Characterization of Monoclonal Antibodies against

HumanApolipoprotein ER. W. MILNE, PH. DOUSTE-BLAZY, and Y. L. MARCEL,Laboratory of Lipoprotein

Metabolism, Clinical Research Institute of Montreal, Montreal H2W11R7,Quebec, Canada

L. RETEGUI, Unite de Medecine Experimentale, International Institute ofCellular Pathology, Brussels, Belgium

A B S T RA C T From a single cell fusion, five stablehybridomas secreting antiapolipoprotein E (apo E)were obtained. The immunoglobulin (Ig)G subclassescontaining the respective monoclonal antibodies wereisolated and were used as the antibody component ina solid-phase radioimmunoassay. The binding of 1251_apo E to the insolubilized antibody was inhibited byunlabeled apo E but not by unlabeled apoproteins A-I,A-II, C-II, and C-III, or by low density lipoproteinimmunodepleted of endogenous apo E. Competitioncurves were obtained with lipoprotein subfractions thathad the same shape as those obtained with purifiedapo E. Apo E levels in normal and hyperlipidemicplasma were well correlated when measured by thefive monoclonal antibodies and polyclonal anti-apoE, although differences in absolute values were ob-served. In normal subjects 34, 10, 20, and 36%of apo Ewas recovered in the very low density lipoprotein,low density lipoprotein, high density lipoprotein,and the d > 1.21-g/ml fractions, respectively, whereasthese values were 34, 7, 12, and 47%, respectively,in type III patients. All antibodies indicated the samesubfraction distribution of apo E. The monoclonalantibodies reacted with all of the isomorphs of apo E.One of the antibodies could be clearly distinguished byits reactivity with chemically modified very low den-sity lipoprotein.

INTRODUCTIONApolipoprotein E (apo E)' is the most actively syn-thetized apolipoprotein in the rat and appears to be of

Dr. Retegui is Fellow of the Consejo Nacional de Investi-gaciones Cientificas y T6cnicas de la Republica Argentina.

Received for publication 3 November 1980 and in revisedform 3 March 1981.

'Abbreviations used in this paper: apo E, B, etc., apolipo-protein E, B, etc.; FBS, fetal bovine serum; HBSS, Hanks'balanced salt solution; HDL, high density lipoprotein; LDL,low density lipoprotein; RIA, radioimmunoassay; VLDL, verylow density lipoprotein.

exclusively hepatic origin (1-3). In humans as in swine,apo B and apo E are respectively responsible forthe binding of low density lipoprotein (LDL) and of aminor high density lipoprotein (HDL) subclass to ahigh-affinity cell-surface receptor on fibroblasts andother cell types (4, 5). The apo E and apo B are recog-nized by the same receptor and, in both cases, bindingis abolished if the lysine or arginine residues of theapoproteins are chemically modified (6). It has beensuggested that structurally similar positively chargedregions on the two apoproteins serve as the recog-nition site for the receptor (7). In addition, apo Eappears to be responsible for the hepatic uptake of asubfraction of homologous (8) and heterologous (9)HDL from rat plasma.

Monoclonal antibodies to apoproteins provide a newapproach to the study of the specificity of the inter-action of lipoproteins and cell surface receptors. Theadvantage of monoclonal antibodies lies in theirspecificity for a single antigenic determinant. It ispossible that with a battery of monoclonal anti-bodies to an apoprotein one could immunochemicallydefine regions of the apoprotein involved in bindingto the cell surface receptor.

Wereport the production and partial characterizationof five monoclonal antibodies against human apo E.

METHODS

Preparation of plasma and plasma lipoproteins. Bloodfrom fasted normal subjects or from fasted type IIA, typeIII, or type IV hyperlipidemic patients of the lipid clinicof the Clinical Research Institute of Montreal was collectedinto EDTA.

Lipoprotein subfractions were prepared in a BeckmanL5-65 ultracentrifuge with a 50.2 Ti rotor (Beckman In-struments, Inc., Spinco Div., Palo Alto, Calif.). The verylow density lipoprotein (VLDL), LDL, and HDL were iso-lated at densities of 1.006, 1.006-1.063, and 1.063-1.21 g/ml,respectively. The infranate, d > 1.21 g/ml, was also retainedfor analysis.

111J. Clin. Invest. © The American Society for Clinical Investigation, Inc. * 0021-9738/81/07/0111/07 $1.00Volume 68 July 1981 111-117

Purification of apo E. The VLDL from normal fastedsubjects was isolated as described above and delipidated withethanol/diethylether (3:2) at a ratio of 25 vol of solvent/volof VLDL. The VLDL precipitate was then subjected to twofurther extractions in diethylether. The moist apolipoproteinwas redissolved in a buffer consisting of 2 mMphosphate,5 Murea, and 10 mMdithiothreitol (pH 7.4). The sample wasapplied to a column of heparin (Upjohn Co., Kalamazoo,Mich.) bound to Sepharose 4B (10), which had been pre-equilibrated with 2 mMNaHPO4 and 5 M urea (pH 7.4).After extensive washing with the same buffer, the boundprotein was eluted with 2 mMNa phosphate buffer (pH 7.4),5 Murea, and 1 MNaCl. The retained fraction was dialyzedagainst 2 mMNa phosphate buffer (pH 7.4), made up to 5 Murea, and reapplied to the heparin-Sepharose column, whichwas conditioned as above. The bound protein was elutedwith a linear gradient 50-300 mMNaCl in 2 mMNa phos-phate buffer (pH 7.4) and 5 Murea. Apo E was eluted between200 and 230 mMNaCl and its puirity checked by alkalinepolyacrylamide gel electrophoresis, by sodium dodecylsulfate-polyacrylamide gel electrophoresis, and by iso-electric focusing on polyacrylamide gel (11). The isolatedapo E did not cross-react with antisera against other apo-proteins.

The major polymorphic species of apo E were isolatedby preparative isoelectrofocusing (pI) of apo VLDL, asdescribed previously (11). The VLDL was prepared from aplasma pool from normal donors. Thus obtained, apo El, E2,E3, E4 had apparent pl of 5.60, 5.65, 5.72, and 5.80, respec-tively; their purity was assessed by analytical isoelectricfocusing and by sodium dodecyl sulfate-polyacrylamide gelelectrophoresis.

Cell lines. The azaquanine-resistant plasmacytoma cellline SP2-0 (12) obtained from Dr. Rolf Kemmler (InstitutPasteur, Paris) was recloned and maintained in an atmos-phere of 8% CO2 in Dulbecco's Modified Eagle's mediumsupplemented with 10% fetal bovine serum (FBS), 15 mMHepes, 50 ,m 2-mercaptoethanol, penicillin-streptomycin,and 1.5,ug/ml azaquanine.

Immunization of mice. BALB/c mice, whose spleens wereto be used for cell fusion were given a primary subcutaneousinjection of 50 ,ug of apo E in complete Freund's adjuvant.After 3 wk, the mice were given a second subcutaneousinjection of 50 ,ug apo E in complete Freund's adjuvant.1 wk later and 3 d before the cell fusion, 2 Ag of apo E wasinjected intravenously.

BALB/c mice were also immunized intraperitonally, ac-cording to the protocol of Tung et al. (13), to induce ascitescontaining anti-apo E antibody.

Cellfusion. The technique used for cell fusion was similarto that described by Kennet et al. (14). A cell suspension inHanks' balanced salt solution (HBSS) was prepared by teasingapart the spleens of two appropriately immunized mice. Thecells were filtered through nylon cloth and centrifuged, andthe erythrocytes lysed by osmotic shock. The cells wereagain centrifuged, filtered through nylon cloth, and washedin HBSS. The SP2-0 cells were washed twice in HBSS,mixed with the spleen cells at a ratio of 1:10, and againwashed in HBSS. The pellet was resuspended in 0.2 ml of 30%polyethylene glycol 1000 (J. T. Baker Chemical Co., Phillips-burg, N. J.), gently swirled for 2 min, and centrifuged for6 min at 200 g. 5 ml of HBSSwas added and the cells gentlyswirled for 3 min followed by centrifugation. 5 ml ofDulbecco's modified Eagle's medium containing 15 mMHepes, 40 ,M 2-mercaptoethanol, 30% FBS and penicillin-streptomycin was added. The pellet was left undisturbed for5 min before gentle resuspension of the cells. The cells werediluted to 20 ml with medium and distributed in two petri

plates. After 24 h the cells were centrifuged and resus-pended in the same medium, which in addition contained0.11 mMhypoxanthine, 0.38 ,uM aminopterine, and 16 ,uMthymidine (HAT) (15). The cells were then distributed in thewells of 11, 96-well microculture plates (Falcon Labware,Div. of Becton, Dickinson & Co., Oxnard, Calif.). On 3 suc-cessive d, 50 ,ul of HAT-containing medium was added toeach well. When sufficient growth had occurred, normally at10-15 d, the supemate were screened for the presence ofanti-apo E antibody.

Screening of supernates. Aliquots of 200 ,u of a solution ofapo E (10 ,ug/ml in 5 mMglycine, pH 9.2) were distributedin Removawells (Dynatech Laboratories, Inc., DynatechCorp., Alexandria, Va.) and left overnight. The followingday the wells were washed in 0.15 MNaCl containing 0.25%Tween 20. An aliquot of hybridoma supernate diluted to 200,ul in phosphate-buffered saline containing 5%FBS was addedto each well and incubated for 3 h at 370C. The wells werewashed in 0.15 MNaCl/0.025% Tween and 200 ,pl of purified125I-anti-mouse Fab, diluted in phosphate-buffered salinecontaining 5%FBS was added and incubated for a further 3 h.The wells were washed and the bound radioactivity deter-mined. To identify the immunoglobulin class and subclass ofthe antibody produced by positive clones, a similar radio-immunoassay was performed but with purified 1251-labeledantibody specific to an individual mouse immunoglobulinclass or subclass substituted for the 125I-anti-Fab. The prepa-ration of the anti-mouse immunoglobulin reagents has beendescribed previously (16).

Maintenance of hybridomas in culture and as ascitictumors. Once positive clones were identified, the cells weretransferred to larger culture vessels and maintained in HAT-containing medium. The concentration of FBS was graduallyreduced from 30 to 10%.

Lightly irradiated (200-500 R) BALB/c mice were injectedintraperitoneally with 5 x 106 cultured hybridoma cells.When sufficient ascitic fluid had accumulated, the mice werekilled and the ascites collected. The ascitic cells were usedto passage the tumor and the fluid was used as a source ofmonoclonal anti-apo E antibody.

Isolation of anti-apo E immunoglobulin. The IgG sub-class containing the anti-apo E antibody was isolated fromthe ascitic fluid of hybridoma-bearing mice by elution fromprotein-A Sepharose 4B (Pharmacia Inc., Uppsala, Sweden)with a stepwise pH gradient (17).

Iodination. 5 ,ug apo E was labeled with 1 mCi of 125I bythe technique of Bolton and Hunter (18). The prelabeledBolton-Hunter reagent was purchased from Amersham Corp.(Arlington Heights, Ill.). The iodinated protein was isolatedby gel filtration on Sephadex G-25. Greater than 90% of theradioactivity was precipitated by trichloroacetic acid and75 to 80% was bound by mouse antihuman apo E. Thespecific activity was -7 ,uCi/,ug.

Labeled apo E was stored at 4°C in phosphate-butfferedsaline containing 5%FBS and used within 14 d. If kept longerthere was a decrease in immunoprecipitable apo E.

Immunoglobulins were labeled by chloramine T (19) andthe free 125I was removed by passage on an anion-exchangeresin AGI x 8 (Bio-Rad Laboratories, Richmond, Calif.).

Radioimmunoassay. The monoclonal antibody IgG frac-tion (see above) was diluted in 5 mMglycine (pH 9.2), and 200-,ul aliquots were distributed in Removawells and leftovernight at room temperature. The dilution of antibody waschosen on the basis of preliminary experiments and theconcentration ranged from 5 to 10 ,g/ml protein depending onthe individual clone. The wells were washed in 0.15 MNaCl/0.025% Tween. The sample to be tested was ap-propriately diluted in phosphate-buffered saline containing

112 R. W. Milne, Ph. Douste-Blazy, L. Retegui, and Y. L. Marcel

5% FBS and mixed with 1251-apo E; the final concentration of1251-apo E being -5 ng/ml. An aliquiot of the mixture (200 ul)was transferred to the precoated wells and incubatedovernight at room temperature. The wells were washed andthe bound radioactivity determined. Under the conditions ofthe test and in the absence of unlabeled apo E -30% of theadded radioactivity was bound. Nonspecific binding (withnormal mouse IgG fixed to the plastic) accounted for <10% ofthe maximum specific binding. Results are expressed as B/Bowhere B = 1251-apo E counts per minute bound- 125I-apo Enonspecifically bound, and Bo= 1251-apo E bound in theabsence of unlabeled apo E- 125-apo E nonspecificallybound.

Chemical modification of VLDL. Protein modification re-actions were adapted from methods described by Mahley et al.(7) and Weisgraber et al. (6). Carbamylation: to 0.5 mgVLDLprotein in 300 ,1 of 0.2 Mborate buffer. (pH 8.8) was added 20mgpotassiuim cyanate. The sample was incubated for 20 h at37°C and the reaction terminated by the addition of 100,ul of FBS.

Redtuctive methylation: to 0.5 mgVLDL in 300 ,ul of 0.2 Msoditum borate (pH 8.8) was added 25 ,ul of a 40-mg/ml solutionof soditum borohybride. The reaction was started by theaddition of 1 pl of aquieouis formaldehyde (37%) and additional1-1I aliquiots were added at 5-min intervals. At 30 min thereaction was either terminated by the addition of 100 ,ul FBS ora flirther aliquiot of soditum borohybride solution was addedand the sequence of formaldehyde additions repeated beforetermnination of the reaction at 60 min with 100 p.l of FBS.

Cyclohexanedione treatment: to 0.5 mg VLDL in 100 ,ul0.15 M NaCl. was added 200 al 0.15 M cyclohexanedionedissolved in 0.2 M borate (pH 8.1). The samples were in-cubated at 37°C for periods from 15 to 120 min. The reactionwas terminated by the addition of 100 pul FBS. The extent ofmodification was not determined.

RESULTS

Cell fusion and monoclonal antibodies. Growthwas observed in 130 wells (12%), which were randomlydistributed over the 11 microculture plates. Eightclones were identified by radioimmunoassay thatsecreted anti-apo E antibody. Two of the clones failedto grow when transferred to larger culture vessels andcould not be saved. Four of the surviving clonesproduced antibody of the IgG, class (1D7, 3B7, 5D9,6C5), and two produced IgG2b antibody (6H7, 7C9).One clone, 5D9, has subsequently stopped secretingantibody.

Cells of each of the positive clones were injectedintraperitoneally into syngenetic lightly irradiated miceand the resulting ascites collected. The IgG subclassthat contained the respective monoclonal antibodywas isolated by Staphylococcus protein A-Sepharose,as described in Methods. The agarose gel isoelectro-focusing patterns of the isolated immunoglobulinfractions are shown in Fig. 1.

Radioimmunoassay of apo E. Each of the mono-clonal antibodies and mouse anti-apo E have beenused as the antibody component in a solid-phaseradioimmunoassay (RIA). Purified human apo E, apoAI, apo AII, apo CII, apo CIII as well as LDL im-

+

.U

'1_ 4

1 2 3 4 5 6

FIGURE 1 Isoelectrofocusing in agarose of 1D7 (1), 3B7 (2),6C5 (3), 6H7 (4), 7C9 (5), and polyclonal mouse antihumanapo E (6). The IgG subclasses that contained the monoclonalantibody had been isolated from the ascitic fluid of hybridoma-bearing mice by elution from protein-A Sepharose. In thecase of the polyclonal anti-apo E, the total IgG fraction wasisolated on protein-A sepharose from the ascitic fluid ofmice immunized with apo E.

munodepleted of endogenous apo E were tested fortheir ability to compete with 125I-apo E. Typicalcompetition curves are shown in Fig. 2. The lowerlimit of detection of purified apo E was in the rangeof 5 to 15 ng protein, dependent upon the individualantibody. Complete inhibition occurred with > 100-500ng apo E, again dependent upon the antibody. At highconcentration, apo AII, apo CII, and apo CIII com-peted with 125I-apo E, which would indicate a probableminor contamination (<1%) of these apoprotein prep-arations with apo E. Porcine and African green monkeyplasma, even at low dilutions, competed poorly with125I-human apo E for the monoclonal and polyclonalantihuman apo E antibody (not shown).

Whendilutions of human plasma or lipoprotein sub-fractions were tested in the RIA, the competition curveswere parallel to that of purified apo E (Fig. 3). Al-though delipidation, heating the sample at 45°C for30 min, or treatment with 8 Murea caused a decreasein measurable apo E in plasma, the shape of the curvewas unchanged (not shown).

The apo E concentration in the plasma of six normal,10 type II, 10 type IV, and 2 type III subjects wasdetermined by RIA using four monoclonal and onepolyclonal mouse antihuman apo E IgG. The meanand standard deviations are given in Table I. Thecorrelation coefficients between determinations madeby different antibodies ranged from 0.93 to 0.99. Inaddition, lipoprotein subfractions of the normal andtype III plasmas were prepared and the apo E concen-trations determined. RIA using different monoclonalantibodies indicated an essentially identical dis-

Monoclonal Antibodies against Apolipoprotein E 113

PROTEIN (ng)

FIGURE 2 Dilutions of purified apo E (-), apo AI (x), apo AII (0), apo CII (0), apo CIII (T),and LDL previously passed over insolubilized mouse antihuman apo E (A) were tested in a RIA fortheir ability to compete with 1251-apo E. The five monoclonal anti-apo E antibodies or polyclonalanti-apo E IgG were used as the antibody component.

PLASMA (nl)

- 6H7

l0 1O2 103LIPOPROTEIN (,g PROTEIN) or APOE (ng PROTEIN)

FIGURE 3 Dilutions of apo E (0), HDL (0), LDL (A), VLDL (A), and plasma (x), were

tested for their ability to compete with '251-human apo E. The five monoclonal antibodies or poly-clonal anti-apo E antibodies were used as the antibody component.

114 R. W. Milne, Ph. Douste-Blazy, L. Retegui, and Y. L. Marcel

1 ID7

0

TABLE IApo E Concentrations in Plasma as Determined in RIA by Means of Different Antibodies

Apo E

Antibody

Polyclonal3B7 6C5 6H7 7C9 anti-apo E

,ugIml

Normal (n =6) 54+11 58+11 54+10 46+9 69+23Type II (n = 10) 60+16 54+17 50+15 49±18 63±17Type III (n = 1) 600 600 520 480 800Type III (n = 1) 250 260 220 200 360Type IV (n = 10) 75±25 58+39 62±33 61±+33 78+33

tribution of apo E. In normal subjects 34, 10, 20, and36% were found in the VLDL, LDL, HDL, and d> 1.21 fractions, respectively, whereas the values were34, 7, 12, and 47% in type III patients.

The within-assay coefficient of variation was 6%when 10 replicate plasma samples were included in theassay. This value represents the mean of three experi-ments. The between-assay coefficient of variation was9% when the same plasma sample was measured 15times in duplicate over a 2-mo period.

The assay for apo E was validated by a series ofparallel determinations of apo E levels by RIA and bydensitometric scanning of apo VLDL analyzed bysodium dodecyl sulfate-polyacrylamide gel electro-phoresis. Apo VLDL from four hyperlipoproteinemicpatients were electrophoresed each at three differentprotein levels, and a linear standard curve wasdrawn from a series of purified apo E standards,electrophoresed, stained, and scanned in the sameconditions. The correlation between the apo E levelsmeasured by RIA and by densitometric scanning was0.999, whereas the RIA-determined levels were 85%ofthose measured by scanning.

Reactivity of monoclonal antibodies with isomorphsof apo E. The isomorphs of apo E, isolated bypreparative isoelectrofocusing, were verified by ana-lytical isoelectrofocusing, which showed that the apoE2 fraction was slightly contaminated with apo E3,and that there was a contamination of the apo E, frac-tion with apo AI. In the latter case the degree ofcontamination was estimated by scanning the analyti-cal gels after staining with Coomassie Blue G250 andthe apo E concentration was appropriately correctedfor this contamination (20).

The individual isomorphs were tested in the RIA fortheir ability to compete with 125I-apo E (Fig. 4). Theantibodies recognize all of the isomorphs. With certainantibodies small differences in reactivities of in-dividual isomorphs were seen as indicated by relativeshifts of the competition curves. Although small, thesedifferences were reproducible.

Reactivity of antibodies with chemically modifiedVLDL. In an attempt to demonstrate that the dif-ferent monoclonal antibodies recognize distinct anti-genic determinants on apo E, the antibodies weretested for their ability to recognize apo E in chemi-cally-modified VLDL. Lysine residues of VLDL weremodified by carbamylation or by reductive methyla-tion, and the modified lipoprotein was tested for itsability to compete in the RIA with 125I-apo-E. The

1.0.

0.8

0.6

0.4-

0.2 .

1.0

0.8

0.6

0.4

0.2

1.0

0.8

0.6

0.4

0.2

ANTI -APO E

3B7

1D7

I I~a

10 102 103 10 102 103APO E (ng PROTEIN)

FIGURE 4 Dilutions of the isomorphs apo E1 (@), apo E2 (x),apo E3 (0), and apo E4 (0) isolated by isoelectrofocusing weretested for their ability to complete with 1251-human apo E. Thefive monoclonal antibodies or polyclonal anti-apo E were usedas the antibody component.

Monoclonal Antibodies against Apolipoprotein E

a

1-

115

TABLE II

Measurable Apo E in Chemically Modified VLDL

Percentage of starting reactivity

Carbamylated Methylated

20 h 30 min 60 min

Experiment Experiment Experiment Experiment Experiment ExperimentAntibody a b a b a b

1D7 42+2 13+4 15+5 8±4 13+3 8+43B7 126+5 166+50 36± 19 38±3 13+4 35+26C5 133+5 186±+11 31+2 37+4 33+6 33±26H7 155+8 158+5 55+5 32±2 50±2 35±37C9 123± 12 236+23 40±4 26+4 37±6 26±4Anti-apo E 89±17 74±16 30±4 14±3 25±4 11±2

residual measurable apo E in modified VLDL ex-pressed as a percentage of the initial concentrationin unmodified VLDL is shown in Table II. Carbamyla-tion of VLDL resulted in an increase in measurableapo E as determined by four of the monoclonal anti-bodies (3B7, 6C5, 6H7, 7C9). On the other hand, thesame treatment caused a large decrease in apo E re-activity with the antibody 1D7. Carbamylation ofVLDL resulted in a small decrease in the apo Edetermined by polyclonal anti-apo E. Reductivemethylation of VLDL decreased the apo E, asmeasured by all of the antibodies. The recognitionby the antibody 1D7 again appeared to be moresusceptible to modification of lysine residues.

When arginine residues of VLDL were modifiedby treatment with cyclohexanedione, the increase inincubation times from 15 to 120 min caused a pro-gressive decrease in measurable apo E content, asdetermined by all of the antibodies (results not shown).

DISCUSSIONFrom a single cell fusion experiment, five stablehybridomas were obtained that secrete antibodiesagainst human apo E. The antibodies show little re-activity against other apoproteins and react with apo Ewhen it is present in intact lipoproteins. Determina-tion of apo E concentrations in plasma by a solid-phase RIA gave values that are similar to those recentlyreported by others (21, 22). Although there was goodcorrelation between the values obtained with in-dividual antibodies, there were differences in theabsolute values measured. This may indicate that whenapo E is present in lipoproteins certain antigenicdeterminants are partially inaccessible to antibody, orthat their conformation is slightly altered. There issome precedent for this idea, as, for example, themajority of antigenic determinants on apo Al arehidden or altered in HDLwhen measured by RIA (23).Nevertheless, methods used to expose determinants of

apo Al (delipidation, urea, etc.) had little effect inthe present study. Furthermore, the same differencesamong the antibodies in absolute apo E determinationswere also observed in lipoproteins isolated by ultra-centrifugation, including the d > 1.21 g/ml infranate,which is relatively poor in lipid. Thus, the apparentlydecreased recognition by certain monoclonal anti-bodies of apo E incorporated in lipoproteins appearsto be independent of the lipid/protein ratio and thesize of the particle. Nevertheless, even relativelysmall amounts of lipid, as might be present in the d> 1.21 infranate, may interfere with the antibody-antigen interaction either by steric hindrance or byslightly altering apo E conformation. Alternatively,certain antibodies may be directed against deter-minants which had been altered during the purificationof apo E on heparin-Sepharose, and their affinity maythus be higher for these altered forms than for the nativedeterminants. On the other hand, certain antibodiesagainst native determinants may poorly recognizealtered determinants of the apo E standard, and thusoverestimate apo E concentrations.

All of the antibodies recognized all of the isomorphsof apo E. Havel et al. (22) have recently shown im-munological identity among the apo E isomorphs bydouble immunodiffusion and identical reactivity inthe radioimmunoassay. With some of the antibodies,however, we did observe small relative shifts of thecompetition curves, which could indicate small dif-ferences in the relative affinities of the antibodiesfor the individual isomorphic species.

Differences among the clones in their ability to reactwith chemically modified proteins were observed.Modification of lysine residues of VLDL by carbamy-lation caused an increase in reactivity with the mono-clonal antibodies 3B7, 6C5, and 6H7. In contrast, theantibody 1D7 recognized carbamylated VLDL verypoorly. The recognition of apo E in VLDL by 1D7 wasalso muich more susceptible to modification of the VLDL

116 R. W. Milne, Ph. Douste-Blazy, L. Retegui, and Y. L. Marcel

by reductive methylation. This wouild suiggest that alysine residtue is present in the aintigenic determinanatrecognized by 1D7 or that the conformiiationi of thisdeterminant is radically altered wheIn lvsine residtueselsewhere in the molectule are modified. In either case,the specificity of 1D7 can be clearly distinguiishled fromthose of the other antibodies.

In concltusion, the five monoclonal anti bodies (le-scribed are specific to apo E. Differenices in theirrelatixve reactivities with apo E in the form of lipo-proteins wouild be con,sistenit with differenit intra-molecular specificities. The specificity of oine of theantibodies (1D7) cani be distinguished by its rie-activity with chemically modified VLDL. Reactivityof individual monoclonial antibodies with chemicalor proteolytic digests of apo E should allow a preciselocalization of antigenic determinants on the apo Emolecule.

ACKNOWLEDGMIENTS

'We are indebted to Dr. J. P. Vaerman and Dr. R. Verderv forvaluable disctussion, anid to Dr. J. Davignon of the ClinicalResearch Instittute of' Nontreal for providing plasmiia samplesfromii his patients. We acknowledge the excelleint technicalassistance of' Ms. Louiise Blanchette, Ms. Dianie Emonid, andMs. Camiiilla V,6zina andcl the fine secretarial hell) of 'Ms. LouliseLalond(le.

This study was stupported 1w grants fromii the MedicalResearch Couincil of Caniada (NIT-4011), fromii the Quel)ecHeart Fouindationi, and fromii the Conseil de la Reclherclhe eiiSante du Quebec.

REFERENCES

1. Marsh, J. B. 1976. Apoproteins of the lipoproteins in a nion-recircutlatinlg perfitisate of rat liv-er.J. Lipid Res. 17: 85-90.

2. Felker, T. E., NI. Fainarui, R. L. Hamiiilton, ancd R. J. Havel.1977. Secretion of the arginiine-rich anid Al apolipopro-teins by the isolated perfuised rat liver. J. Lipid Res. 18:465-473.

3. Wtu, A.-L., and H. G. Windmueller. 1979. Relative coIn-tribtutions bL liver and intestiine to individual plasmliaapolipoproteins in the rat.J. Biol. Clueni. 254: 7316-7322.

4. Brown, MI. S., and J. L. Goldsteini. 1974. Bindinig and(Idegradation of low density lipoproteinis by cutltutredhtuman fibroblasts: comparison of cells from a normiial sub-ject and from a patient with homozygous f:aniiilial hvper-cholesterolemia. J. Biol. Chiewt1. 249: 5153-5161.

5. Bersot, T. P., R. W. Mahlev, MI. S. Brown, anid J. L.Goldstein. 1976. Interaction of swine lipoproteins withthe low density lipoprotein receptor in humiiiani fibroblasts.

J. Biol. C/wei. 251: 2395-2398.6. Weisgraber, K. H., T. L. Innerarity, and R. XV. NMahlev.

1978. Role of the lvsine residtues of plasma lipoproteins inhigh affinity binding to cell suirf:ace receptors on1 humlanlfibroblasts.J. Biol. Chlewii. 253: 9053-9062.

7. Mahlev, R. WV., T. L. Imnerarity, R. E. Pitats, K. H. WNeis-graber, J. H. Brown, aind E. Gross. 1977. Inhibition of

lipoprotein binding to cell stirf'ace receptors of fibroblastsfollowving selective mo(lificationi of' arginvl residtues inatrgininie-richl anid B apoproteins. J. Biol. Ch1ei?w. 252:7279-7287.

8. Quairfordt, S., J. Haniks, R. S. Jonies, and(l F. Shelbuirne.1980. The ptitake of ligh density lipoprotein cholestervlester in the perfilised ratt liver. J. Biol. C/incii. 255:2934 -29:37.

9. Malhlev, R. WV., K. H. XVeisgraber, T. L. lnneraritv, and11. G. Windmiuieller. 1979. Acceleratecl clearanciee of low-density and high-density lipoproteins and retatrdedlclearance of E apoprotei n-containig l ipoprotei us friomiithe plasina of riats iliter mlodificationi of' lysine residuies.Proc. N\at/. Acad. SOi. tU. S. A. 76: 1746- 1750.

10. Slhelbuirne, F. A., andl S. Quiarfordt. 1977. The interactionof heparin with an apoprotein of hulimlani very low densitylipoprotein. J. C/lin. Iltve.st. 60: 944-950.

11. Mlarcel, Y. L., NI. Bergseth, anid A. C. Nestruick. 1979.Preparative isoelectric foeussing of apolipoproteins C acndE fromii hlumiian verv low density lipoproteins. Biochlii.Biop/i ijs. Acta. 573: 175-183.

12. Schliilimian, NI., C. D. Wilde, acndl G. Kohler. 1978. A bettercell linle for mlakinig hvbridomas secreting specific anti-bodies. NatlJre (Loud(.). 276: 269-270.

13. Tung, A. S., S-T. Sni, S. Sato, and A. Nisonoff. 1976.Production of large a-mounts of antibodies in individualmice.J. Iw ,nuniol. 116: 676-683.

14. Kennet, A. H., K. A. Denis, A. S. Tung, acnd N. R. Klinman.1978. Hybrid plasmiiacytoma production: fusions withadult spleen cells, monoclonal spleen fragmenits, neonatalspleen cells, aindl human spleen cells. Curr. To). Micro-biol. 1,nwnmtntiol. 81: 77-91.

15. Littlefield, J. WV. 1964. Selectioni of hybrids fromii matinlgsof fibroblasts in vitro and their presulmedl recoin-binants. Scienice (Wa/i. D. C.). 145: 709-710.

16. Chalon NI-P., R. WV. NIilne, and J.-P. Vaerman. 1979. Ivitr-o immunosuippressive effect of serumi from orallyimmu-niizedi mniee. Etur. J. Ilinwntnol. 9: 747-751.

17. Ey, P. L., S. J. Prauise, andl C. R. Jenkin. 1978. Isolationof pure IgGl, IgG2a aind IgG2b) immuitinoglobulins fromimiouise serumil usinlg Protein A-Sepharose. Iwrwunio-cheminstry. 15: 429-436.

18. Bolton, A. E., and XV. NI. Hunter. 1973. The labellingof proteins to high specific radioactivities by conjuga-tion to a 1251-containing acNlatiing agent: application tothe radioimmunoassay. Biochen. J. 133: 529-539.

19. McConahey, P. J., and F. J. Dixon. 1966. A method oftrace iodination of proteins for immunologic studies.Imit. Arch. Allergy Appl. Iwwiitunlol. 29: 185-193.

20. Warnick, G. R., C. Nlayfield, J. J. Albers, and W. R.Hazzard. 1979. Gel isoelectric focusing method forspecific diagnosis of familial hyperlipoproteinemia type3. Clini. C/lei. 25: 279-284.

21. Blum, C. G., L. Aron, and R. Sciacca. 1980. Radio-immunoassav, studies of human apolipoprotein E. J. Clin.Invest. 66: 1240-1250.

22. Havel, R. J., L. Kotite, J. L. Vigne, J. P. Kane, P. Tun,N. Phillips, and G. C. Chen. 1980. Radioimmunoassayof humilani atrginiie-rich atpolipoprotein, apoprotein E.

J. Clitn. Ive.st. 66: 1351-1362.23. Schonifeld, G., aind B. Pfleger. 1974. The structure of

humlltani high den.sity lipoprotein and the levels of apo-protein A-1 in plassmal as deterimiined by radioimmunio-assay.]. Clin. Inve-st. 54: 236-246.

.Mlonoclonal Antibodies agaitnst Apolipoprotein E 117