An Annexin 2 Phosphorylation Switch Mediates p11-dependent Translocation of Annexin 2 to the Cell Surface*

Annexin 2 is a profibrinolytic co-receptor for plasminogen and tissue plasminogen activator that stimulates activation of the major fibrinolysin, plasmin, at cell surfaces. In human subjects, overexpression of annexin 2 in acute promyelocytic leukemia leads to a bleeding diathesis reflective of excessive cell surface annexin 2-dependent generation of plasmin (Menell, J. S., Cesarman, G. M., Jacovina, A. T., McLaughlin, M. A., Lev, E. A., and Hajjar, K. A. (1999) N. Engl. J. Med. 340, 994–1004). In addition, mice completely deficient in annexin 2 display fibrin accumulation within blood vessels and impaired clearance of injury-induced thrombi (Ling Q., Jacovina, A.T., Deora, A.B., Febbraio, M., Simantov, R., Silverstein, R. L., Hempstead, B. L., Mark, W., and Hajjar, K. A. (2004) J. Clin. Investig. 113, 38–48). Here, we show that endothelial cell annexin 2, a protein that lacks a typical signal peptide, translocates from the cytoplasm to the extracytoplasmic plasma membrane in response to brief temperature stress both in vitro and in vivo in the absence of cell death or cell lysis. This regulated response is independent of new protein or mRNA synthesis and does not require the classical endoplasmic reticulum-Golgi pathway. Temperature stress-induced annexin 2 translocation is dependent on both expression of protein p11 (S100A10) and tyrosine phosphorylation of annexin 2 because annexin 2 release is completely eliminated on depletion of p11, inactivation of tyrosine kinase, or mutation of tyrosine 23. Translocation of annexin 2 to the cell surface dramatically increases tissue plasminogen activator-dependent plasminogen activation potential and may represent a novel stress-induced protein secretion pathway.

The annexins represent a Ͼ50-member family of calcium-dependent, phospholipid-binding proteins of largely unknown function (1). The canonical annexin "fold" (G-X-G-T-(38)-D/E) allows these proteins to shuttle intracellularly between aqueous and membrane compartments in response to fluctuations in calcium concentration. Like most family members, annexin 2 consists of a variable, M r ϳ3,000 amino-terminal "tail" domain and a conserved, M r ϳ33,000 carboxyl-terminal "core" region.
The core specifies membrane binding capability, whereas the tail possesses an invariant tyrosine 23, the pp60-c-src phosphorylation target (2). The annexin 2 tail also mediates heterotetramer formation with the S100 protein p11 (S100A10) that enhances membrane phospholipid binding affinity (3). Annexin 2 was originally discovered as a cellular substrate for the avian sarcoma virus-transforming gene product, pp60-src (4).
Recent evidence indicates that annexin 2 is a constitutive regulator of the fibrinolytic system. Annexin 2 serves as a co-receptor for the plasma-derived fibrinolytic zymogen, plasminogen, and its endothelial cell-derived activator, tissue plasminogen activator (tPA) 1 (5,6). Both ligands assemble at distinct domains under equilibrium conditions and with high affinity, forming a catalytic complex that augments the efficiency (K m /k cat ) of plasmin generation by 60-fold (7). Annexin 2 binding sites for plasminogen and tPA are blocked by interaction with two recognized atherothrombotic agents, lipoprotein(a) and homocysteine, respectively, providing a mechanistic basis for occlusive vascular disease in disease states associated with hyperlipoproteinemia(a) or hyperhomocysteinemia (8,9). In rats, furthermore, pretreatment of the carotid artery with wild-type annexin 2, but not a mutant form of the protein, prevents vessel thrombosis in response to oxidative injury with ferric chloride (10). In human subjects, overexpression of annexin 2 in acute promyelocytic leukemia leads to a hyperfibrinolytic bleeding diathesis reflective of excessive cell surface annexin 2-dependent generation of plasmin (11). In addition, mice completely deficient in annexin 2 display fibrin accumulation within blood vessels and impaired clearance of injuryinduced thrombi (12). Thus, regulated endothelial cell surface expression of annexin 2 may represent a crucial cellular defense mechanism against intravascular thrombosis following vascular injury or stress.
Despite evidence for the central role of annexin 2 in fibrinolysis, the mechanisms that control annexin 2 cell surface expression have not been delineated. Recent evidence indicates that members of the annexin family can be released into the extracellular environment even though they lack the hydrophobic signal sequence typical of classically secreted proteins. Annexin 1 is actively secreted into human seminal plasma (13), and annexin 6 translocates from an intracellular pool to the external basolateral surface of mouse mammary duct epithelium upon onset of lactation (14). Evidence that annexin 2 is similarly expressed on cell surfaces includes the following observations: anti-annexin 2 IgG specifically inhibits binding of plasminogen and tPA to endothelial cells (5), annexin 2-transfected human embryonic kidney HEK293 cells acquire the abil-ity to bind both ligands (5), annexin 2 can be eluted from the cell surface by the calcium-chelating agent EGTA (6), and flow cytometric analysis of fixed nonpermeabilized endothelial, monocytoid, and promyelocytic cells reveals annexin 2 crossreactive material on the cell surface (5,11,15). Nevertheless, the mechanism for extracellular transport of annexin 2 and related polypeptides is unknown.
In the present study, we used a temperature-induced model of cellular stress to examine the stimulated translocation of annexin 2 to the cell surface. Our data show that annexin 2 translocation occurs independently of the classical endoplasmic reticulum-Golgi pathway and does not involve de novo protein synthesis. In contrast, translocation requires the presence of the annexin 2 binding partner p11 (S100A10) and the phosphorylation of annexin 2 at Tyr 23 through a Src-like tyrosine kinasedependent mechanism both in vitro and in vivo. Using both biochemical and quantitative cell biological approaches, we found that brief temperature stress was associated with a doubling or tripling of annexin 2 expression at the cell surface, an increase that would have profound implications for hemostatic balance in humans (11).

EXPERIMENTAL PROCEDURES
Cell Culture and Temperature Stress-Human umbilical vein endothelial cells (HUVECs; 80 -90% confluent) were isolated and propagated as described previously (5) and studied at passage 1-4. For temperature stress, confluent cells were washed twice with HEPES buffered saline (pH 7.4) and incubated in serum-containing medium at 42°C (heat stress) or 37°C (control) in a 5% CO 2 humidified incubator (3 h). Endothelial cell viability after heat stress was verified by trypan blue exclusion and by lactate dehydrogenase release as described previously (16). Inhibitors (cycloheximide, brefeldin A, okadaic acid, sodium orthovanadate, genistein, PP3, PP2, and herbimycin A) were used in each experiment at concentrations that did not produce cytotoxicity as assessed by trypan blue exclusion.
Microinjection and Quantification of Annexin 2-GFP at the Cell Surface-The full-length human annexin 2 cDNA was cloned in-frame with a carboxyl-terminal GFP coding sequence into the eukaryotic expression vector pEGFP-N3 (Clontech) utilizing unique XhoI and KpnI restriction sites. To ascertain the expression of the full-length annexin 2-GFP protein, transient transfection (SuperFect reagent; Qiagen) of transformed HEK293 human embryonic kidney cells was carried out using the above-mentioned cDNA, followed by immunoblot analysis. HUVECs were plated on sterile glass coverslips coated with fibronectin (7 g ml Ϫ1 ; PBS) and allowed to adhere for 6 -8 h. Annexin 2-GFP cDNA stock was prepared in water and subsequently diluted in HKCl microinjection buffer (10 mM HEPES, 140 mM KCl, pH 7.4) to a final concentration of 10 g ml Ϫ1 . Annexin 2-GFP cDNA was introduced into cell nuclei by pressure microinjection using back-loaded glass capillaries and a Narishige micromanipulator (Narishige, Greenvale, NY) (17). Cells were maintained at 37°C in a humidified, 5% CO 2 environment for 60 min after microinjection to allow expression of the exogenous cDNA and then either maintained at 37°C (control) or transferred to 42°C (heat stress) in a humidified incubator for 3 h. Cells were rinsed in PBS and fixed in 2% paraformaldehyde (21°C, 2 min). Annexin 2-GFP at the cell surface was immunolabeled with rabbit anti-GFP IgG (7.5 g ml Ϫ1 ; Novus Biologicals, Littleton, CO), followed by Cy3-conjugated donkey anti-rabbit IgG secondary antibody (1:200 dilution; Jackson Immunoresearch). Images of injected cells expressing annexin 2-GFP were acquired using a Nikon E-600 microscope coupled to a back-illuminated, cooled charge-coupled device camera (CCD1000-PB; Princeton Instruments/Roper Scientific) and transferred to a computer work station running MetaMorph imaging software (Universal Imaging Corp/Molecular Devices). Images were collected using a fluorescein filter (B-2E/C DM 505; Nikon) for direct detection of annexin 2-GFP and a rhodamine filter (G-2E/C DM 565; Nikon) for indirect immunofluorescence detection of cell surface-associated annexin 2-GFP. Images of annexin 2-GFP and anti-GFP labeling were acquired using identical acquisition settings and exposure times for all samples in a single experiment. The integrated cellular fluorescence intensities of total annexin 2-GFP and surface-associated anti-GFP were measured, and the ratio of anti-GFP/annexin 2-GFP intensities was calculated for each group of injected cells.
Plasmin Generation Assays-Rates of plasminogen activation were assessed after incubation of cells with amino-terminal glutamic acid plasminogen (100 nM, 1 h, 4°C) and tPA (10 nM) premixed with the fluorogenic plasmin substrate D-Val-Leu-Lys-7-amino-4-trifluoromethyl coumarin (AFC-81; Enzyme Systems) (125 M, 200 l) (19). Substrate hydrolysis was estimated in duplicate at 4-min intervals in a fluorescence spectrophotometer (SpectraMAX Gemini XS; Molecular Devices) as relative fluorescence units (RFU) at excitation 400 nm and emission 505 nm. Initial rates of plasmin generation were calculated for triplicate experiments using linear regression analysis of plots of RFU versus min 2 (RFU min Ϫ2 ). In the absence of tPA, no significant plasmin generation was observed.
Stable Transfections-Together with a carboxyl-terminal tag containing the c-Myc epitope, the full-length human annexin 2 cDNA was cloned in-frame into the eukaryotic expression vector pcDNA3.1/ Myc-His (Invitrogen) utilizing unique XhoI and KpnI restriction sites. The resulting wild-type vector was subsequently modified by site-directed mutagenesis (QuikChange; Stratagene), substituting the tyrosine (codon TAT) at position 23 with an alanine (codon GCT) resulting in the Y23A mutant. The integrity of both vectors was verified by full-length cDNA sequence analysis. Transformed human embryonic kidney cells (HEK293) were transfected with either wild-type or Y23A annexin 2 mutant cDNA using Superfect reagent (Qiagen). Stable clones were selected and maintained in the presence of G418 (600 g ml Ϫ1 ; Invitrogen). Total cell lysates for immunoblot and immunoprecipitation analysis from these cells were prepared using the protocol mentioned for HUVECs. Anti-myc IgG (Cell Signaling Technology) was used to detect the expression of myc-tagged annexin 2, whereas antiphosphotyrosine (PY 20; Transduction Laboratories) was used to detect phosphotyrosylated annexin 2.
p11 Silencing-Two annealed 64-bp oligonucleotides, each containing 19-nucleotide reverse complement sequences homologous to a portion of the mouse p11 gene ( 141 GGATCCTCTGGCTGTGGAC 159 ), separated by a 9-nucleotide spacer sequence, were subcloned into the BglII and HindIII sites of the 3.2-kb plasmid pSUPER (Oligoengine) containing the H1-RNA promoter (20). A neomycin resistance cDNA cassette was introduced between unique XhoI and KpnI sites, downstream of the HindIII site. The empty pSUPER vector containing the neomycin resistance cassette served as a control. A transformed mouse endothelial cell line (a kind gift of Dr. B. Weksler; Weill Medical College of Cornell University) was transfected with the above-mentioned vectors using Superfect reagent (Qiagen), and single cell clones were selected, expanded, and propagated in the presence of G418 (400 g ml Ϫ1 ; Invitrogen). Total RNA was extracted by using TRIzol reagent (Invitrogen). 1 g of total RNA was reverse transcribed using One-Step reverse transcription-PCR kit (Qiagen) according to the manufacturer's protocol for 22 cycles using mouse p11, annexin 2, and glyceraldehyde-3-phosphate dehydrogenase gene-specific primers, and the products were analyzed by agarose gel electrophoresis.
In Vivo Temperature Stress-Mice (C57Bl/6, 4 -8 weeks old) received intraperitoneal injection with either 0.2% Me 2 SO or herbimycin A in Me 2 SO (200 g kg Ϫ1 ) 2 h before placement in a 2-cm-deep, constant temperature water bath (40.8°C Ϯ 0.2°C) or maintenance at room temperature (24°C) for 45 min (21). After temperature stress, mice were anesthetized by inhalation of methoxyfluorane until tail and toe pinch reflexes were extinguished. After thoracotomy, the right atrial appendage was incised, and the animal was perfused via the left ventricle using a constant infusion pump. The perfusate consisted of 10 ml of ice-cold PBS (2 ml min Ϫ1 ), followed by cell-impermeable sulfo-NHS-SS-biotin (1 mg ml Ϫ1 in PBS, 1 ml min Ϫ1 , 5 ml). Excess biotin was then quenched by infusion of 50 mM NH 4 Cl, followed by ice-cold PBS (1 ml min Ϫ1 ; 5 ml, total volume). The thoracic aorta was harvested, washed twice with ice-cold PBS, and homogenized in 20 mM Tris-HCl, 150 mM NaCl, 5 mM EDTA, pH 8.0, containing 1% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 1 mM sodium orthovanadate, 10 g ml Ϫ1 leupeptin, and 10 g ml Ϫ1 aprotinin (Dounce, Teflon pestle, 10 -12 strokes) before extraction in the same buffer (4°C, 1 h). Cell surface biotinylated proteins were streptavidin (agarose-conjugated)-precipitated from equal protein concentrations of homogenate and analyzed by immunoblot with anti-annexin 2 IgG.

Heat Stress Stimulates Translocation of Annexin 2 to the Surface of Cultured Endothelial Cells in Vitro-
To study regulation of cell surface annexin 2 expression in the human endothelial cell, we used a 3-h period of mild temperature stimulation designed to induce a typical heat stress response.
To determine the effect of this maneuver on cell viability, we utilized the trypan blue exclusion assay, which revealed no significant difference between cells incubated at either 37°C or 42°C. Terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling assay for apoptosis revealed less than 0.0003% immunofluorescence among cells maintained at either 37°C or 42°C, whereas 20 -30% of terminal deoxynucleotidyl transferase-treated positive controls showed strong immunofluorescence. Similarly, neither release of the cytoplasmic enzyme lactate dehydrogenase (9.0 Ϯ 2.4 versus 5.9 Ϯ 3.2 units ml Ϫ1 , S.E.; n ϭ 3) nor release of 51 Cr from preloaded cells (2.5 Ϯ 0.3% versus 0.5 Ϯ 0.4%, S.E.; n ϭ 3) was increased in cells treated at 42°C versus those maintained at 37°C. These data indicated that heat stress failed to induce cell death or cell lysis.
We have previously utilized cell surface elution with the calcium-chelating agent EGTA to dissociate annexin 2 from the endothelial cell surface (16). Whereas annexin 2 levels in cell lysates remained invariant during temperature stress, levels in surface EGTA eluates increased significantly over those in 37°C controls (Fig. 1A). Similarly, among cell surface proteins derivatized with a cell-impermeable biotin analog and precipitated with immobilized streptavidin, temperature stress induced a 2.2 Ϯ 0.2-fold (mean Ϯ S.E.; n ϭ 5) increase in cell surface expression of both annexin 2 and p11 in 42°C samples over 37°C controls (Fig. 1B). Because increased temperature did not induce any change in the recovery of control fluid-phase annexin 2 by biotin-streptavidin precipitation and we saw no difference in detection of another membrane protein, thrombomodulin, we concluded that heat stimulation leads to increased biomass of annexin 2 at the cell surface.
Increased expression of annexin 2 at the cell surface was accompanied by a concomitant increase in cell surface fibrinolytic potential. Temperature stress induced a 2.3 Ϯ 0.2-fold (mean Ϯ S.E.; n ϭ 6) increase in tPA-dependent plasmin generation compared with cells maintained at 37°C (Fig. 1C). Whereas preimmune IgG had no effect on the rate of plasminogen activation, the heat-induced increase in plasmin production was completely inhibited in the presence of anti-annexin 2 IgG, and baseline plasminogen activation at 37°C was reduced by 61.6 Ϯ 3.6% (mean Ϯ S.E.; n ϭ 3). Preimmune IgG, on the other hand, had no effect on this activity. These data indicated that translocated annexin 2 was fibrinolytically active and that all of the heat-induced increment in fibrinolytic potential could be attributed to increased annexin 2 expression.
To better understand the origin of stress-induced cell surface annexin 2, we used human annexin 2 (ϳ37 kDa) tagged with GFP (ϳ25 kDa) at the carboxyl terminus. HEK293 cells transfected with the annexin 2-GFP fusion construct expressed a protein of the predicted apparent molecular mass (ϳ62 kDa) that cross-reacted with antibodies directed against both annexin 2 and GFP ( Fig. 2A). When we examined the effect of heat stress, we observed increased expression of a 62-kDa, anti-GFP-immunoreactive cell surface protein that was increased on cells incubated at 42°C as compared with those incubated at 37°C (Fig. 2B).
To ascertain whether annexin 2 translocated from an intracellular location to the outer leaflet of the plasma membrane of A, immunoblot blot analysis. EGTA surface eluates or whole cell lysates (30 g) from EGTA-stripped HUVECs maintained at 37°C or 42°C for the indicated intervals were analyzed using mouse monoclonal anti-annexin 2 IgG (Transduction Laboratories) and chemiluminescence exactly as described under "Experimental Procedures." B, cell surface biotinylation. HUVECs incubated at 37°C or 42°C (3 h) were surface-labeled with cell-impermeable sulfo-NHS-SS-biotin (ϩ Biotin) or vehicle (Ϫ Biotin) and then lysed. Biotinylated surface proteins were precipitated with streptavidin-conjugated agarose beads and analyzed by immunoblot analysis using anti-annexin 2 and anti-p11 IgG. To control for loading, the same blot was stripped and reprobed with monoclonal IgG directed against the endothelial cell surface protein thrombomodulin. C, tPA-dependent cell surface plasminogen activation. Control (37°C) or heat-stressed (42°C) cells were incubated with plasminogen, washed, and then treated with either preimmune (PI) or anti-annexin 2 rabbit IgG (IM; 20 g ml Ϫ1 ; 1 h) before incubation with tPA and a fluorogenic plasmin substrate. Data are shown in RFU/ min Ϫ2 , with the control sample arbitrarily set at 1.0 (mean Ϯ S.E.; n ϭ 3). individual cells exposed to temperature stress, we expressed GFP-tagged annexin 2 by nuclear injection of cDNAs in human endothelial cells. After a 37°C incubation (60 min), newly synthesized annexin 2-GFP protein was uniformly distributed throughout the cytoplasm of the injected cells. In cells subsequently incubated at 42°C or 37°C, we detected surface-associated annexin 2-GFP protein on paraformaldehyde-fixed cells, using anti-GFP IgG in an indirect immunofluorescence assay. Total GFP-annexin 2 was detected at a separate wavelength via its intrinsic fluorescence. Cell surface annexin 2-GFP-related immunofluorescence was markedly increased in cells subjected to heat stress as compared with cells maintained at 37°C (Fig. 2C). In addition, when the ratio of cell surface GFP to total annexin 2-GFP was analyzed in three independent experiments, we found that 3.4% of control cells (2 of 59) versus 38% of heat-stressed cells (22 of 58) exhibited at least a doubling in the surface/total annexin 2-GFP ratio over the 3-h incubation period (Fig. 2D).
Cell Surface Translocation of Annexin 2 Is Protein Synthesisand Endoplasmic Reticulum-Golgi-independent-The temperature-induced increase in cell surface annexin 2 expression was enhanced rather than impaired by blockade of protein synthesis (Fig. 3). Cycloheximide treatment, at a dose that inhibited 80% of total protein synthesis, potentiated rather than inhibited annexin 2 translocation in both 37°C-and 42°C-treated cells. Furthermore, Northern blot analysis showed no change in steady-state mRNA levels during heat stress (data not shown). In addition, we tested whether an intact endoplasmic reticulum-Golgi-related pathway might be required for annexin 2 release. Temperature-stressed cells treated with brefeldin A, an agent that disrupts the endoplasmic reticulum-Golgi complex, showed no decrease in their ability to direct annexin

FIG. 3. Heat stress-induced surface expression of annexin 2 is enhanced by inhibition of protein synthesis and blockade of the classical secretion pathway.
HUVECs were incubated for 3 h at 37°C or 42°C in the presence or absence of either the mRNA translation inhibitor cycloheximide (10 g ml Ϫ1 ) or the Golgi disruptor brefeldin A (5 g ml Ϫ1 ). Blots were probed for surface biotinylated annexin 2 and then stripped and reprobed with anti-thrombomodulin IgG as a loading control. cDNA were incubated at 37°C (control) or 42°C (heat stress) for 3 h as described under "Experimental Procedures" and subsequently fixed for surface immunostaining in 2% paraformaldehyde. Images are displayed using equivalent intensity ranges. Note that cells subjected to heat stress show a significant increase in surface-labeled Ann 2-GFP protein as compared with controls. D, translocation of annexin 2-GFP to the outer leaflet of the plasma membrane was analyzed quantitatively by measuring the relative fluorescent intensity of surface-associated Ann 2-GFP (immunostained with anti-GFP antibody) to total annexin 2-GFP (direct GFP fluorescence) under control (37°C) and heat stress (42°C) conditions. Data were compiled from three independent experiments.
2 to the cell surface, even though 73% of total protein secretion was blocked (data not shown). In addition, in microinjected cells, we never detected annexin 2-GFP in a pattern consistent with distribution within the endoplasmic reticulum or Golgi (data not shown). These data suggest that temperature stress induces translocation of annexin 2 from the cytoplasm to the cell surface in a protein synthesis-independent and endoplasmic reticulum-Golgi-independent manner.
Cell Surface Translocation of Annexin 2 Requires p11 Expression-p11, a member of the S100 family of calcium-binding proteins, is an annexin 2 ligand that forms a heterotetramer with two annexin 2 monomer subunits in response to changes in intracellular calcium concentration (3). On complex formation with annexin 2 monomer, p11 enhances the phospholipid binding affinity of annexin 2 by directing it to anionic phospholipids on the inner leaflet of the plasma membrane. To study the contribution of p11 to annexin 2 translocation, we stably suppressed p11 protein expression in a transformed mouse endothelial cell line by RNA interference (20) (Fig. 4). We analyzed the effect of small interfering RNA (siRNA) on the p11 mRNA levels by performing reverse transcription-PCR. p11 mRNA levels were found to be reduced by ϳ84% in pSU-PER/p11siRNA cells as compared with the pSUPER cells or the untransfected parental cells. Furthermore, the annexin 2 and glyceraldehyde-3-phosphate dehydrogenase mRNA levels re-mained unchanged, indicating that the siRNA approach was specific and that reduction of p11 mRNA levels did not affect the mRNA levels of annexin 2 (Fig. 4A). In this system, cells stably transfected with the p11 RNA silencing vector (siRNA) showed persistent, ϳ90% suppression of p11 protein expression, whereas annexin 2 protein expression was unchanged (Fig. 4B). In cells transfected with the empty vector, cell surface annexin 2 as well as p11 increased by 2.4-fold in response to temperature stimulation, whereas surface expression of annexin 2 and p11 was markedly reduced at both 37°C and 42°C in cells expressing p11 siRNA (Fig. 4C). These data demonstrate that p11 expression is essential for cell surface translocation of annexin 2.
Heat Stress-induced Translocation of Annexin 2 Requires Phosphorylation of Tyrosine 23-Because targeting of annexin 2 to the plasma membrane renders it increasingly susceptible to phosphorylation (3,22) and annexin 2 is a known substrate for both serine/threonine and tyrosine protein kinases (23), we examined the possibility that a phosphorylation event might regulate the cellular trafficking pattern of annexin 2 (Fig. 5). While a serine/threonine phosphatase inhibitor, okadaic acid, had no effect on heat-induced annexin 2 translocation to the cell surface, the tyrosine phosphatase inhibitor sodium orthovanadate enhanced annexin 2 expression levels on the cell surface by a mean of 2.8 Ϯ 0.5-fold (S.E.; n ϭ 3) (Fig. 5A). Under the same conditions, no change in cell surface thrombomodulin was observed. Orthovanadate augmented the increase in annexin 2 expression seen with heat stress by an additional 1-2-fold. Annexin 2 expression levels were accompanied by commensurate increases in tPA-dependent plasmin generation (Fig. 5B). These data suggested that translocation of functional annexin 2 involves a tyrosine phosphorylation event. To determine whether tyrosine phosphorylation of annexin 2 itself was required for annexin 2 translocation to the cell surface, several experiments were conducted (Fig. 6). We noted, first, that the heat stress-induced increase in cell surface expression of annexin 2 was completely blocked by treatment of cells with either genistein, a general tyrosine kinase inhibitor, or herbimycin A, a pp60-c-src-specific inhibitor (Fig. 6A). At the same time, no change in cell surface thrombomodulin expression was detected. Furthermore, the accompanying enhancement in tPA-dependent fibrinolytic potential was also completely blocked by these agents (Fig.  6B). In addition, PP2, a potent and selective inhibitor of Src kinases, completely blocked the temperature-induced translocation of annexin 2 to the cell surface, whereas an inactive analog, PP3, had no effect (Fig. 6C) (24). Finally, immunoblot blot analysis of post-heat stress anti-phosphotyrosine-precip-itated proteins revealed a 2.5 Ϯ 0.2-fold (mean Ϯ S.E.; n ϭ 3) increase in tyrosine-phosphorylated annexin 2 that was completely abrogated by treatment with either genistein or herbimycin A, when compared with vehicle-treated controls (Fig.  6D). These data indicated that translocation of annexin 2 to the cell surface was completely dependent on the activity of a src-like tyrosine kinase.
To ascertain whether phosphorylation of tyrosine 23, a previously identified target for the pp60-c-src kinase (25), is required for heat stress-induced translocation of annexin 2, we stably transfected HEK293 cells with wild-type and Y23A mutant forms of annexin 2 (Fig. 7A). Whereas cell surface expression of the wild-type myc-tagged protein increased by a mean of 3.3 Ϯ 0.4-fold (S.E.; n ϭ 3) in response to heat stress, we observed no difference in surface expression of myc-tagged mutant annexin 2 in two independent cell lines that expressed the Y23A mutant (Fig. 7A, c), despite the fact that immunoblotting for heat shock protein 70 confirmed a typical heat stress response in these cells (Fig. 7A, a). These data establish that the tyrosine residue at position 23 is essential for heat stress-induced translocation of annexin 2.
Finally, we asked whether tyrosine 23 is the only site of heat stress-induced phosphorylation on annexin 2. Immunoblot analyses of lysates from stably transfected, heat-stressed HEK293 cells expressing myc-tagged wild-type or Y23A mutant annexin 2 were carried out. The results revealed a robust increase in the tyrosine phosphorylation of wild-type annexin 2-expressing HEK293 cells but not Y23A annexin 2-expressing HEK293 cells (Fig. 7B, a). These data suggest that tyrosine 23 is uniquely susceptible to heat-induced phosphorylation.

FIG. 6. Annexin 2 is phosphorylated in response to heat stress.
A, effect of tyrosine kinase inhibitors on annexin 2 surface expression. HUVECs were incubated for 3 h at 37°C or 42°C with or without genistein (100 M) or herbimycin A (1 M). Surface-expressed proteins were analyzed as described in the Fig. 1B legend. B, plasmin generation. The rate of tPA-dependent plasminogen activation was determined as described in the Fig. 1C legend for cells treated as described in A. Values shown are the mean Ϯ S.E. (n ϭ 3). C, effect of pp60 (c-src) inhibition on annexin 2 cell surface expression. HUVECs were incubated for 3 h at 37°C or 42°C with or without 10 M each of PP3 or PP2. Surface EGTA eluates or total cell lysates were analyzed by using anti-annexin 2 IgG. D, effect of tyrosine kinase inhibitors on heat stress-induced phosphorylation of annexin 2. HUVECs were prepared as described in A. Whole cell lysates (1 mg) were immunoprecipitated using anti-phosphotyrosine IgG and blotted using anti-annexin 2 IgG. Identical lysate samples were blotted directly with monoclonal antiannexin 2 IgG to estimate total annexin 2 levels. PY-annexin 2 signifies tyrosine phosphorylated annexin 2.

FIG. 7.
A, heat stress-induced surface translocation of annexin 2 requires tyrosine at position 23. a, total lysates (7 g) from HEK293 cells, stably transfected with myc-tagged wild-type (WT) or mutant (Y23A) annexin 2 and maintained at 37°C or 42°C for 4 h, were immunoblotted with anti-heat shock protein 70 (hsp 70) IgG. b, immunoblotting of total lysates with anti-glyceraldehyde phosphate dehydrogenase (GAPDH) was used as a loading control. c, surface EGTA eluates from the same samples were immunoblotted with anti-myc IgG. B, annexin 2 is phosphorylated on tyrosine at position 23. Myc-tagged annexin 2 was subjected to immunoprecipitation from wild-type-or Y23A mutant-expressing HEK293 cells, using anti-myc IgG, identical to conditions presented in A. The blot was probed with anti-phosphotyrosine IgG (a) and stripped and reprobed with anti-myc IgG to ascertain equal amounts of immunoprecipitated myc-tagged annexin 2 (b). PY-Annexin 2/myc signifies tyrosine-phosphorylated myc-tagged annexin 2.

Systemic Heat Stress Induces Translocation of Annexin 2 in
Vivo-In the intact organism, endothelial cells are subject to a wide variety of stimuli and stresses. To examine the dynamics of annexin 2 endothelial cell surface expression in vivo, we subjected mice to thermal stress (40.8 Ϯ 0.2°C, 45 min) and then isolated a pool of biotinylated aortic endothelial cell surface proteins by precipitation with immobilized streptavidin (Fig. 8). Immunoblot analysis revealed a 7.4 Ϯ 2.7-fold (mean Ϯ S.E.; n ϭ 3) increase in cell surface annexin 2 (Fig. 8A). The stress-induced increase in cell surface annexin 2 was completely inhibited on pretreatment of animals with herbimycin A (200 g kg Ϫ1 , intraperitoneally) 2 h before the application of temperature stress (Fig. 8A). Identical lysate samples were blotted directly with monoclonal anti-annexin 2 IgG to estimate total annexin 2 levels (Fig. 8B). These data reveal that temperature stress in vivo induces tyrosine kinase-related, tyrosine phosphorylation-dependent translocation of annexin 2 to the endothelial cell surface. DISCUSSION Our data demonstrate for the first time that endothelial cell annexin 2 can be translocated from the cytoplasm to the cell surface in response to temperature stress (Fig. 1, A and B; Fig.  2, C and D). We show that this response is independent of the classical endoplasmic reticulum-Golgi pathway and that it does not depend on de novo protein synthesis (Fig. 3). In addition, translocation of annexin 2 is accompanied by a concomitant increase in fibrinolytic potential (Fig. 1C). Temperature-induced annexin 2 translocation is not observed on depletion of intracellular S100A10/p11 protein in response to RNA silencing (Fig. 4). On the other hand, the response is augmented on inhibition of tyrosine phosphatase activity (Fig. 5), and inhibited on blockade of Src-like tyrosine kinase activity (Fig. 6, AϪC). We show, furthermore, that annexin 2 itself is phosphorylated in response to temperature stress (Fig. 6D) and that neither translocation nor phosphorylation occurs when the pp60-c-src target site of annexin 2, Tyr 23 , is mutated (Fig. 7). Finally, we demonstrate that brief in vivo temperature stress is accompanied by a large increase in annexin 2 cell surface expression. As with the in vitro response, this translocation event can be inhibited by pretreatment of the experimental animal with the tyrosine kinase inhibitor herbimycin A (Fig. 8). Together, these data suggest a novel pathway for release of annexin 2 in response to moderate levels of cellular stress.
In mammals, nonclassical secretion pathways have been proposed for more than 20 polypeptides, including fibroblast growth factor (FGF)-1, interleukin 1␤, interleukin 18, platelet-derived growth factor, thioredoxin, mammary-derived growth inhibitor, transglutaminase, ␤-thymosin, lectin L-25, and lectin L-14-1 (26). These structurally and functionally diverse "leaderless" proteins appear to exit the cell by routes that are resistant to Golgi-disrupting agents such as brefeldin A or monensin. Interleukin 1␤, for example, translocates from the cytoplasm against a pH gradient into specialized endolysosomal vesicles that undergo ATP-dependent fusion with the plasma membrane (27). Their release is stimulated by chloroquine, which increases intralysosomal pH, and blocked by nocodazole, which disassembles microtubules (28). FGF-1, on the other hand, employs a nonvesicular, transcription-and translation-dependent mechanism that requires a free cysteine residue and the phosphatidylserine binding domain of FGF-1. FGF-1 appears to complex with both S100A13, a relative of p11, and the p40 extravesicular segment of synaptotagmin (29). Heat stress-induced release of FGF-1 is blocked by the drug amlexanox, which binds S100A13 (30).
Our data suggest a critical role for p11/S100A10 in the translocation of annexin 2. S100 family members appear to represent general effectors for plasma membrane localization of their binding partners. For example, p11 (S100A10) appears to direct the tetrodotoxin-resistant sodium channel Na v 1.8/SNS to the plasma membrane in transfected Chinese hamster ovary cells (31). p11 binds to the carboxyl terminus of the TASK-1 potassium channel, releasing it to the plasma membrane, possibly by masking an endoplasmic reticulum retention signal (32). p11 may also orchestrate the nonlytic egress of the assembled blue tongue virus from infected cells (33). Data presented here suggest that p11 may direct annexin 2 to the inner face of the plasma membrane, where it is phosphorylated by membrane-associated pp60-c-src; p11 then appears to translocate simultaneously with phospho-annexin 2 to the cell surface. Interestingly, both S100A8 and S100A9 are released by protein kinase C-activated monocytes (34) or by activin-overexpressing keratinocytes (35,36), respectively, suggesting that nonclassical translocation may be a general function of S100 protein family members.
The src family kinases undergo activation in response to a wide range of stimuli (37). Biochemical agents include growth factors, such as platelet-derived growth factor, that engage tyrosine kinase receptors; thrombin and endothelin-1, which bind G-protein-coupled receptors; and interleukin-2, which ligates a cytokine receptor. In addition, src kinases can be activated in response to oxidative or radiation stress and by increases in intracellular calcium. A wide range of src substrates contribute to cytoskeletal rearrangement, cell-cell adhesion, cell-substrate adhesion, caveolar structure and function, mitogenic signaling, and RNA processing. Thus, translocation of annexin 2 may take place within the context of a broad program of cellular adaptation.
Brief temperature stress induces alterations in cellular metabolism that include enhanced tyrosine kinase activity, increased intracellular calcium, and stress protein synthesis. Heat stress (38), like hypoxia (39), shear stress (40), and hyperosmolarity (41), is associated with rapid activation of pp60c-src, which can form a stable, herbimycin A-sensitive complex with Hsp90 (42)(43)(44). In the endothelial cell, an abrupt increase in intracellular calcium (45) would be predicted to stimulate association of annexin 2 with p11, as well as mobilization of annexin 2 to the plasma membrane (3), where pp60-c-src phosphorylation could occur (22). Phospho-annexin 2 would couple more avidly with inner leaflet anionic phosphatidylserine (46) and could undergo a conformational change that might potentiate its insertion into the plasma membrane. Under mildly acidic in vitro conditions, the hydrophobicity of annexins 5 and 12 increases dramatically, and annexin 12 has been observed to generate membrane-spanning helices that create a transmem- FIG. 8. Temperature stress increases endothelial cell surface expression of annexin 2 in vivo. Mice were pretreated either with vehicle (0.2% Me 2 SO, intraperitoneally) or a pp60-c-src inhibitor (herbimycin A, 200 g kg Ϫ1 ) 2 h before being subjected to treatment at either 24°C or 41°C for 45 min. A, after perfusion of anesthetized animals with sulfo-NHS-SS-biotin, cell surface biotinylated proteins were precipitated from thoracic aortic homogenates using agarose-immobilized streptavidin. B, homogenates were analyzed for the presence of total annexin 2. Immunoblot blot analyses were conducted using monoclonal anti-annexin 2 IgG. brane pore-like structure (47)(48)(49). A similar "molten globule" mechanism has been postulated for cellular release of FGF-1 (50). Thus, phosphorylation of annexin 2 might induce its penetration into the lipid bilayer and, possibly, its extrusion through the lipid bilayer.
Stress responses in vascular cells may play a pivotal role in the homeostatic response to vascular injury because stress protein expression has been observed in response to ischemia (51), oxidized low density lipoprotein (52), other oxidants (53), and selected cytokines (54). Putative heat shock proteins, such as an hsp 65 cross-reactive antigen, have been observed in heat-treated or tumor necrosis factor ␣-treated endothelial cells (54) and in endothelial cells derived from atherosclerotic lesions (55). Inflammatory cytokines could trigger a similar response in vasculitic arteries. Under such circumstances, augmented expression of annexin 2 and increased surface generation of plasmin could serve adaptively to restore thromboresistance to the injured blood vessel.
A seemingly modest up-regulation in cell surface annexin 2 may have unexpectedly profound implications for health and disease. We previously demonstrated that an ϳ2-3-fold increase in cell surface expression of annexin 2 was associated with a potentially life-threatening bleeding disorder in patients with acute promyelocytic leukemia, possibly due to uncontrolled production of plasmin and depletion of its circulating inhibitor, ␣ 2 -antiplasmin (11). In addition, annexin 2-mediated plasmin activity on endocardial endothelial cells regulates release of active transforming growth factor-␤3 during epithelialmesenchymal transformation during the remodeling of the embryonic chick atrioventricular canal to form heart valves and septae (56). Annexin 2-mediated plasmin activity on monocyte/ macrophage cell surface modulates matrix remodeling and directed migration of these cells (57). Annexin 2-null mice, finally, accumulate intravascular fibrin, are unable to fully eliminate injury-induced arterial thrombi, and show impaired activation of metalloproteinase-9 and -13 (12). Thus, activation of the stimulus-induced annexin 2 phosphorylation switch could modulate not only blood fluidity but also extravascular matrix remodeling and the invasion potential of multiple cell types.