Abstract
We introduce here a new fluorescence microscopy technique for en face analysis of the atherosclerotic fatty streaks (FS). This technique is semiquantitative and has the sensitivity and resolution to map lipids to individual cells in FS less than 100 μm in diameter. New Zealand White rabbits were fed an atherogenic diet for up to 26 weeks. Aortas were fixed in formalin and stained en bloc with the fluorescent dyes Nile red and filipin. Fluorescent staining was validated by correlating microfluorimetric and biochemical measurements of the lipid content in FS. To determine the cell types associated with the different staining patterns, FS were also evaluated by transmission electron microscopy (TEM) and immunohistochemistry (IH). Correlation of microfluorimetry, TEM, IH, and biochemical data indicated that regions rich in non-esterified cholesterol stained with filipin and fluoresced blue owing to accumulations of lipid vessicles and/or cholesterol crystals. Regions rich in neutral and polar lipids stained with Nile red and fluoresced yellow or orange, respectively, owing to accumulations of lipids in both macrophages and smooth muscle cells (SMC). Digital overlays of the filipin and Nile red images revealed that larger lesions (>0.5 mm diameter) had a “nested” distribution of lipids, with a blue (filipin) fringe surrounding an orange (Nile red) fringe surrounding a yellow (Nile red) center.
A
Fluorescence and light microscopy have been instrumental in locating and identifying lipid components of atherosclerotic lesions in animal models and human biopsy tissues (Brown et al. 1992; Seifert 1989; Lupu et al. 1987; Fowler and Greenspan 1985; Kruth 1984a,b, 1985; Kruth and Fry 1984). Immunohistochemical (IH) methods have been used to distinguish populations of smooth muscle cells and macrophages in atherosclerotic lesions (Masuda and Ross 1990; Rosenfeld et al. 1987; Tsukada et al. 1986), and many authors have used electron microscopy to characterize cell types and investigate intracellular aspects of atherosclerosis (Masuda and Ross 1990; Lupu et al. 1987; Rosenfeld et al. 1987). However, techniques such as these, which rely on the evaluation of tissue cross-sections, make the study of entire FS difficult. Moreover, biochemical techniques utilizing extraction procedures do not provide information about the spatial relationships of the lipids and cells composing the FS and lack the sensitivity to study the smallest FS.
We describe here a novel en face fluorescence technique that allows high-resolution mapping of the lipid composition of entire FS. Aortas were collected from New Zealand White rabbits fed an atherogenic diet. After fixation and dual fluorescent staining with filipin and Nile red, digital fluorescence microscopy was used to map areas of lipid accumulation in the aorta. Electron microscopy, immunohistochemistry, and biochemical methods were utilized to validate the staining technique and to identify the cell types and lipid deposits associated with the fluorescently stained regions.
Materials and Methods
Generation of Atherosclerotic Tissues
To generate the widest size range of atherosclerotic lesions, New Zealand White rabbits received an atherogenic diet (Purina Mills; St Louis, MO) of 0.05%, 0.1%, 0.2%, or 0.3% w/w for the first 6 weeks. They were then maintained on 0.15% cholesterol for 14 or 26 weeks. Plasma cholesterol was monitored weekly and averaged over time. The mean (± SD) plasma cholesterol was 596 ± 153 mg/dl for the rabbits sacrificed at 14 weeks and 431 ± 271 mg/dl for those sacrificed at 26 weeks. The rabbits were sacrificed by sodium pentobarbital injection and exsanguination. The aorta was removed, opened lengthwise, pinned out on cork boards, and fixed with phosphate-buffered formalin.
Nile Red/Filipin Fluorescent Staining
Stock solutions of Nile red (Sigma Chemical; St Louis, MO) (0.5 mg/ml in acetone) and filipin (Sigma) (2.5 mg/ml in dimethyl formamide) were prepared and stored at -20C. Pieces of aorta (0.5 X 1.0 cm) containing FS ranging from barely visible to <50 mm2, to those that occupied the entire circumference of the aorta, were trimmed and the adventitia removed by sharp dissection. The pieces were then washed three times for 15 min in PBS w/o Ca+2 or Mg+2, incubated in Nile red/filipin (100 μl Nile red stock + 200 μl filipin stock in 10 ml PBS) for 30 min, washed in PBS three times for 30 min, rinsed with distilled water, laid flat on glass slides, and coverslipped with Crystal Mount (Biomedia; Foster City, CA).
Fluorescence Microscopy
Stained aortas were examined with a Leitz Orthoplan light microscope equipped for epifluorescent illumination. The following filters were used to view stained aortas: 390-490-nm bandpass excitation and 530-550-nm cutoff barrier for Nile red and 340-380-nm bandpass excitation and 430-500-nm cutoff barrier for filipin. Images were recorded digitally from red-green-blue (RGB) video signals generated with a TEC-470 CCD video camera (Optronics Engineering; Golata, CA). RGB images were converted to hue-saturation-intensity (HSI) signals and the hue spectra for each lesion were plotted with ImagePro Plus software (Media Cybernetics; Silver Spring, MD). Finally, hue values were converted to wavelengths (nm) with a standard curve generated with gelatin Wrattan filters (Eastman Kodak; Rochester, NY). Nile red stained neutral lipids yellow (580–596 nm, 590-nm peak fluorescence) and polar lipids orange (597–620 nm, 600-nm peak fluorescence). Filipin-stained non-esterified cholesterol fluoresces blue (460–480 nm, 470-nm peak fluorescence). The Nile red and filipin images were overlaid digitally using OPTIMAS software (OPTIMAS; Edmonds, WA) to create composite images.
Microfluorimetry and Lipid Analysis by Thin-layer Chromatography
To correlate fluorescence intensity with biochemical measurements of lipid content of the lesioned aorta, 4-mm-diameter punch biopsies were taken from the aortas of rabbits after 14 weeks on the atherogenic diet and were imaged for Nile red/filipin staining (X 2 objective) as described above and subsequently processed for lipid analysis. The RGB signals were converted to HSI signals using Image Pro Plus software, the images thresholded for specific emission wavelengths and the fluorescence intensity for putative neutral lipids (Nile red emission 570–590 nm), polar lipids (Nile red emission 590–620 nm), and non-esterified cholesterol (filipin emission 460–500 nm) was measured in arbitrary units. After imaging, the punch biopsies were washed in distilled water, blotted, and subjected to Folch extraction [2:1 choloroform:methanol (v/v)] (Folch et al. 1957). The extracts were dried under nitrogen, brought up in a known volume, and spotted onto sintered silica rods (Chromorods, Type SIII) for thin-layer chromatography/flame ionization detection (TLC/FID). TLC were developed with chloro-form:methanol:water (45:20:2) twice for 3 min, followed by hexane:diethyl ether:formic acid (60:5:0.5) for 36 min. Lipids were measured using an Iatroscan MK-5 TLC/FID Analyzer (RSS; Costa Mesa, CA) and were quantitated against standard curves generated with known lipid standards (Sigma). Fluorimetry and biochemical data were plotted as scatter diagrams and linear regression analysis was performed using Table Curve software (Jandel Scientific; Corte Madera, CA). The statistical significance of the correlations was tested by student's t-test; p<0.05 was accepted as significant.
Immunohistochemistry
Paraffin sections of aorta containing lesions similar in size and location to those stained fluorescently were stained with either anti-smooth muscle α-actin monoclonal antibody ASM-1 (Boehringer Mannheim Biochemicals; Indianapolis, IN) or anti-rabbit alveolar macrophage monoclonal antibody RAM11 (Dako; Carpinteria, CA) for detection of smooth muscle cells and macrophages. The other reagents were purchased as a kit and staining was performed on a TechMate Automated Staining System (BioTek Solutions; Santa Barbara, CA). Serial 5-μm sections were cut from paraffin-embedded segments of aorta, mounted on Capillary Gap Plus slides (BioTek Solutions), and air-dried. Slides were deparaffinized, rehydrated, and steamed for 20 min in heat-induced epitope retrieval buffer. Endogenous peroxidase was blocked with 3% H2O2 in buffer for 15 min. Nonspecific binding was blocked with normal goat serum in PBS buffer for 15 min. The tissue sections were incubated with specific antibodies for 4 hr at room temperature and developed with biotinylated secondary antibodies (ABC) and peroxidase-conjugated streptavidin, using diaminobenzidine as the chromogen. Slides were then counterstained with hematoxylin, dehydrated, cleared, and coverslipped using a xylene-based mounting medium.
Fluorescence microscopy of fatty streaks after staining with Nile red and filipin.
Transmission Electron Microscopy
Lesions similar in size and location to those stained fluorescently were fixed in 3% glutaraldehyde in 0.1 M phosphate buffer, rinsed in buffer, postfixed in 2% buffered osmium tetroxide for 4 hr, dehydrated in a graded series of ethanols, and embedded in Embed 812 (Electron Microscopy Sciences; Ft Washington, PA). Semithin sections were cut on a Reichert Ultracut E, stained with toluidine blue, and lesions selected for thin sectioning. Thin sections were stained with uranyl acetate and lead citrate and examined in a JEOL 1200EX transmission electron microscope.
Results
Punch Biopsies
To ensure that the fluorescent staining accurately reflected the lipid composition of the lesions, punch biopsies (4-mm diameter) from aortas from rabbits collected after 14 weeks on the atherogenic diet were stained with Nile red and filipin, and fluorescent images were stored digitally. Examples of the images for a biopsy are shown in Figures 1A and 1B. Histograms of the fluorescence intensity of Nile red and filipin emission vs wavelength (Figures 2A and 2B) were generated. In the Nile red histogram, identifiable peaks at 600 nm (orange) and 593 nm (yellow) divided the histogram into two regions. The histogram of the filipin emission was monomodal, with a peak at 468 nm (blue).
Histograms showing fluorescence vs emission wavelength of filipin (excitation wavelength 340–380 nm)
Lipids were extracted from 11 punch biopsies after fluorescence spectral analysis and the lipid composition was analyzed by TLC/FID. Figure 2C is the chromatogram from the biopsy illustrated in Figures 1A and 1B and quantitated by spectral analysis in Figures 2A and 2B. Major peaks from cholesteryl-esters (CE), non-esterified cholesterol (C), and phospholipids (PL) were quantitated. Neither Nile red nor filipin was found to interfere with TLC/FID. Correlation analysis of fluorescence intensity and lipid composition is shown in Figure 3. In general, larger lesions had higher levels of fluorescence at all three wavelength ranges evaluated. As expected, larger lesions also tended to have a higher content of CE, C, and PL. In several smaller FS, Nile red and filipin fluorescence were measurable when the lipid content was below the limit of detection of TLC-FID (range 0.05–0.5 μg/biopsy, depending on lipid class), as evidenced by points plotted on the y-axis in Figure 3. Statistically significant correlations were found between filipin fluorescence and C content and between Nile red yellow fluorescence and CE content. Filipin therefore detects regions rich in non-esterified cholesterol and Nile red yellow detects regions rich in esterified cholesterol. Despite the lack of correlation with Nile red orange fluorescence (580–600 nm), a significant correlation was found between total Nile red fluorescence (580–620 nm) and total non-cholesterol lipid (CE + PL) content. This suggests that Nile red orange fluorescence detects areas of mixed lipid composition that are relatively polar in nature and therefore are probably enriched in phospholipids.
Early Stages of FS Formation
For detailed study of the lipid and cellular composition of developing FS, aortas were collected from five rabbits after 26 weeks on an atherogenic diet and were stained with Nile red and filipin. Adjacent areas were processed for immunohistochemistry and transmission electron microscopy to study cell composition or were observed by fluorescence microscopy before staining. This tissue provided areas of normal aorta, very small FS, and well-developed FS, which occupied the entire circumference of the aorta.
The smallest lesions observed were isolated areas (0.01–0.2 mm2) that stained solely with filipin. There was little or no autofluorescence in the emission range of filipin. Filipin staining at the edges of these lesions was ragged and feathery (Figure 1C). TEM of these regions revealed phospholipid vesicles (Figure 4A) in intimal spaces beneath an intact endothelium. In larger lesions, filipin staining was globular and brighter towards the center of the lesion (Figure 1C). TEM revealed cholesterol crystal clefts among densely packed aggregates of lipid vesicles at the bases of larger (> 1-mm2) lesions (Figure 4B).
Correlation analysis of fluorescence intensity and lipid composition.
Transmission electron microscopy (TEM) of a lesion collected after 26 weeks on an atherogenic diet.
The smallest FS that stained with Nile red was as minimal as a few cells in diameter (0.2–1 mm2) and fluoresced orange (Figure 1D). There was little or no autofluorescence in the emission range of Nile red. Staining at the perimeter of larger lesions was also orange, and occasionally corresponded with individual cell profiles (Figure 1F, asterisk). Immunohistochemical studies of small lesions showed positive staining for both macrophages and smooth muscle cells (Figures 5C and 5D). TEM revealed macrophages (Figure 5A) and smooth muscle cells (Figure 5B), with many secondary lysosomes containing phospholipid-rich inclusions. Smooth muscle cells were distinguished from macrophages by nuclear chromatin pattern, cytoplasmic fibers, myofilaments, and the presence of characteristically larger lipid inclusions. Cholesterol crystal clefts were seen occasionally in lysosomal bodies of foam cells (Figure 5A).
Transmission electron microscopy (TEM) of a lesion collected after 26 weeks on an atherogenic diet, showing cells from a region of orange fluorescence by Nile red.
After Nile red staining, in most lesions larger than 1 mm2 the center fluoresced yellow, surrounded by an orange fringe (Figure 1D). Yellow staining tended to be globular and occasionally reflective of individual cell profiles. IH indicated that larger lesions were composed primarily of macrophages and contained few smooth muscle cells (Figures 6C and 6D). Macrophages with many non-membrane-bound lipid droplets, commonly associated with neutral lipids, were seen by TEM (Figure 6A), often in the center and at luminal surfaces of lesions. Smooth muscle cells in these lesions also contained neutral lipid droplets (Figure 6B).
Nile red and filipin fluorescence images were digitally overlaid in Figure 1E. Small patches of filipin staining are interspersed between lesions that stained with Nile red. Regions of orange Nile red staining colocalized with blue filipin staining. Most larger lesions had a “nested” appearance, with bright yellow centers surrounded by orange and blue concentric fringes. A digital overlay of the lesion edge reveals Nile red-stained cell profiles and filipin-stained fringe detail (Figure 1F). Although only relatively small FS have been illustrated, the same pattern was observed at the edges of very large FS.
Discussion
In this study we developed a novel fluorescence microscopy technique for in situ, semiquantitative analysis of lipid deposits in atherosclerotic aortas. The method is based on the ability of the fluorescent dyes filipin and Nile red to selectively stain non-esterified cholesterol and polar and non-polar lipids, respectively. It is useful in routinely fixed tissues and has increased sensitivity and resolution compared to most previously described methodologies. En face preparation allowed visualization of whole lesions from the smallest “pre-lesional” changes to well-developed FS that circumscribed the aorta.
Key to the validity of the lipids maps is that filipin and Nile red stain different lipid moieties specifically. Filipin is a fluorescent probe that has been used extensively to study the distribution of non-esterified cholesterol in a variety of cells and tissues (Siefert et al. 1989; Jaakola et al. 1988; Kruth 1985; Kruth and Fry 1984). Correlation of biochemical and fluorimetry data demonstrated a strong correlation between the intensity of staining with filipin and the content of non-esterified cholesterol in tissue, indicating that the staining technique reliably maps regions rich in non-esterified cholesterol.
Nile red has been used previously to stain neutral lipids in cells and tissues (Frank et al. 1989; Haban and Mesarosova 1989; Brown et al. 1988; Lupu et al. 1987; Fowler and Greenspan 1985). In an aqueous environment, the fluorescence of Nile red is quenched. In contrast, in a non-polar environment Nile red is dequenched and fluoresces yellow. This characteristic allows Nile red to be used as a lipid-specific fluorescent stain. In the present study, a strong correlation was shown between the intensity of Nile red yellow emission (580–596 nm) and tissue cholesteryl ester content, as determined by TLC/FID. Similar results have been reported by Koren et al. (1990), who studied P3888D1 cells incubated with acetylated low-density lipoproteins and correlated Nile red microfluorimetry emission at 589 nm, with cholesteryl ester content measured by gas chromatography. These data indicate that the yellow emission of Nile red reliably maps regions rich in neutral lipids.
Another characteristic of Nile red is its ability to shift its fluorescence emission in response to the polarity of the lipid environment. This is in contrast to dyes such as acridine orange, for which concentration is the predominant factor (Ohkuma and Poole 1978). In a non-polar environment, Nile red fluoresces yellow and in a polar lipid environment the emission peak of Nile red shifts to longer wavelengths, e.g., orange, (Greenspan and Fowler 1985; Greenspan et al. 1985). This polar dependency in the Nile red emission spectra has been used to detect oxidation of LDL (Greenspan and Lou 1993) and to differentiate between neutral lipid and phospholipid-rich regions in cells and tissues (Klinkner et al. 1995; Henault and Killian 1993; Brown et al. 1992; Smyth and Wharton 1992; Hjelle et al. 1991; Bonilla and Prelle 1987). In the present study, although total Nile red emission intensity correlated with total non-cholesterol lipids, we were unable to directly correlate the orange emission of Nile red with phospholipid content. The lack of correlation may be due to the fluorescent quenching that accompanies the red shift in the emission spectrum of Nile red (Greenspan and Fowler 1985). In addition, because of a shift in the excitation maximum of Nile red, it is also possible that the filter system used in this work did not allow imaging the red-orange emission with maximal intensity. Despite the lack of correlation of fluorescence and biochemistry, TEM revealed lipid inclusions rich in phospholipids, as indicated by the presence of phospholipid whorls, in areas that were stained orange by Nile red. These data indicate that although the orange fluorescence of Nile red is not a reliable tool for quantitatively mapping phospholipids, it reliably maps regions containing a relatively polar mix of lipids.
In addition to validating the staining biochemically, we also studied the cell type and lipid inclusions responsible for the staining. TEM revealed that the smallest lesions, which stained solely with filipin, were composed of extracellular accumulations of lipid vesicles. Similar vesicles rich in non-esterified cholesterol have been described previously in the earliest atherosclerotic lesions in cholesterol-fed rabbits (Schwenke and Carew 1989b; Mora et al. 1987; Kruth 1984b) and in humans (Tirziu et al. 1995; Kruth 1984a), and correspond to the areas of increased endothelial permeability (Weinbaum and Chien 1993) and retention of LDL (Schwenke and Carew 1989a). Sometimes referred to as liposomes, these vesicles are believed to be derived from serum lipoproteins (Chao et al. 1994; Mora et al. 1987) and have been co-localized with apolipoprotein B in aortic lesions in rabbits (Mora et al. 1987) and humans (Tirziu et al. 1995). In the present study, TEM also revealed that the central regions of larger lesions (which stained intensely with filipin) contained aggregates of these lipid vesicles admixed with cholesterol crystals. These materials account for the very bright filipin staining seen centrally in larger FS.
Transmission electron microscopy (TEM) of cells from a region of yellow fluorescence by Nile red in a lesion from a rabbit sacrificed after 26 weeks on a high-cholesterol diet.
TEM and IH of the margin of the lesions (which corresponded to the orange fringe) revealed that they were composed of both macrophages and smooth muscle cells that contained lipid inclusions rich in phospholipids. Similarly, TEM and IH of the centers of the larger lesions (which corresponded to the intense filipin and yellow Nile red staining) revealed that they also contained macrophages and smooth muscle cells. However, in this region the cells contained predominantly neutral lipid inclusions. Finding both macrophages and smooth muscle cells in this locations is consistent with previous results showing that early fibrous plaques in rabbits contained relatively equal numbers of smooth muscle cells and macrophages centrally (Tsukada et al. 1986).
In the course of our work to develop the Nile red/filipin staining technique, we conducted a preliminary evaluation of the early stages of FS formation. We found that, in general, as the FS increased in size they also increased in complexity. In addition, although the lesions were heterogeneous, the increase in complexity followed a pattern. The smallest discernible lesions (∼0.01–0.2 mm2) stained solely with filipin. Larger lesions (∼0.2–1.0 mm2) had blue (filipin) fringes with a central region that stained orange with Nile red, and the largest lesions studied (∼2 mm2 up to FS that circumscribed the aorta) had concentric blue and orange fringes surrounding a central region that stained yellow with Nile red. This pattern, which did not follow any changes in distribution of SMC or macrophages, suggests that the complexity of the larger lesions cannot be accounted for by a simple change in cell composition. Rather, our data suggest that complexity results from changes in lipid composition, either in physical form (vesicular vs crystalline) or in chemical form (esterified vs non-esterified cholesterol, polar vs non-polar lipids) as the lesions grow in size. The change in lipid composition is probably the result of metabolic activity of macrophage and smooth muscle cell-derived foam cells: phospholipases, lipases, and acyltransferases working together to rearrange the lipid composition within a lesion (Laposata 1995; Aviram 1993; Brown and Goldstein 1983).
In conclusion, we have developed a new method for the investigation of the atherosclerotic lesions. En face fluorescent staining for neutral lipids, polar lipids, and non-esterified cholesterol discriminates regional differences in lipid composition of the lesions. These regional differences in lipid composition may prove more significant for understanding the evolution of a lesion than the origins of its cell composition. Further study using this dual staining method may reveal characteristic patterns of lipid deposition concurrent with the evolution and progression of atherosclerotic lesions.