(Received for publication, March 6, 1997, and in revised form, June 20, 1997)
From the Nuclear Signaling Laboratory, Division of Biochemistry and Molecular Biology, John Curtin School of Medical Research, Canberra City, A.C.T. 2601, Australia
The retinoblastoma (RB) tumor suppressor is a
nuclear phosphoprotein important for cell growth control and able to
bind specifically to viral oncoproteins such as the SV40 large tumor
antigen (T-ag). Human RB possesses a bipartite nuclear localization
sequence (NLS) consisting of two clusters of basic amino acids within
amino acids 860-877, also present in mouse and Xenopus
homologs, which resembles that of nucleoplasmin. The T-ag NLS
represents a different type of NLS, consisting of only one stretch of
basic amino acids. To compare the nuclear import kinetics conferred by
the bipartite NLS of RB to those conferred by the T-ag NLS, we used
-galactosidase fusion proteins containing the NLSs of either RB or
T-ag. The RB NLS was able to target
-galactosidase to the nucleus
both in vivo (in microinjected cells of the HTC rat
hepatoma line) and in vitro (in mechanically perforated HTC
cells). Mutational substitution of the proximal basic residues of the
NLS abolished nuclear targeting activity, confirming its bipartite
character. Nuclear accumulation of the RB fusion protein was
half-maximal within about 8 min in vivo, maximal levels
being between 3-4-fold those in the cytoplasm, which was less than
50% of the maximal levels attained by the T-ag fusion protein, while
the initial rate of nuclear import of the RB protein was also less than
half that of T-ag. Nuclear import conferred by both NLSs in
vitro was dependent on cytosol and ATP and inhibited by the
nonhydrolyzable GTP analog GTP
S. Using an ELISA-based binding assay,
we determined that the RB bipartite NLS had severely reduced affinity,
compared with the T-ag NLS, for the high affinity heterodimeric
NLS-binding protein complex importin 58/97, this difference presumably
representing the basis of the reduced maximal nuclear accumulation and
import rate in vivo. The results support the hypothesis
that the affinity of NLS recognition by NLS-binding proteins is
critical in determining the kinetics of nuclear protein import.
All passive and active transport into and out of the nucleus
occurs through the nuclear pore complex
(NPC)1 (1-3). Proteins
larger than 45 kDa require a nuclear localization sequence (NLS) (4) to
be targeted to the nucleus. NLSs are defined as the sequences
sufficient and necessary for nuclear import of their respective
proteins (5-7) and are generally functional in targeting heterologous,
normally cytoplasmic proteins to the nucleus. NLS-dependent
protein transport can be divided into two steps. The first is
energy-independent and involves recognition and targeting of the
NLS-containing protein to the NPC by a heterodimeric protein complex
consisting of importin 58/97 (8, 9) or /
(10, 11). The second,
energy-dependent step is the translocation of the
NLS-bearing protein through the NPC into the nucleus (12, 13) and
requires GTP hydrolysis mediated by the GTP-binding protein Ran/TC4
(14-16) and the interacting factor p10/NTF2 (17-19).
We were interested in the nuclear import kinetics of tumor suppressor proteins, one of which is the retinoblastoma protein (p110Rb or RB). RB is frequently altered or deleted in a number of tumors and tumor cell lines, including those derived from retinoblastomas, osteosarcomas, breast cancers and prostate, bladder, and small lung carcinomas (20). It is nuclear (21) with a molecular mass of 105-115 kDa, depending on its phosphorylation state (20). Human RB contains a bipartite NLS (amino acids 860-877), conserved in mammalian and amphibian RB proteins (Refs. 22 and 23 and see Table I). Of the two basic types of NLS, bipartite NLSs comprise two clusters of basic amino acids separated by a 10-12-amino acid spacer resembling the NLS of the Xenopus laevis nuclear phosphoprotein nucleoplasmin (Ref. 24 and see Table I), whereas the other type consists of a single short basic amino acid sequence resembling the T-ag NLS (PKKKRKV132; Ref. 5).
|
This study examines the nuclear import kinetics of fusion proteins carrying the RB NLS in vivo and in vitro at the single cell level and compares results to those for proteins carrying the T-ag NLS. We assess the dependence of nuclear import on cytosolic factors, ATP, and GTPase activity and use an ELISA-based binding assay to determine the binding affinity for the NLS-binding heterodimeric protein complex importin 58/97. We find that the bipartite NLS of RB is much less efficient than the T-ag NLS in terms of both the maximal level and rate of nuclear import, as well as exhibiting a much lower affinity for importin. The results are consistent with the tenet that the affinity of NLS binding by importin is critical in determining the kinetics of nuclear protein import.
Isopropyl--D-thiogalactopyranoside and
the detergent CHAPS were from Boehringer Mannheim, and the sulfhydryl
labeling reagent 5-iodoacetamidofluorescein was from Molecular Probes.
Other reagents were from the sources described previously (26,
28-30).
Cells of the HTC rat hepatoma tissue culture (a derivative of Morris hepatoma 7288C) line were cultured as described previously (26, 31, 32).
Plasmids expressing the
RB-Bip--galactosidase (RB-Bip-
-Gal) and
RB-BipMut-
-galactosidase (RB-BipMut-
-Gal) fusion proteins were
derived by oligonucleotide insertion into the SmaI
restriction endonuclease site of the plasmid vector pPR2 (28). The
resultant fusion proteins express RB amino acids 860-877 fused
NH2-terminal to the Escherichia coli
-galactosidase enzyme sequence (amino acids 9-1023), where
RB-Bip-
-Gal retains the wild type sequence, and the NLS mutated
derivative RB-BipMut-
-Gal possesses Thr-Thr in place of the wild
type amino acids Lys-Arg861. The T-ag
-galactosidase
fusion protein (T-ag-CcN-
-Gal) used in the comparative studies
contains T-ag amino acids 111-135, including the CcN motif (comprising
protein kinase CK2 and cyclin-dependent kinase
phosphorylation sites and the NLS) fused amino-terminal to
-galactosidase amino acids 9-1023 (31, 32).
1 mM isopropyl--D-thiogalactopyranoside was
used to induce expression of
-galactosidase fusion proteins in
E. coli. They were purified by affinity chromatography and
labeled with 5-iodoacetamidofluorescein as described previously (31,
32).
Analysis of nuclear import kinetics at the single cell level was performed using either microinjected (in vivo) or mechanically perforated (in vitro) HTC cells in conjunction with confocal laser scanning microscopy (CLSM) (26-33). In the case of microinjection, HTC cells were fused with polyethylene glycol about 1 h prior to microinjection to produce polykaryons (26, 27, 30). Reticulocyte lysate (Promega) was used as the source of cytosol for the in vitro assay (27, 29, 33). Image analysis of CLSM files using the NIH Image public domain software and curve fitting were performed as described (26, 33).
In in vitro experiments where the ATP dependence of
transport was tested, apyrase pretreatment was used to hydrolyze
endogenous ATP in cytosolic extracts (10 min at room temperature with
800 units/ml) and perforated cells (15 min at 37 °C with 0.2 units/ml) (12, 33), and transport assays were then performed in the absence of the ATP-regenerating system (31, 33), which was otherwise
used. In experiments where the dependence of transport on the
GTP-binding protein Ran/TC4 (34) was tested, cytosolic extract was
treated with 850 µM GTPS (nonhydrolyzable GTP analog) for 5 min at room temperature, prior to use in the in vitro
assay (final GTP
S concentration of 300 µM).
Nuclear accumulation was also examined in vitro in the
presence of a volume of 20 mM Tris (pH 7.0)
containing 10% glycerol and 0.25% CHAPS, which results in permeabilization of the nuclear envelope; accumulation under these conditions only results from binding to nuclear components such as
lamins, chromatin, etc.
An ELISA-based binding assay was
used to examine the binding affinity between importin subunits (mouse
importin 58 and 97 glutathione S-transferase (GST) fusion
proteins expressed as described (8, 9, 30)) and RB or T-ag fusion
proteins (30). Briefly, 96-well microtiter plates were coated with
-galactosidase fusion proteins, blocked with bovine serum albumin,
hybridized with increasing concentrations of importin 58-GST or
precomplexed importin 58/97-GST and then successive incubations carried
out with goat anti-GST primary, and alkaline phosphatase-coupled rabbit
anti-goat secondary, antibodies. After the addition of the substrate
p-nitrophenyl phosphate, A405 was
measured at 5-min intervals for 90 min using a plate reader (Molecular
Devices). Values were corrected by subtracting both
A405 at 0 min and A405 in
wells incubated without importin 58/97-GST complex. To assess importin
binding specifically to the NLSs, quantitation was performed in
identical fashion for
-galactosidase itself and the values
subtracted from those for the respective fusion proteins (30).
To correct for differences in coating, the RB and T-ag fusion proteins
were subjected to a parallel -galactosidase ELISA assay using a
-galactosidase-specific monoclonal antibody together with an
anti-mouse alkaline phosphatase-conjugated secondary antibody and
p-nitrophenyl phosphate (30). As above, values were
corrected by subtracting the absorbance at 0 min. Measurements for
importin binding were ultimately corrected for any differences in
coating efficiencies quantified in the
-galactosidase ELISA, to
enable a true estimate of bound importin to be made (30). Fusion
proteins denatured through preincubation for 10 min at 65 °C were
found to exhibit greatly reduced reactivity in the
-galactosidase
ELISA, as well as an over 100-fold increase in the value of the
KD (apparent affinity constant).
Human RB contains a consensus
bipartite NLS that is conserved in X. laevis and mouse RB
(see Table I and Refs. 22 and 23). To
test whether this sequence in human RB is functional, -galactosidase fusion proteins were derived containing either the wild type human RB
NLS (amino acids 860-877, RB-Bip-
-Gal) or a mutated version (RB-BipMut-
-Gal) in which lysine 860 and arginine 861 were
substituted by threonine residues. This mutation abolishes the
bipartite character of the NLS; similar mutations abolish the nuclear
targeting activity of bipartite NLSs such as those of nucleoplasmin
(24), and the yeast transcription factor SWI5 (26) (see Table I). The
nuclear import kinetics of RB-Bip-
-Gal and RB-BipMut-
-Gal were
measured using in vivo (microinjected cells of the HTC rat
hepatoma line) (26, 32) and in vitro (mechanically
perforated HTC cells) (29, 33) nuclear transport assay systems. The
human RB NLS was capable of targeting the heterologous E. coli protein
-galactosidase (476 kDa) to the nucleus in both
assay systems (Figs. 1 and
2). RB-Bip-
-Gal accumulated in nuclei
maximally to levels 3-4 fold those in the cytoplasm (Fig. 1B and 2B;
Table II). Results were similar to those
for
-galactosidase fusion proteins containing the bipartite NLSs
from SWI5 (26) or human interleukin-5 (hIL-5; Ref. 27) in that
RB-Bip-
-Gal accumulated in the nucleus to a significantly lower
extent than T-ag-CcN-
-Gal; the T-ag NLS is clearly a more potent
targeting signal than the bipartite NLS of RB. The initial nuclear
import rate of the RB-Bip-
-Gal in vivo was also
significantly less (p < 0.0005) than that of
T-ag-CcN-
-Gal (rates of 0.49 and 1.12 Fn/c/min,
respectively, (where Fn/c is defined as the ratio of nuclear
to cytoplasmic fluorescence after the subtraction of fluorescence due
to autofluorescence; see Table II), supporting the idea that the T-ag
NLS is more efficient than the bipartite NLS of RB. The
RB-BipMut-
-Gal protein was completely excluded from the nucleus both
in vivo and in vitro
(Fn/cmax of about 0.5; Figs. 1 and 2B,
left panel; Table II), even up to 12 h in vivo
(not shown), where fusion protein localization was scored using a
histochemical stain for
-galactosidase activity in situ (32). This confirmed the bipartite nature of the RB NLS.
|
To compare the NLSs of RB and
T-ag in more detail, we examined the dependence of nuclear transport on
cellular factors. NLS-dependent nuclear protein import
in vitro is known to be dependent on energy (12) and on the
addition of exogenous cytosol (33-35). The latter supplies the
NLS-binding/NPC-docking dimer of importin 58/97 (11) as well as the
monomeric GTP-binding protein/GTPase Ran/TC4 (34) and interacting
proteins (see Ref. 36), all of which are essential for nuclear
accumulation. In vitro experiments were performed to
investigate the dependence of RB-Bip--Gal nuclear import on ATP and
cytosolic factors (Fig. 2; Table II). In identical fashion to that of
T-ag-CcN-
-Gal, nuclear accumulation of RB-Bip-
-Gal was found to
be dependent on both ATP and exogenous cytosol. In the absence of
either, the Fn/cmax was about 1, indicating no accumulation in the nucleus (Fig. 2; Table II). Similarly, the nonhydrolyzable GTP analog GTP
S inhibited transport (Table II). The
results thus indicated that, like T-ag-CcN-
-Gal, RB-Bip-
-Gal localizes in the nucleus through an energy and cytosolic
factor-dependent pathway inhibitable by GTP analogs,
accumulation thus being an NLS-dependent, active process,
most likely dependent on the importin subunits and the GTPase activity
of Ran/TC4.
The nuclear envelope can be perforated by detergents like CHAPS, under
which conditions molecules can diffuse freely between cytoplasm and
nucleoplasm and accumulation in the nucleus is exclusively due to
binding to nuclear components. No accumulation of either RB-Bip--Gal
or T-ag-CcN-
-Gal was observed in the presence of CHAPS (Table II and
data not shown), indicating that binding in the nucleus is unlikely to
be part of the mechanism of nuclear accumulation conferred by either
NLS (see also Refs. 5-7), although it should be remembered that some
soluble nuclear components may be lost through the CHAPS
permeabilization. That the RB bipartite NLS itself is unlikely to have
a role in binding to nuclear components is supported by other studies
(Ref. 23; see also "Discussion").
The reduced nuclear import of RB-Bip--Gal relative to
the wild type T-ag-CcN-
-Gal fusion protein both in vivo
and in vitro (Figs. 1 and 2) could result from reduced
efficiency of recognition of the NLS by the importin 58/97 complex. To
test this hypothesis, an ELISA-based binding assay (Ref. 30; see
"Materials and Methods") was employed. RB-Bip-
-Gal,
RB-BipMut-
-Gal, and T-ag-CcN-
-Gal fusion proteins were coated
onto microtiter plates and incubated with increasing amounts of
importin 58-GST or importin 58/97-GST complex. Binding was then
quantitated using antibodies specific to GST and an
alkaline-phosphatase-labeled secondary antibody.
For importin 58 alone the apparent dissociation constant
(KD) for both T-ag-CcN--Gal and RB-Bip-
-Gal
was about four times higher than that for the importin 58/97 complex,
confirming that the latter is the high affinity NLS binding component
for both types of NLS (16, 30, 37). The KD of
T-ag-CcN-
-Gal for importin 58/97 (10.3 ± 2.1 nM,
mean ± S.D. for two separate experiments) was about 6.5 times
lower than that for RB-Bip-
-Gal (67.5 ± 16.9 nM)
(Fig. 3). The KD for
importin 58/97 for RB-BipMut-
-Gal could not be determined due to the
lack of binding (<1% the maximal amount of importin 58/97 bound by
RB-Bip-
-Gal), confirming the specificity of the NLS binding assay
(see also Ref. 30). Although the basis was not entirely clear, the
maximal amount of importin bound (the Bmax) also
varied between T-ag-CcN-
-Gal and RB-Bip-
-Gal, whereby the T-ag
NLS maximally bound about three times more importin 58/97 complex than
the bipartite NLS of RB. These results were not attributable to partial
denaturation of the RB-Bip-
-Gal preparation as indicated by the
results for the
-galactosidase ELISA (see also "Materials and
Methods"), which is routinely used to standardize results for
equivalent amounts of native protein (30). We have obtained very
similar results for an IL-5 bipartite NLS containing
-galactosidase
fusion protein (27), so that it seems reasonable to conclude that the
results for RB-Bip-
-Gal do not represent an artifact of either the
RB-Bip-
-Gal protein preparation or the ELISA-based binding assay.
NLS masking within the
-galactosidase tetramer can also not be the
basis of the differences in Bmax, since the four
subunits of
-galactosidase are structurally identical (see Ref. 38);
that is, a reduced Bmax is not interpretable in
terms of differences in the number of importin 58/97 binding sites
(i.e. number of NLSs) (30). The basis of the differences in
Bmax may simply be that
-galactosidase itself
binds significant amounts of importin 58/97 at concentrations of the
latter above 0.1 µM (see Ref. 30 and data not shown), and
since the values for
-galactosidase are always subtracted from those
for the respective fusion proteins, lower affinity NLSs such as that of
RB thereby exhibit ostensibly lower Bmax values
than the high affinity T-ag NLS as a result.
Based on the results for importin 58/97 binding, we conclude that the
relatively low efficiency of the RB bipartite NLS compared with the
T-ag NLS in terms of nuclear import (Figs. 1 and 2) is likely to be the
direct result of its lower affinity for and binding to the importin
58/97 complex. This is consistent with our recent study demonstrating
that the rate of nuclear import in vivo correlates directly
with the respective binding affinity for importin 58/97 of
T-ag-CcN--Gal variants (30).
This study presents the first kinetic analysis of nuclear import
mediated by the RB bipartite NLS. We demonstrate that amino acids
860-877 of human RB (p110Rb) are capable of targeting a
large heterologous protein into the nucleus, both in vivo
and in vitro, and therefore constitute a functional NLS.
Mutation of KR861 to TT abolishes the bipartite character
of the RB NLS, abrogating its nuclear targeting ability both in
vivo and in vitro in analogous fashion to the effect of
similar mutations on other bipartite NLSs (24-27). Nuclear import of
RB-Bip--Gal in vitro requires ATP and the addition of
exogenous cytosol and is inhibited by the nonhydrolyzable GTP analog
GTP
S (Fig. 2B, Table II). RB-Bip-
-Gal thus appears to
be accumulated in the nucleus by a classical NLS-dependent, active pathway, dependent on importin, ATP, and the monomeric GTP-binding protein/GTPase Ran/TC4 (14, 15, 17, 39; see also Ref. 40).
In vitro measurements (Table II) in the absence of an intact nuclear envelope due to the action of the nuclear membrane
permeabilizing detergent CHAPS, demonstrate that nuclear accumulation
is unlikely to be due to binding of the RB fusion protein to nuclear
components (Ref. 23; see also below).
This study embodies the first definitive kinetic comparison of two basic types of NLS, namely the T-ag NLS and the bipartite NLS of RB, both of which, as shown in this study, confer ATP-, cytosolic factor- and GTP-binding protein/GTPase-dependent nuclear accumulation, which appears to be independent of binding to nuclear components. The RB NLS confers significantly lower maximal nuclear accumulation both in vivo and in vitro compared with the T-ag NLS, while the initial import rate of the fusion protein carrying the RB NLS is reduced over 2-fold in vivo. The T-ag NLS is clearly a more efficient NLS than the RB bipartite NLS. Since the results for the RB NLS with respect to in vivo nuclear transport and importin binding properties (see also below) are similar to those for the bipartite NLSs of SWI5 (26) and IL-5 (27), the conclusions reported here with respect to the properties of the RB NLS may be valid for bipartite NLSs in general.
An ELISA-based binding assay (30) was employed to examine the
interaction of the NLS-binding importin 58/97 complex with the T-ag NLS
and the RB bipartite NLS. The specificity and sensitivity of the
binding assay allows the measurement of high affinity (nanomolar) interactions between importin subunits and different NLSs. The dimeric
importin 58/97 complex, rather than the NLS-binding importin 58 subunit
alone, was confirmed through direct affinity determinations to be the
high affinity NLS receptor for both types of NLS (see Refs. 16, 30, and
37). Apart from our previous study (30), measurements of NLS-binding
protein/NLS binding affinity have only been performed for the low
affinity NLS receptor (100 nM and higher; see Ref. 41). Our
assay's specificity is demonstrated by the fact that NLS mutant
derivatives are severely impaired in importin binding as shown in this
and our previous study (30). We found that the T-ag -galactosidase
fusion protein has a 6.5 times higher binding affinity for the importin
58/97 complex than the
-galactosidase fusion protein containing the
RB NLS; a similarly reduced binding affinity for importin 58/97 on the
part of the bipartite NLS of IL-5 supports the idea that the results
for the RB NLS are representative of bipartite NLSs in general and that bipartite NLSs may exhibit lower importin binding affinity than those
of the T-ag NLS type. The lower affinity of the RB bipartite NLS for
the importin 58/97 complex presumably constitutes the basis of the
observed reduced maximal nuclear accumulation and import rate in
vivo conferred by the RB NLS relative to the T-ag NLS. This
implies, as shown previously (Ref. 30 and see Refs. 36 and 42), that
the initial events of nuclear transport, and NLS-binding by importin in
particular, are critical in determining overall nuclear protein import
kinetics. That different members of the NLS-binding protein family,
such as the importins, may have distinct NLS binding affinities (41) is
consistent with this idea.
One aspect that arose from this study in terms of the comparison between the T-ag and RB NLSs was the fact that the results for nuclear import kinetics in vitro did not concur completely with those obtained in vivo. The wild type RB fusion protein exhibited over 50% reduced maximal nuclear accumulation and initial import rates compared with the T-ag fusion protein in vivo, but the differences between the two proteins were not as great in vitro and not significant in the case of the import rate. Clearly, the in vivo and in vitro nuclear transport assays are complementary rather than being identical, the most important differences being the source of the exogenously added cytosol (rabbit reticulocyte lysate), the ATP-regenerating system, and the temperature of incubation (37 °C in vivo as opposed to room temperature in vitro). Theoretically, at least, no transport component/ATP etc. is limiting in vitro, which may explain why the affinity of NLS recognition does not appear to be a significant factor in determining the rate of accumulation in vitro, i.e. why there is no significant difference in the initial rates of fusion proteins containing low affinity (RB) or high affinity (T-ag) NLSs, as opposed to in vivo. The in vivo system is clearly more physiologically relevant with respect to transport kinetics in living cells, while the in vitro system enables the dependence on particular transport components, ATP, etc. to be tested. The level and type of the NLS-binding proteins expressed has recently been shown to vary between cell types (41), so that it seems reasonable to surmise that differences in the nature and amount of NLS-binding proteins and other cytosolic factors compared with in vivo are the basis of the fact that the import rate of RB and T-ag fusion proteins is not significantly different in vitro (see also Ref. 27).
While the conclusions of this study are quite clear with respect to the functional activity of the RB NLS, it should be mentioned that additional sequences may be involved in the nuclear accumulation of the complete RB protein. Cell cycle/phosphorylation-dependent regulation of nuclear association ("tethering") of RB has been reported (43, 44), but our in vitro results here using CHAPS support the idea that the RB bipartite NLS is not likely to be involved in binding to nuclear components. We previously identified a putative CcN motif in RB (36), but the results of Zacksenhaus et al. (23) with respect to mouse p110RB1 imply that this sequence is not essential for RB nuclear localization. The latter study indeed implicates the role of sequences other than the mouse RB bipartite NLS (in exon 25) in nuclear localization, since deletion of exon 22 (within the RB T-ag/E1A binding domain), additional to mutation of the RB bipartite NLS, was found to be necessary to effect an exclusively cytoplasmic location of an NLS mutant. The authors suggest a role for these additional sequences in RB nuclear localization through conferring association with nuclear proteins (23). In the light of our results and those of others (23, 43, 44), it seems reasonable to propose that the bipartite NLS constitutes the primary nuclear entry signal of RB, while other sequences may mediate interactions with nuclear components (23).
In conclusion, this study shows that archetypal representatives of the two basic types of NLS are both accumulated in the nucleus by a classical NLS-dependent, active pathway, but that the RB bipartite NLS is not as efficient in nuclear targeting as the T-ag NLS. The difference is probably due to a lower affinity for the NLS-binding importin 58/97 complex, consistent with our findings for T-ag derivatives (30), indicating a direct correlation between importin binding affinity and the initial nuclear import rate. Reduced affinity for the importin 58/97 complex and reduced in vivo nuclear import conferred by the NLS, compared with the properties conferred by the T-ag NLS, may be a common feature of bipartite NLSs.
We thank Joanna Reid for helpful discussions and Lyndall J. Briggs and Patricia Jans for skilled technical assistance.