Enhanced vascular permeability facilitates entry of plasma HDL and promotes macrophage-reverse cholesterol transport from skin in mice.

Reverse cholesterol transport (RCT) pathway from macrophage foam cells initiates when HDL particles cross the endothelium, enter the interstitial fluid, and induce cholesterol efflux from these cells. We injected [(3)H]cholesterol-loaded J774 macrophages into the dorsal skin of mice and measured the transfer of macrophage-derived [(3)H]cholesterol to feces [macrophage-RCT (m-RCT)]. Injection of histamine to the macrophage injection site increased locally vascular permeability, enhanced influx of intravenously administered HDL, and stimulated m-RCT from the histamine-treated site. The stimulatory effect of histamine on m-RCT was abolished by prior administration of histamine H1 receptor (H1R) antagonist pyrilamine, indicating that the histamine effect was H1R-dependent. Subcutaneous administration of two other vasoactive mediators, serotonin or bradykinin, and activation of skin mast cells to secrete histamine and other vasoactive compounds also stimulated m-RCT. None of the studied vasoactive mediators affected serum HDL levels or the cholesterol-releasing ability of J774 macrophages in culture, indicating that acceleration of m-RCT was solely due to increased availability of cholesterol acceptors in skin. We conclude that disruption of the endothelial barrier by vasoactive compounds enhances the passage of HDL into interstitial fluid and increases the rate of RCT from peripheral macrophage foam cells, which reveals a novel tissue cholesterol-regulating function of these compounds.


Materials
Evans blue dye, histamine, serotonin, bradykinin, the mast cell-degranulating compound 48/80 (48/80), the histamine H1 receptor (H1R) antagonist pyrilamine, and the histamine H2 receptor (H2R) antagonist ranitidine were purchased from Sigma Aldrich. [1 ␣ ,2 ␣ (n)- 3 H]cholesterol (specifi c activity 40-60/mmol) was purchased from Perkin Elmer. Animals C57BL/6JOlaHsd mice (age 9-12 weeks) were purchased from Harlan Laboratories (Venray, Holland). ApoA-I-defi cient (apoA-I-KO) mice on the C57BL/6J background (age 20-30 weeks) were kindly provided by Dr. Jesús Osada (University of Zaragoza, Spain) and LDL receptor-defi cient (LDLr-KO) mice (age 9-16 weeks) on the C57BL/6J background were obtained from Jackson Laboratory (Bar Harbor, ME; Stock Number: 002207). Mice were housed in groups of fi ve per cage, allowed to access regular chow rodent diet (Teklad Global, Harlan Laboratories) and water ad libitum and were maintained in an automatic 12/12 h dark-light cycle. All experiments were conducted in compliance with national guidelines for the care and use of research animals, and the experimental protocols were approved by the Animal Experiment Board of Finland (ELLA) and by the Institutional Animal Care Committee of the Institut de Recerca de l'Hospital de la Santa Creu I Sant Pau, Spain.

Skin permeability test and measurement of the histamine content of skin
The fur on the dorsal back was shaved 2 days prior the experiment to prevent skin irritation on the experimental day. For the permeability test, mice were given an intraperitoneal injection of Evans blue dye in saline (2%; 4 ml/kg) ( 16 ), and the dye was allowed to circulate for 4 h. After 30 min of Evans blue administration, mice received subcutaneous injection of histamine (250 µg/kg in 100 µl saline) or vehicle alone in the rostral back. The degree of Evans blue extravasation was quantitated by measuring the formed blue-colored circle at 0.5, 1, 2, 3, and 4 h and documenting it by photography. The skin content of histamine was measured after 1 h of subcutaneous injection of histamine or vehicle. Mice were euthanized, and a piece of skin (area of 1 cm 2 ) was dissected from the injection site, weighted, and chopped up with scissors. The fi nely cut pieces were placed in 500 µl of distilled water in Eppendorf tubes and then subjected to fi ve freeze-thaw cycles (liquid nitrogen-warm tap water). TCA was then added to the tubes (fi nal concentration 5%), after which they were thoroughly vortexed, incubated at +4°C for 30 min, and fi nally centrifuged at 10,000 g for 10 min. The clear supernatant was collected, the histamine content was determined fl uorometrically ( 17 ), and the results were expressed as nanograms of histamine per area and per weight of skin.

Measurement of mast cell activation in skin
To induce degranulation of mast cells in the skin, 48/80 (2.5 mg/kg) was subcutaneously injected to mice. Mast cells were defi ned to be activated when extracellular granules were found in the immediate vicinity of the cells. For quantifi cation of the number of activated mast cells, a piece of skin was cut from the treated skin sites and submerged in the tissue section medium Tissue-Tek OCT (Sakura Finetek Europe). The skin samples were then rapidly frozen in liquid nitrogen and stored at Ϫ 80°C to be further sectioned into 5 µm slices. The slices were stained in 0.1% Toluidine Blue, and the number of activated mast cells was cholesterol to HDL initiates RCT in all tissues, but only the tiny fraction of the total body RCT that originates in the cholesterol-loaded macrophage foam cells [macrophage-RCT (m-RCT)] located in the intima is directly involved in atherosclerosis ( 5 ). It currently appears that the dynamics of the cholesterol fl ux induced from macrophage foam cells to feces by HDL particles, and their main apolipoprotein constituent apoA-I, is critical for their cardioprotective function ( 6 ). Indeed, the capability of HDL to induce regression of atherosclerosis has been demonstrated in clinical studies in which infusion of apoA-I-containing reconstituted HDL reduced the atheroma volume or improved the plaque characterization score in patients who had suffered an acute coronary syndrome ( 7,8 ).
The skin is the largest organ of the body, and its cholesterol content has been directly associated with an increased carotid intima-media thickness in individuals without diagnosed cardiovascular disease, suggesting that measuring skin cholesterol content may be useful when attempting to identify individuals with subclinical asymptomatic atherosclerosis (9)(10)(11). Importantly, because deposition of cholesterol in skin xanthomas and in early atherosclerotic lesions has been suggested to share common pathogenic mechanisms ( 12 ), a close resemblance can be postulated between skin and the intimal layer of human arteries, two sites where foam cells are formed. In cells that reside in peripheral tissues such as the skin, RCT is initiated by the cholesterol effl ux-inducing abilities of the HDL particles present in the cutaneous interstitial fl uid. Notably, it has been recently found that the lymphatic vasculature, by facilitating removal of HDL from interstitial fl uids, stimulates cholesterol clearance from the skin and the arterial wall and, accordingly, enables RCT from these sites ( 13,14 ). These observations support the notion that an impaired outward fl ow of HDL from the interstitial fl uid to lymph shares common features in the skin and the arterial wall and so infl uences cholesterol balance in these two tissues.
We have previously shown that an increase in the endothelial permeability of dorsal skin sites in rats induced by histamine increases the transendothelial passage of plasma LDL to the affected sites ( 15 ). Here, we hypothesize that an increase in the transendothelial passage of circulating HDL into tissue sites in which foam cells are present would enhance the local interstitial pool of HDL and promote m-RCT. To test this hypothesis, we evaluated the effect of increased vascular permeability in the skin on the rate of RCT from J774 foam cells injected into the dorsal subcutaneous layer of the skin in mice. We found that increasing the subcutaneous levels of infl ammatory mediators capable of increasing endothelial permeability, namely histamine, serotonin, and bradykinin, locally stimulated uptake of HDL by the skin and also accelerated the rate of m-RCT from skin to feces. The present results reveal that sole stimulation of HDL transfer from plasma to peripheral interstitial fl uids is capable of promoting m-RCT in vivo. was expressed as a fraction of the injected dose. In other experiments, the m-RCT rate from skin to feces was evaluated in wildtype mice upon triggering release of endogenous histamine in skin by subcutaneous dose of 48/80 (2.5 mg/kg) and in histamine-treated LDLr-KO mice. We also evaluated the effect of histamine in the setting of low HDL levels. In this experiment, apoA-I-KO mice were treated with the studied subcutaneous dose of histamine followed by an intravenous injection of HDL (100-150 µg protein, 95-145 nmol cholesterol) isolated from wild-type mice, which partly ( ‫ف‬ 20%) and transiently reconstituted the basal levels of HDL-cholesterol in these mice, which were 0.6-0.8 mM. Groups of mice that received only subcutaneous histamine or only intravenous HDL acted as reference groups.
Gene expression of cholesterol effl ux transporters in cultured J774 macrophages and determination of macrophage cholesterol effl ux to mouse serum in vitro J774 cells were cultured in RPMI 1640 medium supplemented with 10% FBS, 100 U/ml penicillin, and 100 U/ml streptomycin (BioWhittaker, Verviers, Belgium), seeded in 12well plates at 5 × 10 5 cells/well, and grown for 3 days. To mimic the in vivo fl ow and ensuing dilution and predicted decline in concentration of the subcutaneous injected compounds (histamine, serotonin, bradykinin) occurring in the interstitial fl uid of the skin to which the labeled macrophages were exposed, the J774 cells were incubated with decreasing concentrations of each compound for various intervals of time, up to 3 h. Accordingly, the initial concentration of each compound in the incubation medium was the same as that injected to the mice; after 5 min of incubation, the compound concentration was progressively reduced within intervals of 10 min to achieve up to a 100fold reduction at 30 min of incubation. After such a 30 min preconditioning period, the incubation was prolonged until 3 h at this low concentration. Macrophages incubated in only medium (nonconditioned cells) acted as controls. The preconditioned cells were then collected for quantitative RT-PCR analysis. Total RNA was isolated using RNeasy Plus Mini Kit (Qiagen, Hilden, Germany). cDNA was generated using Oligo(dT)23 (Sigma-Aldrich, St. Louis, MO) and M-MLV Reverse Transcriptase, RNase H Minus, Point Mutant (Promega, Madison, WI) and was subjected to quantitative RT-PCR amplifi cation using the TaqMan Master Mix (Applied Biosystems, Foster City, CA). Specifi c TaqMan probes (Applied Biosystems) were used for each gene: Mm00442646_m1 for Abca1 , Mm00437390_m1 for Abcg1 , and Mm00450236_m1 for Scarb1 [scavenger receptor class B type I (SR-BI)]. Gapdh was used as the reference gene (Mm99999915_ g1). Reactions were run on a CFX96 TM Real-Time System (Bio-Rad, Hercules, CA) according to the manufacturer's instructions. The relative mRNA expression levels were calculated using the ⌬ ⌬ Ct method.
In a parallel experiment, in vitro cholesterol effl ux from [ 3 H] cholesterol-labeled J774 macrophage foam cells was determined after the cells had been preconditioned for 30 min with decreasing concentrations of the vasoactive compounds, as described above. For this purpose, aliquots of the normal pool of mouse serum (2.5% fi nal concentration in medium) were added to the preconditioned macrophages in media containing the respective vasoactive compound at the fi nal (100-fold reduced) concentration, and cholesterol effl ux was determined during a 3 h incubation period. Effl ux from nonconditioned macrophages to mouse serum in the absence of vasoactive compounds acted as the reference. Fractional cholesterol effl ux (%) was calculated as dpm medium /(dpm cells + dpm medium ) × 100, as described before ( 20 ). Effl ux values in the absence of serum were subtracted from those in the presence of serum. determined microscopically at 20× and expressed as percent of the total number of mast cells, as described previously ( 18 ).

Stimulation of the vascular permeability of mice
The targeted site of skin for the multiple subcutaneous injections was confi ned to a skin area of a diameter of about 1 cm in the rostral back between the scapulae. Mice received subcutaneous injections of either histamine (250 µg/kg), serotonin (500 µg/kg), or bradykinin (500 µg/kg) in 100 l of saline, or vehicle alone as control. To further stimulate vascular permeability, the same dose of each respective compound was administered after 24 h of the initial dose. To block the H1R or the H2R, or both, mice were intraperitoneally injected with either pyrilamine (20 mg/kg), ranitidine (20 mg/kg), or with both histamine receptor antagonists, respectively, 1 h prior to histamine injection. Each experiment with the vasoactive compounds was repeated at least two times, and the number of mice ranged from 10 to 15 per treatment, as indicated in the fi gure legends.  dpm/mouse). Minimal volume of blood (ranging from 10 to 50 l) was drawn from the saphenous vein at 2 min, 30 min, 60 min, 3 h, 24 h, and 48 h. Plasma decay curves for the tracer were normalized to the serum radioactivity determined at the 2 min time point after injection of the radioactive tracer. Mice were euthanized at 1 h, 24 h, and 48 h, and samples of skin were dissected from the site of the subcutaneous injection. Tissues were rinsed with saline, fi nely chopped, and incubated overnight in 1 ml of 2 N KOH at 37°C. Radioactivity in serum and tissue extracts was measured by liquid scintillation counting (LSC).

Evaluation of m-RCT from skin to feces
We determined the in vivo m-RCT rate from skin to feces within a 48 h period, as previously described ( 20 ). In brief, [ 3 H] cholesterol-labeled J774 mouse macrophages (3-5 × 10 6 cells; 2-5 × 10 6 dpm per mouse) were subcutaneous injected into the rostral back area of mice, and immediately thereafter, the various vasoactive compounds were injected into an adjacent skin site. To further stimulate vascular permeability, the dose of each vasoactive compound was administered two times within a 48 h period, as described above. After the treatments, mice were maintained for 48 h in individual mesh bottom cages to quantitatively collect the fecal material. Finally, mice were exsanguinated by cardiac puncture under terminal isofl urane anesthesia and livers removed after rinsing with saline. Total serum [ 3 H]cholesterol radioactivity and HDL-associated [ 3 H] cholesterol radioactivity were determined by LSC after precipitation of apoB-lipoproteins with phosphotungstic acid and magnesium chloride (Roche Diagnostics GmbH, Germany), as previously described ( 20 ). Total hepatic and fecal lipids were extracted with isopropanol:hexane. The lipid-containing layer was collected, solvent was evaporated, and [ 3 H]tracer radioactivity was measured by LSC. The [ 3 H]tracer detected in fecal bile acids was determined in the remaining aqueous portion of the fecal extracts. The quantity of [ located within the target skin area. After 48 h of injection of the cells, the transfer of macrophage-derived [ 3 H]cholesterol to feces was evaluated ( Fig. 1 ). In comparison with control mice that received only vehicle, treatment with histamine signifi cantly increased the content of radiolabeled cholesterol in serum and HDL and also tended to increase it in the liver ( Fig. 1A, B ). More importantly, the quantity of macrophage-derived [ 3 H]cholesterol excreted into feces during the 48 h period was 2-fold higher in these mice in which the immediate vasoactive histamine stimulus was repeated at 24 h ( Fig. 1C ). We further evaluated whether the effect of histamine on m-RCT involved the H1R and/ or H2R, which have been shown to stimulate the transendothelial transport of LDL from plasma to skin of rats in vivo ( 15 ). For this purpose, mice received 1 h before the histamine dose an intraperitoneal injection of either pyrilamine or ranitidine to antagonize H1R and H2R, respectively, and the fecal recovery of the macrophage-derived 3 H-radioactivity was measured. As shown in Fig. 2 , the stimulatory effect of histamine on the m-RCT from skin to feces was signifi cantly reduced when the H1R antagonist

Other analyses
Total cholesterol, HDL-cholesterol, and triglyceride concentrations in mouse serum were measured enzymatically with commercial kits (Roche Diagnostics, Basel, Switzerland).

Statistical analysis
GraphPad Prism 4.0 software (GraphPad, San Diego, CA) was used to perform all statistical analyses. Unpaired t -test with Welch's correction was used to evaluate statistical differences between two groups. One-way ANOVA with Bonferroni's multiple comparison test was used to compare data from three or more groups. A P value <0.05 was considered statistically signifi cant. All data are expressed as the mean ± SEM.

Subcutaneous injection of histamine rapidly increases the transport of Evans blue from blood to skin
We fi rst evaluated the time dependency of subcutaneous dose of histamine (250 g/kg), a well-known vasodilator agent, to induce vascular permeability in the skin of mice. For this purpose, mice received an intraperitoneal injection of Evans blue dye and 30 min later subcutaneous injections of histamine or saline into the interscapular area of the back. Histamine induced a fast (within 10 min) local permeability reaction refl ected by the appearance of a blue-colored area in the skin around the injection site, whereas injection of saline alone was without effect (n = 5 mice/group). The diameters of the colored circles in the histamine-treated mice were 16 ± 2.4 mm (at 10 min) and 18 ± 3.3 mm (at 30 min). One hour after histamine injection, it was no more possible to measure accurately the magnitude of the generated area because of extensive diffusion of the dye in the dorsal skin. Moreover, at that time the histamine contents in the treated skin areas were equally low in histaminetreated and vehicle-treated mice (0.39 ± 0.02 and 0.38 ± 0.04 ng/mg of skin, respectively; n = 6 mice/group), refl ecting fast removal of the injected histamine from the injection site.

Histamine markedly stimulates fecal excretion of [ 3 H] cholesterol from subcutaneously injected macrophage foam cells
To test whether the histamine-mediated increase of vascular permeability in the skin was capable of modifying the RCT rate from macrophages located in the subcutaneous layer of the skin, [ 3 H]cholesterol-labeled macrophages were subcutaneously injected, and immediately thereafter, histamine was injected into the same tissue site. In preliminary experiments, we observed that a single dose of histamine induced only a nonsignifi cant increase in the overall m-RCT rate to feces relative to the vehicle group (0.72 ± 0.23 vs. 0.48 ± 0.06% of the injected dose, P = 0.32, n = 8 mice/group) within the 48 h period. This failure could be related to the rapid and only transient local effect of histamine in the skin. Thus, in the following experiments, mice received a second subcutaneous dose of histamine at 24 h after the fi rst dose, all various injection sites being closely was administered, and only slightly reduced when the H2R antagonist was administered. When administered together, the two antagonists exhibited a stronger effect, which supported the notion that the H2 antagonist had a small additive effect when combined with the H1 antagonist. Yet, no statistical difference was found between the values obtained when blocking only H1R or blocking both H1R and HR2. Overall these results indicated that the histamine-dependent effect on m-RCT was mainly mediated by H1R.

Histamine increases infl ux of HDL to skin
To get insight into the mechanism by which histamine induced m-RCT from the treated skin, we evaluated the capability of the locally injected histamine to induce the infl ux of [ 3 H]HDL from plasma into the skin within 48 h (i.e., the time period in which m-RCT rate was measured). Because the tissue content of HDL is determined at any moment by the magnitudes of the simultaneously occurring infl ow into and the outfl ow from the extracellular fl uid in the treated tissue site, it is not possible to determine the exact magnitude of the vascular leakage in the treated skin site at various time points. Thus, we measured the total content of radioactivity in the studied skin site, which is an approximation of the concentration of the labeled HDL in the extracellular fl uid of the tissue site and also the determinant of the local cellular cholesterol effl ux-inducing capacity of HDL in vivo. For this purpose, we isolated mouse HDL, radiolabeled the protein component of the isolated HDL, and intravenously injected the [ 3 H]HDL into mice immediately after the animals had received subcutaneous injection of histamine or vehicle. After 1 h, 24 h, and 48 h, the mice were euthanized, the treated sites of skin were collected, and their contents of 3 H-radioactivity were measured. As shown in Fig. 3A , the local effect on skin permeability induced by subcutaneous injection of histamine did not modify the clearance kinetics of the circulating pool of [ 3 H]HDL. Importantly, a signifi cant 3-fold increase in the fl ux of [ 3 H]HDL into the histamine injection skin site was observed within 1 h ( Fig.  3B ). Because, to promote m-RCT, the HDL particles that entered the skin site must leave the site, we measured the levels of [   ]bile acids). Data represent the mean ± SEM from two independent experiments and were normalized as percent of control (vehicle). Mean value obtained in the vehicle group was 0.13 ± 0.01, and it was set as 100%. Statistical signifi cance is denoted as * P < 0.05, ** P < 0.01; n = 10 mice/group for vehicle and control, and n = 4-5 mice/group for the histamine antagonist groups. subcutaneous injections of either serotonin (500 µg/kg), bradykinin (500 µg/kg), or vehicle, and the infl ux of intravenously injected [ 3 H]HDL to the treated skin sites was measured at 1 h and 24 h. The results indicated that, comparable to histamine, both serotonin and bradykinin stimulated within 1 h the infl ux of [ 3 H]HDL to the injected skin site, whereas after 24 h of injection the [ 3 H]HDL content in the skin decreased to the level of the control group ( Fig. 4A ). Neither serotonin nor bradykinin affected the kinetics of the circulating pool of [ 3 H]HDL ( Fig. 4A ,  inset). Next, we evaluated the effect of subcutaneous administration of serotonin and bradykinin on m-RCT to feces. For this purpose, either compound was subcutaneous injected immediately after the injection of the [ 3 H] cholesterol-labeled macrophages, and the transfer of the 3 H-tracer to feces was determined. Like histamine, both compounds were administered twice, 24 h apart, and feces were collected over 48 h. As shown in Fig. 4B, C , administration of either compound stimulated signifi cantly the excretion of [ 3 H]cholesterol and the appearance of the 3 H-tracer in the bile acid fraction in feces, so the total fecal 3 H-radioactivity signifi cantly increased. Altogether, the data revealed that, besides histamine, other physiological vasoactive compounds also have the potential to increase indicated that in the m-RCT experiment the initial dose of histamine must have stimulated the passage of HDL through the skin over a period of 24 h and that the second dose must have exerted an additive effect further increasing the availability of HDL particles in the skin site. Thus, the two doses must have triggered two similar 24 h periods during which an enhanced fl ow of endogenous (unlabeled) HDL through the treated skin site provided twice the conditions for a more effi cient RCT initiation from tissue macrophages. These results are compatible with a moderate nonsignifi cant increase of m-RCT observed with one histamine dose and the more robust stimulatory effect observed with the two doses.

Serotonin and bradykinin stimulate HDL uptake by skin and promote RCT from skin macrophages to feces
To further test the hypothesis that the histamineinduced increase in the infl ux of HDL into the skin was due to increased vascular permeability, we repeated the above experiment using two other natural vasoactive compounds. We selected serotonin and bradykinin, which, like histamine, are endogenous infl ammatory mediators that have been demonstrated to also rapidly increase vascular permeability in rodents ( 15,21 ). Thus, mice received Fig. 4. Serotonin and bradykinin increase the infl ux of HDL to skin and stimulate m-RCT from skin to feces. The infl ux of HDL to skin and decay curves for the tracer in plasma were evaluated up to 24 h in mice that received subcutaneous injections of serotonin (500 µg/kg), bradykinin (500 µg/kg), or vehicle alone immediately followed by intravenous injection of a labeled HDL preparation, as described in Fig. 3 . Samples of skin were dissected from the site of subcutaneous injection at 1 h and 24 h, and blood samples analyzed for the decay curves (inset) were collected at similar time points as described in Fig. 3 . Radioactivity was measured in the tissue extracts, and the results expressed as nanograms of HDL per milligram of tissue wet weight (A). To evaluate m-RCT, mice received subcutaneous injection of [ 3 H] cholesterol-labeled J774 macrophage foam cells immediately followed by subcutaneous injection of serotonin (500 µg/kg), bradykinin (500 µg/kg), or vehicle alone to the dorsal skin site. Treatments with serotonin and bradykinin were repeated 24 h after the initial injection to the respective group of mice. The rate of m-RCT from skin to feces ([ 3 H]cholesterol + [ 3 H]bile acids) was determined, as described in Fig. 1 , in serotonin-treated (B) and bradykinin-treated (C) groups. Data represent the mean ± SEM from two to three independent experiments. Statistical signifi cance is denoted as * P < 0.05, ** P < 0.01, *** P < 0.001; n = 10-12 mice/group. from wild-type mice immediately after the initial subcutaneous injection of histamine. The single dose of HDL had only a short-lasting effect on the basal low HDL levels, and, accordingly, after 48 h the plasma lipids were not changed ( Table 3 ). Measurement of the transfer of macrophage-derived cholesterol to feces indicated that administration of merely histamine to apoA-I-KO mice the skin-originated m-RCT resulting from an increase in the permeability to HDL of the vascular bed surrounding the macrophages.

Vasoactive compounds do not affect cholesterol effl ux from cultured macrophages
We next evaluated the potential effects of the vasoactive compounds on the gene expression of cholesterol effl ux transporters in macrophages and the ability of cultured J774 macrophage foam cells to effl ux cholesterol to mouse serum. To generate preconditioned cells, we incubated the J774 macrophages with decreasing concentrations of the compounds for up to 3 h (see Materials and Methods) but did not fi nd any changes in the gene expression levels of the ABCA1, ABCG1, or SR-BI transporter ( Fig. 5A ). Of the two histamine receptor antagonists, the H2R antagonist ranitidine induced overexpression of macrophage ABCA1 relative to the control cells (1.51 ± 0.1 vs. 1.0 ± 0.1, P = 0.03), which may partially explain its less marked effect in blocking the stimulatory effect of histamine on m-RCT (see Fig. 2 ). Importantly, the effi ciency of the preconditioned macrophages to effl ux cholesterol was not affected when the cells were further incubated for 3 h in mouse serum-containing medium ( Fig. 5B ). Analysis of serum lipids in mice treated with histamine and histamine receptor antagonists indicated that administration of the various compounds did not affect circulating lipid levels, particularly HDL-cholesterol, indicating that the treatments had not modifi ed the cholesterol effl ux capacity of the serum ( Table 1 ). Neither did serotonin infl uence serum lipids; however, bradykinin treatment decreased triglyceride levels without having any effect on HDLcholesterol levels ( Table 2 ). Altogether, these results provided strong supportive evidence for the claim that a local increase in the vascular permeability inducing enhanced entry of HDL particles into skin, rather than modifi cation of the intrinsic effi ciency of the macrophage cholesterol effl ux pathways, was the driving mechanism promoting m-RCT in vivo when skin was treated with the vasoactive compounds.

Histamine administration to apoA-I-defi cient mice increases m-RCT from skin to feces in the presence of exogenous HDL
Because the stimulatory effect of increasing vascular permeability on m-RCT depended on the increased passage of HDL particles to the skin site where the macrophage foam cells were present, it seemed very likely that under conditions of low serum HDL levels the effect of the vasoactive compounds on m-RCT would be reduced. We predicted that in this experimental setting a transient enrichment of serum with exogenous HDL would result in enhanced passage of HDL through the leaky endothelium to the skin site, thereby stimulating the m-RCT rate. We tested these predictions in apoA-I-KO mice, which present 2-fold lower levels of HDL-cholesterol relative to the wild-type mice ( Table 3 ). The basic m-RCT protocol used in this work was applied with the exception that the mice received an intravenous injection of HDL isolated  Figs. 1 and 4 , and was progressively reduced within intervals of 10 min, to achieve a fi nal 100-fold reduction at 30 min of incubation, which was prolonged until 3 h. Macrophages incubated in the absence of vasoactive compounds acted as control. The relative mRNA expression of ABCA1, ABCG1, and SR-BI genes was determined by RT-PCR in the preconditioned macrophages derived from triplicate wells. The signal of control macrophages was set at a normalized value of 1 arbitrary unit, and GAPDH was used as internal control (A). Cholesterol effl ux from [ 3 H]cholesterol-labeled J774 macrophage foam cells to mouse serum (2.5% in medium) was assessed after the preconditioning incubation step of the cells in the presence of the vasoactive compounds at their fi nal concentrations, as described above. Nonpreconditioned macrophages incubated in the absence of vasoactive compounds and in the presence of serum alone acted as control. Fractional cholesterol effl ux (%) was expressed as dpm medium / (dpm cells + dpm medium ) × 100 (B). All data represent the mean ± SEM from four replicate wells. Mice were treated with histamine (250 µg/kg), the H1R antagonist pyrilamine (20 mg/kg), the H2R antagonist ranitidine (20 mg/kg), or vehicle alone. The data presented correspond to serum obtained at the fi nal time point (48 h) of the experiment shown in Fig. 2 . Data represent the mean ± SEM. resulted in a slight, nonsignifi cant increase of m-RCT ( Fig. 6 ), which may have been caused by the low residual levels of apoA-I-defi cient HDL in the serum of these mice. Neither did intravenous administration of HDL alone (i.e., without histamine) produce any effect on m-RCT. However, when exogenous HDL was administered together with histamine, the m-RCT rate increased signifi cantly. Altogether, these data showed that, on a low-HDL background, increase of circulating HDL levels and triggering of local vascular leakage were both required to enhance passage of plasma HDL into the skin and accelerate the initiation of m-RCT from the subcutaneously located macrophages.

Histamine administration does not modify m-RCT from skin to feces in the presence of high levels of circulating non-HDL lipoproteins
We speculated that if apoB lipoproteins, instead of HDL, were the predominant class of lipoproteins in the mouse plasma, the benefi cial effect on m-RCT of increasing the vascular permeability could be challenged. Thus, following our basic protocol, we evaluated the effect of histamine on the rate of m-RCT from skin in the atherosclerosisprone LDLr-KO mice, which exhibit normal HDL levels but 10-fold higher levels of non-HDL (apoB lipoproteins) relative to the wild-type ( Table 4 ). We found that, under such hypercholesterolemic background, histamine did not increase the transfer of macrophage-derived radioactivity to serum, liver, or feces ( Fig. 7 ). However, the magnitude of the recovered macrophage-derived radioactivity refl ected high transfer rates to serum, and particularly to the nonHDL fraction ( Fig. 7A ), which accounted for ‫ف‬ 80% of the total radioactivity in the serum of both vehicle-and histamine-treated mice. The high levels of the 3 Htracer recovered in feces suggested that LDLr-KO mice might exhibit high m-RCT rates already under basal conditions, that is, ‫ف‬ 4-fold higher total radioactivity in feces (0.45% of injected dose) relative to the wild-type mice (0.11% in Fig. 1C , and 0.10% in Fig. 4B, C ). The fi nding that histamine did not have any further effect on such apparently accelerated basal m-RCT rate in LDLr-KO mice rather suggested that histamine and hypercholesterolemia may have shared a common stimulatory mechanism, such as the vascular leakage, which has been, indeed, found to occur in hypercholesterolemic mice ( 22,23 ).

Endogenous histamine released from activated mast cells increases m-RCT from skin
Mast cells are infl ammatory cells, which, when activated to degranulate, release histamine and several other vasoactive compounds in vivo ( 24 ). Because mast cells are the major source of histamine in the subcutaneous layer of skin ( 25 ), we evaluated whether their local activation could infl uence the rate of m-RCT. For this purpose, m-RCT was measured following our basic protocol in wildtype mice that had received subcutaneous injection of the noncytotoxic mast cell-specifi c degranulating compound 48/80 1 h before the [ 3 H]cholesterol-labeled J774 macrophages were injected into the same area of skin. We found that 48/80 effectively induced mast cell activation in the skin, which, when compared with vehicle, resulted in exocytosis of the cytoplasmic granules at the site of injection ( Fig. 8A , right, upper, and lower panels, respectively) and could be quantitatively assessed by determining the numbers of degranulated mast cells at the injection site ( Fig.  8A , left, upper panel). The observed reduction of histamine content in the tissue site in which the mast cells had been activated agrees well with the known diffusion and rapid escape of the released histamine from the mast cell activation site ( Fig. 8A , bottom left). Importantly, we found that a single injection of 48/80 that induced degranulation of ‫ف‬ 70% mast cells in the treated skin area was sufficient to increase the m-RCT rate ( Fig. 8B ). Treatment of mice with 48/80 did not modify serum lipids ( Table 5 ), nor did addition of 48/80 to cultured [ 3 H]cholesterollabeled J774 macrophage foam cells affect the ability of the cells to effl ux cholesterol to mouse serum ( Fig. 8B ,  inset). These data demonstrate that activation of skin mast cells with ensuing release of histamine and other vasoactive compounds is capable of stimulating the basal rate of RCT from nearby macrophage foam cells in vivo.

DISCUSSION
A given concentration of lipoprotein particles in the interstitial fl uid results from the balance determined by the Mice received subcutaneous injections of serotonin (500 µg/kg), bradykinin (500 µg/kg), or vehicle alone. The data presented correspond to serum obtained at the fi nal time point (48 h) of the experiments shown in Fig. 4B, C . Data represent the mean ± SEM. Statistical signifi cance relative to the vehicle group is denoted as a P < 0.001. mouse plasma. Importantly, this effect was suffi cient to increase the rate of RCT originating in vivo from macrophages located within the treated skin area. Because none of the vasoactive compounds studied here affected the serum HDL levels or the rate of cholesterol effl ux from the cultured J774 macrophages, the stimulation of m-RCT in vivo must have been due to increased entry of circulating HDL into the treated skin, rather than due to an increased cholesterol effl ux potential of the cholesterol acceptors or donors per se. These fi ndings support a novel independent role for the rate of the passage of plasma HDL to the interstitial fl uids in regulating m-RCT rate from peripheral tissues and, accordingly, the cholesterol content of macrophages in these body compartments.
The passage of lipoproteins across the normal intact endothelium is a regulated process that occurs mainly by transcellular routes ( 28 ), while the ultrafi ltration of circulating lipoproteins at sites of endothelial activation where vascular permeability is increased has been found to be a paracellular process that occurs at the site of gaps between the endothelial cells ( 29 ). It has been established that infl ammatory mediators such as histamine, bradykinin, serotonin, and other vasoactive compounds, upon binding to their receptors on endothelial cells, disrupt the organization of the interendothelial junctions, thus leading to opening of the endothelial barrier ( 30 ). The size of the gaps can be considerable; for example, it was reported that substance P may induce gaps in the interendothelial junctions of endothelial cells in vessels of the rat trachea that range from 100 to 400 nm and have a half-life of ‫ف‬ 1.9 min ( 31 ). Histamine, a vasoactive compound primarily released by activated mast cells, and in small quantities also by other cells, such as the endothelial cells themselves, is the most studied compound in in vitro models of vascular permeability. Importantly, the increase of vascular permeability induced by histamine has resulted in immediate extravasation of LDL from plasma to skin in rats and hamsters ( 15,32 ). The rapid extravascular accumulation of LDL most likely results from the sieving of lipoproteins ratio of particles entering via the capillaries and those removed via the lymphatic vasculature ( 26,27 ). Although it is likely that modifi cation of the vascular permeability, similar to variations in the lymphatic fl ow ( 13,14 ), may locally modify the concentration of HDL particles in the interstitial fl uid and so infl uence the initiation of RCT from peripheral macrophages, a direct role for an increase in vascular permeability as a regulator of m-RCT has not been experimentally proven before. Here we evaluated the RCT rate from cholesterol-labeled macrophages injected into the highly vascularized subcutaneous skin layer of mice in which infl ux of circulating HDL was enhanced by treatment with various vasoactive compounds. Because skin also contains various types of cells secreting vasoactive compounds, notably the mast cells, skin was considered to be a suitable tissue where HDL concentration could also be raised by naturally occurring endogenous compounds capable of increasing the vascular permeability. We demonstrated that sole administration of histamine, serotonin, or bradykinin, each capable of increasing vascular permeability facilitated the entry to the treated skin of circulating HDL particles, the main lipoprotein component in Mice received subcutaneous injection of histamine (250 µg/kg), intravenous injection of HDL (4-6 mg protein/kg), or both. Mice that received vehicle alone acted as control group. The data presented correspond to serum obtained at the fi nal time point (48 h) of the experiment shown in Fig. 6 . Data represent the mean ± SEM.   Fig. 7 . Data represent the mean ± SEM. could also mean that the second dose of histamine had an additive effect, which was qualitatively different from that achieved with the initial dose (e.g., on the lymph fl ow). Indeed, complex and even opposite effects of histamine on the lymphatic function have been reported. Thus, histamine appeared to stimulate the lymphatic vessel contractile activity ( 36 ), to act as an endothelium-derived relaxing factor in mesenteric lymphatic vessels ( 37 ), or to disrupt the integrity of the barrier function of lymphatic endothelial cells, thereby potentiating leakage of solutes from lymph back to the interstitial space ( 38 ). Whether the second dose of histamine had also activated the lymphatic SR-BI transporter recently found to regulate the entry of HDL into the lymphatic vessels ( 14 ) remains unknown.
In the present studies, we use wild-type mice fed a standard chow diet as our experimental model. However, in contrast to mice, which are HDL animals, in humans most of the cholesterol is transported in LDL, and, moreover, severe hypercholesterolemia due to increased LDL levels is associated with accumulation of cholesterol in peripheral tissues, particularly in the intimal layer of arteries generating atherosclerotic lesions and also in the skin generating xanthomas ( 39 ). In this context, it is of particular relevance that the genetic absence of apoA-I in humans also results in the formation of skin xanthomas, so revealing a critical role of this apolipoprotein in the regulation of skin cholesterol balance ( 40 ). Notably, the present results support the view that an increased passage of apoA-I-containing HDL to the interstitial fl uid of the skin is capable of locally accelerating m-RCT from cholesterol-loaded cells and so potentially able to decrease the accumulation of cholesterol found in skin xanthomas. In an attempt to mimic the human hypercholesterolemic phenotype, we also evaluated the effect of increasing vascular permeability in the LDLr-KO mouse, which exhibits high non-HDL/HDL ratio and is considered an animal model of familial hypercholesterolemia (FH) ( 41 ). Surprisingly, the basal rate of m-RCT already appeared to be increased in the untreated LDLr-KO mice relative to the untreated wild-type mice (compare the vehicle groups in Fig. 7C and Figs. 1C and 4B, C ) and treatment with histamine failed to have any further additive effect on the basal rate in the KO mice. Because hypercholesterolemia weakens the endothelial junctions and increases vascular permeability in mice ( 22,23 ), histamine may have failed when acting on an endothelium already leaky. To our best knowledge, no published data on the effi ciency of m-RCT in LDLr-KO mice are available. Thus, we can only speculate that high m-RCT rates in the LDLr-defi cient mice ( Fig. 7C ) might have been caused by an enhanced passage of plasma HDL through the skin followed by an alternative m-RCT route that includes transfer of cholesterol to LDL and uptake of the cholesterol-enriched LDL by the hepatic LRP receptor ( 42 ). In addition, effl ux of macrophage cholesterol to LDL could predominate when its concentration exceeds that of HDL ( 43 ). The induction of high rates of m-RCT in a mouse model of FH observed in the present study and also in cholesterol-fed wild-type mice ( 44 ) would seem to challenge the concept that the rate of m-RCT predicts atherosclerosis susceptibility, at least in the mouse. through water pores of ‫ف‬ 22 nm in radius, up to the size of LDL ( 32 ). Such transient increase in endothelial permeability induced by histamine may be prolonged by thrombin via a mechanism that involves modulation of the intracellular levels of calcium ions, as it was shown in human endothelial cell monolayers ( 33 ).
Because the permeability-enhancing effect of histamine is rapid and transitory, we administered two subsequent doses, which, by opening the endothelial gaps twice, facilitated the infl ux of suffi cient amounts of circulating HDL into the skin and increased m-RCT rate over a 48 h period. Under these experimental conditions, the stimulatory effect of histamine on m-RCT appeared to be mediated mainly by the H1R, with the H2R exerting a small additive effect. Consistent with our fi ndings, activation of H1R, rather than H2R, has been found to enhance vascular permeability for endogenous LDL in cholesterol-fed apoE Ϫ / Ϫ mice without altering the plasma lipoprotein levels ( 34 ). It has been also shown that histamine increases the passage of LDL through monolayers of human arterial endothelial cells, having its peak at 2 h but rapidly decreasing to the control level within 4-6 h, and, moreover, that this effect can be blocked by the H1R antagonist pyrilamine, rather than by the H2R antagonist cimetidine ( 35 ). The requirement of two doses of histamine for the m-RCT response  ( 25,45 ). Of note, mast cells are the only resident cells in the rodent skin tissue, which upon activation acutely secrete preformed fully active neutral serine proteases ( 46 ). The We fi nally tested our permeability hypothesis in a model of local infl ammation by activating mast cells in the skin of wild-type mice to degranulate and to secrete a wide cocktail of vasoactive mediators, notably histamine, . Statistical signifi cance are denoted as * P < 0.05, *** P < 0.001; n = 5-8 mice/group. Cholesterol effl ux from cultured [ 3 H]cholesterol-labeled J774 macrophage foam cells (see insert in B) to mouse serum (2.5% in medium) was assessed in the presence of 48/80 (50 µg/ml in medium). Fractional cholesterol effl ux (%) was expressed as dpm medium /(dpm cells + dpm medium ) × 100.  Fig. 8B . Data represent the mean ± SEM. mast cell-derived chymases, particularly the mouse mast cell protease-4 (mMCP-4), are capable of degrading mouse HDL in the extravascular fl uids of tissues in which mast cells have been activated, as has been observed during systemic mast cell activation in a mouse model of systemic anaphylaxis ( 47 ). When considering the above background information, it is likely that the actual impact of mast cell activation on HDL functionality in vivo will depend on the balance between the vasoactive and proteolytic effects, which may vary when tested in different experimental settings. We can infer that the present fi ndings disclosed a scenario in which an acute endogenous release of a mixture of vasoactive compounds increasing the infl ux of HDL to the skin site was capable of overriding the inhibitory effect of the coreleased proteases on an expanded interstitial pool of HDL. It is likely that the transient and highly local proteolytic effect of the granulebound proteases is challenged by their rapid phagocytosis by neighboring cells, such as fi broblasts, endothelial cells, and particularly macrophages at the skin site artifi cially enriched with these cells ( 48,49 ).
The present study is the fi rst to link an increased endothelial permeability with increased rates of RCT from peripheral tissue macrophages. It was recently reported that the function of lymphatic vessels clearing HDL from skin remains as the dominant conduits for m-RCT also in tissues surrounding an implanted tumor in mice characterized by increased vascular permeability ( 13 ). A more detailed characterization of the relative roles played by the vascular permeability and lymphatic drainage regulating coordinately the inward and outward fl ow of HDL on the RCT rate deserves further investigation. Finally, we envision that the proatherogenic effect of increased vascular permeability that is associated with increased infl ux of LDL ( 15 ) is counteracted by simultaneously increased infl ux of HDL, particularly of the functionally highly competent and very small-sized HDL particles, such as pre ␤ 1 -HDL. Because an increase in the vascular permeability will induce an increased passage of both LDL and HDL, and we lack pharmacological means of selectively preventing the passage of LDL, maintaining the proper balance between the proatherogenic LDL and the antiatherogenic HDL in the circulation remains the main clinical strategy when aiming to reduce the formation of macrophage foam cells or trying to accelerate their regression.