Characterization of Monoclonal AntibodiesAgainst Human Low Density Lipoprotein

Ross W. Milne, Richard Theolis, Jr., Roy B. Verdery,and Yves L Marcel

Seven monoclonal antibodies against human low density lipoprotein (LDL) havebeen characterized as to their specificity and ability to interfere with the LDL pathwayin cultured human flbroblasts. The Immunoreactivity with LDL of two of the antibodies(2D8 and 4G3) was particularly sensitive to modification of lysine and arginlne resi-dues In LDL. Cotitration experiments indicated that the antibodies 3A8 and 3A10 mayreact with the same determinant and that five antibodies (5E11, 3A8, 3A10, 4G3 and3F5) recognized determinants that were grouped in the same region of the molecule.The two other antibodies (1D1 and 2D8) reacted with determinants distant from thisregion. When tested In molar excess relative to LDL, Fab fragments of 5E11, 3A8,3A10,4G3 and 3F5 (but not 1D1 or 2D8) were capable of blocking the binding of 1 2 5 I -LDL to the LDL receptor and Interfering with LDL suppression of cholesterol synthe-sis in cultured fibroblasts. Increasing the concentration 10-fold did not change theresults significantly. Based on these results we have proposed a map of the determi-nants as they would appear in LDL.(Arteriosclerosis 3:23-30, January/February 1983)

L ow density lipoprotein (LDL) is the principal carri-er of plasma cholesterol in humans, and LDL

levels have been positively correlated with the de-velopment of atherosclerosis.

Apolipoprotein B (apo B) is the major protein spe-cies of LDL, but the elucidation of its structure hasproven to be a most difficult task mainly because ofits total insolubility in a delipidated form which maybe related to high oxidation sensitivity.1 While widelyvarying molecular weights for apo B have been re-ported in recent years, it is now thought to have amolecular weight in excess of 200.000.2'3 The LDLreceptor present on the surface of mammalian cellsbinds certain plasma lipoproteins by recognizingsites which are present on apo B as well as on apoE.4'5 The binding of LDL to the LDL receptor initiatesa series of events which has been called the LDLpathway.6'7 The LDL is taken into the cell where it ishydrolyzed in lysosomes. The released cholesterolinhibits endogenous cholesterol synthesis at thelevel of the rate-limiting enzyme, hydroxymethyl glu-

From the Laboratory of Lipoprotein Metabolism, Clinical Re-search Institute of Montreal, Montreal, Quebec, Canada.

This study was supported by grants from the Medical ResearchCouncil of Canada (MA-6804 and MT-4011), the Quebec HeartFoundation and the Conseil de la Recherche en Sante duQuebec.

Address for reprints: Dr. Ross Milne, Clinical Research Instituteof Montreal, Laboratory of Lipoprotein Metabolism, 110 Pine Ave-nue West, Montreal H2W 1R7, Quebec, Canada.

Received May 20, 1982; revision accepted September 13,1982.

taryl coenzyme A reductase (HMG CoA reductase),stimulates cholesterol esterification, and inhibitssynthesis of the LDL receptor itself.

Monoclonal antibodies have proven useful in thestudy of protein structure and in the identification offunctional domain of proteins. With the aid of mono-clonal antibodies against human LDL, we are tryingto identify and eventually purify and characterize theregions of apo B recognized by the LDL receptor.Here we describe seven monoclonal antibodiesagainst LDL which are differentiated on the basis oftheir cross-reactivities with chemically modified LDLand their abilities to interfere with the LDL pathway.In addition, we describe a novel adaptation of thecotitration method8 which allows a tentative mappingof the antigenic determinants.

MethodsPreparation of LDL

Blood from normal subjects was collected intoEDTA and the red blood cells were removed by cen-trifugation. Lipoprotein subfractions were preparedin a Beckman L5-65 ultracentrifuge with a 50.2 Tirotor (Beckman Instruments Incorporated, SpincoDivision, Palo Alto, California). LDL were isolated bysuccessive ultracentrifugations9 at 4° between den-sities of 1.030-1.050 g/ml, dialyzed exhaustivelyagainst phosphate-buffered saline (PBS) containing1 mM EDTA, sterilized by ultrafiltration and stored at4°C.

23 by guest on November 7, 2015http://atvb.ahajournals.org/Downloaded from

24 ARTERIOSCLEROSIS VOL. 3, No 1, JANUARY/FEBRUARY 1983

Production of Monoclonal Antibodies

Female 8-week-old BALB/c mice (Jackson Labo-ratories, Bar Harbor, Maine) whose spleens were tobe used for cell fusion were given a primary intraperi-toneal injection of 100 fig of LDL in completeFreund's adjuvant (CFA). After 3 weeks, the micewere given a second subcutaneous injection of 50 figof LDL in CFA. One week later and 3 days before thefusion, 2 fig of LDL was injected intravenously. Thetechnique for cell fusion has been previously de-scribed.10 Following fusion, the cells, suspended inDulbecco's Modified Eagle's Medium (DMEM), sup-plemented with 15 mM Hepes, 40 fiM 2-mercap-toethanol, penicillin-streptomycin, 30% fetal bovineserum (FBS), 0.11 mM hypoxanthine, 0.38 fiM amin-opterine and 16 fiM thymidine were distributed in576 microculture wells (Costar, Cambridge, Massa-chusetts). When sufficient growth had occurred, nor-mally at 10 to 15 days, the supernatants werescreened by a solid phase radioimmunoassay(RIA)10 for the presence of anti-LDL antibodies usingRemovawells (Dynatech Laboratories Incorporated,Dynatech Corporation, Alexandria, Virginia) whichhad been coated overnight at room temperature with200 fi\ of a solution of LDL (50 fig LDL protein/ml in5 mM glycine pH 9.2). The cells in the positive wellswere recloned by limiting dilutions in 96 well micro-culture plates at a cell density that produced growthin approximately 10% of the wells. To identify theimmunoglobulin class and subclass, a sample of hy-bridoma supernatant, concentrated by precipitationwith 50% saturated ammonium sulphate was testedby double immunodiffusion against antisera specificfor mouse immunoglobulin (Ig) classes and sub-classes.11 Selected subclones were injected intra-peritoneally into BALB/c mice (approximately 5 x106 hybridoma cells/mouse) and the resulting asciticfluid was used as a source of anti-LDL antibodies.

Isolation of Anti-LDL Immunoglobulin

The IgG subclass containing the anti-LDL anti-body was isolated from the ascitic fluid of hybridoma-bearing mice by elution from protein-A Sepharose4B (Pharmacia Incorporated, Uppsala, Sweden)with a step-wise pH gradient.12

Preparation of Fab Fragments

Papain digestion of isolated IgG was carried outaccording to the procedures of Gorini et al.13 Theundigested IgG and Fc fragments were removed bypassage on Protein-A Sepharose 4B. The anti-LDLactivity of the Fab fragments was measured by asolid phase radioimmunoassay. We pipetted 200 (i\of LDL (50 fig LDL protein/ml in 5 mM glycine pH 9.2)into each Removawell and left it overnight. The fol-lowing day the wells were washed in 0.15 M NaCIcontaining 0.025% Tween 20. We then added 200 fi\aliquots of serial dilutions of anti-LDL Fab in PBScontaining 5% FBS to the wells and left them for 3

hours at room temperature. The wells were washedas before and 200 fi\ of purified 125I antimouse Fab11

in PBS with 5% FBS was added and kept at roomtemperature for a further 3 hours. The wells werewashed again and the bound radioactivity deter-mined. The concentration of anti-LDL Fab, whichresulted in bound radioactivity three times that ob-tained with an equivalent concentration of normalmouse Fab, was used as a measure of the antibodyactivity.

Cotltratlon of Anti-LDL Antibodies

The method of Fisher and Brown8 was modified forsoluble antigens and 200 fi\ of LDL (25 fig LDL pro-tein/ml in 5mM glycine pH 9.2) was used to coatRemovawells as already described. Serial dilutionsof either individual antibodies (ascitic fluid) or a 1 to 1(vol/vol) combination of two different antibodies wereprepared in PBS/5% FBS. Then 200 fi\ aliquots ofthe dilution were added to the Removawells and therest of the assay was carried out as previously de-scribed for measuring antibody activity in the Fabpreparations.

lodlnatlon of Proteins

For RIA, iodination of IgG (50 fig) and LDL (25 fig)with 0.5 mCi of 125I (Amersham Corporation, Arling-ton Heights, Ilinois), was carried out according to themethod of Mellman and Unkeless.14 Free 125I wasremoved by passage on AGIX8 resin (Bio-Rad Labo-ratories, Richmond, California). Labelled proteinswere stored in PBS containing 0.1% bovine serumalbumin and 0.02% sodium azide. Immediately afterlabelling, more than 90% of the radioactivity wasprecipitated by 10% trichloroacetic acid. In the caseof 125I-LDL, the trichloroacetic acid nonprecipitableradioactivity increased upon storage. Consequently,the 12SI-LDL was repassed on AGIX8 just before anassay and was used within 14 days of the labelling.For binding experiments, LDL were iodinated by themethod of Bilheimer et al.15

Radioimmunoassay of Apo B

The RIA for apo B was essentially the same as thatfor apo E.10 We distributed 200 fi\ of the IgG fractiondiluted in 5 mM glycine in Removawells and left itovernight. The concentration of IgG chosen on thebasis of preliminary experiments ranged from 5 to 20/xg/ml depending upon the individual clone. Thewells were washed with 0.15 M NaCI with 0.025%Tween 20. The sample to be tested was diluted inPBS containing 5% FBS and mixed with 12SI-LDL; thefinal concentration of 125I-LDL was approximately 10ng/ml. We transferred 200 fi\ of the mixture to theprecoated wells and incubated this overnight at roomtemperature. The wells were washed as before andthe bound radioactivity was determined. Under theconditions of the test and in the absence of theunlabelled LDL, 20% of the added radioactivitywas bound. The intraassay coefficient of variationwas 6%.

by guest on November 7, 2015http://atvb.ahajournals.org/Downloaded from

MONOCLONAL ANTIBODIES AGAINST APO B Milne et al. 25

Chemical Modification of LDL

Carbamylation, reductive methylation, and cyclo-hexanedione modification of LDL were carried out asdescribed by Mahley et al5 and Weisgraber et al.18

The extent of modification was estimated from theamino acid analysis.

Cyclohexanedione Treatment

We added 12 mg LDL in 1 ml 0.15 M NaCI, 0.01%EDTA (pH 7) to 2 ml 0.15 M cyclohexanedione in0.2 M sodium borate (pH 8) and incubated this for2 hours at 35°C. The reaction was terminated bydialysis for 24 hours against 0.15 M NaCI, 0.01%EDTA, pH 7.

Carbamylation

We mixed 2 mg LDL in 1 ml 0.15 M NaCI, 0.01%EDTA (pH 7) with 0.5 ml of 0.2 M sodium borate,pH 8. Then we added 30 mg of sodium cyanate andincubated the mixture for 2 hours at 37°C. Thesample was dialyzed for 36 hours against 0.15 MNaCI, 0.01% EDTA, pH 7.

Reductive Methylation

We added 1 mg of sodium borohydride to 2 mgLDL in 1.5 ml 0.2 M sodium borate (pH 8). The reac-tion was started by the addition of 1 /xl of aqueousformaldehyde (37%); additional 1 /il aliquots wereadded at 5-minute intervals. At 30 minutes the reac-tion was terminated by passage on a column of Se-phadex G-25 (Pharmacia Incorporated, Uppsala,Sweden). The fractions containing the LDL werepooled and dialyzed against 0.15 M NaCI 0.01%EDTA (pH 7) for 24 hours.

LDL Binding to Human Flbroblasts

Human skin fibroblasts from a forearm skin biopsyand the fibroblast strain GM 1915 (Human GeneticMutant Cell Repository, Camden, New Jersey) froma homozygous familial hypercholesterolemic (HFH)patient have been characterized previously.17

For binding experiments, cells were grown in 35mm dishes (Costar, Cambridge, Massachusetts)containing 2 ml of DMEM supplemented with 10%(vol/vol) FBS. Before reaching confluency, cellswere rinsed with 1 ml of sterile PBS and incubated for48 hours in DMEM containing 10% (vol/vol) lipopro-tein-deficient serum (Medium A). Before the experi-ment, the cells were cooled at 4°C for 30 minutes,Medium A was replaced by 0.7 ml of cold Medium Asupplemented with HEPES pH 7.4 and 125I-LDL or1Z5l-LDL-Fab complexes and the cells were thenincubated for 4 hours at 4°C on a rotary shaker(30 oscillations per minute).

For preparation of 125l-LDL-Fab complexes,12SI-LDL (2.5 /xg/ml) and Fab fragments were incu-bated overnight at 4°C in Medium A. The washingprocedure and determination of bound 125I-LDL were

performed as described by Innerarity et al.18 Specificbinding was measured by subtracting the amount ofbound 126I-LDL in the presence of 500 /xg/ml ofhuman LDL.4

Determination of the Effect of Antibodies onthe LDL-Medlated Suppression of CholesterolSynthesis

For cholesterol synthesis experiments, cells wereplated in 100 mm Petri dishes (Falcon Labware, Ox-nard, California) in 10 ml DMEM supplemented with5% (vol/vol) pooled human serum and 100 units/mlpenicillin and streptomycin. After reaching 80% to90% confluency, the monolayers were rinsed with5 ml PBS, changed to 10 ml DMEM containing 5%(vol/vol) lipoprotein-deflclent serum (Medium B).17

After 24 hours of incubation at 37°C, Medium B wasremoved and replaced by 5 ml of fresh Medium Bcontaining 5 /imol of 14C-acetate (2 /xCi//imol) (NewEngland Nuclear, Boston, Massachusetts) and ei-ther no LDL, 20 /xg/ml LDL (pool of four subjects), orLDL-Fab complexes. After 24 hours the medium wasremoved, the cells were rinsed twice with 5 ml PBSand were harvested for lipid analysis. For prepara-tion of LDL-Fab complexes, LDL (final concentration140 fj.g/m\) and Fab fragments at various concentra-tions were allowed to react in PBS containing 1 mMEDTA for 24 hours at 4°C. LDL Fab complexes werediluted seven times in Medium B and sterilized byultrafiltration before addition to the cells.

Cholesterol synthesis as measured by 14C-acetateincorporation into cholesterol was determined bydigitonin precipitation19 of nonsaponifiable lipids aspreviously described.17

Protein Determination

Protein was measured according to Lowry et al.20

using bovine serum albumin as a standard.

ResultsCell Fusion

Growth occurred in all wells that were seeded and47 were positive for anti-apo B. These were reclonedby limiting dilution and seven were chosen for de-tailed study based on their titre. Three hybridomassecreted antibody of the IgGi subclass (1D1, 2D8,5E11) and four secreted lgG2a antibody (3A8,3A10,3F5, 4G3). The antibodies showed no reactivity inRIA against apo A-l, apo A-ll, apo C-ll, apo C-lll, orapo E but did react with delipidated denatured apo B.

Cross-Reactivities of Antibodies withChemically Modified LDL

The cross-reactivities of antibodies with chemical-ly modified antigens provides a method for distin-guishing differences in the intramolecular specific-ities of the antibodies.21 To detect differences in the

by guest on November 7, 2015http://atvb.ahajournals.org/Downloaded from

26 ARTERIOSCLEROSIS VOL 3, No 1, JANUARY/FEBRUARY 1983

Table 1.LDL

Antibody

1D12D83A83A103F54G35E11

Measurable Apo B In Chemically Modified

Percentage of starting

Cyclohex-anedione

163±135 ±0.6

105±1397 ±2.952 ±4.6

1 ±0.12108±2.9

Methyl-atlon

8.8 ±0.9NDND

2.6 ±0.22.3 ±0.1

ND3.4 ±0.5

reactivity

Carbamyl-ation

51 ±6.45.9 ±1.050 ±4.055 ±5.046 ±2.37.5 ±0.963 ±2.9

Values are means of triplicate determinations ± 1 so.ND = not detected (< 0.5% of the starting reactivity).

specificities of the anti-LDL monoclonal antibodies,we tested chemically modified LDL for its ability tocompete with 125I-LDL in a solid phase RIA. The re-sults from one experiment expressed as the percent-age of starting reactivity are shown in table 1. Cyclo-hexartedlone treatment of LDL, which in thisexperiment modified 40% of the arginine residues,resulted in a complete loss of reactivity with anti-bodies 2D8 and 4G3 and a 50% loss of reactivity with3F5. In contrast, reactivity was increased with 1D1and unchanged with 3A8, 3A10, and 5E11. Reduc-tive methylation of LDL (60% modification) caused

an almost total loss of reactivity with all the anti-bodies. Carbamylated LDL (12% modification) re-tained about one-half of its starting reactivity with1D1, 3A8,3A10,3F5, and 5E11, whereas more than90% of reactivity with 2D8 and 4G3 was lost. A simi-lar pattern of reactivity was seen in three independ-ent, identical experiments.

Cotitration of Antibodies

Fisher and Brown8 have recently described amethod of cotitration of two monoclonal antibodies todetermine if the two monoclonal antibodies reactwith the same or different antigenic determinants oncell surfaces. We have adapted this technique forsoluble antigens. When a solid phase RIA was car-ried out using increasing dilutions of monoclonalantibodies, the binding curves obtained were char-acterized by a plateau at low dilutions (figure 1). Weobserved in preliminary experiments that the heightof the plateau was a function of the LDL concentra-tion used to coat the plastic (data not shown). Thisindicates that, in the plateau region, the LDL is alimiting reactant, a prerequisite for the cotitrationexperiments.

Figure 1 shows an example of the two types ofresults which were obtained when different anti-bodies were mixed in a 1 to 1 ratio before beingadded to the wells. The combination of 3A10 and

60 r

o 40

eo

Q

OCD 20

* i ft-8 I

8 32 128 2 8RECIPROCAL OF ANTIBODY DILUTION « JO'3

B

32 128

Figure 1. Cotitration of monoclonal antibodies against LDL. Monoclonal antibodies were cotitratedusing a solid phase RIA as described in the Methods section either mixed in a 1 to 1 ratio or separately.A. 2D8 A — A , 3A10 • — • , 2D8 + 3A10 o—o. B. 3A8 A — A , 3A10 • — • , 3A8 + 3A10 o—o.

by guest on November 7, 2015http://atvb.ahajournals.org/Downloaded from

MONOCLONAL ANTIBODIES AGAINST APO B Milne et al. 27

Table 2. Cotttratlon of Monoclonal Antt-LDL Antibodies

Antibody

1D12D83A83A103F54G35E11

1D1

28.8 ± 1.8

2D8

42.8 ±1.233.9 ±1.3

3A8

40.5±0.648.0 + 0.629.7 + 0.7

3A10

41.0 ±0.749.2 ±0.629.6 + 0.528.8 ± 0.9

3F5

46.2 ±0.753.8 ±1.340.9 + 2.140.6 ± 0.930.2 ±2.7

4Q3

52.5 ±0.959.8 ±0.545.5 ±1.445.3 ±1.246.0 ±2.043.5 ±2.6

5E11

45.7+1.446.7 ±1.735.8 + 1.139.7 ±1.251.7+1.153.2 ±0.0329.8 ±0.06

CPM (x 10"3) bound in the plateau region of the curve ± 1 SD.

2D8 (figure 1 A) gave a plateau which was signifi-cantly higher than that obtained with either of theantibodies individually. This would indicate that 3A10and 2D8 react with different determinants. The factthat the binding was less than additive may indicatethat there are certain steric restrictions on the bind-ing, perhaps at the level of the 12SI rabbit antimouseFab. In contrast, when the antibodies 3A8 and 3A10(figure 1 B) were cotitrated, the maximum radioactiv-ity bound was not more than that bound when the twoantibodies were titrated separately. This would indi-cate that the antibodies recognized the same orclosely, spatially associated determinants on theLDL.

Table 2 summarizes the results of one cotitrationexperiment. It is clear that the antibodies 1D1 and2D8 have specificities which are unique. Apparently,3A8 and 3A10 recognize the same determinantwhich is probably close to that recognized by 4G3but distant from the determinant recognized by 3F5.The bound counts obtained with a mixture of 3F5 and4G3 were not significantly higher than those seenwith 4G3 alone. In this experiment, 5E11 seems tohave a specificity different from that of the other anti-bodies. However, in two subsequent experiments,the differences between 5E11, 3A8, and 3A10 werenot significant. Otherwise the results from the threeexperiments were identical.

The Effect of Monoclonal Antibodieson the LDL Pathway

To evaluate the relationship between the determi-nants recognized by the individual monoclonal anti-bodies and the site on apo B recognized by the LDLreceptor, we tested the ability of their Fab fragmentsto block 125I-LDL binding to fibroblasts. The 125I-LDLconcentration (2.5 /xg/ml) was determined to be opti-mal in preliminary experiments. The Fab fragmentswere prepared from an IgG preparation from asciticfluid that contained, in addition to the specific mono-clonal antibody, varying amounts of "nonantibody"immunoglobulin; as papain digestion of mouse IgG isassociated with a loss in antibody activity,22 the con-centrations of Fab for the binding experiments werechosen on the basis of their respective antibody ac-

tivities in an RIA against LDL. The most active frag-ment in RIA, 1D1, Was tested at the equivalent of a3 M excess relative to apo B (assuming molecularweights of 5 x 104 for Fab and 5 x 105 for apo B)and the others, at concentrations corresponding totheir respective activities in RiA relative to that of1D1. Under these conditions, the Fab fragments of3A8, 3A10, 4G3, and 5E11 eliminated virtually allspecific binding of 12Sl-i_DL, whereas 1D1 and 2D8had no effect (figure 2). Fab fragments of 3F5 de-creased binding by about 65%. When the Fab frag-ments were tested at a 10-fold higher concentration(a nominative 30 M excess of active Fab fragmentsrelative to apo B), there was no correspondingchange in their respective abilities to block LDL bind-ing, indicating that the maximum effect had beenreached and that the differences among the anti-bodies were qualitative (figure 2).

To confirm the results of 125I-LDL binding studies,we tested the Fab fragments for their ability to inter-fere with LDL-mediated suppression of cholesterolsynthesis in cultured fibroblasts. Incorporation of14C-acetate into cholesterol was measured in the ab-sence of exogenous LDL or in the presence of LDL orLDL-Fab complexes as described In the Methodssection. The LDL concentration of 20 ptg/ml was de-termined to be optimal in preliminary experiments.17

As in the binding experiments, 1D1 was tested at athree-fold and a 30-fold molar excess relative to LDLand the others, at concentrations proportional to theirrespective activities in RIA relative to 1D1. Whentested at a three-fold molar excess, the Fab frag-ments of 1D1, 2D8, and normal mouse IgG had noeffect on the LDL-suppression of cholesterol synthe-sis (figure 3). In contrast, 3A8, 3A10, 3F5, 4G3, and5E11 clearly reduced LDL-mediated suppression.When the Fab fragments were tested at 10-fold high-er concentrations, there was no corresponding in-crease in their abilities to reduce LDL suppression(data not shown). The presence of 20 /ig/ml of ex-ogenous LDL caused a 30% decrease in cholesterolsynthesis in HFH fibroblasts (data not shown). Thiswould indicate a significant receptor-independentcomponent of LDL uptake under the conditions of theexperiments which could account for the inability ofthe antibodies to totally eliminate the effects of LDLon cholesterol synthesis.

by guest on November 7, 2015http://atvb.ahajournals.org/Downloaded from

28 ARTERIOSCLEROSIS VOL 3, No 1, JANUARY/FEBRUARY 1983

I20

tooeno

GUJQ_

80

60O

40OoLLO

20

1 1

T -r

NO FAB 101 208 3A8 3A10 3F5 4G3 5E1I

Figure 2. The effect of anti-LDL Fab fragments on the binding of 125I-LDL to cultured fibro-biasts. The 125I-LDL was tested at a concentration of 2.5 jug/ml. The protein concentrations ofFab were: 1D1,0.6 /u,g/ml; 2D8,0.8 ju,g/ml; 3A8,3.1 ^g/ml; 3A10,50 /ng/ml; 3F5,2.5 jug/ml; 4G3,0.3 /ug/ml; and 5E11,1.0 /xg/ml. These represent a nominative three-fold molar excess relativeto apo B (open bars). The cross-hatched bars represent the effect of 10-fold higher concentra-tions of Fab. The data are the means of triplicate dishes ± 1 SD.

Zo<Q:Oa.oo2

OOr

oou.o9<

in

50

10 -

I

*̂1 00n

00

en

T

T

O

en

T

m

en

en

J L .

UJm

-X.

•Q

au.c

Figure 3. The effect of Fab fragments on suppression of cholesterol synthesis by LDL. Theresults are expressed as percentages of 14C-acetate incorporation into cholesterol by culturedhuman fibrobiasts in the absence of exogenous LDL. The control suppressed state and all Fabfragments were tested in the presence of 20 /ug/ml of LDL. On the basis of their relative activitiesagainst LDL in a solid phase RIA, the protein concentrations of Fab used were: 1D1, 5 /xg/ml;2D8,9.6 /Lig/ml; 3A8,120 /Ltg/ml; 3A10, 5 /xg/ml; 3F5, 50 ju.g/ml; 4G3,15 /u,g/ml; and 5E11, 5 /utg/ml. Normal mouse Fab (nFab) was tested at a concentration of 50 /xg/ml. The data are themeans of triplicate dishes ± 1 SD. The preparation of Fab fragments is different from that infigure 2.

by guest on November 7, 2015http://atvb.ahajournals.org/Downloaded from

MONOCLONAL ANTIBODIES AGAINST APO B Milne et al. 29

DiscussionWe have characterized seven monoclonal anti-

bodies to human apo B with respect to their individ-ual specificities and their capacities to interfere withthe inhibition of the LDL pathway.

Several antibodies are distinguished by their char-acteristic reactions with chemically modified LDL.Most notable are the antibodies 2D8 and 4G3 whichshow virtually no reactivity with cyclohexanedione-treated LDL. This would suggest that an arginineresidue may be present in their respective antigenicdeterminants or, alternatively, that the conformationof the determinants are altered by modification ofarginine residues elsewhere in the molecule. Thedeterminants recognized by 2D8 and 4G3 also ap-pear more susceptible to carbamylation; a reactionwhich mainly modifies lysine residues. These experi-ments were designed solely to differentiate amongthe intramolecular specificities of the antibodies. Anyextrapolation of these results to the observation thatmodification of arginine and lysine residues of LDLeliminated binding to the LDL receptor5 is not justi-fied at this point.

The cotjtration experiments allowed a further dif-ferentiation of the antibody specificities. We proposethat the results in table 1 represent three basic typesof relationships between individual antibodies. Thefirst type is that in which the two antibodies react withdeterminants that are far enough apart so that thetwo antibodies can bind to the same molecule ofantigen with little mutual interference. An example ofthis would be the relationship of 2D8 and 3A10. Atthe other extreme is the case in which the two anti-bodies bind to the same antigenic determinant, aswould appear to occur with 3A8 and 3A10. The thirdtype of relationship would be that in which the anti-bodies recognize different antigenic determinantswhich are either overlapping or close enough togeth-er to prevent their mutual binding to the same mole-cule of antigen. We propose that the relationship of4G3 to 3A8/3A10 and to 3F5 represents an exampleof this. While 4G3 cannot be distinguished from ei-ther 3A8/3A10 or 3F5, it is clear that 3A8/3A10 and3F5 bind to different determinants. Thus, the deter-minant recognized by 4G3 may lie between thoserecognized by 3A8/3A10 and 3F5 respectively. Inter-estingly, myoglobin, a molecule of 153 amino acidresidues which possesses five antigenic determi-nants, can fix a maximum of four of the correspond-ing antibodies at equivalence.21

With the results in table 1 we have constructed atheoretical "map" of the antigenic determinants asthey may be present on apo B (figure 4). Since 2D8and 1D1 have specificities different from each otherand from all the other antibodies, their placement isarbitrary. As antigenic determinants are a function ofthe three-dimensional structure on the antigen, fig-ure 4 is proposed as a representation of the spatialrelationship of the determinants and not necessarilyof their positions in the primary amino acid se-

2D8

LDl RECEPTOR RECOGNITION SITE

I I3A8

3F5 4G3 3AK) 5EH IDf

Figure 4. The proposed spatial relationships of antigenicdeterminants on apo B recognized by a series of anti-LDLmonoclonal antibodies. The relative positions of the deter-minants (identified in the figure by their corresponding anti-bodies) was predicted from cotitration experiments. Theplacement of the determinants 1D1 and 2D8 was arbitrary.The tentative identification of the region recognized by theLDL receptor was based on the ability of Fab fragments tointerfere with the LDL receptor pathway.

quence. Atassi23 has demonstrated that in lysozyme,spatially adjacent residues that are far apart in thepeptide chain can together form an antigenic deter-minant. Moreover, antigenic determinants tend to beat the surface of the molecule and to occur at bendsin the three-dimensional structure.21 Recently, Conti-Tronconi et al24 have used monoclonal antibodies topartially map immunochemically an acetylcholine re-ceptor. They have predicted the relative positions ofantigenic determinants from estimations of the com-position of antigen-antibody complexes.

When tested in molar excess, Fab fragments ofseveral antibodies were capable both of blockingbinding of 126I-LDL to the LDL receptor and of inter-fering with the inhibition by LDL of cholesterol syn-thesis in cultured fibroblasts. These effects wereindependent of the Fab concentrations from a three-fold to a 30-fold molar excess relative to apo B whichdemonstrate saturation by the antibodies. There isan excellent correlation between the effects of theindividual antibodies as shown at two sequentialsteps in the LDL pathway; this would indicate that theantibodies act by blocking receptor-mediated LDLbinding and the subsequent metabolic events in thepathway. The determinants that were grouped to-gether by the cotitration experiments (3F5, 4G3,3A8, 3A10, and 5E11; see figure 4) are also thosethat are capable of interference with the LDL path-way. This would suggest that the site on apo B recog-nized by the LDL receptor is in this region of themolecule. The fact that 3F5 is intermediate in itsability to block both LDL binding and LDL-mediatedsuppression of cholesterol synthesis compared tothe other antibodies may be a reflection of its relativedistance from the actual binding site.

Due to its insolubility and its large size, apo B hasremained poorly characterized. With the aid ofmonoclonal antibodies against LDL, we have suc-ceeded in differentiating a series of distinct antigenicdeterminants on apo B and have presented a tenta-tive map of these determinants as they may appearin LDL. A number of the determinants appear to besituated in the region of apo B recognized by the LDL

by guest on November 7, 2015http://atvb.ahajournals.org/Downloaded from

30 ARTERIOSCLEROSIS VOL 3, No 1, JANUARY/FEBRUARY 1983

receptor. This theoretical map is designed to serveas a working model and will no doubt be changedand expanded as other monoclonal antibodies arecharacterized and as the reactivities of the anti-bodies with proteolytic fragments of apo B are deter-mined. Monoclonal antibodies offer a promising newapproach to the mapping of functional domainson LDL and to the determination of the structure ofapo B.

AcknowledgmentsWe are indebted to Philip Weech, Sonia Goldstein, John Chap-

man, and Thomas Innerarity for valuable discussion and MalurSairam and Karl Weisgraber for help with amino acid analyses.We thank Louise Blanchette and Camilla Vezina for technicalassistance and Louise Lalonde for secretarial help.

References1. Lee DM, Valente AJ, Kuo WH, Maeda H. Properties of apoli-

poprotein B in urea and in aqueous buffers. The use of glu-tathione and nitrogen in its solubilization. Biochim BiophysActa 1981;666:133-146

2. Steele JCH, Reynolds JA. Molecular weight and hydrodyna-mic properties of apolipoprotein B in guanidine hydrochlorideand sodium dodecyl sulphate solutions. J Biol Chem 1979;254:1639-1643

3. Kane JP, Hardman DA, Paulus HE. Heterogeneity of apoli-poprotein B: Isolation of a new species from human chylomi-crons. Proc Natl Acad Sci USA 1980;77:2465-2469

4. Brown MS, Goldstein JL. Familial hypercholesterolemia:Defective binding of lipoproteins to cultured fibroblasts asso-ciated with impaired regulation of 3- hydroxy -3- methylglu-taryl coenzyme A reductase activity. Proc Natl Acad Sci USA1974;71:788-792

5. Mahley RW, Innerarity TL, Pitas RE, Weisgraber KH,Brown JH, Gross E. Inhibition of lipoprotein binding to cellsurface receptors of fibroblasts following selective modifica-tion of arginyl residues in arginine rich and B apoproteins. JBiol Chem 1977;252:7279-7287

6. Goldstein JL, Brown MS. The low-density lipoprotein path-way and its relation to atherosclerosis. Ann Rev Biochem1977;46:897-930

7. Brown MS, Kovanen PT, Goldstein JL. Regulation of plas-ma cholesterol by lipoprotein receptors. Science 1981; 212:628-635

8. Fisher AG, Brown G. A rapid method for determining wheth-er monoclonal antibodies react with the same or differentantigens on the cell surface. J Immunol Methods 1980;39:377-385

9. Havel RJ, Eder HA, Bragdon JH. The distribution andchemical composition of ultracentrifugally separated lipopro-teins in human serum. J Clin Invest 1955;34:1345-1355

10. Milne RW, Douste-Blazy Ph, Retegul L, Marcel YL. Char-acterization of monoclonal antibodies against human apolipo-protein E. J Clin Invest 1981 ;68:111-117

11. Chalon M-P, Milne RW, Vaerman J-P. In vitro immuno-suppressive effect of serum from orally immunized mice. EurJ Immunol 1979;9:747-751

12. Ey PL, Prowse SJ, Jenkln CR. Isolation of pure lgG1, lgG2aand lgG2b immunoglobulin from mouse serum using protein-A Sepharose. Immunochemistry 1978; 15:429-436

13. Gorlnl G, Medgysl GA, Doria G. Structural localization ofisotypic markers on mouse myeloma IgG globulins. J Im-munol 1969;103:1125-1131

14. Mellman IS, Unkeless JC. Purification of a functional mouseFc receptor through the use of a monoclonal antibody. J ExpMed 1980;152:1048-1069

15. Bllhelmer DW, Eisenberg S, Levy Rl. The metabolism ofvery low density lipoproteins. 1. Preliminary in vitro and in vivoobservations. Biochim Biophys Acta 1972;260:212-221

16. Weisgraber KH, Innerarity TL, Mahley RW. Role of lysineresidues of plasma lipoproteins in high affinity binding to cellsurface receptors on human fibroblasts. J Biol Chem 1978;253:9053-9062

17. Verdery RB, Theolis R. Regulation of sphingomyelin longchain base synthesis in human fibroblasts in culture: Role oflipoproteins and the low density lipoprotein receptor. J BiolChem 1982;257:1412-1417

18. Innerarity TL, Pitas RE, Mahley RW. Receptor binding ofcholesterol-induced high-density lipoproteins containing pre-dominantly apoprotein E to cultured fibroblasts with muta-tions of the low-density lipoprotein receptor locus. Biochemis-try 1980;19:4359-4365

19. Kandutsh AA, Saucier SE. Prevention of cyclic and triton-induced increases in hydroxymethylglutaryl coenzyme A re-ductase and sterol synthesis by puromycin. J Biol Chem1969;244:2299-2305

20. Lowry OH, Rosebourgh NJ, Farr AL, Randall RJ. Proteinmeasurement with the folin phenol reagent. J Biol Chem1951;193:265-275

21. Atassl MZ. Antigenic structure of myoglobin: The completeimmunochemical anatomy of a protein and conclusions relat-ing to antigenic structures of proteins. Immunochemistry1975;12:423-438

22. Retegul LA, Milne RW, Camblaso CL, Masson PL. Therecognition by monoclonal antibodies of various portions of amajor antigenic site of human growth hormone. Mol Immunol1982;19:865-875

23. Atassl MZ. Precise determination of the entire antigenicstructure of lysozyme: Molecular features of protein antigenicstructures and potential of "surface simulation" synthesis —A powerful new concept for protein binding sites. Immuno-chemistry 1978:15:909-936

24. Conti-Tronconl B, Tzartos S, Llndstrom J. Monoclonalantibodies as probes of acetylcholine receptor structure. 2.Binding to native receptor. Biochemistry 1981 ;20:2181-2191

Index Terms: monoclonal antibodies • cholesterol synthesis • low density lipoprotein • apolipoprotein B

by guest on November 7, 2015http://atvb.ahajournals.org/Downloaded from

R W Milne, R Theolis, Jr, R B Verdery and Y L MarcelCharacterization of monoclonal antibodies against human low density lipoprotein.

Print ISSN: 1079-5642. Online ISSN: 1524-4636 Copyright © 1983 American Heart Association, Inc. All rights reserved.

Greenville Avenue, Dallas, TX 75231is published by the American Heart Association, 7272Arteriosclerosis, Thrombosis, and Vascular Biology

doi: 10.1161/01.ATV.3.1.231983;3:23-30Arterioscler Thromb Vasc Biol. 

http://atvb.ahajournals.org/content/3/1/23the World Wide Web at:

The online version of this article, along with updated information and services, is located on

  http://atvb.ahajournals.org//subscriptions/

online at: isArteriosclerosis, Thrombosis, and Vascular Biology Information about subscribing to Subscriptions:

  http://www.lww.com/reprints

Information about reprints can be found online at: Reprints: 

document. Question and AnswerPermissions and Rightspage under Services. Further information about this process is available in the

which permission is being requested is located, click Request Permissions in the middle column of the Webthe Copyright Clearance Center, not the Editorial Office. Once the online version of the published article for

can be obtained via RightsLink, a service ofArteriosclerosis, Thrombosis, and Vascular Biologypublished in Requests for permissions to reproduce figures, tables, or portions of articles originallyPermissions:

by guest on November 7, 2015http://atvb.ahajournals.org/Downloaded from