Leukotriene B4 pathway activation and atherosclerosis in obstructive sleep apnea.

Leukotriene B(4) (LTB(4)) production increases in obstructive sleep apnea syndrome (OSA) and is linked to early vascular remodeling, the mechanism of which is unknown. The objective of this study was to to determine the molecular mechanisms of LTB(4) pathway activation in polymorphonuclear cells (PMNs) and early vascular remodeling in OSA and the specific contribution of intermittent hypoxia (IH). PMNs were isolated from 120 OSA patients and 33 healthy subjects and used for measurements of LTB(4) production, determination of mRNA and protein expression levels, or exposed for four cycles of in vitro IH. PMNs derived from OSA patients exhibited increased LTB(4) production, for which apnea-hypopnea index was an independent predictor (P=0.042). 5-Lipoxygenase-activating protein (FLAP) mRNA and protein increased significantly in PMNs from OSA patients versus controls and were associated with carotid luminal diameter and intima-media thickness. LTB(4) (10 ng/ml) increased IL-6 (P=0.006) and MCP-1 (P=0.002) production in OSA patient monocytes. In vitro exposure of PMNs from controls to IH enhanced FLAP mRNA levels (P= 0.027) and induced a 2.7-fold increase (P=0.028) in LTB(4) secretion compared with PMNs exposed to normoxia. In conclusion, upregulation of FLAP in PMNs in response to IH may participate in early vascular remodeling in OSA patients, suggesting FLAP as a potential therapeutic target for the cardiovascular morbidity associated with OSA.

Subjects were enrolled regardless of previous cardiovascular medical history, hence refl ecting routine clinical activity. They were nonoverlapping with cardiovascular disease-free subjects included in our previous study ( 4 ).
After the nocturnal polysomnography, peripheral blood was sampled. Plasma glucose and serum triglycerides levels were measured on an automat (Modular 700, Roche, Meylan, France). Serum insulin was measured by radioimmunometric sandwich assay (CIS bio international, Gif-Sur-Yvette, France). Serum highsensitivity C-reactive protein (hs-CRP) was measured by automated immunonephelometry (Behring Nephelometer II Analyzer, Dade Behring, Germany). Carotid ultrasonography was performed in 29 OSA patients and four controls as described ( 1 ).

PMN preparation
Human PMNs were isolated as previously described ( 4 ) and used for one of the following protocols: 1 ) PMNs were resuspended in PBS, pH 7.4 containing 0.133 g/l CaCl 2 and 0.1 g/l Mg 2+ for A23187 stimulation (68 OSA patients and 19 controls); 2 ) resuspended in RNAlater (Ambion) for RT-PCR (25 OSA patients and nine controls); or 3 ) resuspended in PBS with protease inhibitors (Sigma) for Western blot (12 OSA patients and fi ve controls). LTB 4 production by PMNs was measured on stimulation with A23187 ( 4 ). Viability exceeded 98% by Trypan blue exclusion method. PMNs (2 × 10 6 cells/ml) were incubated for 15 min at 37°C with 10 µmol/l A23187 or vehicle (PBS). In some experiments, PMNs were incubated with the 5-LO inhibitor AA861 (10 µM) or the FLAP inhibitor MK886 (10 µM) for 30 min before A23187 stimulation. Incubations were stopped by centrifugation In addition to monocytes and macrophages, PMNs mediate the onset and progression of atherosclerosis ( 14 ). As the principal source of LTB 4 , the capacity of PMNs to produce LTB 4 is measured by ex vivo production of LTB 4 in response to calcium ionophore; LTB 4 production by stimulated PMNs rises in patients with a history of myocardial infarction ( 15 ) and in nonobese cardiovascular disease-free OSA patients ( 4 ).

LTB 4 production in A23187-treated PMNs
Recent fi ndings suggest that intermittent hypoxia (IH) induces LTB 4 activation in monocytes in vitro and that the LTB 4 pathway contributes in the development of atherosclerosis in chronic IH-exposed ApoE Ϫ / Ϫ mice ( 16 ). However, the effects of OSA on LTB 4 pathway transcription and atherosclerosis have not been examined in humans.
The aims of the present study were: 1 ) to examine PMN production of LTB 4 in a large cohort of OSA patients refl ecting routine clinical practice; 2 ) to determine the molecular mechanisms of LTB 4 pathway activation and early vascular remodeling in OSA; 3 ) to study the consequences of LTB 4 pathway activation in OSA in terms of paracrine effect on monocytes, another cellular type greatly involved in atherogenesis; and 4 ) to explore the contribution of the hypoxic component of OSA on LTB 4 pathway activation in isolated PMNs exposed to in vitro IH.

Patients
This study was approved by the local ethics committee per the Declaration of Helsinki. All participants gave written informed consent. Patients were referred to the Grenoble University Hospital sleep laboratory for suspicion of OSA. Controls were healthy volunteers who were free of infl ammatory and sleep disorders. All subjects underwent a full polysomnography, which was scored blinded to biological parameters, and biochemical measures were analyzed blinded to these results. A fl ow chart detailing patient inclusion in the different experiments is shown in Fig. 1 .
The exclusion criteria were: cancer, infectious or autoimmune disease, diabetes mellitus, disease that potentially affected blood from healthy subjects were analyzed by Western blot and membranes probed with rabbit GAPDH antibody (AbCam) to confi rm that loading control did not differ between samples.

In vitro exposure of PMNs from healthy subjects to IH
Purifi ed PMNs from eight healthy subjects underwent four cycles of IH using a modifi ed protocol ( 20 ). In a hypoxia chamber with atmospheric pressure maintained, a 35 min hypoxic period (95% N 2 and 5% CO 2 ) was followed by 25 min of reoxygenation (95% O 2 and 5% CO 2 ), after which the cells were resuspended in RNAlater (Ambion) for LTB 4 pathway RT-PCR analysis (n = 4) or in PBS for A23187 1 µM stimulation (n = 6). Control-PMNs from the same donors were maintained in normoxic conditions for the same durations.

Isolation of monocytes and cytokine production in response to LTB 4
Monocytes were isolated using standard methods ( 21 ). Briefl y, after separation by dextran sedimentation and centrifugation through a discontinuous fi coll, mononuclear cells were placed on plastic tissue culture dishes (Falcon 3003) precoated with pooled human serum for 15 min at 37°C. After 2 h culture at 37°C in RPMI 1640 with 10% pooled human serum, nonadherent cells were removed. Plastic adherent cells (monocytes) were collected by scraping , washed twice in RPMI, and suspended in 2 ml RPMI to be counted.

FLAP, 5-LO, LTA 4 H, and BLT 1 and BLT 2 receptor mRNA levels in PMNs
Total mRNA was isolated from PMNs using the RNeasy kit (Qiagen) as described ( 18 ) and reverse-transcribed using Superscript II (Invitrogen, Carlsbad, CA) with random hexamers per the manufacturer's instructions. Quantitative TaqMan PCR was performed on a 7900HT using primer/probe pairs designed with Assay-On-Demand™ (both Applied Biosystems), as indicated in supplementary Table I . Data were normalized to cyclophilin A and GAPDH mRNA and expressed as 2

FLAP and LTA 4 H proteins expression in PMNs from OSA
PMNs were subjected to three freeze/thaw cycles in liquid nitrogen and ultracentrifuged for 30 min at 100,000 g The supernatant (cytosolic extract) was collected, and the pellet was resuspended in 500 µL PBS. After ultracentrifugation at 100,000 g for 30 min, the pellet was resuspended in RIPA containing protease inhibitors (Sigma). Protein concentration was measured by Bradford assay.
Ten micrograms of proteins (cytosolic and membrane extract) was resolved by 12% SDS-PAGE; transferred onto nitrocellulose membranes; blocked with 5% milk powder in TBS, pH 7.4, containing 0.1% Tween 20; probed with polyclonal anti-FLAP for membrane extract or goat anti-LTA 4 H for cytosolic extract (200 ng/ml, Santa Cruz Biotechnology) and peroxidase-conjugated secondary anti-rabbit or -goat (1:25,000 Jackson ImmunoResearch Laboratories); and detected by ECL as previously described ( 19 ). In separate experiments, cytosolic fractions from PMNs derived by AHI. A23187-induced LTB 4 production increased with OSA severity, rising signifi cantly in severe OSA patients versus controls ( Table 1 ). Pretreatment with AA861 or MK886 inhibited A23187-mediated LTB 4 synthesis (data not shown).

Confounders of LTB 4 production in OSA patients
LTB 4 production correlated signifi cantly with AHI and percentage of time spent with SaO 2 < 90% ( Table 2 ). LTB 4 concentrations were unrelated to age, body mass index (BMI), or metabolic variables. Gender, smoking status, and different drug treatments (lipid-lowering, antihypertensive, or antiplatelet treatments) did not infl uence LTB 4 concentration. A multiple-linear regression including pharmacological treatments, AHI, glycemia, and LDL-cholesterol (variables with P -value < 0.2 in the univariate analysis) indicated that AHI ( P = 0.042) was an independent predictor of log LTB 4 concentrations but this model merely explained 14% of the variance.

Increased FLAP mRNA expression in OSA patient PMNs
Supplementary Table II shows the baseline characteristics of the nine controls and 25 OSA patients included in the RT-PCR experiments. Among the patients included in the mRNA study, 13 were severe, eight were moderate , and four were mild OSA patients. FLAP mRNA levels were signifi cantly higher in PMNs derived from OSA patients versus controls ( Fig. 3A ). Conversely, 5-LO, LTA 4 H ( Fig. 3A ), Human Cytokine Panel, R & D Systems, Minneapolis, MN) on a Bioplex 200 (Bio-Rad Laboratories, Hercules, CA) using Luminex xMAP™ Technology (Luminex Corporation, Austin, TX,) as described ( 22 ).

Statistical analysis
Statistical analyses were performed using NCSS97 (Kaysville, UT). Continuous data were expressed as median and 10th and 90th percentiles. Noncontinuous data were expressed as numbers and percentages and compared by chi-square test. When necessary, LTB 4 was log-transformed to normalize data or appropriate nonparametric tests were used (Spearman correlation coefficient, Kruskall-Wallis method, and Mann-Whitney U test). LTB 4 production between OSA subgroups stratifi ed by AHI and controls was compared by Kruskall-Wallis method and subsequent pairwise comparisons were made by nonparametric Bonferroni multiple comparison test. The impact of pharmacological treatments and polysomonographic parameters on log-LTB 4 concentrations was analyzed by multiple regression. Differences between chemokine and cytokine concentrations at baseline and after challenge with LTB 4 were analyzed by Wilcoxon signed-rank test. P < 0.05 was considered signifi cant. Table 1 shows the baseline characteristics of subjects in whom A23187-induced stimulation of LTB 4 was measured, stratifi ed    was examined by Western blot. Supplementary Table III shows the baseline characteristics of the fi ve controls and 12 OSA patients included in the Western blot experiments. In line with the PCR results, FLAP increased in the membrane fraction of PMNs derived from OSA patients versus controls, whereas cytosolic LTA 4 H did not significantly differ ( Fig. 3B ). FLAP expression correlated with AHI ( r = 0.536, P = 0.03) but not with RDI, min SaO 2 , and mean SO 2 . FLAP ( r = 0.664, P = 0.01) and LTA 4 H ( r = 0.677, P = 0.0098) levels were associated with BMI.

Specifi c effect of IH on the LTB 4 pathway
In vitro exposure of PMNs from healthy subjects to IH consisting of four cycles of 35 min hypoxia followed by 25 min reoxygenation increased FLAP and LTA 4 H mRNA levels ( Fig. 4A ) versus PMNs under normoxic conditions. In contrast, IH did not alter 5-LO ( Fig. 4A ) or BLT 1 or BLT 2 receptor mRNA (data not shown). As shown in Fig. 4B , LTB 4 secretion in response to A23181 challenge was signifi cantly 2.7-fold enhanced in PMNs exposed to IH (ng/ml/2.10 6 cells): 1.6 ± 0.4 (normoxia conditions) versus 3.3 ± 0.4 (hypoxic conditions).

Associations of the LTB 4 pathway with atherosclerosis
The correlations between PMN mRNA and protein levels for FLAP and LTA 4 H with measures collected at carotid artery sonography are shown in Table 3 . FLAP mRNA and protein in circulating PMNs correlated signifi cantly with mean luminal diameter and mean intima-media thickness (IMT) of common carotid arteries ( Table 3 ). Similarly, and BLT 1 and BLT 2 receptor mRNA (data not shown) did not differ signifi cantly between the groups. Although FLAP mRNA and OSA severity did not correlate signifi cantly, there was a trend correlation of FLAP mRNA with percentage of time spent with SaO 2 < 90% ( r = 0.358, P = 0.0613). FLAP mRNA was unrelated to BMI ( P = 0.276).

Increased FLAP protein expression in OSA patient PMNs
Because FLAP and LTA 4 H mRNA levels rose in OSA patients, protein expression of these LTB 4 pathway components

DISCUSSION
Our results point to an important role of the LTB 4 pathway in PMNs for atherosclerosis associated with OSA. Transcriptional alterations of the LTB 4 pathway in PMNs, with a major contributing role of IH, may represent a potential molecular mechanism of leukotriene-induced atherogenesis in OSA. In particular, this study is the fi rst to correlate subclinical atherosclerosis with expression levels of the LTB 4 pathway in PMNs and to show that IH increased LTB 4 production in PMNs associated with greater FLAP mRNA and protein expression in OSA patients. Collectively, these data suggest an important role of the LTB 4 pathway in sleep apnea-related atherosclerosis.
Increased LTB 4 production in A23187-stimulated PMNs has previously been demonstrated in patients with a history of myocardial infarction ( 15 ) and in cardiovascular diseasefree OSA patients ( 4 ). The present study showed, for the fi rst time, increased LTB 4 production in A23187-stimulated PMN 5-LO mRNA was associated with carotid luminal diameter and IMT ( Table 3 ).
Although LTA 4 H mRNA levels in PMNs did not correlate with these markers of early vascular remodeling, LTA 4 H protein in PMNs was associated with right carotid luminal diameter and left IMT ( Table 3 ). BLT 1 and BLT 2 mRNA in PMNs and monocytes was unrelated to these markers (see supplementary Table IV).

Paracrine effects of LTB 4 pathway activation in OSA patients
LTB 4 (10 ng/ml) increased IL-6 and MCP-1 production in OSA patient monocytes ( Fig. 5 ). Conversely, LTB 4 did not alter TNF-␣ or RANTES secretion from monocytes. , and expressed as fold change compared with NX) B/A23187-mediated LTB 4 production by PMNs exposed to in vitro IH or NX.  chemotactic and proliferative effects of LTB 4 on vascular smooth muscle cells ( 25 ). The association of transcriptional levels of the LTB 4 pathway in PMNs with atherosclerosis and vascular remodelling is in line with PMNs being a major source of LTB 4 production. However, because monocytes/macrophages are major effectors in atherosclerosis, it is also important that our results indicate that LTB 4 production in PMNs may act in a paracrine way to induce proinfl ammatory IL-6 and MCP-1 in monocytes. These fi ndings are supported by the upregulation of leukotriene B 4 receptors by IH in THP-1 cells ( 16 ). Similarly, proatherosclerotic MCP-1 levels rise in monocytes after sleep in severe OSA patients ( 26 ) and IH-increased MCP-1 expression is markedly attenuated by BLT 1 receptor antagonist ( 16 ). Thus, LTB 4induced proinfl ammatory monocyte signaling might be a link between PMN-derived LTB 4 and atherosclerosis.
Although our results link IH in OSA patients and LTB 4 pathway activation in PMNs, leading to LTB 4 -induced proinfl ammatory effects in monocytes and atherosclerosis, it must be acknowledged that atherosclerosis is a multifactorial disease. Due to confounders and cardiovascular risk factors in our patients, we cannot exclude that LTB 4 pathway activation by IH may not be the sole mechanism of vascular remodeling. Notably, LTA 4 H expression correlated with BMI but not oxygen desaturation, whereas FLAP expression was infl uenced by oxygen desaturation and obesity as reported ( 5,18 ). Similarly, FLAP mRNA and protein levels were signifi cantly higher in PMNs derived from OSA patients versus controls, which supports the suggested mechanism, but our data demonstrated only trend correlation of FLAP mRNA with percentage of time spent with SaO 2 < 90%. In addition, because the protein analysis did not include an internal control for each sample, it cannot be fully excluded that subtle differences in protein loading may have infl uenced the evaluation of protein levels. Finally, the presence of cardiovascular risk factors in our PMNs from severe OSA patients presenting cardiovascular comorbitities as seen in clinical practice and its correlation with AHI and percentage of total sleep time with SaO 2 < 90%.
Although the present and previous ( 4, 15 ) studies demonstrate LTB 4 pathway activation in PMNs and suggested an association with cardiovascular disease and OSA, the molecular mechanisms have remained largely unexplored. In the present study, the increase in LTB 4 production in OSA was associated with an increased mRNA expression of FLAP, suggesting transcriptional activation of LTB 4 pathway components in OSA patients, which was also translated into higher protein levels. Our in vitro exposure of PMNs to IH, which increased LTB 4 production and induced FLAP upregulation, suggests that IH directly participates in LTB 4 pathway activation in OSA. The latter results are consistent with recent fi ndings in another cell type, showing that in vitro exposure of monocyte THP-1 cells to IH increases expression levels of LTB 4 synthesizing enzymes ( 23 ). Although IH in vitro might not refl ect the repetitive desaturationreoxygenation sequences in OSA, hypoxic conditions in vitro are linked to NF-kB and HIF1 ␣ activation, delaying PMN apoptosis ( 20 ) and upregulating FLAP ( 24 ). LTB 4 pathway activation in OSA may help to explain the association between OSA and atherosclerosis. In the present study, the expression levels of LTB 4 synthesizing enzymes correlated with fi ndings on carotid ultrasound, linking, for the fi rst time, transcription of the LTB 4 pathway in peripheral leukocytes with atherosclerosis and early vascular remodeling. For example, 5-LO transcript levels were higher in PMNs derived from subjects with carotid atherosclerotic plaques compared with those derived from subjects without atherosclerosis. The latter fi nding is in line with the proinfl ammatory and proatherogenic effects of LTB 4 , which have been previously established ( 8,9 ). In addition, the correlation between FLAP and 5-LO expression in PMNs with carotid wall hypertrophy, measured as IMT, is consistent with the direct population may contribute to blunt the correlation between hypoxia severity and FLAP mRNA levels. In particular, our observations support that obesity is a major confounding factor of the infl ammatory state in OSA patients. However, considering BMI and other potential cardiovascular comorbidities, AHI remained an independent predictor of LTB 4 production in OSA patients by multivariate analysis.
In summary, we have demonstrated activation of the LTB 4 pathway in PMNs from OSA patients through transcriptional upregulation, which correlated with carotid atherosclerosis and IMT. LTB 4 -induced proinfl ammatory monocyte signaling might be a link between PMN-derived LTB 4 and atherosclerosis. Together with the upregulation of FLAP and LTA 4 H in PMNs exposed in vitro to IH, our data provide evidence that IH is a major feature of OSA involved in LTB 4 pathway activation underlying vascular remodeling and atherosclerosis. Our results implicate LTB 4 pathway, notably FLAP, as a therapeutic target in OSA-associated metabolic and cardiovascular morbidity.