Green fluorescent protein-tagging reduces the nucleocytoplasmic shuttling specifically of...

Green fluorescent protein-tagging reduces thenucleocytoplasmic shuttling specifically ofunphosphorylated STAT1Thomas Meyer1, Andreas Begitt2 and Uwe Vinkemeier2

1 Abteilung Psychosomatische Medizin und Psychotherapie, Philipps-Universitat Marburg, Germany

2 Abteilung Zellulare Signalverarbeitung, Leibniz-Institut fur Molekulare Pharmakologie, and Freie Universitat Berlin, Germany

Cytokines and growth factors modulate the trans-

criptional activity of their target cells. To this effect,

cytokine-specific signal transducers of the signal trans-

ducers and activators of transcription (STAT) family

of transcription factors are activated at membrane-

bound receptors and subsequently translocate into the

nucleus [1,2]. ‘Activation’ is a multistep process leading

to the tyrosine phosphorylation and subsequent dime-

rization of STATs that enables high affinity and

sequence-specific recognition of DNA [3–5]. In the

nucleus, the activated STAT dimers stimulate or

repress gene transcription by directly binding to

Keywords

FRAP; GFP; nuclear export; nuclear import;

STAT1

Correspondence

U. Vinkemeier, Abteilung Zellulare

Signalverarbeitung, Leibniz-Institut fur

Molekulare Pharmakologie, FU Berlin,

Robert-Rossle-Str. 10, 13125 Berlin,

Germany

Fax: +49 30 94793 179

Tel: +49 30 94793 171

E-mail: [email protected]

(Received 21 August 2006, revised

1 December 2006, accepted 6 December

2006)

doi:10.1111/j.1742-4658.2006.05626.x

Fluorescence recovery after photobleaching (FRAP) and related techniques

using green fluorescent protein (GFP)-tagged proteins are widely used to

study the subcellular trafficking of proteins. It was concluded from these

experiments that the cytokine-induced nuclear import of tyrosine-phos-

phorylated (activated) signal transducer and activator of transcription 1

(STAT1) was rapid, while the constitutive shuttling of unphosphorylated

STAT1 was determined to be inefficient. However, unrelated experiments

came to different conclusions concerning the constitutive translocation of

STAT1. Because these discrepancies have not been resolved, it remained

unclear whether or not unphosphorylated STAT1 is a relevant regulator of

cytokine-dependent gene expression. This study was initiated to examine

the influence of GFP-tagging on the nucleocytoplasmic shuttling of phos-

phorylated and unphosphorylated STAT1. In accordance with previous

findings our results confirm the undisturbed rapid nuclear import of GFP-

tagged activated STAT1. However, we reveal an inhibitory influence of

GFP specifically on the constitutive nucleocytoplasmic cycling of the

unphosphorylated protein. The decreased shuttling of unphosphorylated

STAT1-GFP significantly reduced the activation level while nuclear accu-

mulation was prolonged. Importantly, despite unimpaired nuclear import

of activated STAT1 the transcription of a STAT1-dependent reporter gene

was more than halved after GFP-tagging, which could be linked directly to

reduced nucleocytoplasmic shuttling. In conclusion, it is demonstrated that

GFP-based techniques considerably underestimate the actual shuttling rate

of unphosphorylated native STAT1. The results confirm that the activation

of STAT1 and hence its transcriptional activity is proportional to the

nucleocytoplasmic shuttling rate of the unphosphorylated protein. More-

over, our data indicate that GFP-tagging may differently affect the

mechanistically distinct translocation pathways of a shuttling protein.

Abbreviations

FRAP, fluorescence recovery after photobleaching; GFP, green fluorescent protein; IFN, interferon; LMB, leptomycin B; NES, nuclear export

signal; STAT, signal transducer and activator of transcription.

FEBS Journal 274 (2007) 815–826 ª 2007 The Authors Journal compilation ª 2007 FEBS 815

cognate binding sites in the promoters of cytokine-

responsive genes [6]. By light microscopy this sequence

of events is observable and appears as the nuclear

accumulation of STAT molecules within minutes of cy-

tokine stimulation [7]. The accumulation phase can last

for up to a few hours before the prestimulation distri-

bution is gradually restored. Initially, the STATs were

considered cytoplasmic proteins that enter the nuclear

compartment only in response to cytokine stimulation.

This simple concept has been amended recently and a

more variegated model of STAT functioning was pro-

posed [8]. In the revised model the STATs are des-

cribed as nucleocytoplasmic shuttling proteins that

enter the nucleus irrespective of cytokine stimulation

or tyrosine phosphorylation. The constitutive import

activity of STAT1 occurs independent of transport

receptors (also termed karyopherins) via direct con-

tacts with nucleoporins in the nuclear pore. The same

mechanism applies also for nuclear export, although

carrier-dependent transport via CRM1 (also known as

exportin 1) contributes [9]. Activated STAT1, however,

is barred from further karyopherin-independent shut-

tling [9,10]. Yet, dimerization triggers the exposure

of noncanonical import signals, and the concomitant

binding of transport receptors results in nuclear import

[11–13]. Nuclear export of STAT1, on the other hand,

requires its tyrosine dephosphorylation [10,14].

The possibility to use genetically encoded fluorescent

markers such as green fluorescent protein (GFP) in

combination with laser bleaching has allowed to

directly follow the redistribution of molecules in living

cells [15,16]. This approach was used to study the

intracellular trafficking of STAT1, and no significant

differences were reported between tagged and endo-

genous native molecules [17–19]. These experiments

demonstrated that the inducible nuclear translocation

of phosphorylated STAT1-GFP caused the rapid nuc-

lear import and accumulation, whereas the constitutive

nucleocytoplasmic shuttling of unphosphorylated

STAT1-GFP in unstimulated cells occurred at a very

low rate. It was thus concluded that these results were

applicable also to the endogenous native STAT1. Yet,

unrelated experiments and theoretical considerations

indicated that efficient shuttling also of unphosphoryl-

ated STATs is required for efficient cytokine signalling

[10,20,21]. These discrepancies have not been resolved

to date.

Here, we analyze the influence of GFP-tagging on

phosphorylated and unphosphorylated STAT1. In

accordance with previous results we find that the trans-

port of phosphorylated STAT1 was not affected by the

GFP domain. Contrary, the nucleocytoplasmic shut-

tling of unphosphorylated STAT1 was considerably

reduced after GFP-tagging. Accordingly, the phos-

phorylation and DNA binding decayed significantly

faster. Moreover, the reduced translocation of unphos-

phorylated STAT1-GFP diminished the transcription

of a reporter gene by 60%, even though the nuclear

accumulation was prolonged. Thus, these analyses

demonstrate that GFP-based techniques considerably

underestimate the actual shuttling rate of unphosphory-

lated native STAT1. The results confirm the functional

importance of efficient translocation of nonphosphory-

lated STAT1 for the cytokine-dependent transcription.

Results

GFP-tagging prolonged the duration of STAT1

nuclear accumulation, but not its build-up phase

Carboxy-terminal fusion proteins of STAT1 with GFP

are generally regarded to behave indistinguishably

from the endogenous wild-type protein. At first, we

compared the accumulation kinetics of wild-type

STAT1 and GFP-tagged STAT1 transcription factors

upon stimulation of cells with interferon c (IFNc).Interferon induces the transient accumulation of

STAT1 molecules within the nuclear compartment,

which lasts for a few hours before the prestimulation

distribution is restored [1]. As previously reported [17],

we found that the nuclear import rate of STAT1 is

independent of the expression of the GFP domain in

IFNc-stimulated cells, because the build-up of nuclear

accumulation did not differ between the native and the

GFP-tagged variant. Already 5 min after exposure of

the cells to IFNc both native STAT1 in untransfected

cells and STAT1-GFP in cells expressing the GFP

fusion lost their predominantly cytoplasmic resting dis-

tribution (Fig. 1A, panels a,g) and began to accumu-

late in the nucleus (Fig. 1A, panels b,h). Ten minutes

after IFNc addition to the cells, STAT1 is concentra-

ted within the nucleus, irrespective of the expression of

the GFP domain (Fig. 1A, panels c,i). After 60 min

exposure to IFNc the cytoplasm appeared depleted of

either STAT1 immunoreactivity or GFP fluorescence,

indicating that the cytokine-induced nuclear import of

tyrosine-phosphorylated STAT1 is not protracted by

the expression of the GFP fusion (Fig. 1A, panels d,j).

In contrast to the build-up of nuclear accumulation,

the decay of STAT1 accumulation in the nucleus dif-

fered considerably between the tagged and untagged

variants. Four hours after a one-hour stimulation with

IFNc, the accumulation phase of native STAT1 was

nearly over (Fig. 1A, panel e) and an additional two

hours later the distribution of unstimulated cells was

regained (Fig. 1A, panel f). GFP-tagged STAT1

GFP-tagging of STAT1 T. Meyer et al.

816 FEBS Journal 274 (2007) 815–826 ª 2007 The Authors Journal compilation ª 2007 FEBS

protein, however, remained largely nuclear also four

hours after the initial IFNc stimulation (Fig. 1A,

panel k). Even after another two hours STAT1-GFP

was still predominantly nuclear (Fig. 1A, panel l). Pro-

longed nuclear accumulation of STAT1-GFP was seen

not only with HeLa cells, but also with 293T cells or

2fTGH cells (not shown). In addition, we transfected

STAT1-deficient U3A cells with wild-type or GFP-

tagged STAT1. Subsequently, the cells were fixed and

stained identically with a STAT1 antibody to observe

the nucleocytoplasmic translocation of both STAT1

variant proteins by indirect immunofluorescence micro-

scopy (Fig. 1B). As indicated by the staining intensi-

ties, both STAT1 variant proteins were expressed at

comparable levels. Moreover, as seen before in HeLa

cells (Fig. 1A), also in U3A cells the nuclear accumula-

tion of untagged STAT1 was terminated sooner. Wild-

type STAT1 accumulated in the nucleus only for

about one hour after interferon stimulation (Fig. 1B,

panels b,c). For GFP-tagged STAT1, however, the

accumulation phase lasted about twice as long

(Fig. 1B, panels i–k), before a pancellular distribution

was reached (Fig. 1B, panel l). Taken together, our

results confirm that the nuclear entry of activated

STAT1 did not differ between native and GFP-tagged

STAT1. The duration of nuclear accumulation, on the

other hand, was markedly prolonged by the fusion to

GFP.

The prolonged nuclear accumulation of

GFP-tagged STAT1 was associated with reduced

tyrosine phosphorylation and DNA binding

Because only unphosphorylated STAT1 can exit the

nucleus, the prolonged nuclear accumulation of

STAT1-GFP may be caused by defective tyrosine

dephosphorylation. This was examined in HeLa cells

transiently expressing STAT1-GFP. As can be seen

Fig. 1. GFP-tagging results in a prolonged nuclear accumulation of STAT1. (A) Shown is the time course of nuclear accumulation of either

endogenous or GFP-tagged STAT1 in interferon (IFN) c-stimulated HeLa cells, as determined by immunocytochemistry using a STAT1-speci-

fic antibody (left) and direct fluorescence microscopy (right), respectively. Cells were either left untreated (a, g) or treated with 5 ngÆmL)1

IFNc for 5 min (b, h), 10 min (c, i) or 60 min (d, j) before fixation. In (e, f, k, l) the cells were pretreated with IFNc for 60 min followed by

incubation for additional 4 h (e, k) or 6 h (f, l) in IFNc-free medium. Shown are fluorescence micrographs demonstrating the intracellular local-

ization of STAT1 and STAT1-GFP, respectively, and the corresponding Hoechst-stained nuclei. Note the rapid build-up of both endogenous

and GFP-tagged STAT1 within the nucleus, which occurred with similar kinetics, and the delayed nuclear export of STAT1-GFP as compared

to the wild-type protein. (B) STAT1-deficient U3A cells transiently expressing wild-type STAT1 (left) or STAT1-GFP (right) were left untreated

(a, g) or stimulated with IFNc for 60 min. Subsequently, the cells were fixed and stained right away (b, h), or the cells were fixed after con-

tinued incubation in the absence of interferon for the times indicated. Shown are indirect immunofluorescence results with a STAT1-specific

antibody.

T. Meyer et al. GFP-tagging of STAT1


from the western blot of Fig. 2A (lower panel), the

expression levels of endogenous STAT1 and recom-

binant STAT1-GFP were comparable. Stimulation of

the cells with IFNc induced the comparable tyrosine

phosphorylation of native and tagged STAT1. How-

ever, the phosphorylation signal decayed faster for

GFP-tagged STAT1 (Fig. 2A). A quantitative analysis

of the western blotting results clearly demonstrated

this outcome (Fig. 2B). Because high-affinity DNA

binding requires the tyrosine phosphorylation of

STAT1, we also analyzed the time course of DNA

binding. As is shown in Fig. 2C, endogenous and

STAT1-GFP bound to DNA predominantly as ho-

modimers, although a minor proportion of hetero-

dimers was also observed (Fig. 2C, asterisk). As

expected from the time course of STAT1 activation

shown in Fig. 2A, the DNA binding activity of

STAT1-GFP decayed faster than the DNA binding of

the endogenous STAT1 (see the quantitation,

Fig. 2D). We concluded that the fusion of STAT1

with GFP did not impair the DNA binding and the

tyrosine dephosphorylation reaction. Hence, the pro-

longed nuclear presence of STAT1-GFP was not due

to enhanced activation, but the result of nuclear

retention of unphosphorylated molecules.

Protein microinjection and pharmacological

export inhibition revealed that GFP-tagging

impairs the nucleocytoplasmic translocation

specifically of unphosphorylated STAT1

To directly analyze the influence of GFP-tagging on

the nucleocytoplasmic translocation of STAT1, we per-

formed microinjection studies with HeLa cells using

recombinant STAT1 and STAT1-GFP purified from

baculovirus-infected insect cells. Co-injected chromo-

phor-labelled bovine serum albumin failed to cross the

nuclear envelope of injected cells and thus served as a

A B

D

C

Fig. 2. Kinetics of tyrosine phosphorylation (A, B) and DNA binding activity (C, D) of untagged and GFP-tagged STAT1. Equal numbers of

HeLa cells coexpressing endogenous STAT1 and the recombinant STAT1-GFP (Fig. 1A) were either left untreated (– IFNc) or stimulated with

IFNc for the indicated times (+ IFNc, 1–4 h). (A) Shown is a western blot of whole cell extracts with a phospho-specific STAT1-Tyr701 anti-

body (upper), and a reprobing with anti-STAT1 IgG (lower). The positions of native STAT1 (lower mark) and the GFP fusion (upper mark) are

indicated. In (B) the normalized signal intensities for tyrosine-phosphorylated STAT1 [(signal Tyr-phosphorylated STAT1) ⁄ (signal unphosphoryl-

ated STAT1)] of three independent experiments were densitometrically quantified and plotted. (C, D) Protein extracts prepared from the

same cells as described in (A) were subjected to DNA binding analysis. Extracts were incubated with radiolabelled M67 probe and separated

on 4% nondenaturing polyacrylamide gels. The positions of homodimers of endogenous STAT1 (lower mark), of homodimers of recombinant

STAT1-GFP (upper mark), and of heterodimers thereof (asterisk) are indicated at the right margin of the gel. The arrowhead marks an unspe-

cific band. Included are whole cell extracts from STAT1-negative U3A cells (lane 1) and two supershift experiment with HeLa cell extracts in

the presence of anti-GFP (lane 2) and anti-STAT1 IgG (lane 3), respectively. Results from quantification of three independent experiments

are shown in (D).



marker for the injection site. At first, STAT1 and

STAT1-GFP were injected in the cytoplasm, followed

by the stimulation of cells with IFNc for 30 min or

60 min to trigger tyrosine phosphorylation. Thereafter,

cells were fixed and processed for immunocytochemis-

try (wild-type), or the GFP epifluorescence was

observed directly (STAT1-GFP). As is shown in

panels A–D of Fig. 3, both STAT1 variants rapidly

accumulated in the nucleus, irrespective of the GFP

domain. This result confirms earlier reports of unper-

turbed nuclear import of activated STAT1-GFP

[17,18].

Fig. 3. Microinjection of purified, recombin-

ant STAT1 reveals impaired nucleocytoplas-

mic translocation of GFP-tagged STAT1.

Unphosphorylated wild-type STAT1 or the

GFP fusion thereof were purified from bacu-

lovirus-infected Sf9 cells and injected at con-

centrations of 1 mgÆmL)1 into either the

cytosol (A–J) or the nucleus (K–P) of resting

HeLa cells grown on poly L-lysine-coated

glass coverslips. As a marker for the injec-

tion site 0.2 mgÆmL)1 tetramethylrhodamine

isothiocyanate- (T-) or fluorescein isothiocya-

nate-labelled bovine serum albumin (F-BSA)

was coinjected. In (A–D) the injected cells

were stimulated with 5 ngÆmL)1 IFNc and in

(E–P) the cells were left untreated. At the

indicated times after injection the cells were

fixed and stained with anti-STAT1 IgG C-24

(A, B, E–G, K–M) or the GFP fluorescence

was observed directly (C, D, H–J, N–P). For

each condition about 20 cells were success-

fully microinjected, one of which is shown.



Next, we examined the impact of the GFP domain

on the nuclear import of unphosphorylated STAT1

(Fig. 3E–J). The experimental set up was the same as in

panels A–D, but the stimulation of cells with interferon

was omitted. As expected, unphosphorylated STAT1

rapidly entered the nucleus, leading to a nearly pancel-

lular distribution already after 30 min (Fig. 3E) that

did not change much for another 90 min (Fig. 3F,G).

GFP-tagged STAT, however, had a reduced rate of

nuclear import. After 30 min only a small amount of

the injected GFP variant was detectable in the nucleus

(Fig. 3H), and even after another 30 min or 90 min the

concentration of STAT1-GFP remained lower in the

nucleus than in the cytoplasm (Fig. 3I,J).

Similar results were obtained for the nuclear export

of unphosphorylated STAT1. Wild-type or GFP-

tagged STAT1 was microinjected into the nuclei of

resting cells and the appearance of STAT1 in the cyto-

plasm was monitored (Fig. 3K–P). As early as 15 min

after nuclear delivery of STAT1, significant amounts

of the wild-type protein were detected outside of the

nucleus (Fig. 3K) and after 30–60 min the pancellular

distribution was reached (Fig. 3L,M). In contrast, nuc-

lear export of the GFP fusion protein was significantly

reduced, as it took more than 30 min to detect measur-

able fluorescence signals outside the injected nucleus

(Fig. 3N,O). Even after 60 min the majority of the

injected STAT1-GFP was still restricted to the nucleus,

indicating that the fusion to GFP had diminished the

nuclear export rate (Fig. 3P).

The microinjection studies with purified exogenous

recombinant proteins shown in Fig. 3 indicated that

import of GFP-tagged STAT1 was reduced in unstimu-

lated cells. We wanted to confirm this conclusion also

for endogenous STAT1 variants expressed in transfected

cells. However, two obstacles complicate the analysis of

STAT1 nuclear import in unstimulated cells. Most

importantly, there are no drugs available to induce or

inhibit the constitutive cytokine-independent nucleocy-

toplasmic transport of STAT1. Additionally, STAT1 is

only slightly more concentrated in the cytoplasm of un-

stimulated cells, making it difficult to clearly discern

changes in the nucleocytoplasmic distribution. In order

to overcome these limitations, we placed a canonical

nuclear export signal (NES) at the C-terminus of

STAT1, thus generating STAT1-NES and STAT1-NES-

GFP. Due to their increased nuclear export rates both

NES mutants displayed an exclusively cytoplasmic

localization in resting HeLa cells (Fig. 4A,D). Notably,

the incubation with leptomycin B (LMB) rapidly

Fig. 4. Analysis of the nuclear import of unphosphorylated STAT1. Shown is the distribution of STAT1 that expresses a nuclear export signal

(NES) alone (STAT1-NES; A–C) or in combination with GFP (STAT1-NES-GFP; D–F). Untreated cells (A, D) display the cytoplasmic accumula-

tion of both STAT1 variant proteins. Inactivation of the NES receptor CRM1 with leptomycin B (LMB) for 60 min (B, E) or 120 min (C, F)

results in the nuclear translocation of STAT1. The STAT1 distribution was detected immunocytochemically in fixed HeLa cells by using anti-

STAT1 IgG C-24 and an appropriate Cy3-conjugated secondary antibody. Note that LMB treatment did not result in the accumulation of

STAT1 in the nucleus, thus indicating continued CRM1-independent nuclear export. In Figs 4 and 5 the exposure time was reduced to sup-

press detection of the endogenous STAT1 (see Experimental procedures).



incapacitates the NES receptor CRM1 [22]. This, in

turn, triggers the collapse of STAT1-NES cytoplasmic

accumulation. Thus, we generated an inducible system

to analyze the influence of GFP-tagging on the cyto-

kine-independent nuclear import of STAT1 (Fig. 4).

Cells expressing STAT1-NES or STAT1-NES-GFP

were fixed one or two hours after the addition of LMB,

and STAT1 was detected immunocytochemically. In

accordance with the protein microinjection results

shown in Fig. 3, we found protracted nuclear import of

GFP-tagged STAT1 also in transfected cells. While the

GFP fusion protein was still predominantly cytoplasmic

one hour after the addition of LMB (Fig. 4E), the un-

tagged STAT1-NES had already reached the pancellular

distribution at this time point (Fig. 4B).

Conversely, the NES-variants of STAT1 can also be

used to analyze nuclear export. For this purpose cells

expressing the NES mutants were treated with IFNc in

addition to LMB, which resulted in their comparable

nuclear accumulation after 60 min (Fig. 5B,F). The

cells were then incubated for another two or four

hours in the presence of LMB, before the cells were

fixed and immunostained for STAT1. As was seen in

Fig. 1A and 1B for the endogenous and transfected

wild-type protein in comparison to STAT1-GFP, also

STAT1-NES returned to the cytoplasm much more

quickly than the respective GFP-tagged variant. The

non-GFP-tagged STAT1-NES displayed a pancellular

distribution after 4 h (Fig. 5D), whereas the respective

GFP-tagged mutant was still predominantly nuclear at

this time point (Fig. 5H). Taken together, the results

of Figs 3–5 demonstrate the inhibitory influence of

GFP-tagging on both the nuclear import and export

specifically of unphosphorylated STAT1. Based on the

time required to achieve pancellular distribution, it

appeared that the translocation rates were reduced at

least by a factor of 2.

GFP-tagging decreases the transcriptional

activity of STAT1 by reducing its

nucleocytoplasmic shuttling

We have demonstrated here that nucleocytoplasmic

shuttling of GFP-tagged STAT1 is slow in comparison

to native STAT1. Hence the cytokine sensitivity of

STAT1-GFP was reduced accordingly (Fig. 2). Finally,

Fig. 5. Analysis of the nuclear export of unphosphorylated STAT1. HeLa cells expressing STAT1-NES (A–D) or STAT1-NES-GFP (E–H) were

left unstimulated or stimulated for 60 min with 5 ngÆmL)1 IFNc and 10 ngÆmL)1 LMB to induce nuclear accumulation. Cells were then fixed

immediately (A, B, E, F), or the incubation was continued for 2 h (C, G) or 4 h (D, H) in IFNc-free medium. Detection of STAT1 localization

was by immunocytochemistry, and the nuclei were stained with Hoechst dye. Note the indiscriminate nuclear accumulation resulting from

IFNc and leptomycin B treatment, and the slow return to the cytoplasm of the GFP-tagged variant.



we therefore examined the influence of the GFP fusion

on the STAT1-dependent activation of a luciferase

reporter gene. As shown in Fig. 6A, the gene induction

by wild-type STAT1 was more than twice as efficient

as by the GFP fusion protein, despite similar expres-

sion levels of both proteins. Because GFP was fused to

the C-terminal transactivation domain of STAT1, tran-

scription could be compromised not only by the effects

of GFP on nucleocytoplasmic shuttling, but also by

reduced recruitment of transcription cofactors due to

the presence of additional residues close to the tran-

scription activation domain. In order to discriminate

between these possibilities, we included in this analysis

the mutant STAT1-NES. As we have demonstrated

previously, the addition of a nuclear export signal to

STAT1 resulted in comparable nuclear export rates

([21], see also Fig. 4) and hence minimized the activa-

tion differences among STAT1 variant proteins [21].

A

C

B

Fig. 6. Diminished nucleocytoplasmic shuttling of STAT1 is associated with a decrease in transcriptional activity. (A) STAT1-negative U3A

cells were cotransfected with plasmids coding for either wild-type STAT1 or STAT1-GFP together with an IFNc-inducible luciferase reporter

gene construct and a plasmid for constitutive expression of b-galactosidase. Luciferase activity was determined 24 h post-transfection in

unstimulated cells (grey bars) or after stimulation with IFNc for 6 h (black bars). The luciferase expression induced by STAT1 and STAT1-

NES were set to 100%. Error bars represent standard deviations for six independent experiments normalized to the expression of b-galac-

tosidase. (B) Gel shift analysis with wild-type STAT1, STAT1-NES and STAT1-NES-GFP and a M67 STAT1-binding site. U3A cells expressing

the STAT1 variant proteins were treated with IFNc for 60 min or left unstimulated. Soluble proteins were extracted sequentially from the

cytosol and the nucleus, and the combined extracts were used for further analyses. Extracts from IFN-stimulated cells were analyzed first

by western blotting to determine the concentration of activated STAT1 (not shown). Subsequently, the IFN-treated extracts were normalized

(see Experimental procedures), such that an identical quantity of each activated STAT1 variant was incubated with the radioactive DNA

probe. Shown is the autoradiogram of the native PAGE. (C) U3A cells were cotransfected with plasmids encoding STAT1-NES or STAT1-

NES-GFP and reporter genes as described in (A). The cells were stimulated with IFNc in the absence (black bars) or presence (grey bars) of

leptomycin B. Gene induction in unstimulated cells was barely detectable both without (white bars) or with (not shown) leptomycin. The

reporter gene activity was analysed as described (A). The inserts in (A) and (C) show western blotting results of whole cell extracts prepared

from cells used for the reporter gene analyses. The extracts were probed with a pan-STAT1 antibody (top) and reprobed with an antibody

against b-actin (bottom).



Moreover, the addition of the export signal did not

diminish DNA binding of STAT1 (Fig. 6B) Remark-

ably, however, when the export differences were equal-

ized by the inclusion of the export signal, the

expression of the GFP domain no longer reduced

the transcriptional activity of STAT1 (Fig. 6C). In the

presence of LMB, which specifically inactivates the

exportin CRM1 and hence blocks the export enhance-

ment conferred by the NES (Fig. 4), the transcriptional

activity of GFP-tagged STAT1 again reached only

about 50% of the untagged STAT1 (Fig. 6C). We

therefore concluded that the observed inhibitory effect

of GFP on gene transcription could be attributed

primarily to its effects on STAT1 nucleocytoplasmic

shuttling.

Discussion

The results presented here demonstrated that the nucleo-

cytoplasmic translocation of the STAT1-GFP fusion

protein differed from the endogenous wild-type protein.

Our data indicate an inhibitory influence of GFP-tag-

ging specifically on the constitutive nucleocytoplasmic

shuttling of unphosphorylated STAT1. We used several

independent experimental approaches to exclude stain-

ing differences and overexpression artifacts as the cause

of the reduced nucleocytoplasmic shuttling. We used

C-terminally GFP-tagged STAT1 in our experiments,

which was reported before to be indistinguishable from

the wild-type protein [17,18,23]. However, only the ini-

tial phases of the cytokine-induced nuclear accumula-

tion were considered in those studies. Indeed, our data

confirm that the cytokine-induced nuclear import of

activated STAT1 and the onset of nuclear accumulation

are not disturbed by the presence of the GFP domain at

the STAT1 C-terminus. Yet, the nuclear accumulation

of STATs is highly dynamic and dependent on their

continuous nuclear export with subsequent rephosph-

orylation at the cell membrane. Accordingly, the acti-

vation level is immediately reduced when the

shuttling rate falls. Consistent with this model,

impaired cycling caused by GFP-tagging resulted in

diminished activation and DNA binding. As expec-

ted, this also adversely affected the transcriptional

activity. While addition of the GFP domain to the

STAT1 C-terminal transactivation domain did not

inhibit transcriptional activation during stimulation of

cells with IFNc per se, the inefficient cycling never-

theless resulted in a strongly decreased transcriptional

yield. These results underscore the physiological

importance of continuous nucleocytoplasmic cycling

for the cytokine-dependent gene regulatory functions

of STAT1.

GFP-tagging is used extensively to examine the

intracellular mobility and nucleocytoplasmic shuttling

of proteins. For STAT1, some of these studies indicate

the relatively slow shuttling before the stimulation of

cells with cytokines, and the rapid nuclear import of

tyrosine-phosphorylated molecules during cytokine sti-

mulation leading to nuclear accumulation [17–19]. The

results presented in this work urge caution concerning

the assignment of results that were obtained with

GFP-tagged STAT1 to the native protein. Our data

indicate that the constitutive shuttling rates of STAT1-

GFP were reduced at least by a factor of two, while

the nuclear import of activated STAT1-GFP was not

measurably affected. We cannot conclude from our

data that the constitutive and cytokine-induced trans-

locations of native STAT1 occur with identical rates.

However, the results presented here indicate that the

translocation rates of STAT1 before and after cytokine

stimulation are much more similar than suggested by

fluorescence recovery after photobleaching (FRAP)

analyses with GFP-tagged molecules. This conclusion

is supported also by experiments performed with a

STAT1 mutant that has lost the ability to interact with

DNA [10]. Although there was no obvious difference

in the nuclear import of the activated mutant in com-

parison to wild-type STAT1, there was nevertheless no

clearly discernible nuclear accumulation [10]. However,

if nuclear import was stimulated substantially by IFN

treatment whereas the constitutive shuttling remained

at a slow rate, the outcome should be nuclear accumu-

lation. Because this is not the case, increased nuclear

import is unlikely to account for the transient cyto-

kine-induced accumulation of STAT1 in the nucleus.

Rather, the nuclear retention of activated molecules

appears to play a prominent role. We thus conclude

that translocation analyses by photobleaching tech-

niques in combination with GFP-tagging do not

properly reflect the dynamic redistribution of unphos-

phorylated wild-type STAT1, because the nuclear

envelope represents a major diffusion barrier for the

constitutive cycling of STAT1-GFP. Photobleaching

techniques, however, are well suited to examine the

mobility of STAT1 in the cytoplasm or the nucleo-

plasm, because the STAT1-GFP mobility, with or

without ligand stimulation, is comparable to that of

freely diffusible GFP [18].

It is interesting to note that the predominantly car-

rier-independent cycling of unphosphorylated STAT1

was strongly affected by GFP, while the carrier-

dependent nuclear import of tyrosine-phosphorylated

STAT1 was not perturbed. Previous work has shown

that the transport rate of protein cargo is dependent

predominantly on its size and surface properties



[24–26]. Unphosphorylated and phosphorylated

STAT1 probably are very similar in terms of their

molecular mass, as there is evidence that both exist as

stable dimers (molecular mass � 170 kDa) [27].

Unphosphorylated STAT1 enters the nucleus without

the assistance of karyopherins via direct interactions

with the nuclear pore [9,28]. Two regions at the

STAT1 C-terminus have been implicated in the carrier-

independent nuclear translocation, the linker domain

and the C-terminal transactivation domain [9,21]. In

addition, the serine phosphorylation of the transactiva-

tion domain was shown to enhance nuclear export of

STAT1 [21]. However, as STAT1-GFP is highly ser-

ine-phosphorylated (not shown), it is likely that the

presence of the GFP protein interferes with the bind-

ing of STAT1 to the nuclear pore proteins, thus redu-

cing its ability to engage in productive interactions.

Phosphorylated STAT1 dimers, on the other hand,

cannot traverse the nuclear pore on their own, but

require both importin a and importin b in order to

enter the nucleus [9,12,13,23]. The karyopherins may

function as chaperones during translocation through

the pores. Irrespective of the presence of the GFP

domain, these large proteins (molecular mass

>58 kDa) are likely to provide extensive interfaces

with the nuclear pore that are necessary for the pas-

sage of activated STAT1.

It is well known that genetically encoded fluorescent

tags can perturb the behavior of the acceptor protein.

This study demonstrates that predominantly carrier-

independent translocation mechanisms are altered by

the addition of a GFP-fluorescent tag, while carrier-

dependent import was not affected. An increasing

number of signaling proteins is known to use both car-

rier-dependent as well as carrier-independent mecha-

nisms during nucleocytoplasmic cycling [29]. In those

cases the mechanistically distinct translocation events

have to be analyzed separately and both without and

with GFP-tagging, in order to identify the possible

limitations of GFP-tagged derivatives.

Experimental procedures

Plasmids

Mammalian expression plasmids encoding wild-type human

STAT1 and STAT1 cDNA fused carboxy-terminally to

green fluorescent protein were described [10,30]. The

construct pSTAT1-NES-GFP was generated by PCR ampli-

fication of pGST-NES-GFP using the primer pair 5¢-ATA

TATGGATCCAGATAAAGATGTGAATGAG-3¢ and

5¢-CGCCCCGACACCCGCCAACACCC-3¢ and Vent

DNA polymerase (New England Biolabs, Frankfurt am

Main, Germany). The plasmid pGST-NES-GFP encodes

residues 367–427 of human STAT1, which confers NES

activity [28]. The PCR product was subsequently digested

with BamHI and NotI and inserted into the corresponding

sites of pSTAT1-GFP. The ATG start codon of GFP was

mutated to TAG to prevent expression of the GFP domain,

thus generating pSTAT1-NES. Site-directed mutagenesis

was done with the Quik-Change kit (Stratagene, Amster-

dam, the Netherlands) and specific primers. Recombinant

STAT1 proteins were purified from Sf9 insect cells that had

been infected with baculovirus transfer vectors (pFastBac)

encoding wild-type STAT1 or STAT1-GFP [9].

Cell culture and DNA transfections

Human HeLa cells, 393T cells, and U3A cells were grown

in Dulbecco’s modified Eagle’s medium (DMEM) supple-

mented with 10% fetal bovine serum (Biochrom, Berlin,

Germany) and 1% penicillin ⁄ streptomycin in a humidified

7% CO2 atmosphere. For immunocytochemical studies the

cells were passaged on poly l-lysine-coated glass coverslips

in 12-well plates. The next day the cells were transiently

transfected with 0.8 lgÆwell)1 expression plasmid by the

Lipofectamine method according to the manufacturer’s

instructions (Invitrogen, Karlsruhe, Germany). Twenty-four

hours post-transfection the cells were either stimulated with

5 ngÆmL)1 IFNc (Biomol, Hamburg, Germany) for the

indicated times or left unstimulated. For CRM1 inhibition

the cells were exposed to 10 ngÆmL)1 leptomycin B (Sigma,

Munich, Germany).

Purification of recombinant proteins

STAT1 and STAT1-GFP, both of which contained a C-ter-

minal Strep-tag, were expressed in baculovirus-infected Sf9

cells and purified from cell lysates using a Strep-Tactin col-

umn [9]. Both STAT1 variants were concentrated by ultra-

filtration in NaCl ⁄Pi.

Microinjection

Purified recombinant STAT1 proteins were injected at a

concentration of 1 mgÆmL)1 into either the cytosol or nuc-

leus of HeLa cells. As a marker for the injection site,

0.2 mgÆmL)1 BSA coupled to fluorescein isothiocyanate or

tetramethylrhodamine isothiocyanate (both from Sigma)

was present in the injection solution. Microinjections were

performed with plastic capillaries (Femtotips; Eppendorf,

Hamburg, Germany) and the Transjector 5246 attached to

the Micromanipulator 5171 (Eppendorf). Injection of GFP-

expressing cells was monitored under an inverted micro-

scope (Axiovert 25) equipped with UV light emission (Zeiss,

Oberkochen, Germany). Typically, 30 cells were injected

within 10 min using a maximal pressure of 40 hPa. After



the indicated incubation periods at 37 �C the cells were

fixed for 10 min at )20 �C in cold methanol. Injected

STAT1 was detected with monoclonal antibody C-136 as

described [9].

Immunocytochemistry

Indirect immunofluorescence microscopy of untransfected

cells and cells transiently expressing recombinant STAT1

was done with affinity-purified rabbit polyclonal STAT1-

specific antibody C-24 (Santa Cruz Biotechnology, Heidel-

berg, Germany). Briefly, after methanol-fixation, cells were

treated for 45 min with 25% fetal bovine serum ⁄NaCl ⁄Pi.

Incubation with anti-STAT1 IgG C-24 diluted 1 : 1000 in

25% fetal bovine serum ⁄NaCl ⁄Pi was for 45 min at room

temperature. After three washes with NaCl ⁄Pi the cells were

incubated for an additional 45 min with Cy3-labelled goat

anti-rabbit IgG (Jackson ImmunoResearch, Cambridge,

UK). For direct detection of GFP fusion proteins, cells

were fixed for 10 min at room temperature in 3.7% para-

formaldehyde in NaCl ⁄Pi. Subsequently, the nuclei were

stained for 5 min with 1 lgÆmL)1 Hoechst 33258 (Sigma)

and the samples were mounted in fluorescence mounting

medium (Dako, Hamburg, Germany).

Fluorescence microscopy

Fluorescence microscopy was performed using a Zeiss Axio-

plan 2 imaging microscope equipped with the appropriate

fluorescence filters. Fluorescence images acquired with a 63·oil immersion lens were taken with an AxioCam CCD cam-

era and were processed with the Zeiss axiovision and

Adobe photoshop (Adobe Systems, San Jose, CA, USA)

packages. For Figs 4 and 5, the exposure time used for

recording the STAT1 fluorescence was reduced to 100 ms

(500 ms were used in Fig. 1A) to avoid overexposure.

Western blotting

Transfected HeLa or U3A cells growing on a 6 cm dish

were lysed with 150 lL whole cell lysis buffer (50 mm

Tris ⁄HCl, pH 7.4, 280 mm NaCl, 0.5% Nonidet P-40,

0.2 mm EDTA, 2 mm EGTA, 10% glycerol, 10 mm b-gly-cerol phosphate, 50 mm NaF, 1 mm Na3VO4, 2 mm dithio-

treitol, 0.1 mm phenylmethylsulfonyl fluoride, supplemented

with Complete protease inhibitors from Roche, Penzberg,

Germany) on ice for 30 min. The extracts were centrifuged

at 16 000 g and 4 �C for 5 min and the supernatants boiled

in SDS sample buffer. The proteins were resolved on 7%

SDS-polyacrylamide gels and transferred to nitrocellulose.

The membranes were blocked with 4% BSA in Tris-buffered

saline plus 0.1% Tween-20 (TBS-T) for 45 min and incuba-

ted for an additional 45 min with a polyclonal antibody

exclusively reacting with tyrosine-phosphorylated STAT1

(Cell Signaling, Frankfurt am Main, Germany), which was

diluted 1 : 1000 in TBS-T. Then the membranes were

washed three times in TBS-T and relevant bands were detec-

ted by incubating with horseradish peroxidase-conjugated

secondary antibody (Dako) using the chemiluminescence

reaction kit (Amersham Biosciences, Freiburg, Germany).

The membranes were stripped off bound peroxidase activity

by incubation in stripping buffer (62.5 mm Tris ⁄HCl, 2%

SDS, 0.7% 2-mercaptoethanol) for 40 min at 50 �C. The

membranes were finally incubated with the pan-STAT1 anti-

body C-24 and re-exposed to secondary antibodies. Alter-

natively, the blots were reprobed with a monoclonal b-actinantibody (Sigma) as the loading control.

Gelshift and reporter gene assays

Electromobility shift assays using whole cell extracts and

reporter gene assays were performed as previously described

[10,30]. IFN-treated extracts used in Fig. 6B were normal-

ized by the addition of U3A whole cell extracts, such that

identical amounts of activated STAT1 were incubated

with the radioactive probe. For supershift assays the

STAT1-specific antibody C-24 or an anti-GFP polyclonal

antibody (obtained from rabbits immunized with purified

bacterially expressed GFP-GST fusion protein) was used at

a concentration of 40 lgÆmL)1.

Acknowledgements

The authors thank Melanie Lange for expert technical

support. This work was financed in part by grants from

the Deutsche Forschungsgemeinschaft and the Bundes-

ministerium fur Bildung und Forschung (BioFuture).

References

1 Levy DE & Darnell JE Jr (2002) Stats: transcriptional

control and biological impact. Nat Rev Mol Cell Biol 3,

651–662.

2 Greenlund AC, Morales MO, Viviano BL, Yan H,

Krolewski J & Schreiber RD (1995) Stat recruitment by

tyrosine-phosphorylated cytokine receptors: an ordered

reversible affinity-driven process. Immunity 2, 677–687.

3 Horvath CM, Wen Z & Darnell JE Jr (1995) A STAT

protein domain that determines DNA sequence recogni-

tion suggests a novel DNA-binding domain. Genes Dev

9, 984–994.

4 Shuai K, Stark GR, Kerr IM & Darnell JE Jr (1993) A

single phosphotyrosine residue of STAT91 required for

gene activation by interferon-gamma. Science 261,

1744–1746.

5 Shuai K, Horvath CM, Huang LH, Qureshi SA, Cow-

burn D & Darnell JE Jr (1994) Interferon activation of

the transcription factor Stat91 involves dimerization



through SH2–phosphotyrosyl peptide interactions. Cell

76, 821–828.

6 Decker T, Kovarik P & Meinke A (1997) GAS ele-

ments: a few nucleotides with a major impact on cyto-

kine-induced gene expression. J Interferon Cytokine Res

17, 121–134.

7 Schindler C, Shuai K, Prezioso VR & Darnell JE Jr

(1992) Interferon-dependent tyrosine phosphorylation of

a latent cytoplasmic transcription factor. Science 257,

809–813.

8 Meyer T & Vinkemeier U (2004) Nucleocytoplamic

shuttling of STAT transcription factors. Eur J Biochem

271, 4606–4612.

9 Marg A, Shan Y, Meyer T, Meissner T, Brandenburg

M & Vinkemeier U (2004) Nucleocytoplasmic shuttling

by nucleoporins Nup153 and Nup214 and CRM1-

dependent nuclear export control the subcellular distri-

bution of latent Stat1. J Cell Biol 165, 823–833.

10 Meyer T, Marg A, Lemke P, Wiesner B & Vinkemeier

U (2003) DNA binding controls inactivation and

nuclear accumulation of the transcription factor Stat1.

Genes Dev 17, 1992–2005.

11 Sekimoto T, Nakajima K, Tachibana T, Hirano T &

Yoneda Y (1996) Interferon-c-dependent nuclear import

of Stat1 is mediated by the GTPase activity of

Ran ⁄TC4. J Biol Chem 271, 31017–31020.

12 Sekimoto T, Imamoto N, Nakajima K, Hirano T &

Yoneda Y (1997) Extracellular signal-dependent nuclear

import of Stat1 is mediated by nuclear pore-targeting

complex formation with NPI-1, but not Rch1. EMBO J

16, 7067–7077.

13 Fagerlund R, Melen K, Kinnunen L & Julkunen I

(2002) Arginine ⁄ lysine-rich nuclear localization signals

mediate interactions between dimeric STATs and impor-

tin a5. J Biol Chem 277, 30072–30078.

14 Haspel RL, Salditt-Georgieff M & Darnell JE Jr (1996)

The rapid inactivation of the nuclear tyrosine phos-

phorylated Stat1 depends upon a protein tyrosine phos-

phatase. EMBO J 15, 6262–6268.

15 Axelrod D, Koppel DE, Schlessinger J, Elson E &

Webb WW (1976) Mobility measurement by analysis of

fluorescence photobleaching recovery kinetics. Biophys J

16, 1055–1069.

16 Patterson GH, Knobel SM, Sharif WD, Kain SR & Pis-

ton DW (1997) Use of the green fluorescent protein and

its mutants in quantitative fluorescence microscopy.

Biophy J 73, 2782–2790.

17 Koster M & Hauser H (1999) Dynamic redistribution

of STAT1 protein in IFN signaling visualized by

GFP fusion proteins. Eur J Biochem 260,

137–144.

18 Lillemeier BF, Koster M & Kerr IM (2001) STAT1

from the cell membrane to the DNA. EMBO J 20,

2508–2517.

19 Koster M, Frahm T & Hauser H (2005) Nucleocytoplas-

mic shuttling revealed by FRAP and FLIP technologies.

Curr Opin Biotechnol 16, 28–34.

20 Swameye I, Muller TG, Timmer J, Sandra O & Kling-

muller U (2003) Identification of nucleocytoplasmic

cycling as a remote sensor in cellular signaling by data-

based modeling. Proc Natl Acad Sci USA 100, 1028–

1033.

21 Lodige I, Marg A, Wiesner B, Malecova B, Oelgeschla-

ger T & Vinkemeier U (2005) Nuclear export determines

the cytokine sensitivity of STAT transcription factors.

J Biol Chem 280, 43087–43099.

22 Kudo N, Wolff B, Sekimoto T, Schreiner EP, Yoneda

Y, Yanagida M, Horinouchi S & Yoshida M (1998)

Leptomycin B inhibition of signal-mediated nuclear

export by direct binding to CRM1. Exp Cell Res 242,

540–547.

23 McBride KM, Banninger G, McDonald C & Reich NC

(2002) Regulated nuclear import of the STAT1 trans-

cription factor by direct binding of importin-a. EMBO

J 21, 1754–1763.

24 Feldherr CM & Akin D (1990) The permeability of the

nuclear envelope in dividing and nondividing cell cul-

tures. J Cell Biol 111, 1–8.

25 Ribbeck K & Gorlich D (2002) The permeability barrier

of nuclear pore complexes appears to operate via hydro-

phobic exclusion. EMBO J 21, 2664–2671.

26 Lyman SK, Guan T, Bednenko J, Wodrich H & Gerace

L (2002) Influence of cargo size on Ran and energy

requirements for nuclear protein import. J Cell Biol

159, 55–67.

27 Braunstein J, Brutsaert S, Olson R & Schindler C

(2003) STATs dimerize in the absence of phosphoryla-

tion. J Biol Chem 278, 34133–34140.

28 Meyer T, Begitt A, Lodige I, van Rossum M & Vinke-

meier U (2002) Constitutive and IFN-c-induced nuclear

import of STAT1 proceed through independent path-

ways. EMBO J 21, 344–354.

29 Xu L & Massague J (2004) Nucleocytoplasmic shuttling

of signal transducers. Nat Rev Mol Cell Biol 5, 209–219.

30 Begitt A, Meyer T, van Rossum M & Vinkemeier U

(2000) Nucleocytoplasmic translocation of Stat1 is regu-

lated by a leucine-rich export signal in the coiled-coil

domain. Proc Natl Acad Sci USA 97, 10418–10423.



Green fluorescent protein-tagging reduces the nucleocytoplasmic shuttling specifically of...

Documents

Transcript of Green fluorescent protein-tagging reduces the nucleocytoplasmic shuttling specifically of...