Differences in the uptake of modified low density lipoproteins by tissue cultured endothelial cells

Low density lipoprotein (LDL) is the major carrier of cholesterol and is absorbed by receptor-mediated endocytosis (Brown & Goldstein, 1979). In addition there is a separate ‘scavenger’ pathway of LDL uptake in macrophages and endothelial cells (Goldstein, Ho, Basu & Brown, 1979; Stein & Stein, 1980). The scavenger pathway recognizes only anionic LDL produced by modification of the side-chain amino groups of arginine and lysine residues in the apo-protein. Anionic LDL has been synthesized using malondialdehyde (Haberland, Fogelman & Edwards, 1982), succinic anhydride (Haberland, Oleh & Fogelman, 1984) and, most commonly, acetic anhydride producing acetyl-LDL (Ac-LDL; Basu, Goldstein, Anderson & Brown, 1976).

Voyta, Via, Butterfield & Zetter (1984) have used this cellular specificity of the native and scavenger pathways to isolate pure endothelial cells from bovine aorta and adrenal capillary in vitro. The basis of their method is the incorporation of a fluorescent, lipophillic probe 1,1′-dioctadecyl-1-3,3,3′,3′-tetramethyl-indo-carbo-cyanine perchlorate (Dil; Pitas, Innerarity, Weinstein & Mahly, 1981) into the lipid core of the lipoprotein. On degradation of the Dil-Ac-LDL by the lysosomal enzymes the Dil accumulated in the intracellular membranes rendering them fluorescent. Voyta et al. (1984) proposed that Ac-LDL uptake may be a general endothelial cell marker.

However, in vivo, Pitas et al. (1981) found differential uptake of acetylated and acetoacetylated LDL by endothelial cells of different organs (Pitas, Boyles, Mahley 8z Bissell, 1985). For instance, endothelial cells of the sinusoids of liver, bone marrow and adrenal were labelled but those of the heart, testes, kidney, brain, adipose tissue and duodenum were not. Furthermore, Pitas et al. (1985) attempted to quantify the uptake of Ac-LDL in vivo and found less-intense staining with aortic endothelial cells than was observed by Voyta et al. (1984) in vitro.

There thus appears some doubt as to whether there is uptake of Ac-LDL by all types of endothelial cells. We have attempted to clarify the position by investigating Ac-LDL uptake in vitro by a variety of endothelial cells. We also report on the effect of cell density and phase of cell cycle on Ac-LDL uptake.

Human LDL (d = 1·019—l·063gml⁻¹) and lipoprotein-deficient serum (LPDS, d> l·21g ml⁻¹) were isolated from the pooled sera of six healthy, fasting individuals by sequential ultra-centrifugation (Havel, Eden & Bragden, 1955). Acetylated LDL was prepared as previously described (Basu et al. 1976) and repeatedly dialysed against LDL buffer (0·5 M-NaCl, 0·1% (wN) EDTA, pH 7·4) before use. The Ac-LDL was characterized by electrophoresis in Tris-tricine buffer (0·08M:0·024M, respectively, pH8·4) in 0·5% agarose maintained on gel bond. Native LDL was also run as a standard. Protein was visualized with Coomasie Blue and lipid with methanolic fat red, 7B (Gurr).

LDL and Ac-LDL were labelled with the fluorescent probe (Dil; Molecular Probes, Junction City, N.Y.) as described by Pitas et al. (1981). Briefly, to each sample of lipoprotein (1 mg in 2 ml of LPDS) was added 50μl of Dil (3mgml⁻¹ in dimethylsulphoxide (DMSO)). The solution was gently mixed, filter-sterilized using 0·45 μm filters and incubated at 37°C for 18 h. The density of the mixture was then raised to 1·063gml⁻¹ with solid KBr and the fluorescent lipoprotein was isolated by centrifugation (20 h; 60 000 g).

The Dil concentration of the lipoprotein was determined by extraction with methanol and determining its fluorescence against standard concentrations of Dil using a Perkin-Elmer model MPF-2A fluorescent spectrophotometer (excitation 520 nm; emission 570 nm). The total cholesterol and triglyceride contents were obtained by enzymic methods. LDL protein concentration was determined by rocket immuno-electrophoresis against an LDL antibody. All preparations of LDL were expressed in terms of LDL protein and batches of Dil labelled LDL and Ac-LDL had molar ratios of Dil:protein of from 250:1 to 400:1.

Capillary endothelial cells from cow brain white matter (BCEC) and bovine aorta (BAEC) were obtained and cultured as previously described (Keeganet al. 1982; Kumar et al. 1984). Using an indirect immunofluorescent technique (LF.) the cells were stained with anti-bovine factor VIII serum (a gift from Dr E. Kirkby) and were found to be strongly positive (Kumar et al. 1984; see also Fig. 2). MY7 monoclonal antibody (Coulter Corporation, Hialeah, Florida, U.S.A.) was also used to examine BCEC by an indirect LF. technique. MY7, which specifically reacts with monocytes and granulocytes failed to stain BCEC.

BAEC were used for labelling experiments within six to eight passages of establishing the primary culture and were judged to be confluent when a continuous monolayer of cobblestone appearance was present. Subconfluent cells were those occupying 50% of the available area of the culture flasks.

These cells were obtained from bovine adrenal cortex using a slight modification of the method of Folkman, Haudenschild & Zetter (1979). The tissue was minced and treated with 0·2% collagenase (Boehringer), in 0·5% bovine serum albumin in phosphate-buff ered saline (PB S; 0 · 1 M, pH 7·0), for 45 min at room temperature. The tissue was filtered (110μm Nylex filter), centrifuged (1000 g, 10 min), washed with DMEM containing 10% foetal calf serum (FCS) and plated onto Falcon dishes coated with 1% gelatine. The cells were subsequently cultured in DMEM containing 10% FCS, 100 μg ml⁻¹ of heparin and endothelial cell growth factor (2 μg ml⁻¹; kindly provided by Dr A. Schreiber) and were used as primary cultures.

Mouse peritoneal macrophages were prepared and cultured according to Adams (1979). Five days after intraperitoneal injection of 10 mg of thioglycollate broth in saline, the animals were killed and the peritoneum was washed out with cold complete medium 199. The effluent was plated in Leighton tubes. More than 95% of cells were macrophages as shown by non-specific esterase staining.

Cells were labelled with Dil containing lipoproteins as fallows. The medium was removed from the cells and replaced with fresh medium containing either Dil-LDL or Dil-Ac-LDL at a concentration of 10μgml⁻¹. The cells were incubated for 16h at 37°C and then trypsinized, washed and harvested. The single-cell suspension was stored at 4°C before cell sorting.

Hoechst 2.2342 (Calbiochem-Behring) is a quantitative label for DNA enabling cells to be sorted into G ₂M or G₁S, according to their DNA content. Cells were suspended in PBS and Hoechst 22342 (30 μg ml⁻¹ in DMSO) added. The cells were incubated for 1 h, washed and resuspended in PBS for cell sorting. Cells were labelled with Hoechst immediately before sorting was carried out.

This was performed on a Becton-Dickinson FACS IV cell sorter with single-cell suspensions. Cells and collection tubes were chilled in ice before and during sorting. The sampling gates were set using stained and unstained cells and the scatter gates were set to remove clumps of cells. Cells were sorted and collected under sterile conditions into PBS, and then transferred to tissue culture flasks and grown as described above.

Cells were initially sorted into G ₂M and G₁S populations using the fluorescence of Hoechst 22342 (excitation 357 run; emission 418 run and above) and then analysed with respect to the fluorescence of Dil using an argon laser (excitation 514nm; emission above 550nm). In all cases the same number of cells (30 000) was analysed.

The specificity of uptake of LDL and Ac-LDL by BAEC and BCEC was qualitatively investigated by incubating the cells with Dil-labelled lipoproteins. BAEC produced a marked fluorescence with Dil-Ac-LDL (Fig. 1) and negligible fluorescence with Dil-LDL. The fluorescence was granular and intracellular. In contrast, BCEC from brain failed to produce staining with either lipoprotein. To determine whether this failure to endocytose Ac-LDL was a general property of microvessel-derived endothelial cells or an organ-specific property of capillary endothelium, the uptake of Ac-LDL by bovine adrenal capillary cells in vitro was examined. Strong fluorescence was observed in these cells (Table 1). The endothelial origin of all three cell types was confirmed by the presence of factor VIII antigen. We eliminated thy possibility of contamination by monocytes using MY7 monoclonal antibody and phase-contrast microscopy. Almost 100% of the examined cells contained factor VIII antigen (Fig. 2). A very intense staining of macrophages was obtained with Dil-Ac-LDL and only faint fluorescence with Dil-LDL (Table 1).

BAEC were incubated with Dil-LDL and Dil-Ac-LDL when subconfluent and confluent. The cells were subsequently analysed by flow cytometry and the relative fluorescence of the same number of cells (30000 in each case) was determined. The results are listed in Table 2 as the means of six determinations with standard deviations. Cell density had a marked effect on LDL uptake (Table 2), and subconfluent cells accumulated approximately 34% more Ac-LDL compared to confluent cell lines.

Cells were sorted into G ₂M (20—27·5% of the total) and G\S (72·79%) using Hoechst 22342. Cells were then analysed to determine Dil uptake (Table 3). There was no difference in the uptake of Dil-LDL. In contrast, cells in G₂M contained significantly higher amounts of Dil-Ac-LDL than those in G₂S.

It has been shown that endothelial cells from microvessel sources, including spleen, ovarian interstitia and adrenal have receptors for Ac-LDL (Voyta et al. 1984; Pitas et al. 1985). We were able to confirm the uptake of Ac-LDL by both BAEC and BAdEC. In contrast, BCEC failed to stain with either LDL or Ac-LDL. These results agree with those of Pitas et al. (1985), who found that brain microvessel endothelial cells failed to take up Ac-LDL in vivo. It is known that cells from brain capillary differ from those from other sources in a number of ways, including the possession of glutamyl-transpeptidase (Orlowski, Sessa & Green, 1974), and the 5F2 antigen (White et al. 1983) and the absence of xanthine oxidase (Jarasch et al. 1981) and the PAS IV antigen (Greenwait & Mather, 1985). The failure of brain capillary endothelial cells to absorb Ac-LDL adds a further means of distinction and indicates that Ac-LDL uptake may not be a universal marker for endothelial cells.

Little is known of the effect of contact inhibition on the Ac-LDL receptor in endothelial cells although there have been a number of studies of the LDL receptor in fibroblasts and smooth muscle cells. Confluent BAEC were found to bind two-to three-fold less LDL than subconfluent and internalized hardly any (Vlodovasky, Fielding, Fielding & Gospadarowicz, 1978). BAECs were shown to have fewer LDL receptors when contact inhibited (Coetzee, Stein & Stein, 1979). Our results (Table 2) also confirm that when the same number of aortic ceils were analysed subconfluent cells had a fluorescence with LDL of 9–2 compared with 3–41 for confluent cells(a ratio of 2–7:1; P< 0-005). This ratio is in the same range as that found by Vlodovasky et al. (1978).

However, in comparison, our results for Ac-LDL showed a much less marked inhibition of uptake by confluence. BAEC had a fluorescence of 136·5 when sub-confluent and 101·7 when confluent (a ratio of 1·34:1; P<0·05). Stein & Stein (1980) used [¹²⁵I] Ac-LDL to investigate the binding of Ac-LDL to BAEC. They found a ratio for subconfluent:confluent of 0·66:1. They did, however, use lipo-protein-deficient serum to grow their cells and this may not have led to complete down-regulation of receptors. It is probable that the conditions we used in our experiments were nearer to those experienced by contact-inhibited cells in vivo.

As regards the effect of cell cycle, cells in either G\S or showed no difference in LDL uptake (Table 3). Comparison of the uptake of Ac-LDL by cells in different phases of the cell cycle showed that cells in G I internalized 45% more Ac-LDL than those in G\S (PC0-025).

Our results raise the possibility of using Ac-LDL as a drug-targeting agent against endothelial cells of solid tumours. Unlike all normal adult endothelia, the tumour vasculature undergoes rapid cell division, with labelling indices up to 1000 times greater than that of other vessels (Denekamp, 1984). Encapsulation of cytotoxic drugs into Ac-LDL, as has already been achieved for LDL (Firestone et al. 1984), would result in their differential uptake by certain endothelia. Use of phase-specific lipid-soluble agents, such as vinca derivatives, would confine their effect to dividing endothelium. The modest bias toward uptake by dividing endothelial cells shown here could further increase this selectivity.

This work was supported by the Wellcome Trust and D.C.W. is in receipt of a grant from the MRC.