Evaluation of a high-content screening fluorescence-based assay analyzing the pharmacological modulation of lipid homeostasis in human macrophages


  • Part of this work was presented at the 10th Leipziger Workshop “Systems Biology and Clinical Cytomics,” April 7–9, 2005, Leipzig, Germany.



For understanding cholesterol and phospholipid efflux pathways there is a need for cellular fluorescence-based high-content screens (HCS) to investigated the cholesterol and phospholipid content in human macrophages.


Making use of fluorescence imaging based on HCS we have developed a tool to evaluate new agents that can act as inducers of cholesterol efflux. The fluorescence assay is based on the different staining patterns of cholesterol-loaded (E-LDL) and deloaded (HDL3) differentiated monocytes by the saturated, fluorescene lipid probe (1,2-dimyristoyl-sn-glycero-3-phospho-ethanolamine)-tetramethyl-rhodamin.


Morphologic examination and statistical evaluation of the staining pattern such as gray value, threshold area, shape factor and the spot size distribution provides evidence for a significant pattern change when cholesterol enriched and cholesterol depleted differentiated monocytes were imaged. © 2006 International Society for Analytical Cytology

ATP binding cassette transporter 1 (ABCA1) gene transcription is regulated by oxysterols such as 27-OH-cholesterol that act as cofactors for the nuclear receptor heterodimer liver X receptor/retinoid X receptor (LXR/RXR). We have intensively investigated cholesterol and phospholipid efflux pathways mediated by ABCA1, which is induced in cholesterol-loaded human macrophages (1). We are currently developing novel fluorescence-based high-content screens (HCS) using fluorescence imaging methods to evaluate small molecules that target LXR or RXR as potential regulators of ABCA1 expression and inducers of cholesterol efflux or alternatively inhibit cholesterol influx. Differentiated monocytes were loaded with enzymatically modified LDL (E-LDL) and subsequently incubated for 16 h in the absence or presence of HDL3, followed by staining either with the saturated, fluorescent lipid probe (1,2-dimyristoyl-sn-glycero-3-phospho-ethanolamine)-tetramethyl-rhodamin (DMPE-TMR) or other raft-marker fluorophores. A recent study indicated a high partition coefficient of DMPE-TMR into membrane domains that are in the liquid-ordered phase and most probably represent the in vivo correlate of Triton-resistant membrane domains (2). To identify targets as markers for apo AI-mediated cholesterol efflux, we have assayed several fluorescent probes that stain cellular structures, which are possibly involved in cholesterol efflux and can be monitored in terms of changes in intracellular quantification and distribution.


A possibility to detect cholesterol-rich lipid membrane microdomains is based on DMPE-TMR (Avanti Polar Lipids, Inc. Alabaster, AL), a phospholipid possessing an unsaturated side chain, and therefore associated with highly lipid-ordered membrane domains such as rafts. The fluorophore is able to specifically intercalate with these domains and also allows the visualization of the phagosome compartment.

We also tested the hydrophobic dye Nile Red (Sigma-Aldrich GmbH Steinheim, Germany) that has been shown to stain lipid droplets in macrophages and adipocytes. However, depending upon the hydrophobicity of the cytoplasmatic environment, the emission maxima can vary over a range of more than 60 nm and therefore precise quantification may be difficult.

In addition, we have investigated different fluorescent conjugates of bacterial toxins in cholesterol-loaded and deloaded macrophages. They are able to bind to galactosyl moieties, which are constituents of plasma membrane microdomains enriched in sphingolipids and cholesterol. Commercially available lipid-raft labeling kits based on the Alexa dye series include green, orange, and red fluorescent probe conjugate of cholera toxin subunit B (CT-B) (Molecular Probes—Invitrogen Labeling & Detection Eugene, OR). This CT-B conjugate binds to the pentasaccharide chain of the plasma membrane ganglioside GM1, which selectively partitions into lipid rafts.


Since information from single cell analysis, as often performed in the laboratory, does not provide results of good statistical significance, sufficient numbers of cells have to be analyzed in order to obtain a reliable averaged result, e.g., in metabolic studies of cells. Moreover, misinterpretation of results due to a potential heterogeneity of the cell subpopulation has to be avoided. For this purpose we are using the cell-based HCS System Discovery-1 (Universal Imaging, Downingtown, PA), which is a conventional high-end fluorescence inverse microscope with sixfold objective-revolver consisting of a multiSpec-imager for emission beam splitting and a high end CoolSNAP-HQ monochrome CCD camera (Photometrics Inc., Tucson, AZ). The system is controlled by a high-end MetaMorph imaging system (Universal Imaging, Downingtown, PA) for measurements and data handling, and can perform fast well-scanning using an autofocus option. The assay experiments were performed after the elutriated monocytes had been cultured in 96-well polystyrene plates with macrophage serum-free medium, supplemented with monocyte-colony stimulating factor (MCSF, human 50 ng/ml). Incubation for up to 7 days at 37°C was performed to induce phagocytic differentiation as described previously (3). On the fourth day, E-LDL (40 μg/ml) (non-oxidatively modified, enzymatically degraded LDL) was added to the cells for 48 h to generate foam cells resembling lipid-laden cells of the arteriosclerotic lesion. On day 6, the foam cells were deloaded of cholesterol with HDL3 (100 μg/ml) for 48 h. On days 1, 4, 6, and 7 cells were harvested. Cells were fixed in 2% paraformaldehyde for 15 min at room temperature and incubated first with a 50 mM solution of glycine, followed by an incubation for 30 min at 4°C with the respective fluorescent marker. For quantitative image analysis, fluorescent images from samples were taken using the Discovery1 system. The fluorescent objects were selected by the threshold function and evaluated in nine regions per well by quantification of pixel intensity, object size, and shape factor. Statistical evaluation was performed based on at least five wells per different treated cells.


A typical MetaMorph assay function automated image analysis gave the following results. In unloaded macrophages, areas of less than 1 μm size possessing a weak and homogeneous peripheral fluorescence pattern were observed (image a in Fig. 1A). Since previous data obtained with a similar fluorophore suggested a raft size of about 1 μm (2, 4), these areas may represent single lipid microdomains. Because of the resolution of the microscope of 0.4 μm, it cannot be excluded that the spots represent clusters of closely located domains. In contrast to this, after cholesterol loading, the overall fluorescence intensity increased and larger confluent areas with a diameter of ∼3–8 μm were visible (image b in Fig. 1A), indicating that cholesterol loading promotes the confluence of relatively small rafts into larger domains.

Figure 1.

A: Visualizing DMPE-TMR staining of plasma membrane microdomains in macrophages under E-LDL loading and HDL3-induced deloading (cholesterol efflux). A: Represents unloaded macrophages (a) showing a homogeneous distribution of membrane lipids apparently possessing a weak and homogeneous fluorescence; (b) cholesterol-loaded cells that were subsequently incubated for 48 h with E-LDL showing a raft association and the formation of large membrane near compartment; (c) HDL3 deloading of cholesterol rich macrophages results in disruption of large microdomains. B: Results from image analysis. Calculated pixel gray values, the percentage of the threshold area, inverse shape factor, and the spot size distribution of DMPE-TMR-stained macrophages when treated with MCSF, followed by E-LDL load and HDL3 deload. Error bars in percentage gave the following deviations: gray value ± 12%, threshold area ± 15%, shape factor ± 22%, and for the size distribution 19% of the mean. For all measured cellular parameters it was found that all values increased after E-LDL loading (P < 1.9 × 10−2) as compared with MCSF treatment and decreased after HDL3 deloading (P < 1.3 × 10−5) as compared with the E-LDL loaded specimen. In these experiments, data of n = 10 randomly selected cells were analyzed using two-tailed paired Student's t-test.

In Figure 1B typical graphs show the average pixel gray value, the percentage of the threshold area as well as the inverse shape factor and the spot size distribution.

Cholesterol enrichment may increase the fluorophore association with the raft domains. In contrast, HDL3 incubation of cholesterol-loaded cells reversed this effect and reduced the overall fluorescence intensity, demonstrating that HDL3 dissociates lipid microdomains at the plasma membrane in vivo (image c in Fig. 1A).

Statistical evaluation of the HCS measurements provides evidence for the existence of compositionally distinct membrane microdomains in primary human cells. Apo AI/ABCA1-dependent and HDL3-mediated lipid efflux modify not only the lipid composition but also the raft distribution in macrophages. This effect is accompanied by an increase of ABCA1-dependent cholesterol and choline phospholipid efflux that is activated by Apo AI or HDL3.

To study the impact of compounds that target LXR or RXR in order to activate the ABCA1-promoter and enhance cholesterol efflux, several substances of interest were included into the incubation scheme. Preliminary data of the fluorescence-based quantitative assay for small molecules modulating the ABCA1-mediated cholesterol transport in macrophages based on a high-content screening platform and conventional 96-well plates still have a high coefficient of variation, and it is necessary to emend processing and assay performance to improve statistical significance. However, our data provide evidence that the sensitivity of the assay system is satisfactory for a large-scale screening.