AMP-activated Protein Kinase-regulated Phosphorylation and Acetylation of Importin (cid:1) 1 INVOLVEMENT IN THE NUCLEAR IMPORT OF RNA-BINDING PROTEIN HuR*

Nuclear import of HuR, a shuttling RNA-binding protein, is associated with reduced stability of its target mRNAs. Increased function of the AMP-activated protein kinase (AMPK), an enzyme involved in responding to metabolic stress, was recently shown to reduce the cytoplasmic levels of HuR. Here, we provide evidence that importin (cid:1) 1, an adaptor protein involved in nuclear import, contributes to the nuclear import of HuR through two AMPK-modulated mechanisms. First, AMPK triggered the acetylation of importin (cid:1) 1 on Lys 22 , a process dependent on the acetylase activity of p300. Second, AMPK phosphorylated importin (cid:1) 1 on Ser 105 . Accord-ingly, expression of importin (cid:1) 1 proteins bearing K22R or S105A mutations failed to mediate the nuclear import of HuR in intact cells. Our results

but the export factor involved in HNS-dependent transport is currently unknown (35). Although a role for HuR in the export of target mRNAs awaits to be definitively demonstrated, evidence accumulated thus far strongly supports such a function. Therefore, the existence of multiple export pathways for HuR would ensure the rapid and effective export of HuR target mRNAs. Recently, transportin 1 (Trn1; also known as Karyopherin/Kap␤2) as well as the highly similar transportin 2 (Trn2) were shown to participate in the nuclear import of HuR (36). In a series of elegant in vitro studies, the HNS was found to mediate the import of HuR by transportins (35,36).
In vivo, the levels of cytoplasmic HuR have been shown to be potently regulated by the activity of the AMP-activated protein kinase (AMPK), an enzyme that participates in the cellular response to metabolic stress (37). Believed to function as a "low fuel warning system" of the cell, AMPK activity is strongly elevated by conditions that elevate the cellular AMP:ATP ratio, such as exposure to metabolic poisons like arsenite and azide, depletion of growth factors or glucose, and treatment with the pharmacological agent AICAR (which mimics the effect of AMP) (38). In turn, active AMPK phosphorylates a number of metabolic enzymes causing both a global inhibition of biosynthetic pathways, thus conserving energy, and a global activation of catabolic pathways, thus generating more ATP (reviewed in Ref. 39). AMPK activity is ubiquitous, although different isoforms of its catalytic (␣) and regulatory (␤ and ␥) subunits exhibit tissue-specific distribution, as well as preferential localization in different subcellular compartments (e.g. ␣2 subunits are partly nuclear, whereas ␣1 subunits are only found in the cytoplasm, etc.). Recently, we found that decreased AMPK activity led to an elevation in cytoplasmic HuR levels; conversely, AMPK activation through interventions such as treatment with AICAR and ectopic expression of a constitutively active isoform of AMPK caused a reduction in cytoplasmic HuR (40). The lowering of cytoplasmic HuR was accompanied by a reduction in complexes of HuR bound to target mRNAs encoding proliferative proteins (cyclins A and B1), a decrease in the stability of such mRNAs, a lessening in the expression of the corresponding protein products, and an ensuing inhibition of cell growth (40).
Given that HuR shuttles between the nucleus and cytoplasm (26,27,29,30) and given that AMPK-triggered reductions in cytoplasmic HuR levels profoundly influence the stability of target mRNAs (40), we set out to investigate the mechanisms underlying the AMPK-mediated nuclear import of HuR. In this investigation, we have identified importin ␣1 as a key cytoplasmic HuR ligand mediating this process. Importin ␣1 functions as an adaptor that associates with importin ␤, which transports bound cargoes through the nuclear pore complex (NPC). We provide additional evidence in support of a role for AMPK in dually modifying importin ␣1: AMPK triggered the acetylation of importin ␣1 through AMPK-stimulated activation of acetylase p300, and AMPK directly phosphorylated importin ␣1, both in vivo and in vitro. Importantly, overexpression of wild-type full-length importin ␣1, but not point mutants of importin ␣1 lacking the phosphorylation and acetylation sites, readily promoted the nuclear import of HuR in intact cells. Together, our data indicate that AMPK induces the acetylation and phosphorylation of importin ␣1 and strongly suggest that dual modification of importin ␣1 is required for the nuclear import of HuR.

EXPERIMENTAL PROCEDURES
Cell Culture, Treatment, Transfection, and Infection-Human colorectal carcinoma RKO cells were cultured in minimum essential medium (41), and human embryo kidney fibroblasts 293 cells (ATCC) were cultured in Dulbecco's modified essential medium, each supplemented with 10% fetal bovine serum (Hyclone, Logan, UT) and antibiotics. Sodium butyrate, trichostatin A, and AICAR were from Sigma, and calpain inhibitor I, N-Ac-Leu-Leu-norleucinal (ALLN) was from Calbiochem (La Jolla, CA). Adenoviruses expressing either the control gene GFP (AdGFP) or a constitutively active isoform of the AMPK ␣1 subunit (Ad(CA)AMPK) (42) were amplified and titered in 293 cells using standard methodologies. The infections were carried out in serum-free Dulbecco's modified Eagle's medium for 4 h. Infection efficiency of RKO cells was determined by infection with AdGFP at various plaque-forming units/cell and assessment of the percentage of GFP-expressing cells 48 h later. For Ͼ90% infection, 100 plaque-forming units/cell was required, in accordance with the low infection rates of RKO cells (41); 100 and 20 plaque-forming units/cell were used in all infections of RKO and 293 cells, respectively. Cytoplasmic, nuclear, and whole cell fractions were prepared as described (24).
Constructs and Recombinant Proteins-All of the purified recombinant proteins were produced in Escherichia coli. His-tagged (C-terminal) importin ␣1 was purified by nickel-nitrilotriacetic acid affinity chromatography followed by ion exchange and gel filtration chromatography. GST-Imp␤ and GST-Kap␤2 were purified by glutathione-Sepharose affinity chromatography, cleaved with either Precission protease (Imp␤) or Tev protease (Kap␤2), and Imp␤ and Kap␤2 were further purified by ion exchange and gel filtration chromatography (43).
Pulse Labeling of Endogenous Protein-RKO cells (200,000 cells/ 6-cm dish) were incubated in methionine-and cysteine-free medium containing 5% dialyzed fetal bovine serum for 30 min and then for an additional 20 min with 900 Ci of L-[ 35 S]methionine and L-[ 35 S]cysteine (NEG-072 Expre 35 S 35 S Protein labeling mix; PerkinElmer Life Sciences) per well and scraped into 500 l of ice-cold phosphate-buffered saline. After centrifugation (1,800 rpm, 4°C, 1 min), the pellets were shock-frozen in liquid nitrogen and resuspended in 100 l of TSD lysis buffer (50 mM Tris, pH 7.5, 1% SDS, and 5 mM DTT). One hundred-g protein aliquots were brought to a final volume of 1.2 ml in TNN buffer (50 mM Tris, pH 7.5, 250 mM NaCl, 5 mM EDTA, 0.5% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, 2 ng/l aprotinin, and 2 ng/l leupeptin) and precleared with protein A/G (1:1)-Sepharose mix (Sigma) for 1 h at 4°C. Following a brief centrifugation, the supernatants were incubated for 1 h at 4°C with 0.4 g of either a monoclonal antibody that recognizes HuR (Santa Cruz Biotechnology) or IgG1 (BD Biosciences Pharmingen) and precipitated using 50 l of protein A/G (1:1) mix (Sigma) for 1 h at 4°C. Following washes in TNN buffer, the IP material was resolved by electrophoresis in SDS-containing 12% polyacrylamide gels, transferred onto polyvinylidene difluoride filters, and visualized using a PhosphorImager (Molecular Dynamics).
AMPK and HDAC Activity Assays-AMPK was assayed as described (38,40). Briefly, AMPK was immunoprecipitated from 5 g of cell lysate by incubation with 1 g of anti-␣1 and 1 g of anti-␣2 polyclonal antibodies, and AMPK activity in the IP complexes was determined by phosphorylation of peptide HMRSAMSGLHLVKRR (SAMS (38)). Synthetic peptides encompassing Ser 77 ( 71 QGTVNWSVDDIVKGI 85 ), Ser 105 ( 98 QAARKLLSREKQ PPID 113 ), and Ser 179 ( 172 ASPHAHISEQAVWAL-G 187 ) were obtained from AnaSpec, Inc. (San Jose, CA) and used in phosphorylation assays employing AMPK as kinase, following the procedures described above for the SAMS peptide. Phosphorylated substrates were measured by scintillation counting. HDAC activity was measured with a histone deacetylase assay kit from Upstate Biotechnology Inc., following the manufacturer's instructions; histone H4 was used as a substrate.
Acetylation of Importin ␣1 by p300 -Following infection of RKO cells with either AdGFP or Ad(CA)AMPK, p300 was immunoprecipitated from 100 g of nuclear lysate (prepared as previously described (24)) using 3 g of an anti-p300 monoclonal antibody (Upstate Biotechnology, Inc.). p300 bound to G protein-coupled Sepharose beads was incubated with a reaction mixture containing 50 mM Tris-HCl, pH 8.0, 10% glycerol, 0.1 mM EDTA, 1 mM DTT, 0.5 g of purified His-Imp␣1 protein, 20 M acetyl-CoA, and 1 Ci of [ 3 H]acetyl-CoA for 30 min at 30°C in a shaking incubator. The reaction mixtures were then size-separated by 12% SDS-PAGE, and radiolabeled acetylated proteins were visualized using a phosphorimaging device (Molecular Dynamics).
Protein Phosphorylation Assays-Active AMPK was immunoprecipitated from 200 g of lysate prepared from 293 cells that had been infected with Ad(CA)AMPK, using an anti-c-Myc antibody and G protein-coupled Sepharose beads. For in vitro phosphorylation assays, the proteins that were either immunoprecipitated from RKO cells or purified from bacteria were incubated with AMPK (bound to G proteincoupled Sepharose beads) in the presence of 10 Ci of [␥-32 P]dATP and kinase buffer containing 50 mM Hepes, 1 mM DTT, 0.02% Brij-35, 6.25 mM MgCl 2 , and 0.25 mM AMP for 10 min at 30°C. Phosphorylated products were size-fractionated by 15% SDS-PAGE and transferred onto polyvinylidene difluoride membranes. Phosphorylated products were visualized using a PhosphorImager (Molecular Dynamics).
For in vivo phosphorylation assays, cultured cells were incubated with 0.5 mCi/ml H 3 32 PO 4 ( 32 P i ), for 3 h under standard conditions (37°C, 5% CO 2 ), and phosphorylation of either p300 or importin ␣1 was detected by IP using antibodies from Upstate Biotechnology Inc. or BD Biosciences, respectively.
Solution Binding Assays-Solution binding assays were performed essentially as described (46). Two g of either MBP-HuR or His-importin ␣1 immobilized on 10 l of protein G-Sepharose beads using either anti-MBP or anti-His antibodies, respectively, were incubated with 2 g of either Imp␤, Kap␤2, MBP (New England Biolabs, Beverly, MA), MBP-HuR, Imp␣1, or a combination of these proteins for 45 min at 25°C. The proteins bound to products immobilized on beads, as well as proteins remaining unbound, were visualized after 12% SDS-PAGE and staining with Coomassie Blue dye.
Immunofluorescence-293 cells were seeded on coverslips and were transfected with the plasmids indicated. Forty-eight h after transfection, the cells were fixed for 15 min in phosphate-buffered saline containing 4% paraformaldehyde and permeabilized for 15 min in phosphate-buffered saline containing 0.4% Triton X-100. After incubation for 16 h in blocking buffer (phosphate-buffered saline containing 2% bovine serum albumin and 0.1% Tween 20), the coverslips were incubated for 1 h at 1:500 dilution of mouse anti-HuR (Santa Cruz Biotechnology.) prepared in blocking buffer. Following washes with PBS containing 0.1% Tween 20, the samples were incubated for 1 h with a mixture of horse anti-mouse Texas Red (1:200; Jackson Laboratories) and Hoechst 33342 (1:5,000; Molecular Probes). After washes with PBS containing 0.1% Tween 20, the coverslips were mounted in Vectashield (Vector Laboratories, Burlingame, CA) and visualized with an Axiovert 200 M microscope (Zeiss; 63ϫ lens) using separate channels for the analysis of red fluorescence and blue fluorescence. The images were processed with the AxioVision 3.0 program (Zeiss). Representative photographs from three independent experiments are shown.

Reduced Cytoplasmic HuR Levels after AMPK Activation-
Treatment of human colorectal carcinoma RKO cells with the AMP analog AICAR, a strong and highly specific activator of AMPK (38) (Fig. 1A), markedly reduced the levels of cytoplasmic HuR (Fig. 1B). Evidence that the AMPK-imposed reduction in cytoplasmic HuR abundance was not due to increased cytoplasmic degradation of HuR was obtained by using inhibitors of the proteasome and proteases. As shown in Fig. 1C, pretreatment with ALLN (which inhibits the activity of the proteasome, calpain, and cathepsins) failed to prevent the reduction in cytoplasmic HuR levels elicited through treatment with AICAR (Fig. 1C); use of proteasome inhibitor lactacystin or protease inhibitors ALLM or PD150606 similarly failed to alter the reduction in cytoplasmic HuR levels triggered by AICAR (not shown). In addition, the AMPK-imposed reduction in cytoplasmic HuR abundance was not due to decreased translation, as observed by assessing nascent HuR levels. De novo synthesis of HuR was monitored by pulse labeling (for 20 min) of RKO cells in the presence of L-[ 35 S]methionine and L-[ 35 S]cysteine, followed by detection of the newly synthesized HuR through IP using anti-HuR antibodies (Fig. 1D). These two sets of data lent support to the notion that HuR abundance in the cytoplasm was primarily governed via nucleocytoplasmic transport and not changes in HuR translation or cytoplasmic stability. In keeping with earlier studies demonstrating that increased AMPK activity reduces cytoplasmic HuR levels (40) and studies that HuR shuttles between the nucleus and the cytoplasm (26,27,29,30), these findings supported the existence of an AMPKtriggered effect on nuclear transport of HuR. Given that AMPK-imposed reductions in cytoplasmic HuR levels profoundly impacted on the half-lives of HuR target mRNAs (40), we sought to examine the regulatory mechanisms whereby AMPK regulated the nuclear import of HuR.
Reduction in Cytoplasmic HuR Levels by AMPK Involves Changes in Protein Acetylation-Recently, the acetylase p300 (CBP) was identified as a downstream target of AMPK (47). To study whether changes in protein acetylation were linked to the nuclear import of HuR, RKO cells were treated with two potent inhibitors of HDAC activity, sodium butyrate (SButyr) and trichostatin A (TSA), and cytoplasmic HuR levels were subsequently monitored by Western blotting. As shown in Fig.  2A, cytoplasmic HuR abundance was reduced to ϳ40 and 20% of the levels seen in untreated RKO populations, respectively, although whole cell HuR levels were essentially unchanged (Total). As anticipated, both SButyr and TSA strongly lowered histone deacetylase activity (Fig. 2B, top panel), but neither drug was capable of modulating AMPK activity, either alone or in combination (Fig. 2B, bottom panel), revealing that a deacetylase activity was not upstream of AMPK regulatory pathways.
To find out whether regulated protein acetylation was downstream of AMPK in the AICAR-triggered pathway or whether the two pathways (protein acetylation, AMPK activation) were instead contributing independently to the nuclear import of HuR, experiments were carried out using AICAR and HDAC inhibitors simultaneously. To find a greater reduction in cytoplasmic HuR levels with the combined treatment (AICAR together with HDAC inhibitors) than with either of the single treatments would indicate the existence of two distinct, parallel pathways of reducing cytoplasmic HuR, one mediated by AMPK, the other mediated by acetylases. Instead, as shown in Fig. 2C, treatments with either AICAR alone or SButyr plus TSA were each capable of reducing cytoplasmic HuR levels, but combined treatment with AICAR, SButyr, and TSA did not cause a further reduction in cytoplasmic HuR levels beyond AICAR. Assessment of the levels of ␤-actin and ␤-tubulin served to monitor equal loading and transfer of cytoplasmic and total protein samples, respectively. D, RKO cells were treated for 6 h in the presence of the indicated concentrations of AICAR, and then incubated for an additional 20 min in the presence of L-[ 35 S]methionine and L-[ 35 S]cysteine, whereupon nascent HuR was visualized by IP as described under "Experimental Procedures"; the IgG1 lane represents the control IP material. The samples were resolved by electrophoresis in SDS-containing 12% polyacrylamide gels. Radiolabeled HuR signal is shown.

FIG. 1. AMPK activity and subcellular localization of HuR in RKO cells treated with AICAR.
A, 6 h after exposure of RKO cells to the indicated doses of AICAR, whole cell lysates were prepared, and AMPK activity was assayed after IP with a polyclonal antibody recognizing the AMPK ␣1 and ␣2 subunits, using the synthetic peptide SAMS as substrate (described under "Experimental Procedures" and in Ref. 38). The data represent the means and S.E. of three separate experiments. B, Western blot analysis to assess HuR levels in whole cell (Total, 10 g), cytoplasmic (40 g), and nuclear (10 g) lysates prepared from RKO cells that were treated for 6 h with the indicated concentrations of AICAR. Assessment of the levels of the cytoplasm-specific ␤-tubulin and nucleus-specific HDAC1 after sequential stripping and reprobing of the filters served to verify the quality and equal loading of the cytoplasmic and nuclear preparations, respectively. All Western blotting signals were measured by densitometry scanning; HuR abundance in each compartment following AICAR treatment was calculated after normalization against the values of control proteins (either ␤-tubulin or HDAC1) and is presented as a percentage of remaining signals relative to those measured in untreated samples of each lysate group. C, Western blot analysis to monitor HuR levels in either cytoplasmic or whole cell (Total) lysates prepared from RKO cells that were either pretreated for 1 h with 10 M of the protease/proteasome inhibitor calpain inhibitor I, N-Ac-Leu-Leu-norleucinal (ALLN) or left untreated and then incubated for 6 h in the presence of the indicated doses of

FIG. 2. Effect of histone deacetylase inhibitors on the cytoplasmic localization of HuR and AMPK activity.
A, RKO cells were treated with histone deacetylase inhibitors SButyr (15 mM) or TSA (1 M), and cytoplasmic (40 g) and whole cell (Total, 10 g) lysates were used to monitor HuR expression levels. Relative HuR abundance was calculated as explained in the legend of Fig. 1; ␤-actin signals served to normalize for differences in loading and transfer of protein samples. B, 6 h after treatment of RKO cells with 2 mM AICAR, 15 mM SButyr, 1 M TSA, or combinations of these drugs, the lysates were prepared, and HDAC activity ("Experimental Procedures") and AMPK activity (described in the legend of Fig. 1) were assayed. The data represent the mean of two experiments, each yielding similar results. C, RKO cells were treated for 6 h with 2 mM AICAR, 15 mM Sbutyr, or 1 M TSA, alone or in combination, and HuR abundance was assessed in cytoplasmic (40 g) and whole cell (10 g) lysates. Relative HuR expression levels were calculated as explained in the legend of Fig. 1; ␤-actin signals served to normalize for differences in the loading and transfer of protein samples. Ctrl, control. those attained by AICAR alone. Together, these observations support the notion that protein acetylation is implicated in the cellular localization of HuR and further suggest that protein acetylation is a downstream event of the AMPK-controlled pathway of HuR localization.
HuR Associates Specifically with Importin ␣1-The above findings prompted a further analysis of components of the nuclear import machinery potentially involved in transporting HuR into the nucleus following AMPK activation. To identify the specific factor(s) that might mediate the import of HuR into the nucleus, we carried out a series of IP reactions coupled with Western blot analyses to investigate potential associations between HuR and other proteins. This survey included a number of transport proteins such as CRM1, NTF2, and members of the karyopherin ␤, importin ␤, and importin ␣ families. It also included several negative control proteins, including proteins involved in the structure and function of the nuclear pore complex (such as Ran, Ran-binding protein, and nucleoporins) and nuclear proteins such as HDACs, given the aforementioned evidence that cytoplasmic HuR levels were influenced by altered protein acetylation.
The potential interactions between these proteins and HuR were first assayed by IP using an anti-HuR antibody, followed by Western blotting using antibodies that specifically recognized the proteins of interest (Fig. 3A, left column, Panel). These interactions were further tested by carrying out individual IP reactions with antibodies recognizing each of the proteins listed and then performing Western blot analysis of the IP material to test for HuR presence (Fig. 3A). These studies revealed that importin ␤, karyopherin ␤2, and importin ␣1 were associated with HuR, although importin ␤ appeared to bind less strongly with HuR. Fig. 3B depicts illustrative IPcoupled Western blot analyses performed to obtain the data summarized in Fig. 3A.
Further evidence that importin ␣1 associated with HuR was obtained by using recombinant proteins in solution binding assays that were visualized by Coomassie Blue staining. Incubations were carried out initially using recombinant Histagged importin ␣1 (His-Imp␣1) immobilized on beads through an antibody recognizing the histidine tag. As shown, both importin ␤ (Imp␤) and HuR (expressed and purified as a maltosebinding protein-HuR fusion protein, MBP-HuR), but not MBP alone, were shown to interact with His-Imp␣1 (Fig. 4A). Conversely, testing of MBP-HuR immobilized on beads through an anti-MBP antibody revealed that recombinant His-Imp␣1 associated with MBP-HuR (Fig. 4B); Imp␤ and Kap␤2 were also found to associate with MBP-HuR, revealing that these proteins may also be capable of binding MBP-HuR in vitro, in keeping with previous findings (35,36). Control incubations using an N-terminal fragment of c-Jun linked to GST showed no binding to MBP-HuR. In each case, control incubations using beads and antibody revealed no unspecific binding (not shown). Importin ␣1 has been described as an adaptor protein that associates with transport receptor protein importin ␤, thereby delivering its cargo into the nucleus (48,49). Thus, the following set of experiments sought to examine whether AMPK might regulate importin ␣1 function and thereby influence HuR import.
Importin ␣1 Is Acetylated by p300 in an AMPK-dependent Manner in RKO Cells-First, we investigated the possibility that AMPK might regulate importin ␣1 acetylation, an event that has been shown to promote its interaction with importin ␤ (45). In this regard, two recent studies provided important leads: nuclear acetylase p300 is a target of phosphorylation by AMPK (47), and importin ␣1 is a substrate of acetylation by p300 (45). In agreement with these reports, we found that 293 cells transfected with a p300-expressing construct (pCMV␤-p300) displayed strongly elevated levels of acetylated importin ␣1, as well as increased acetylation of histone H4, a known substrate of p300 (50, 51) (Fig. 5A).
Based on these pieces of evidence, we set out to examine whether acetylation of importin ␣1 were dependent on the AMPK activity levels of the cell. To this end, we first prepared cells exhibiting elevated AMPK activity through infection with an adenovirus that expresses a constitutively active isoform of the ␣ subunit (Ad(CA)AMPK), whereas control populations were prepared by infection using a GFP-expressing adenovirus (AdGFP) (42). Forty-eight h after infection of RKO cells, AMPK activity was 2.8-fold higher in Ad(CA)AMPK-infected cells than FIG. 3. Interaction of HuR with nuclear proteins and with proteins involved in nucleocytoplasmic transport. A, whole cell lysates prepared from RKO cells were subjected to two types of analyses. First (IP: HuR; IB: Panel column), HuR was immunoprecipitated along with associated proteins, whereupon the presence of nuclear proteins and proteins associated with the nucleocytoplasmic transport system (left column) was detected by performing Western blot (WB) analysis with the corresponding specific antibodies. Second (IP: Panel; IB: HuR column), IP reactions were performed using a collection of antibodies recognizing each of the proteins listed in the left column, followed by Western blot analysis to detect HuR in the immunoprecipitated material. The details are provided under "Experimental Procedures." Western blot signals: Ϫ, undetectable; ϩ/Ϫ, weak; ϩ, moderate; ϩϩ, prominent; ϩϩϩϩ, very strong; ND, not done. B, Western blot analysis to illustrate IP-coupled Western blot analyses used to generate the list in A. Two hundred g of whole cell lysate was used per each immunoprecipitation (IP) reaction, and the immunoprecipitated material was then used to perform Western blotting analysis to detect the presence of the proteins indicated. IP followed by WB analyses were carried out two to five times; representative Western blotting signals are shown. IB, immunoblot.
in AdGFP-infected control populations (Fig. 5B), in keeping with earlier studies using these adenoviral vectors (40). p300 activity in each infection group was then tested by performing IP of p300 and then using the pelleted material to test acetylase activity (i.e. its ability to add a [ 14 C]acetyl group onto a substrate). As shown, the p300 immunoprecipitate was not capable of acetylating substrate MBP-HuR (or substrate Kap␤2), 2 but it readily acetylated importin ␣1 (His-Imp␣1), in agreement with earlier findings (45); moreover, importin ␣1 acetylation was enhanced when using material immunoprecipitated from Ad(CA)AMPK-infected cultures (Fig. 5C). Increased phosphorylation of endogenous p300 was seen after IP of p300 from Ad(CA)AMPK-infected cells that had been incubated with 32 P i (Fig. 5D). Ad(CA)AMPK-infected cells also displayed an increase in the acetylation of histone H4 and importin ␣1, further demonstrating that elevated AMPK activity was linked to the increased acetylation of p300 tar- , or plasmid pCMV␤-p300 (p300), expressing wild-type p300, the abundance of p300, total importin ␣1, acetylated importin ␣1, total histone H4, and acetylated histone H4, was assessed by Western blotting using 20 g of whole cell lysate. Antibodies and procedures are described under "Experimental Procedures." B, AMPK activity in cells that had been infected with either an adenovirus expressing a constitutively active isoform of the ␣ subunit of AMPK (Ad(CA)AMPK) or control AdGFP adenovirus was calculated as explained in the legend of Fig. 1A. C, top panel, endogenous p300 was immunoprecipitated from RKO cells that had been infected with either Ad(CA)AMPK or AdGFP and then used for in vitro acetylation of 0.5 g of purified His-importin ␣1 (His-Imp␣1) or the control protein MBP-HuR. The protein molecular weight markers are indicated. Bottom panel, the presence of p300, His-Imp␣1, and MBP-HuR in the reaction materials was monitored by Western blotting. D, RKO cells infected with either Ad(CA)AMPK or control AdGFP were incubated with 32 P i for 3 h, whereupon phosphorylated p300 was detected by immunoprecipitation of p300, separation of IP material by 12% SDS-PAGE, and visualization using a PhosphorImager. 24 h after infection of 293 cells using either Ad(CA)AMPK or control AdGFP, the abundance of p300, total importin ␣1, acetylated importin ␣1, total histone H4, and acetylated histone H4 was assessed by Western blotting using 20 g of whole cell lysate. The antibodies and procedures are described under "Experimental Procedures." FIG. 4. Importin ␣1 and HuR interact in vitro. A, 2 g each of purified recombinant Imp␤, MBP-HuR, or MBP were incubated with immobilized His-Imp␣1 and used in solution binding assays (described under "Experimental Procedures"). B, 2 g each of purified recombinant Imp␤, Kap␤2, His-Imp␣1, and GST-c-Jun were incubated with immobilized MBP-HuR. All bound and unbound fractions, prepared as described under "Experimental Procedures," were size-fractionated by 12% SDS-PAGE and detected by Coomassie Blue staining. H.C., heavy chain; MWM, molecular weight marker; *, truncated MBP-HuR product. gets (Fig. 5D). As previously reported (40), infection of RKO cells with Ad(CA)AMPK caused a reduction in cytoplasmic HuR levels (not shown).
Importantly, overexpression of p300 also caused a decrease in cytoplasmic HuR levels, as determined by both Western blotting (Fig. 6A) and immunofluorescence (Fig. 6B). These findings strongly support the hypothesis that p300-mediated events influence the subcellular compartmentalization of HuR.
Importin ␣1 Is Phosphorylated by AMPK in Vitro and in Vivo-Next, we examined whether importin ␣1 was a target of AMPK phosphorylation. We first obtained active AMPK through infection of RKO cells with Ad(CA)AMPK and then used the immunoprecipitated enzyme in in vitro phosphorylation assays (described under "Experimental Procedures"). As shown in Fig. 7A, AMPK was capable of phosphorylating purified His-Imp␣1 but not control proteins Imp␤ or MBP-HuR. AMPK was also capable of phosphorylating endogenous importin ␣1, as demonstrated in phosphorylation reactions using immunoprecipitated endogenous importin ␣1 as substrate; control IP reactions employing karyopherin ␤2 or HuR as substrates did not produce phosphorylated bands (Fig. 7B). Finally, evidence supporting the notion that AMPK was capable of phosphorylating importin ␣1 in vivo was obtained through infection of RKO cells using Ad(CA)AMPK. As shown, importin ␣1 immunoprecipitated from Ad(CA)AMPK-infected popula-tions had incorporated considerably more 32 P i than control AdGFP-infected populations (Fig. 7C).
Wild-type Importin ␣1, but Not Phosphorylation-or Acetylation-defective Mutants, Can Promote the Nuclear Localization of HuR-Additional studies were aimed at studying the effect of importin ␣1 mutants on the subcellular localization of HuR. In particular, we sought to map the residue of the importin ␣1 protein that was phosphorylated by AMPK and investigate the influence of importin ␣1 phosphorylation on the cytoplasmic abundance of HuR. First, several importin ␣1 N-terminal deletion mutants (Fig. 8A) were expressed in E. coli as His-tagged proteins. IP of His-tagged products from crude bacterial lysates was followed by Western blotting (Fig. 8B, IPϩWB), then kinase assay (Fig. 8B, 32 P signal) to determine their suitability as substrates of phosphorylation by AMPK. As shown, full-length FIG. 6. Decreased cytoplasmic localization of HuR in cells overexpressing p300. A, Western blot analysis to assess HuR levels in whole cell (Total, 10 g), cytoplasmic (Cytopl., 40 g), and nuclear (10 g) lysates prepared from 293 cells 48 h after transfection with 20 g of either plasmid pCMV␤ (V) or plasmid pCMV␤-p300 (p300). Assessment of the levels of the cytoplasm-specific ␤-tubulin and nucleus-specific HDAC1 after sequential stripping and reprobing of the filters served to verify the quality and equal loading of the cytoplasmic and nuclear preparations, respectively. Western blotting signals were measured by densitometry scanning; HuR abundance, calculated after normalization against the values of control proteins (either ␤-tubulin or HDAC1), is presented as the percentage of remaining signals relative to those measured in untreated samples of each lysate group. Shown are representative Western blotting signals; experiments were performed in triplicate. B, detection of HuR by immunofluorescence in 293 cells 48 h after transfection with either plasmid pCMV␤ (V) or plasmid pCMV␤-p300 (p300). Left panels, HuR immunofluorescence; right panels, Hoechst staining to visualize nuclei. Representative photographs from three independent experiments are shown.

FIG. 7. AMPK phosphorylates importin ␣1 in vivo and in vitro.
A, constitutively active AMPK, prepared from 293 cells that had been infected with Ad(CA)AMPK, was prepared by IP using an anti-Myc antibody. The immunoprecipitated material was then used in in vitro phosphorylation assays using purified His-Imp␣1 as well as control proteins Imp␤ and MBP-HuR as substrates. B, endogenous importin ␣1 and constitutively active AMPK were immunoprecipitated independently, and the immunoprecipitated materials incubated to assess the phosphorylation of importin ␣1 in vitro. In vitro phosphorylation assays (A and B) were carried out as described under "Experimental Procedures"; the phosphorylated products were visualized through electrophoresis of the kinase reaction mixtures in SDS-containing 12% polyacrylamide gels that were subsequently transferred onto filters and subjected to both PhosphorImager scanning (top panel) and Western blotting (WB, bottom panel) to detect the proteins indicated. C, in vivo phosphorylation of importin ␣1. 293 cells infected with either Ad-AMPK(CA) or AdGFP were incubated with 1 mCi/ml 32 P i , whereupon whole cell lysates were immunoprecipitated (IP) and importin ␣1 abundance and phosphorylation status were analyzed by Western blotting (WB, top panel) and PhosphorImager scanning (bottom panel), respectively.
importin ␣1 (Imp␣1), as well as a deletion construct lacking the first 62 amino acids (Imp␣1(⌬62)), were readily phosphorylated by AMPK. However, an importin ␣1 truncated variant lacking the first 205 amino acids (Imp␣1(⌬205)) could not be phosphorylated, indicating that a putative phosphorylation site existed within positions 62 and 205 of the importin ␣1 protein. The preferred AMPK phosphorylation site, derived from six AMPK target proteins and further refined using synthetic peptides (reviewed in Ref. 39), has been reported to be either a serine or a threonine residue within the following amino acid context: Hyd-(Xaa, Bas)-Xaa-Xaa-Ser/Thr-Xaa-Xaa-Xaa-Hyd (where Hyd (Ϫ5) represents amino acids carrying a basic side chain (Leu, Met, Ile, Phe, or Val), and Bas, which can be located at both positions Ϫ4 and Ϫ3, represents amino acids carrying a basic side chain (Arg Ͼ Lys Ͼ His)). Although AMPK can phosphorylate synthetic peptides on both threonine and serine residues, all in vivo targets identified thus far are phosphorylated on serine residues. Scanning of the importin ␣1 protein revealed one such AMPK phosphorylation motif encompassing Ser 105 and only one additional motif encompassing Ser 179 . Initial indications that Ser 105 might be a target of AMPK-mediated phosphorylation came from studies in which synthetic peptides were tested. A peptide comprising Ser 105 and flanking amino acids, 98 ARKLLSREKQ 112 , was readily phosphorylated by AMPK (Fig. 8C). In fact, phosphorylation levels were greater than 50% of those seen when using the archetypal SAMS peptide. By contrast, peptides encompassing other regions of importin ␣1 apparently lacking target AMPK phosphorylation sites, such as peptide 71 QGTVNWSVDDIVKGI 85 , which comprised Ser 77 and surrounding amino acids, and peptide 172 AS-PHAHISEQAVWALG 188 , which comprised Ser 179 and surrounding amino acids, were poor targets of AMPK-mediated phosphorylation (Fig. 8C). An additional demonstration that Ser 105 was a target of phosphorylation by AMPK came from the generation and testing of point importin ␣1 mutants (see Fig. 10).
To test the influence of importin ␣1 modifications on the cytoplasmic abundance of HuR, we first prepared several importin ␣1 deletion constructs for analysis in cells. Our initial attempts to carry out import assays in permeabilized cells, the preferred methodology to study nuclear import, were inconclusive; our efforts to modify (phosphorylate and acetylate) bacterially produced importin ␣1 through post-translational modification reactions rendered a pool of importin ␣1 in which the extent and stability of these modifications could not be confirmed with certainty and also inevitably generated degradation by-products that yielded false positive results on import assays (not shown). Therefore, we opted for studying this process by investigating the influence of importin ␣1 overexpression in mammalian cells. The effects of importin ␣1, expressed either as the full-length wild-type protein or as truncated or point-mutant proteins, were assessed by monitoring cytoplasmic HuR levels by Western blotting and by immunofluorescence. cDNA inserts encoding Imp␣1, Imp␣1(⌬62), and Imp␣1(⌬205) (Fig. 8A) were subcloned into mammalian expression vector pcDNA3.1/V5-His-TOPO and expressed as His- tagged products in 293 cells by transient transfection; expression levels of these proteins were monitored by Western blot analysis using an anti-His antibody (Fig. 9A). Compared with the relative abundance of cytoplasmic HuR in vector-transfected cells (vector), full-length importin ␣1 overexpression (Imp␣1) was capable of promoting marked decreases in cytoplasmic HuR levels (to between 20 -30% of the levels seen in vector-transfected populations; Fig. 9B). By contrast, overexpression of Imp␣1(⌬205), which lacks residues Ser 105 and Lys 22 , failed to substantially reduce HuR cytoplasmic levels compared with vector-transfected samples. The finding that overexpression of Imp␣1(⌬62), which lacks residue Lys 22 but contains Ser 105 , allowed recovery of 72% of cytoplasmic HuR argued that acetylation at this site is necessary for maximal importin ␣1-mediated transport of HuR into the nucleus. No changes in either total or nuclear HuR levels were detected in any of the transfection groups (Fig. 9B), as anticipated (24).
Analysis of the subcellular distribution of HuR by immunofluorescence (Fig. 11A) was in precise agreement with the Western blotting data (Fig. 10C). Overexpression of Imp␣1, but not vector, Imp␣1(K22R), Imp␣1(S105A), or Imp␣1(K22R, S105A), effectively led to an enhanced HuR import, as revealed by the marked reduction in cytoplasmic HuR in these populations. Similar findings were obtained in transient transfections to express the deletion mutants described in Fig. 9 (data not shown). Importantly, only wild-type importin ␣1 appeared to interact with HuR, as revealed by co-IP analysis using an anti-HuR antibody followed by Western blot analysis of the ectopically expressed (HA-tagged) importin ␣1 proteins present in the IP material using an anti-HA antibody. As shown in Fig.  11B, wild-type importin ␣1 appeared to interact with HuR, but no such interactions were detected with any of the point mutants tested. These findings lend support to the notion that wildtype importin ␣1, but not importin ␣1 mutants that cannot be phosphorylated or acetylated, are capable of binding HuR.
From the data presented here, a model is proposed whereby AMPK elicits a dual modification of importin ␣1 (Fig. 12): it phosphorylates importin ␣1 directly (Ser 105 ) and acetylates importin ␣1 indirectly, through phosphorylation of p300, which in turn acetylates importin ␣1 (Lys 22 ). Importin ␣1 bearing mutations in these modification sites have an impaired ability to promote the nuclear localization of HuR.

DISCUSSION
In this study, we sought to investigate the mechanisms whereby AMPK mediated the previously reported nuclear import of HuR (40). Our findings strongly suggest that importin ␣1 participates in the nuclear import of HuR and further support a direct role for AMPK in modulating this function by eliciting a dual modification of importin ␣1. Using both in vivo and in vitro approaches, AMPK was found to directly phosphorylate importin ␣1 on residue Ser 105 . In addition, AMPK indirectly caused importin ␣1 acetylation, an effect that relied on AMPK-mediated phosphorylation of acetylase p300 (47). Accordingly, mutated importin ␣1 proteins lacking the phosphorylation site, the acetylation site, or both sites exhibited an impaired ability to mediate the nuclear import of HuR in intact cells. FIG. 9. Influence of overexpression of full-length importin ␣1 or truncated importin ␣1 on cytoplasmic HuR levels. A, 24 h after transfection of 293 cells with His-tagged importin ␣1 constructs derived from vector pcDNA3.1/V5-His-TOPO, full-length importin ␣1 (Imp␣1) or N-terminally truncated importin ␣1 mutant proteins (Imp␣1(⌬62) and Imp␣1(⌬205)) were detected using an anti-His antibody. B, 24 h after transfection of RKO cells with the constructs indicated, Western blot analyses were carried out to assess HuR levels in cytoplasmic (Cytopl., 40 g), nuclear (10 g), and whole cell (Total, 10 g) lysates.
Verification of the quality and equal loading of the protein preparations was carried out by assessing the levels of ␤-tubulin (for cytoplasmic lysates) and HDAC1 (for nuclear and whole cell lysates) after stripping and reprobing the filters. Western blotting signals in the cytoplasmic preparations were measured by densitometry scanning; HuR abundance was calculated after normalization against the intensity values of ␤-tubulin and indicated as a percentage of remaining signals relative to those measured in vector-transfected samples. The experiment was repeated three times, and representative Western blots are shown. Cellular macromolecules are transported between the nucleus and the cytoplasm through the nuclear pore complex (the NPC, a large structure composed of ϳ30 proteins named nucleoporins), mediated by the action of several families of soluble transport molecules. One such family of transport molecules, represented by NTF2, is involved in the nuclear import of the small GTPase Ran (52,53), whereas another comprises the TAP/NXF p15/NXT protein dimer, which participates in the nuclear export of mRNA (reviewed in Ref. 54). However, the largest family of transport factors is the family of proteins known as importins/exportins or karyopherins. In mammalian cells, many members of this family have been shown to be involved in nuclear import (such as importin ␤, karyopherin ␤2, transportin 2, transportin-SR, importin ␣1, importin ␣5, importin ␣7, and importin ␣9), whereas others have been shown to mediate nuclear export (like CAS, CRM1, exportin-t, exportin 4, and exportin 5), and yet others been reported to mediate both import and export (like importin 13) (36,49,55).
Members of the importin ␤ family have been most extensively implicated in the "classical" import pathway of cytoplasmic import substrates (cargoes). In addition to receptor molecules such as those comprising the importin ␤ family, adaptor proteins such as those in the importin ␣ family (including importins ␣1, ␣3, ␣4, ␣5, ␣6, and ␣7 in humans (44)) mediate the recognition of nuclear localization signals such as the classical NLS but require importin ␤ to be transported through the NPC. Although some evidence exists that importin ␣ may cross the NPC independently of importin ␤ (56), there are no reports to date that cargo-bound importin ␣ can enter the nucleus freely through the NPC. Our findings further support the view that importin ␤ may be present in complexes that also comprise importin ␣1 and HuR, given the association of these three proteins in immunoprecipitated material from cell lysates (Fig.  3) and their interaction in solution binding assays (Fig. 4). These interactions suggest that HuR is a nuclear import cargo for the importin ␣1-importin ␤ pathway. An NLS of HuR had previously been mapped to its HNS, which comprises residues 205-237 (27), and the HNS has been shown to be the NLS for karyopherin ␤2 and transportin 2 nuclear import pathways (35,36). The identity of the HuR NLS that binds importin ␣ has not been determined, and no strict monopartite or bipartite NLS sequences have been identified. A candidate bipartite-like NLS, 205 RRFGGPVHHQAQRFRF 220 was first suggested by Fan and Steitz (27), but its fusion with pyruvate kinase failed to localize to the nucleus, suggesting that it is likely not an efficient NLS. Another sequence that resembles a monopartite classical NLS, albeit suboptimal, is 320 KTNKSHK 326 , found at the C terminus of HuR. Recently, the repertoire of sequences FIG. 10. Influence of overexpression of wild-type importin ␣1 and point-mutated importin ␣1 on cytoplasmic HuR levels. A, His-tagged importin ␣1 constructs were prepared from vector pcDNA3.1/V5-His-TOPO to express either wild-type importin ␣1 (Imp␣1) or importin ␣1 proteins bearing point mutations as shown. B, 24 h after transfection of 293 cells with the constructs indicated, in vitro phosphorylation was assessed by using immunoprecipitated (CA)AMPK (obtained by performing anti-c-Myc IP from Ad(CA)AMPK-infected 293 cells; "Experimental Procedures") and His-tagged proteins were detected by Western blotting. C, twenty-four h after transfection of 293 cells with the constructs indicated, Western blot analyses were carried out to assess HuR levels in cytoplasmic (Cytopl., 40 g) and whole cell (Total, 10 g) lysates. To verify the quality and equal loading of the protein preparations, the blots were stripped and reprobed to detect ␤-tubulin; an anti-Myc antibody was used to monitor the evenness of tagged protein expression in the transfected populations. Western blotting signals in the cytoplasmic preparations were measured by densitometry scanning; HuR abundance was calculated after normalization against the intensity values of ␤-tubulin and indicated as a percentage of the remaining signals relative to those measured in vector-transfected samples. The experiment was repeated three times; representative Western blots are shown. that bind in an NLS-like manner to importin ␣ has been expanded to include the MNRRKIAMPKRRMAFK sequence of nucleoporin Nup2p, which binds in a different but overlapping site from the classical NLS (57). Therefore, it is conceivable that importin ␣ binds cargo ligands containing NLS other than the well known mono-and bipartite classical NLSs.
While this work was in progress, Gü ttinger et al. (36) as well as Rebane et al. (35) reported that Trn1/Kap␤2 and Trn2 can mediate the nuclear import of HuR. Using in vitro import assays, they found that the Trn/Kap␤2 proteins were able to import HuR, whereas importin ␤ together with importin ␣5 were not. The Trn/Kap␤2 pathways and the pathway described in the present investigation likely represent multiple parallel and co-existing mechanisms of HuR nuclear import. Supporting this notion are other prominent examples of proteins whose nuclear import occurs via multiple import pathways (58 -61). Second, as in the study by Gü ttinger et al., importin ␣5 did not appear to bind to HuR, and according to our data, neither did other ␣ importins except importin ␣1. In this regard, other proteins like RCC1, STAT-1, and STAT-2 are known to be imported via single ␣ importin isoforms (44,(62)(63)(64). Finally, it remains to be assessed whether a lack of Kap␤2 in living cells results in a transport defect for HuR in vivo (36). Our findings that overexpression of wild-type importin ␣1 in living cells enhanced the nuclear translocation of HuR, whereas abolishing phosphorylation and/or acetylation of importin ␣1 blocked this effect, provide strong evidence that importin ␣1 is a functionally relevant import factor in vivo. Moreover, our results shed light into the functional significance that two previously reported modifications of importin ␣1, phosphorylation and acetylation (45,65), may have in the import of a specific substrate. It is important to note that by preventing importin ␣1 phosphorylation and/or acetylation in living cells, we were able to reduce, although not completely block, the nuclear translocation of HuR. We therefore propose that the import of HuR, similarly to the previously reported import of RCC1 (66), can be mediated both by a pathway dependent on a specific ␣ importin and by importin ␣-independent pathway(s). Importin ␣ has an N-terminal importin ␤-binding domain that also includes an autoinhibitory region, followed by an NLS-binding domain comprising 10 armadillo repeats that begin at amino acid ϳ70 and a C-terminal domain that is necessary for binding its export factor CAS (48,67). Our data could provide the basis for a novel regulatory model whereby phosphorylation and/or acetylation of importin ␣1 may increase its affinity for importin ␤ and also increase its affinity for nuclear import cargoes such as HuR. AMPK triggered the acetylation of residue Lys 22 , which resides within the importin ␤-binding domain of importin ␣1. In the crystal structure of free and autoinhibited mouse importin ␣, Lys 22 is disordered and not observed (68), and it is distant in sequence from the pseudo-NLS autoinhibitory sequence (residues 44 -54) in the N-terminal region of importin ␣. Therefore, acetylation of Lys 22 may not affect importin ␣ autoinhibition in the free protein. How- FIG. 11. Immunofluorescence detection of cytoplasmic HuR levels in 293 cells transfected with either full-length or truncated importin ␣1 proteins. A, HuR was detected by immunofluorescence 48 h after transfection of 293 cells with pcDNA3.1/V5-His-TOPO plasmids either lacking an insert (V) or carrying wild-type importin ␣1 (Imp␣1) or importin ␣1 proteins bearing point mutations (Imp␣1(K22R), Imp␣1(S105A), or Imp␣1(K22R,S105A)). Left panel, HuR immunofluorescence; right panel, Hoechst staining to visualize nuclei. Representative photographs from three independent experiments are shown. B, lysates from 293 cells that had been transfected 48 h earlier with the HA-tagged constructs described in A were subjected to IP using anti-HuR antibodies. The presence of the ectopically expressed wild-type or point mutant importin ␣1 proteins (all of them exhibiting the same molecular weight) in the IP materials was subsequently detected by Western blot analysis using an anti-HA antibody. H.C., heavy immunoglobulin chain.
FIG. 12. Proposed model of AMPK-triggered HuR nuclear import through its AMPK-mediated dual modification of importin ␣1. We propose that AMPK promotes the nuclear import of HuR via mechanisms involving the transport protein importin ␣1. AMPK achieves importin ␣1-mediated nuclear import through modification of importin ␣1 on two residues. One modification is the acetylation of importin ␣1, plausibly through AMPK-mediated phosphorylation of p300 (a modification that activates p300), which in turn acetylates importin ␣1 on residue Lys 22 . The second modification is the direct phosphorylation by AMPK on Ser 105 of importin ␣1. We further propose that importin ␤ mediates the import of HuR-importin ␣1 complexes into the nucleus. ever, the side chain of Lys 22 participates in hydrophobic interactions with the side chain of Trp 472 of importin ␤ (69). Consequently, acetylation of the Lys 22 side chain, which removes its positive charge and also increases its hydrophobicity, could result in increased affinity between importin ␣ and importin ␤. Ser 105 , the importin ␣ residue that is phosphorylated in response to AMPK stimulation, is in the armadillo domain situated at the C terminus of helix H3 of the first armadillo repeat (68,70). In the crystal structure of mouse importin ␣ bound to the bipartite NLS of nucleoplasmin, the Ser 105 side chain is near the Lys 170 side chain of the NLS (71). Phosphorylation of Ser 105 should increase ionic interactions between Ser 105 and the NLS, thus increasing the affinity of importin ␣ for its cargo. Taken together, acetylation and phosphorylation of importin ␣ on Lys 22 and Ser 105 could result in more efficient formation of importin ␣-importin ␤-cargo complexes, respectively, and thus further facilitate nuclear import by this pathway. Taken together, we propose that under basal growth conditions, HuR import into the nucleus is mediated by several transport factors, like karyopherin ␤2/transportin 1, transportin 2, and importin ␣1-importin ␤. Upon stimulation of AMPK, importin ␣1 is modified by acetylation and phosphorylation, resulting in a further increase in the efficiency of the importin ␣1-importin ␤ pathway to increase transport of cargoes such as HuR into the nucleus.
It will be important to investigate whether importin ␣1bound HuR is associated with target mRNAs in vivo. The data obtained from in vitro binding assays indicate that it is possible for HuR to bind to importin ␣1 without mRNA (Fig. 4). In fact, an intriguing hypothesis that remains to be tested is whether the presence of a target mRNA might actually inhibit the association of HuR with importin ␣1. A less plausible possibility, although also worthy of future examination, is that HuR might enter the nucleus bound to a target mRNA, possibly as a means of controlling its stability or preventing its translation. Experiments are underway to test these various scenarios by combining IP and real time reverse transcription-PCR assays to test for the presence of endogenous HuR target mRNAs in importin ␣1-HuR complexes. In this regard, IP of importin ␣1 under conditions that preserved protein-RNA interactions did not reveal any specific enrichment of HuR target mRNAs in the immunoprecipitated material. 2 These preliminary findings would support the notion that any HuR present in complexes with importin ␣1 would likely not be bound to mRNAs.
The findings presented here that AMPK influences importin ␣1-mediated events may have potentially important implications beyond the regulation of HuR function. Changes in the subcellular distribution of other importin ␣1 cargo molecules, which typically contain classical NLS, should be re-examined in light of a potential role for AMPK in regulating their nucleocytoplasmic transport. In addition, importin ␣7 and to a lesser extent importin ␣3, also appeared to be adequate in vitro targets of AMPK-mediated phosphorylation. 3 This finding was not surprising, given the conservation of sequences around the importin ␣1 Ser 105 ( 100 ARKLLSREKQ 109 ), importin ␣3 ( 95 ARKLLSSDRN 104 ), and importin ␣7 ( 104 FRKLLSKEPS 113 ). Although neither importin ␣3 nor importin ␣7 appeared to form bona fide complexes with HuR, their target cargoes may likewise be subject to nuclear import in an AMPK-governed fashion. However, the functional consequences of phosphorylation of ␣ importins remains to be examined in greater detail, given a recent report that mutation of the conserved Ser 56 phosphorylation site of Drosophila importin ␣2 did not appear to influence its in vivo function in oogenesis (72). Assessment of the influence of importin ␣1 mutants on the nuclear import of other ␣ importin cargoes (such as STAT-1, STAT-2, tissue transglutaminase, DNA helicase Q1, the serum-and glucocorticoidinducible kinase Sgk1, p53, lymphoid enhancer factor-1, etc.) was hampered by the need to treat cells with various agents prior to analysis, the lack of availability of suitable antibodies, or the fact that the cargo molecules in question were preferentially imported by importins other than importin ␣1. Further work is undoubtedly needed to establish whether other importin ␣1 cargoes might also be regulated by AMPK in the same fashion as HuR. The functional redundancy of importin ␣ orthologs will make this task a challenging one.
The regulation of AMPK on gene expression has been best studied for the yeast AMPK homolog SNF1. SNF1 phosphorylates various transcription factors that jointly up-regulate the expression of glucose-repressed genes. In mammalian cells, AMPK has been proposed to regulate the expression of genes important for energy conservation at a time of low fuel availability (39). Although the direct effectors of this regulation have not been fully elucidated, a growing number of nuclear proteins, including transcription factors, have been identified as targets of phosphorylation by AMPK (73). In light of the results from this study, we propose that AMPK effects on importin ␣1, and possibly other importin ␣ proteins, further contribute to altering patterns of expressed genes by regulating the import of proteins that modulate gene expression at multiple levels. Such targets may include NLS-containing transcription factors, as well as other RNA-binding proteins known to undergo nuclear import or nucleocytoplasmic shuttle (74 -77). Given the abundance of NLS, both classical and nonclassical (such as M9), in proteins destined for nuclear import, elucidation of the specificity of the cargoes that utilize the AMPK and importin ␣1 pathway remains an important matter for future study.
In summary, we have described a novel function for AMPK as regulator of the nuclear localization of HuR, and we propose that such an import process is mediated by importin ␣1. Our data provide an important specific example of how AMPK may regulate gene expression in mammalian cells, a function of AMPK that had been previously postulated based on the direct effect of the yeast SNF1 homolog on transcription patterns. In mammalian cells, AMPK-driven modifications of importin ␣1 function may thus dictate the subcellular presence of various NLS-bearing RNA-binding proteins as well as transcription factors and thereby orchestrate transcriptional and post-transcriptional events underlying changes in gene expression patterns.