| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Department of Biology (J.A.N., K.E.F., T.A.S., B.S.W., L.A.A.)
College of William and Mary Williamsburg, Virginia, 23187
Department of Zoology (C.F.B., J. M., L.A.A.) University
of Canterbury Christchurch, New Zealand 8001
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
(TR) exhibits
a dual role as an activator or repressor of gene transcription in
response to thyroid hormone (T3). Our studies
show that TR, formerly thought to reside solely in the nucleus
tightly bound to DNA, actually shuttles rapidly between the nucleus and
cytoplasm. The finding that TR shuttles reveals an additional
checkpoint in receptor control of gene expression. Using
Xenopus oocyte microinjection assays, we show that there
are two coexisting mechanisms for nuclear entry of TR. First,
nuclear import of TR (molecular mass 46 kDa) was not
sensitive to general inhibitors of signal-mediated transport,
indicating that TR can enter the oocyte nucleus by passive
diffusion. Second, when TR was tagged with
glutathione-S-transferase, import of the fusion protein
(molecular mass 73 kDa) was completely blocked by these inhibitors,
demonstrating that an alternative, signal-mediated import pathway
exists for TR. Nuclear retention of TR in oocytes is enhanced in
the presence of T3, suggesting that more
intranuclear binding sites are available for the ligand-bound receptor.
Using mammalian cells, we show that shuttling of green fluorescent
protein (GFP)-tagged and untagged TR is inhibited in both chilled
and energy-depleted cells, suggesting that there is an
energy-requiring step in the nuclear retention/export process. Nuclear
export of TR is not blocked by leptomycin B, a specific inhibitor of
the export receptor CRM1, indicating that TR does not require the
CRM1 pathway to exit the nucleus. Dominant negative mutants of TR with
defects in DNA binding and transactivation accumulate in the cytoplasm
at steady state, illustrating that even single amino acid changes in
functional domains may alter the subcellular distribution of TR. In
contrast to TR, nuclear export of its oncogenic homolog v-ErbA is
sensitive to leptomycin B, suggesting that the oncoprotein follows a
CRM1-mediated export pathway. Acquisition of altered nuclear export
capabilities may contribute to the oncogenic properties of v-ErbA. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
1, TRß1, and TRß2, exhibit similar ligand-dependent regulation
of gene activity, whereas a fourth isoform, TR2, produced by
differential splicing of the TR primary transcript, lacks the
ligand-binding and transactivation domains (1). In addition,
v-erbA, a mutated viral derivative of the cellular
locus encoding TR1 (c-erbA), has acquired oncogenic
properties. The v-ErbA oncoprotein is unable to bind ligand and acts as
a constitutive dominant repressor of transcription regulated by TR in
mammalian and avian cells (2).
Most members of the nuclear receptor superfamily require ligand binding
for functional activity. For example, the unliganded glucocorticoid
receptor resides in the cytoplasm, while the liganded receptor
undergoes rapid nuclear translocation and promotes gene activation (3),
and, although unliganded receptors for progesterone, estrogen, and
retinoic acid are primarily intranuclear, ligand binding is required
for these receptors to interact with target genes (4, 5). TR is unusual
in that it is bound to target genes both in the presence and absence of
ligand (1). The dual role of TR as a repressor (or activator in some
cases) of specific genes in the absence of ligand and an activator (or
repressor) of these same genes in the presence of ligand implies
constitutive nuclear localization. Most studies, although limited in
their resolution, have provided evidence in favor of this restricted
subcellular distribution for TR (1, 2, 6, 7, 8, 9, 10); however, when
overexpressed, some TR1 (11) and TRß1 (12) may localize to the
cytoplasm.
The nucleus forms a discrete compartment in eukaryotic cells. This allows gene expression to be regulated by altering the nucleocytoplasmic distribution of transcription factors in response to external stimuli. To fully understand the cellular response to T3, investigation of all levels of receptor control is essential, including mechanisms for transport of TR across the nuclear envelope and its subsequent nuclear retention.
In recent years, there has been a tremendous increase in our mechanistic understanding of how macromolecules enter and exit the nucleus. After synthesis in the cytoplasm, nuclear proteins are imported into the nucleus exclusively through large proteinaceous structures of approximately 125 MDa in size called nuclear pore complexes (NPCs) (13, 14). The majority of nuclear proteins contain a short amino acid motif called a nuclear localization signal (NLS), which interacts with soluble receptor and NPC-associated proteins in a multistep, energy-dependent import process. There is a diversity of import receptors that recognize different classes of NLS (13).
In addition to serving as a gated entry point for large proteins, the NPC also provides a passive diffusion channel for ions and metabolites and, in principle, for proteins smaller than 60 kDa (13). However, in most cases, small proteins, like larger proteins, enter the nucleus by an energy-dependent and receptor-mediated process (13, 15). There are only a few known exceptions to this general rule (16, 17, 18, 19, 20, 21). It was originally thought that the process of translocation through the central channel of the NPC was the energy-requiring step in nuclear import. Recent data suggest, however, that translocation occurs in the absence of energy and that cytoplasmic hydrolysis of GTP by Ran during receptor recycling may be the only energy-consuming event in nuclear transport (13).
A number of proteins, including steroid hormone receptors, shuttle between the cytoplasm and nucleus (3, 4, 5, 22). Nuclear export, like nuclear import, occurs by multiple pathways (13). The balance between nuclear import and export of transcription factors provides an additional level of control in the regulation of gene expression (22). For example, an NLS that is necessary for nuclear import may not be sufficient for nuclear retention of a regulatory protein, unless activation of import signals is coupled to the suppression of export signals (22, 23). Nuclear retention of shuttling proteins, and small proteins that enter and exit the nucleus by passive diffusion, may also depend upon the availability of intranuclear binding sites (3, 18, 21, 24, 25, 26).
In the present study, we explored the molecular mechanisms regulating
nuclear localization and retention of TR1 (hereafter referred to as
TR for simplicity). To further investigate the complexity of the
cellular response to T3, we used a complementary
approach of Xenopus oocyte microinjection and transient
transfection in conjunction with interspecies heterokaryons.
Unexpectedly, we found that nuclear import of TR in
Xenopus oocytes can proceed by two different coexisting
mechanisms, either through a passive diffusion pathway, or a
signal-mediated process. Importantly, although TR accumulates in the
nucleus at steady state, the receptor shuttles rapidly between the
cytoplasm and nucleus in both Xenopus oocytes and mammalian
cells. We show that two dominant negative mutants of TR, with defects
in DNA binding and transactivation, accumulate in the cytoplasm at
steady state, illustrating that even single amino acid changes in
functional domains may lead to a dramatic shift in the subcellular
distribution of TR. The oncoprotein v-ErbA, which is found in both the
cytoplasm and nucleus at steady state, is also a shuttling protein, but
in contrast to TR and other nuclear receptors, v-ErbA follows a
CRM1-mediated export pathway.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
in
Xenopus Oocytes, we began by using
Xenopus oocyte microinjection assays. Xenopus
oocytes are well suited to addressing these questions for a number of
reasons. First, oocyte microinjection assays have contributed
significantly to advances in understanding of transcriptional
regulation by TR (27, 28), Second, since it is possible to inject
labeled proteins into the cytoplasm or nucleus of Xenopus
oocytes and manually dissect both subcellular compartments, this assay
offers the possibility of studying both nuclear import and export
processes in quantitative terms. Finally, the low levels of endogenous
TR present in oocytes are unlikely to compete with microinjected TR. TR
mRNA and TR protein are present at all stages of oogenesis, but the
majority of maternal transcripts are not translated and are stored for
use later in embryogenesis (27, 29).
It was anticipated, based on the literature regarding TR localization
and function in avian and mammalian cells (1, 2, 6, 7, 8, 9, 10), that TR
would be entirely localized to the nucleus of Xenopus
oocytes; however, we found that microinjected TR localized in both
the cytoplasm and nucleus. After microinjection into the oocyte
cytoplasm, only approximately 40% of in vitro-translated
35S-labeled rat TR was localized to the oocyte
nucleus (Fig. 1
, lane 2). A successful
cytoplasmic injection was scored by the absence of a red nucleus
(colored by hemoglobin present in the reticulocyte lysate) (see
Materials and Methods). To ensure that nonsaturating
concentrations of 35S-TR were injected, the
nucleocytoplasmic distribution of 50, 100, and 200 pg of
35S-TR per oocyte was analyzed. No difference
was observed in the nucleocytoplasmic distribution patterns within this
range of concentrations (data not shown); thus 100 pg per oocyte were
used in subsequent experiments. To determine the optimal time at which
to perform nucleocytoplasmic transport assays, the kinetics of nuclear
import for TR were analyzed at intervals between 2 h and
12 h (data not shown). Steady state was reached at approximately
6 h post injection.
|
was not entirely
localized to the nucleus is that fewer intranuclear binding sites are
available for rat TR in Xenopus oocytes. To rule out the
possibility that our use of a heterologous assay was skewing the
localization pattern, we also analyzed the nucleocytoplasmic
distribution of cytoplasmically injected
35S-labeled Xenopus TR (Fig. 1
, lanes 3 and 4). The distribution was comparable for the two receptors,
indicating that our results cannot simply be explained as a
species-specific difference.
Since TR is not entirely localized to the nucleus in Xenopus
oocytes, these large cells provide an excellent system for
reconstituting regulatory networks affecting nuclear retention of
TR. We thus sought to determine whether the nucleocytoplasmic
distribution pattern differs between liganded and unliganded receptor.
No detectable levels of T3 are present in
Xenopus oocytes (27). In the absence of
T3, on average 39% of cytoplasmically injected
35S-TR was localized to the oocyte nucleus
(Table 1 and Fig. 2A). In the presence of
T3, the amount of 35S-TR
localized to the nucleus after cytoplasmic injection increased
significantly to 51% (P < 0.001, Table 1; Fig. 2A, compare lanes 1–2 with 3–4). Ligand-enhanced nuclear retention was
also observed for nuclear-injected 35S-TR; in
the absence of T3, only 62% of TR was
retained in the nucleus, compared with 85% retention in the presence
of T3 (P < 0.001, Table 1; Fig. 2A, compare lanes 5–6 with 7–8). A successful nuclear injection was
scored by the presence of a red nucleus (colored by hemoglobin present
in the reticulocyte lysate) (see Materials and Methods).
This significant increase in the amount of
35S-TR localized to the nucleus in the
presence of ligand suggested that a greater number of intranuclear
binding sites is available for interaction with
T3-bound receptor.
|
|
distribution was attributable to
hormone binding, rather than a nonspecific effect, oocytes were
cytoplasmically injected with 35S-labeled
ribosomal protein L5, and nuclear localization was assayed in the
presence or absence of T3. As expected,
incubation in T3 did not result in an increase in
the amount of 35S-labeled ribosomal protein L5
localized to the oocyte nucleus (Fig. 2B, compare lanes 1–2 with
3–4); in fact, a slight reduction in nuclear ribosomal protein L5 was
observed (Table 2). As ribosomal protein
L5 does not bind T3, this confirms that enhanced
nuclear retention of TR is a specific effect, attributable to ligand
binding.
|
Can Occur by Passive Diffusion in
Xenopus Oocytes occurs,
several criteria for distinguishing passive diffusion from
signal-mediated import were assessed (13): 1) Is TR import energy-
and temperature-dependent or independent? 2) Is TR import inhibited
by wheat germ agglutinin (WGA), a lectin that binds to NPC proteins,
thereby inhibiting signal-mediated nuclear import without affecting
passive diffusion? 3) Is TR import competed by excess NLS-bearing
substrates? Competition would indicate that nuclear entry shares
saturable components with a certain NLS receptor-mediated pathway.
First, we tested whether nuclear import of TR in Xenopus
oocytes is temperature-dependent. Surprisingly, import of in
vitro-generated 35S-TR was not inhibited
in chilled oocytes (P > 0.1, Table 1; Fig. 3A, compare lanes 1–2 with 3–4),
suggesting that TR can enter the oocyte nucleus by passive diffusion
through the NPCs. A possible alternative explanation for these findings
is that partitioning of TR to the nuclear fraction represents TR bound
to the cytoplasmic face of the NPC, but unable to carry out the second,
temperature-dependent step of signal-mediated translocation through the
NPC into the interior of the nucleus (30). To ensure that
35S-TR had, in fact, accumulated within the
interior of the nucleus in chilled oocytes, isolated nuclei were washed
extensively with 1% Triton X-100 to release any proteins associated
with the outer nuclear membrane. Amounts of
35S-TR extracted from nuclei of either chilled
or warm oocytes dissected in the presence of detergent were comparable
to those from nuclei dissected in the absence of detergent (data not
shown). These observations suggest that, in both chilled and warm
oocytes, the majority of 35S-TR is located in
the interior of the nucleus and that 35S-TR
does not accumulate at the cytoplasmic face of the NPCs.
|
; Fig. 3, compare lanes 8 and 10).
Competence of L5 import was restored when oocytes were further
incubated for 6 h at 18–20 C (compare lanes 8 and 12).
A passive diffusion mechanism of TR import is further supported by
our next finding that import is energy-independent. In oocytes treated
with apyrase, which depletes cellular ATP (and GTP), import of TR
was not significantly inhibited (P > 0.1, Table 1;
Fig. 3B, compare lanes 1–2 with 3–4). In contrast, import of the
control protein, ribosomal protein L5, was significantly inhibited
(P < 0.001, Table 2; Fig. 3B, compare lanes 6 and
8).
Treatment with apyrase does not distinguish between a requirement for
ATP vs. GTP hydrolysis during nuclear import. In most cases
studied so far, signal-mediated protein import is functionally linked
to the Ran-GTPase cycle (13); however, there are exceptions. For
example, nuclear import of the U1A and U2B'' spliceosome proteins
requires ATP hydrolysis, but is Ran-independent (33). Similarly,
nuclear import of I
B (inhibitor of B) occurs by an
ATP-dependent and Ran-independent pathway (34). Typically, to
distinguish between ATP-dependent and GTP-dependent import pathways,
nonhydrolyzable NTP analogs are tested for their ability to support
import. However, these analogs are highly toxic to oocytes, resulting
in irreversible changes in cell morphology and rapid mortality (35).
Thus, to indirectly test whether TR nuclear import is functionally
linked to GTP hydrolysis, we assessed the effect of perturbation
of the Ran-GTP-ase system. For this purpose we used both wild-type
Ran and a mutated derivative in which the glutamine at position 69 has
been substituted for leucine (RanQ69L) (36). Because this mutant is
deficient for GTP hydrolysis activity, import processes that are either
linked to, or dependent on, Ran-mediated GTP hydrolysis should be
inhibited by the addition of RanQ69L. Both wild-type Ran and mutant Ran
can be microinjected into Xenopus oocytes without subsequent
loss of viability (37).
The import potential of TR was determined after cytoplasmic
injection of 35S-TR together with wild-type
Ran or RanQ69L. Neither excess wild-type Ran nor RanQ69L had a
significant effect on 35S-TR import
(P > 0.1, Table 1; Fig. 3C, compare lanes 1–2 and
3–4), suggesting that TR import does not require Ran-mediated GTP
hydrolysis. In contrast, nuclear import of
35S-labeled ribosomal protein L5 was inhibited to
a significant degree in the presence of RanQ69L (P <
0.01, Table 2; Fig. 3C, compare lanes 6 and 8). The findings for L5 are
consistent with the observation that complex formation of this small
ribosomal protein with import receptors is strongly inhibited in the
presence of RanQ69L (32).
Next we assessed whether specific interaction with the NPC is required
for nuclear import of TR, by testing for sensitivity to WGA. WGA
binds to N-acetylglucosamine-containing proteins present in
the NPCs and inhibits signal-mediated import. Passive diffusion is not
inhibited, since WGA does not physically occlude the nuclear pores
(31). TR import was not significantly inhibited in oocytes
preinjected with WGA (P > 0.1, Table 1; Fig. 3D, compare lanes 2 and 4). As expected for a protein imported via a
NPC-mediated pathway, ribosomal protein L5 import was significantly
inhibited in oocytes preinjected with WGA (P < 0.001,
Table 2; Fig. 3D, compare lanes 6 and 8). WGA inhibition reduced the
percentage of nuclear L5 from 33% to 5%; a similar reduction to 4%
was observed for energy depletion assays.
Signal-mediated transport pathways are saturable; thus, as an
additional criterion for characterizing the import pathway of TR,
competition assays with histone H1 and ribosomal protein L5 were
performed. The rationale for these competition experiments is that
import kinetics of TR would be at a reduced level if histone H1 (15)
or L5 (31, 32), acting as direct competitors of signal-mediated
processes, were imported by the same mechanisms as TR. Such a
reduction would be indicative of competition for shared transport
factors. Members of the importin ß superfamily of transport receptors
account for all interactions with the NPC required for passage into the
nucleus. In some cases, importin ß-type receptors need an adapter
molecule to interact with NLS-bearing cargo (13). For example, import
of proteins that carry a classical NLS is mediated by an importin
/importin ß heterodimer, whereas histone H1 is imported by an
importin 7/importin ß heterodimer (13, 15). Thus, even if a different
adapter is used, competitor cargo may still compete for use of importin
ß and docking sites at the NPC.
To ascertain histone H1 concentrations at which inhibition of
signal-mediated transport occurred, 35S-labeled
ribosomal protein L5 was competed against excess histone H1 at two
intracellular concentrations. Both low and high concentrations of
histone H1 were shown to inhibit the nuclear import of
35S-L5 (P < 0.001, Table 2; Fig. 3E, compare lanes 14, 16, and 18). In contrast,
35S-TR import was not inhibited by coinjection
with histone H1, at either low or high intracellular concentrations
(P > 0.1, Table 1; Fig. 3E, compare lanes 2, 4, and
6).
Nuclear import of 35S-labeled ribosomal
protein L5 was inhibited by equimolar amounts of unlabeled L5
(P < 0.001, Table 2; Fig. 3E, compare lanes 20 and
22). In contrast, coinjection with 35S-labeled
ribosomal protein L5 did not lead to an alteration in the
nucleocytoplasmic distribution of 35S-TR
(P > 0.1, Table 1; Fig. 3E, compare lanes 8 and 10).
The lack of competition observed with TR by ribosomal protein L5
suggests that independent mechanisms of transport are in operation for
these two proteins. This is of particular interest since the NLS of
ribosomal protein L5 is able to interact directly with multiple import
receptors, including importin , importin ß, transportin, importin
5 (RanBP5), and importin 7 (RanBP7) (32, 38).
In summary, we have shown that 35S-TR import
in Xenopus oocytes is temperature- and energy-independent
and is not blocked by the dominant negative RanQ69L protein, WGA, or
the presence of excess NLS-bearing substrates. This lack of sensitivity
to general inhibitors of signal-mediated import is consistent with a
passive diffusion mechanism for nuclear entry of TR.
Nuclear Retention of TR Is Temperature Dependent in
Xenopus Oocytes
By definition, the steady state levels attained by passive
diffusion would be expected to be equal in both directions through the
NPC; however, we observed differences in the percentage of TR in the
nucleus after cytoplasmic vs. nuclear injection (Table 1 and
Fig. 2A), suggesting a more complex scenario. Thus, having shown that
nuclear import of TR can occur by simple diffusion in
Xenopus oocytes, we next sought to ascertain whether nuclear
export of TR also can follow a passive diffusion pathway. To this
end, export competence of nuclear-injected
35S-TR was assessed in chilled oocytes.
Nuclear retention of TR was significantly increased by chilling:
86% of TR was retained in the nucleus of chilled oocytes, compared
with 62% in oocytes at physiological temperature (P <
0.001, Table 1; Fig. 4, compare lanes
3–4 with 5–6). Lastly, nuclear retention of the control protein,
ribosomal protein L5, was also increased by low-temperature incubations
(P < 0.001, Table 2; Fig. 4, compare lanes 9–10 with
11–12). In summary, these data suggest that there is a
temperature-dependent step in the nuclear export/retention pathway of
TR.
|
in
Xenopus Oocytes translocation through the
NPC in Xenopus oocytes. Thus, we next asked the question
whether alternative pathways for nuclear import of TR exist. We tested
whether import of TR tagged with
glutathione-S-transferase (GST) is sensitive to transport
inhibitors. GST-TR interacts specifically with target DNA sequences
in an electrophoretic gel mobility shift assay (data not shown),
suggesting that the tag does not interfere with DNA binding. The
GST-TR fusion protein was used because this molecule should be too
large (molecular mass 73 kDa) to enter the nucleus by passive
diffusion. When a similar experiment was performed with GST-tagged
catalytic subunit of cAMP-dependent protein kinase, the fusion protein
was restricted to the cytoplasm while the wild-type subunit freely
diffused through the NPC (16).
First, we showed by Western blot analysis that GST alone remains
localized exclusively in the oocyte cytoplasm after cytoplasmic
injection (Fig. 5A). When fused to TR,
however, the full-length fusion protein accumulated in the oocyte
nucleus after cytoplasmic injection (Fig. 5B, lane 5). Four predominant
bands were present in the injectate, including full-length GST-TR
(73 kDa) and three degradation products (Fig. 5B, lane 1).
Interestingly, the largest degradation product present in the sample of
injected GST-TR did not localize to the nucleus (Fig. 5B, compare
lanes 4 and 5). When Western blots were probed with an antibody
specific for the extreme C terminus of TR, this band was not
recognized by the antibody, suggesting that the shorter fusion protein
lacks the C-terminal sequence of TR (data not shown). This
degradation product served as a convenient internal control for the
site of injection and for fractionation.
|
was inhibited in chilled
oocytes (
75% inhibition, n = 4), with transport competence
restored by subsequent incubation at 20 C (Fig. 5C, compare lanes 2, 4,
and 6). Reversible inhibition suggests that chilling blocked GST-TR
import by inhibiting the activity of specific components of the
transport machinery, rather than by preventing import by way of
nonspecific cellular damage. Import was completely inhibited by
treatment with apyrase to deplete cellular ATP (n = 3; Fig. 5D, compare lanes 4 and 6) and was inhibited to varying degrees by WGA
(65% inhibition, n = 4; Fig. 5E, compare lanes 2, 4, and 6)
and competitor histone H1 (35% inhibition, n = 6; Fig. 5F, compare lanes 2 and 4). Unlike the ligand-enhanced nuclear retention
observed for 35S-TR, nuclear accumulation of
the GST-TR fusion protein was not significantly enhanced in the
presence of T3 (10% increase, n = 3;
Fig. 3G, compare lanes 2 and 4). It is not known, however, whether
GST-TR binds T3 with the same affinity as the
native receptor.
Interestingly, nuclear import of GST-TR was not inhibited in the
presence of RanQ69L (data not shown). It is possible that in these
assays, RanQ69L was not injected at high enough concentrations to
out-compete endogenous Ran. Alternatively, GST-TR may follow an
ATP-dependent, Ran-independent nuclear import pathway; or, since GTP
hydrolysis by Ran is not required for passage of cargo into the nucleus
(39), GST-TR may have undergone one round of import and then been
trapped in the nucleus, yielding a distribution similar to the pattern
achieved by nucleocytoplasmic shuttling of TR. Further experimentation
under defined conditions (i.e. in vitro nuclear
import assays using permeabilized mammalian cells) will be required to
distinguish between these possibilities.
In summary, these findings show that TR has the necessary sequence
information for nuclear targeting of a fusion protein that is too large
to enter the nucleus by passive diffusion. Import of the GST-TR
fusion protein was temperature and energy-dependent, and blocked by WGA
and competitor NLS-bearing substrate. This sensitivity to general
inhibitors of signal-mediated import is consistent with a
signal-mediated mechanism for nuclear entry of GST-TR. Taken
together with the results described in the preceding sections, these
data suggest that a signal-mediated import pathway, although not
absolutely necessary, contributes to nuclear entry of normal TR in
Xenopus oocytes.
Nuclear Localization of Green Fluorescent Protein (GFP)-Tagged
TR in Mammalian Cells
Our finding that TR can undergo nucleocytoplasmic shuttling in
Xenopus oocytes was intriguing but could be interpreted as
simply an unusual species- or cell-specific difference. To follow up on
these observations, we investigated the nucleocytoplasmic distribution
of GFP-tagged TR in transfected mammalian
cells.
GFP has been effectively used as a tag to investigate transport and
localization of a variety of proteins, including the nuclear receptors
for glucocorticoids (40), estrogen (41), androgen (42), and
mineralocorticoids (43), and, while this study was underway, one of the
other major isoforms of the thyroid hormone receptor, TRß1
(12). To ensure that the GFP tag did not significantly alter the
functional characteristics of TR, we evaluated the DNA binding,
transactivation, and nucleocytoplasmic distribution characteristics of
GFP-TR.
First, we showed by electrophoretic gel mobility shift assay that
GFP-TR interacts specifically with target DNA sequences,
demonstrating that the tags do not interfere with DNA binding (data not
shown). Second, we examined the transcriptional activity of GFP-TR
in transfected COS-1 cells, using a CAT reporter system under control
of a TRE (Fig. 6). The levels of CAT
protein produced were lower for ligand-dependent reporter activation by
GFP-TR compared with untagged TR; however, it is not possible to
make a direct comparison between the two assays. Untagged TR and
GFP-TR constructs are under control of different promoters (Rous
sarcoma virus and cytomegalovirus, respectively); thus a difference in
the levels of CAT protein may simply reflect lower levels of expression
of GFP-TR compared with untagged TR. Importantly, what we did
demonstrate is that GFP-TR, like native (untagged) TR, stimulates
transcription from a TRE-CAT reporter construct in the presence of
T3 (Fig. 6).
|
in
both living cells and fixed specimens. In contrast to the distribution
of TR in Xenopus oocytes, GFP-TR appears to localize
entirely to the nucleus in mouse cells (Fig. 7, A and B). GFP-TR is
distributed diffusely throughout the nucleus in both live (Fig. 7A) and
fixed (Fig. 7B) preparations. No alteration in the nucleocytoplasmic
distribution of GFP-TR in living cells was observed in response to
treatment with ligand; complete localization to the nucleus was
maintained in the presence or absence of T3
concentrations ranging from 10-9
M to 10-3
M (Fig. 7A and Table 3). As expected, when not fused to TR,
GFP was distributed throughout the cell (Fig. 7C). The nuclear
localization characteristics of GFP-TR are thus comparable to those
described in the literature for native TR (6, 7, 8, 9, 10). Taken together
with our finding that GFP-TR is transcriptionally competent and
hormone responsive, we can therefore conclude that GFP-TR is a valid
probe to study the subcellular trafficking of
TR.
|
|
appears to be
exclusively nuclear in NIH/3T3 cells, it is possible to examine nuclear
export characteristics, but not nuclear import by this type of assay.
To determine whether nuclear retention of GFP-TR was temperature-
and energy-dependent, cells were incubated under chilled and
energy-depleted conditions for 3–4 h, in the presence of cycloheximide
to prevent de novo protein synthesis. Under these
conditions, in the majority of cells GFP-TR remained entirely
localized to the nucleus (Fig. 7, A and B, and Table 3). These data are
consistent with our findings in Xenopus oocytes and indicate
that some aspect of nuclear retention occurs by an active process.
GFP-Tagged TR and Native TR Undergo Nucleocytoplasmic
Shuttling in Mammalian Cells
Although it is well established that nuclear receptors for steroid
hormones undergo nucleocytoplasmic shuttling (3, 4, 5, 40, 42), it is not
known whether all nuclear receptors for nonsteroid hormones share this
property. The dual role of TR as a repressor of specific genes in the
absence of ligand and an activator of these same genes in the presence
of ligand had led to the view that TR is exclusively in the nucleus and
is a stable constituent of chromatin (2). Having validated the use of
GFP-tagged TR as a probe and having completed analysis of receptor
localization in single cells, we returned to our primary question: can
TR shuttle in mammalian cells? To this end, we used a very powerful
and elegant method—a transient transfection interspecies heterokaryon
assay (44). The strength of this method lies in its sensitivity; it is
possible to detect rapid nucleocytoplasmic shuttling of proteins that
by other methods would appear to be retained in the nucleus at all
times (Fig. 8A).
|
(Fig. 8B, lower panel) and native TR (Fig. 8B, upper
panel) both rapidly accumulated in the human nuclei in
heterokaryons, demonstrating export of TR from the mouse nucleus and
reimport into the human nucleus. In mouse and human nuclei, both
GFP-TR and untagged TR appeared to distribute diffusely
throughout the nucleus. In nuclei with less intense fluorescence, the
receptor appeared to be excluded from nucleoli. Only exogenous TR
expressed in successfully transfected mouse cells was stained with the
anti-TR antibody; neither cell line contains detectable levels of
endogenous TR (Fig. 8B, middle panel).
GFP-TR (molecular mass 76 kDa) is above the size limits for passive
diffusion. Thus, the possibility exists that, in the single-cell
analysis described in the previous section, GFP-TR was retained in
the nucleus, while the smaller untagged receptor would have been
capable of nuclear export by simple diffusion. To investigate the
overall mechanism for nucleocytoplasmic shuttling, we performed
heterokaryon assays under conditions that would permit passive
diffusion but inhibit active transport processes. In both chilled and
energy-depleted cells, neither GFP-TR nor untagged TR accumulated
in the human nuclei in heterokaryons (Fig. 9). Chilled and energy-depleted
heterokaryons appeared smaller, slightly rounded up, and the actin
filaments were less distinct (Fig. 9). Since an extended period of
chilling or energy depletion (>90 min) had detrimental effects on
heterokaryon viability (data not shown), we limited our analyses to
periods of 60–90 min. Conceivably, this time frame may be insufficient
to reveal relatively slow nuclear export. However, we feel this is
unlikely since shuttling occurs rapidly under physiological conditions
(within 1 h). Furthermore, in single-cell analyses in which
prolonged treatment was possible, after 4 h of energy depletion or
chilling no cytoplasmic accumulation of TR was observed (Fig. 7).
Inhibition of TR shuttling suggests that release from intranuclear
binding sites is temperature- and energy-dependent and/or export occurs
by a signal-mediated process. These findings are consistent with our
results in Xenopus oocytes, where nuclear retention of
microinjected TR also was shown to be temperature-dependent.
|
in Xenopus oocytes as shown in the
present study, and in CV1 cells overexpressing GFP-TRß1 (12), it was
of interest to determine whether the presence or absence of ligand
would influence the direction and extent of nucleocytoplasmic shuttling
of TR in heterokaryons. We showed, however, that there was no
apparent difference in the rapidity or extent of nuclear export from
mouse nuclei and import into human nuclei in heterokaryons: shuttling
of GFP-TR and untagged TR occurred in the presence and absence of
T3 (Fig. 9).
In summary, our data from mammalian cell studies show that TR
accumulates in the nucleus at steady state but, surprisingly, the
receptor shuttles rapidly between the nucleus and cytoplasm. Shuttling
of TR is energy- and temperature-dependent, but apparently not
ligand-dependent. These findings reveal an additional checkpoint in
control of gene expression by TR and pave the way for future studies
on the role of nucleocytoplasmic shuttling in gene regulation.
Nuclear Export of TR Is Not Mediated by the Export Receptor
CRM1
To begin to elucidate the mechanism for TR shuttling, we sought
to ascertain whether TR nuclear export is facilitated by the export
receptor CRM1 (13). For this purpose, we prepared heterokaryons by
fusing mouse cells transfected with GFP-TR with untransfected human
cells. Subsequently, heterokaryons were treated with the CRM1-specific
inhibitor, leptomycin B (3, 45, 46). In both the presence and absence
of leptomycin B, GFP-TR rapidly accumulated in the human nuclei in
heterokaryons (Fig. 10, upper
left panels), suggesting that TR does not require the CRM1
pathway to exit the nucleus.
|
, upper right panels).
Dominant Negative Mutants of TR Have Altered Subcellular
Distribution Patterns
To further investigate the shuttling and nuclear retention
properties of TR, we assessed the relative importance of DNA-binding
and transactivation domains on the nucleocytoplasmic distribution
pattern of TR. For this purpose, we employed three dominant negative
mutants of TR: C122A, G121A, and v-ErbA. C122A is a synthetic
DNA-binding deficient mutant of TRß. In this mutant, alanine was
substituted for the coordinating cysteine at residue 122 in the
recognition -helix of the first zinc finger in the DNA-binding
domain (27, 48). In G121A, another synthetic mutant derived from TRß,
alanine was substituted for glycine at residue 121 of the first zinc
finger. The G121A mutant is able to bind DNA but is defective in
transcriptional activation activity (48). Due to an N-terminal fusion
with retroviral gag sequences, v-ErbA, the viral oncogenic homolog of
TR, lacks the first 12 amino acids of TR. In addition, the viral
protein lacks nine amino acids close to the C terminus and has 13 amino
acid changes: 2 in the DNA-binding domain and the other 11 distributed
along the molecule in the hinge region (domain D) and ligand-binding
domain (2, 49). The oncoprotein is unable to bind ligand and acts as a
constitutive repressor of transcription regulated by TR.
In transient transfection assays in NIH/3T3 cells, both untagged and
GFP-tagged DNA-binding mutant C122A and transactivation mutant G121A
show either diffuse whole cell staining at steady state, or exclusion
from the nucleus with punctate cytoplasmic foci or discrete aggregates
particularly around the nuclear periphery (Fig. 10, lower left
panels, and data not shown). Interestingly, these distribution
patterns are very similar to the patterns described for a TR mutant
in which the entire D domain was deleted (8). The subcellular
distribution of the virally derived oncogenic homolog of TR was
similar to the distribution pattern of the synthetic mutants tested.
GFP-tagged and untagged v-ErbA localized both in the nucleus and the
cytoplasm, showing either diffuse whole cell staining or a distinct
punctate nucleocytoplasmic distribution at steady state (Fig. 10, lower right panels, and data not shown). These observations
for GFP-v-ErbA are in agreement with previous studies of the
distribution of native v-ErbA in transfected cells, in which the
oncoprotein also was detected in both the nucleus and cytoplasm (9, 50). Similarly, in Xenopus oocytes,
35S-labeled C122A and v-ErbA both showed a
nucleocytoplasmic distribution; approximately 40% of C122A and 20% of
v-ErbA accumulated in the nucleus after cytoplasmic microinjection
(data not shown).
In summary, the data presented here show that even single amino acid changes in functional domains of TR may lead to a dramatic shift in the balance between nuclear retention and cytoplasmic accumulation. These findings highlight the importance of DNA-binding and transcriptional activation activity in nuclear localization of TR.
Nuclear Export of the Oncoprotein v-ErbA is CRM1-Dependent
To follow up on our observation that nuclear export of TR is
leptomycin B-insensitive, we tested the effect of the drug on the
subcellular distribution of the TR mutants described in the preceding
section, using single-cell analysis. There was no change in the
nucleocytoplasmic distribution of TRß mutants G121A and C122A after
treatment with leptomycin B (Fig. 10, lower left panels),
suggesting that the transactivation mutant and the DNA-binding mutant
do not require the CRM1 pathway to exit the nucleus.
In contrast, nuclear export of both untagged and GFP-tagged v-ErbA was
sensitive to leptomycin B (Fig. 10, lower right panels, and
data not shown). In control cells, the oncoprotein v-ErbA is
distributed in both the nuclear and cytoplasmic compartments; however,
in cells treated with leptomycin B there was a dramatic shift in the
distribution pattern. Upon treatment with the CRM1-specific inhibitor,
v-ErbA was entirely localized to the nucleus in 100% of transfected
cells (Fig. 10, lower right panels), suggesting that the
oncoprotein was imported into the nucleus and subsequently trapped in
the nuclear interior by the block to the export pathway. These striking
findings demonstrate that not only is v-ErbA a shuttling protein but,
in contrast to TR and other nuclear receptors, v-ErbA follows a
CRM1-mediated export pathway.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
. Our
data provide new insights into the molecular mechanisms regulating
nuclear localization and retention of TR. We find that nuclear
import of TR proceeds by one of two coexisting mechanisms in
Xenopus oocytes, a passive diffusion pathway or
signal-mediated translocation through the NPC. We have also determined
that, although TR accumulates in the nucleus at steady state, the
receptor can shuttle rapidly between the cytoplasm and nucleus in both
Xenopus oocytes and mammalian cells. Although there is an
energy-requiring step in the nuclear retention/export pathway, nuclear
export of TR is not mediated by the export receptor CRM1. Dominant
negative variants of TR do not accumulate in the nucleus at steady
state, highlighting the importance of intact DNA-binding,
hormone-binding, and transactivation domains for nuclear retention. The
oncoprotein v-ErbA has acquired the ability to follow a CRM1-mediated
pathway, suggesting that acquisition of altered nuclear export
capabilities has contributed to the oncogenic conversion of TR.
Alternative Pathways for Nuclear Import
We showed, by the following three lines of evidence, that nuclear
import of 35S-labeled TR in Xenopus
oocytes can occur by simple passive diffusion: 1) import was energy
and temperature-independent; 2) import was not inhibited by WGA; 3)
import was not competed by excess NLS-bearing substrates, indicating
that TR nuclear entry does not share saturable components with certain
NLS receptor-mediated pathways. Some exceptions to the typical
signal-mediated pathway for nuclear entry have been reported. For
example, calmodulin (CaM) appears to be imported by simple diffusion,
with CaM-binding proteins in the nucleus acting as a sink for nuclear
retention of CaM upon an increase in intracellular free
Ca2+ (21); nuclear accumulation of the U1
snRNP-specific protein C is due to diffusion and retention in the
nucleus upon incorporation of the protein into the U1 snRNP (18); and,
import characteristics of the free catalytic subunit of cAMP-dependent
protein kinase are consistent with simple diffusion (16).
With our subsequent observation that nuclear retention of untagged
TR in both Xenopus oocytes and transfected mammalian
cells was temperature-dependent, it became apparent that simple
diffusion was not sufficient to explain all aspects of TR
translocation through the NPC. This finding suggested that a
signal-mediated pathway is followed for transport out of the nucleus
and/or that release from intranuclear binding sites is an active
process. Different energy requirements for bidirectional transport
across the nuclear envelope have been reported for steroid hormone
receptors. For this case, import is temperature- and energy-dependent,
while export is not inhibited by energy depletion or chilling (46, 51).
Before export, however, the glucocorticoid receptor must be released
from association with the nuclear matrix, and this step in the nuclear
export pathway is energy-dependent (51).
Finally, we showed that nuclear import of a GST-TR fusion protein in
Xenopus oocytes is sensitive to transport inhibitors,
suggesting that an alternative signal-mediated import pathway can be
followed by TR. There is precedence for such alternative import
pathways. A recent report shows that there are two coexisting
mechanisms for nuclear entry of mitogen-activated protein kinase (MAPK)
(45 kDa): one is passive diffusion and the other is active transport,
the latter being dependent upon the formation of a MAPK dimer (20). It
is possible that the signal-mediated import mechanism is required for
rapid nuclear translocation (20). TR is known to carry out its
function primarily as a heterodimer, in association with the retinoid X
receptor (RXR) both in vitro and in vivo
(1, 2). It is possible that, similar to MAPK, RXR/TR heterodimers
or TR/TR homodimers are formed predominantly in the cytoplasm
and, since they are too large for passive diffusion, are imported by a
regulated, signal-mediated pathway. In this model, since the two
pathways coexist, TR monomers could utilize either an active or
passive mode of nuclear entry. The inter-oocyte variability in response
to general inhibitors of signal-mediated transport (see Table 1) may
reflect partial inhibition of TR nuclear accumulation due to the
general inhibitors blocking access to the signal-mediated pathway.
In addition to heterodimerization with RXR, TR also interacts
with histone deacetylases and corepressors such as N-CoR (nuclear
receptor corepressor) and SMRT (silencing mediator for retinoid and
thyroid hormone receptors) in the absence of T3,
and coactivators such as members of the p160 family and p300/CBP in the
presence of T3 (1, 28). It will be of interest to
determine whether binding of TR to transcriptional corepressors and
coactivators, and subsequent interaction of TR with the chromatin
infrastructure, are important determinants for the mode of nuclear
import, and the nuclear retention and shuttling characteristics of
TR.
It remains to be determined whether alternative import pathways are
available in mammalian cells. Given that GFP-TR enters mammalian
cell nuclei, we can conclude that a signal-mediated pathway is
followed, since the fusion protein is above the size limits for
diffusion. However, fusion protein import could occur by facilitated
diffusion, which requires interactions with the NPC, but is
energy-independent (13). Since both untagged and GFP-tagged TR
rapidly accumulate in the nucleus at steady state and export is
temperature- and energy-dependent, there was no detectable cytoplasmic
population of TR before, during, or after chilling or
energy-depletion experiments. Therefore, it was not possible to
determine the general mechanism of nuclear import using single-cell or
heterokaryon transient transfection assays. Studies using alternative
approaches to investigate the nuclear import pathway followed by TR
in mammalian cells are underway. In summary, data pr
