12/15-Lipoxygenase Activity Mediates Inflammatory Monocyte/Endothelial Interactions and Atherosclerosis in Vivo*

We have shown that the 12/15-lipoxygenase (12/15-LO) product 12S-hydroxyeicosatetraenoic acid increases monocyte adhesion to human endothelial cells (EC) in vitro. Recent studies have implicated 12/15-LO in mediating atherosclerosis in mice. We generated transgenic mice on a C57BL/6J (B6) background that modestly overexpressed the murine 12/15-LO gene (designated LOTG). LOTG mice had 2.5-fold elevations in levels of 12S-hydroxyeicosatetraenoic acid and a 2-fold increase in expression of 12/15-LO protein in vivo. These mice developed spontaneous aortic fatty streak lesions on a chow diet. Thus, we examined effects of 12/15-LO expression on early events leading to atherosclerosis in these mice. We found that, under basal unstimulated conditions, LOTG EC bound more monocytes than B6 control EC (18 ± 2 versus 7 ± 1 monocytes/field, respectively; p < 0.0001). Inhibition of 12/15-LO activity in LOTG EC using a 12/15-LO ribozyme completely blocked monocyte adhesion in LOTG mice. Thus, 12/15-LO activity is required for monocyte/EC adhesion in the vessel wall. Expression of ICAM-1 in aortic endothelia of LOTG mice was increased severalfold. VCAM-1 expression was not changed. In a series of blocking studies, antibodies to α4 and β2 integrins in WEHI monocytes blocked monocyte adhesion to both LOTG and B6 control EC. Inhibition of ICAM-1, VCAM-1, and connecting segment-1 fibronectin in EC significantly reduced adhesion of WEHI monocytes to LOTG EC. In summary, these data indicate that EC from LOTG mice are “pre-activated” to bind monocytes. Monocyte adhesion in LOTG mice is mediated through β2 integrin and ICAM-1 interactions as well as through VLA-4 and connecting segment-1 fibronectin/VCAM-1 interactions. Thus, 12/15-LO mediates monocyte/EC interactions in the vessel wall in atherogenesis at least in part through molecular regulation of expression of endothelial adhesion molecules.

itors decrease the ability of macrophages to oxidize LDL (8,(13)(14)(15)(16). Cathcart and co-workers (15) found that 15-LO activity in monocytes produces superoxide that mediates oxidation of LDL. 15-LO protein has been localized to aortic atherosclerotic lesions in rabbits and humans (17,18) and is responsible for production of oxidized lipid adducts localized within atherosclerotic plaques (16,19). We have shown previously that 12S-HETE induces monocyte adhesion to human aortic EC in vitro (20).
To understand the mechanisms by which 12/15-LO-generated eicosanoids modulate events leading to atherogenesis in vivo, we produced transgenic mice that modestly overexpressed the murine 12/15-LO gene. In vivo, these mice produced 2-fold elevations in levels of 12S-HETE and 13S-HODE, the major eicosanoid products of the enzyme. The transgenic mice developed spontaneous aortic fatty streak lesions on a rodent chow diet. Thus, we examined the role of 12/15-LO activity in mediating monocyte/endothelial interactions, key early events in atherogenesis in vivo.
Generation of Mice-12/15-Lipoxygenase transgenic mice (designated LOTG) were generated on a C57BL/6J (B6) background using a 15.5-kb mouse genomic clone as shown in Fig. 1A. This genomic clone contains the full-length murine 4.5-kb leukocyte-type 12-LO gene (4). The genomic clone also comprises ϳ7 kb upstream and 2 kb downstream of the murine 12/15-LO gene to include additional proximal promoter and enhancer elements. The genomic clone was excised from FIXII (Stratagene) using NotI and injected into C57BL/6J mouse blastocysts. Two founder lines were generated: 16F and 41F. Mice from both founder lines were viable and fertile. We have not observed phenotypic differences between male and female LOTG mice. The genotyping strategy for these mice utilized the T7 and T3 polymerase sites located just inside the NotI restriction site of the genomic clone (Fig.  1A). The following primers were used to identify positive mice: T7 (5Ј-ctctaatacgactcactata-3Ј) and T7409 (5Ј-tgttgaccaaatcaccgaga-3Ј); and T3 (5Ј-caattaaccctcactaaagg-3Ј) and T3271 (5Ј-atgccaccctacagaaatgcta-3Ј). The T7/T7409 primers yield a 420-bp PCR product; the T3/ T3271 primers yield a 333-bp product. PCR conditions were as follows: 94°C for 2 min; followed by 40 cycles at 94°C for 15 s, 54°C for 30 s, and 72°C for 45 s; with a final extension time of 5 min at 72°C. Results of the genotyping for the founder mice using the T3/T3271 and T7/ FIG. 1. Generation of murine 12/15-LO transgenic mice. A, schematic of the genomic construct used to generate LOTG mice. The genomic clone was excised from the FIXII vector using the NotI restriction sites. This generated a linear genomic clone that included ϳ7 kb upstream and 2 kb downstream of the 4.5-kb murine 12/15-LO gene. Exons 1-14 are illustrated. Excision of the genomic clone using NotI allowed inclusion of the T7 and T3 sequences that were used in the screening strategy to identify positive founder mice. B, identification of two founder lines (16F and 41F) using a PCR strategy. PCR was performed on genomic DNA isolated from rodent tail clips as described under "Experimental Procedures." Shown are a 333-bp PCR product using the T3/T3271 primers and a 407-bp product using the T7/T7409 primers. The NotI-excised genomic clone was analyzed as a positive control by PCR for identification purposes. C, Southern analysis of genomic DNA isolated from both transgenic founder lines. Genomic DNAs from B6 control and 16F and 41F transgenic mice were separated on a 1% agarose gel. The genomic DNA was probed with a 32 P-labeled full-length mouse 12/15-LO cDNA probe. L, ladder.
T7409 primers are shown in Fig. 1B. From the founder mice, these PCR products were sequenced and found to correspond to the 12/15-LO genomic clone sequence. Southern analysis of genomic DNA from C57BL/6J control mice and 16F and 41F transgenic mice showed increased 12/15-LO copy number in both transgenic lines compared with the control line (Fig. 1C). The full-length mouse 12/15-LO cDNA was used as a probe for Southern analysis. In some studies, we used 12/ 15-LO knockout mice on a C57BL/6J background (21). These mice were a kind gift of Dr. Colin D. Funk (University of Pennsylvania).
Eicosanoid Measurement in Urine-Lipids were extracted from urine as described previously using Varian C 18 Bond-Elut columns in series (10). 8S-HETE was added as an internal standard. The fluorescence derivatives of the free fatty acids were formed using 8 mg of 2-(2,3naphthalimino)ethyltrifluoromethanesulfonate dissolved in 0.5 ml of acetonitrile. The reaction mixtures were dried with nitrogen, resuspended in 0.4 ml of acetonitrile, diluted with 0.6 ml of water, and applied to a third Bond-Elut column. The fatty acid derivatives were eluted with 0.6 ml of ethyl acetate, evaporated under nitrogen, and resuspended in 100 l of methanol for HPLC analysis. HPLC separation and analysis were performed using a Waters C 18 symmetry column and eluting isocratically for 100 min in 61% solvent B (solvent B ϭ 50% methanol/tetrahydrofuran ϩ 0.1% acetic acid, and solvent A ϭ 0.1% acetic acid) at a flow rate of 1.2 ml/min following related protocols of Roman and co-workers (22). Peaks were detected fluorometrically at an excitation wavelength of 259 nm and an emission wavelength of 394 nm. The area ratio of sample HETE area to internal standard (8-HETE) area was plotted against nanograms of HETE injected, and unknown sample HETE values were calculated from their area ratios. HETEs (5,12, and HODEs (9,13-HODE) were base line-separated using this elution protocol. HETE and HODE measurements in urine were normalized to grams of creatinine. Fig. 2 illustrates the HPLC elution profile of HETE and HODE standards (panel A) as well as representative HPLC chromatograms from B6 control and LOTG mice (panel B).
Aortic Atherosclerosis Measurements-Male C57BL/6J mice and male mice from the 41F line (8 weeks old) were fed a rodent chow diet for 20 weeks. After 20 weeks, mice were euthanized. The organs were perfused with 10% formalin. The intact aorta from the heart to the iliac artery (ϳ3 mm from the bifurcation) was removed. The adventitia was removed, and the aorta was pinned and stained with Sudan IV. Quantification of aortic lesion area was performed according to the methods of Daugherty and Whitman (23).
Flow Cytometry Analysis of Endothelial Adhesion Molecule Expression-Mouse EC at passage 4 were collected in phosphate-buffered saline by gentle scraping using a cell scraper. Cells (150,000/sample) were analyzed for each antibody. Cells were incubated for 30 min at 4°C with a 1:100 dilution of fluorescein isothiocyanate (FITC)-labeled primary antibodies for mouse adhesion molecules (FITC-labeled antimouse ICAM-1, BD Biosciences catalog no. 553252; and FITC-labeled anti-mouse VCAM-1, BD Biosciences catalog no. 553332) or isotype control antibodies (FITC-labeled hamster IgG1, BD Biosciences catalog no. 553971; and FITC-labeled rat IgG2a, BD Biosciences catalog no. 553929). After incubation, cells were washed three times with phosphate-buffered saline and fixed in paraformaldehyde. Samples were analyzed at the University of Virginia Flow Cytometry Core using a BD Biosciences FACSCalibur TM flow cytometer. FITC fluorescence was collected in the FL1 channel through a 530/30-nm band-pass filter. Analyses were performed using FL1 channel histograms, from which the mean fluorescence intensity channel was calculated for each sample.
Isolation of Mouse Aortic Endothelial Cells-Aortic EC from C57BL/ 6J, LOTG, and 12/15-LO knockout (designated LOKO) mice were harvested from mouse aortas under sterile conditions following the methods outlined previously (24). Mouse EC cultures were cultured in Dulbecco's modified Eagle's medium containing 15% heat-inactivated fetal bovine serum, 60 g/ml endothelial cell growth supplement, and 100 g/ml heparin and were used in experiments from passages 2 to 5.
Immunoblotting for 12/15-LO Protein-MAEC from B6 control and LOTG mice were harvested in 1ϫ cell lysis buffer (Cell Signaling, Inc.) in the presence of a protease inhibitor mixture (Sigma). 75 g of total EC protein was analyzed by 4 -12% SDS-PAGE in MOPS running buffer (Invitrogen) and transferred to nitrocellulose. Pierce BLOTTO buffer was used as a blocking agent. Membranes were probed with a 1:1000 dilution of rabbit polyclonal antibody to murine 12/15-LO. 12/ 15-LO protein was detected using a 1:2000 dilution of horseradish peroxidase-conjugated anti-rabbit IgG and chemiluminescence. Bands were normalized to tubulin (1:5000 dilution of antibody) and quantitated by densitometry.
Mouse Monocyte Isolation-Whole blood was obtained via retro-orbital bleeding from 15 heterozygous CX3CR1-GFP mice. These mice were generated by replacing the coding region of CX3CR1 (the fractalkine receptor) with GFP. Thus, these mice express GFP only in monocytes and in a subset of natural killer cells. GFP expression in the monocytes allowed us to purify monocytes using sterile fluorescence cell sorting. To isolate the monocytes, whole blood was collected into heparin and diluted with an equal volume of 0.9% saline. 6 ml of diluted blood was carefully layered over 3 ml of NycoPrep one-step 1.077/265 animal separation solution into 15-ml tubes. Blood was centrifuged at 600 ϫ g for 15 min. Mononuclear cells were harvested from the interface between the plasma layer and the NycoPrep solution. Mononuclear cells were washed with Hanks' balanced salt solution, centrifuged at 400 ϫ g for 15 min, and resuspended in 1 ml of Hanks' balanced salt solution. For cell sorting, monocytes were incubated for 30 min at 4°C with allophycocyanin-labeled anti-NK1.1 antibody to label natural killer cells. After incubation, cells were rinsed with Hanks' balanced salt solution, resuspended in 1 ml of Hanks' balanced salt solution, and sorted using a BD Biosciences FACSVantage SE TM cell sorter by forward and side scatter gating for monocytes ( Fig. 3A) and then selecting for GFP hi /NK1.1 lo -expressing cells (Fig. 3B). Monocyte recovery was only ϳ50%; however, purity based upon morphology was Ͼ92%. For the adhesion assay, isolated mouse monocytes were labeled with calcein O,OЈ-diacetate tetrakis(acetoxymethyl) ester according to the manufacturer's instructions, and 20,000 monocytes/well were incubated with MAEC for 45 min at 37°C. Unbound monocytes were rinsed, and adherent monocytes localized within a 10 ϫ 10 grid at ϫ40 magnification were counted by epifluorescence microscopy.
Quantitative Real-time PCR for Murine ICAM-1 and Murine 12/15-LO-Organ tissues were collected from control and LOTG mice from the 41F line and quick-frozen in liquid nitrogen for 12/15-LO mRNA measurements. MAEC were freshly isolated from aortas and cultured as described above in 100-mm cell dishes for 12/15-LO mRNA and ICAM-1 mRNA measurements. Total cellular RNA was obtained from either MAEC or frozen tissue using TRIzol (Invitrogen) according to the manufacturer's instructions. Reverse transcription of 2 g of total RNA was performed in a total volume of 25 l using Thermoscript RT (Invitrogen) and random hexamers. The RNA was treated with DNase I at room temperature for 10 min. The reaction was stopped by addition of EDTA. For quantitative PCR analysis, the resulting cDNA was diluted 1:5, and 2 l was used for each PCR. Reagents from the QIAGEN real-time PCR kit containing SYBR Green were used for quantitative PCRs. The PCR conditions for ICAM-1 were as follows: 95°C for 10 min; 95°C for 4 min; followed by 40 cycles at 95°C for 15 s, 60°C for 30 s, and 72°C for 30 s; followed by a final extension at 81°C for 15 s. For 12/15-LO and cyclophilin, the PCR conditions were as follows: 95°C for 10 min; 95°C for 4 min; followed by 40 cycles at 95°C for 15 s, 62°C for 30 s, and 72°C for 30 s; followed by a final extension at 81°C for 15 s. The primers for murine 12/15-LO were 5Ј-ctctcaaggcctgttcagga-3Ј (forward) and 5Ј-gtccattgtccccagaacct-3Ј (reverse). The primers for murine ICAM-1 were 5Ј-agatcacattcacggtgctg-3Ј (forward) and 5Ј-cttcagaggcaggaaacagg-3Ј (reverse). The primers for murine cyclophilin were 5Јtggagagcaccaagacagaca-3Ј (forward) and 5Ј-tgccggagtcgacaatgat-3Ј (reverse). Data were analyzed and are presented based upon the relative FIG. 2. HPLC analysis of eicosanoid production in 12/15-LO transgenic mice. Urine was collected from mice using metabolic cages, and HPLC analysis of eicosanoid production was performed as described under "Experimental Procedures." A, an HPLC chromatogram illustrating elution profiles of a standard mixture of HETEs (He) and HODEs (Ho). B, representative elution profiles of eicosanoids present in the urine of a B6 control mouse and a LOTG mouse. expression method (31). This formula for calculation is follows: relative expression ϭ 2 Ϫ(S⌬CTϪC⌬CT) , where ⌬C T is the difference in threshold cycle between the gene of interest (either 12/15-LO or ICAM-1) and the housekeeping gene (cyclophilin), S is LOTG mice, and C is B6 control mice.
Statistical Analyses-Comparisons between groups were performed by analysis of variance (ANOVA). Data are represented as the means Ϯ S.E. of eight mice (unless otherwise noted in the figure legends). All comparisons were made using Fisher's Least Significant Difference procedure so that multiple comparisons were made at the 0.05 level only if the overall F-test from ANOVA was significant at p Ͻ 0.05.

12/15-LO Transgenic Mice Have Increased Levels of 12S-HETE and 13S
-HODE in Vivo-LOTG mice had significant elevations in levels of urinary 12S-HETE and 13S-HODE, the primary eicosanoid products of the 12/15-LO enzyme (Fig. 4). Levels of 12S-HETE were increased from 200 ng/g of creatinine in B6 control mice to 500 ng/g of creatinine in LOTG mice (41F line) (p Ͻ 0.0001). Levels of urinary 13S-HODE were found to be 10 ng/g of creatinine in B6 control mice and 50 ng/g of creatinine in LOTG mice (41F line) (p Ͻ 0.0001). Levels of urinary 12S-HETE and 13S-HODE were also elevated in mice from the 16F line (289 ng/g of creatinine for 12S-HETE and 17 ng/g of creatinine for 13S-HODE; p Ͻ 0.001), although the increase in eicosanoid production in this transgenic line was much more modest. Overall, there was no apparent shunting to the 5-LO pathway in the LOTG mice, as 5S-HETE levels in LOTG mice remained similar to B6 control levels. We have previously reported that 5S-HETE does not induce monocyte adhesion to the endothelium (20). Tissue distribution of 12/ 15-LO mRNA in LOTG mice was normal, with expression found in aorta, brain, heart, kidney, liver, and pancreas. Expression of 12/15-LO mRNA was increased 2-fold in aorta, liver, and pancreas and 3-4-fold in brain, heart, and kidney of LOTG mice compared with B6 mice (Fig. 5).
Endothelial Cells from 12/15-LO Transgenic Mice Are "Preactivated" to Bind Monocytes-We have shown that nanomolar concentrations of 12S-HETE activate human aortic EC to bind monocytes (20). Based upon these initial data in human aortic EC, we wondered if aortic EC from LOTG mice would show increased monocyte adhesion. We recently developed a technique in the laboratory for isolation of primary MAEC (24). EC were isolated from B6 and LOTG mice and used under basal unstimulated conditions. As shown in Fig. 6, there was a significant increase in monocyte adhesion to unstimulated LOTG EC from both founder lines (4 Ϯ 1 monocytes/field for 16F mice and 7 Ϯ 1 monocytes/field for 41F mice versus 2 Ϯ 1 monocytes/ field for B6 control mice; p Ͻ 0.01). In this experiment, we also compared adhesion of WEHI 78/24 cells (a mouse monocyte cell line) with that of primary mouse monocytes. Mouse peripheral blood monocytes are difficult to isolate, so we routinely use the WEHI mouse monocytes in adhesion assays. Both WEHI monocyte binding and mouse peripheral blood monocyte binding to EC were higher in LOTG mice than in B6 control mice (p Ͻ 0.001) (Fig. 6). These data indicate that 1) unstimulated basal EC isolated from LOTG mice from both founder lines displayed increased adhesion; 2) WEHI 78/24 cells bound B6 control and LOTG EC in a similar manner compared with primary mouse monocytes, although the background adhesion was slightly higher with the WEHI cells; and 3) the 41F line showed significantly higher adhesion than the 16F line, presumably due to the slightly higher activity of 12/15-LO in this line (Figs. 1, 2 (Fig. 7). In addition, we found a coordinated induction of 12/15-LO protein in aortic EC from LOTG mice (line 41F) compared with B6 control mice (p Ͻ 0.009) (Fig. 7). These data reflect a 2.5-fold increase in 12/ 15-LO protein expression in aortic EC from LOTG mice of the 41F line.

12/15-LO Transgenic Mice Develop Aortic Atherosclerosis on a Chow
Diet-Male mice from the 41F line were fed a chow diet for 15 weeks. Aortas were isolated from the heart to the iliac artery (ϳ3 mm from the bifurcation) and quantified for ather-osclerosis by en face staining of the aorta according to the methods of Daugherty and Whitman (23). Laboratory strains of mice do not develop significant aortic fatty streak lesions when maintained on chow diets (33)(34)(35). LOTG mice developed spontaneous aortic atherosclerotic lesions on a chow diet (Fig. 8). The lesions resembled typical fatty streaks with subendothelial accumulations of lipid and were most prominent around the aortic arch and intercostal areas (data not shown).
Decreased Monocyte Adhesion to Endothelial Cells from 12/ 15-LO Knockout Mice-We also isolated EC from LOKO mice, which are on a C57BL/6J background and have a global deletion of the murine 12/15-LO gene (5,21). We examined monocyte adhesion to LOKO EC and to EC that were stimulated with 250 g/ml mildly oxidized LDL for 4 h. As shown in Fig. 9, WEHI monocyte adhesion to EC from LOKO mice was significantly lower compared with B6 control or LOTG mice from line 41F (3 Ϯ 1 monocytes/field for LOKO mice versus 6 Ϯ 1 monocytes/field for B6 control mice versus 18 Ϯ 3 monocytes/field for LOTG mice; p Ͻ 0.0001). All EC groups responded to oxidized LDL, although LOTG EC bound more monocytes than B6 control or LOKO EC in all cases. TNF-␣ was used as a positive control to indicate maximal binding in the adhesion assay.

12/15-LO Products Directly Stimulate Monocyte/Endothelial
Adhesion-To test the hypothesis that 12/15-LO activity directly mediates monocyte adhesion, we inhibited expression of 12/15-LO in LOTG EC using an adenovirus expressing a ribozyme to 12/15-LO as well as the pharmacological 12/15-LO inhibitor CDC and then measured adhesion. CDC blocks platelet 12-LO and 12/15-LO expression in EC (36). The DNA/RNA hammerhead ribozyme was generated to recognize the first 7 bp of the porcine and murine 12/15-LO mRNA sequences (30).

FIG. 4. Transgenic mice have increased production of 12/15-LO eicosanoid products in vivo.
Urine samples from B6 control (CTR) and LOTG line 16F and 41F mice were collected overnight using metabolic cages. Lipids were extracted from urine using Bond-Elut columns, and eicosanoids were measured by fluorescence HPLC as described under "Experimental Procedures." Eicosanoid values were normalized to grams of creatinine present in urine. Values that were significantly higher than B6 control values for mice by ANOVA are indicated (*, p Ͻ 0.001; #, p Ͻ 0.01). Data represent the means Ϯ S.E. of six mice/group.

FIG. 5. Tissue distribution of 12/ 15-LO in control and transgenic mice.
Tissues were harvested from B6 control and LOTG line 41F (TG) mice and quickfrozen in liquid nitrogen. Total cellular RNA was isolated, and real-time PCR for 12/15-LO was performed as described under "Experimental Procedures." All values were normalized to cyclophilin based upon relative expression analysis for quantitative PCR. Values for B6 control mice were set to 1 for relative expression analysis. Values that were significantly higher than B6 control values for each tissue by Student's t test are indicated (*, p Ͻ 0.01; **, p Ͻ 0.005). Data represent the means Ϯ S.E. of four mice/group.

FIG. 6. Aortic endothelial cells from 12/15-LO transgenic mice have increased monocyte adhesion.
Unstimulated aortic EC from B6 control (CTR) and LOTG line 16F and 41F mice were used in an adhesion assay as described under "Experimental Procedures." Either calcein-labeled WEHI 78/24 cells (50,000/well) or calcein-labeled peripheral blood mouse monocytes (MouseMonos; 20,000/well) were used in the adhesion assay. Adherent monocytes localized within a 10 ϫ 10 grid at ϫ40 magnification were counted by epifluorescence microscopy. Values that were significantly higher than B6 control values by ANOVA are indicated (#, p Ͻ 0.001; *, p Ͻ 0.01).
We inserted the ribozyme into an adenoviral vector for transfection into primary EC and have previously used this adenoviral ribozyme construct to block 12/15-LO expression in porcine EC (30). Blocking the 12/15-LO pathway in LOTG mice using the 12/15-LO ribozyme or CDC significantly reduced monocyte adhesion (Fig. 10). To also test the hypothesis that 12/15-LO products can directly stimulate monocyte adhesion, B6 control and LOKO EC were treated for 4 h with 100 nM 12S-HETE and then used in an adhesion assay. This concentration of 12S-HETE is believed to be within the physiological range found in blood (see "Discussion") (37)(38)(39)(40)(41)(42). We performed dose-response studies with exogenous 12S-HETE on monocyte/EC adhesion and found, within a dose range of 1-500 nM 12S-HETE, that 100 nM 12S-HETE was the lowest concentration that provided a significant increase in monocyte adhesion to B6 control EC (data not shown). As shown in Fig. 11, addition of 100 nM 12S-HETE to B6 control and LOKO EC signifi-cantly increased monocyte adhesion. The enantiomer 12R-HETE, which is not a product of the 12/15-LO pathway, had no effect on monocyte adhesion. Taken together, these data illustrate that products of the 12/15-LO enzyme are primary mediators of monocyte/endothelial interactions in the vessel wall.
12/15-LO Activity Regulates Expression of Endothelial ICAM-1-To determine how 12/15-LO activity regulates monocyte/EC interactions, we measured expression of key molecules that are known mediators of monocyte adhesion. By flow cytometry, we found that ICAM-1 was increased dramatically on the surface of LOTG endothelium; however, levels of VCAM-1 were unchanged (Fig. 12). The mean fluorescence intensity of ICAM-1 expression on B6 control endothelium was significantly lower compared with expression on LOTG endothelium (B6 control mean fluorescence intensity of 51 versus LOTG mean fluorescence intensity of 189), indicating a dramatic increase in ICAM-1 protein expression on LOTG endothelium. These data suggest regulation of ICAM-1 expression by 12/ 15-LO products. To further test the hypothesis that 12/15-LO could modulate ICAM-1 expression, we quantified ICAM-1 mRNA abundance in LOTG and B6 control EC. Using quantitative PCR, we found a significant 4.5-fold increase in ICAM-1 mRNA (Fig. 13) in LOTG EC and little or no expression of ICAM-1 mRNA in LOKO EC, suggesting that 12/15-LO can directly mediate ICAM-1 expression on the endothelium.
Adhesion in LOTG Mice Is Mediated through VCAM-1, CS-1 Fibronectin (FN), and ICAM-1-Because we found that ICAM-1 expression was increased in LOTG mice, we wanted to determine whether ICAM-1 is the primary mediator of monocyte/EC interactions in these mice. We performed monocyte adhesion assays in the presence of blocking antibodies to either mouse ICAM-1 or ␤ 2 integrin. Blocking ICAM-1 in EC or ␤ 2 integrins in monocytes significantly reduced by 60% (but did not completely inhibit) monocyte adhesion to LOTG EC. Thus, we looked at the contribution of other candidate adhesion molecules to mediating adhesion. We have shown previously that 12S-HETE causes increased deposition of CS-1 FN on the apical surface of human aortic EC (20). Currently, there are no available reagents to measure CS-1 FN expression in the mouse. To indirectly examine the role of CS-1 FN in mediating monocyte adhesion to MAEC, we used a peptide that specifically blocks CS-1-mediated monocyte adhesion (43). This peptide blocks the LDV-binding site for CS-1 FN on VLA-4 (43). As shown in Fig. 14, monocyte adhesion to LOTG EC was decreased by 50% in the presence of the CS-1 FN blocking peptide. Use of a blocking antibody to VCAM-1 (44) also reduced adhesion by 30% (Fig. 14). Use of blocking antibodies to both ␣ 4 and ␤ 2 integrins (44) in monocytes completely prevented adhesion. Counterligands for VLA-4 (␣ 4 ␤ 1 integrin) in EC are VCAM-1 and CS-1 FN (20,45). Counterligands for ␤ 2 integrins include ICAM-1 and ICAM-2. Thus, monocyte adhesion in LOTG mice is mediated through ␤ 2 integrin and ICAM-1 interactions as well as VLA-4 and CS-1 FN/VCAM-1 interactions.

DISCUSSION
Recent studies have defined a role for 12/15-LO products in mediating inflammation and atherosclerosis (7-9, 11-13, 21, 27, 46, 47). However, the mechanisms through which these 12/15-LO eicosanoids act to mediate inflammation and atherosclerosis are still not well understood. Generation of 12/15-LO knockout mice by Funk and co-workers has contributed significantly to our understanding of 12/15-LO (5-7, 9, 12, 21, 32). Often, however, deletion of a key arachidonic acid pathway in mice can result in shunting of arachidonic acid into other eicosanoid pathways, as is the case with 12/15-LO knockout mice (32), thus making data interpretation somewhat difficult. To define the mechanisms by which 12/15-LO activity mediates atherosclerosis, we generated transgenic mice that expressed the murine 12/15-LO gene. Thus, we were able to use the mice as a novel tool to directly examine the direct role of 12/15-LO activity in mediating monocyte/endothelial interactions in vivo. These mice had a 2-fold increase in 12/15-LO protein, which resulted in a modest, yet significant increase in 12/15-LO products in vivo (Fig. 4). We believe that the increased eicosanoid production we measured in vivo was a direct result of overexpression of the 12/15-LO gene in the mice and represents systemic production of these relevant eicosanoid products. However, we cannot rule out that these eicosanoid products were generated indirectly through activation of another pathway, such as through phospholipase A 2 action on cellular phospholipids. Nevertheless, we found no obvious shunting of arachidonic acid to other eicosanoid pathways in the LOTG mice, suggesting that this production was specific to overexpression of the 12/15-LO enzyme.
Our study illustrates that 12/15-LO is a primary mediator of monocyte/endothelial interactions in the vessel wall. This is the first study that indicates that 12/15-LO products directly mediate monocyte adhesion to the endothelium through regulation of expression of key adhesion molecules in vivo. In LOTG mice, we found a significant increase in monocyte adhesion to aortic EC (Fig. 6). This increase in monocyte adhesion occurred even with basal unstimulated EC, suggesting that the LOTG EC were pre-activated to bind monocytes. There was little or no monocyte adhesion to aortic EC from LOKO mice (Fig. 9). Direct inhibition of endothelial 12/15-LO using a 12/15-LO ribozyme construct or a pharmacological 12/15-LO inhibitor blocked monocyte adhesion to LOTG EC (Fig. 10). Finally, FIG. 9. Transgenic EC respond to oxidized LDL. Aortic EC from LOTG line 41F, LOKO, or B6 control mice were incubated in the absence or presence of 250 g/ml oxidized LDL (OXLDL) for 4 h at 37°C, followed by an adhesion assay. TNF-␣ (10 units/ml) was incubated with B6 control MAEC for 4 h as a positive control to show maximal binding. Adherent WEHI monocytes localized within a 10 ϫ 10 grid at ϫ40 magnification were counted by epifluorescence microscopy. *, significantly higher than B6 control values by ANOVA (p Ͻ 0.001); #, significantly higher than unstimulated values for that group by ANOVA (p Ͻ 0.005); **, significantly lower than B6 control values (p Ͻ 0.009). Data represent the means Ϯ S.E. of 10 experiments using six mice/group.  11. Addition of 12S-HETE to EC from 12/15-LO knockout mice restores monocyte adhesion. Aortic EC from B6 control or LOKO mice were incubated for 4 h at 37°C with 100 nM 12S-HETE, 10 units/ml TNF-␣ as a positive control, or 100 nM 12R-HETE as a negative control. Monocyte adhesion was performed as described under "Experimental Procedures" using WEHI monocytes. *, significantly higher than B6 control values (p Ͻ 0.001); #, significantly higher than LOKO values (p Ͻ 0.001); **, significantly lower than B6 control values (p Ͻ 0.009). Data represent the mean Ϯ S.E. of three experiments using six mice/group. addition of nanomolar levels of 12S-HETE (37,39) to EC from LOKO mice significantly increased monocyte adhesion (Fig.  11). Taken together, these studies all illustrate the importance of endothelial 12/15-LO products in directly mediating monocyte/endothelial interactions in the vessel wall.
We chose to use nanomolar concentrations of 12S-HETE in some experiments based upon previously reported measure-ments of levels of 12S-HETE in human plasma and serum (40 -42). In human plasma, levels of 12S-HETE have been reported to be in the range of 1-800 nM, depending on the method of quantification. In stimulated serum in vitro (through activation of platelets and leukocytes), this value increases to the low micromolar range (42). However, the exact concentration of active 12S-HETE is not known, as 12S-HETE can FIG. 13. ICAM-1 mRNA expression is increased in 12/15-LO transgenic mice and decreased in 12/15-LO knockout mice. Aortas were isolated from B6 control (CTR), LOTG line 41F, and LOKO mice. EC mRNA was obtained as described under "Experimental Procedures." Total RNA was reverse-transcribed as described under "Experimental Procedures." EC cDNA was used in a quantitative PCR with primers specific for murine ICAM-1. Data were normalized to cyclophilin as a control. The relative expression of 12/15-LO mRNA in B6 control mice was set to 1. *, significantly higher than B6 control values (p Ͻ 0.0001); **, significantly lower than B6 control values (p Ͻ 0.01). readily bind to plasma proteins. Nevertheless, we used 100 nM to reflect a modest concentration of 12S-HETE that may be relevant to that found in the circulation.
The reasons for the observed "basal" activation of LOTG EC could be as follows: 1) an increase in adhesion molecule (VCAM-1, ICAM-1) expression on the EC surface, 2) an increase in CS-1 FN expression on the EC surface, and 3) and increase in endothelial chemokine production. We found a significant increase in ICAM-1 expression, but little or no increase in VCAM-1 expression, on endothelia of LOTG mice. We also found a significant decrease in ICAM-1 mRNA levels in LOKO mice. These data indicate that 12/15-LO somehow regulates ICAM-1 production. The mechanisms for this regulation are as yet unknown, but may relate to direct effects of eicosanoids on ICAM-1 mRNA expression. We found that blocking ICAM-1 in EC or its counterligand ␤ 2 integrin in monocytes significantly reduced monocyte adhesion to EC. Blocking ICAM-1 in EC reduced monocyte adhesion by 50%, and blocking ␤ 2 integrin reduced monocyte adhesion by ϳ60%. These data suggest that ICAM-1 is important in 12/15-LO-mediated monocyte adhesion. Use of a blocking antibody to VCAM-1 reduced adhesion to EC by ϳ30% (Fig. 13). However, blocking CS-1 FN on the EC surface using a peptide specific for the CS-1 FN-binding site on VLA-4 significantly reduced monocyte adhesion to EC by 60%. Blocking VLA-4 (by blocking binding sites for both VCAM-1 and CS-1 FN) reduced monocyte adhesion to LOTG EC by 80%. Additional studies will be needed to define the exact contributions of CS-1 FN versus VCAM-1 to mediating monocyte/endothelial adhesion in LOTG mice. We also need to examine the role of ICAM-2 in mediating monocyte adhesion because it is also a ligand for ␤ 2 integrins. Nevertheless, these studies reveal that ICAM-1/␤ 2 integrin interactions as well as VCAM-1/CS-1 FN/␣ 4 integrin interactions both play a role in 12/15-LO-mediated monocyte adhesion to EC in LOTG mice. We are currently in the process of examining the role of 12/ 15-LO activity in the modulation of endothelial chemokine production in LOTG mice, as this also contributes to monocyte/ endothelial interactions.
We cannot yet rule out contributions of the platelet 12-LO enzyme to mediating monocyte/endothelial interactions in vivo. The platelet 12-LO pathway also generates 12S-HETE and is localized within circulating platelets. In our studies using isolated primary cultures of aortic EC from LOTG and LOKO mice, the platelet 12-LO pathway would not be a significant factor. However, in vivo, platelet 12-LO activity may be important for atherogenesis. Studies defining the role of platelet 12-LO in atherogenesis are currently underway in the laboratory.
In summary, overexpression of murine 12/15-LO in mice increases monocyte/endothelial interactions. These monocyte/ endothelial interactions in LOTG mice are caused by endothelial CS-1 FN and VCAM-1 interactions with VLA-4 in monocytes and by endothelial ICAM-1 and ␤ 2 integrin interactions in monocytes. LOTG mice have significant up-regulation of endothelial ICAM-1 expression, suggesting that 12/15-LO regulates ICAM-1 production. LOTG mice develop spontaneous aortic fatty streak lesions on a rodent chow diet. Thus, the LOTG mouse is a new model that should provide novel mechanistic information regarding the role of 12/15-LO in mediating atherosclerotic vascular disease in vivo.