Abstract
The scavenger receptors CLA-1/SR-BI and CD36 interact with native and modified lipoproteins and with some anionic phospholipids. In addition, CD36 binds/transports long-chain free fatty acids. Recent biochemical evidences indicates that the rabbit CLA-1/SR-BI receptor can be detected in enterocytes, and previous studies showed the presence of mRNA for both CLA-1/SR-BI and CD36 in some segments of the intestinal tract. These findings prompted us to study their respective localization and distribution from the human stomach to the colorectal segments, using immunohistochemical methods. Their expression in the colorectal carcinoma-derived cell line Caco-2 was analyzed by Northern blotting. In the human intestinal tract, CLA-1/SR-BI was found in the brush-border membrane of enterocytes from the duodenum to the rectum. However, CD36 was found only in the duodenal and jejunal epithelium, whereas enterocytes from other intestinal segments were not stained. In the duodenum and jejunum, CD36 co-localized with CLA-1/SR-BI in the apical membrane of enterocytes. The gastric epithelium was immunonegative for both glycoproteins. We also found that CLA-1/SR-BI mRNA was expressed in Caco-2 cells and that its expression levels increased concomitantly with their differentiation. In contrast, the CD36 transcript was not found in this colon cell line, in agreement with the absence of this protein in colon epithelium. The specific localization of CLA-1/SR-BI and CD36 along the human gastrointestinal tract and their ability to interact with a large variety of lipids strongly support a physiological role for them in absorption of dietary lipids.
F
The lipoprotein receptor CLA-1, also known as scavenger receptor BI (SR-BI) (Calvo and Vega 1993), is mostly expressed in liver and steroidogenic tissues, where it binds native high-density lipoproteins (HDLs) and mediates selective uptake of HDL cholesteryl ester (Acton et al. 1996). More recently, its capacity to transfer free cholesterol (Stangl et al. 1999) and phospholipids (Urban et al. 2000) into cells has been demonstrated. In rabbit, CLA-1/SR-BI is also expressed in enterocytes, and anti-CLA-1/SR-BI antibodies (Hauser et al. 1998) block the uptake of cholesterol from small unilamelar vesicles to brush-border membrane vesicles and to Caco-2 cells, showing that CLA-1/SR-B1 might mediate intestinal uptake of dietary lipids. Further support for this conclusion arises from data showing that binding of lipoproteins to the brush-border membrane-resident receptor inhibits the uptake of cholesterol (Boffelli et al. 1997a,b; Hauser et al. 1998).
CD36 is a plasma membrane glycoprotein structurally related to CLA-1/SR-BI, which is expressed in platelets (Li et al. 1993), monocyte/macrophages (Talle et al. 1983; Han et al. 1997), capillary endothelial cells (Knowles et al. 1984; Swerlick et al. 1992), erythroblasts (Edelman et al. 1986), mammary and retinal epithelial cells (Greenwalt et al. 1990), and adipocytes (Abumrad et al. 1993). Like CLA-1/SR-BI, CD36 interacts with native and modified lipoproteins and some anionic phospholipids (Rigotti et al. 1995; Calvo et al. 1997, 1998; Gu et al. 1998). Particularly interesting is the fact that CD36, in addition to binding oxidized LDL, is able to take up the lipoprotein (Nozaki et al. 1995). Moreover, CD36 binds/transports long chain fatty acids (Abumrad et al. 1993).
The regiospecificity of SR-BI and CD36 has been characterized along the gastrocolic axis of the murine species. In mouse, SR-BI is expressed in the proximal intestine where cholesterol absorption occurs (Repa et al. 2000). The fatty acid transporter (FAT), the murine analogue of CD36, has been reported to be expressed in the brush border of enterocytes of proximal intestine, where it could promote uptake of dietary long-chain fatty acid (Poirier et al. 1996). However, no information is yet available on the distribution of CD36 in human intestine.
Although Northern blotting experiments have revealed the expression of both CD36 and CLA-1/SR-BI in human intestine (Calvo et al. 1997), neither their respective distribution along the intestinal tract nor their presence in the brush-border membrane of human enterocytes has been previously investigated. In this study we examined the localization and expression pattern of CLA-1/SR-BI and CD36 along the human gastrointestinal tract, as well as their relative mRNA expression levels in Caco-2 cells, a colon carcinoma-derived cell line extensively used as a model to study intestinal lipoprotein synthesis, uptake, and secretion (Levy et al. 1995).
Materials and Methods
Primary Antibodies
The monoclonal antibody (MAb) FA6.152 (Edelman et al. 1986) used to detect CD36 was obtained from Immunotech (Marseille, France). This MAb, originally described as recognizing the thrombospondin receptor, has been broadly used for recognizing CD36 by different methods as recommended by the manufacturers (Huh et al. 1995) and as a specific inhibitor of the binding of oxidized LDL to CD36 (Ohgami et al. 2001). The rabbit polyclonal antibody 1336, reactive with human CLA-1/SR-BI, was raised as previously described (Calvo et al. 1997). The specificity of this antibody has been previously tested by immunocytochemistry, on transfection of CLA/SR-B1 in non-expressing cells, and by immunoprecipitation and Western blotting experiments (Calvo et al. 1997; Hauser et al. 1998).
Tissue Samples
Tissue samples were obtained from patients who had undergone surgical resection of different areas of the digestive tract. Normal tissue found adjacent to the resected pathological tissues was used in all cases, as ascertained by routine hematoxylin/eosin staining, with the ongoing approval of the internal review board. Immediately after resection, samples were frozen in isopentane or fixed with 4% paraformaldehyde in PBS for 6–24 hr at 4C. These fixed samples were embedded in paraffin according to conventional methods (Gonzalez–Santander et al. 1997). Initially, we obtained samples from 35 patients, but 15 samples were discarded because of poor morphology or detection of abnormalities, such as inflammatory infiltrates, epithelial atrophy, or extensive necrotic areas.
Streptavidin–Peroxidase Methods
Frozen and paraffin sections (5 mm thick) were mounted in silanized slides and allowed to dry overnight before immunohistochemical staining. Paraffin was removed with xylene. Sections were hydrated and endogenous peroxidase activity was inhibited by incubation with 3% H2O2 for 10 min and 0.3% H2O2 in methanol for an additional 20 min. Sections were then washed in Tris-buffered saline (TBS) and incubated in 3% normal serum, goat serum for CLA-1/SR-B1 staining, or horse serum for CD36, with 0.05% Triton X-100 in TBS, pH 7.5, at room temperature (RT) for 30 min. Then the sections were incubated for 12 hr at 4C with the primary antibodies, anti-CLA-1 (1:200) and anti-CD36 (4 μg/ml) diluted in TBS. After washing twice in TBS, sections were incubated with the secondary antibodies for 1 hr at RT. The biotinylated secondary antibodies used were goat anti-rabbit IgG (1:200) for CLA-1/SR-B1 and horse anti-mouse IgG (1:200) for CD36 (both from Vector Labs; Burlingame, CA). Sections were washed in TBS and incubated with the streptavidin–peroxidase complex (Zymed Labs; San Francisco, CA) for 30 min and washed in TBS followed by Tris-HCl buffer, pH 7.6. The peroxidase activity was revealed using 3′-diaminobenzidine tetrahydrochloride (DAB) as chromogen (Sigma; St Louis, MO). The reaction product of DAB was intensified with nickel nitrate (10 μl of an 8% solution of nickel nitrate in 1 ml DAB–H2O2 solution) to obtain a dark black color of immunostained antigens (Lobo et al. 2000a,b). This method provides a better signal-to-noise ratio with no evident distortions in the relative intensities of the signals from sites with different levels of expression. Thereafter, the sections were dehydrated in ethanol, mounted in Eukitt (O. Kindler; Freiburg, Germany) without being counterstained, and were observed under a light microscope.
CLA-1/SR-B1 localization in the stomach (
Localization of CLA-1/SR-B1 (
Immunofluorescence Methods
Sections were processed and incubated with the primary antibodies as described above and then incubated with rhodamine (1:150)- or FITC (1:60)-conjugated secondary antibodies (Boehringer Mannheim; Mannhein, Germany) for 60 min in darkness. After washing in TBS, the sections were mounted using Mowiol (Sigma–Aldrich Quimica; Madrid, Spain) and observed in a Zeiss epifluorescence microscope.
Control Experiments
Although the specificity of the antibodies has been previously described (see above) to further assess their specificity in the immunohistochemical procedures performed in this study the following negative controls were systematically performed for both CLA-1/SR-BI and CD36 staining: (a) omitting the primary specific antibodies, (b) using rabbit preimmune serum or normal mouse serum instead of the primary antibodies, and (c) incubating with an inappropriate secondary antibody after the incubation with the primary antibodies at optimal titers.
Northern Blotting Analysis
Total RNA was extracted from Caco-2 cells, collected at Days 0, 3, 6, 9, and 12 after reaching confluence, using the acid guanidine thiocyanate method (Chomczynski and Sacchi 1987). Samples of 20 μg of total RNA were analyzed by Northern blotting. Ethidium bromide staining of 28S and 18S rRNAs was used as loading control. Probes for human CD36 and CLA-1/SR-BI were as described (Calvo et al. 1997). For apoE, a 0.4-kb EcoRI-BamHI fragment derived from the fourth exon was used as a probe (Mehran et al. 1997).
Results
The distribution of CLA-1/SR-BI and CD36 along the gastrocolic and crypt-to-villus axes of the human gastrointestinal mucosa has been investigated. Both of the immunohistochemical methods we used, streptavidin–peroxidase (Figures 1 and 2) and immunofluorescence (Figure 3), led to similar results. The use of the polyclonal antibody anti-CLA-1/SR-BI in formalin-fixed, paraffin-embedded samples and in frozen tissues provided identical results. However, although CD36 was detected in paraformaldehyde-fixed, paraffin-embedded samples, immunolabeling for CD36 on unfixed frozen sections provided the best signal-to-noise ratio. In addition, and as an internal positive control for CD36 immunostaining, this protein was clearly detected in microvascular endothelial cells of the gastrointestinal tract (Figure 3) whereas the endothelium of large vessels was unstained, as has been previously described (Knowles et al. 1984; Han et al. 1997). No staining was found in any of the negative controls performed for both CLA-1/SR-BI and CD36 proteins (data not shown).
In stomach, neither CLA-1/SR-BI nor CD36 was detected in epithelial, parietal, mucous and endocrine cells (Figure 1A). In contrast, both CLA-1/SR-BI and CD36 were found in the brush-border membrane of enterocytes at specific segments of the intestinal tract, but not in the nucleus, cytoplasm, or basolateral plasma membrane of enterocytes. Intestinal goblet cells, Paneth cells, and enteroendocrine cells were immunonegative for both glycoproteins. Interestingly, regional differences in the distribution of these glycoproteins along the gastrointestinal tract were noted. Thus, while the duodenal epithelium showed positive staining for both glycoproteins (Figures 1B-1D, 2A–2E, and 3A–3C), differences in the immunoreactivity along the crypt-to-villus axis were observed. CLA-1/SR-BI was expressed in the brush-border membrane of enterocytes all along the epithelium, from the crypt to the villous tip. Surface enterocytes were the most intensely stained cells for CLA-1/SR-BI. However, CD36 labeling was restricted to the brush border of the enterocytes located in the upper two thirds of the villus and it was absent from crypt cells (Figure 3A–3C). On the other hand, CLA-1/SR-BI was also detected on the luminal surface of Brunner's gland cells, where no immunoreactivity for CD36 was observed (Figure 3D and 3E). Moreover, a similar distribution pattern for CLA-1/SR-BI and CD36 in the jejunal mucosa was observed (Figures 1E, 2F–2H, and 3F). A negative immunoreaction for CD36 was observed in the epithelium of the ileum, colon, and rectum (Figure 3G and 3H), where the immunoreactivity for CLA-1/SR-BI glycoprotein was intense (Figure 1F–1K). In these intestinal segments, CLA-1/SR-BI was found in the apical membrane of enterocytes all along the intestinal epithelium. As observed in small intestine, the immature undifferentiated cells from the base of the crypts showed a positive immunoreaction for CLA-1/SR-BI, whereas those areas of the colon and rectum epithelium mainly composed of goblet cells were unstained (Figure 1K).
Sections of the duodenum from the villous tip (
We also examined the expression of CD36 and CLA-1/SR-BI mRNA in human Caco-2 cells, an enterocytic cell line derived from a colon adenocarcinoma that expresses CLA-1/SR-BI. This cell line differentiates on reaching confluence, forming a polarized monolayer with an apical brush-border membrane, similar to that of the enterocyte. Expression of apoE mRNA was used as a positive control for differentiation (Wilson and Rudel 1994). As shown in Figure 4, mRNA expression of CLA-1/SR-BI, like that of apoE, increased with the differentiation of the cells. However, no CD36 transcript was detected in this cell line, even after long exposures of the autoradiographs. Integrity of the CD36 probe was checked with mRNA obtained from the CD36-positive monocytic cell line THP-1 (data not shown). Absence of expression of CD36 was additionally confirmed by the lack of staining of Caco-2 cells with the anti-CD36 antibody FA6.152 (data not shown). These data indicate that the expression pattern of CD36 and CLA-1/SR-BI in the colon-derived Caco-2 cell line was consistent with that observed for both glycoproteins in colon enterocytes.
Discussion
In this study we have demonstrated the localization of CLA-1/SR-BI and CD36 receptors in the brush-border membrane of human enterocytes. Both proteins are co-localized in the enterocytes that cover the surface of the duodenum and jejunum villi. Moreover, and unlike CD36, the expression of CLA-1/SR-BI is not restricted to the duodenal and jejunal villous cells but is also present in the intestinal crypts, Brunner's glands, and in the rest of the intestinal tract, ileum, colon and rectum, where the localization of this glycoprotein is demonstrated for the first time. Accordingly, Caco-2 cells, a colon-derived cell line, expressed mRNA for CLA-1/SR-BI but not CD36.
Northern blotting analysis of apo-E, CLA-1/SR-B1, and CD36 mRNA expression along the differentiation of Caco-2 cells. Equal amounts (20 μg) of total RNA, based on the intensity of the ethidium bromide staining of 18S and 28S rRNAs, were run in an agarose gel, blotted onto nitrocellulose, and hybridized with human apo-E, CLA-1/SR-B1, and CD36 probes. The figure shows a representative experiment of the two that were performed with similar results, in which an increase in the mRNA for both apo-E and CLA-1/SR-BI from 0 to 12 days after plating was observed. The estimation of the optical density for each line compared with the controls showed increases of 79%, 85%, 91%, and 87% for apo-E, and 30%, 51.6%, 61%, and 49% for CLA-1/SR-BI, at 3, 6, 9, and 12 days, respectively.
The broad ligand-binding specificity of both CD36 and CLA-1/SR-BI (Acton et al. 1996; Calvo et al. 1997, 1998; Gu et al. 1998), together with their co-localization in the brush-order membrane of the human enterocytes that cover the surface of the duodenum and jejunal villi, the intestinal regions most actively implicated in cholesterol absorption (Borgström 1960; Arnesjö et al. 1969; reviewed in Wilson and Rudel 1994), strongly suggest the involvement of these receptors in the uptake of intestinal lipids. These features provide a framework that encourages the examination, at a molecular level, of the biochemical and genetic evidence that supports the existence of protein-facilitated mechanisms for intestinal cholesterol absorption by the enterocyte. Among them are (a) the ability of small molecules (pamaqueside and SCH 48461) to act as potent inhibitors of cholesterol absorption at doses below those required to bind cholesterol in a 1:1 molar ratio (Salisbury et al. 1995; Morehouse et al. 1999), (b) the poor intestinal absorption of plant sterol molecules in spite of their structural similarity to cholesterol (Miettinen and Gylling 1999), and (c) the observed mouse strain-specific differences and interindividual variation in the efficiency of cholesterol absorption (Grundy 1983; Kirk et al. 1995; Carter et al. 1997).
How the intestinal absorption of lipids other than cholesterol takes place is still a matter of debate. It has been shown that the uptake of cholesterol esters, triacylglycerols, fatty acids, and phospholipids by small intestine brush-border membrane vesicles is completely inhibited by proteolytic treatment (Hauser et al. 1998). Therefore, both CD-36 and CLA-1/SR-BI multiligand receptors could modulate not only the absorption of cholesterol but also the uptake of different lipids by a protein-mediated mechanism similarly to what has been postulated for SR-BI and cholesterol (Hauser et al. 1998).
From our experimental observations it is also difficult to ascribe specific and/or differential roles for CD36 and CLA-1/SR-BI. Simultaneous expression of CD36 and CLA-1/SR-BI in the proximal intestine might serve either to guarantee, by acting redundantly, a vital function and/or to allow a fine-tuned modulation of lipid absorption. Interestingly, it has recently been reported that in SR-BI deficient mice, intestinal cholesterol absorption is normal, suggesting the necessity of other molecules able to compensate for the loss of intestinal absorptive activity (Mardones et al. 2001). Selective expression of CD36 in the proximal intestine is consistent with the observed regional expression of intestinal genes involved in the absorption of nutrients (lipids and non-lipids) in the diet (Shaw–Smith and Walters 1997). In addition, the specific ability of CLA-1/SR-BI to transfer free cholesterol (Stangl et al. 1999) and phospholipids (Urban et al. 2000) into cells and that of CD36 to take up oxidized LDL (Nozaki et al. 1995) and long-chain fatty acids (Abumrad et al. 1993), together with the regiospecificity of cholesterol absorption, suggests that both proteins could be involved in and/or responsible for the absorption of cholesterol. However, although it is clear that CLA-1/SR-BI and CD36 are structurally related, the differences shown by both proteins in selective uptake of HDL cholesteryl ester (CE) (Gu et al. 1998; Stangl et al. 1998; Connelly et al. 1999) indicate that the two receptors differ in some functions, e.g., CLA-1/SR-BI but not CD36 mediates selective uptake of cholesterol ester. In this regard, CD36 could mediate only the binding of HDL, while CLA-1/SR-BI could specifically mediate CE transfer. Finally, the selective expression of CLA-1/SR-BI in the intestinal crypts of the small intestine and in large intestine, where the absorption of lipids is quantitatively much less important, suggests that CLA-1/SR-BI should have other function(s) in addition to lipid transport or serving as a safeguard molecule for those lipids not taken up in the proximal intestine. In this regard, putative interindividual variation of CLA-1/SR-BI expression in colon could be related to the development of colon diseases caused by unabsorbed lipid metabolites (Vonk et al. 1997).
The results of the present study, together with previous evidence, demonstrate the existence of a family of lipid-binding proteins localized in the brush-border membrane of the enterocytes that could mediate the absorption of lipids. Further investigations are necessary to determine the functions of each protein in the different intestinal segments where they are expressed. These studies could provide new therapeutic strategies that, acting on these transporters, would permit regulation of the absorption of lipids, a physiological function directly related to hypercholesterolemia and obesity, two major risk factors for atherosclerosis.
Footnotes
Acknowledgements
Supported by grants from Comisión Interministerial de Ciencia y Tecnología (PM98/0063 to R.B.), Comunidad Autónoma de Madrid (08.2/0004/1998 to R.B.), and Fondo de Investigación Sanitaria (97/0389 to A.M.H. and 99/1046 to M.A.V.). E. Teixeiro is a fellow of the CAM.