Unexpected Consequences of Deletion of the First Two Repeats of the Ligand-binding Domain from the Low Density Lipoprotein Receptor EVIDENCE FROM A HUMAN MUTATION*

Heterozygosity for a 5-kilobase (kb) deletion of the first two ligand-binding repeats (exons 2 and 3) of the low density lipoprotein (LDL) receptor (R) gene (LDL-R (cid:68) 5kb) confers familial hypercholesterolemia (FH). The FH phenotype is unexpected based on previous site- directed mutagenesis showing that deletion of exons 2 and 3 resulted in little or no defect in LDL-R activity. In the present study, we took unique advantage of the ability to distinguish the LDL-R (cid:68) 5kb from the normal receptor on the basis of size, in order to resolve this ap- parent discrepancy. Fibroblasts from heterozygotes for the LDL-R (cid:68) 5kb displayed 50% of normal capacity to bind LDL and (cid:98) -VLDL, apparently due to lower receptor number. Cellular mRNA for the (cid:68) 5kb allele was at least as abundant as that for the normal allele. Immunoblot- ting and cell binding assays with anti-LDL-R antibody IgG-4A4 demonstrated normal synthesis and transport of the (cid:68) 5kb receptor. Ligand blotting demonstrated that the (cid:68) 5kb receptor displayed minimal or no ability to bind LDL or (cid:98) -VLDL. Thus, in contrast to transfected cell lines, in human fibroblasts, the first two cysteine- rich repeats of the LDL-R

The low density lipoprotein (LDL) 1 receptor (R) binds and catabolizes apolipoprotein E-containing chylomicron and VLDL remnants and LDL. In the liver, LDL-R functions to remove these lipoproteins from plasma for eventual excretion of the cholesterol into the bile. In peripheral cells, it functions to provide the cell with cholesterol needed for membrane synthesis. The LDL-R contains five major structural domains: a seven-repeat, cysteine-rich ligand binding domain encoded by exons 2-6, an epidermal growth factor-precursor homology domain (exons 7-14), a glycosylation domain (exon 15), a mem-brane-spanning domain (exon 16), and a cytoplasmic tail (exon 17) (1). The LDL-R is one of the few proteins for which knowledge of the structure-function relationship has been generated both from site-directed mutagenesis and from numerous naturally occurring human mutations.
Mutations in the LDL-R gene resulting in a dysfunctional receptor cause a codominantly inherited disorder of plasma cholesterol catabolism known as familial hypercholesterolemia (FH). Human LDL-R mutations have been assigned to five classes of defects based on their phenotypic effects on the receptor protein (1). We have previously described a deletion of approximately 5 kb, which removes exons 2 and 3 of the LDL-R gene (LDL-R ⌬5kb) (2). In site-directed mutagenesis experiments, deletion of the first repeat (exon 2) has no effect on the binding or internalization of LDL or ␤-VLDL or recycling of receptors in transfected mammalian cells (3). Simultaneous deletion of exons 2 and 3 has resulted in a receptor which binds LDL 70% as well as the normal receptor and which binds ␤-VLDL equally as well (4). These results have led to the suggestion that the first two repeats of the LDL-R ligandbinding domain are not necessary for LDL-R function.
Some studies have shown that the clinical phenotype resulting from LDL-R mutations correlates with biochemical phenotype or class (5,6). As such, one would expect that heterozygosity for the LDL-R ⌬5kb would result in relatively mild or no expression of familial hypercholesterolemia (FH). However, taking advantage of genetic founder effects among French Canadians, we have observed that plasma total and LDL cholesterol levels among 8 probands for this deletion are indistinguishable from those in heterozygotes for a null LDL-R allele. In the context of a clinical genetic study of a kindred with the 5-kb LDL-R gene deletion (7), we noted that heterozygote (HTZ) fibroblasts displayed consistently 50 -60% the maximal receptor activity of normal cells. This was again unexpected based on the apparent activity of the LDL-R lacking exons 2 and 3 in transfected cells.
Normally, the biochemical consequences of LDL-R mutations in cells from carriers are difficult to study in the absence of a homozygote. Unlike the case for the vast majority of described mutations of the LDL-R, the LDL-R ⌬5kb deleted protein is distinguishable from the normal receptor on the basis of size. This situation offers a unique opportunity to examine and compare the mRNA, protein, and ligand binding to the mutant receptor with those of the normal receptor in vivo within the same cell. The present study provides evidence for important differences in the consequences of deletion of exons 2 and 3 from the LDL-R gene as assessed by site-directed mutagenesis and by analysis of heterozygous fibroblasts.

MATERIALS AND METHODS
Characterization of the FH Phenotype-Plasma total and LDL cholesterol were measured as described (7). The presence of the French-Canadian LDL-R Ͼ10and 5-kb gene deletions was determined by Southern blotting as described (2).
Cell Surface Binding of Lipoproteins and Antibodies-Human skin fibroblasts were obtained from normal subjects and from FH patients heterozygous for the ⌬5kb deletion. Receptor-negative fibroblasts were obtained from an FH patient homozygous for the Ͼ10-kb deletion (8). Lipoprotein-deficient serum (LPDS) and LDL were isolated from human plasma (9). ␤-VLDL were isolated from blood of cholesterol-fed rabbits (10). Lipoproteins were iodinated with carrier-free 125 I-sodium (Amersham, Oakville, ON) with the IODOGEN (Pierce) method (11). Monoclonal anti-LDL-R antibody IgG-4A4 was isolated from hybridoma cells (American Type Culture Collection, Rockville, MD) as described (12), purified with MAC ® Protein A Capsules (Amicon) as recommended by the manufacturer, and iodinated with IODOBEADS ® (13) (Pierce). Following a 48-h incubation of the cells in 10% LPDS, cell surface binding at 4°C of 125 I-LDL, 125 I-␤-VLDL, and 125 I-IgG-4A4 were performed as described (9). Estimates of receptor number and affinity were calculated by the method of Scatchard.
LDL-R Immunoblotting-Cell protein extracts were prepared generally as described (14). Cells were washed with phosphate-buffered saline containing 1.5 mM phenylmethylsulfonyl fluoride. After centrifugation at 10,000 ϫ g for 30 s, cells were lysed in 180 l of 50 mM Tris-maleate, pH 6.5, 2 mM CaCl 2 , 1% Triton X-100, and 1.5 mM phenylmethylsulfonyl fluoride for 20 min. Cellular debris was pelleted by centrifugation at 10,000 ϫ g for 5 min, and the lysates were stored at Ϫ70°C. Thirty to fifty g of cell lysate was subjected to SDS-6% polyacrylamide gel electrophoresis on a Minigel apparatus (Bio-Rad) at 3.5 mA/gel at 4°C. No reducing agent or heat was used. In some experiments, 40 g of protein extracts were treated for 1 h at room temperature with 0.016 unit of neuraminidase (Bio-Rad, Mississauga, Ontario) in 50 mM sodium citrate, pH 4.5. Proteins were transferred to nitrocellulose membranes (Amersham, Oakville, Ontario) in 25 mM Tris, 192 mM glycine, 20% methanol at 115 V for 1 h with refrigeration (15). LDL-R protein was detected as follows. Nitrocellulose membranes were incubated in buffer A (Tris-buffered saline, pH 7.6, 2 mM CaCl 2 , 0.1% Tween 20) containing 5% dried non-fat milk for 1 h at room temperature. Membranes were incubated overnight at 4°C in buffer A with monoclonal anti-LDL-R antibody IgG-C7 (Amersham) (1 g/ml) or IgG-4A4 (25 ng/ml), washed in buffer A, 1 ϫ 15 min, 3 ϫ 5 min, incubated 1 h in buffer A with anti-mouse horseradish peroxidase-labeled antibody (Amersham) (1/5000), and washed 1 ϫ 15 min and 3 ϫ 5 min with buffer A. Detection of the immunoreactive bands was performed with the ECL ® chemiluminescence system (Amersham) as recommended by the manufacturer. Intensity of bands corresponding to LDL-R were quantified after densitometry of the autoradiograph with the IS-1000 Digital Imaging System version 1.97 (Alpha Innotech Corp.). Each blot was exposed to autoradiographic film for varying lengths of time (30 s to 5 min), and determination of relative band intensities was accomplished by averaging scans from 3 exposures from which the signal increased linearly with exposure time.
Ligand Blotting of LDL-R-Fibroblast culture, cell protein extract preparation, electrophoresis, and blotting were performed as for LDL-R immunoblotting. Ligand blotting was performed essentially as described (16). Membranes were incubated for 1 h at 37°C in buffer B (10 mM Tris-HCl, 150 mM NaCl, 2 mM CaCl 2 , pH 8.0) containing 50 mg/ml bovine serum albumin (Sigma) followed by an incubation for 2 h at room temperature in buffer B with 125 I-LDL or 125 I-␤-VLDL. Blots were washed in buffer B for 15 min, followed by 3 ϫ 30 min washes. Membranes were subjected to autoradiography.
LDL-R mRNA Quantification-Fibroblast RNA was isolated with the RNAzol (Cinna Biotecx, Friendswood, TX) method (17). Oligonucleotides were synthesized by the solid phase triester method on a Pharmacia LKB Gene Assembler Plus DNA synthesizer. The ratio of fibroblast and lymphocyte LDL-R mRNA to that for ␤-actin was quantified by RT-PCR and fluorescence DNA detection as described (7). In LDL-R ⌬5kb HTZ cells, the relative expression of the deleted versus the normal LDL-R allele was determined by amplification of the cDNA with fluorescent primers 5Ј-cgccgcggcggggactgcag-3Ј and 5Ј-agttttcctcgtcagtttgtc-3Ј, yielding polymerase chain reaction fragments of 638 and 441 nucleotides for the normal and deleted allele, respectively. The latter oligonucleotide was labeled with Tamra ® fluorescent dye (Applied Biosystems, Foster City, CA). Aliquots of the reactions were subjected to 6% polyacrylamide gel electrophoresis and fluorescence detection and quantification on a model 373A Automated DNA Sequencer ® (Applied Biosystems). The HincII polymorphism in exon 12 of the LDL-R gene (18) was detected in cDNA as follows. cDNA from fibroblasts or lymphocytes was amplified with 100 ng each of primers 5Ј-ctcagtggccgcctctactgggtag-3Ј and 5Ј-ctgtgaggcagctcctcatgtccctg-3Ј in the presence of 10 M fluorescein-12-labeled dUTP (Boehringer Mannheim, Laval, Quebec). Aliquots of the reactions were digested with HincII. Fluorescence peak areas corresponding to DNA fragments of molecular weight 140 and 204 base pairs, corresponding to the presence or absence of the HincII site, were quantified by integration with 672 GeneScanner ® software (Applied Biosystems).

Phenotypic Expression of the LDL-R ⌬5kb in Vivo and in
Vitro-Plasma total and LDL cholesterol concentrations in 8 probands heterozygous for the LDL-R ⌬5kb were not significantly different from those in heterozygotes for a deletion in the LDL-R gene which results in a null allele (8) (not shown).
Binding experiments with 125 I-LDL at 4°C and Scatchard analysis (Fig. 1A) revealed that the apparent defect in LDL-R activity in LDL-R ⌬5kb fibroblasts was due to an apparently lower number of binding sites (94 versus 159 ng of ligand/mg of cellular protein for LDL-R ⌬5kb HTZ and normal fibroblasts, respectively) with no difference in receptor affinity (2.96 and 3.12 g/ml) compared to normal subjects. A similar experiment with ␤-VLDL as ligand revealed a defect in binding by ⌬5kb fibroblasts of approximately 50% that was also associated with lower receptor number (205 versus 457 ng/mg) and with higher affinity (0.42 versus 0.83 g/ml) (Fig. 1B).
LDL-R mRNA in LDL-R ⌬5kb Fibroblasts-Although LDL-R activity in LDL-R ⌬5kb fibroblasts was only 50 -60% that of normal cells, the ratio of LDL-R mRNA to that for ␤-actin in fibroblasts and in lymphocytes was similar in carriers and non-carriers (7). To compare the relative amount of LDL-R mRNA corresponding to the deleted allele with that of the normal allele, primers surrounding the deletion were designed (see "Materials and Methods"). Surprisingly, in fibroblast-derived cDNA from 5 related and 2 unrelated heterozygotes for the LDL-R ⌬5kb, the deleted allele was consistently overexpressed compared to the normal allele by approximately 50% (Fig. 2). To see if the apparent overexpression of the deleted allele could result from more efficient amplification of a shorter DNA fragment, the ⌬5kb and normal alleles were distinguished with a HincII polymorphism (18) in exon 12 of the LDL-R gene. Haplotype analysis in a kindred (7) containing LDL-R ⌬5kb heterozygotes revealed that the LDL-R ⌬5kb allele did not contain the polymorphic HincII site (data not shown). Digestion of the LDL-R exon 12 amplified from cDNA of LDL-R ⌬5kb HTZ fibroblasts in which the normal LDL-R allele contained the HincII site revealed approximately 50% greater undigested than digested fragment (data not shown). Thus, low receptor activity in LDL-R ⌬5kb HTZ fibroblasts cannot be attributed to defective transcription of the ⌬5kb allele.
LDL-R Immunoblotting-Because the recognition epitope of IgG-C7 is the first repeat of the ligand-binding domain (3), this antibody was not expected to recognize the LDL-R ⌬5kb. Consistent with this, only one band of apparent molecular mass of 140 kDa (average value from 9 independent experiments) was observed when blots of protein extracts from normal and LDL-R ⌬5kb HTZ fibroblasts were probed with this antibody (Fig. 3, left, LDL-R IgG-C7). When similar blots were probed with monoclonal anti-LDL-R antibody IgG-4C4, raised against the carboxyl terminus of the LDL-R (3), only one band was seen for normal fibroblasts, while two bands were seen in LDL-R ⌬5kb HTZ fibroblasts (Fig. 3, left, LDL-R IgG-4A4). The second band, not revealed with IgG-C7, appears at approximately 127 kDa, consistent with the deletion of amino acids encoded by 2 and 3 resulting in the loss of 83 amino acids (approximately 10 kDa).
The relative amounts of LDL-R protein was assessed by densitometry in 4 independent experiments. The amount of total LDL-R protein was similar in normal and ⌬5kb HTZ fibroblasts (Fig. 3, lower right). The amount of protein corresponding to the normal and LDL-R ⌬5kb allele was approximately equal in fibroblasts from 4 LDL-R ⌬5kb HTZ (Fig. 3, left, LDL-R IgG-4A4, and lower right panel). No LDL-R protein was detected with extracts from receptor-negative fibroblasts with either antibody (Fig. 3, left, ⌬10kb HMZ). Thus, the LDL-R ⌬5kb protein appears to be synthesized in normal amounts.
Treatment of cell protein extracts from LDL-R ⌬5kb HTZ fibroblasts with neuraminidase reduced the apparent size of the normal receptor from 147 to 134 kDa (Fig. 4), consistent with previous reports of 10 -15-kDa reduction (19,20). A reduction of apparent molecular mass was also observed for the ⌬5kb protein, from 134 to 117 kDa. The difference in size between the normal and deleted receptor was similar before and after neuraminidase treatment. Thus, no defect in glycosylation of the ⌬5kb receptor was detected.
Cell Surface Binding of 125 I-IgG-4A4 to LDL-R ⌬5kb HTZ and Normal Fibroblasts-To address the possibility that low LDL-R activity in LDL-R ⌬5kb HTZ fibroblasts could be attributable to defective transport of the ⌬5kb receptor to the cell surface, binding of anti-LDL-R antibodies to fibroblasts was studied. The number of binding sites for 125 IgG-4A4, which recognizes both the normal and ⌬5kb receptor, did not differ in normal and LDL-R ⌬5kb HTZ fibroblasts (13 ng/mg cell protein, Fig. 5), despite lower affinity for the antibody in ⌬5kb HTZ fibroblasts (0.43 versus 0.14 g/ml). A similar result was obtained in a second experiment, in which specific binding of 1 and 5 g of 125 IgG-4A4 by LDL-R ⌬5kb HTZ fibroblasts (measured in the presence of a 50-fold excess of unlabeled antibody) was 56 and 75% of normal (data not shown). Binding of 1 and 5 g of 125 IgG-4A4 to LDL receptor-negative fibroblasts was 19 and 21% of that in normal fibroblasts.
Ligand Blotting of LDL-R-Ligand blotting of fibroblast protein extracts with 125 I-LDL resulted in a band of approximately 140 kDa in normal fibroblasts (Fig. 6, lanes 2 and 6) but none in LDL-R-deficient fibroblasts (Fig. 6, lanes 1, 5, and 10). Ligand blotting of cell protein from LDL-R ⌬5kb HTZ fibroblasts revealed the same band as seen in normal subjects, but of lower intensity, and, in only 1 of 4 instances, a faint band of lower molecular mass (Fig. 6, lane 9), corresponding to the ⌬5kb form of the LDL-R. In 2 of 3 independent preparations, no binding of 125 I-LDL to the ⌬5kb receptor was detectable (Fig. 6, lanes 3, 4,  7, and 8), even after 6 days of exposure. Similar experiments with 125 I-␤-VLDL revealed interaction of this ligand with the deleted receptor which was detectable but weak (Fig. 6B). Immunoblotting performed in parallel on these same protein extracts revealed immunoreactive protein corresponding to both the normal and ⌬5kb receptor in equal amounts. Thus, the ⌬5kb receptor displays little or no ability to bind LDL or ␤-VLDL under these experimental conditions. DISCUSSION Site-directed mutagenesis experiments have shown minimal loss of LDL and ␤-VLDL binding, respectively, from an LDL receptor lacking exons 2 and 3 encoding the first two of seven ligand binding repeats (3,4). Based on this information it is surprising that in French Canadians, with the exception of carriers of the ⑀2 allele (7), heterozygosity for a 5-kb deletion of the LDL-R gene (LDL-R ⌬5kb) encompassing exons 2 and 3 is associated with plasma LDL and total cholesterol levels which are equally as elevated as those associated with heterozygosity for an LDL-R null allele. The present study sought to resolve this apparent contradiction between in vivo phenotype and in vitro consequences of the LDL-R deletion.
Among possibilities to explain the association of the LDL-R ⌬5kb with FH were decreased mRNA or protein synthesis, slow transport to the cell surface, or poor affinity of the receptor for LDL. The first two possibilities were ruled out by measurements of normal levels of mRNA and normal levels of LDL-R protein in LDL-R ⌬5kb HTZ fibroblasts. The receptor appeared to be glycosylated normally, implying normal processing (21). LDL-R ⌬5kb HTZ fibroblasts bound similar amounts of anti-LDL-R antibody IgG-4A4 as did normal cells, also suggesting normal transport of the ⌬5kb receptor to the cell surface. However, ligand blotting of cell protein extracts from LDL-R ⌬5kb HTZ fibroblasts revealed little or no LDL binding to the ⌬5kb receptor. Thus, the apparent reduction of 50% in receptor num-  triangles) and LDL-R ⌬5kb HTZ (⌬5Kb, circles) fibroblasts. After a 48-h incubation with 10% LPDS, cells were incubated with 1 ml of medium containing the indicated concentrations of 125 I-IgG-4A4 (485 cpm/ng). After 3 h at 4°C, the total radioactivity bound to cells was determined as described under "Materials and Methods." Shown is specific binding, obtained by subtraction of binding in the presence of a 20-fold excess of unlabeled antibody from the total. Nonspecific binding averaged 40 -50% of total. Due to anomalously high nonspecific binding to NL fibroblasts at 1 g/ml, this point was not included in the curve. Inset shows Scatchard analysis. ber in LDL-R ⌬5kb HTZ fibroblasts compared to normal is due not to the absence of the receptor on the cell surface, such as in a class 2 receptor defect, but to the inability of the receptor to bind LDL, i.e. a class 3 defect. The deleted receptor was also defective in binding apolipoprotein E, as evidenced by 50% of normal maximal cell surface binding of ␤-VLDL in ⌬5kb HTZ fibroblasts and by weak interaction with the deleted receptor on ligand blots. This result is also in contrast to those observed in the transfected receptor lacking exons 2 and 3, which displayed no defect in ␤-VLDL uptake and to data which suggest that LDL-R repeat 5 mediates binding of apolipoprotein E-containing lipoproteins (22).
Conversely to the present study, theoretically severe mutations in the LDL receptor gene do not always result in the FH phenotype. A class 2B defective LDL-R receptor (i.e. synthesized but not displayed on cell surface) was reported (23,24) in which heterozygous parents of the affected homozygous child did not express consistent or significant hypercholesterolemia. An LDL-R gene deletion of approximately 10 kb, FH-Tonami-2 (25), eliminating exons 2 and 3, has been found in 10 Japanese families with hypercholesterolemia and is associated with cholesterol levels lower than those of typical FH patients, including two heterozygous family members with normal plasma cholesterol levels (26). According to the present data, such cases of milder than expected FH phenotype are most likely explained by up-regulation of the normal allele in some cases rather than by residual function of the defective LDL-R.
Several possibilities could explain the discrepancy between the consequences of deleting LDL-R exons 2 and 3 described herein and those previously described (4). The present ϳ5-kb genomic deletion of the LDL-R gene between introns 1 and 3 (2) is predicted to result in an in-frame creation of an Ala residue at the expense of Val 2 and Pro 84 ; i.e. an amino-terminal sequence of Ala 1 -Ala-Pro 85 . In the site-directed mutagenesis study (4), the deletion was of residues 1-83; i.e. an aminoterminal sequence of Pro 84 -Pro 85 . The possible functional sig-nificance of this difference is not clear but seems unlikely to account for differences in the ligand binding ability of the two deleted receptors. Other examples of discrepancies between apparent effects of gene deletions as assessed by in vitro studies and by phenotypic expression of a naturally occurring deletion is seen when domain 3 (O-linked sugar domain) is deleted in vitro by site-directed mutagenesis resulting in no defect in receptor activity (27), while a homozygote for such a mutation expresses FH (28,29). A variant of lipoprotein lipase containing an Asn2913 Ser substitution which is functionally mildly abnormal in vitro (30) is associated with type IV hypertriglyceridemia in French Canadians (31). In the case of lipoprotein lipase, the unexpectedly profound clinical effect of heterozygosity for a mildly defective variant may be attributable to a dominant negative mechanism, wherein the defective variant would interfere with lipoprotein lipase dimerization, which is necessary for function. Although the LDL-R is present on the cell surface as a monomer, one possible locus for a dominant negative effect of heterozygosity for a defective receptor could be in receptor clustering prior to internalization (14, 32).
However, such an explanation for the defective LDL-R activity observed in LDL-R ⌬5kb HTZ fibroblasts is unlikely based on the observation that the ⌬5kb receptor is poorly able to bind ligand. A more likely explanation for the discrepancies between results obtained from site-directed mutagenesis studies and HTZ fibroblasts is the amount of receptor expressed under each set of circumstances. Overexpression of proteins in transfected cells has been observed to result in unphysiological phenomena, such as secretion of immature forms of apolipoprotein A-I (33) or constitutive activity of sterol regulatory element binding proteins 1 and 2 (34). Thus, prediction of in vivo phenotypic effects of gene mutations from their functional effects in transfected cells may be complicated by the unphysiologically high levels of expression.
It has been estimated that 15-30% of "isolated" O-linked carbohydrate is located on the amino-terminal half of the receptor (27), more specifically, within the ϳ40-kDa ligand-binding domain (35). The absence of glycosylation in this domain in a monensin-resistant cell line has been shown to reduce LDL-R affinity for LDL by approximately 75% (35). Thus, loss of Olinked carbohydrate may at least partially explain the absence of ligand binding of the LDL-R ⌬5kb receptor observed in the present study. Although similar decreases in molecular mass after neuraminidase treatment between the LDL-R ⌬5kb and normal protein in the present study may imply that significant O-linked glycosylation does not occur in the first two repeats of the LDL-R, it is questionable whether a difference would be detectable. O-Linked sialic acid and galactose residues are expected to contribute approximately 25 kDa to the molecular mass of the LDL-R (27). Therefore, if the carbohydrate was evenly distributed among repeats, the expected loss of molecular size after neuraminidase treatment due to glycosylation of the first two repeats is 3 kDa. Thus, it is possible that the effect of deletion of exons 2 and 3 on LDL-R activity is attributable to loss of carbohydrate and subsequent loss of receptor affinity for LDL. As such, possible differences in glycosylation patterns between human fibroblasts and transfected CHO cells may contribute to differences between the present and a previous study (4) on the functional consequence of the absence of these two repeats.
Another potentially interesting explanation for the surprisingly severe effects of the LDL-R ⌬5kb is a regulatory effect of a gene deletion that is not apparent in cells transfected with cDNA. Thus, one possibility is that deletion of a liver-specific enhancer in introns 1 or 2 causes a liver-specific regulatory FIG. 6. Ligand blotting with 125 I-LDL to solubilized LDL receptors from normal and LDL-R ⌬5kb HTZ fibroblasts. Ninety g of detergent-solubilized fibroblast protein extracts were electrophoresed on 6% SDS-polyacrylamide gels. After transfer to nitrocellulose, membranes were probed with 17 g/ml 125 I-LDL (100 -300 cpm/ng) (top) or 35 g/ml 125 I-␤-VLDL (80 cpm/ng) (bottom). Shown are autoradiograms. Molecular mass (kDa) is indicated at the right; bands at 140 and 127 represent the normal and ⌬5kb form of the LDL-R, respectively. Top, extracts from receptor-negative (lanes 1, 5, and 10), normal (lanes 2 and 6), and LDL-R ⌬5kb HTZ fibroblasts (lanes 3, 4, and 7-9); bottom, from receptor-negative (lane 2), normal (lanes 3 and 4), and ⌬5kb HTZ (lanes 5-10) fibroblasts.
defect. An LDL-R gene deletion of exons 2 and 3 similar to the LDL-R ⌬5kb has been reported to result from Alu recombination (36). Alu sequences have been known to act as enhancers (37) or repressors (37). In the present study, however, LDL-R mRNA corresponding to the ⌬5kb allele was consistently higher in fibroblasts and lymphocytes than that of the normal allele, suggesting deletion of an element which may act as a repressor, at least in these cell types. Further studies will explore the regulatory consequences of the LDL-R ⌬5kb.