with lamin A, which restored nuclear localization of ING1 (Fig. 3d). As ING1 levels seemed to be greater in cells with intact (Fig. 3a) or restored (Fig. 3d) lamin A, we next asked whether ING1 levels were altered in LMNA–/– or HGPS cells. Cell fractionation showed that ING1 levels are decreased several fold in LMNA–/– (Fig. 4a) and in different HGPS cells (Fig. 4b) cells. We also confirmed a previous report that Rb levels are reduced in LMNA–/– cells19 (Fig. 4a) and showed here that Rb levels were also reduced in HGPS cells (Fig. 4b).
ING proteins have roles in apoptosis, with both forced overexpres-sion8 and knockout in normal cells24 increasing cell death. This is con-sistent with both hyperacetylation and hypoacetylation of histones H3 and H4, which are targeted by ING18,25,26, increasing susceptibil-ity to apoptosis. As loss of interaction with lamin affected subcellular localization of ING1, and previous studies have implicated localiza-tion in ING1 function8,27, we asked whether ING1, which was unable to interact with lamin A, could affect apoptosis as efficiently as intact
a d
fT C N T C N
b
c
e
32.525
16.5
16.5
83
8316.5
2532.5
16.525
32.5
47.5
32.5
32.5
16.5
47.5
47.5
32.547.5
62
83
62
8362
83
62
8362
8362
8362
62
62
32.547.5
62
47.562
62
Lamin ALamin C
Lamin A
Lamin A/C IP
Lamin C
Lamin ALamin C
IgG(H)
ING1
ING1 IP ING1 IP
ING1 IP
ING1IgG(L)LID+NLS
IgG(H)
IgG
IgG
IgG
ING1 LID
Lamin A
Lamin A (H)Lamin A (M)
Lamin A (H)
Lamin A (M)
Lamin C
Actin
Lamin A 50
Lamin A 50 (H)
LID+NLSNLS
ING1ING1 LIDLID+NLS
Lamin A
LID+NLS
415–664
1–4471–406
Lamin A1–4471–406
NLS
V ING1 LID LID+NLS NLS
V ING1 LID LID+NLS NLS
V 1–664 1–406 1–447 415–664
INPUT LYSATE
ING1
H2B
CBB
CBB
Lamin AIg
G
Pro
g1
Pro
g2
Hs6
8
Pro
g1
Pro
g2
Hs6
8
Prog
1
Prog
2
LMN
A+/
+
LMN
A–/
–
LMNA+/+ LMNA–/–
Hs6
8
Mr(K)
Mr(K)
Mr(K)
Mr(K)
Mr(K)
Mr(K)
Figure 2 ING1 binds to lamin A through a LID. (a) Lysates from HEK293 cells transfected with the constructs indicated were immunoprecipitated with ING antibodies and immunoprecipitates were probed with lamin antibodies. Lower panels show blots of lysates with lamin and ING antibodies to confirm expression levels. (b) Reciprocal immunoprecipitations using lamin A/C antibody to immunoprecipitation and ING antibodies to blots confirm that interactions occur between lamin A and the full-length and LID forms of ING1. (c) The lamin interaction domain of ING1 binds to the N terminus of lamin A. Myc-tagged constructs encoding the indicated fragments of lamin A were co-transfected into HEK293 cells with the LID+NLS and lysates were immunoprecipitated with ING1 antibodies. Amino acids 1–406 of lamin A were sufficient for binding to ING1. ExactaCruz secondary antibody that does not recognize denatured IgG bands was used in this experiment. (d) Associations between endogenous ING1 and lamins in
normal Hs68 human fibroblasts and HGPS fibroblasts. The Prog1 HGPS cell strain showed very weak associations between ING1 and lamins and ING1 failed to interact with the truncated lamin A Δ50 (progerin) expressed from one allele in this strain. The actin blot served as a loading control. (e) The LID+NLS polypeptide binds lamin A but not progerin or lamin C in vitro. For this far western blotting assay, His-tag purified LID+NLS protein was used to probe membranes containing lysates from the indicated cell types that were electrophoresed, blotted and renatured. The lower panels show lamin A expression and Commassie Brilliant Blue (CBB) staining of cell lysates. (f) ING1 levels in total (T), cytoplasmic (C) and nuclear (N) fractions of syngeneic wild-type (LMNA+/+) and lamin A knockout murine fibroblasts. Blotting for lamin A and histone H2B confirmed the knockout status of LMNA–/– cells and the purity of cytoplasmic and nuclear fractions. CBB shows total protein in samples.
nature cell biology volume 10 | number 11 | november 2008 1335
ING1. Second, we asked whether blocking lamin A–ING1 interac-tions would affect apoptosis. ING1 lacking LID (ING1∆LID) had a smaller effect on apoptosis than intact ING1, whereas transfection of the LID was almost as effective as intact ING1 in inducing apoptosis (Supplementary Information, Fig. S2). Overexpressing ING1 efficiently induced apoptosis in LMNA+/+, but not in LMNA–/– cells, as assayed by TUNEL (Fig. 4c, d) and sub-G1 DNA content (Fig. 4e). Furthermore, ING1∆LID was less effective in inducing apoptosis, whereas the 3×LID construct was effective in LMNA+/+ but not in LMNA–/– cells (Fig. 4f). ING1 protein lacking the LID region also showed decreased nuclear localization, compared with intact ING1 (Fig. 4g), possibly accounting for its reduced effectiveness in inducing apoptosis.
HGPS and LMNA–/– cells show changes in nuclear envelope integrity and nuclear architecture10,21. ING1 expression induced a flattened cell phenotype, which was not seen with the ING1ΔLID construct in which irregularities in nuclei were clear (Fig. 5a). Expression of the LID+NLS protein also induced profound changes in nuclear architecture, including lobulation of nuclei. Quantification of these effects using a blind experi-mental protocol showed morphological changes, including thickening of the nuclear lamina, irregular nuclear shapes and multiple nuclei in
57% of LID transfected cells, but only 9% of cells transfected with con-trol vector (Fig. 5b, left panel). As ING1 altered nuclear morphology and induced chromatin condensation (arrow in Fig. 5a), reminiscent of senescence and early stages of apoptosis28, we tested the ability of ING constructs to induce the formation of multiple nuclei in a time-course experiment designed to minimize changes due to apoptosis and to sub-tle, secondary effects. By this measure, the LID was even more effective in inducing morphological abnormalities at a highly significant rate, compared with the wild-type (P < 0.0005) and LID-deleted forms of ING1 (Fig. 5b, right panel).
HGPS and LMNA–/– cells show envelope lobulation and loss of het-erochromatin at the nuclear membrane periphery22,23 and ING proteins alter chromatin structure8. Comparison of the nuclear envelope of nor-mal human fibroblasts with HGPS fibroblasts, using electron micro-scopy, showed that the nuclear envelopes of HGPS cells were indeed abnormal and that overexpression of ING1ΔLID induced marked irregularities in the shape of the nucleus (Supplementary Information, Fig. S3). Examination of the structure and distribution of peripheral heterochromatin in cells expressing ING1ΔLID showed that it caused discontinuities in the distribution of peripheral heterochromatin,
b
a LMNA+/+ Embryonic stem cells
Lamin A
Lamin ALamin A
DAPI
DAPIDAPI
ING1
ING1
Merge
Merge
LMNA–/–
LMNA–/–
+Lamin A
LMNA–/–
+Lamin A
Hs68
Prog1
Prog2
ING1
ING1 DAPI
d
c
Figure 3 ING1 localization in different model cell systems. (a) Fluorescent localization of ING1 and DNA in wild-type and LMNA–/– mouse fibroblasts. (b) Colocalization study of ING1 and lamin A in Hs68, Prog1 and Prog2 fibroblasts. Staining for ING1 (red), lamin A (green) and DNA (blue), and deconvolution of images show primarily nuclear (Hs68) and mixed nuclear and cytoplasmic (Prog1 and Prog2) localization of ING1. (c) The R1 line of murine ES cells was
stained for the expression of lamin A (green), ING1 (red) and DNA (blue). The arrow indicates an ES cell that has begun to differentiate and express lamin A, and also shows nuclear accumulation of ING1. (d) Re-introduction of lamin A in LMNA–/– cells by ectopic expression. Arrows highlight individual cells showing differing lamin A staining patterns, with cells expressing lamin A localizing ING1 to the nucleus. Scale bars represent 20 µM (a, c, d) and 50 µM (b).
1336 nature cell biology volume 10 | number 11 | november 2008
compared with cells overexpressing green fluorescent protein (GFP). This is consistent with previous reports regarding alterations in HGPS cells21–23 and the roles of ING112 and ING213 in cell senescence. In some
cells overexpressing ING1ΔLID, loss of peripheral heterochromatin was evident (arrows in Supplementary Information, Fig. S1e) and dif-ferent degrees of distension of the outer nuclear membrane were seen
c
LMNA+/+
LMNA–/–
+GFP
+GFP–ING1
+GFP
+GFP–ING1
TUNELGFPDAPI
d
GFP GFP–ING1 GFP GFP–ING1
TUN
EL
pos
itive
cel
ls
(per
cent
age)
LMNA+/+ LMNA–/–
25
20
15
10
5
0
e
GFP GFP–ING1 GFP GFP–ING1
LMNA+/+ LMNA–/–S
ub-G
1 D
NA
con
tent
(per
cent
age)
25
20
15
10
5
0
V
f
TUN
EL
pos
itive
cel
ls
(per
cent
age)
LMNA+/+ LMNA–/–
25
30
20
15
10
5
0ING1 ∆LID 3xLID V ING1 ∆LID 3xLID
g
ING1
ING1
ING1∆LID
ING1∆LID
DAPI
a
LMNA+/+ LMNA–/–
LMNA+/+ LMNA–/–
32.5
47.5
83
ING1
Rb
Actin
ING
1lev
el
3.0
2.5
2.0
1.5
1.0
0
ING
1lev
el
3.0
2.5
2.0
1.5
1.0
0Hs68 Prog 1 Prog 2 Prog 2
Rb
leve
l
4.0
3.0
2.0
1.0
0Hs68 Prog 1 Prog 2 Prog 2
b
32.5
47.5
83
ING1
Rb
Actin
Hs68 Prog1 Prog 2 Prog 5Mr(K)
Figure 4 Lamin A affects ING1 levels and biological activity. (a) Lysates from wild-type and syngeneic LMNA–/– cells were probed for expression of ING1 and Rb, which was previously reported to be reduced by about 8-fold in lamin A knockout cells19. The graph shows that in three independent experiments, ING levels, as estimated by scanning densitometry, were reduced by an average of 3-fold in LMNA–/– cells (mean ± s.d., P < 0.02) (b) Lysates from normal diploid fibroblasts and different strains of HGPS fibroblasts probed for ING1 and Rb expression showed that both ING1 and Rb levels were markedly reduced in HGPS cells, compared with control fibroblasts (mean ± s.d., P < 0.05 (ING1); P < 0.005 (Rb)). (c) LMNA–/– or LMNA+/+ cells were transfected with GFP or a GFP–ING1 co-expression construct and analysed for GFP expression (green) or apoptosis (red) using the TUNEL assay 36 h
after transfection. (d) Data from three independent experiments as shown in panel c (mean ± s.d., P < 0.005 for LMNA+/+ versus LMNA–/–). (e) Assays as in panel d, but analysed by flow cytometry (mean ± s.d., P < 0.05). (f) Cells with or lacking lamin A expression, transfected with the indicated expression constructs and analysed for apoptosis 36 h later using the TUNEL assay. ING1 did not effectively induce apoptosis in the absence of lamin A expression, although some induction was seen for ING1 in LMNA–/– cells, regardless of the absence or presence of the LID, suggesting a major lamin A-dependent and a minor lamin A-independent component (mean ± s.d., n = 3, P < 0.001). (g) Transfected ING1 localizes primarily to the nucleus, whereas ING1 lacking the LID localizes to both the nuclear and cytoplasmic compartments. Scale bars represent 120 µM (c) and 20 µM (g).
nature cell biology volume 10 | number 11 | november 2008 1337
(horizontal arrow in Supplementary Information, Fig. S1h). Expression of constructs interfering with ING1–lamin A interaction also induced expression of senescence-associated β-galactosidase (SA-βGal), simi-lar to staining observed in senescent, but not in low passage (young) Hs68 fibroblasts (Fig. 5c). This is consistent with the altered chromatin structure seen in senescing cells29 and in cells with mutant lamin A23,30. Examination of p21 and p16 cyclin-dependent kinase inhibitor levels further confirmed that the LID and 3×LID were inducing a phenotype reminiscent of senescence (Fig. 5d).
These data identify a LID unique to the ING family of chromatin inter-acting proteins that mediates a direct interaction between lamin A and ING1. The LID and lamin A interact through the amino terminus of lamin A and although insolubility of sub-fragments of this region did not allow precise mapping of the region on lamin A, it is in the rod domain through which the lamins form dimers16. Progerin, which lacks the final 50 amino acids of lamin A does not bind ING1 at physiological levels, whereas lamin C, which differs from lamin A in the last 97 amino acids (567–664 of lamin A) is recovered in immunoprecipitations. The reason
for this is not obvious, given that interaction was mapped to the amino terminus of lamin A. This observation suggests that the tail region of the lamins has profound effects on the ability of the rod domain to inter-act with ING1, and suggests the possibility that ING proteins may bind dimers or multimers, rather than monomers of lamin in vivo. However, far western analyses (Fig. 2e) suggest that interaction with monomers is possible in vitro. Alternatively, a lack of progerin recovery in immuno-precipitation-western analyses may be due to differential solubility of progerin and lamin A. Studies using mouse ES and LMNA–/– cells con-firm the HGPS results and indicate that lamin anchors the ING1 protein within the nucleus. This is consistent with ING1 localization through mitosis, which mirrors the localization of lamin A31,32 (Supplementary Information, Fig. S4), indicating that lamin A and ING1 occupy the same cellular compartments.
The PHD regions of ING proteins interact specifically with methylated forms of histone H3 (refs 1–4) and ING proteins function as stoichiometric components of HAT and HDAC complexes7. Our data showing that lamin A binds and targets ING1, and regulates its cellular levels and activities,
c
db
1 2 3 4 1 2 3 4
a
1 2 3 4 5
1
GFP
LID
ING1∆LID ING1∆LID
ING1
ING1
GFP DAPI
DAPI
DAPI
DIC
4
52
3 SA
-βG
al (p
erce
ntag
e)
1009080706050403020100
Hs68293
p21
p16
Actin
16.5
16.5
47.5
Vector LID 3xLID Vector LID 3xLID
Ab
norm
al n
ucle
i (p
erce
ntag
e)
Mul
tiple
nuc
lei
(per
cent
age)
P < 0.0005
P < 0.005
P < 0.05
P < 0.0005
P < 0.0005
P < 0.005
406070
4050
20100
30
30
20
10
Prog1 (29)
Hs68 (95 MPD) MPD)
Hs68 (29 MPD) GFP Control
LID+NLS
Figure 5 Overexpression of ING constructs affects nuclear morphology and induces senescence. (a) ING1 constructs were co-transfected with GFP into HEK293 cells and morphology of GFP-positive cells was examined 48 h later. Transfection with intact ING1 initially induced a flattened cell morphology as seen in the DIC image and in formation of pycnotic nuclei and other features typical of apoptosis, highlighted by the arrow in the expanded panel to the right and previously reported by many groups8. Transfection with the ING1ΔLID construct resulted in abnormal nuclear morphology including multi-nucleated cells but little indication of highly condensed regions of chromatin as seen for ING1. (b) The percentage of transfected cells showing abnormal or multiple nuclei was estimated by visual inspection 48 h after transfection of HEK293 cells with constructs encoding GFP (1), ING1 (2), ING1ΔLID (3) or the LID+NLS (4). Abnormal nuclei were defined as those having irregular
shapes with multi-lobulation or multiple nuclei and did not include cells with condensed chromatin alone, as shown for ING1 transfection in a (data are mean ± s.d.). (c) Subconfluent primary Hs68 cells at low and high passage, the Prog1 HGPS strain (29 MPD), HEK293 cells transfected with a control GFP construct or HEK293 transfected with the LID+NLS construct were fixed and stained for 28 h to detect SA-βGal activity as a measure of senescence. Scale bars are 80 µm. The graph shows the percentage of cells expressing SA-βGal where n = 100 for each condition. (d) Established HEK293 or primary Hs68 human fibroblast cells transfected with an empty vector, the LID construct or the triple LID construct were harvested 36 h later and protein lysates were blotted for the expression of the p21 or p16 senescence markers. Blotting with actin served as a loading and transfer control. Scale bars represent 50 µM (left DIC) or 20 µM (DAPI) (a) and 120 µM (g).
1338 nature cell biology volume 10 | number 11 | november 2008
indicate that ING proteins function as bridges between chromatin and the nuclear lamina. This is consistent with ING proteins contributing to altered lamin function in HGPS, which shows altered histone H3 modification on lysine residues (H3K9, 20, 27; ref. 30) and in muscular dystrophy, where lamin perturbation causes discontinuities in peripheral heterochromatin10. Supplementary Information, Fig. S5 shows a model of how ING tethering in the nucleus leads to normal function in HAT and HDAC complexes. Supplementary Information, Fig. S6 shows how expressing normal ING1, the LID and ING1∆LID could lead to disruption of the ING1–lamin A interaction and similar phenotypes, respectively. ING1 increases histone H3 and H4 acetylation25 and a hallmark of HGPS is loss of peripheral het-erochromatin22,23. In addition to effects on ING1, lamin A also localizes and regulates the levels of Rb in LMNA–/– cells19, and we show a reduction in HGPS cells (Fig. 4b). Rb is another tumour suppressor with strong links to chromatin regulation, cell senescence and tumorigenesis. These observa-tions suggest that altered levels of lamin A, which can induce a progeroid phenotype independently33, and expression of mutant forms of lamin A may disrupt chromatin structure and affect apoptosis by effects on Rb and ING1. This may occur through the LID, a sequence unique to the ING proteins, as both ING1 and Rb associate in the mSIN3 HDAC complex6,34. Determining whether there are relationships between expression of the many family members and splicing isoforms of ING proteins35, their rela-tive affinities for lamin A and effects on histone acetylation, will further clarify the roles of the lamin A–ING protein interaction in the process of cell senescence12,13, heterochromatin formation29 and linkage to HGPS and the nuclear lamina30.
METHODSMass spectrometry. The His-tagged ING1–TriLID+NLS construct was trans-fected into HEK293 cells using calcium phosphate. The pcDNA4/TO/Myc–HisA vector was used as a negative control in parallel. After 24 h, cells were lysed under non-denaturing conditions in 800 µl of lysis buffer containing 10 mM imida-zole. Cell lysates were then loaded onto Ni-NTA spin columns (Qiagen) equili-brated with lysis buffer. After three washes with 15 mM imidazole, His-tagged protein complexes were eluted with elution buffer containing 250 mM imida-zole. Buffer exchange was performed before submitting samples for LC/MS/MS using Microcon columns (Millipore). LC/MS/MS analyses were performed by the Southern Alberta Mass Spectrometry (SAMS) Facility at the University of Calgary and the database search engine Mascot was used to identify binding partners (http://www.matrixscience.com/home.html).
Flow cytometry. For analysis of apoptosis, HEK293, LMNA–/– and LMNA+/+, MEF cells were co-transfected with GFP and various ING1b constructs, and collected on ice 48 h later. After centrifugation, pelleted cells were resuspended in 500 µl PBS, fixed in 500 µl of ice-cold 2% formaldehyde at 4 °C for 1 h and resuspended in 500 µl PBS. Samples were permeabilized with 1.5 ml ice-cold 95% ethanol, treated with 1 mg ml–1 RNaseA in PBS for 30 min at room temperature and DNA was stained with 2.5 µg propidium iodide. The cell cycle was analysed using a Becton-Dickinson FACScan flow cytometer at the University of Calgary Flow Cytometry Core Facility.
Cell culture and fractionation. Human Hs68 primary diploid fibroblasts (ATCC CRL-1635), HEK293 HEK cells (ATCC CRL-1573), mouse ES cells (R1, a gift from E. Rattner and D. Rancourt (University of Calgary, Canada), skin fibroblasts from Hutchinson-Gilford progeria syndrome (HGPS) patients (Coriell AG11513/Prog1, AG00989/Prog2, AG06917/Prog5), SK-N-SH cells (ATCC HTB-11) and murine LMNA+/+ and LMNA–/– MEFs were used in this study with culture condi-tions as described previously10,23. Unless otherwise specified, the mean population doubling (MPD) levels for HS68 cells used in this study were between 35 and 38. The Prog1 HGPS fibroblasts were used at passages 18–20; Prog2 fibroblasts at passages 25–28, Prog5 at 21–25 and MEFs at passages 19–22. For cell fractionation assays, cytoplasmic and nuclear components of cells were separated in 0.5% NP-40 buffer with an 18-G needle. After centrifugation, supernatants were collected for
cytoplasmic proteins, and pellets containing nuclei were extracted with NP-40 buffer to minimize contamination with cytoplasm. Nuclei were lysed with SDS sample buffer to generate nuclear fractions.
Immunoprecipitation-western and far western blotting. Cells were lysed and sonicated on ice in 1 ml RIPA buffer12 containing protease inhibitor cocktail tablets (Roche). For initial analyses of protein complexes, a similar lysis buffer was used except that it also included 0.25% each of the non-ionic detergents NP40, Tween 20 and Triton X-100, with a final wash on ice in buffer lacking detergents to minimize interference in MS. Lysates were clarified by centrifugation and 30 µl aliquots were used to test expression levels of proteins. Immunoprecipitation was performed by incubating Protein G–sepharose beads saturated with fresh anti-ING1, anti-Myc monoclonal antibodies (SAS facility), or lamin A/C antibody (Santa Cruz sc-7292), with cell lysates for 4 h at 4 °C. For analysis of ING1–ING5 interactions with lamin A, chicken anti-ING antibodies were raised against unique regions of each ING protein, verified not to cross react and were detected with rabbit anti-IgY. The mouse anti-Flag antibody M2 (Sigma) was used as a non-specific IgG control. Anti-lamin A/C (Santa Cruz sc-7292), anti-lamin A (Santa Cruz sc-20680), anti-actin (Santa Cruz sc-32251), and anti-histone H2B (Cell Signaling 2722) were used for western blotting at dilutions of 1:300, 1:500, 1:200, and 1:500 respectively.
For far western blotting, cell lysates were electrophoresed on denaturing SDS–PAGE gels and transferred to nitrocellulose. Membranes were washed with PBS to remove SDS, rinsed in binding buffer (10% glycerol, 0.1 M NaCl, 0.02 M Tris pH 7.6, 1 mM EDTA and 0.1% Tween-20) and blocked in 2% non-fat milk in binding buffer overnight at 4 °C. Membranes were subsequently incubated with Ni-NTA-purified His-tagged LID+NLS protein (10 µg ml–1) in binding buffer with 2% non-fat milk for 1 h at room temperature and washed with PBS-Tween 20. Anti-Myc-tagged antibody 9E10 and goat anti-mouse HRP-conjugated sec-ondary antibody were used to detect bound protein.
Immunofluorescence and electron microscopy. Cells on glass coverslips were transfected two days before fixation. Cells at 70–80% confluence were fixed with 3.5% paraformaldehyde for 10 min and permeabilized with 0.5% Triton X-100 (Sigma). Blocking was with 3.5% BSA in PBS for 1 h at room temperature. Human and mouse ING1 proteins were visualized after incubation for 1 h at 37 °C with the Cab1 mouse monoclonal antibody that recognizes the NLS of ING1 and lamin A rabbit polyclonal antibody (Santa Cruz sc-20680; 1:200). Secondary antibod-ies were goat anti-rabbit Alexa Fluor 488 and goat anti-mouse Alexa Fluor 568 (1:1000; Invitrogen). After washing, mounting (Biomeda) and staining with the DNA-specific dye DAPI (1 µg ml–1), cells were visualized using a Zeiss Axiovert 200 microscope and AxioVision 4.5 software. Digital deconvolution was with VayTek Microtome software. TUNEL assays were performed according to recom-mendations from the supplier (TMR red in situ death detection kit, Roche).
For transmission electron microscopy, cells grown in 35-mm tissue culture dishes and treated as described in the text were fixed with 2% glutaraldehyde in 0.2 M Na cacodylate buffer for 30 min. Samples were then washed twice in cacodylate buffer, post-fixed for 1 h with 2% osmium tetroxide in cacodylate buffer, dehy-drated in ethanol and embedded in Polybed 812 resin. Thin cross-sections were cut with a diamond knife on a Reichert ultramicrotome. Sections were stained with uranyl acetate and lead citrate, and examined on a Hitachi H-7000 TEM micro-scope at the Microscopy and Imaging Facility of the University of Calgary.
Note: Supplementary Information is available on the Nature Cell Biology website.
AcKnowlEdgEMEntSWe thank E. White for lamin constructs, J. Quarrie and D. Rancourt for ES cells, the SACRI Microscopy Facility for cell imaging, D. Boland and S. Law of the SAS for ING antibodies, D. Schreimer for LC/MS/MS advice and L. Robertson for help with flow cytometry. This study was supported by grants from the CIHR to K.R., from the NSERC to J.B.R. and to B.B. from the NIH. K.R. is a Scientist of the AHFMR, M.A.S. holds studentships from AHFMR and the Alberta Cancer Board (ACB) and H.S. and P.B. hold an ACB studentship.
coMPEting FinAnciAl intEREStSThe authors declare no competing financial interests.
Published online at http://www.nature.com/naturecellbiology/ reprints and permissions information is available online at http://npg.nature.com/reprintsandpermissions/
nature cell biology volume 10 | number 11 | november 2008 1339
1. Shi, X. et al. ING2 PHD domain links histone H3 lysine 4 methylation to active gene repression. Nature 442, 96–99 (2006).
2. Pena, P. V. et al. Molecular mechanism of histone H3K4me3 recognition by plant homeodomain of ING2. Nature 442, 100–103 (2006).
3. Martin, D. G. et al. The Yng1p plant homeodomain finger is a methyl-histone binding module that recognizes lysine 4-methylated histone H3. Mol. Cell Biol. 26, 7871–7879. (2006).
4. Palacios, A. et al. Solution structure and NMR characterization of the binding to meth-ylated histone tails of the plant homeodomain finger of the tumour suppressor ING4. FEBS Lett. 580, 6903–6908 (2006).
5. Loewith, R, Meijer, M, Lees-Miller, S, Riabowol, K, & Young, D. Three yeast proteins related to the human candidate tumor suppressor p33ING1 are associated with histone acetyltransferase activities. Mol. Cell Biol. 20, 3807–3816 (2000).
6. Kuzmichev, A, Zhang, Y, Erdjument-Bromage, H, Tempst, P, & Reinberg, D. Role of the Sin3-histone deacetylase complex in growth regulation by the candidate tumor suppressor p33 (ING1). Mol. Cell Biol. 22, 835–848 (2002).
7. Doyon, Y. C. et al. ING tumor suppressor proteins are critical regulators of chromatin acetylation required for genome expression and perpetuation. Mol. Cell 21, 51–64 (2006).
8. Soliman, M. A. & Riabowol, K. After a decade of study-ING, a PHD for a versatile family of proteins. Trends Biochem Sci. 32, 509–519 (2007).
9. Eriksson, M. et al. Recurrent de novo point mutations in lamin A cause Hutchinson-Gilford progeria syndrome. Nature 423, 293–298 (2003).
10. Burke, B. & Stewart, CL. The laminopathies: the functional architecture of the nucleus and its contribution to disease. Ann. Rev Genomics Hum Genet. 7, 369–405 (2006).
11. He, G. H., Helbing, C. C., Wagner, M. J., Sensen, C. W. & Riabowol, K. Mol. Biol Evol. 22, 104–116 (2005).
12. Garkavtsev, I, & Riabowol, K. Extension of the replicative life span of human diploid fibroblasts by inhibition of the p33ING1 candidate tumor suppressor. Mol. Cell Biol. 17, 2014–2019 (1997).
13. Pedeux, R. et al. ING2 regulates the onset of replicative senescence by induction of p300-dependent p53 acetylation. Mol. Cell Biol. 25, 6639–6648 (2005).
14. Scott, M. et al. UV-induced binding of ING1 to PCNA regulates the induction of apop-tosis. J. Cell Sci. 114, 3455–3462 (2001).
15. Wang, J, Chin, M.Y, & Li, G. The novel tumor suppressor p33ING2 enhances nucleotide excision repair via inducement of histone H4 acetylation and chromatin relaxation. Cancer Res. 66, 1906–1911 (2006).
16. Zastrow, M. S., Vicek, S. & Wilson, K. L. Proteins that bind A-type lamins: integrating isolated clues. J. Cell Sci. 117, 979–987 (2004).
17. Taniura, H., Glass, C. & Gerace, L. A chromatin binding site in the tail domain of nuclear lamins that interacts with core histones. J. Cell Biol. 131, 33–44 (1995).
18. Ozaki, T. et al. Complex formation between lamin A and the retinoblastoma gene prod-uct: identification of the domain on lamin A required for its interaction. Oncogene 9, 2649–2653 (1994).
19. Johnson, B. R. et al. A-type lamins regulate retinoblastoma protein function by promot-ing subnuclear localization and preventing proteosomal degradation. Proc. Natl Acad Sci. USA 101, 9677–9682 (2004).
20. Helsens, K., Martens, L., Vandekerckhove, J. & Gevaert, K. MascotDatfile: an open-source library to fully parse and analyse MASCOT MS/MS search results. Proteomics. 7, 364–366 (2007).
21. Broers, J. L., Ramaekers, F. C., Bonne, G., Yaou, R. B. & Hutchison, C. J. Nuclear lamins: laminopathies and their role in premature ageing. Physiol Rev. 86, 967–1008 (2006).
22. Sullivan, T. et al. Loss of A-type lamin expression compromises nuclear envelope integrity leading to muscular dystrophy. J. Cell Biol. 147, 913–920 (1999).
23. Goldman, R. D. et al. Accumulation of mutant lamin A causes progressive changes in nuclear architecture in Hutchinson-Gilford progeria syndrome. Proc. Natl. Acad. Sci. U. S.A 101, 8963–8968 (2004).
24. Coles, A. H. et al. Deletion of p37Ing1 in mice reveals a p53-independent role for Ing1 in the suppression of cell proliferation, apoptosis, and tumorigenesis. Cancer Res. 67, 2054–2061 (2007).
25. Vieyra, D. et al. Human ING1 Proteins Differentially Regulate Histone Acetylation. J. Biol Chem. 277, 29832–29839 (2002).
26. Taverna, S. D. et al. Yng1 PHD finger binding to H3 trimethylated at K4 promotes NuA3 HAT activity at K14 of H3 and transcription at a subset of targeted ORFs. Mol. Cell 24, 785–796 (2006).
27. Nouman, G. S. et al. Down regulation of nuclear expression of the p33 (ING1b) inhibi-tor of growth protein in invasive carcinoma of the breast. J. Clin Pathol. 56, 507–511 (2003).
28. Vieyra, D. et al. Ing1 isoforms differentially affect apoptosis in a cell age-dependent manner. Cancer Res. 62, 4445–4452 (2002).
29. Narita, M. et al. Rb-mediated heterochromatin formation and silencing of E2F target genes during cellular senescence. Cell 113, 703–716 (2003).
30. Shumaker, D. K. et al. Mutant nuclear lamin A leads to progressive alterations of epigenetic control in premature aging. Proc. Natl. Acad. Sci. U. S. A 103, 8703–8708 (2006).
31. Broers, J. L. et al. Dynamics of the nuclear lamina as monitored by GFP-tagged A-type lamins. J. Cell Sci. 112, 3463–3475 (1999).
32. Moir, R.D, Yoon, M., Khoun, S. & Goldman, R. D. Nuclear Lamins A and B1: Different Pathways of Assembly during Nuclear Envelope Formation in Living Cells. J. Cell Biol. 151, 1155–1168 (2000).
33. Candelario, J., Sudhakar, S., Navarro, S., Reddy, S. & Comai, L. Perturbation of wild-type lamin A metabolism results in a progeroid pgenotype. Aging Cell 7, 355–367 (2008).
34. Lai, A. et al. RBP1 recruits the mSIN3-histone deacetylase complex to the pocket of retinoblastoma tumor suppressor family proteins found in limited discrete regions of the nucleus at growth arrest. Mol. Cell Biol. 21, 2918–32 (2001).
35. Walzak, A. A., Veldhoen, N, Feng, X, Riabowol, K. & Helbing, C. C. Expression profiles of mRNA transcript variants encoding the human inhibitor of growth tumor suppressor gene family in normal and neoplastic tissues. Exp. Cell Res. 314, 273–85 (2008).
1340 nature cell biology volume 10 | number 11 | november 2008
Figure S1 Co-precipitation of ING1-ING5 with Lamin A. Lysates from HEK293 cells transfected with ING1-ING5 or GST expression constructs were immunoprecipitated with preimmune non-specific antibody (NS Ab) or
anti-lamin A (α-LMNA) and precipitates were blotted with chicken a, α-ING1 b, α-ING2 c, α-ING3 d, α-ING4 e, α-ING5 or f, α−ING1 and visualized with α-IgY-HRP.
Figure S2 Interaction with lamins regulates ING1 induction of apoptosis. GFP and ING1 constructs were cotransfected into HEK293 cells, which were harvested 48h later and stained with propidium iodide. Cells with a sub-G1 DNA content (a measure of apoptosis) are highlighted by the arrows. The LID induced apoptosis by itself at a similar level to wild type ING1, while ING1ΔLID promoted apoptosis less effectively. Induction of apoptosis
by increasing linkage between lamin A and HAT/HDAC complexes (ING1 transfection) or by blocking linkage (ING1ΔLID and LID transfections) is consistent with levels of lamin A being important for normal cell metabolism since both decreases21,22 and increases35 in lamin A levels induce sensitivity to apoptosis. Values presented are the average of three independent assays.
Figure S3 Effects of disrupting ING1-Lamin A interactions on nuclear membrane structure. EM analyses of Hs68, HGPS and ING1-transfected cells. a-b, Compared to Hs68 (a) cells, invaginations highlighted by the arrowheads were frequently observed in the nuclear envelope of Prog1 fibroblasts (b) as previously described for HGPS cells22-23, which were also larger in size suggesting that they were entering senescence despite being at a lower passage level that the other primary fibroblast strains. c-h, HEK 293 cells transfected with GFP (c,f) or ING1ΔLID (d,e,g) were fixed and stained 48h after transfection. In contrast to the GFP control which showed morphology similar to untransfected cells with a largely continuous layer of peripheral heretochromatin (compare panels a & c), the LID deletion ING1 construct frequently induced nuclear membrane invagination, lobulation or
multiple nuclei, consistent with disruption of the lamin A-ING1 interaction and subsequent altering of the distribution of peripheral heterochromatin contributing to nuclear membrane instability. Panels f-h show higher magnification micrographs that highlight discontinuities in the peripheral heterochromatin near the nuclear envelope (arrowheads) that have been reported in HGPS23 and in lamin A knockout22 cells. The arrows in panels d & e compare regions of the NE showing differences in peripheral chromatin density and the horizontal arrow in panel h shows distention of the outer nuclear membrane. All analyses were initially done using a blind experimental protocol in which microscopy was done and recorded for coded samples. Magnifications used in panels a-d are the same and the scale bar equals 2µm. The scale bar = 2µm for panel e and 0.3 µm for panels f-h.
Figure S4 Subcellular localization of ING1 during mitosis. Fibroblasts were fixed and stained for ING1 proteins (green) and DNA (DAPI, blue). a, ING1 proteins are nuclear during interphase and b, early prophase but start dissociating from nuclear structures in c, pro-metaphase and remain distributed throughout the cell in d, metaphase and e, anaphase. They begin reassociating with nuclear structures in f, telophase. This is consistent
with ING1 binding lamin A/C since transit from the nucleus to cytoplasm it is similar to the dynamics of lamin A/C depolymerization. Entry into the reforming nucleus appears to slightly precede lamin A/C repolymerization during mitosis which is believed to occur after cytokinesis and staining appears in particulate sites consistent with binding a subpopulation of lamin A. The scale bar represents 5 µm.
Figure S5 The subcellular localization of ING1 by lamin A potentiates ING1 function. Model for how tethering of ING proteins by Lamin A is needed to stabilize ING proteins and localize them in the nucleus for use as stoichometeric components of HAT and HDAC complexes. Nuclear transport and chromatin tethering by interaction with lamin A is followed by recruitment of acetylation and/or deacetylation complexes and altering histone acetylation levels and chromatin structure. If this function is altered by loss of Lamin A interaction and subsequently decreased levels
of ING and Rb proteins, the expected phenotype would be an altered state of chromatin compaction and distribution, with subsequent altered susceptibility to apoptosis, both of which are seen in HGPS cells and in response to altered levels of both Rb and ING proteins. ING binding to trimethylated histone H3, stoichometric residence in HAT and HDAC complexes, ING1 phosphorylation, binding to 14-3-3 and exit from the nucleus and transport by karyopherins have all been previously described (reviewed in reference 8).
Figure S6 Disrupting ING1-Lamin A interactions alters histone acetylation. a, under normal conditions ING1 binds to lamin A, promoting interaction with trimethylated histone H3K4. ING1 then recruits HDAC complexes to modify adjacent lysine residues such as H3K14. b, Overexpression of ING1 to suprastoichometric levels results in individual molecules of ING1 binding to lamin A/HDAC complexes via the LID and to histone H3 via the PHD, resulting in loss of HDAC
targeting. Increased acetylation of both histone H3 and histone H4 has been reported upon overexpression of ING1 by needle microinjection25. c, Expression of the LID region blocks the ability of ING1 to interact with lamin A, resulting in loss of HDAC targeting. d, ING1 lacking the LID partially localizes to the nucleus and that fraction that does partially prevents interaction between ING1 and lamin A/HDAC complexes resulting in a similar but less obvious phenotype.
