Integration of the non-genomic and genomic actions of estrogen. Membrane-initiated signaling by steroid to transcription and cell biology.

Estrogen binds to receptors that translocate to the plasma membrane and to the nucleus. The rapid, non-genomic actions of this sex steroid are attributed to membrane action, while gene transcription occurs through nuclear receptor function. However, gene transcription can also result from estrogen signaling initiated at the membrane, but the relative importance of this mechanism is not known. In vascular endothelial cells (EC), estradiol (E(2)) activates several kinase cascades, including phosphatidylinositol 3-phosphate (PI3K)/Akt, a signaling pathway that impacts EC biology. We determined here by DNA microarray that 40-min exposure to E(2) significantly increased 250 genes in EC, up-regulation that was substantially prevented by the PI3K inhibitor, LY294002. This coincided with maximum E(2)-induced PI3K activity at 15-30 min. An important vascular gene strongly up-regulated by E(2) in our array produces cyclooxygenase-2 (Cox-2). In cultured EC, E(2) induced both Cox-2 gene expression and new Cox-2 protein synthesis by 40 and 60 min, respectively, and rapidly stimulated the secretion of prostaglandins PGI(2) and PGE(2). The up-regulation of gene expression reflected transcriptional transactivation, shown using Cox-2 promoter/luciferase reporters in the EC. Soluble inhibitors or dominant negative constructs for PI3K and Akt prevented all these actions of E(2). Functionally, EC migration was induced by the sex steroid, and this was significantly reversed by NS-398, a Cox-2 inhibitor. Gene transcription and cell biological effects of estrogen emanate from rapid and specific signaling, integrating cell surface and nuclear actions of this steroid.

Estradiol (E 2 ) 1 and other steroid hormones are traditionally considered to transactivate target genes after binding nuclear receptors (1). However, E 2 also has rapid, non-genomic effects (2)(3)(4), and these have recently been attributed to cell membrane-initiated signaling. At the cell surface, a small population of ER binds E 2 and activates G proteins (5)(6)(7). Multiple signaling pathways are then rapidly stimulated by E 2 in target cells that express endogenous ER␣ and ER␤, and these pathways have been linked to discrete cellular actions of the steroid (8 -11). In this respect, a truncated MTA1 protein was recently found to be highly expressed in aggressive breast cancer (12). This protein sequesters ER away from the nucleus and strongly reduces E 2 -activated transcription yet promotes increased ERK signaling and aggressive behavior of the tumor. It is proposed therefore that the integration of cell surface and nuclear signaling impacts overall cell biology (13).
Signaling from the membrane leads to the post-translational modification of important structural and functional proteins in the cell. In EC, E 2 activates the p38-MAPKAP-2 kinase pathway; MAPKAP-2 phosphorylates and modifies the function of heat shock protein 27 (14). This important modification leads to the morphological preservation and survival of the EC and stimulates primitive capillary tube formation. In breast cancer, E 2 acts as a cell survival factor, in part by preventing chemotherapy or radiation-induced JNK activation (9). JNK phosphorylates and inactivates Bcl2 and Bcl-xl, leading to the assembly of the apoptosome and caspase-mediated cell death. By preventing JNK activation and Bcl2/Bcl-xl phosphorylation, E 2 rescues the breast cancer cells (9). This provides a mechanism for the ability of E 2 to oppose therapeutic interventions in this malignancy.
In addition to post-translational protein effects, E 2 is recognized to stimulate transcription through signaling typically initiated at the membrane. As precedent for this effect of E 2 , growth factor tyrosine kinase receptors (insulin-like growth factor-1 receptor and epidermal growth factor receptor) exist in the plasma membrane and signal through common kinase cascades to gene transcription. E 2 effects may involve the G protein-initiated, signaling-induced synthesis or activation of transcription factors. E 2 stimulates c-fos through ERK-or PI3Kdependent pathways (15)(16)(17), the BCL-2 gene through the modification of Sp-1 transcription factor (18), and the prolactin gene through ERK (19). E 2 activation of PI3K in EC results from the membrane ER-p85 regulatory subunit (PI3K) association, and this signaling to nitric oxide production rescues rats from ischemia-reperfusion injury (20). However, insight is largely lacking as to the full extent and importance of the specific integration of membrane signaling and nuclear effects of the sex steroid.
We therefore identified the transcriptional targets in EC that are rapidly up-regulated from E 2 signaling through the activation of PI3K. We also describe in depth the ability of E 2 -induced PI3K activation to up-regulate one specific gene, Cox-2, leading to the production of the enzyme, secretion of products of Cox-2 activation, and EC migration. This approach can be used to identify programs of gene activation that result from membrane-initiated steroid signaling (MISS) 2 by E 2 .

MATERIALS AND METHODS
PI3K Activity Assay-EC were incubated with/without 10 nM E 2 and 10 M LY294002 for up to 6 h. Cells were then lysed, and the lysates pelleted then dissolved in SDS sample buffer, boiled, separated, and transferred onto nitrocellulose. Phosphorylated Akt was detected using phospho-specific monoclonal antibodies (Santa Cruz Biotechnology) and the ECL Western blot kit. Equal samples from the cells were also immunoprecipitated, and immunoblots of the precipitated kinase protein from each experimental condition were determined to show equal gel loading. All experiments were repeated two or three times.
DNA Arrays-Human umbilical vein endothelial cells were incubated without or with 10 nM E 2 with or without the specific PI3K inhibitor, LY294002 (10 M), or LY294002 alone, for 40 min. For validation, the experiment was repeated a second time. Total RNA was extracted using TRIzol reagent (Invitrogen, Carlsbad, CA) and by RNeasy columns (Qiagen, Valencia, CA). Total RNA was adjusted to 1 g/l, and first-strand cDNA, followed by double-stranded cDNA was synthesized from poly(A) ϩ mRNA by the Microarray Facility at the University of California, Irvine. This was done using the SuperScript double-stranded cDNA synthesis kit (Invitrogen) and poly(T) nucleotide primers that contained a sequence recognized by T7 RNA polymerase. A portion of the resulting double-stranded cDNA was used as a template to generate biotin-tagged cRNA from an in vitro transcription reaction (IVT) using the BioArray High-Yield RNA transcript labeling kit (T7) (Enzo Diagnostics, Farmingdale, NY). 15 g of the biotintagged cRNA was fragmented to strands of 35-200 bases in length following prescribed protocols (Affymetrix GeneChip Expression Analysis Technical Manual). 10 g of the fragmented cRNA was hybridized with rotation at 45°C for 16 h to probe sets present on an Affymetrix human U95a array (Affymetrix, Santa Clara, CA). The arrays were automatically washed and stained with streptavidin-phycoerythrin. Probe arrays were then scanned on a Hewlett-Packard GeneArray scanner. Affymetrix Microarray Suite 5.0 was used to quantify and analyze the average difference in intensities between conditions of the experiment, for each represented gene. The comparisons between conditions were outputted as -fold increase or decrease, or no change, and data were compared with determine which genes were E 2 -responsive in a PI3K-dependent fashion. This output was then inserted into a stringent, Bayesian-based statistical analysis program (Cyber-T) (21), available as a web interface at the University of California at Irvine. The upper 2.5% of genes identified as being significantly different by Cyber-T (and in agreement with the Affymetrix analysis) are presented in table form.
RT-PCR and Reporter Assays-Validation of differential expression was performed by RT-PCR for four genes identified in the microarray and was standardized to glyceraldehyde-3-phosphate dehydrogenase. cDNA was prepared from 3 g of total RNA isolated from EC, primed with random hexamers (Invitrogen), and reverse-transcribed with Superscript II (Invitrogen) per the manufacturer's instructions. The primers were prepared by Invitrogen for the following sequences; Cox-2, TGGGAAGCCTTCTCTAACCTCTCCT and CTTTGACTGTGG-GAGGATACATCTC; CREM, TGGAAACAGTTGAATCACAG; and CTACTAATCTGTTTTGGGAG; EGR2, CAGTACCCTGGTGCCAGCTG and TGTGGATCTCTCTGGCACGG; JUN-B; CCGGATGTGCACGAAA-ATGGAACAG and ACCGTCCGCAAAGCCCTCCTG; glyceraldehyde-3phosphate dehydrogenase; ACCACAGTCCATGCCATCAC and TCCA-CCACCCTGTTGCTGTA. PCR reactions were performed using 200 nM of primers with 2 l of cDNA in 50 l of Platinum PCR supermix (Invitrogen). After an initial denaturation step of 94°C for 4 min, 25-35 cycles of 94°C for 30 s, 55-61°C for 45 s, 72°C for 30 -45 s, and a final extension at 72°C for 10 min. The PCR products were separated by electrophoresis in a 1.2% agarose gel and visualized by ethidium bromide staining. The cycle number and annealing temperature was adjusted for each of the genes amplified.
For reporter assays, BAEC were transiently transfected with 10 g of the PGL-3 plasmid containing 1.8 kb of the human Cox-2 promoter driving a luciferase reporter fusion protein (22), kindly provided by Dr.
David Dixon (Vanderbilt). The liposome-mediated transfection was carried out as previously described (6,9), and the cells were recovered in serum overnight, then synchronized without serum over 24 h before experiments (in phenol red-free medium). Cells were then incubated for 6 h with10 nM E 2 with or without either LY294002 (10 M), a soluble inhibitor of NFB (SN50M), or the corresponding inactive control (SN50), each at 20 M. All inhibitors were from Calbiochem (San Diego, CA). Assays were quantified by luminometer, and the results were adjusted for expression of co-transfected Renilla luciferase. Triplicate determinations per condition were carried out in each of two experiments. The combined data were analyzed by analysis of variance plus Schefe's test, at a p Ͻ 0.05 level of significance.
For PGI 2 (measured as 6-keto-PGF1␣) and PGE 2 secretion studies, BAEC were incubated for 0 -240 min with E 2 , to determine time of peak secretion. This was found to be 30 min, as subsequent time points reflected both secretion and accumulation. Additional studies were then conducted at 30 min. BAEC were incubated with/without 10 nM E 2 , with or without inhibitors, including NS-398 (5 M) (Cox-2 inhibitor), or SC-560 (20 nM) (Cox-1 inhibitor). Each condition was run in duplicate. The incubation media were collected and concentrated by lyophilization, and prostaglandin concentrations were determined by enzyme- . Cells were incubated with 10 nM E 2 with or without 10 M LY294002 for the times indicated, and Akt phosphorylation as a function of PI3K activity was determined by Western blot. Immunoblot of total Akt protein is shown below each condition. A representative study of two completed is shown.

FIG. 2. Estrogen-induced genes in endothelial cells, dependent
upon phosphatidylinositol 3-kinase signaling. EC were incubated/ not incubated with 10 nM E 2 with or without LY294002 (PI3K inhibitor), or LY294002 alone, for 40 min. Total RNA from the four conditions was then extracted for purposes of microarray analysis after cDNA/cRNA synthesis (see "Materials and Methods"). The figure is the output from CyberT analysis, a stringent, Bayesian-based statistical analysis program. Duplicate determinations for all conditions were utilized for the analysis. Genes in red are the upper 2.5% of all genes positively regulated by E 2 and mostly inhibited by LY294002.
linked immunosorbent assay (Cayman Chemical). The study was repeated three times, with the results reflected in the bar graph.
EC Migration Studies-EC were grown to monolayer confluence on six-well plates and synchronized for 24 h in the absence of serum. A "wound" was created by scraping the monolayer with a single-edge razor blade, and cells were removed to the left of the wound. Serum-free Dulbecco's modified Eagle's medium containing 10 nM E 2 with or without NS-398, SC-560, LY294002, wortmannin, or alone as control was added to separate dishes of wounded EC for 24 h at 37°C. The cells were then fixed in 3.7% formaldehyde and assessed for migration (14). BAEC migration was measured using an image analyzer system composed of an inverted microscope and a 20-to 24-inch digitizing board (Jandel Scientific, Corte Madera, CA) attached to a computer. The Sigma Scan program (Jandel) was used for analysis of measurements of the distance traveled by the cells within the calibrated area adjacent to the wound. Five measurements in each well were taken, and results from three separate experiments contributed to create the bar graph.

RESULTS AND DISCUSSION
Estrogen Stimulates PI3K Activity in EC-E 2 induced substantial Akt phosphorylation by 5 min, reflecting PI3K activation, because the phosphorylation was totally prevented by a PI3K inhibitor (Fig. 1). The peak activity occurred at 15 min and lasted for the 6-h duration of the experiment. We therefore carried out our array studies to assess genes rapidly induced by E 2 via PI3K signaling, based upon these results. E 2 Rapidly Induces Many Genes via PI3K Activation-EC were incubated with 10 nM E 2 or without steroid (control) for 40 min, in the presence or absence of LY294002, a PI3K inhibitor. cRNA from each experimental condition was used for microarray gene analysis, as delineated under "Materials and Methods." The DNA array hybridization pattern was analyzed by both Affymetrix statistical software and the CyberT Bayesianbased program, and 250 genes were identified as being significantly up-regulated by E 2 (upper 2.5% of all genes) (Fig. 2). This occurred in both E 2 -inducible and PI3K-reversible fashion. In comparing control cRNA (no treatment) to cRNA from cells treated with LY294002 alone, no differences were detected. We list the genes that were 1) up-regulated in this fashion by more than 2-fold and 2) identified to have some known function (Table I). In contrast, few genes were downregulated by E 2 in the quiescent EC, and none depended upon PI3K/Akt activation, whereas several genes were stimulated by E 2 and further enhanced upon PI3K inhibition. Genes discussed in the text are given in boldface in Table I.
As might be predicted from the time chosen, many transcription factors were rapidly up-regulated. These included the fos, myc, and jun genes previously known to be stimulated by estrogen (15,26,27), which thereby validate our results. Here we extend the findings and implicate PI3K action in protooncogene up-regulation by E 2 . This sex steroid was recently shown to activate c-fos transcription in MCF-7 cells, via a PI3K/Akt pathway, targeting the serum response factor motif in the proximal fos promoter (16). We also found stimulation of many transcription factors not previously known to be regu- Genes significantly up-regulated by estradiol in a PI3K-dependent fashion Genes that were induced at least 2-fold in endothelial cells after 40-min exposure to E 2 but were not significantly induced after exposure to E 2 and Ly294002 are listed. Differentially expressed genes were identified by both Affymetrix analysis software and the Cyber T (Bayesian) statistical analysis program. lated by E 2 ; these genes are therefore implicated in further E 2 transcriptional action. In some situations, linked gene programs could be tentatively identified, based upon the existing literature. As an example, bone morphogenetic protein 2 (Table  I, cytokines) stimulates osteoblast precursor-cell differentiation in part via up-regulating the AREB6 transcription factor (29). Estrogen induces osteoblast differentiation (30), and we found that E 2 stimulates both genes here, via a PI3K-induced mechanism. Another transcriptional target for AREB6 is the Na ϩ /K ϩ -ATPase gene (31). E 2 is known to stimulate the activity of this enzyme (32), potentially linking these observations to upstream signaling. As shown here, the HZF2 transcription factor is induced by E 2 in PI3K-dependent fashion (Table I). HZF2 has been reported to be up-regulated by nitric oxide (NO) (33), and E 2 strongly and rapidly stimulates NO production in EC in PI3K-dependent fashion (20). NO induction by E 2 prevents the deleterious blood vessel response to ischemia-reperfusion injury (20). Together, these results potentially identify a linked cell biological program in EC. Several members of the EGR family of transcription factors were identified as induced by E 2 in our array. Egr-1 is up-regulated (and important) in the response to acute and chronic vascular injury, where it may serve a protective function (34,35). E 2 mitigates the acute injury response to carotid angioplasty (36), perhaps in part through inducing Egr-1 in the EC, as shown here.
Signaling molecules were also rapidly induced by E 2 , including both kinases and phosphatases. Steroid and glucocorticoid-inducible kinase activity is known to be a target for PI3K/Akt signaling (37) and has roles in both steroid-induced memory (38) and sodium transport (39), which are both functions of E 2 (40,41). We also found that, via PI3K, E 2 up-regulates Cot/ (Tpl-2), a transformation-associated factor and serine/threonine kinase. Akt phosphorylation of Cot induces NFB-dependent transduction (42), important for various functions, including cell survival. E 2 potently inhibits hypoxia-induced EC apoptosis (14).
Genes coding for structural proteins, cytokines, or enzymes were identified as being stimulated by E 2 . Two members of the NGFI-B subfamily of nuclear orphan receptors, TR3 and MINOR, were rapidly and strongly induced by E 2 via PI3K. A cytokine gene, CYR61, was previously demonstrated to be upregulated by E 2 in breast cancer but through an unknown mechanism (43). In EC, we implicate signaling via PI3K. Upregulation of the PTTG gene by estrogen (found here) was previously shown to contribute to the pathogenesis of E 2induced pituitary tumor formation and the stimulation of bFGF secretion (44). E 2 and bFGF are recognized angiogenesis factors (14,45), and therefore a potential linkage in EC of PTTG-induced bFGF could be important for the recognized neovascularization function of E 2 . The PLAP (phospholipase A2-activating protein) gene was found to be stimulated in our array (Table I). This protein has been reported to contribute to up-regulate the Cox-2 gene and stimulate prostaglandin E 2 (PGE 2 ) production (46). Based on our results, we propose a Integration of the Non-genomic and Genomic Actions of Estrogen linkage of E 2 -induced, PI3K-dependent up-regulation of PLAP, contributing to E 2 -induced Cox-2 up-regulation and PG secretion/production (see below). Thus, many relevant genes are rapidly induced by this sex steroid in response to one signal pathway typically initiated at the plasma membrane.
Confirmation by RT-PCR of E 2 -induced Gene Up-regulation-To confirm the array studies, we carried out semi-quantitative RT-PCR for several of the genes identified (Fig. 3). For the four genes examined (Cox-2, JunB, CREM, and EGR2), there was increased expression in EC after 40-min exposure to 10 nM E 2 (lane 2), compared with the control (no E 2 ) (lane 1). Furthermore, this increase was significantly abrogated by addition of the PI3K inhibitor, LY294002 (lane 3), whereas the inhibitor alone had no effect compared with control levels (lane 4). The results provide validation of the array data for these specific genes.
Cox-2 Gene and E 2 /ER Interactions Result from Rapid Signaling by the Steroid-One of the genes strongly up-regulated by E 2 in our EC gene array codes for the Cox-2 enzyme (Table  I). Cox-2 activity gives rise to pGI 2 and PGE 2 production, important for various aspects of vascular function (47,48). To explore the interactions between E 2 signaling through PI3K to Cox-2 in greater depth, we first confirmed the array results by RT-PCR (Fig. 3). We then further investigated transcriptional regulation by E 2 . We therefore expressed in BAEC a plasmid containing a 1.8-kb human Cox-2 promoter driving a luciferase reporter. This was significantly responsive to 10 nM E 2 , in a PI3K-dependent fashion (Fig. 4).
One known target for PI3K/Akt signaling is the activation of the NFB transcription factor (49,50). We found that a soluble NFB inhibitor (but not its inactive control) completely reversed the E 2 stimulation of the Cox-2 promoter (Fig. 4). These data support the microarray studies and the idea that the up-regulation of the Cox-2 gene by E 2 is transcriptional. Furthermore, this E 2 action depends on signaling via PI3K and NFB. NFB binding sites are present within the human Cox-2 promoter at Ϫ580 and Ϫ358. Supporting this mechanism, we found that E 2 significantly activated both 0.8 and 0.4 kb Cox-2 promoters driving luciferase reporter constructs, suggesting that the NFB binding site at Ϫ358 in particular is important. 3 Previous studies indicate that PI3K/Akt can regulate Cox-2 mRNA production or stability in positive or negative fashion, dependent upon the stimulus and cellular context (51,52).
Cox-2 Protein Synthesis and Prostaglandin Secretion Are Stimulated by E 2 -Cox-2 protein synthesis was then determined by Western blot. After 1-h exposure of EC to E 2 , the Cox-2 protein was nearly 3-fold increased in relevant E 2 concentration-responsive fashion (Fig. 5A). The PI3K inhibitors, LY 294002 and wortmannin, each caused an 80% reduction in Cox-2 protein synthesis, as did ICI 182,780, an ER antagonist. Expression of dominant negative constructs for the p85 subunit of PI3K (DN-PI3K) or Akt (DN-Akt) also resulted in substantial inhibition of E 2 -induced Cox-2 protein expression. The dominant negative constructs had no effects alone (data not shown).
In preliminary studies, we determined a time course for PGE 2 and PGI 2 secretion in response to E 2 . Compared with basal secretion (0 time), E 2 stimulated an initial 2-fold PGE 2 release at 10 min (first point assessed), reaching a maximum increase at 30 min and plateauing thereafter (data not shown). Based upon these results, BAEC were then incubated with 10 nM E 2 with or without inhibitors of ER, PI3K, and Cox-1 or Cox-2 for 30 min. As seen in Fig. 5B, E 2 stimulated a 13-fold increase of PGE 2 secretion, 75% prevented by ICI 182,780. Co-incubation of the cells with NS-398 (a specific Cox-2 inhibitor) reversed the E 2 effect by 86%. A specific Cox-1 inhibitor (SC-560) also provided a 33% inhibition of the E 2 effect, suggesting that PGE 2 synthesis was mainly dependent upon the Cox-2 enzyme. It is recognized that PGE 2 and PGI 2 can both result from either Cox-1 or Cox-2 enzymatic action, but in a given cell type or in response to a specific stimulus, one cyclooxygenase activity may predominate over the other (53). Importantly, the PI3K inhibitors, wortmannin and LY 294002 each prevented E 2 -induced PGE 2 secretion by 87%. This is consistent with the ability of E 2 to activate the Cox-2 gene in a PI3K-dependent fashion, as identified by microarray and reporter studies. Interestingly, the PGE 2 receptor was also found in our array to be up-regulated by E 2 via PI3K (Table I).
As for PGI 2 , E 2 induced a 4-fold increase in secretion (Fig.  5B). This was 60% reversed by ICI 182,780, 58% by the Cox-2 inhibitor, and 36% by the Cox-1 inhibitor. The PI3K inhibitors also reduced the E 2 induction of PGI 2 by 60%. Our results greatly extend the observations of others that E 2 can rapidly stimulate the secretion of PGI 2 from endothelial cells (54,55). Here we show transcriptional up-regulation of the Cox-2 gene, increased protein production, and stimulation of both PGE 2 and PGI 2 secretion, in a PI3K-dependent fashion. An observation that is relevant for EC is that Cox-2 and cAMP-signaling enhances angiogenesis through the induction of vascular endothelial growth factor (56). Cox-2 and cAMP are up-regulated by FIG . 5. A, E 2 stimulates the production of Cox-2 protein in EC via PI3K and Akt signaling. EC were incubated with several concentrations of E 2 for 60 min, and Cox-2 protein was determined by Western blot of immunoprecipitated cell lysate. In some conditions, LY294002 or ICI182780 was added 30 min prior to 10 nM E 2 , or the cells were first transfected to express pMT2-AH-Akt (DN.Akt) or pcDNA3-delta p85 (DN.PI3-K). Control and E 2 with or without LY conditions were carried out in cells transfected with pcDNA3. A representative study is shown, and the bar graph reflects three experiments combined. *, p Ͻ 0.05 for control versus E 2 ; ϩ, p Ͻ 0.05 for 10 nM E 2 versus E 2 plus inhibitor. B, secretion of PGE 2 (left) and PGI 2 (6-keto prostaglandin F1alpha) (right) in response to E 2 . EC were incubated with 10 nM E 2 with or withoutPI3K inhibitors, or with E 2 ϩ NS-398 (Cox-2 inhibitor) or SC-560 (Cox-1 inhibitor) for 30 min (soluble inhibitors were added 30 min prior to E 2 ). The bar graph represents three combined experiments. E 2 signaling from the membrane (2, 6, and here). This may be significant to E 2 -modulated induction and developmental function of vascular endothelial growth factor, in the formation and permeability of the blood vessels of the ovary and uterus (56,57).
EC Migration-We then examined possible roles for Cox-2derived prostaglandins and E 2 in mediating EC migration. E 2 is known to induce several aspects of angiogenesis, including EC migration (14), and Cox-2 also importantly participates in these processes (28). Cultured EC were "wounded," and the migration of EC across this wound barrier was determined after 24-h exposure to various conditions. As seen in Fig. 6, 10 nM E 2 (panel B) induced a 3-fold increase in migration area compared with control EC (panel A). The effect of E 2 was reduced 75% by a Cox-2 inhibitor (panel C), but not significantly by a Cox-1 inhibitor (panel D). The effect of E 2 was also prevented by LY294002 (panel E) and by wortmannin (PI3K inhibitor) (panel F). The inhibitors by themselves had no effects on EC migration (data not shown). Thus, we link the Cox-2 up-regulation in EC, induced by E 2 and PI3K signaling, to a cell biological outcome.
Increasingly, the integration of membrane and nuclear actions of steroids is recognized. An important mechanism demonstrated here is that steroids rapidly induce through kinase signaling many genes coding for transcription factors. We determined that the identified transcription factor gene promoters usually lack any form of estrogen response elements. Transcription factor up-regulation presumably then leads to the induction of additional genes that regulate steroid-induced cellular function. However, we also show that, by the same signaling mechanism, steroids rapidly up-regulate genes coding for enzymes or signaling molecules. The protein products of these genes both directly impact cell functions (i.e. cell migration) and induce additional transcription. Through membraneinitiated steroid signaling (MISS), we propose that estrogen affects overall cellular processes by post-translationally modulating the functions of existing proteins (9,14) and by activating discrete programs of gene expression. Cultured EC were cut with a surgical blade, and all cells were removed on the left side of the wound with a scraper. The remaining cells were then cultured overnight with/without 10 nM E 2 with or without PI3K and Cox-2 inhibitors. Panel A is control cells (no E 2 or serum); panel B is E 2 . Panel C is E 2 ϩ NS-398, panel D is E 2 ϩ SC-560, panel E is E 2 ϩ LY294002, and panel F is E 2 ϩ wortmannin (50 nM). The inhibitors alone had no effects (data not shown). The bar graph below the composite reflects three experiments combined. *, p Ͻ 0.05 for control versus E 2 ; ϩ, p Ͻ 0.05 for E 2 versus E 2 plus inhibitor.