Stat3 Isoforms, α and β, Demonstrate Distinct Intracellular Dynamics with Prolonged Nuclear Retention of Stat3β Mapping to Its Unique C-terminal End*

Two isoforms of Stat3 (signal transducer and activator of transcription 3) are expressed in cells, α (p92) and β (p83), both derived from a single gene by alternative mRNA splicing. The 55-residue C-terminal transactivation domain of Stat3α is deleted in Stat3β and replaced by seven unique C-terminal residues (CT7) whose function remains uncertain. We subcloned the open reading frames of Stat3α and Stat3β into the C terminus of green fluorescent protein (GFP). Fluorescent microscopic analysis of HEK293T cells transiently transfected with GFP-Stat3α or GFP-Stat3β revealed similar kinetics and cytokine concentration dependence of nuclear accumulation; these findings were confirmed by high throughput microscope analysis of murine embryonic fibroblasts that lacked endogenous Stat3 but stably expressed either GFP-Stat3α or GFP-Stat3β. However, although time to half-maximal cytoplasmic reaccumulation after cytokine withdrawal was 15 min for GFP-Stat3α, it was >180 min for GFP-Stat3β. Furthermore, although the intranuclear mobility of GFP-Stat3α was rapid and increased with cytokine stimulation, the intranuclear mobility of GFP-Stat3β in unstimulated cells was slower than that of GFP-Stat3α in unstimulated cells and was slowed further following cytokine stimulation. Deletion of the unique CT7 domain from Stat3β eliminated prolonged nuclear retention but did not alter its intranuclear mobility. Thus, Stat3α and Stat3β have distinct intracellular dynamics, with Stat3β exhibiting prolonged nuclear retention and reduced intranuclear mobility especially following ligand stimulation. Prolonged nuclear retention, but not reduced intranuclear mobility, mapped to the CT7 domain of Stat3β.

Two isoforms of Stat3 (signal transducer and activator of transcription 3) are expressed in cells, ␣ (p92) and ␤ (p83), both derived from a single gene by alternative mRNA splicing. The 55-residue C-terminal transactivation domain of Stat3␣ is deleted in Stat3␤ and replaced by seven unique C-terminal residues (CT7) whose function remains uncertain. We subcloned the open reading frames of Stat3␣ and Stat3␤ into the C terminus of green fluorescent protein (GFP). Fluorescent microscopic analysis of HEK293T cells transiently transfected with GFP-Stat3␣ or GFP-Stat3␤ revealed similar kinetics and cytokine concentration dependence of nuclear accumulation; these findings were confirmed by high throughput microscope analysis of murine embryonic fibroblasts that lacked endogenous Stat3 but stably expressed either GFP-Stat3␣ or GFP-Stat3␤. However, although time to half-maximal cytoplasmic reaccumulation after cytokine withdrawal was 15 min for GFP-Stat3␣, it was >180 min for GFP-Stat3␤. Furthermore, although the intranuclear mobility of GFP-Stat3␣ was rapid and increased with cytokine stimulation, the intranuclear mobility of GFP-Stat3␤ in unstimulated cells was slower than that of GFP-Stat3␣ in unstimulated cells and was slowed further following cytokine stimulation. Deletion of the unique CT7 domain from Stat3␤ eliminated prolonged nuclear retention but did not alter its intranuclear mobility. Thus, Stat3␣ and Stat3␤ have distinct intracellular dynamics, with Stat3␤ exhibiting prolonged nuclear retention and reduced intranuclear mobility especially following ligand stimulation. Prolonged nuclear retention, but not reduced intranuclear mobility, mapped to the CT7 domain of Stat3␤.
Stat3␤ was isolated initially in the mouse using a yeast twohybrid strategy with the N terminus of c-Jun as bait (6). The CT7 domain was presumed responsible for this interaction; however, subsequent studies revealed that residues within the coiled-coil and DNA binding domains of Stat3, common to both isoforms, were found to be responsible for c-Jun interaction, thereby leaving unresolved the functional role of the CT7 domain if any.
To gain additional insight into the unique aspects of Stat3␤ at the single cell level and how the CT7 domain may contribute to them, we subcloned the open reading frames of Stat3␣ and Stat3␤ into the C terminus of GFP and transfected them into HEK293T and mouse embryo fibroblast (MEF) cells; endogenous Stat3␣ and Stat3␤ were deleted within MEF cells using Cre-lox technology. Cytokine stimulation of the resultant cell lines revealed similar kinetics of nuclear accumulation for GFP-Stat3␣ and GFP-Stat3␤. However, cytoplasmic reaccumulation after ligand removal, although rapid for GFP-Stat3␣ (t1 ⁄ 2 ϭ 15 min), was markedly delayed for GFP-Stat3␤ (t1 ⁄ 2 Ͼ 180 min). Prolonged nuclear retention of Stat3␤ was accompanied by reduced intranuclear mobility and increased immobile fraction. Interestingly, prolonged nuclear retention but neither reduced intranuclear mobility nor increased immobile fraction mapped to the CT7 domain of Stat3␤.
Plasmid-GFP-Stat3␣ and GFP-Stat3␤ expression vectors were constructed by subcloning the full-length cDNA of human Stat3␣ and Stat3␤ in frame 3Ј of GFP contained within pEGFP-C1 (Clontech) using the appropriate enzyme digestion fragments. The truncated Stat3␤ mutant construct (deletion of the C-terminal 7 amino acids; Stat3␤⌬) was generated by substituting a stop codon for that encoding the first in the 7-amino acid sequence using QuikChange II site-directed mutagenesis kit (Stratagene). All constructs were confirmed by DNA sequencing. An acute phase response element (APRE)-luciferase reporter construct, which has four copies of APREs, was kindly provided by Dr. I. Matsumura (Osaka, Japan). The fulllength G-CSF receptor cDNA was obtained and used as described previously (22).
Luciferase Assay-For transient transfections, MEFs were grown in 6-well (35-mm diameter) tissue culture plates to 50 -70% confluence. Twelve hours later, the cells were transiently transfected with the indicated expression vectors and reporter genes using the GeneJuice transfection reagents according to the manufacturer's instruction as previously reported (23). The amounts of plasmid DNA used per well were 250 ng of reporter vector, 1 g of expression vector, and 250 ng of ␤-galactosidase expression vector (Promega, Madison, WI) as transfection control. Luciferase activity was measured in a luminometer (Luminoskan Ascent, Labsystems, Franklin, MA), expressed in arbitrary units, and normalized according to the transcriptional efficiency from ␤-galactosidase expression. Each point is the mean of at least three independent experiments.
Gel Shift Assay-HEK293T cells were transiently transfected in 6-well plates using 2 g of plasmid. Forty-eight hours later, cells were either not treated or treated for 30 min with G-CSF (100 ng/ml). Whole cell extracts were prepared, and gel shift assays were performed as described previously (23). Briefly, 20 g of whole cell extract protein was incubated with [␥-32 P]adenosine triphosphate (ATP; PerkinElmer Life Sciences) end-labeled duplex oligonucleotide high affinity serum-inducible element probe at 37°C for 2 h in the following binding buffer: 20 mM Hepes, pH 7.4, 50 mM KCl, 1 mM 2-mercaptoethanol, 10% glycerol, 1 g of poly(dI-dC) (Amersham Biosciences), and 100 g of bovine serum albumin. Protein-DNA complexes were separated on 4.5% polyacrylamide gels equilibrated in 0.25ϫ Tris borate-EDTA buffer (TBE). Gels were dried and exposed to PhosphorImager plates, and images were developed and quantitated using a PhosphorImager (Amersham Biosciences) and ImageQuant software.
Immunoblotting-Whole cell lysates were prepared in lysis buffer (20 mM Hepes, pH 7.9, 420 mM NaCl, 20 mM NaF, 1 mM Na 3 VO 4 , 1 mM Na 4 P 2 O 7 , 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 20% glycerol, 0.5 mM phenylmethylsulfonyl fluoride). Equivalent amounts of total cellular protein were electrophoresed on 7.5% SDS-polyacrylamide gel and transferred to polyvinylidene difluoride membrane (Millipore). Probing of polyvinylidene difluoride membranes with primary antibodies and detection of horseradish peroxidase-conjugated secondary antibodies by enhanced chemiluminescence was as indicated (Amersham Biosciences). Antibodies used in this study were as follows: Stat3 monoclonal antibody obtained from Transduction Laboratories (Lexington, KY) and ␤-actin monoclonal antibody obtained from Abcam Inc.
Generation of MEF Cells Stably Expressing GFP-Stat3␣ or GFP-Stat3␤-Stat3 ⌬/⌬ MEFs were transfected with GFP-Stat3␣ or GFP-Stat3␤ constructs (3 g) using DNA carrier GeneJuice reagent as reported previously (23). After transfection overnight, the cells were fed with fresh medium and incubated for 24 h. Forty-eight hours after transfection, cells were split and cultured in selection medium containing G418 (1 mg/ml; Invitrogen) at different cell concentrations to promote growth of isolated colonies. After 10 days, 50 isolated colonies of GFP-Stat3␣ and GFP-Stat3␤ were harvested using sterile cloning discs (PGC Scientific Corp.) with trypsin-EDTA (Invitrogen); each colony was replated into a single well of a 24-well plate. After 3 weeks of growth in G418-containing medium, each colony was screened for GFP expression by fluorescence microscopy. Several colonies were expanded, analyzed, and sorted for low and uniform levels of GFP expression by the Beckman Altra cell sorter. Sorted cells with levels of GFP-Stat3 isoform expression similar to endogenous levels in Stat3 fl/fl MEF cells (Fig. 3) were selected for further analysis. MEF/GFP-Stat3␤⌬ cells were generated in a similar fashion with the exception that after selection in G418, cells were sorted four times to obtain cells with low and uniform levels of expression that on immunoblot analysis were similar to endogenous Stat3 levels in Stat3 fl/fl MEF cells (Fig. 3).
A Beckman/Q3DM IC-100 high throughput microscope (HTM) was used to automate fluorescent image acquisition and analysis of the nuclear translocation, as described (24). The system was used to scan multiple fields and to acquire and analyze each of the cells in the images. Algorithms were generated to determine fractional localized intensity in the nucleus (FLIN) for each set of cell images, as described (25). Nuclear masks were generated by applying a nonlinear least-squares optimized image filter to create marked object-background fluorescence corrected by estimating and subtracting the mean background image intensity. The correlated channel mask was computed as an intersection between the threshold-correlated channel image (the threshold level is dynamically computed using a proprietary background level estimation method) and a circle of user-defined radius from the nuclear centroid. Different gates (area, wiggle, and absolute) were set to measure the majority of cells and to avoid overexpression or lower expression level cells and very large or small cells. The same settings were used for examining FLIN in all experiments.
Fluorescence Recovery after Photobleaching-Prior to live cell imaging, MEF/GFR-Stat3␣, -␤, and -␤⌬ cells (2 ϫ 10 5 cells/ dish) were incubated in 23-mm glass bottom Delta T plates (Bioptechs Inc.) until 80 -90% confluent. After culturing for 24 h and treatment without and with IL-6 (200 ng/ml) and sIL-6R (250 ng/ml) for 30 min, cells were examined using an LSM 510 confocal microscope (Carl Zeiss) equipped with a 63ϫ (numerical aperture 1.4) objective. Cells were maintained at 37°C using a Bioptechs Delta Controller, and fresh medium without or with IL-6 and sIL-6R was cycled over the cells. Five single imaging scans were acquired prior to bleaching. A bleach pulse of 2 s was delivered using a 488-nm laser set for GFP maximum power for 20 iterations (laser output, 75%); immediately after the bleach, the intensity remaining was 40 -50% of the original fluorescence. A single z section was imaged before and at time intervals following the bleach. Fluorescence intensity of the region of interest was determined using LSM software. Data were exported to Excel for analysis as previously described (26). LSM images were exported as TIF files. All of the quantitative data for fluorescence recovery after photobleaching (FRAP) recovery kinetics represent means Ϯ S.D. from at least 18 cells imaged in two independent experiments. The immobile fraction for each cell analyzed was calculated by subtracting the mean of the last 10 single image scans from the mean of the prebleach single image scans and then dividing this value by the mean of the prebleach single image scans and multiplying by 100.
Statistical Analysis-Unless indicated otherwise, data presented are the mean Ϯ S.E.; differences between means were assessed for significance using analysis of variance or Student's t test, where appropriate.

RESULTS
The Addition of GFP to the N Terminus of Stat3␣ or Stat3␤ Does Not Change Its DNA Binding or Transcriptional Activity-To study the cellular dynamics of Stat3␣ and Stat3␤ at the single cell level, the open reading frames for each were inserted into the C terminus of GFP (Fig. 1). Transient transfection of these constructs into HEK293T cells revealed expression of proteins of the anticipated size ( Fig. 1) that bound DNA with activity similar to their untagged counterparts upon activation with two cytokine/receptor systems known to activate Stat3, G-CSF/G-CSFR (Fig. 1b) and IL-6/IL-6R (Fig. 1c). As shown previously for their untagged counterparts (27,28), GFP-Stat3␤ displayed a higher level of DNA binding without and with cytokine/receptor addition than GFP-Stat3␣. To further characterize the functional properties of GFP-Stat3␣ and GFP-Stat3␤ compared with their untagged counterparts, we examined their transcriptional activity by luciferase reporter assay in MEF/Stat3 ⌬/⌬ cells in which the endogenous Stat3 gene had been deleted using Cre-lox technology (21). MEF/Stat3 ⌬/⌬ cells were co-transfected with G-CSFR, a luciferase reporter construct containing four copies of the APRE from the rat ␣ 2 -macroglobulin promoter and Stat3␣, Stat3␤, or both expression plasmids and treated without or with G-CSF. G-CSF-induced activation of the APRE-luciferase reporter was observed in MEF/Stat3 ⌬/⌬ cells co-transfected with Stat3␣ but not in cells co-transfected with Stat3␤ (Fig. 2a), similar to results obtained by others using similar reporter constructs in COS-7 cells (8,29). Furthermore, Stat3␣-mediated activation of this reporter was inhibited by about 65% in cells co-transfected with Stat3␤ (Fig. 2a), similar to findings using reporter constructs containing the pIRE contained within the ICAM-1 promoter (8). Experiments using GFP-tagged Stat3 isoforms completely recapitulated these findings (Fig. 2b). Thus, the GFP tag did not alter the DNA binding or transcription activities of Stat3␣ and Stat3␤ in MEF cells.
Examination of Nuclear Accumulation and Cytoplasmic Reaccumulation of Stat3␣ and Stat3␤-Stat3␣ has been demonstrated to shuttle continuously between the cytoplasm and nucleus in unstimulated cells (30 -38) and to accumulate in the nucleus following cytokine stimulation due to decreased nuclear export (30); the dynamics of Stat3␤ at the single cell level and how it compares to Stat3␣ has not been examined. To address this, we transiently co-transfected HEK293T cells with plasmids containing G-CSFR and either GFP-Stat3␣ or GFP-Stat3␤ and examined 30 -50 cells by fluorescent microscopy before and after incubation with G-CSF (0, 1, 3, 10, 30, and 100 ng/ml) for 30 min ( Table 1). The percentage of cells transfected with GFP-Stat3␣ that demonstrated predominantly nuclear localization increased from 4% cells incubated without G-CSF to 90% of cells incubated with G-CSF at 100 ng/ml; similar results were obtained at each corresponding concentrations of G-CSF for cells transfected with GFP-Stat3␤. To assess the kinetics of nuclear translocation of GFP-Stat3␣ versus GFP-Stat3␤, HEK293T cells co-transfected with G-CSFR and GFP-Stat3␣ or GFP-Stat3␤ were incubated without or with G-CSF (100 ng/ml) for 0, 1, 3, 10, 30, and 100 min. The percentage of cells transfected with GFP-Stat3␣ that demonstrated predominantly nuclear localization increased from 4% of cells at time 0 to 95% of cells at both 30 and 100 min; similar results were obtained at corresponding time points examined for cells transfected with GFP-Stat3␤.
To confirm and extend these findings in cells that express only GFP-tagged Stat3␣ or Stat3␤ at levels similar to endogenous levels, MEF/Stat3 ⌬/⌬ cells were stably transfected with the GFP-Stat3␣ or GFP-Stat3␤ plasmid constructs, and surviving colonies were sorted to obtain Ͼ90% GFP-positive cells that express GFP-Stat3 at low to moderate levels. Immunoblot analysis of representative colonies established that levels of GFP-  To examine the kinetics of cytoplasmic reaccumulation of Stat3 isoforms, cells were removed from IL-6/sIL-6R and examined by HTM at 30 -60-min intervals (Fig. 3). FLIN decreased rapidly in MEF/GFP-Stat3␣ cells following cytokine/receptor removal, returning to unstimulated levels within 30 min and demonstrated a time to half-maximal cytoplasmic reaccumulation (t1 ⁄2 ) of 15 min. In contrast, FLIN of MEF/GFP-Stat3␤ cells decreased only slightly after cytokine/receptor removal and demonstrated a cytoplasmic reaccumulation t1 ⁄2 of Ͼ180 min.

Examination of the Nuclear Mobility of Stat3␣ and Stat3␤-
We have previously demonstrated that aberrant nuclear sequestration of chimeric RAR␣-containing transcription factors found in acute promyelocytic leukemia cells is associated with their reduced nuclear mobility (26). To determine if reduced mobility of Stat3␤ may contribute to its nuclear retention, we performed FRAP using MEF/ GFP-Stat3␣ and MEF/GFP-Stat3␤ cells ( Fig. 4 and Table 2). Only cells expressing low levels of protein were examined by FRAP to avoid artifacts of overexpression as previously described (31).
Release from a large aggregate or complex could explain the ligandstimulated increase in mobility of Stat3␣ and would be accompanied by an decrease in the immobile fraction (32). Conversely, formation of an aggregate or binding to a large complex could explain the ligandmediated decrease in mobility of Stat3␤ and would be accompanied by an increased immobile fraction. Consistent with these predictions (Fig. 4 and Table 2), ligand/receptor stimulation of MEF/GFP-Stat3␣ cells resulted in a 13% decrease in the immobile fraction of Stat3␣. In contrast, ligand stimulation of MEF/GFP-Stat3␤ cells resulted in a 32% increase in the immobile fraction of Stat3␤. Thus, in addition to prolonged nuclear retention, Stat3␤ exhibited reduced nuclear mobility in both unstimulated and stimulated cells. Furthermore, in contrast to Stat3␣, the mobility of Stat3␤ decreased and the immobile fraction increased with activation. These results are consistent with the hypothesis that reduced intranuclear mobility, accom-  NOVEMBER 30, 2007 • VOLUME 282 • NUMBER 48 panied by increased immobile fraction, may contribute to prolonged nuclear retention of Stat3␤.

Nuclear Retention of Stat3b
The C-terminal 7-Amino Acid Domain of Stat3␤ Is Responsible for Prolonged Nuclear Retention but Not Reduced Nuclear Mobility or Increased Immobile Fraction-To determine the contribution of the CT7 domain of Stat3␤ to prolonged nuclear retention, reduced nuclear mobility, and increased immobile fraction, we used MEF/Stat3 ⌬/⌬ cells to generate cells that stably express GFP-labeled Stat3␤ in which the CT7 domain was deleted (MEF/GFP-Stat3␤⌬) (Fig. 5). Gel shift assays demonstrated that GFP-Stat3␤⌬ bound DNA with affinity similar to GFP-Stat3␤ (Fig. 1c, lanes 7-10), consistent with previous reports examining this Stat3 construct untagged with GFP (33). HTM analysis revealed that FLIN values of MEF/GFP-Stat3␤⌬ cells before and after stimulation with IL-6/sIL-6R␣ were similar to MEF/GFP-Stat3␤ cells (Fig. 5). However, FLIN in MEF/ GFP-Stat3␤⌬ cells returned to unstimulated levels within 30 min following cytokine/receptor removal with a cytoplasmic reaccumulation t1 ⁄ 2 of 15 min, similar to GFP-Stat3␣. These results indicate that the CT7 domain is responsible for prolonged nuclear retention of Stat3␤. In contrast, FRAP results showed that GFP-Stat3␤⌬ retained mobility features of Stat3␤, including reduced mobility in unstimulated cells (FRAP t1 ⁄ 2 ϭ 1.42 Ϯ 0.40 s) and the property of further reduced mobility in stimulated cells (FRAP t1 ⁄ 2 ϭ 2.89 Ϯ 0.67 s; p Ͻ 0.05). In addition, similar to Stat3␤, the immobile fraction of Stat3␤⌬ also increased 43% with ligand stimulation (Fig. 5 and Table 2). These results indicate that although the CT7 domain is responsible for prolonged nuclear retention of Stat3␤, it is not responsible for its reduced intranuclear mobility or increased immobile fraction. Furthermore, these findings suggest that reduced intranuclear mobility and increased immobile fraction do not contribute to prolonged nuclear retention; rather, the mechanism of prolonged nuclear retention is mediated through a distinct mechanism involving the CT7 domain.

DISCUSSION
Although much has been learned regarding the molecular and cellular biology of Stat3 in the past 15 years, more remains to be discovered regarding its intracellular trafficking and intranuclear dynamics, especially whether or not these features differ between isoforms and the domains within each responsible for any differences observed. Several studies examining intracellular trafficking of Stat3␣ have been reported (30 -38, 42). Fluorescent microscopic studies of cells stained with antibody or expressing Stat3␣ tagged with GFP or its variants have demonstrated localization of Stat3␣ within the cytoplasm as well as the nucleus in a diffuse pattern within resting cells (34 -39), similar to our findings with GFP-Stat3␣. Basal co-localization has been attributed to constitutive shuttling of Stat3␣ between the cytoplasm and nucleus. Basal shuttling of Stat3␣ into the nucleus has been attributed to interaction between residues 150 and 162 within the coiled-coil domain with importin-␣3 (40). Basal shuttling of Stat3␣ out of the nucleus mapped to two nuclear export sequences, one contained within residues 404 -414 within the DNA-binding domain and the other within residues 524 -535 of the linker domain; basal shuttling of Stat3␣ out of the nucleus is blocked by the CRM1 (chromosome region maintenance 1) inhibitor leptomycin B (36). Early cell fractionation studies (reviewed in Ref. 1) and most studies using fluorescent microscopic imaging (30, 34,  Nuclear Retention of Stat3b strated accumulation of Stat3␣ in the nucleus with ligand stimulation. Ligand-mediated nuclear accumulation has been attributed to reduced nuclear export (30) and to interaction of Stat3␣ arginine residues 214 and 215 in the coiled-coil domain with importin-␣5 and, to a lesser extent, importin-␣7 (30,40,42). Most groups, including ours, have reported that Stat3␣ is distributed diffusely within the nucleus, excluding the nucleolus, in ligand-stimulated cells, although Herrmann et al. (37) described Stat3␣ within nuclear bodies (0.2-1 m in size; 10 -70 nuclear bodies/nucleus) distinct from splicing factor compartments and promyelocytic leukemia nuclear bodies. Reaccumulation of Stat3␣ in the cytoplasm after ligand stimulation was blocked by leptomycin B and was attributed to a nuclear export signal sequence located within the C-terminal portion of the coiled-coil domain at residues 306 -318 (36). A more recent report suggested that Arg 214 and Arg 215 of Stat3␣ inhibited the interaction between Stat3␣ and CRM1 (38), since mutation of these residues to alanine increased binding of CRM1 and nuclear export. One group has previously reported the results of FRAP studies of Stat3␣ (37) within the nucleoplasm of resting HepG2 and COS-7 cells transiently transfected with Stat3-YFP and within nuclear bodies of ligand-stimulated cells. Whereas they did not report times to half-maximal recovery of fluorescence, they observed a smaller immobile fraction (ϳ23%) in the nucleoplasm of resting cells compared with the immobile fraction with nuclear bodies of ligand-stimulated cells (ϳ49%); FRAP of Stat3␣ within the nucleoplasm of stimulated cells was not reported. We did not observe nuclear bodies in our studies in either stably transfected MEF cells or transiently transfected HEK293T cells perhaps because we were especially careful to avoid cells with excess overexpression, which we previously demonstrated can lead to artifacts in localization and mobility (26,31).
Ours is the first report to examine intracellular trafficking and nuclear dynamics of Stat3␤ and to compare these findings with Stat3␣. Similar to GFP-Stat3␣, GFP-Stat3␤ was localized to both the cytoplasm and nucleus with a diffuse distribution in resting HEK293T and MEF cells. GFP-Stat3␤ also exhibited FLIN similar to GFP-Stat3␣. Cytokine stimulation revealed kinetics and cytokine concentration dependence of nuclear accumulation for GFP-Stat3␤ similar to GFP-Stat3␣; the distribution of GFP-Stat3␤ also was similar to GFP-Stat3␣ being distributed in a diffuse pattern within the nucleus and excluded from the nucleolus. However, although cytoplasmic reaccumulation was rapid for GFP-Stat3␣ (t1 ⁄ 2 ϭ 15 min), it was markedly delayed for GFP-Stat3␤ (t1 ⁄ 2 Ͼ 3 h). In addition, the intranuclear mobility of the GFP-Stat3␤ in unstimulated cells was slower than GFP-Stat3␣ without stimulation. Also, in contrast to Stat3␣, Stat3␤ was slowed further following ligand simulation. Ligand-mediated changes in the immobile fraction were consistent with ligand-mediated changes in intranuclear mobility. Deletion of the unique CT7 domain eliminated prolonged nuclear retention of Stat3␤; however, it did not change its intranuclear mobility or percentage immobile fraction.
Studies to determine the mechanism through which the CT7 domain mediates prolonged nuclear retention of Stat3␤ are under way. Thus far, BLAST analysis (43) did not reveal the presence of a similar amino acid sequence in any nuclear protein in the protein data base, suggesting that the mechanism of CT7-mediated Stat3␤ nuclear retention may be unique. Also, the addition of the CT7 domain to the C terminus of GFP-Stat3␣ to generate GFP-Stat3␣␤ did not result in its prolonged nuclear retention following transient transfection of HEK293T cells (data not shown), indicating that nuclear retention by the CT7 domain is context-dependent.
Although nuclear export of Stat1 appears to be linked to dephosphorylation (44), results with Stat3 are less clear (36,38). To examine whether or not differences in Stat3 dephosphorylation correlated to differences in nuclear export, we transiently transfected HEK293T cells with each GFP-tagged Stat3 construct and examined cells for cytoplasmic-to-nuclear and nuclear-to-cytoplasmic translocation. Each construct behaved in a manner similar to stably transfected MEF cells (supplemental Fig. 1). Each GFP-tagged Stat3 construct translocated to the nucleus within 30 min following ligand stimulation; although Stat3␣ and Stat3␤⌬ recycled to the cytoplasm within 120 min, Stat3␤ did not. Immunoblotting of whole cell extracts from each revealed no differences in Tyr 705 phosphorylation status, indicating that the differences in nuclear retention observed did not correlate to Tyr 705 phosphorylation status.
When phosphorylated on Tyr 705 to equivalent degrees, Stat3␤ binds DNA with 20 -50-fold more avidity than Stat3␣ in vitro (27) and with 10 -20 times more avidity in vivo (28). Previous FRAP studies of Stat1 have demonstrated that a Stat1 mutant with enhanced DNA binding also demonstrated reduced mobility compared with wild-type Stat1 (45). These results together with our studies demonstrating that GFP-Stat3␤⌬ binds DNA with avidity similar to GFP-Stat3␤ and retains the mobility features of GFP-Stat3␤ suggest that the distinct nuclear mobility features of Stat3␤ are the result of its increase DNA binding avidity. Attempts to generate MEF cells stably expressing GFP-tagged Stat3␤ mutated to decrease DNA binding (GFP-Stat3␤Dm1 and GFP-Stat3␤Dm2) were unsuccessful, perhaps due to protein misfolding and degradation (see Fig. 1c) resulting in toxicity. However, transient transfection into HeLa cells revealed impaired ligand-stimulated nuclear accumulation of GFP-Stat3␤Dm2 (Fig. 6). Ligand-induced DNA binding has previously been demonstrated to control nuclear accumulation of Stat1 along with its nuclear mobility (45).
Stat3␣ and Stat3␤ demonstrate both distinct and overlapping functions in gene transcription (6, 46 -48), normal biology, and pathophysiology. The finding of prolonged nuclear retention may help explain some of these observations. Examination of mice deficient in Stat3␤ or expressing only Stat3␤ (49,50) revealed that Stat3␤ protected against necrosis of renal tubules, thymus, and liver in the late (Ͼ96 h) phase of endotoxemia, down-modulated a subset of endotoxin-inducible genes in the liver, and protected mice from embryonic lethality observed at 6.5-7 days in mice deficient in both Stat3 isoforms (51). Stat3 activation contributes to oncogenesis in many tumor cell systems (52,53) and has attracted considerable attention as a target for cancer therapy (54 -56). Stat3␣ is the oncogenic isoform, since a constitutively active form of Stat3␣ alone was capable of inducing NIH-3T3 to form foci that formed tumors in nude mice (57), whereas Stat3␤ antagonized the oncogenic activity of Stat3␣ (58, 59) downstream of v-Src. It is tempting to speculate that the late protection in endotoxemia, the protection from embryonic lethality and/or the anti-oncogenic effect of Stat3␤ may be attributable to CT7-mediated prolonged nuclear retention of Stat3␤ alone or heterodimerized with other STAT protein, such as Stat5A, Stat5B, or Stat1. Last, Stat3 is the most pleiotropic member of the STAT protein family. Our finding that only Stat3␤ among ␤ isoforms of STAT family members contains a functional domain distinct from its ␣ counterpart may help explain this.