Extracellular Release of the Atheroprotective Heat Shock Protein 27 Is Mediated by Estrogen and Competitively Inhibits acLDL Binding to Scavenger Receptor-A

HSP27 Overexpression Protects Against Atherogenesis in Female Mice

Using a mouse model of inflammatory atherosclerosis, we sought to determine whether overexpression of human HSP27 (HSP27^o/e) is protective against atherogenesis in apoE^−/− mice that are prone to atherosclerosis when fed a diet supplemented with cholesterol.¹⁹ apoE^−/−HSP27^o/e and apoE^−/− mice were placed on a high-cholesterol diet for 4 weeks and euthanized at 10 weeks of age. Mean body weight and length, as well as total serum cholesterol levels, were similar for these 2 groups of mice (see supplemental Figure I). The percentage aortic lesion area, measured by quantitative histomorphology of oil red O–stained en face specimens was 35% reduced in apoE^−/−HSP27^o/e versus apoE^−/− female mice (n=9/group; P<0.001; Figure 1). The male apoE^−/− HSP27^o/e and apoE^−/− mice (n=6/group) had similar percentage aortic lesion areas (12.9±1.2% versus 13.5±1.0% lesion area compared to total arch area, respectively; P=0.69). Serum HSP27 levels measured using an ELISA were low in all apoE^−/−HSP27^o/e mice fed a normal chow diet (Figure 2A). However, after 4 weeks of a high-fat diet, female apoE^−/−HSP27^o/e mice had more than 10-fold higher circulating levels of HSP27 compared to male mice (P≤0.05). This difference in serum HSP27 levels was not apparent until after 2 weeks of ingestion of a cholesterol-enriched diet (eg, serum HSP27 levels after 2 weeks in females: 536±308 pg/mL versus males: 123±81 pg/mL; P=0.456). When circulating HSP27 levels were compared to total en face lesion area, there was a remarkable inverse correlation between serum HSP27 levels and atherosclerotic lesion area (r²=0.78; P<0.001, Figure 2B) in both males and females. When only females were used for correlation analysis, the relationship between circulating HSP27 and atherosclerotic lesion area was even stronger (r²=0.9; P<0.001).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. HSP27 overexpression reduces atherosclerotic lesion size in female mice. Total aortic en face atherosclerotic lesion area was analyzed in mice overexpressing human HSP27 (apoE^−/−HSP27^o/e) (A, C) and compared to their apoE^−/− littermates (B, D); pictured here are female mice. Quantification of lesion area/total aortic arch area (E) demonstrated a 35% reduction in lesion burden in HSP27^o/e apoE^−/− compared to apoE^−/− female mice (*P<0.001). No difference was observed in the male mice.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. A, HSP27 serum levels increase in response to an atherogenic diet. Serum HSP27 levels from mice fed a normal chow diet (left columns) and from the same mice 4 weeks after ingesting a high-fat atherogenic diet (right columns) (*P≤0.001). B, Regression analysis comparing HSP27 serum levels with aortic en face lesion area in male and female mice HSP27 overexpressing mice.

HSP27 is Secreted In Vitro in Response to Estrogen and acLDL

As serum HSP27 levels inversely correlate with aortic lesion area in mice fed a cholesterol-enriched diet, and is >10-fold higher in females than in males, we determined in vitro whether HSP27 is released on stimulation with estradiol (E2) or atherogenic acetylated low-density lipoprotein (acLDL). Human macrophages (U937) were plated in replicates at a density of 1×10⁶ per well and treated with estradiol or acLDL. Conditioned media was collected and before analysis for secreted HSP27, overall cell viability was measured using an LDH-release assay and revealed no difference in cell viability or membrane integrity between any of the treatments (data not shown). Treatment of macrophages with estrogen (E2) for 24 hours caused a dose-dependent increase in HSP27 release into the media compared to controls (Figure 3A).^20,21 Estrogen-induced HSP27 release also increased over time, with maximum secretion after 24 hours (Figure 3B). Macrophages were subjected to increasing concentrations of acLDL (1 to 100 μg/mL) for 24 hours. HSP27 protein was detected by Western blot in conditioned media from cells treated with 100 μg/mL acLDL (Figure 3C). The addition of acLDL to the media containing estrogen caused a further increase in HSP27 secretion when compared to estrogen or acLDL treatment alone, indicating that these 2 mechanisms of secretion may act synergistically (Figure 3D). On treatment with estrogen, examination of intracellular protein levels revealed that HSP27 protein levels increased slightly in response to estrogen treatment, suggestive of an intracellular pool of HSP27 that is secreted without necessitating de novo protein synthesis (Figure 3E). On treatment with acLDL, there was a concomitant dose-dependent increase in intracellular HSP27 expression, then an apparent decrease corresponding to increased HSP27 protein release into the media (Figure 3F).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. HSP27 is secreted from macrophages in response to estrogen and acLDL. A, Human U937 macrophages in culture were treated with 10% charcoal-stripped FBS with or without increasing concentrations of 17β-estradiol (E2) for 24 hours, and conditioned media was subjected to Western blotting using an antibody against HSP27. HSP27 secretion into the media increased in a dose-dependent manner with E2. *P≤0.05 compared to control. B, U937 cells were treated with or without 100 nmol/L E2 for 1, 3, and 24 hours, and conditioned media showed increasing HSP27 released into the extracellular space over time. *P≤0.05 compared to control; #P≤0.05 compared to all treatment groups. C, U937 cells were treated in culture with increasing concentrations of acLDL (0 to 100 μg/mL). HSP27 was detected in the conditioned media on treatment with 100 μg/mL acLDL (top panel). *P≤0.05 compared to control. D, U937 cells were treated with 100 nmol/L E2 or 100 μg/mL acLDL, or both together for 24 hours and conditioned media was analyzed by Western blot. HSP27 in the conditioned media increased with treatment of E2 and further increased with treatment of acLDL. An equal number of cells and volume of conditioned media was analyzed for each treatment condition for all conditioned media experiments. *P≤0.05 compared to control; #P≤0.05 compared to all treatment groups. E, Intracellular protein from U937 cells treated with increasing concentrations of E2 and probed using an antibody to HSP27. Equal protein loading was confirmed using Ponceau S staining and α-actin blotting (see supplemental Figure III). F, Intracellular protein from U937 cells treated with increasing concentrations of acLDL, and probed using an antibody to HSP27. Equal protein loading was confirmed using Ponceau S staining and α-actin blotting (see supplemental Figure III).

To examine the pathway of HSP27 secretion, we used 2 independent experiments: the first used human U937 macrophages treated with E2 and immunolabeled for HSP27 and a marker of the lysosomal membrane (LAMP1); second, mouse J774 macrophages were transfected with a fluorescently tagged HSP27 (HSP27-ECFP), and treated with acLDL in the presence of Lysotracker, which labels acidic organelles (ie, lysosomes) in live cells. Using both methods, we localized intracellular HSP27 in macrophage lysosomes after treatment with E2 or acLDL. Specifically, human macrophages treated with 100 nmol/L E2 overnight displayed colocalization of HSP27 (red) and LAMP1 (green; Figure 4A). Similarly, mouse macrophages transfected with fluorescently-tagged HSP27 (green) treated with 100 μg/mL acLDL and incubated in the presence of Lysotracker (red) displayed HSP27 colocalization within lysosomes after treatment for 1 hour and 24 hours (merged image, Figure 4B). Without estrogen treatment, HSP27 showed minimal colocalization with the lysosome (Figure 4A and 4B, bottom row). Cells transfected with empty ECFP alone with or without acLDL treatment did not show colocalization with the lysosome, indicating that HSP27-ECFP was not simply degraded and targeted to the lysosome (data not shown). These results indicate that HSP27 is found within secretory lysosomal-like vesicles in macrophages under conditions which stimulate its secretion (eg, on treatment with E2 or acLDL).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4. A, U937 cells were treated with 100 nmol/L E2 overnight and immunolabeled using antibodies to LAMP1 (green, a marker for the lysosomal membrane) and HSP27 (red). Merged image (yellow) shows colocalization of HSP27 in the lysosome (magnification, 100×). Images were obtained with sequential scanning to avoid bleed through, and arrows indicate locations which are positive for LAMP1 in the green channel and where no staining is observed in the red channel. B, J774 cells transfected with HSP27-ECFP (green) were treated with 100 μg/mL acLDL 24 hours in the presence of Lysotracker red. Colocalization is seen as a yellow color (magnification, 100×). Images were obtained with sequential scanning to avoid bleed through, and arrows indicate locations which are positive for HSP27-ECFP in the green channel and where no staining is observed in the red channel. Experiments were performed in triplicate for all treatment conditions. See supplemental Figure IV for additional controls.

Extracellular HSP27 Binds the Scavenger Receptor-A and Prevents acLDL Uptake

Given that HSP27 is secreted not only in vitro by atherogenic lipids, but in vivo in response to high-fat diet, the next step was to determine whether extracellular HSP27 is involved in cholesterol trafficking and thus potentially the progression of atherosclerosis. Other groups have shown that members of the heat shock protein family (eg, HSP70) bind a variety of cell surface receptors, including toll-like receptors and scavenger receptors.^22–25 We investigated whether extracellular HSP27 binds the SR-A—an important receptor for the uptake of atherogenic lipids and the progression of atherosclerosis.^26,27 Immunolabeling studies reveal that recombinant extracellular HSP27 is capable of binding the surface of macrophages, and colocalizes with SR-A (Figure 5A). In the presence of fucoidan, a specific competitor for SR-A, HSP27 binding to SR-A was reduced, indicating that this interaction is specific.²⁸ Furthermore, in macrophages from SR-A^−/− mice, HSP27 binding was absent (Figure 5B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5. Extracellular HSP27 colocalizes with the SR-A on the surface of macrophages. A, Recombinant HSP27 (5 μg/mL) was administered to J774 macrophages for 2 hours at 4°C. Immunolabeling of HSP27 (green) and SR-A (red) was visualized using confocal microscopy (100×). Cells were also treated with an SR-A specific competitive ligand (fucoidan, 10 μg/mL) before administration of HSP27 (bottom panel). Colocalization is seen as a yellow color. B, Macrophages from SR-A^+/+ (see supplemental Figure VB) and SR-A^−/− mice were harvested and incubated with recombinant HSP27 (5 μg/mL) for 2 hours at 4°C. Immunolabeling was performed as described above. C, HSP27-ECFP (or empty ECFP control) secreted in response to acLDL was applied to naïve macrophages at 4°C for 2 hours (see supplemental Figure VC). Cells were crosslinked with increasing concentrations of DTSSP (0.1 to 2 mmol/L) and immunoprecipitated using an antibody to ECFP. Lane containing nonimmunoprecipitated cell lysate represents 25% to 30% of input (see supplemental Figure VD). Experiments were performed in triplicate for all treatment conditions.

Next, we explored whether endogenous HSP27 secreted in response to acLDL is capable of binding SR-A. Conditioned media from macrophages transfected with HSP27-ECFP (or empty ECFP) and treated with 100 μg/mL acLDL was applied to naïve, untreated macrophages at 4°C for 2 hours to allow HSP27-ECFP to bind the cell surface. Cells were then treated with DTSSP, a reversible, membrane-impermeable cross-linking agent, to cross-link HSP27 to the cell surface. Immunoprecipitation was carried out using antibodies to the fluorescent tag (anti-ECFP) and cross-linked proteins were reduced to reverse cross-linking, then separated on an SDS-PAGE gel and subjected to immunoblotting. Using antibodies to SR-A, we demonstrated that HSP27-ECFP is secreted from cells treated with acLDL and binds the SR-A under various concentrations of DTSSP (Figure 5C). Macrophage whole cell lysates were also probed and displayed a ≈80 kDa band corresponding to SR-A in these cells.

As foam cell formation is a hallmark of atherogenesis, and prevention of lipid uptake by macrophages may serve to reduce lesion development and vessel wall inflammation, we hypothesized that HSP27 binds the SR-A to prevent uptake of atherogenic lipids such as acLDL to attenuate foam cell formation. Mouse macrophages were cultured in vitro in the presence of fluorescently labeled acLDL for 6 hours, and acLDL uptake was measured using a fluorometer before being normalized for total cell number (n=10 wells per treatment; experiments performed in triplicate). When extracellular HSP27 was added to the culture media, there was a 41% reduction in specific acLDL uptake by macrophages (P≤0.05; Figure 6; normalized to nonspecific uptake in the presence of the SR-A competitor fucoidan). To investigate whether a reduction in overall lipid uptake occurs in vivo in the presence of HSP27, peritoneal macrophages from HSP27^o/eapoE^−/− and apoE^−/− mice were harvested and stained with Oil Red O, which stains intracellular lipids red. Semiquantitative analysis was performed and the total number of lipid-laden cells were counted and expressed as a percentage of total cell number by 2 independent blind reviewers. Only 3% of macrophages from HSP27^o/eapoE^−/− were considered lipid-laden, whereas 21% of cells from apoE^−/− mice were considered lipid-laden (Figure 6B and supplemental Table I). Moreover, using Western blotting we noted that the reduction in macrophage acLDL uptake by HSP27 attenuated acLDL-induced release of IL-1β, a potent proinflammatory cytokine (Figure 7A). Extracellular HSP27 also increased the released of the antiinflammatory cytokine IL-10 (Figure 7B). Previous reports involving human monocytes suggest that HSP27 can induce the release of IL-10 but not TNF-α, and hence agree with our observations that extracellular HSP27 is primarily an antiinflammatory signaling protein.²⁹ These results suggest that extracellular HSP27 is capable of reducing foam cell formation and the accompanying inflammation.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6. A, Extracellular HSP27 inhibits uptake of acLDL by macrophages via the SR-A. Fluorescently labeled acLDL (5 μg/mL) was added to macrophages in the presence (right column) or absence (left column) of recombinant HSP27 (5 μg/mL). Cells were also treated with an SR-A specific competitor (fucoidan, 10 μg/mL) before administration of acLDL or HSP27. Cells were harvested, and fluorescent acLDL uptake was measured using a fluorometer and normalized to cell number. Nonspecific uptake in the presence of fucoidan was subtracted to define specific acLDL uptake; at least 10 samples were analyzed per group and comparisons were made using a 1-way ANOVA; *P=0.005 compared to acLDL alone. B, Peritoneal macrophages from apoE^−/− and HSP27^o/eapoE^−/− mice were plated, fixed, and stained for Oil Red O. Cells were counted and those that were “lipid-laden” (deep red color) were expressed as a % of total cell number. HSP27^o/eapoE^−/− macrophages contained only 3% lipid-laden cells compared to apoE^−/− macrophages which contained 21% lipid-laden cells (see supplemental Table I).

• Download figure
• Open in new tab
• Download powerpoint

Figure 7. Extracellular HSP27 results in release of cytokines involved in the inflammatory response. Fluorescently labeled acLDL (5 μg/mL) was added to J774 macrophages in the presence or absence of recombinant HSP27 (5 μg/mL). Conditioned media was collected after 30 minutes and 4 hours, and 40 μL was subjected to Western blotting for IL-1β (A) and IL-10 (B). Addition of HSP27 caused a decrease in the acLDL-induced secretion of IL-1β into the extracellular space. An equal number of cells and volume of conditioned media was analyzed for each treatment condition for all conditioned media experiments. Experiments were performed in triplicate for all treatment conditions.

HSP27 Overexpression Reduces Cell Adhesion and Migration

To further investigate how HSP27 might be protective against the development of atherosclerosis, peritoneal macrophages were harvested from apoE^−/−HSP27^o/e and apoE^−/− mice after a high-fat diet. Cells were plated in culture on a collagen matrix and allowed to adhere for 2 hours. There was a 53% reduction in cell adhesion in apoE^−/−HSP27^o/e macrophages compared to apoE^−/− (P≤0.001; Figure 8A and 8B). Macrophages were also maintained overnight in a transwell migration chamber and allowed to migrate toward a 10% FBS gradient. There was a 42% reduction in cell migration in apoE^−/−HSP27^o/e macrophages compared to apoE^−/− (P≤0.01; Figure 8C). No cell migration was detectable in the absence of FBS (data not shown). Taken together, macrophages that overexpress HSP27 display reduced adherence and migration in vitro, suggesting that in vivo these cells are less likely to incorporate into vascular lesions and exacerbate disease.

• Download figure
• Open in new tab
• Download powerpoint

Figure 8. HSP27 overexpression results in decreased macrophage adhesion and migration. A, Peritoneal macrophages harvested from apoE^−/−HSP27^o/e and apoE^−/− mice were plated on type I collagen. After 2 hours incubation, the cells were washed and fixed. B, Cell nuclei were stained with Hoechst 33258 and the number of cells per high power field (HPF) was manually counted. C, Peritoneal macrophages as in A were subject to CytoSelect migration assay for 24 hours and quantified as total number of cells migrated toward 10% FBS. Experiments were performed in triplicate for all treatment conditions.