Fluorescence Techniques Using Dehydroergosterol to Study Cholesterol Trafficking

Abstract

Cholesterol itself has very few structural/chemical features suitable for real-time imaging in living cells. Thus, the advent of dehydroergosterol [ergosta-5,7,9(11),22-tetraen-3β-ol, DHE] the fluorescent sterol most structurally and functionally similar to cholesterol to date, has proven to be a major asset for real-time probing/elucidating the sterol environment and intracellular sterol trafficking in living organisms. DHE is a naturally-occurring, fluorescent sterol analog that faithfully mimics many of the properties of cholesterol. Because these properties are very sensitive to sterol structure and degradation, such studies require the use of extremely pure (>98%) quantities of fluorescent sterol. DHE is readily bound by cholesterol-binding proteins, is incorporated into lipoproteins (from the diet of animals or by exchange in vitro), and for real-time imaging studies is easily incorporated into cultured cells where it co-distributes with endogenous sterol. Incorporation from an ethanolic stock solution to cell culture media is effective, but this process forms an aqueous dispersion of dehydroergosterol crystals which can result in endocytic cellular uptake and distribution into lysosomes which is problematic in imaging DHE at the plasma membrane of living cells. In contrast, monomeric DHE can be incorporated from unilamellar vesicles by exchange/fusion with the plasma membrane or from DHE-methyl-β-cyclodextrin (DHE-MβCD) complexes by exchange with the plasma membrane. Both of the latter techniques can deliver large quantities of monomeric dehydroergosterol with significant distribution into the plasma membrane. The properties and behavior of DHE in protein-binding, lipoproteins, model membranes, biological membranes, lipid rafts/caveolae, and real-time imaging in living cells indicate that this naturally-occurring fluorescent sterol is a useful mimic for probing the properties of cholesterol in these systems.

1. Historical Perspective

In order to resolve the dynamics and factors governing cholesterol trafficking within cells, it is important to recognize that the cholesterol molecule has very few polar constituents (Fig. 1A). Consequently, cholesterol is poorly soluble in aqueous environments such as cytosol as evidenced by critical micellar concentrations of cholesterol and other sterols being very low, near 20–30 nM (1–5). The physiological impact of cholesterol’s low aqueous solubility is that spontaneous cholesterol desorption from the cytofacial leaflet of the plasma membrane or lysosomal membrane into the cytosol for transfer to intracellular sites for esterification, oxidation, or secretion is extremely slow (t_1/2 3 hours-days) (6–9). Therefore, it is important to resolve factors that influence the solubility of cholesterol in the cytoplasm as well as to understand how the architecture and properties of cholesterol in the plasma membrane govern cholesterol movement from serum lipoproteins/albumin into the membrane exofacial leaflet, across the bilayer, and from the membrane cytofacial leaflet into the cytoplasm either as vesicles or to cholesterol binding proteins.

Fig. 1. Structures of cholesterol and fluorescent cholesterol analogs.

A. Naturally-occurring sterols: cholesterol and fluorescent sterol dehydroergosterol (DHE); B. Photo-activatable cholesterol analog, (FCBP); C. Cholesterol-rich microdomain preferring cholesterol analogs: Dansyl-cholesterol, (DChol); BODIPY-cholesterol analog (LZ260); D. Cholesterol-poor microdomain preferring cholesterol analogs: BODIPY-coprostanol analog 3; BODIPY-cholesterol analog LZ110a; 22-NBD-cholesterol.

Multiple physical approaches in model or fixed cell systems have been used to resolve the properties of cholesterol in membranes and the influence of cholesterol on membrane structure. Membrane cholesterol organization has been examined using electron microscopy to visualize the membrane at magnifications up to 100,000X (10). X-ray diffraction has been applied to membranes to determine cholesterol orientation and depth as well as determine lipidic crystal structures (11, 12). By using deuterated molecules and cooled neutron beams, neutron scattering profiles of phospholipid/cholesterol bilayers have been measured and neutron scatter off of the deuterium atom has been used to accurately determine cholesterol’s location within the membrane (11, 12). Electron scattering has had limited use due to high vacuum requirements but has yielded some packing orientation of cholesterol-rich domains (11–13). Calorimetry or thermal analysis resolves the effects of cholesterol level on lipid phase transitions in model membranes of various lipid compositions (14). Unfortunately the complexity of lipid compositions in biological membranes broadens/abolishes lipid phase transitions—thereby limiting the usefulness of calorimetry. By placing the membranes in a strong magnetic field with exposure to electromagnetic radiation, nuclear magnetic resonance (NMR) measurements involving the interaction with the nonzero spin of certain nuclei such as ¹H and ¹³C have provided not only detailed structural information (e.g. order parameter) on cholesterol but also its interaction strengths with phospholipid constituents of the membrane (12, 15, 16). While ¹H and ¹³C labeled cholesterols provide probes with structure most closely resembling that of cholesterol, it is difficult to incorporate large amounts of these labeled sterols into biological membranes and the time-scale of NMR measurements precludes resolution of cholesterol dynamics on the biological time-scale (17). Another technique, electron spin resonance (EPR) involves the exploitation of the paramagnetism of certain molecules with unpaired spins, usually ¹⁴N (15). Due to the nature of the low natural paramagnetism of lipids, EPR probes are synthesized with spin labels such as nitroxide free radicals (e.g. dimethyl nitroxide referred to as doxyl). The properties of doxyl labeled sterol analogs such as SL-cholestane, SL-cholesterol, and SL-androstane have recently been examined in POPC large unilamellar vesicles (LUVs). Using EPR, it was found that the SL-cholestane was moved deeper within the membrane as a result of the bulky doxyl replacing the OH group but with correct membrane orientation similar to the SL-cholesterol which had the doxyl replacing two methyl groups in the cholesterol tail (16). However, EPR showed that the SL-androstane molecule was upside down in membrane orientation. This was due to the hydroxyl at the terminal end of the lipid chain and the replacement of the OH by the doxyl group (16). Other techniques to examine cholesterol architecture in membranes have included radiolabeled methods, Raman scattering, and absorption techniques including infrared and optical rotation dispersion/circular dichroism. While these approaches have been very useful for studies of model membrane systems, purified biological membrane fractions in vitro, or membranes of fixed cells, they have not provided much insight into cholesterol dynamics in living cells.

Obviously, a great deal of effort has been made to design both suitable cholesterol analogs and the respective experimental methods to gain insights into the lipid domain structure and dynamics of biological membranes. Nevertheless, there is concern whether the behavior of the cholesterol analogues truly reflects that of cholesterol. Most of the probes have been chosen for the respective studies based on their close structural and organizational resemblance to the cholesterol molecule in both natural and artifical membranes. Until recently it was almost universally accepted that cholesterol represents the unique and essential building block of the cell membrane which defines, to a great extent, the normal function of biological organisms including man and most animals. Under this paradigm even small alterations in the chemical structure of cholesterol would be predicted to be incompatible with animal life (rev. in (18). But is cholesterol really essential for normal cell organization and function? How stringent are the stereo- and physico-chemical criteria for the membrane sterol molecule that creates a platform (environment) for membrane proteins? Several studies suggest that many aspects of the original paradigm—i.e. compatibility with life—may be met by sterols other than cholesterol: (i) Simple animal cells such as yeast (Candida tropicalis, Saccharomyces cerevisiae) and Red Sea sponge (Biemna fortis) synthesize DHE (Fig. 1A) as a major membrane sterol component (19–21). DHE differs from cholesterol by having an ergosterol side-chain (i.e. branched methyl and double bond), an ergosterol B ring (two double bonds), and a C ring with an additional double bond that maintains the extended cyclopentanophenanthrene ring structure (Fig. 1A); (ii). DHE completely replaces the requirement for cholesterol as the only sterol source in the diet of the worm C. elegans (19); (iii) Mouse L-cell fibroblasts lack the enzyme 3β-hydroxysteroid-Δ²⁴-reductase (desmosterol reductase, Dhcr24) (22). Desmosterol differs from cholesterol in having an extra double bound in the side chain. When cultured in chemically-defined medium the L-cells synthesize desmosterol, replace membrane cholesterol with desmosterol, and grow normally despite the absence of cholesterol (22, 23); (iv). Mouse L-cells cultured in chemically-defined medium containing dehydroergosterol (DHE) accumulate DHE which replaces as much as 90% of endogenous membrane sterol without adverse effects on membrane phospholipid or fatty acid composition, sterol/phospholipid ratio, activity of cholesterol sensitive enzymes in the plasma membrane, or cell growth (24). Similar observations have been made with cultured human fibroblasts and MDCK cells (25–27); (v). Most of the cholesterol in the developing and early neonatal rat retina can be replaced by desmosterol without alteration in function (28); (v) Ablation of the enzyme 3β-hydroxysteroid-Δ²⁴-reductase (desmosterol reductase, Dhcr24) in mice is not lethal and such mice exhibit only a mildly altered phenotype evidenced by disturbances occurring in steroid homeostasis (18, 29). The development of these viable “cholesterol-free mice”, where almost all of the cholesterol is replaced by desmosterol, shows that there is not an absolute requirement for cholesterol to maintain life (18, 29). It should be noted, however, that the same mutation in humans causes severe abnormalities, likely due to the inability of human embryos to access maternal cholesterol which is in contrast to mouse embryogenesis where maternal cholesterol is available (18). Taken together, these data suggest that the membranes of mammalian and other animal cells can tolerate small changes in the structure of the cholesterol side (desmosterol, DHE) and ring structure (DHE) and remain viable. However, not all small changes in the cholesterol structure are equally well tolerated. For example, loss of the Dhcr7 gene in mice (Smith-Lemli-Opitz syndrome in humans) results in accumulation of 7-dehydrocholesterol and very short-lived mice (rev. in (29). Much further work needs to be done to establish the exact substitutions/changes in cholesterol structure that can be accommodated to maintain viability.

2. Advent of fluorescent sterols

More recently fluorescence detection has been widely used to study cholesterol not only in vitro, but also in real-time imaging of fluorescent sterols (Fig. 1B-F) in living cultured cells or in vivo. Fluorescent sterols, like other fluorescent probes, provide ease of handling with very high level sensitivity of detection, relatively short detection times, and multiplexing of several probes (17). These characteristics of fluorescence detection have made it very beneficial in utilizing small quantities of fluorescent sterols not only for use with traditional spectrofluorometers but also in newly developed systems. In particular, fluorescent sterols and other probes have brought about the widespread use of laser scanning confocal microscopy (LSCM) and multiphoton laser scanning microscopy (MPLSM) to simultaneously study the interaction of lipids like cholesterol and proteins/receptors in real-time within living cells.

With rare exception, the fluorophores present in most commercially available fluorescent sterol (Avanti Polar Lipids, Alabaster, AL; Invitrogen, Carlsbad, CA; Sigma, St. Louis, MO) as well as those synthesized in individual laboratories are synthetic chemical tags that have been used to localize cholesterol in membranes. This includes the weakly fluorescent photoactivatable free cholesterol benzophenone FCBP that photocross-links to proteins or lipids (Fig. 1B), the cholesterol-rich microdomain preferring cholesterol analogs such as Dansyl-cholesterol (6-Dansyl-cholestanol) and BODIPY-cholesterol analog LZ260 (Fig. 1C), and the cholesterol-poor microdomain preferring BODIPY-coprostanol, BODIPY-cholesterol analog LZ110a, 22-NBD-cholesterol (Fig. 1D), 25-NBD-cholesterol (not shown), etc. (16, 30–38). Although there is considerable discussion regarding the solubility and orientation of these synthetic fluorescent sterol probes in membrane bilayers, especially 22-NBD-cholesterol and 25-NBD-cholesterol, at low concentrations some are thought to mimic the distribution and orientation of cholesterol (16, 35–37). However, relatively few have been examined with regards to their ability to be bound by intracellular cholesterol-binding proteins (e.g. SCP-2, ADRP) or their real-time uptake, distribution, and efflux dynamics in living cells. One exception is 22-NBD-cholesterol which has been shown to be not only bound with high affinity (nM K_ds) by SCP-2 and ADRP, but with orientation within the binding site similar to cholesterol—suggesting that this probe may be suitable for study of intracellular cholesterol trafficking (4, 39–44). Indeed, 22-NBD-cholesterol has proven useful for real-time imaging of HDL-mediated cholesterol uptake, efflux, and intracellular trafficking in living cells such as L-cell fibroblasts as well as rat macrophages and lymphocytes (43, 45, 46). Sterol carrier protein-2 (SCP-2) overexpression facilitated (i.e. decreased t_1/2 of efflux) of protein mediated NBD-cholesterol cytosolic efflux (43). Concomitantly SCP-2 overexpression decreased the slower, vesicular cholesterol transfer—likely by binding and sequestering ligands (i.e. phosphatidylinositides, fatty acyl CoAs) that regulate vesicular trafficking (43, 47–53).

However, there is a small group of naturally fluorescent lipids that includes sterols such as dehydroergosterol (DHE, Fig. 1A) as well as some fatty acids (parinaric acids), retinoids (retinol, retinoic acid), carotenes, and steroids (estriol). Because of the tremendous interest of biologists and membraneologists in cholesterol-rich microdomains (also called lipid rafts or caveolae) over the past 15 years (rev. in (54–63), DHE has emerged as a popular, potentially less perturbing fluorescent sterol probe since it does not contain additional bulky fluorescent groups added to the sterol structure. Although this ultraviolet (UV) light absorbing and fluorescent sterol (i.e. DHE) was first chemically synthesized over 50 years ago (64), it was its detection at high level in membranes of eukaryotes such as yeast Candida tropicalis (19) and Red Sea sponge Biemna fortis (21) that suggested its potential usefulness as a probe for cholesterol. Consequently, the focus of the remainder of the current review is on the use of naturally fluorescent DHE as a probe of cholesterol behavior in sterol-protein interactions, lipoproteins, model membranes, biological membranes, and lipid rafts/caveolae. Since cholesterol itself has very few structural/chemical features suitable for UV/VIS or optical imaging, the advent of DHE with many properties mimicking or very similar to those of cholesterol, has proven to be a major asset for probing/elucidating real-time sterol dynamics by video and MPLSM imaging in living cells. For additional insights, the reader is referred to earlier reviews on DHE, synthetic fluorescent sterols, spin labeled sterols, photoreactive sterols, and chemical tags of cholesterol (65–68).

3. Effect of Sterol Structure and Purity on Sterol Architecture and Dynamics in Membranes

Despite the fact that DHE is a naturally occurring fluorescent sterol appearing in sponge and yeast, it is not generally obtained from these sources. Instead, DHE from commercial (Sigma, St. Louis, MO) as well as laboratory sources is chemically synthesized by an adaptation of a 50 year old method (64, 69–71). However, synthetic DHE prepared by this established method, especially that obtained from commercial sources, may contain a significant sterol impurity, varying from batch to batch, and as high as 20–40% (43, 70, 71). It is important to note that not only the quantity, but also the chemical structure of the sterol and the presence of cholesterol oxidation products are known to significantly affect the structure and/or function of model membranes (rev. in (54, 55, 70, 72, 73), biological membranes (13, 74–80), lipid raft/caveolae (58, 81–88), and cholesterol-requiring plasma membrane receptors (66, 89–91). Because of this sensitivity it is essential that fluorescent sterols such as DHE not contain other sterols (e.g. ergosterol starting material or side products produced during DHE synthesis) or oxidation products. Such sterols/oxidation products may be non-fluorescent, potentially toxic (e.g. oxidized sterols), alter membrane structure, and/or not faithfully reflect the properties of cholesterol as accurately as pure DHE. This is even more important for real-time imaging of fluorescent DHE in living cells where such impurities may compete with DHE for uptake and/or may be toxic. Therefore, the following text deals with procedures for removing unreacted substrate (ergosterol), identifying side reaction or oxidation products, removing impurities, and modifying the synthesis procedure to prevent these artifacts. The DHE synthesis procedures described below tested ergosterol substrate obtained from several manufacturers (Steraloids, Wilmington, NH; Aldrich Chemical Co., Milwaukee, WI; Fluka Chemie, Steinheim, Switzerland; Sigma Chemical, St Louis, Mo), but all other reagents were in common from the same manufacturer: “Purified Plus” methanol, chloroform, and ethyl ether were from Burdick-Jackson (Muskegon, MI); “Baker Analyzed” 99⁺% pure glacial acetic acid was from Mallinckrodt Baker (Phillipsburg, NJ), 99⁺% pure acetic anhydride was from VWR (Atlanta, GA), and anhydrous mercuric acetate was from Fisher Scientific (Pittsburgh, PA).

3.1. Identification of potential impurities obtained through the classical DHE synthesis procedure

DHE was synthesized from ergosterol by a classical procedure established over 50 years ago (64, 69). Purity of DHE was determined by high performance liquid chromatography as described (69). The HPLC elution profile was monitored as follows: absorbance at 205 nm (all constituents), absorbance at 325 nm (DHE), and fluorescence emission at 375 nm (DHE). When the elution profile of commercially available DHE was monitored at 205 nm absorbance, at least two major peaks were obtained with retention times of 11.75 min and 15.3 min, respectively (Fig. 2A). The latter peak comprised at least 20% of total sterol To determine which of these peaks represented DHE, the elution profile was monitored by absorbance at 325 nm (Fig. 2B) and fluorescence emission at 375 nm (Fig. 2C). Only the larger peak near 11.75 min absorbed at 325 nm (Fig. 2B) and fluoresced (Fig. 2C). The smaller peak at 15.3 min comprising about 20% of total sterol was not DHE since it did not absorb or fluoresce at wavelengths typical of DHE. HPLC chromatograms (not shown) of DHE prepared by our laboratory from only one lot of ergosterol (Steraloids), but not from subsequent lots from the same or other sources, yielded pure DHE (>99%). All other preparations yielded DHE which was highly (20–40%) impure. To further resolve the impurity, complete absorbance spectra were obtained. The absorption spectrum of >99% pure DHE showed maxima at 310, 325 nm, and 342 nm (Fig. 3A). In contrast, absorbance spectra of impure synthetic DHE exhibited several additional peaks at 235, 242, and 251 nm (Fig. 3B), more clearly revealed by a difference spectrum (Fig. 3C). The latter absorbance maxima were not consistent with the starting material, ergosterol, which exhibited absorption maxima at much longer wavelengths 271, 281, and 293 nm (not shown). It is important to note that HPLC elution profiles, whether monitored at 205 nm or 242 nm (wavelength at which the impurity absorption was highest), were unable to clearly distinguish whether the impurity was unreacted ergosterol or an as yet unidentified compound with similar retention time (not shown). While mass spectra of the impure DHE detected DHE at (M+1)⁺ of 395.3 and (M-OH)⁺ of 377.5 as expected (Fig. 4A), an additional component appeared with (M+1)⁺ of 397.3 and (M-OH)⁺ of 379.3 (Fig. 4A). Surprisingly the APCI mass spectrum of this additional component was identical to that of ergosterol (Fig. 4B)—suggesting that the impurity was possibly an isomer of ergosterol. Indeed it was reported nearly 60 years ago that small amounts of nitric acid or mercuric nitrate catalyze not only the formation of the tetraene DHE but also a side product, i.e., mercurated triene, of ergosterol due to double bond rearrangement (92, 93). The spectral characteristics (i.e., maxima at 235, 242, and 251 nm) of the DHE impurity were the same as those of an isomer of ergosterol (5,7,22-cholestatrien-24β-methyl-3β-ol ) known as ergosterol D (64, 94). Ergosterol D is a triene [i.e. (7,9(11),22-cholestatrien-24β-methyl-3β-ol)] which differs from ergosterol (i.e. 5,7,22-cholestatrien-24β-methyl-3β-ol) in that the double bonds in the B ring are each displaced one carbon atom position.

Fig. 2. HPLC analysis of DHE.

A-C. DHE synthesized by established method as described in Methods. D-F. DHE synthesized by an improved method as described in Methods. Reversed phase HPLC was performed and elution of organic compounds was detected by absorbance at 205 nm (A, D). Elution of DHE was specifically detected by absorbance at 325 nm (B,E) and by fluorescence emission at 375 nm upon excitation at 325 nm (C,F). The peak with retention time at 11.75 min was due to DHE while that at 15.3 min was due to a non-fluorescent sterol.

Fig. 3. Absorption spectra of DHE.

A. Pure (~99%) DHE synthesized by an improved method as described in Methods. B. Commercially available DHE (Sigma). C. Difference spectrum was calculated by subtraction of (A) from (B). This spectrum shows the triene impurity alone. Note: it was necessary to use absorbance spectra to characterize the impurity, since the emission and excitation spectra determined the impurity was non-fluorescent (not shown). D. UV absorbance spectrum of ergosterol acetate. E. UV absorbance spectrum taken after 16 h dehydration of ergosterol acetate. F. UV absorbance spectrum of the resultant DHE acetate with impurity from the dehydration step.

Fig. 4. Atmospheric pressure chemical ionization (APCI) mass spectra of the DHE and ergosterol.

Mass spectral analysis was performed with a Thermo Finnigan LCQ Deca ion trap liquid chromatograph mass spectrometer (Thermo Finnigan, San Jose, CA) in atmospheric pressure chemical ionization (APCI) ion mode with sample flow rates of 200 μl/min near the quadrupole ion trap mass analyzer involving ultra-high purity helium gas. Samples were brought up in 1 μMolar solutions of equal volumes of MeOH and H₂O where both the methanol and H₂O were HPLC grade Burdick and Jackson (Muskegon, MI) purchased from VWR (Atlanta, GA). A. Mass spectrogram of impure DHE (Fig. 2A-C). The daughter ion resulting from the parent ion minus a hydroxyl group was indicated by (M-OH)⁺ while plus a proton was indicated by (M+1)⁺. The impurity was observed as a daughter ion resulting from the parent ion losing a hydroxyl group (M-OH)⁺ and gaining a proton (M+1)⁺. B. Mass spectrogram of commercially available ergosterol (99% pure), with the observed (M+1)⁺ and (M-OH)⁺ daughter ions. Note that the ergosterol had the same mass as the triene impurity.

Based on the above data, a general scheme of DHE synthesis and formation of the ergosterol D isomer was established (Fig. 5). This scheme suggests that the synthesis of pure DHE involved steps I, II (but not IIA), and III (Fig. 5). The presence of some as yet unknown factor (“impurity”) in starting material ergosterol, however, resulted in rearrangement in the double bonds in the B ring of ergosterol (essentially a competing side reaction) as shown in step IIA (Fig. 5). This side reaction product was an isomer of ergosterol, ergosterol D [i.e. (7,9,22-cholestatrien-24β-methyl-3β-ol)] distinct from the original ergosterol (5,7,22-cholestatrien-24β-methyl-3β-ol) and DHE. In further support of this scheme, the absorbance spectrum of each product during DHE was monitored (Fig.5). The UV absorbance spectrum of ergosterol acetate (step I, Fig. 5) exhibited three absorption maxima (Fig. 3D) essentially identical to pure ergosterol (not shown). Hence, ergosterol acetate formation did not alter UV absorbance spectrum of ergosterol. In Fig. 3E is shown the UV absorbance spectrum of the intermediate product taken 16 h into the 18 h dehydration step (step II, Fig. 5). This spectrum was broad with vibrational bands that were no longer distinct (cf. Fig. 3D). Fig. 3F clearly showed that during formation of DHE acetate (step II in Fig. 5), the side reaction product ergosterol D acetate (step IIA in Fig. 5) was concomitantly formed. Finally, saponification (step III, Fig. 5) of the DHE acetate and the ergosterol D acetate and recrystallization resulted in a UV absorbance spectrum (Fig. 3F) consistent with presence of both DHE (longer wavelength peaks) and ergosterol D (peaks between 230 and 260 nm) as indicated.

Fig. 5. General scheme of synthesis of DHE.

Step I is the conversion of ergosterol to ergosterol acetate. Step II is the dehydration of the ergosterol acetate to form a new double bond in the C ring. Step IIA is the competing side reaction which causes formation of the triene impurity in DHE. Note that this triene impurity had the same molecular mass as ergosterol because it was an isomer of ergosterol, i.e. ergosterol D. Step III is the saponification of the DHE acetate to yield DHE product.

3.2. Removal of impurities from DHE: Modified DHE synthesis procedure

The fact that HPLC, absorbance spectra, and mass spectral analysis did not detect any other organic contaminant in ergosterol starting material suggested that a water soluble inorganic contaminant might be a contributing factor to the formation of the non-fluorescent ergosterol D side reaction product. Therefore, in an effort to remove potential water-soluble contaminants, the ergosterol was dissolved in chloroform, washed at least three times with an equal amount of deionized water, dried under N₂, dissolved in hot methanol and recrystallized three times, and then dried under N₂. HPLC analysis to detect all constituents (absorbance at 205) revealed only a single peak was detectable, retention time 11.7 min (Fig. 2D). Likewise, when DHE was specifically monitored by absorbance at 325 nm (Fig. 2E) and fluorescence emission at 375 nm (Fig. 2F) in each case only a single peak was detectable, again with retention time of 11.7 min. HPLC chromatograms and absorbance spectra indicated that DHE synthesized by the modified method was ≥99% pure. Furthermore, APCI mass spectroscopy detected only a single peak with >99% (not shown).

4. Monomeric vs Crystalline DHE

Many of the spectroscopic properties of DHE as a fluorescent sterol probe have been elucidated in organic solvents (64, 65, 95) and in aqueous buffers (69, 70, 96). In organic solvents such as ethanol, DHE exhibits three UV excitation maxima (311, 324, and 340 nm) and three red-shifted emission maxima (354, 371, and 390 nm) characteristic of the DHE monomer. Interestingly, however, in aqueous buffers DHE exhibits a prominent spectral enhancement of the emission maxima near 404 nm and 426 nm as a possible result of excimeric emission—reflecting DHE in microcrystalline structures, but with only a slight (about 5 nm) red shift in both absorbance maxima and emission maxima (70). For instance, the respective excitation and emission maxima at 324 nm and 370 nm in ethanol shifted to 329 nm and 375 nm in aqueous such that the difference in the similar vibronic levels was maintained at 46 nm. This was consistent with solvent polarity (i.e dielectric constant) or pH having only a very minor effect on DHE and that charge separation does not occur. Many different organic solvents were used in a previous study to show that the Stokes shift varied between 3867 cm⁻¹ and 4000 cm⁻¹ (95). The DHE excitation and emission of the aqueous dispersion was similar to spectra of DHE polycrystalline powder; and both forms also exhibited similar lifetimes and corresponding fractions in a dual lifetime analysis (96).

Advantage has been taken of the distinction in the spectral characteristics to differentiate monomeric and microcrystalline (mostly in lysosomes) forms of DHE in living cells visualized by multiphoton imaging (70). In order to determine quantitatively the amount of the crystalline component and monomeric, the DHE was examined by fluorescence spectroscopy wherein the ratio of fluorescence intensity measurements at the wavelengths 426 nm and 373 nm was computed. Under monomeric conditions, such as the case in an organic solvent (e.g. ethanol) or low concentrations of DHE in model membranes (e.g. large unilamellar vesicles or LUVs), the ratio was measured as I₄₂₆/I₃₇₃ ~0.3 and would increase to approximately I₄₂₆/I₃₇₃ ~3.4 for an aqueous dispersion of DHE in a buffered solution with pH. 7.4 (70). Thus, the spectral properties of DHE give important insights on the cholesterol organization in membranes or organelles (e.g. lysosomes).

5. DHE Properties in Lipoproteins, Cholesterol-Binding Proteins, and Peptides

DHE has been directly incorporated into lipoproteins in vitro (97–99) or in vivo by feeding animals (rats, rabbits) dietary DHE followed by isolation of serum lipoproteins (100, 101). These studies showed that DHE was localized in the surface monolayer surrounding the neutral lipid cores of triglyceride/cholesteryl ester rich lipoproteins similarly as cholesterol. In addition, DHE and a similar synthetic fluorescent cholesterol analog (cholestatrienol, CTE) have been used to examine the structure of the sterol binding site of cholesterol-binding proteins such as sterol carrier protein-2 (SCP-2) (2, 39, 102–105) and liver fatty acid binding protein (L-FABP) (2, 106, 107). SCP-2 and L-FABP generally bound one mole of fluorescent sterol/mole protein. The fluorescent DHE or CTE were highly ordered in the binding pockets as indicated by markedly increased polarization as compared to aqueous buffer. The intermolecular distance between the DHE or CTE fluorophore and Trp or Tyr residues of these proteins measured by a fluorescence resonance energy transfer (FRET) in these studies was in the range of 16–19 Å, indicating close molecular interaction (108). Finally, FRET between Trp of mellitin (peptide from bee venom) and DHE incorporated in model membranes indicated that mellitin bound to the membrane, bound mellitin Trp was in close proximity to DHE (i.e. a few angstroms), and DHE was non-randomly distributed within the model membrane (109). Taken together these findings were consistent with DHE as a useful probe for examining the interactions and location of cholesterol in cholesterol-binding proteins and lipoproteins.

6. DHE Properties in Membranes

6.1. Model Membranes

The usefulness of DHE and cholestatrienol (CTE) as fluorescent cholesterol analogs in model membranes in a series of investigations spanning several decades demonstrated the usefulness of these probes to monitor cholesterol structural (polarization, limiting anisotropy, order) and dynamic (lifetime, rotational rate) properties in membranes (12, 65, 70, 95, 96, 108, 110–120). The lifetime studies resolved for the first time at least two DHE domains in model membranes, one less sensitive to the aqueous than the other, and that these domains were dependent on the lipid composition, temperature, and other properties of the membrane (54, 119). This possibility was further supported by model membrane studies of DHE exchange which provided kinetic evidence for the existence of multiple sterol containing domains in membranes (118, 121–124). DHE was especially useful for these studies because the spontaneous exchange of DHE between model membranes was very similar to that of radiolabeled cholesterol (117). Further, these dynamic and kinetic pools of DHE represent lateral sterol-rich and –poor domains rather than pools in the outer vs. inner leaflet because: (i) the transbilayer distribution of sterol across model membrane bilayers is uniform, unless membranes are very highly curved (rev. in (12)), (ii) the transbilayer migration rate of DHE is very rapid (much faster than kinetic exchange rates between membranes) and is very similar to that of cholesterol (108), (iii) Dynamic (lifetime) and DHE kinetic exchange studies determined that the presence of a large and a small DHE pool, (iv) DHE kinetic exchange studies determined that the large sterol pool corresponded to very-slowly (days) exchangeable sterol, while the small pool (generally <10%) was rapidly exchangeable. The pool size and half-times of exchange were dramatically affected by the presence of acidic phospholipids and, even more so by the presence of sphingolipids in the model membranes. The sterol carrier protein SCP-2 rapidly facilitated DHE transfer between model membranes by decreasing the half-time of exchange and increasing the size of the rapidly exchangeable domain. This action of SCP-2 was mechanistically due not only to the presence of a sterol binding site in this protein, but also the membrane interaction domain comprised on a positively charged face of an N-terminal amphipathic α-helix (125–127). These studies demonstrated that the properties of DHE and/or CTE closely reflected those of cholesterol in model membranes.

6.2 Suitability of DHE as a Probe for Cholesterol in Biological Membranes

DHE is non-toxic to animal cells as shown by its presence in membranes of yeast (Candida tropicalis, Saccharomyces cerevisiae) and Red Sea sponge (Biemna fortis) as well as its ability to completely replace cholesterol as the only sterol source in the diet of animals such as C. elegans or cultured L-cells (19–21, 24, 128). Unlike most synthetic fluorescent tagged sterols which can only be incorporated at very low amounts into cell membranes, DHE taken up by cultured cells can replace nearly 90% of total cholesterol without adverse effect on cell viability, plasma membrane structure, plasma membrane phospholipid composition or sterol/phospholipid ratio, or plasma membrane sterol sensitive enzymes (24). Plasma membrane receptors such as the oxytocin receptor are exquisitely sensitive to the structure of the sterol with which it interacts. In cholesterol-depleted membranes, only cholesterol and DHE are effective in reconstituting activity of this highly sterol-sensitive plasma membrane protein (89–91). In cultured cells DHE codistributes with cholesterol in plasma membranes, microsomes, mitochondria, and lysosomes (24, 129, 130).

6.3 Delivery of DHE into Cultured Cell Membranes

DHE may be delivered to living cells for uptake and incorporation into plasma membranes by several methods: (i) While relatively slow, the simplest is delivery as micro-crystalline DHE from a stock solution in ethanol over a 2 day period as described previously (24, 70). DHE can enter the cells by exchange between the micro-crystals and the cell surface plasma membrane, a process that may be facilitated by albumin in the medium (131–134). Concomitantly, the cultured cells also phagocytose DHE micro-crystals from the medium as shown by MPLSM colocalization with lysosomal markers (70). Over time the endocytosed DHE micro-crystals may be digested/disrupted in the lysosome or outside the lysosome by intracellular cholesterol-binding proteins such as SCP-2. Consequently, most DHE delivered in this fashion is codistributed similarly as cholesterol in all subcellular membranes including plasma membranes, plasma membrane cholesterol-rich lipid rafts, plasma membrane cholesterol-poor non-rafts, lysosomes, endoplasmic reticulum, and mitochondria (6, 7, 24, 26, 27, 129, 135–137). The presence of DHE micro-crystals in MPLSM images of cells cultured with DHE micro-crystals can be reduced by pulse labeling the cells with DHE micro-crystals for several hours followed by washing and incubation in DHE free medium for several days. (ii) An even slower, but more physiological method is to deliver DHE complexed to serum lipoproteins (100, 101). In this method the DHE is dissolved in organic solvent which is used to coat the inside of a tube or (to increase surface area) a tube containing celite (diatomaceous earth), the solvent is evaporated, serum lipoproteins are added in buffer with antimicrobial agent, incubated overnight with shaking at 37^oC, followed by sedimentation to remove particulates and filtration to assure sterility. (iii) A faster, but less efficient method to deliver DHE as a component of large unilamellar vesicles (LUVs) over a 1 day period as described previously (70). The DHE present in the LUVs enters the cell either by exchange with the cell plasma membrane (potentially facilitated by albumin in the medium) or, depending on the charge of the LUV, by facilitated fusion with the cell plasma membrane. While the LUV method completely avoids the presence of micro-crystalline DHE, less total DHE is incorporated into the cells since incubation with larger amounts of LUVs concomitantly results in incorporation large amounts of carrier phospholipid which itself may alter cell membrane properties depending on the specific phospholipid polar head group (e.g. choline, ethanolamine, etc) and esterified fatty acids (saturated, monounsaturated, polyunsaturated). (iv) The most rapid method for delivery of large amounts of DHE into cells is to deliver DHE as dehydroergosterol-methyl-β-cyclodextrin (DHE-MβCD) complexes prepared by adding DHE to an aqueous solution of MβCD (3mM DHE and 30mM MβCD). This mixture was overlaid with N₂, continuously vortexed under light protection for 24 h at room temperature, and filtered through a 0.2 μm filter to remove insoluble material and DHE crystals. Then, 20 μg of DHE was added to the cells in the form of DHE-MβCD complexes and allowed to incubate for 45 minutes at room temperature in PBS. Prior to imaging, cells were washed three times with PBS. With this technique DHE enters the cell by exchange between the DHE-MβCD complexes and the cell plasma membrane. The cyclodextrin-based method for DHE delivery to the cell may proceed, in addition to the clathrin mediated pathway, through the caveolae-dependent endocytosis due to non-selective cholesterol loading onto the plasma membrane by cyclodextrin. The major drawback of this method is that large excess of MβCD by itself can deplete cells of cholesterol while very high quantities of DHE-MβCD at high molar ratios can increase total sterol content. Therefore, care must be taken to optimize the DHE-MβCD molar ratio and concentration in the medium to maintain unaltered total sterol content in the cell and cell membranes.

In summary, the above methods of DHE delivery differ in rapidity, the amount of DHE that can be delivered, and potentially DHE distribution in the cell may differ from one loading method or another. The latter may be an advantage of a particular method if it is desirable to preferentially label a specific cellular compartment. Alternately, if uniform distribution of DHE is desired, this potential problem may be minimized if the cells are incubated over sufficient time for the DHE to equilibrate between all cellular compartments.

6.4 DHE Architecture in Purified Biological Membranes

DHE has been used to probe the structural (polarization, limiting anisotropy, order), dynamic (lifetime, rotational rate), and kinetic (exchange) environment of cholesterol in plasma membranes (6, 7, 54, 55, 135, 138–141), microsomes (6, 7), mitochondria (6, 7, 142, 143), and lysosomes (8, 129). DHE has revealed several important structural and dynamic aspects of the cholesterol microenvironment in biological membranes.

First, in contrast to model membranes, DHE was found to be asymmetrically distributed across the plasma membrane bilayer (rev. in (12, 144, 145). DHE revealed that the cytofacial leaflet of plasma membranes from L-cell fibroblasts (12, 24, 146, 147), erythrocytes (144), and brain synaptosomes (145, 148–151). While other methods (filipin staining, exchange, cholesterol oxidase accessibility, neutron diffraction, etc) have also been used to measure transbilayer cholesterol distribution, many have significant limitations including accessibility of membrane cholesterol (filipin, cholesterol oxidase), production of a potential perturbant (cholesterol oxidase produces cholesterone), and poor dynamic resolution (neutron diffraction) (rev. in (12). While measurements of transbilayer sterol distribution by different methods may therefore not necessarily agree, multiple methods (DHE, NBD-cholesterol quenching, and exchange assays) indicate that the cytofacial leaflet of erythrocytes has more cholesterol than the exofacial leaflet (144, 152). Likewise, several methods (DHE, cholesterol oxidase) indicate that the synaptosomal plasma membrane has more cholesterol in the cytofacial leaflet than the exofacial leaflet (12, 145, 148–150). The transbilayer distribution of sterol is a major determinant of the transbilayer fluidity gradient present in these membranes, the cytofacial cholesterol-rich leaflet is more rigid (less fluid) than the relatively cholesterol-poor exofacial leaflet (rev. in (146, 147, 149, 153–159). The transbilayer migration rate of DHE in model membranes and plasma membranes is relatively fast, t_1/2 of sec-min depending on the membrane examined (rev. in (12, 108, 144, 145). Rapid transbilayer migration of DHE is also supported by the finding that uptake of DHE into and non-vesicular transport across cultured hepatocyte-derived cells (i.e. from basolateral to canalicular plasma membrane) is also very rapid, i.e. t_1/2 of 1–3 min (160). The finding of rapid transbilayer sterol movement suggested that the asymmetric transbilayer distribution in the plasma membrane was not due to restricted movement of cholesterol from one leaflet to the other, but was instead a property of the lipid bilayer itself. DHE transbilayer distribution has broad applicability in studies of the action of anesthetics, cholesterol lowering drugs (statins), alcohol (chronic and acute), lupus erythematosus, aging, apoE, and sterol carrier proteins all of which can alter plasma membrane or synaptosomal plasma membrane transbilayer sterol distribution and/or transbilayer fluidity (26, 135, 136, 145, 149, 150, 154, 161–172).

Second, the DHE lateral distribution in both model and biological membranes was not uniform, but instead reflected sterol-rich and poor domains (rev. in (54, 55). The exchange kinetics of DHE and radiolabeled cholesterol between model membranes was indistinguishable (117). For a kinetic sterol exchange assay between biological membranes, cellular organellar membranes can be isolated from cells supplemented with DHE to act as donors and without supplementation of DHE to act as acceptors. Not only exchange kinetics but also lifetime analysis resolved multiple sterol domains; and cholesterol binding/transfer proteins such as SCP-2 (and less so L-FABP) rapidly facilitated DHE transfer from these purified membrane fractions to other intracellular membranes in vitro (54, 55). With regards to cholesterol movement between purified intracellular organelle membrane fractions, the most rapid spontaneous DHE transfer occurred from mitochondrial membranes to microsomal membranes, mitochondrial to lysosomal and to plasma membranes, and plasma membranes to microsomal membranes (173). Least rapid spontaneous DHE transfer occurred from lysosomal to plasma membranes. However, in the presence of the SCP-2, the DHE transfer was markedly enhanced in the approximate order of plasma membrane to microsomal membranes and plasma membranes to lysosomal membranes (173). This order was followed by enhanced transfer from mitochondrial membranes to microsomal membrane, mitochondrial membranes to plasma membranes, and a nearly equal transfer between mitochondrial and plasma membranes. Other than the transfer between mitochondrial and plasma membranes, the reverse transfer was markedly slower in both spontaneous and SCP-2 mediated (173). Thus, in vitro exchange dynamics of DHE revealed that: (i) spontaneous sterol transfer was very slow, but vectorial; (ii) spontaneous sterol transfer from lysosomes and lysosomal membranes was extremely slow, despite the fact that this is the route of LDL-receptor mediated cholesterol uptake; (iii) cholesterol-binding proteins such as SCP-2 rapidly enhanced the extent and directional transfer of sterol transfer from lysosomes and lysosomal membranes to the plasma membrane.

6.5. Physiological Significance of in vitro Studies with DHE and other Fluorescent Sterols: Cultured Cells and Animals

The significance of the above findings in living cells was confirmed with DHE and other fluorescent sterols in a variety of cultured cells including L-cells, CHO cells, macrophages, and hepatocyte derived cell lines (39, 43, 45, 160, 174–176). Studies with transfected L-cells showed that uptake and efflux of fluorescent sterols was highly dependent on the expression of cholesterol binding proteins (e.g. such as SCP-2) in the cytosol. Furthermore, transhepatocyte transfer of DHE was very rapid and not vesicular. Since hepatocytes have very high levels of intracellular cholesterol binding proteins (e.g. SCP-2, L-FABP) taken together such data suggested that rapid transhepatocyte cholesterol trafficking from the basolateral to canalicular membrane for efflux was facilitated by these proteins in the cytosol. The physiological impact of cholesterol binding proteins such as SCP-2 and L-FABP on cholesterol uptake, esterification, and excretion into bile was confirmed by studies with SCP-2 transgenic (i.e. overexpressing or antisense cDNA treated) mice and rats (177–179) and gene-targeted mice wherein SCP-2/SCP-x (180–182), SCP-x (183), or L-FABP (184–189) were ablated.

Studies with radiolabeled cholesterol were consistent with many of the findings shown with the fluorescent DHE and other sterols in cultured cells (43, 45, 172, 190–197). In one study, it was shown that overexpression of L-FABP in L-cell fibroblasts enhanced the transfer or radiolabled cholesterol from the plasma membrane to the endoplasmic reticulum for esterification (194). Transfer was inhibited by drugs that bound/competed with cholesterol for the L-FABP binding site. In another study L-cells were transfected with the cDNA that encodes for the 13.2 kDa and 15 kDa SCP-2 proteins respectively. In nearly all tissues and cells examined the 15 kDa SCP-2 undergoes complete N-terminal post-translational cleavage to yield the mature 13.2 kDa SCP-2 (rev. in (173). Despite the fact that only the 13.2 kDa protein was found in both cases, different effects on the intracellular trafficking of cholesterol were observed (191). In early experiments using radiolabeled cholesterol, 15 kDa SCP-2 transfected L-cells showed an enhanced uptake of [³H]cholesterol after 4 hours including a resultant increase in esterification of the exogenous radiolabeled cholesterol as shown by the 30% increase in [³H]cholesteryl ester levels with no enhancement in the 13.2 kDa SCP-2 transfected L-cells as compared to the control cells. Both the 15 kDa and the 13.2 kDa SCP-2 transfected L-cells showed a significant increase (15 and 11-fold, respectively) in the initial [³H]cholesteryl ester synthesis rate compared to control cells after treatment with sphingomyelinase. This experiment was important in order to examine the effect of SCP-2 expression on the intracellular movement of cholesterol from the plasma membrane to the endoplasmic reticulum for esterification (192). The resulting data corresponded with the aforementioned fluorescent in vitro assays where greater enhancement occurred in the direction of exchange of DHE from plasma membrane to microsomal vesicle (6, 173). The increased rate in ester synthesis resulted in a 1.5 increase in the [³H]cholesteryl ester levels, while at saturation (1 hr treatment), the radiolabeled cholesteryl ester levels were increased 1.6 and 1.3 fold for 15 kDa and 13.2 kDa SCP-2 overexpression (192). Inhibition of microsomal cholesteryl ester synthesis by drug treatment in the sphingomyelinase treated cells showed no differences. The overexpression of 13.2 kDa SCP-2 did not elevate cholesterol homeostasis of free cholesterol while elevating esterified cholesterol levels (192). Treatment with [³H]oleic acid revealed that the 15 kDa SCP-2 overexpression resulted in a specific increase of cholesterol esterification as opposed to the 13.2 kDa SCP-2 overexpression which resulted in esterification into triacylglycerols(192). Using [³H]cholesterol and the fluorescent 22-NBD-cholesterol in conjunction with high density lipoproteins (HDL), L-cells overexpressing SCP-2 were shown to inhibit the HDL mediated efflux of cholesterol of 61% and 157%, respectively (43). However, most of the inhibition occurred in the slower component while the faster component of the efflux pool (protein-mediated) was somewhat less affected. Clearly, both the radiolabeled cholesterol uptake and HDL-mediated 22-NBD-cholesterol efflux studies in living cells reveal that the effects of SCP-2 expression on cellular cholesterol transport become increasingly significant over longer time periods.

Overexpression of SCP-2 also had an inhibitory effect on the in vitro enhancement of SCP-2 on the exchange of DHE and cholesterol between purified lysosomal membranes as seen by the initial rates (8). Lipid analysis of lysosomal membranes revealed a significant decrease in the cholesterol to phospholipid ratio as a result of a small decrease in the cholesterol mass with a corresponding increase in phosphatidylserine; however, lyso-bis-phosphatidic acid (LBPA), which is involved in lysosomal cholesterol trafficking, was dramatically decreased by nearly 3-fold. Furthermore, sterol exchanges involving lysosomal membranes isolated from normal CWN human fibroblasts and NPC1 human fibroblasts revealed an apparent similar effect (25). The exchange of sterol from the fast kinetic pool was increased dramatically (2-fold increase in initial rate) upon addition of SCP-2 in vitro for the NPC1 lysosomal membranes when compared to the control CWN lysosomal membranes as determined by DHE polarization exchange assays. The decreased expression of SCP-2 in NPC1 fibroblasts, as seen in murine models of Niemann-Pick type C disease(198), allowed for a higher exchange rate of the fast kinetic pool as compared to CWN fibroblasts with larger normal expression of SCP-2. In fact, the lysosomal membranes from CWN human fibroblasts showed much slower spontaneous initial rate of sterol exchange as well. However, the trend was reversed in the slower but larger kinetic pool, where higher levels of SCP-2 caused a higher rate of sterol transfer in vitro. Both cell types exhibited similar sterol kinetic pool fractions for the isolated lysosomal membranes (25) despite differences in sterol transfer rates. With the exception of the pool fraction sizes, sterol transfer characteristics examined as a function of expression of SCP-2 in lysosomal membranes isolated from control and transfected murine fibroblasts appeared consistent with those seen in CWN and NPC1 human fibroblasts (8) possibly through membrane lipid domain alteration and/or involvement with LBPA (199). Exchanges between lysosomal donors to mitochondrial acceptors revealed similar results whereas SCP-2 overexpression removed the fast kinetic pool but enhanced the slower kinetic pool both in rate and in pool size (173). In the reverse exchange (mitochondrial donor to lysosomal acceptors), no fast kinetic pool was detected in the control. However, SCP-2 overexpression still reduced the early exchange of sterol as seen by the 6-fold reduction in initial rate by shifting most of the sterol into the non-exchangeable pool (173). In part, this might explain the large pool of non-exchangeable sterol observed in exchange assays using mitochondrial membrane vesicles isolated from steroidogenic MA-10 Leydig cells (143).

Thus, modulation of directional cholesterol transport and cholesterol esterification as well as membrane lipid composition has been demonstrated to occur as a function of expression levels of L-FABP or SCP-2, possibly through the interaction of proteins/receptors involved with cholesterol trafficking and lipid sensing/signaling. The resulting evidence has been obtained with assays using different labeled sterols including the fluorescent cholesterol analog DHE.

7. Plasma Membrane Sterol-rich Microdomains: in vitro studies

Based on findings with delivery of dansyl-cholesterol, the method and duration of fluorescent sterol delivery (e.g. micro-crystals, lipoproteins, LUV, or cyclodextrin complexes) may affect the distribution of fluorescent sterol to lipid rafts/caveolae or non-rafts (see Section 6.3 and ref. (32). When transformed cells (L-cells, MDCK cells) or primary mouse hepatocytes were cultured with DHE (micro-crystals or LUV) under conditions to maximize DHE incorporation and equilibration, biochemical fractionation by affinity chromatography of purified plasma membrane vesicles yielded sterol-rich lipid rafts/caveolae, caveolae, and lipid rafts, respectively (26, 27, 70, 136). DHE was codistributed similarly as cholesterol into sterol-rich and –poor domains (26, 27, 70, 82, 102, 135, 136, 172, 182, 196). Lipid raft/caveolae associated receptors such as the oxytocin receptor are sensitive to the structure of the sterol with which it interacts and both cholesterol and DHE, but not other sterols, are effective in reconstituting activity of this highly sterol-sensitive plasma membrane protein in cholesterol-depleted lipid rafts/caveolae (89–91). DHE mobility (fluorescence polarization) differed significantly in lipid rafts/caveolae from non-rafts. While spontaneous DHE transfer from lipid rafts/caveolae was relatively slow, that from non-rafts was essentially non-detectable. SCP-2 dramatically enhanced DHE transfer from lipid rafts/caveolae, but not from non-raft domains. Further examination of the plasma membrane cholesterol partitioning revealed that sterol transfer, as resolved using DHE in fluorescence polarization kinetic assays, from caveolae/raft domains with introduction of SCP-2 was enhanced as compared to plasma membrane vesicles (136). Principally, the initial rate of the caveolae/raft fractions was increased by a factor of 5 with a larger portion of the sterol arranged in an exchangeable kinetic pool as compared to the overall plasma membrane fraction (136). A possible mechanism for this enhancement may involve: (i) SCP-2 directly binding cholesterol as shown by fluorescent sterol binding assays and cross-linking by photoactivatable cholesterol (Fig. 1B) (39–41, 44); (ii) SCP-2 directly interacting with caveolin-1 within the plasma membrane as observed from in vitro (coIP, CD) and in vivo (two hybrid, double immunofluorescence FRET, double immunogold EM) assays that revealed an average molecular interaction distance of ~48Å was measured (200). Specifically, SCP-2 has been shown to directly interact with the N-terminal sequence of amino acids of caveolin-1 (201). Finally, DHE transfer from lipid rafts/caveolae to serum lipoproteins was remarkably specific for the type of lipoprotein/apoprotein, while that to non-rafts was very slow and not specific (27). The importance of resolving cholesterol structure and dynamics in these cholesterol-rich plasma membrane microdomains is underscored by the fact that lipid rafts/caveolae function not only in reverse cholesterol transport, but also in cell recognition, signaling, immune function, and potocytosis (rev. in (58, 60, 82, 84, 85, 87). Furthermore, lipid rafts/caveolae were shown to mediate the action of potential bioterror toxins and serve as entry portals for a host of bioterror pathogens (rev. in (82, 202).

8. Plasma Membrane Sterol-rich Microdomains: real-time imaging of DHE in plasma membranes of living cells

Recent improvements in imaging microscopy allowed the direct, real-time visualization of DHE in living cells by conventional (DHE delivered as micro-crystals) or video (DHE delivered as cyclodextrin complexes) UV fluorescence microscopy (39, 160, 174–176) and by MPLSM (DHE delivered as LUV or micro-crystals) multiphoton laser scanning microscopy (71, 82, 203, 204). Analysis of intracellular localization and distribution of DHE at the plasma membrane can be accomplished after segmentation of specific regions of interest such as cellular organelles and performed on a cell by cell basis (70, 203). Previously, two techniques involving image segmentation of the plasma membrane of cells labeled with DHE and multiple fluorescent probes (i.e. Nile Red, ECFP-Mem, and DiO) were validated and compared (203): (i) subtraction and sliding window and (ii). rank statistic-based methodology. The subtraction technique was performed with the DHE and Nile Red colocalization experimental results by subtracting the DHE channel from the Nile Red channel. This procedure worked very well with Nile Red as it only very weakly stained the PM as compared to the significantly brighter labeling by the DHE. Noise was reduced by using a sliding window analysis technique where window size, intensity thresholding, and the signal to noise ratio was optimized. In the rank stastic-based technique, two small windows are created in the image and the pixel intensities in each window ranked from the lowest to highest; then a comparison is made by using Miller’s rank statistic. The solution is optimized based upon intensity thresholding. The segmented plasma membrane regions were smoothed using geometric moments functions based on the intensity distribution within the segmented PM. Subsequently, geometric reference and spatial intensity measurements were calculated. Spatial statistical analysis was performed using pixels with DHE intensities higher than the median intensity to test for complete spatial randomness (CSR). Monte Carlo simulations were run using the narrowed data set (peak intensities—relative higher concentrations) as well as the complete data set (203).

8.1. Multiphoton imaging of DHE incorporated into L-cells from ethanolic stock diluted in aqueous medium

L-cell fibroblasts grown on chambered coverglass were incubated with the fluorescent sterol analog, DHE, from ethanolic stock solution as described above and imaged by multi-photon microscopy. Despite multiple washings of the cells, images from a single channel covering the range 350–450 nm revealed very bright patches of fluorescence emission (Fig. 6A). By monitoring the ratio of the 455 nm region to 375 nm region, the areas that were either mostly micro-crystalline or mostly monomeric were distinguished (70). The micro-crystalline DHE phagocytosed by the cells was largely found to be in lysosomes (70) but breakdown of the microcrystals occured over time (~2-3 days) such that monomeric DHE was distributed throughout the cell into the plasma membrane, lipid storage droplets, endoplasmic reticulum, and other organellar membranes. This was also observed in organelles isolated from cells incubated for several days on DHE from ethanolic stock solution (6, 8, 70, 102, 129). As a result of the increased localized concentration of micro-crystalline DHE and its enhanced excimeric fluorescence, difficulties arise in detecting low to moderate amounts of monomeric DHE (Fig. 6A:inset). This is due to the intense emission arising from crystalline DHE which can cause saturation of the PMTs (70, 204). Plasma membrane sterol can be detected by imaging cells that do not have residual large amounts of crystalline DHE.

Fig. 6. Real time multiphoton imaging of DHE in living L-cell fibroblasts.

L-cells were plated onto chambered coverglass incubated with DHE by three different techniques and imaged by multi-photon laser scanning microscopy using a MRC-1024MP system with external non-descanned detection. Excitation occurred at a wavelength of 900 nm using a femto-second modelocked titanium:sapphire laser and the emission was detected through a D400/100 and UV 440LP dichroic filter. A. Image of L-cells that were incubated for two days with 20μg DHE from ethanolic stock solution added to 1 mL media (ethanol < 0.5%) wherein large accumulation of crystalline DHE occurred. The high intensities (crystalline DHE) was kept under saturation levels and within the instrumental dynamic range; thus lowering the sensitivity of detection of monomeric DHE within the plasma membrane. B. Image of L-cells that were incubated for one day with 20μg DHE in the form of large unilamellar vesicles (65:35 mol % POPC:DHE). C. Image of L-cells that were incubated for 45 minutes with 20μg dehydroergosterol methyl-β-cyclodextrin complexes. Insets: Regions that included a section of the plasma membrane and outlined in a white rectangle were magnified 2.5X. Bars = 20 μm.

8.2. Multiphoton imaging of DHE incorporated into L-cells from LUV diluted in medium

DHE in the form of LUVs was taken up by living cells (Fig. 6B) in largely monomeric form as indicated by the emission peak ratios (70). There was almost no appearance of DHE micro-crystals (70). As a result, sensitivity of detection can be increased by using an emission dichroic/filter combination with broader bandwidth and higher photomultiplier gain. Thus, DHE with its low quantum yield can easily be detected at the plasma membrane (Fig. 6B: inset) and in intracellular compartments as had been observed from previous spectroscopic studies(8, 70, 102, 135, 172). Signal saturation may occur as a result of accumulation of unesterified or esterified monomeric DHE in lipid storage vesicles (Fig. 6B), distinct and much less intense than microcrystalline DHE observed in lysosomes upon culturing cells with DHE micro-crystals. MPLSM images of cells supplement with DHE as LUV revealed that DHE was non-randomly distributed into sterol-rich and poor- domains within the plasma membrane of living L-cells (70, 203, 204).

8.3. Multiphoton imaging of DHE incorporated into L-cells from DHE-MβCD complexes diluted in aqueous medium

L-cell incubation with DHE-MβCD complexes also yielded the monomeric DHE form as visualized by real-time multi-photon imaging in a confocal slice through the cell (Fig. 6). Within 45 minutes, heterogeneous regions of the plasma membrane were strongly labeled with DHE (Fig. 6C:inset), evidenced by the strong patchy outline at the cellular periphery. Other intracellular membranes were weakly labeled throughout the cell (Fig. 6C). In many of the cells, morphological structures such as microvilli and filopodia were also visible as a result of the strong signal from the fluorescent DHE. The non-random distribution of DHE in to sterol-rich and poor- domains at the plasma membrane was consistent with other studies wherein DHE was supplemented as LUV to the cells (70, 203, 204). This would suggest that supplementation of cells with DHE-MβCD complexes also non-selectively loaded the different domains in the plasma membrane.

8.4. Image analysis of DHE distribution after multiphoton imaging

While MPLSM and visual inspection of DHE at the plasma membrane of cells supplemented with DHE as micro-crystals, LUV, or DHE-MβCD complexes revealed that DHE was non-randomly distributed into sterol-rich and poor- domains, a mathematical framework substantiating this observation requires additional analysis (70, 203, 204).

Image analysis originally consisted of intensity measurements using region of interests and the use of colocalization, ratiometric, and fluorogram techniques. Further structured segmentation of the plasma membrane has permitted examination of the distribution of DHE through the use of inferential statistics, utilizing hypothesis testing and correlation determination, and has opened up an opportunity to do distributional modeling (71, 203). Segmenation analysis of the MPLSM images of cells supplement with DHE as LUV confirmed that DHE was indeed non-randomly distributed into sterol-rich and poor- domains within the plasma membrane of living L-cells (203, 204).

9. Summary and Discussion

The naturally-occurring fluorescent sterol DHE (19, 21) has proven a powerful probe for use in investigating membrane structure and cholesterol domain dynamics in vitro (8, 54, 55, 65, 70, 102, 172), for examining cholesterol-protein interactions in vitro (2, 39, 96, 101, 102, 108, 205), and for first time visualization in real-time of the distribution of cholesterol in the plasma membranes in living cells (70, 203, 204). Successful application of the fluorescent DHE to investigation of the function and organization of sterols in membranes, especially lipid raft/caveolae microdomains of living cells, requires the use of highly purified DHE. This is due to the fact that lipid rafts/caveolae are highly sensitive not only to cholesterol content (58, 60, 82, 84, 85, 87, 88), but also to sterol structure (89), and sterol oxidation. The present review yielded new insights into this problem and potential applications of DHE imaging in living cells.

The presence of a non-fluorescent sterol impurity was observed not only in commercially available DHE, but also in DHE synthesized in the present laboratory from ergosterol obtained from multiple commercial sources. The appearance of the non-fluorescent sterol impurity was somewhat variable (20–40%) depending on the commercial source and batch of the ergosterol substrate but without dependence on any other reagent used for the synthesis. While both HPLC and APCI mass spectroscopy suggested that the impurity was either ergosterol or an ergosterol isomer, the unique UV spectral absorption characteristics of this impurity provided evidence that the non-fluorescent sterol impurity was an isomer of the initial ergosterol (5,7,22-cholestatrien-24β-methyl-3β-ol). This isomer, known as ergosterol D [i.e., 7,9(11),22-cholestatrien-24β-methyl-3β-ol], was not present in the ergosterol starting material but potentially arose from the rearrangement of double bonds in the B ring of the ergosterol during the dehydration step (Step IIA, Fig. 5). Purifying the ergosterol created a non-detectable yield of the ergosterol D impurity during DHE synthesis. While the nature of the water soluble component facilitating formation of the side reaction ergosterol D isomer was not identified, it has been reported that small amounts of nitric acid or mercuric nitrate catalyze the formation of the tetraene DHE as well as a side product, i.e., mercurated triene, and so such impurities may be possible contaminants in the original ergosterol (64).

Recently, real-time multiphoton imaging of the naturally-occurring fluorescent sterol DHE in the plasma membrane of living cells (70, 82, 203, 204) was used to examine sterol distribution directly with high optical sectioning capability (206–209). While subcellular distribution (plasma membrane, lysosome, lipid droplet, etc) of DHE was highly dependent on the method of delivery (microcrystals, LUV, MβCD), DHE at the plasma membrane was found to be distributed non-randomly into sterol-rich and sterol-poor regions in plasma membranes of living cells with the size range of sterol-rich clustering domains estimated to be from 200 nm (limit of optical microscopy) to 565 nm (70, 71, 82, 203). Due to the diffraction limit of resolution, there is uncertainty about whether these regions, as visible under laser scanning microscopy, are contiguous lateral domains (70, 71, 203) or represent artificial enhancement of the intensity as a result of microvillar/filipodia extensions and/or folding (210, 211) of the plasma membrane.

Video imaging studies involving HepG2 cells had resolved regions in the size range of 2000–3000 nm and had shown evidence of sterol-rich and sterol-poor distribution in the plasma membrane as well as canalicular microvilli of the polarized cells (210). It was concluded, however, that these were not regions of enrichment but represented microvilli (210). This same conclusion was reached following subsequent video imaging studies involving methyl-β-cyclodextrin complexes using a pulse chase method and the cell lines Hep G2, J774, and TRVb1 (211) where artifactual enhancement of the fluorescent emission was the result of rough surface topology and cell protrusions (211). However, other studies using filipin, which labels all cholesterol in the plasma membrane without differentiating between sterol-rich or sterol-poor, did not reveal any enhancement in intensity as a result of plasma membrane ruffling or tubule sizing issues (212, 213).

Differences in results obtained by video imaging may be due to several factors: strong photobleaching of the fluorescent sterol as a result of video imaging techniques, saturation of all the plasma membrane with fluorescent sterol for video imaging, and video imaging’s weak resolution in the Z-axis facilitating the need for deconvolution. In contrast, the MPLSM imaging studies used a loading methodology wherein only small amounts (non-saturable) of DHE were introduced into the plasma membrane over longer period of time to minimize the initial perturbations within the membrane (70, 71, 203). With multi-photon excitation used in MPLSM, less extraneous photobleaching of the fluorescent sterol occurred with fluorescence emission from only the excitation volume at a radial resolution on the order of 300 nm while shorter dwell times minimized photodamage. The combination of probing sterol organization and distribution with low amounts of fluorescent sterol together with the resolution enhancement resulting from low yield in the 3-photon excitation volume element, cross-interference from the rough surface topology was reduced.

Available data from many experimental studies of the plasma membrane in general suggest cholesterol is the driving force for microdomain formation and cholesterol is not uniformly or homogeneously distributed but that the plasma membrane of living cells consists of areas of cholesterol segregation (regions that are cholesterol-rich and cholesterol-poor) (56, 214). Biochemical studies also support this concept and demonstrate that purified cholesterol-rich microdomains isolated from cultured cell (L-cell, MDCK, primary hepatocytes) represent nearly one-third of the plasma membrane, are rich in cholesterol as well as saturated/monounsaturated fatty acylated phospholipids, and are comprised of physically distinct, liquid-ordered membrane phases intermediate between fluid liquid-crystalline and rigid gel phases (26, 82, 102, 136, 172, 182, 215–217). Studies with purified microdomains from L-cell and MDCK plasma membranes showed that both exogenous (e.g. HDL) and endogenous (e.g. SCP-2) cholesterol binding proteins preferentially donate or extract cholesterol from cholesterol-rich, but not cholesterol-poor, microdomains (26, 27, 136, 137, 172). The microdomain/lipid raft concept, despite controversy in details, provides a framework for biologists studying localization and function of membrane protein receptors, transporters, and downstream signaling molecules that regulate uptake of cholesterol (86, 87, 218–239), fatty acids (240–242), glucose (243–254), and other activities (216, 255–260).

Abbreviations

DHE

dehydroergosterol

DHE-MβCD

DHE-methyl-β-cyclodextrin

SP

spin-labeled

BODIPY

4,4-difluoro-4-bora-3a,4a-diaza-s-indacene

NBD

N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino

Dansyl

5-dimethylaminonapthalene-1-sulfonyl

CTE

cholestatrienol

DiO

3,3'-dioctadecyloxacarbocyanine perchlorate

ECFP

enhanced cyan fluorescent protein

BHT

butylatedhydroxytoluene

APCI

(atmospheric pressure chemical ionization) mass spectroscopy

HPLC

high performance liquid chromatography

LUV

large unilamellar vesicle

SCP-2

sterol carrier protein-2

L-FABP

liver fatty acid binding protein

MDCK

Madin-Darby canine kidney

EPR

electron spin resonance