Up-regulation of the KLF2 Transcription Factor by Fluid Shear Stress Requires Nucleolin*

We have previously characterized the regulation of the KLF2 transcription factor gene by describing an induction complex that binds to and regulates its promoter. In the present study, by using DNA affinity chromatography and mass spectrometry, we have identified nucleolin as an additional protein that binds to a palindromic response region in the KLF2 promoter. The presence of nucleolin on the KLF2 promoter in macrophages was verified by electrophoretic mobility shift assays. Interestingly, in mouse and human endothelial cell lines, electrophoretic mobility shift assays and chromatin immunoprecipitation analyses indicated that nucleolin binds the KLF2 promoter only upon application of fluid shear stress. Pretreatment of the endothelial cells with LY294002, a specific inhibitor of phosphatidylinositol 3-kinase (PI3K), blocked the shear stress-induced binding of nucleolin to the promoter, demonstrating its PI3K-dependent regulation. Additionally, nucleolin exhibited dynamic flow-specific, PI3K-dependent alterations in size. Anti-nucleolin antibodies interacted with a 110-kDa form in static endothelial cells and with several catalytic forms that changed in abundance after the application of shear stress. Immunoprecipitation experiments demonstrated that fluid flow induced the interaction of nucleolin with the p85 regulatory subunit of PI3K. Finally, introduction of small interfering RNAs targeting the nucleolin genetic sequence selectively reduced nucleolin expression and was sufficient to block the induction of KLF2 by shear stress. These data support a general role for nucleolin in gene regulation and identify it as a novel factor involved in regulation of KLF2 expression.

Since its discovery, KLF2 (Krüppel-like factor 2) (also known as lung Krüppel-like factor) has proven vital to developmental and cellular functions in several distinct tissue types. The gene is highly conserved between mouse and human homologues and is expressed in similar tissues (lung, heart, skeletal muscle, pancreas, and placenta) in both animals (1). The KLF2 protein is a C2/H2 zinc finger DNA-binding protein of 354 amino acids with an approximate molecular mass of 38 kDa. Studies have demonstrated that KLF2 plays critical functional roles in adipocytes, erythrocytes, and T lymphocytes. Consistent with the role of Krüppel-like factors in late stage differentiation, KLF2 prevents the differentiation of pre-adipocytes in mature adipocytes (2,3). It is also required for primitive erythropoiesis through the regulation of embryonic ␤-like globin genes (4). In T cells, KLF2 has been shown to be an antiapoptotic and antiproliferative factor capable of maintaining T cell quiescence (5,6). Several tissue-specific downstream target genes of KLF2 have been identified, including peroxisome proliferator-activated receptor-␥ in adipocytes (2) and p21 and interleukin-2 in T cells (6,7).
The function and regulation of KLF2 in the vasculature has recently been the subject of intense research. During development, KLF2 is involved in an endothelium-mediated transcriptional pathway necessary for vessel stabilization (8,9). Interestingly, in vitro and in vivo experiments with vascular endothelial cells indicate that KLF2 exhibits sustained induction under laminar shear stress (10). This is especially important in light of the fact that shear stress is protective against the development of certain vascular diseases, particularly atherosclerosis (11)(12)(13). KLF2 expression is found throughout the aorta, except at areas of decreased shear stress (branches and bifurcations) (14), which are more prone to develop atherosclerotic lesions (15,16). KLF2 may, therefore, function as a flow-inducible athero-protective factor in the endothelium. This conclusion is supported by in vitro studies demonstrating that KLF2 increases endothelial nitric-oxide synthase expression (14,17,18), is antithrombotic (19), decreases vascular cell adhesion molecule-1 expression, and inhibits the effects of proinflammatory cytokines (17). Furthermore, KLF2 is down-regulated by atherosclerotic inflammatory cytokines like tumor necrosis factor ␣ (20) and contributes to some of the beneficial effects of 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors (statins) upon the vasculature (21,22). The cumulative evidence indicates that the regulatory mechanisms of the KLF2 gene represent a potential therapeutic target for the prevention or treatment of atherosclerosis.
The regulation of the KLF2 gene is complex. Several transacting factors, chromatin modifications, and three signaling pathways have been implicated to date. These processes are largely focused on a 30-bp region of the proximal promoter located between Ϫ138 and Ϫ111 nucleotides from the start site of transcription (23). This region, which is evolutionarily conserved between mouse and human promoters, is critical for KLF2 expression in every cell type tested thus far. In endothelial cells, it mediates the induction of KLF2 expression by fluid shear stress through the binding of flow-influenced nuclear factors (24). We have recently demonstrated that p300/CREB-binding protein-associated factor (PCAF) 3 and heterogeneous nuclear ribonucleoprotein D (hnRNP-D) bind this region specifically upon the application of fluid shear stress to a mouse endothelial cell line (18). The binding of these factors is dependent upon the activity of the phosphatidylinositol 3-kinase (PI3K) pathway and is correlated with histone H3 and H4 acetylation to facilitate chromatin remodeling. Similarly, tumor necrosis fac-* This work was supported by National Institutes of Health Grant R01 HL57281 and by American Heart Association Predoctoral Fellowship 0415180B. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 These authors contributed equally to this work. 2  tor-␣ has been found to repress KLF2 expression in endothelial cells by activating histone deacetylase 4, which blocks binding of the myocyte enhancement factor 2 transcription factor to this region (20). Myocyte enhancement factor 2 has also been found to bind the Ϫ138/Ϫ111 promoter region in T cells (25). We have also demonstrated that this same region binds several other factors to induce KLF2 expression in macrophages, including hnRNP-D, hnRNP-U, PCAF, and p300 (26). The full complement of trans-acting factors is not known, however; nor is the complete mechanism of activation fully understood.
In the present study, we identify nucleolin as an additional factor that binds to the Ϫ138/Ϫ111 region of the KLF2 promoter. Nucleolin is a multifunctional phosphoprotein with diverse roles in both the nucleus and the cytoplasm. Several studies have shown that nucleolin interacts with transcription factor complexes, including LR1 and E47 (27,28). It also possesses helicase and chromatin remodeling activities, which could facilitate transcription (29,30). Here, we demonstrate that nucleolin binds the KLF2 promoter in macrophages and endothelial cells. Additionally, we have found that the PI3K pathway is required for nucleolin binding in endothelial cells and that nucleolin is required for induction of the KLF2 gene by fluid shear stress.
Preparation of Nuclear Extracts-Nuclear extracts were prepared as described (31). Protein concentration was determined by BCA assay (Bio-Rad), and the extract was stored at Ϫ80°C.
DNA Affinity Chromatography and Mass Spectrometry-DNA affinity chromatography and mass spectrometry were performed as described previously (26) with the exception that complimentary strands of the 62-bp region of the KLF2 promoter (nucleotides Ϫ157 to Ϫ95) were annealed and cross-linked to Sepharose beads. Following washing and incubation in DNA binding buffer, 10 mg of nuclear extract was incubated with the beads for 30 min on ice. The beads were washed 10 times with 10 ml of ice-cold binding buffer, and the bound proteins were eluted in a buffer consisting of 20 mM acetate buffer (pH 5.0), 0.5% Nonidet P-40, 2.5 M KCl, 10% glycerol, and 1 mM dithiothreitol. The affinity-eluted fraction was analyzed by mass spectrometry without separation on polyacrylamide gels.
Electrophoretic Mobility Shift Assays (EMSAs) and Chromatin Immunoprecipitation Assays (ChIPs)-EMSAs were performed as described previously (26), except that KCl concentration was reduced to 75 mM. EMSA nucleolin supershifts with EOMA nuclear extract were standardized using antibody provided by France Carrier (Baltimore, MD). 5 g of nuclear extract was used with the antibody. For EMSAs with HUVEC nuclear extract, 15 g of extract was used, and supershifts were performed with mouse monoclonal anti-nucleolin antibody (clone 4i51; Abcam). ChIP assays were performed with anti-nucleolin polyclonal antibody (Novus Biologicals, Littleton, CO) as follows. DNA/protein interactions in static or shear stress cells were fixed by formaldehyde cross-linking (1%) for 10 min with continuous stirring. The cross-linking reaction was stopped by the addition of 0.125 M glycine. The cells were collected by centrifugation, washed with ice-cold PBS (with protein inhibitor mixture), and placed in 5 ml of Buffer A (100 mM Hepes, pH 7.9, 1.5 mM MgCl 2 , 50 mM KCl, 1 M aprotinin, 1 M phenylmethylsulfonyl fluoride, 1 M leupeptin) for 10 min on ice. Next, the cells were lysed by glass pestle, the nuclei were collected by centrifugation, and the cross-linked chromatin was sonicated by four 15-s pulses with a 550 Sonic Dismembrator (setting of 3). The cross-linked soluble chromatin was collected by centrifugation and diluted in radioimmune precipitation buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.05% Nonidet P-40, 1 mM EDTA, and protein inhibitor mixture). Chromatin was blocked with 20 l of salmon sperm DNA, bovine serum albumin, poly(dI-dC), and protein A/G-Sepharose beads for 4 h at 4°C. Immunoprecipitations were done with the specific antibody overnight at 4°C. The DNA-antibody complexes were pulled-down with protein A/G beads (1 h at 4°C). The beads were washed three times with 1.5 ml of binding buffer, followed by three washes each with low salt buffer (50 mM Tris-HCl, pH 8.0, 0.5% Triton X-100, 150 mM NaCl, 1 mM EDTA, 0.1% deoxycholate) and high salt buffer (50 mM Hepes, pH 7.5, 250 mM LiCl, 1 mM EDTA, 1% Nonidet P-40, and 0.7% deoxycholate). Bound chromatin was eluted with 75 l of elution buffer (50 mM sodium bicarbonate, 0.5% SDS; 15 min with continuous stirring). Cross-linking was reversed by heating at 65°C for 4 h in the presence of 0.3 M NaCl. The mixture was treated with Proteinase K, and the DNA fragment was purified by Qiagen columns.
The purified DNA was amplified by radioactive PCR using the following primer pairs to amplify a 139-bp segment encompassing the KLF2 promoter shear stress response region: forward, GGCTTGAG-GAGCGCAGTCCGGGCTCCCGCA; reverse, CCGGGCTAGGAG-GCGTCGACGGAAACGCGT.
Fluid Shear Stress Experiments-A CELLMAX QUAD artificial capillary cell culture system (Spectrum Laboratories, Rancho Dominguez, CA) was used as described previously (24). EOMA cells and HUVECs were exposed to shear stress at 19 dynes/cm 2 . For flow experiments with HUVECs, the artificial capillaries were coated in 10 g/ml human fibronectin (Sigma) before introduction of cells. For experiments with the PI3K signal transduction inhibitor LY294002 (Calbiochem), the compound was dissolved in Me 2 SO 4 and added to the flow system at a concentration of 40 mol/liter, accounting for the total volume of liquid in the flow module. Cells treated with an equal volume of Me 2 SO 4 were used as a negative control (vehicle alone). Immediately following the addition of LY294002, the medium was allowed to circulate over the cells at a shear stress level of Ͻ1 dyne/cm 2 for 30 min before increasing the level of shear stress to 19 dynes/cm 2 .
Immunoprecipitation and Immunoblotting-For immunoprecipitation of HA-tagged nucleolin, HUVECs were transiently transfected with the expression vector (kindly provided by Leslyn Hanakahi (Baltimore, MD)) using Fugene 6 (Roche Applied Science). All cells were isolated in nonionic cells lysis buffer consisting of 20 mM Tris (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 0.5% Nonidet P-40, 2.5 mM sodium pyrophosphate, 1 mM ␤-glycerolphosphate, 1 mM Na 3 VO 4 , and protease inhibitor mixture (Sigma). Cells were also sonicated three times for 5 s each. Immunoprecipitations were performed in cell lysis buffer with 250 -500 g of cell extract incubated for 4 h with rabbit anti-hnRNP-D (AUF1) antibody (Upstate Biotechnology, Inc.), mouse anti-nucleolin monoclonal antibody (Clone 4E2; Stressgen Bioreagents, Victoria, Canada), rabbit anti-hemagglutinin probe (Y-11; Santa Cruz Biotechnology, Inc., Santa Cruz, CA), or rabbit anti-p85 polyclonal antibody (Cell Signaling Technology, Inc., Beverly, MA). Lysates were precleared with 50 l of protein A/G-agarose beads prior to the addition of primary precipitating antibody. Antibody complexes were precipitated by the addi-tion of protein A/G-agarose beads (Santa Cruz Biotechnology) for 3 h to overnight. Precipitated complexes were washed for 5 min in high stringency buffer (0.1% SDS, 1% deoxycholate, 0.5% Triton X-100, 20 mM Tris-HCl (pH 7.5), 120 mM NaCl, 25 mM KCl, 5 mM EDTA, 5 mM EGTA, 0.1 mM dithiothreitol), followed by a 10-min wash in high salt wash buffer (high stringency buffer plus 1 M NaCl), followed by a 10-min wash in low salt wash buffer (2 mM EDTA, 10 mM Tris-HCl (pH 7.5), 0.5 mM dithiothreitol). 40 l of 1ϫ Laemmli sample buffer was added to the pellet, followed by vortexing and heating to 100°C for 5 min. Supernatant was collected and run on a 10% polyacrylamide gel for immunoblotting. The immunoblotting procedure has been described previously (18). An additional mouse monoclonal anti-nucleolin antibody (clone 4i51; Abcam, Cambridge, UK) was used when indicated. GAPDH was used as a loading control by stripping and reprobing blots with mouse anti-GAPDH antibody (Advance ImmunoChemical, Long Beach, CA.).
Small Interfering RNA (siRNA)-mediated Gene Knockdown-A pool of four human siRNA duplexes targeting human nucleolin was purchased from Dharmacon, Inc. (Lafayette, CO). HUVECs at ϳ70% confluence were transiently transfected with 100 nM siRNA using Lipofectamine 2000 (Invitrogen), according to the manufacturer's instructions. Functional nontargeting siRNA-negative control and lamin A/C siRNA-positive control were also purchased from Dharmacon. Cells were exposed to fluid flow (12 h at 19 dynes/cm 2 ) 24 h after transfection. The percentage reduction in nucleolin protein levels (compared with functional nontargeting siRNA-transfected controls) was determined by densitometric analyses of immunoblots (normalized to GAPDH levels) using ImageQuant 5.0. Anti-lamin A/C antibody for probing immunoblots for siRNA-positive control was purchased from Cell Signaling Technology (Beverly, MA).

RESULTS
Nucleolin Binds to the KLF2 Promoter-Our laboratory has previously characterized the cis-acting element and binding proteins of the KLF2 gene. To further define the mechanism by which the KLF2 gene is regulated, we used a DNA affinity chromatography procedure similar to that used previously (26), except that a slightly larger region of the KLF2 promoter was cross-linked to the beads (62 bp between Ϫ157 and Ϫ95 nucleotides from the start site of transcription). This was done because our previous work had shown that, although trans-acting factors bind in the 30-nucleotide core element between Ϫ138 and Ϫ111 bp, a slightly larger region encompassing adjacent nucleotides is necessary for stabilization of the activating complex and full activation of the promoter (18,24). Nuclear extract from the mouse WR19M.1 macrophage cell line was incubated with the cross-linked beads, followed by extensive washing. The WR19M.1 cell line was used in these initial experiments, because it has a high level of KLF2 expression and is an excellent source for the purification of factors that bind the KLF2 promoter (26). A high salt/detergent elution buffer was required to elute the bound protein factors.
Washing fractions from the column were retained and used along with negative (no nuclear extract; lane 1) and positive (input nuclear extract; lane 2) controls in EMSAs to verify the retention of nucleotide binding in the eluted fraction (Fig. 1). No binding was present in the washes or in the fraction eluted with high salt buffer (2.5 M KCl; lanes [3][4][5][6][7][8]. 0.5% Nonidet P-40 was added to the buffer to effectively elute the proteins from the column, which retained binding activity (lane 9). The fact that detergent was required suggested that the proteins bind the KLF2 promoter with high affinity. Furthermore, two separate DNA-binding complexes were eluted from the column (Bound 1 and Bound 2).
Eluted fractions were subsequently pooled, dialyzed, and concentrated. The concentrated samples were subjected to mass spectrometry by liquid secondary ionization-MS-MS at the University of Cincinnati Mass Spectrometry core facility (available on the World Wide Web at www.chembus.uc.edu/ massnew/maintbl.asp).
Several proteins were detected based on these criteria, one of which was chosen for analysis in this study. The rest of the identified proteins are under investigation. One list of observed mass fingerprints (Table 1) fitted to the predicted mass fingerprint of nucleolin. To confirm that nucleolin is present in the KLF2 promoter trans-activating complex,

TABLE 1
Mass spectrometry analysis of nucleolin MS spectra were obtained from the eluted fraction of the DNA affinity chromatography following desalting. Peptides were determined by peptide mass fingerprint analysis of relevant mass peaks. Those found to correspond with a high degree of certainty to mouse nucleolin (SwissProt database accession number P09405) are listed. EMSAs were performed with WR19M.1 nuclear extract, Ϫ138/Ϫ108 radiolabled probe, and anti-nucleolin polyclonal antibody ( Fig. 2A). As noted in Fig. 1, the EMSA revealed two separate DNA-binding complexes (lane 2). Competition with unlabeled Ϫ138/Ϫ108 probe depleted binding of both complexes, demonstrating specific DNA-binding proteins within this region (lane 3). The addition of the anti-nucleolin antibody immunodepleted the binding complexes seen in the positive control, suggesting that the presence of nucleolin in both complexes is essential for the complex to bind (lane 4). Immunodepletion was also noted previously when EMSAs were conducted with anti-hnRNP-D antibody (18,26). Rabbit IgG did not effect complex formation (lane 5). The binding of nucleolin in WR19M.1 macrophages was verified by ChIP assay using anti-nucleolin polyclonal antibody and primers to amplify across the Ϫ138/Ϫ108 KLF2 promoter region by polymerase chain reaction (data not shown). EMSAs were also performed with nuclear extract from the mouse endothelial EOMA cell line using a probe of the shear stress response region (Ϫ138/Ϫ108) of the KLF2 promoter (Fig. 2B). Extract was used from cells exposed to 24 h of pulsatile fluid shear stress at 19 dynes/cm 2 and from cells kept under static conditions. As with the EMSAs using macrophage cell extract, two DNA binding complexes were noted under both static conditions (lane 2). Interestingly, an additional DNA-binding complex appeared under shear conditions (Shear; lane 2; arrow). Competition with unlabeled probe depleted binding of all the complexes, demonstrating specificity (lanes 3). Under static conditions, the addition of an anti-nucleolin antibody slightly immunodepleted the faster migrating complex (Static; lane 4). This suggested that, in EOMA cells, this particular anti-nucleolin antibody was reacting with a nonspecific, constitutively present, or non-shear stress-specific, component of the faster migrating complex.

Peptide Observed mass Predicted mass Position in nucleolin
Importantly, under flow conditions, the addition of anti-nucleolin antibody immunodepleted (designated by an asterisk) the shear stressspecific binding (arrow) seen in the positive (no antibody) control (Shear; lane 4). This demonstrated the presence of nucleolin in the flow-specific DNA binding complex. The specificity of nucleolin antibody supershift was checked by using rabbit IgG, which did not produce any effect on the EMSA binding pattern (lane 5). Additionally, an immunoblot of EOMA cell extract showed a slight decrease in nucleolin levels at the standard molecular mass of 110 kDa after 6 h of flow, exactly the point at which shear stress increases KLF2 expression in the EOMA cell line (24). After 12 h of fluid shear stress, the level of nucleolin returned to static levels (Fig. 2C).
In order to determine in which DNA binding complex nucleolin resides and to verify the shear stress-specific DNA binding activity of nucleolin, EMSAs were performed with HUVEC nuclear extract under static and shear conditions. Although the pattern of binding was slightly different from that of EOMA cells, the application of flow greatly enhanced the presence of two DNA binding complexes (Fig. 2D,  arrows). The addition of an anti-nucleolin monoclonal antibody immunodepleted the top, slowly migrating complex under flow conditions (lane 8). The same antibody had no effect on the faster migrating complex or the basal level of binding found under static conditions (lanes 4  and 8). Therefore, nucleolin is a component of the slower migrating, DNA-specific complex.
KLF2 expression is greatly up-regulated (ϳ15-fold) when fluid shear stress is applied to this cell line (24). Therefore, the EMSA data demonstrated that nucleolin may function as a flow-specific factor to induce KLF2 expression in endothelial cells. Ectopic expression of human nucleolin via transient transfection of an HA-tagged nucleolin expression vector (HA-nucleolin) into HUVEC endothelial cells failed to transactivate the KLF2 promoter under static conditions or to superactivate the promoter under flow conditions (data not shown). This finding was identical to the lack of promoter activation previously found with ectopic hnRNP-D transfection (18). One possible reason that ectopic expression of nucleolin and hnRNP-D failed to trans-activate the KLF2 promoter is that, given their abundance within the cell and their multitude of regulated functions, their DNA binding activities represent only a small fraction of the total nucleolin and hnRNP D within a cell. Therefore, their DNA binding levels may not be affected by overexpression (32).

Flow-inducible Binding of Nucleolin to the KLF2 Promoter in Endothelial Cells is PI3K-dependent-
We have reported that the activity of the PI3K signal transduction pathway is required for binding of shear stress induction factors to the KLF2 promoter (18). Therefore, we sought to determine if the flow-specific binding of nucleolin to the KLF2 promoter was also PI3K-dependent. EMSAs were repeated with nuclear extract from EOMA cells exposed to fluid shear stress (as above) in the presence of the specific PI3K inhibitor LY294002 or Me 2 SO (vehicle) control. As reported previously, Me 2 SO did not affect formation of the shear stress complex, but LY294002 treatment greatly attenuated nuclear binding (Fig. 3A). To verify the flow-specific binding of nucleolin on the KLF2 promoter and to assess the effect of PI3K inhibition on said binding, standard ChIP assays were performed with a different anti-nucleolin polyclonal antibody than used in the EMSA experiments. PCR primers were used to amplify across the shear stress response region of the KLF2 promoter. In EOMA cells, nucleolin binding was not detectable under static conditions or when flow was applied in the presence of the PI3K inhibitor LY294002 (Fig. 3B). Nucleolin binding was readily observable, however, under flow conditions, with and without vehicle control (Me 2 SO). To account for any species, cell line, or endothelial cell heterogeneity, the experiment was repeated with nuclear extract from primary human HUVECs. As found in the EOMA cell line, nucleolin bound the KLF2 promoter only upon application of fluid stress in the presence of a functional PI3K pathway (Fig. 3C). The ChIP assays confirm that nucleolin binds the endothelial KLF2 promoter in a shear stress-specific, PI3K-dependent manner.
Fluid Flow Results in Catalytic Forms of Nucleolin-Having established that the binding of nucleolin to the KLF2 is flow-specific, we investigated the post-translation regulation of nucleolin in response to shear stress. In Western blots of HUVEC extract, we routinely observed that the anti-nucleolin polyclonal antibody immunodetected several species of various molecular sizes, including polypeptide bands of 110, 95, 85, 70, 65, and 50 kDa. Almost all of the forms increased in abun-dance upon the application of the flow, the sole exception being a 70-kDa form that decreased (Fig. 4A). Importantly, the sizes of the catalytic nucleolin forms noted in our experiments correspond to previous reports of multiple nucleolin forms (33)(34)(35)(36)(37). The formation of these catalytic forms was blocked when the cells were pretreated with the PI3K inhibitor LY294002.
To verify the smaller forms as species of nucleolin, an anti-nucleolin monoclonal antibody (clone E42), known to react with multiple catalytic forms, was used (35,38). In response to 12 h of fluid flow, the monoclonal antibody also detected a 95-kDa form that increased and a 70-kDa form that decreased (Fig. 4B). A second anti-nucleolin monoclonal antibody (clone 4i51) detected the 95-and 50-kDa forms specifically in extract from cells subjected to shear stress for 6 h (Fig. 4C). This antibody also detected a decrease in the standard 110 kDa at 6 h of flow, as noted in Fig. 2C. The monoclonal antibodies did not react with the exact same nucleolin species, possibly because those catalytic species lack the particular monoclonal epitope.
To account for the effects of species or endothelial cell heterogeneity on flow-induced nucleolin catalysis, an immunoblot shear stress time course was conducted using mouse EOMA whole cell extract. A similar, although not identical, pattern of catalysis occurred (Fig. 4D). A shear stress-specific polypeptide of ϳ75 kDa formed within 5 min of the onset of shear stress (arrow; lane 2) and persisted throughout the duration of the experiment (24 h; lane 5). An additional nucleolin band of ϳ70 kDa appeared after 24 h of shear stress (asterisk; lane 5). A 50-kDa form was noted under both static and shear conditions.
As with HUVECs, PI3K inhibition via LY294002 treatment blocked formation of all the various species (lane 6), whereas Me 2 SO vehicle control had no effect (lane 7 ). Notably, the mouse EOMA cells produced different sizes of the shear stress-specific forms than the human HUVECs. This is probably a reflection of species and/or endothelial cell type specificity in the exact forms of nucleolin generated in response to shear stress.
Nucleolin Interacts with PI3K and hnRNP-D-Nucleolin has been reported to directly bind the p85 regulatory subunit of PI3K upon activation of the CR2 cell surface receptor on human B lymphocytes (34). Considering the dependence of nucleolin binding upon PI3K activity in endothelial cells, we sought to determine if a nucleolin-PI3K interaction was occurring in our experiments. Immunoprecipitation experiments were performed following transient transfection of HA-nucleolin into HUVECs and subsequent exposure of the cells to 12 h of fluid flow in the presence or absence (vehicle alone) of LY294002. Immunoprecipitation with anti-HA antibody followed by immunoblotting with anti-p85 antibody demonstrated that nucleolin binds the p85 subunit (Fig. 5A). A basal level of binding was present under static conditions (lanes 3 and 4), but an increase in p85 association occurred in response to fluid shear stress (lane 5). Importantly, inhibition of PI3K activity via LY294002 treatment reduced the nucleolin/p85 interaction to basal levels (lane 6). For verification, the experiment was repeated in reverse: immunoprecipitation with anti-p85 antibody followed by immunoblotting with monoclonal anti-nucleolin antibody (Fig. 5B). Under static conditions, a basal level of p85/ nucleolin interaction was again observed (lane 3). This interaction increased under flow conditions (lane 4) but was reduced to a basal level upon inhibition of PI3K via LY294002 treatment (lane 5). This  Protein complexes were precipitated with anti-HA antibody, followed by 10% polyacrylamide gel electrophoresis and immunoblotting (IB) with anti-p85 antibody. Blots were stripped and reprobed with anti-HA antibody as an internal control. B, IP of static or shear-stressed (12 h at 19 dynes/cm 2 ) HUVECs with anti-p85 antibody, followed by immunoblotting with anti-nucleolin monoclonal antibody. Blots were stripped and reprobed with anti-p85 antibody for internal control. C, IP of static or shear stressed HUVECs, in the presence of LY294002 (40 mol/liter) or Me 2 SO, with anti-nucleolin monoclonal (clone 4E2) antibody followed by immunoblotting with anti-hnRNP-D antibody. Blots were stripped and reprobed with anti-nucleolin antibody. D, IP of HUVECs (under static, shear-stressed, or shear-stressed plus LY294002 conditions) with anti-hnRNP-D antibody, followed by immunoblotting with anti-nucleolin antibody. Blots were stripped and reprobed with anti-hnRNP-D antibody.
finding explains, in part, the mechanism by which the PI3K is involved in the mechano-activation of the KLF2 promoter.
In B-lymphocytes, nucleolin interacts with hnRNP-D to form the LR1 transcriptional coactivator complex (32). Since both nucleolin and hnRNP-D also bind the KLF2 promoter, we hypothesized that they may interact in endothelial cells. Immunoprecipitations of HUVEC extract with anti-hnRNP-D antibody, followed by immunoblotting with antinucleolin antibody, indicated that hnRNP-D and nucleolin interact under both static and shear conditions, irrespective of PI3K inhibition (Fig. 5C). Reversal of the immunoprecipitating and immunoblotting antibodies verified a constitutive level of interaction (Fig. 5D). These findings indicated that nucleolin and hnRNP-D can interact in endothelial cells.
siRNA-mediated Reduction of Nucleolin Blocks the Induction of KLF2 by Fluid Shear Stress-The flow-specific binding of nucleolin to the KLF2 promoter in endothelial cells suggested that a significant reduction in cellular nucleolin levels might be sufficient to block the induction of KLF2 transcription by shear stress. To determine the contribution of nucleolin to the flow induction of the KLF2 gene, HUVECs were transfected with a pool of four RNA oligonucleotides targeting the nucleolin genomic sequence. A specific reduction in nucleolin protein levels of 78.5 Ϯ 4.6% (n ϭ 5) was achieved, compared with cells transfected with a pool of functional nontargeting siRNAs (F.N.T. in Fig. 6A). When the nucleolin-targeting siRNAs were introduced prior to fluid shear stress exposure, they reduced nucleolin mRNA levels and prevented the flow activation of KLF2 expression, as determined by reverse transcription PCR (Fig. 6B). KLF2 mRNA was reduced to a basal transcript level comparable with static cells. Functional nontargeting siRNA had no effect on mRNA levels. This finding indicates that nucleolin is essential to mediate the response of the KLF2 promoter to fluid shear stress.

DISCUSSION
We have identified nucleolin as a major component of KLF2 gene regulation. Furthermore, this is the first demonstration that nucleolin responds to fluid shear stress. These findings are supported by 1) KLF2 promoter DNA affinity chromatography and mass spectrometry with WR19M.1 macrophage extract, 2) verification of binding in macrophages and endothelial cells with EMSAs and ChIPs, 3) ChIPs demonstrating that, in endothelial cells, nucleolin binds the KLF2 promoter in a flow-specific PI3K-dependent manner, 4) co-immunoprecipitation experiments indicating that nucleolin interacts with additional factors involved in KLF2 gene regulation (p85 and hnRNP-D), and 5) siRNA-mediated gene knockdown experiments demonstrating that nucleolin is required for the induction of KLF2 expression by fluid flow.
Nucleolin is an abundant and ubiquitous cellular protein that is highly phosphorylated and possesses a diverse array of functions (39). It plays a role in ribosomal DNA transcription (40), ribosomal assembly (41), chromatin decondensation (29), and mRNA stability (42,43). It has also been identified as a transcriptional repressor (33) and a component (along with hnRNP-D) of the B cell transcription factor LR1 (28,32). Nucleolin has also been found to interact with the basic helix-loop-helix coactivator E47 (27). Interestingly, we have also verified, via ChIP assays, the presence of E47 on the KLF2 promoter in macrophages. 4 It seems, therefore, that nucleolin functions as a transcriptional regulator through its interactions with specific additional proteins.
The apparent molecular mass of nucleolin is ϳ104 -110 kDa in SDS-PAGE immunoblotting (40,44). However, the catalytic proteolysis of nucleolin is a well recognized phenomenon (39). In response to a specific stimulus, smaller forms of various molecular sizes are frequently observed (45)(46)(47). The exact size and predominance of a given species depends both on cell type and stimulus, but the smaller forms do exhibit independent functional specificity (35, 44) (e.g. the 95-kDa species of nucleolin that has been found to interact with the p85 regulatory subunit of PI3K) (34). The production of these forms probably results from the high susceptibility of nucleolin to post-translational regulation (48). Nucleolin is highly phosphorylated, and phosphorylation at specific residues by the flow-activated signal transduction pathway may facilitate catalysis. The exact reasons nucleolin is regulated in this manner are unknown. It has been postulated that this is the mechanism by which nucleolin can perform so many disparate functions, separating specialized domains from nonspecific regions (39).
The function(s) of nucleolin in the KLF2 promoter trans-activation complex is possibly related to its helicase activities. Nucleolin can unwind DNA-DNA duplexes and DNA-RNA duplexes and interacts directly with DNA topoisomerase I (49,50). This is noteworthy, considering that we have determined in vitro that the AT-rich tripartite palindrome motif, to which nucleolin binds, forms single-stranded secondary cruciform structures within the GC-rich area in which it resides. 4 Antibody specific for cruciform DNA structures was able to supershift an oligonucleotide probe consisting of the first 157 nucleotides of the KLF2 promoter in electrophoretic mobility shift assays. The proximal promoter of KLF1 (EKLF) failed to bind the cruciform antibody. In 4 J. P. Huddleson, N. Ahmad, and J. B. Lingrel, unpublished observations. FIGURE 6. Inhibition of nucleolin by siRNA blocks the induction of KLF2 mRNA by fluid shear stress. A, HUVECs were transfected with functional nontargeting siRNAs (F.N.T., negative control), lamin A/C siRNA (positive control), or nucleolin siRNA. Whole cell lysate was isolated 48 h after transfection, and 30 g was run on 10% polyacrylamide gel. Blots were probed with anti-nucleolin polyclonal antibody, stripped, and reprobed with anti-lamin A/C antibody and then stripped and reprobed with anti-GAPDH antibody. B, the effect of nucleolin reduction on flow-mediated induction of KLF2 was determined by reverse transcription-PCR. HUVECs were transfected with functional nontargeting siRNA or nucleolin siRNA and allowed to recover overnight. Cells were then exposed to 12 h of fluid shear stress (19 dynes/cm 2 ). RNA was isolated and reversed transcribed, and PCR was performed (30 cycles) with gene-specific primers for KLF2, nucleolin, or GAPDH. addition, two-dimensional agarose gel electrophoresis of a plasmid containing the KLF2 proximal promoter region generated a DNA pattern indicative of secondary DNA structures. Small palindromes, such as those found in the KLF2 promoter tripartite palindrome motif, have a strong tendency to form cruciform structures (51,52). Interestingly, nucleolin binds AT-rich single-stranded DNA with greater affinity (K D ϭ 0.6 nM) than it binds double-stranded DNA, and single-stranded DNA is found to some degree in DNA secondary structures (53,54). Nucleolin has also been reported to bind long stem-loop structures (55). Therefore, the interaction of nucleolin with secondary DNA structures within the tripartite palindrome motif of the Krüppel-like factor promoter may facilitate activation.
As mentioned above, six total factors in three different cell types have now been implicated in regulating the KLF2 promoter: hnRNP-U, hnRNP-D, PCAF, p300, myocyte enhancement factor 2, and nucleolin. We have preliminary evidence of an additional two. 4 None of these proteins are tissue-specific. In general, therefore, the evidence suggests that the KLF2 promoter is regulated by ubiquitous or non-cell-specific factors but that the binding of these proteins is restricted to a specific stimulus or signal transduction pathway(s). For example, the binding of nucleolin to the KLF2 promoter in endothelial cells is flow-specific, and the specificity of the binding interaction results from activation of the PI3K-dependent mechano-transduction pathway. Further elucidation of the factors and mechanisms that regulate KLF2 induction in diverse cell types will help our understanding of the larger biological processes that require its expression.