Mannose 6-phosphate receptor-mediated uptake of modified low density lipoprotein results in down regulation of hydroxymethylglutaryl-CoA reductase in normal and familial hypercholesterolemic fibroblasts.

Monophosphotetramannosyl-1-deoxymannitol-1-yl-low density lipoprotein (Man-6-P-LDL) was prepared by covalent attachment of the pentasaccharide omega-(6-phospho-tetra(alpha 1-3)mannosyl(alpha 1-2)mannose to amino groups on low density lipoprotein. Normal human fibroblasts were shown to specifically bind, internalize, and degrade 125I-labeled Man-6-P-LDL. Specificity for the mannose 6-phosphate (Man-6-P) receptor was demonstrated by competitive displacement with cold Man-6-P-LDL, Man-6-P, or mannose. No displacement was seen with cold LDL. Kd is estimated to be less than or equal to 2 X 10(-9) M. Degradation of 125I-labeled Man-6-P-LDL in familial hypercholesterolemic fibroblasts showed the same time course and specificity as observed in normal fibroblasts. Man-y-P-LDL was also able to deliver cholesterol to the cytosol where down regulation of the enzyme 3-hydroxy-3-methylglutaryl CoA reductase was observed in both normal and familial hypercholesterolemic fibroblasts. Down regulation could be blocked by Man-6-P in both cell lines. The possible uses of agents such as Man-6-P-LDL as research probes and therapeutic tools directed to specific cell types are discussed.

Studies on the receptor-mediated transport of a number of hormones, toxins, and other proteins into cells have revealed several common features which may have general application to all protein transport systems (for review see Ref. 1). In many cases binding to a cell surface receptor involves a discrete portion of the protein molecule spatially distinct from the site on the molecule eliciting a physiological response once entry has been achieved. This generalized bifunctional scheme may he applied not only to bacterial toxins and certain peptide hormones (where the response observed is due to some enzymatic or other action of a peptide (2-4), but also includes transport proteins in which the physiological effector is nonprotein in nature (e.g. cholesterol transport by low density lipoproteins (5), vitamin B12 transport via transcobalamin I1 complex (6), etc.). Clear demonstration of the separation of functional components is often possible by physical separation (3,7,8), selective inhibitors (9,10) or promoters (9), covalent modification (11), and genetic selection (12).
If the functions are truly separate one should he able to modify receptor specificity by alterations on one portion with-* The costs of publication of this article were defrayed in part by the payment af page charges. This artisle must therefore be hereby marked "ad~lertizement" in accordance with 18  out affecting the other. Utilizing this concept, a number of hybrid proteins have now been constructed with varying degrees of efficacy (13)(14)(15)(16)(17)(18)(19). The receptor-binding specificity of one protein is covalently attached to a second protein molecule bearing its own enzymatic or other biological specificity. These hybrid proteins may serve as a new class of cell typespecific pharmacological reagents (1, 13). In addition they may have utility for probing the biochemical steps in the entry process.
While the use of protein-protein hybrids as probes of transport mechanisms is a relatively new development, the ultimately more refined use of small ligands as the recognition moiety for binding has been exploited successfully for many years especially in the study of carbohydrate-specified catabolic systems for glycoproteins (20). The work reported here uses this technique of coupling an oligosaccharide chain to a protein as a means of specifying an alternate receptor-mediated route of entry. We report here the first use of this method to incorporate a new recognition signal into a transport protein such that receptor specificity is altered. Thus, it is possible to achieve transport via an unique receptor-mediated process utilizing a route independent of that normally used, while maintaining physiological effector function intact.
We have reported elsewhere the use of this method to produce a cell type-specific toxin, Man-6-P-ricin (21).
By using appropriate sugars to block the receptor-mediated uptake of a variety of glycoproteins, a number of carbohydrate recognition systems involved in uptake have been defined. Receptor-mediated endocytosis of glycoproteins involves primary recognition by the terminal sugar residue (nonreducing) in one of the oligosaccharide chains of the glycoprotein. Carbohydrate recognition systems have been described showing specificity for galactose (e.g. asialoglycoproteins in rat and avian hepatocytes (22,23)), N-acetylglucosamine or mannose (avian hepatocytes, liver sinusoid cells, and macrophages (23-2 6 ) ) , fucose (lactoferrin in mouse hepatocytes (27)) and mannose &phosphate (lysosomal hydrolases in human fibroblasts (28,29)). Rogers and Kornfeld (20) followed by Lee and coworkers (30,31) and others (26) have constructed a number of semi-synthetic "neoglycoproteins" which have proved to be essential for a definition of the Carbohydrate specificity of the hepatic clearance systems recognizing galactose or mannose/ N-acetylglucosamine. Further confirmation of the cell type specificity of these systems has been obtained by electron microscope autoradiographic studies (32).
Extensive study on the mechanism of uptake of low density lipoproteins into cultured human fibroblasts has resulted in a proposed description of the defect in type IIa hyperlipidemia (familial hypercholesterolemia) (for review see Ref. 5). This heritable disease apparently results from the absence of an LDL' receptor on the surface of fibroblasts and a number of other cell types. The mutant cells are thus unable to bind LDL through the normal high afflnity route of wild type fibroblasts. This inability to bind LDL specifically thus prevents internalization, degradation, and subsequent release of cholesterol and down regulation of the enzyme 3-hydroxy-3methylglutaryl coenzyme A reductase.
In this report we show that the failure of familial hypercholesterolemic fibroblasts to take up LDL can be circumvented in tissue-cultured fibroblasts by supplying monophosphotetramannosyl-1-deoxymannitol-1-yl-LDL. By coupling LDL to a naturally occurring oligosaccharide, o-(6-phospho)-tetra(d-3)mannosyl-(cul-2)mannose, uptake of LDL into the lysosome is achieved via the route normally used for the recapture of lysosomal hydrolases (as indicated by the ability to block uptake completely with Man-6-P and the inability of cold LDL to block entrance or degradation of 1251-labeled Man-6-P-LDL): Lipoprotein thus transported is rapidly degraded in both normal and familial hypercholesterolemic fibroblasts with a resultant down regulation of HMG-CoA reductase in both cell types.
Another chemically modified LDL derivative has been reported to elicit such response in fibroblasts, e.g. cationized LDL in which the lipoprotein was covalently altered by blocking a large number of carboxyl groups with N,N-dimethyl-1,3diamine (33). The uptake of cationized LDL is a nonsaturable process over the reported concentration range (and hence may represent uptake as a result of low affinity binding to a very large number of surface membrane anionic sites). To contrast this with the present study, LDL coupled to a phosphomannan oligosaccharide has lost the ability to bind to the LDL receptor but has the newly acquired ability to bind specifically to a finite number of lysosomal hydrolase receptors.
We have thus modified a physiological protein, LDL, by attachment of mannose-6-phosphate. Man-6-P-LDL binds to a distinct physiological receptor, enters the cell, and is thus delivered to the lysosome independently of the LDL receptor. In the lysosome, degradation of the lipoprotein particle results in expression of the normal physiological function of LDL, namely down regulation of the enzyme HMG-CoA reductase.

ZsoLation of LDL
All LDL used in this study was between density 1.006 to 1.06 and was prepared by the density flotation method of Have1 et al. (34). Fresh plasma collected in Na2/EDTA (4 mM) was centrifuged for 15 min at 4000 X gmaX to remove any cellular debris. The supernatant plasma was centrifuged at 4°C for 16 h at 50,000 rpm in a Beckman 60 Ti rotor after which lipoproteins of less than solvent density (1.006) had concentrated in a layer at the top of the tube, with an intennediate clear zone and then a second layer of heavier lipoproteins. Tubes were sliced in the middle of the clear zone using a tube slicer (Nuclear Supply Co., Washington, D.C.) and the top layer discarded. The infranatant was adjusted back to the original volume by addition of NaCl(O.85% w/v) and then the density was raised to 1.06 by addition of solid KBr (7.98 g/100 ml) and recentrifuged as before. The tubes were sliced and the supernatant recovered, adjusted back to the original volume with KBr solution (density = 1.06), and recentrifuged.
The abbreviations used are: LDL, low density lipoprotein; Man-6-P-LDL, monophosphotetramannosyl-l-deoxymannitol-l-yl-low density lipoprotein; Glc-LDL, maltosyl-1-deoxysorbitol-1-yl-LDL; PMs, w-(6-phospho)-tetra(cul-3)mannosyl(al-2)mannose; Man-6-P, Again the tubes were sliced, and the LDL recovered in the supernatant was dialyzed for 24 h against four changes of l liter of saline/ EDTA. Following dialysis LDL was concentrated on an Amicon UM 30 ultrafilter to approximately 30 mg of protein per ml and stored at 04°C. Maximum storage time was 2 months. Lipoprotein-depleted serum was obtained by density flotation of human serum at density 1.21. All lipoproteins were removed by slicing out the supernatant layer. The infranatant was then adjusted to the original volume and dialyzed against three changes of 2 liters of saline/EDTA (pH 7.4) over a 24-h period. Serum thus obtained was filter sterilized and stored at -2O' C.

Preparation of w-(6-Phospho)-tetra(al-3)mannosyL(al-2)mannose
Phosphomannan obtained from Hansenula holstii NRRL Y-2448 was a g f t from Dr. Morey E. Slodki, Northern Regional Research Laboratories, Peoria, IL. Preparation of the phosphorylated pentasaccharide was essentially as already described (35). The structure of this pentasaccharide has been shown to be 6-P-Man(al-3)Man(al-3)Msn(al-3)Man(al-Z)Man (35,36). Phosphomannan (5.4 g) was swollen in 200 ml of KC1 by heating to 100°C on a hot plate for 30 min with stirring. After cooling the pH was adjusted to 2.5 by addition of 2 N HC1, and the solution was covered with a watch glass and placed in a boiling water bath for 6 h. After cooling the pH was adjusted to 6.9 with NaOH (1 N), and the faintly cloudy solution was centrifuged for 5 min at 4000 X gmaX. The volume of the supernatant was readjusted to 200 ml by addition of water. Barium acetate (2 g) was then dissolved in the supernatant, and 1 N NaOH was added to pH 9.5. Addition of 20 ml of absolute ethanol followed by chilling on ice effected precipitation of unhydrolyzed phosphomannan and a high molecular weight acid-resistant core. After centrifugation as before, 400 ml of cold absolute ethanol was added to precipitate the barium salt of PMs which was recovered by centrifugation as before. The precipitate so obtained was dissolved in water and submitted once again to ethanol fractionation. The final product was dissolved in water and lyophilized for storage as the Ba2+ salt.
Prior to use in coupling to LDL, 250 mg of PM5 was dissolved in 2 ml of acetic acid (0.1 N) and chromatographed on Sephadex G-25 superfine (95 X 1.5 cm) with 0.1 N acetic acid as eluant. Fractions (2 ml each) were collected and analyzed for carbohydrate content. In a typical experiment 3 to 5% of the applied carbohydrate (elution volume 90 ml) eluted before the major peak. This was believed to be residual core phosphomannan and (PM& resulting from incomplete hydrolysis. The major peak (elution volume 110 ml to 130 m l ) was pooled and acetic acid removed on a rotary evaporator at 4OOC. The Ba'+ salt was exchanged for Na+ by passing over a column of Dowex 50 (H' form, 10 X 2 cm) and titrating with NaOH (1 N) to pH 8.5. PMs (Na') so obtained was concentrated on a rotary evaporator at 4OoC to approximately 400 mg/ml for use in the coupling experiment. Hex0se:phosphate ratio of this peak was 5.0 to 5.51.0. PMs groups per 500,000 daltons of LDL protein or approximately to 18% of the available lysines (based on reported 57 lysine per 100,000 molecular weight apo B (38)). NaCNBH3 freshly obtained (Aldrich) was adequate for these purposes. Older stock was purified according to the method of Borch et al. (39). Ethyl acetate was substituted for ether as the solvent of choice in preparation of the dioxanate.'

Iodination of Man-6-P-LDL
Labeling of Man-6-P-LDL with lZ5I was accomplished by the iodine monochloride method of Macfarlane (40). To Man-6-P-LDL (1 mg) in 130 pl of saline/EDTA in a 1.5-ml Eppendorf tube was added 80 yl of glycine/NaOH ( (42,43). Cells were grown in dishes as described above. At the beginning of the experiment, the medium was removed by aspiration and replaced with 2 ml of fresh medium containing the indicated amounts of lz51-labeled Man-6-P-LDL and the various competitors for displacement 2. Degradation-After the indicated incubation period, 1 ml of medium was removed from the dish and added to 0.5 ml of fetal calf serum, then 0.5 ml of 50% trichloroacetic acid was added, the tube was mixed and incubated for 30 min at 4"C, and then centrifuged at 1506 x g for 10 min.  (48).

Studies on Man-6-P-LDL in Normal Human Fibroblasts-"'I-Labeled
Man-6-P-LDL is shown to be rapidly bound to the surface of normal human fibroblasts, reaching a steady state after about 1 h (Fig. 1). Similarly, accumulation of 1251labeled Man-6-P-LDL in a nonheparin-releasable compartment begins immediately and then approaches steady state. Trichloroacetic acid-soluble counts released to the extracellular medium show an initial lag of about 1 h and then a rapid increase indicating that the artificial hybrid constructed is rapidly degraded by normal human fibroblasta. Since degradation continues to increase with time while binding and internalization level off we conclude the surface receptors are maintained at a steady state concentration during the course of internalization.
Recycling of receptors or unmasking of existing receptors may account for this process.
The processes of binding, uptake, and degradation are shown to be both saturable and specific for the Man-6-P receptor (Fig. 2, A, B, C) since lZ51-labeled Man-6-P-LDL can be displaced by excess cold Man-6-P-LDL but not by LDL. Dependence on the Man-6-P receptor-mediated uptake is further shown by the displacement of label with both mannose B-phosphate and mannose. K, for mannose B-phosphate is approximately lo-" M and for mannose approximately 10-l M. These values are in agreement with results obtained for inhibition of uptake of lysosomal hydrolases (28,29). When Man-6-P-LDL was assayed for its ability to competitively inhibit uptake of cu-r,-iduronidase in tibroblast cultures, K, was determined to be 2 x 10e7 M (49). While an accurate dissociation constant Kd cannot be obtained from our data, we estimate Kd 5 2 x lo-' M.3 This is close to Kd = W9 M (50,51) reported for various lysosomal hydrolases. The absence of displacement by cold LDL indicates that Man-6-P-LDL does not depend on the LDL pathway for any step in the process leading to degradation and must, therefore, be wholly depend-  Cells were plated at 2 X 105/dish in 60-mm culture plates as described under "Experimental Procedures" and grown for 6 days in complete medium (+IO%, FCS). On Day 6 the medium was removed and replaced with 2 ml of medium containing 6% lipoprotein-depleted serum. After 24 h, the medium was replaced with 2 ml of fresh lipoprotein-depleted medium containing 0.7 pg/ml of '251-labeled Man-6-P-LDL (75 cpm/ng, 39,000 cpm/dish) f 100 pg/ml of Man-6-P-LDL and incubated at 37°C in a 5% COS atmosphere. Degradation, at various times the medium was removed and lZ5I-labeled trichloroacetic acid-soluble material was determined by precipitating with 5% trichloroacetic acid. Binding, after removal of medium, the dish was washed twice rapidly with 2-ml portions of cold Buffer A and then incubated for 1 h at room temperature with 2 ml of Hepes plus heparin (Hepes (10 mM), heparin (IO mg/ml), and NaCl (50 mM), pH 7.4). An aliquot (1 ml) of this buffer was used for determining heparinreleasable counts. Internalization, nonheparin-releasable material was released by dissolving the cell monolayer in 2 ml of NaOH (0.1 N). Values are corrected for cell protein. Correction for specific binding, internalization, and degradation was made by subtracting counts obtained in the presence of excess cold Man-6-P-LDL. At 24 h, nonspecific binding, internalization, and degradation were, respectively, 22, 9, and 10% of the total counts obtained. Lipoprotein concentrations are based on protein molecular weight of 500,000 per mol of LDL. Error bars indicate +I S.D. M , binding (heparinreleasable counts); t " l , internalization (nonheparin-releasable counts); A-A, degradation (trichloroacetic acid-soluble counts). ent on the alternative pathway provided by mannose 6-phosphate attachment. This is entirely in agreement with published data showing loss of LDL binding when the primary amino groups of lysine were blocked (52). Using both LDL and 25-hydroxycholesterol(O.6 pg/ml) plus cholesterol (15 pg/ml) as positive controls for down regulation of HMG-CoA reductase, we examined the dependence of down regulation on concentration using a fixed incubation time (12 h). Both Man-6-P-LDL and LDL exhibit a similar dose-dependent effect on the level of enzyme measured in fibroblasts. In the experiment shown in Fig. 3, loss of 50% of enzyme activity occurs at approximately 8 X M for Man-6-P-LDL compared to 1 x lo-@ M for LDL. Various preparations of Man-6-P-LDL were able to down regulate the enzyme to varying extents. The best preparations of Man-6-P-LDL were equipotent with LDL in this assay. This variation in batches may reflect some loss of cholesterol from the lipoprotein particle as a result of the prolonged incubation at 37°C in the coupling reaction.
In the presence of 10" M Man-6-P a shift in the dose response curve is seen for Man-6-P-LDL but not for native LDL. The ability of Man-6-P to provide almost complete blockade of the Man-6-P-LDL-mediated down regulation of HMG-CoA reductase is a further piece of evidence supporting our conclusion that Man-6-P-LDL can no longer be recognized by the receptor for native LDL but must now rely entirely on the recognition signal provided by the attached oligosaccharide.
LDL incubated with PMs but in the absence of NaCNBHs was shown to behave identically with native LDL (Table I) indicating that covalent attachment of the ligand is necessary to specify entrance. Glc-LDL, at a concentration predicted to reduce enzyme activity by 50%, was unable to effect the down regulation of HMG-CoA reductase in agreement with other reports that LDL modified in such a way as to lose a substantial proportion of its amino groups is no longer recognized by the LDL receptor (52). lemic Fibroblasts-HMG-CoA reductase activity in normal human fibroblasts could be induced a t any state of growth simply by replacing normal media with medium containing 6% lipoprotein-depleted serum. In contrast, GM 2000 fibroblasts (from a familial hypercholesterolemic individual) were shown to exhibit maximum HMG-CoA reductase activity 36 to 48 h after seeding into plates in the presence or absence of lipoproteins in the medium! This narrow window of activity necessitated a change in experimental procedure to accommodate this cell line, and so all subsequent experiments were performed between 36 and 48 h after seeding. 1251-Labeled Man-6-P-LDL is rapidly degraded by both normal and familial hypercholesterolemic fibroblasts assayed at low cell density (Fig. 4). Either cold Man-6-P-LDL or Man-6-P could be used to compete for the degradation of '251-labeled Man-6-P-LDL, whereas cold LDL has no effect. This result in familial hypercholesterolemic fibroblasts is clear evidence that the LDL receptor is not required for internalization of Man-6-P-LDL. Man-6-P-LDL for both normal and familial hypercholesterolemic fibroblasts. The results for these representative experiments show the down regulation in familial hypercholesterolemic fibroblasts which was always slightly less than that observed for normal fibroblasts assayed on the same day. The down regulation can be inhibited by lo-' M Man-6-P and thus can be clearly seen to involve the Man-6-P receptor. (The higher Man-6-P concentration to inhibit this process is required to overcome the extremely large dose of Man-6-P-LDL (up to 100 pg/ml) used in these studies compared with the concentration of 1251-labeled Man-6-P-LDL used in the tracer studies (0.1 to 0.5 pg/ml).)

Studies on Man-6-P-LDL in Familial
Our data demonstrates the use of a n oligosaccharide marker to covalently modify a protein with a known physiological role in such a way that transport to the interior of the cell is effected via an alternate receptor, leaving the physiology of the protein intact (as measured by intracellular response).
The rate of internalization and degradation of 1251-labeled  Cells were plated at 2 X lo5 cells/dish in 2.5 ml of medium containing 6% lipoprotein-depleted serum. After 36 h of growth, medium was removed and replaced with fresh lipoprotein-depleted medium containing 0.2 p g / d of '251-labeled Man-6-P-LDL (250 cpm/ng) and the indicated competitor. Degradation to Trichloroacetic acid-soluble material was determined at various times as described in Fig. 1. All values were corrected by subtracting trichloroacetic acid-soluble counts from the input medium. t " . ,   Fig. 4. After 36 h of growth medium was replaced with fresh lipoprotein-depleted medium containing 100 pg/ml of lipoprotein & Man-6-P (10 mM), incubated for 10 h, and assayed for HMG-CoA reductase activity as described in Fig.   3. m, normal cells; 0, familial hypercholesterolemic cells. A , effect of LDL on down regulation of HMG-CoA reductase. B , inhibition of Man-6-P-LDL-mediated down regulation of HMG-CoA reductase. achieved is fully predicted by the bifunctional model for transport proteins. What role, if any, the LDL portion of Man-6-P-LDL has in achieving eventual localization in the lysosome remains to be answered. Bifunctionality may extend beyond the transported protein to the receptor itself. Since there are clearly two functions, binding and transport, it will be of interest to determine whether these functions reside in the same molecules or require interaction between several membrane components. Other work in the laboratory using oligosaccharides linked to ricin may be useful in dissecting out these interrelationships between the multiple functions of a receptor and the transported protein (21).
Man-6-P-LDL o r other lipoprotein hybrids designed to make use of this approach can be applied to the study of the various levels of control of lipoprotein synthesis and secretion. The rate of LDL-apoprotein B synthesis has been shown to be significantly higher in familial hypercholesterolemic homozygotes than in normals while the fractional catabolic rate was lower (54). Observations on the synthesis of cholesterol in liver biopsies in familial hypercholesterolemic individuals indicate oversynthesis or defective suppression of synthesis in response to cholesterol feeding (55,56). In contrast there is no increase in the incorporation of [I4C]acetate into cholesterol when whole body synthesis is measured (57), nor is there increased synthesis of cholesterol in skin biopsies of familial hypercholesterolemic patients (58). The evidence points to a complex series of control mechanisms interacting to produce elevated plasma LDL in familial hypercholesterolemic individuals. Hybrid lipoproteins similar to Man-6-P-LDL can be constructed and may possibly cast some light on these relationships. Man-6-P-LDL could be used to deliver cholesterol to the fibroblasts of familial hypercholesterolemic individuals if this is found to be of therapeutic use. A similar means might be used to regulate hepatic cholesterol synthesis. In fact the hybrid approach permits the design of a variety of agents for delivery to many specific cell types.
We have demonstrated here one example of this new class of physiological reagents in which receptor specificity may be altered while effector function is maintained. This approach may prove useful in the construction of research probes and therapeutic agents for delivery to specitk cell types.