Published Online: 1 November, 1987 | Supp Info:
`http://doi.org/10.1084/jem.166.5.1351
`on July 18, 2018
`jem.rupress.org
`Downloaded from
`
`COMPARISON OF THE EFFECTOR FUNCTIONS OF HUMAN
`IMMUNOGLOBULINS USING A MATCHED SET OF
`CHIMERIC ANTIBODIES
`
`BY MARIANNE BRUGGEMANN,* GARETH T. WILLIAMS,1
`CAROL I. BINDON,* MICHAEL R. CLARK,* MATTHEW R. WALKER,:t
`ROY JEFFERIS,:t HERMAN WALDMANN,* AND MICHAELS. NEUBERGER§
`
`From the *Division of Immunology, Department of Pathology, Addenbrooke's Hospital,
`Cambridge CB2 2QQ; the :l:Department of Immunology, University of Birmingham,
`Birmingham B 15 2T]; and the 1Medical Research Council Laboratory of Molecular Biology,
`Cambridge CB2 2QH, United Kingdom
`
`The five lg classes have distinct biological roles. The IgG subclasses also show
`marked differences in their ability to mediate a variety of effector functions. A
`detailed comparison of the properties of the human lg classes and subclasses is
`not only of interest for relating the functions of antibodies to their structures
`but is also of great importance for the implementation of therapy based upon
`immunological intervention. Indeed, this second aspect has become particularly
`significant as the development of techniques for the production of chimeric
`antibodies (1-3) should ensure that immunological intervention is now likely to
`make use of mAbs that have human effector functions; several cell lines have
`already been established that secrete chimeric antibodies directed against human
`cancer cells ( 4-6).
`Much of our knowledge of the properties of human Igs has been obtained
`from the study of myeloma proteins (reviewed in references 7-9). However,
`generally myeloma proteins do not bind identified antigens and, moreover,
`different myeloma proteins differ not only in their heavy chain class/subclass but
`also in their light chains and variable regions. As the initiation of antibody
`effector activity is usually a consequence of antigen binding and is indeed
`influenced by the quality of that binding, previous studies on myeloma proteins,
`although valuable, may not provide a sufficient picture of antibody effector
`function for therapeutic purposes. To carry out a detailed and controlled
`comparison of the effector functions of the different human C" regions, we have
`established a panel of cell lines that secrete a matched set of human chimeric
`antibodies. These antibodies are directed against the hapten 4-hydroxy-3-nitro(cid:173)
`phenacetyl (NP). 1 This specificity for a known hapten has allowed us to compare
`the effector functions of the IgG subclasses not only when interacting with
`soluble antigen but also when interacting with cell-bound antigen. This has
`enabled us to determine the efficacy with which different subclasses lyse their
`
`This work was supported by grants from the Medical Research Council and Wellcome Biotechnology.
`1 Abbreviations used in this paper: ADCC, antibody-dependent cell-mediated cytotoxicity, NP,
`nitrophenacetyl; NIP, 5-iodo-4-hydroxy-3-nitrophenyl.
`J. EXP. MED.© The Rockefeller University Press • 0022-1007 /87 /11/1351/11 $2.00
`Volume 166 November 1987 1351-1361
`
`1351
`
`

`

`1352
`
`EFFECTOR FUNCTIONS OF HUMAN CHIMERIC ANTIBODIES
`
`target cells by means of both cell-mediated and complement-dependent mecha(cid:173)
`msms.
`
`Materials and Methods
`Plasmids. Plasmid pSV-VNr (Fig. 1) contains a single Barn HI site for the insertion of
`restriction fragments containing CH exons; this plasmid, as well as pSV-VNrHE, has been
`previously described (3). Derivatives of pSV-V NP with Xba I or Hind III sites in place of
`the Barn HI site were created by inserting octameric linkers into the Barn HI site that had
`been blunted using Escherichia coli DNA polymerase I Kienow fragment. Construction of
`the Hind III vector required prior destruction of two other Hind III sites in pSV-VNr by
`Kienow treatment. The origin and manipulation of the DNA inserts specifying the
`different C" genes is described in the legend to Fig. 1.
`Cell Lines 2 and Transfection. J558L plasmacytoma cells (10) were obtained from Dr.
`S. L. Morrison (Dept. of Microbiology, College of Physicians and Surgeons, Columbia
`Univ., New York) and were grown in DME containing IO% FCS. Transfection was
`performed by spheroplast fusion (11) in the case of the IgG2-, IgG4-, and IgE-secreting
`cell lines, and by electroporation for the other constructs. For electroporation (12), 2 X
`107 cells that had been washed and resuspended in 0.2 ml cold PBS were mixed with
`linearized plasmid DNA (15 µgin 20 µl H 2O) and placed in a I cm X 0.4 cm X 4.5 cm
`plastic microcuvette that had been equipped with aluminum electrodes. The sample was
`given 15 2-kV pulses at 1-s intervals from an Apelex power supply, left on ice for 30 min,
`and then resuspended in 50 ml DME/10% FCS/gentamycin before transferring to 24-
`well plates. Selective medium containing mycophenolic acid was applied 24 h later (11).
`Anti-NP antibody in the culture medium was assayed by RIA (13) and cells were cloned
`by limiting dilution.
`Protein Purification and Analysis. For antibody purification, cells were allowed to grow
`to saturation in 2 liters of DME/10% FCS. The culture medium was filtered, supplemented
`to 0.05% in NaN 3 , and passed over a 5-iodo-4-hydroxy-3-nitrophenacetyl (NIP)-caproate
`Sepharose affinity column. This column was prepared according to a procedure provided
`by M. Cramer (University of Cologne, Cologne, Federal Republic of Germany). 80 ml of
`Sepharose CL-4B (Pharmacia Fine Chemicals, Uppsala, Sweden) that had been washed
`and resuspended in an equal volume of ice-cold 2 M Na2CO 3, was activated by mixing
`with 4 ml of 0.5 g/ml CNBr in acetonitrile. After extensive washing, the activated
`Sepharose was mixed with 40 ml of bis(3-aminopropyl)amine that had been brought to
`pH 10 by the slow addition of 12 M HCl at 4°C. After gentle mixing overnight at 4°C,
`the derivatized Sepharose was washed in H2O, resuspended in cold 3% NaHCO3 and
`mixed with 60 mg of NIP-caproate-0-succinimide (Cambridge Research Biochemicals,
`Harston, Cambridge) that had been dissolved in 1 ml dioxane. After mixing overnight at
`4°C, the NIPcap-Sepharose was washed with 0.1 M glycine/HCI, pH 3, followed by PBS
`and then stored in PBS/0.05% NaN 3. After samples had been passed through the column
`and extensive washing, antibody was eluted from the sorbent with 0.5 mM NIP-caproate
`or NP-caproate in PBS and dialyzed extensively. Antibody samples were centrifuged
`extensively to remove aggregates.
`Protein samples were analyzed on SDS/PAGE gels (14). Samples were also purified on
`NIP-caproate Sepharose from cells labeled with [14C]-L-lysine in the presence or absence
`of 10 µg/ml tunicamycin as previously described (15).
`Serologi-cal Assays. Serologic analysis was performed in an ELISA assay as follows: wells
`of flat-bottomed microtiter plates were coated overnight at 4 "C with l 00 µ1 purified anti(cid:173)
`NP antibody diluted to 1 µg/ml in 0.2 M NaHCO3, pH 9.6, or with purified human
`paraproteins of serologically defined class, subclass, and allotype. After washing with
`PBS/0.05% Tween 20 (PBS/Tween), I 00 µI of a 1: I 00 dilution of mAb ascitic fluid in
`
`2 The cell lines will be deposited in the European Centre of Animal Cell Cultures, PHLS Centre
`for Applied Microbiology and Research, Porton Down, Salisbury, Wilts.SP4 OJG, United Kingdom,
`where they may be obtained by scientists affilitated with nonprofit research in pursuance of an
`academic research program.
`
`

`

`BRUGGEMANN ET AL.
`
`1353
`
`PBS/Tween was added to each well and incubated at 3 7 ° C for 2 h. After extensive
`washing with PBS/Tween, I 00 µI of a I :4,000 dilution in PBS/Tween of polyclonal sheep
`anti-mouse lg antibody conjugated to horseradish peroxidase (BDS Biologicals Ltd.,
`Birmingham, United Kingdom) was added and incubated for a further 2 hat 37°C before
`repeated washing in PBS/Tween. Bound sheep antibody was revealed by addition of the
`substrate (2.2 mM o-phenylene diamine). The reaction was quenched after 15 min by the
`addition of 50 µI 12.5% H 2SO4 and the OD295 was measured using a Titertec Miniscan
`(Flow Laboratories, Irvine, Scotland). Positive reactions were scored where the OD295 was
`0.5 OD units above control values.
`Protein A Binding. Purified antibodies ( I 00 µg/ml in PBS) were coated onto the wells
`of microtiter plates (Cooke; Dynatech Laboratories, Inc., Alexandria, VA) before blocking
`with PBS/3% BSA. The wells were then incubated with [125I)protein A (50,000 cpm per
`well; Amersham International, Amersham, United Kingdom), which had been diluted in
`citrate/phosphate buffer of the appropriate pH. The wells were washed at the same pH.
`Parallel incubations were also performed in which the plates were washed at the various
`pHs and the immobilized antibody developed with radio-iodinated monoclonal anti-.\ 1
`antibody Lsl 36 (13) in PBS/I% BSA. This confirmed that the pH dependence observed
`in the protein A binding assays was indeed a result of the pH dependence of anti(cid:173)
`body /protein A interactions rather than artefact due to washing the chimeric antibody
`off the microtiter plate.
`Clq Binding and Hemolysis Assays. Human Clq that had been radioiodinated by the
`lactoperoxidase method was a gift from Dr. N. Hughes-Jones and B. Gorick (MRC Centre,
`Cambridge, United Kingdom). Human erythrocytes were coupled with NIP-kephalin (gift
`of Dr. U. Weltzien, Max Planck Institute For Immunology, Freiburg, Federal Republic
`of Germany) as previously described ( I 6), and, if required, were labeled with sodium
`51 Cr)chromate (Amersham International) as previously described (17). For Clq binding
`[
`assays, washed NIP-red cells (20 µI at 109 cells/ml) in PBS/I% BSA were coated with
`saturating amounts of antibody (IO µI at 200 µg/ml) and then supplemented with
`125I)Clq (IO µI at 10-60 µg/ml). After rotation at 37°C for I h, samples were centrifuged
`[
`in microfuge tubes through 150 µI of oil of density l .028 (made by mixing four parts di(cid:173)
`n-butyl phthalate with one part dinonyl phthalate). The cell pellets were separated by
`clipping off the bottom of the tube and the radioactivity in the bound and free fraction
`was determined. Controls were performed without added antibody; nonspecific binding
`was always <I%.
`For hemolytic complement assays, NIP-human red cells were labeled with [51Cr](cid:173)
`chromate, washed, and 50 µI of cells (-10 9/ml) were mixed in microtiter wells with 50 µI
`of appropriate dilutions of the chimeric antibodies. After IO min at room temperature,
`100 µI of diluted human complement was added to give a final concentration of 20%.
`After a 30-min incubation at 37°C, the cells and supernatant were separated by centri(cid:173)
`fugation (IO0 g, 2 min). Samples incubated with no antibody were used to calculate the
`spontaneous 51Cr release. The percentage specific 51Cr release is calculated as: Percent
`release = 100 X [(Test release - spontaneous release)]/[(Total radioactivity) -
`(sponta(cid:173)
`neous release)).
`Antibody-dependent Cell-mediated Cytotoxicity. This was performed essentially as de(cid:173)
`scribed (18) but with modifications. Cells (2 X 106
`) of the human T cell line HPB-ALL
`were labeled in 100 µI of medium with 50 µCi of sodium [51Cr)chromate for 30 min at
`37°C, and 2 µI of NIP-kephalin (IO0 µg/ml) was then added and the incubation was
`continued for another 15 min. These target cells were then washed with Hepes-buffered
`Iscove's modified DME containing I% BSA. The effector cells were obtained from healthy
`donors by venipuncture and, after defibrination with glass beads, mononuclear cells were
`isolated by centrifugation on Ficoll-Hypaque (19) and cultured overnight in Iscove's
`MDM/5% FCS. Antibody-dependent cell-mediated cytotoxicity (ADCC) was measured by
`mixing labeled target cells (50 µI at 4 X 105/ml) with dilutions (100 µI) of the chimeric
`antibodies and then supplementing with the effector cells (50 µI at 1.2 X I 0 7 /ml). The
`cells were pelleted (200 g, IO min) and incubated at 37°C for 4 h. The radioactivity in
`the supernatant was then measured. Assays were performed in triplicate and controls
`
`

`

`FIGURE I. Structure of plasmids. (A) The
`pSV-VNP vector. The mouse IgH enhancer (E)
`is located upstream of the leader (L) and V
`exons of the V NP variable region. The family
`of vectors have a Barn HI, Hind III, or Xba I
`site located downstream of V NP for the insertion
`of germline CH regions. The transcriptional
`orientations of the L/V NP, gpt, and ampicillin
`resistance (amp) genes are marked with arrows.
`(B) The CH region inserts. The exons for the
`3' end of the mRNA of the membrane forms
`of the heavy chains are not depicted, although
`they are probably present in their entirety in
`the 'Y but not in the µ,, £, or a constructs. Only
`restriction sites at the ends of the CH gene
`inserts are depicted. The CH inserts originate:
`µ, Xba I fragment of phage XC75R (20); -yl,
`Hind III fragment of phage Charon 4A
`Hig-yl0 (21); -y2 and -y4, Hind III-Barn HI
`fragments of cosig8 (22), which were obtained
`as Barn HI-Barn HI fragments after cloning of
`the genomic Hind III fragments in pUCI 9; -y3
`was obtained as a 7.4-kb Hind III fragment
`from either cosigl (22) yielding an IgG.G3m(g)
`antibody or from phage XEZZ-y3 (23) yielding
`an IgG.G3m(b) antibody. The a2 insert origi(cid:173)
`nated from cosig 10 (22) and was obtained as a
`Hind III fragment from a plasmid subclone
`whose insert extended from the Pst I site 5' of
`CHI to the Bgl II site 3' of CH3. Plasmid pSV(cid:173)
`VNpH£ has been described previously (3).
`
`1
`
`2
`
`3
`
`1111 .D
`
`4
`
`<ll
`X
`
`"O
`C:
`::i:
`
`"O
`C:
`::i:
`
`E
`"' CD
`
`E
`"'
`CD
`
`E
`<ll
`CD
`
`1kb
`
`Q)
`
`~. h 2 3
`~. _h_ 2 3
`~. h 2 3
`
`"l?
`
`1
`
`h 2 3
`
`E 1
`
`2 3
`
`4
`
`Ill
`-y2 I I Ill
`111111
`Ill
`~ ••• 1
`J!
`
`1354
`
`EFFECTOR FUNCTIONS OF HUMAN CHIMERIC ANTIBODIES
`
`A
`
`©
`
`pSV-VNP
`
`gpt -
`
`amp -
`
`Barn or Xba or H,nd
`
`B
`
`µ,
`
`<ll
`.D
`X
`
`Sµ
`
`'Y 1
`
`-y3
`
`')'4
`
`f
`
`0/ 2
`
`1
`
`2
`
`3
`
`"'
`Ill CD
`
`performed without effector cells (no lysis was then observed). The specific 51Cr release
`was calculated as described above.
`
`Results
`Cell Lines Expressing Human Chimeric Antibodies. The basic plasmid that was
`used for the construction of the chimeric heavy chain genes (pSV-V NP) is depicted
`in Fig. 1. The plasmid contains the VH gene of a mouse anti-NP antibody.
`Upstream of the V H promoter is the mouse IgH enhancer element; downstream
`ofV His a unique Barn HI restriction site. Different human CH genes were inserted
`either into this Barn HI site or into derivatives of pSV-V NP in which the Barn HI
`site was converted to a Hind III or Xba I site by insertion of linkers. The exact
`CH fragments inserted are described in Fig. 1. In the case of -y3, two constructs
`were assembled with different -y3 genes that originated from different sources,
`having been cloned from different individuals.
`The gpt marker present in plasmid pSV-VNP allows stably transfected cells to
`be selected by virtue of their resistance to the drug mycophenolic acid. The
`plasmids were introduced into the mouse plasmacytoma J558L as described in
`Materials and Methods. This plasmacytoma secretes a A1 light chain but expresses
`no heavy chain of its own. The V,. of the endogenous light chain complements
`the V NP of the transfected heavy chain to yield an NP-specific antibody; such
`antibodies display a slightly greater affinity for the iodinated derivative, NIP,
`than for NP itself.
`Antibodies were purified to homogeneity from the culture medium of cloned
`
`

`

`BRUGGEMANN ET AL.
`+Tm
`+Tm
`M E A2 M E A2 G3G1 G2G4G3G1 G2G4
`
`A
`
`B
`
`1355
`
`M E A2 G3G102 G4
`
`69-
`
`45-
`
`205-
`
`150·
`
`116·
`97·
`
`-h
`
`69· -
`
`FIGURE 2. Analysis of chimeric antibodies by SDS-PAGE. (A) Antibodies from cells biosyn(cid:173)
`thetically labeled in the presence or absence of tunicamycin (Tm) were analyzed on I 0% SDS(cid:173)
`p AGE gels after reduction. (B) Antibodies purified from culture supernatants were analyzed
`without reduction on a 7% SDS-P AGE gel. The positions of molecular mass (kD) markers are
`indicated.
`
`transfectants; yields were typically in the range 2 to 10 mg/liter although
`sometimes up to 30 mg/liter was achieved.
`Analysis of Antibodies on SDS/PAGE. To characterize the chimeric antibodies,
`purified samples were reduced and analyzed by SOS-PAGE. The mobilities of
`the heavy chains were much as predicted on the basis of their DNA sequences
`except forµ, f, and a 2 • We have previously shown (3) that the chimeric E heavy
`chains become heavily glycosylated. We therefore resorted to biosynthetic label(cid:173)
`ing to compare the antibody normally secreted by the transfectants with that
`made in the presence of the glycosylation inhibitor tunicamycin. The results
`(Fig. 2A) demonstrate the IgM, IgE, and IgA2 secreted by the J558L transfec(cid:173)
`tants are indeed heavily glycosylated on their heavy chains. A much smaller
`amount of glycosylation is apparent on the 'Y heavy chains. The sizes of the heavy
`chains of the unglycosylated antibodies are exactly as predicted from the DNA
`sequences (the -y3 heavy chains standing out from the other 'Y chains because of
`the long hinge).
`To examine whether the secreted antibodies were correctly assembled, unre(cid:173)
`duced samples were analyzed on 7% SOS-PAGE gels (Fig. 2B). The IgGs and
`IgE exhibit mobilities consistent with their having the expected H 2L2 structures
`whereas the IgM is clearly of very large molecular weight and, as expected for
`the pentameric form, scarcely enters the gel. IgA2 gives the most complex
`pattern and appears to contain H, HL, H 2L 2 , and (H 2L2)z forms; this may reflect
`the secretion of some noncovalently linked molecules.
`Serological Characterization. For potential therapeutic applications, it is clearly
`important to establish whether chimeric human antibodies secreted by a mouse
`plasmacytoma are indeed homologous to human mAbs. While the analysis on
`SOS-PAGE indicated that the chimeric antibodies resembled their human coun(cid:173)
`terparts as regards the molecular weights of both the native and the unglycosy(cid:173)
`lated heavy chains, we decided to extend this characterization to include a wide
`
`

`

`1356
`
`EFFECTOR FUNCTIONS OF HUMAN CHIMERIC ANTIBODIES
`
`I HINGE
`
`ZG4-IgG3
`
`DOMAIN
`
`I INTER
`We10 ~g1
`or,*
`veg*
`F7c*
`
`Cy2
`
`JD312
`G7c
`JL512
`NL16
`GOMl
`GON2
`GB7B
`HP6031
`ASS
`F10F
`JD79
`HP6080
`
`IgGl
`IgGl
`IgG2
`IgG2
`IgG4
`non-IgG2
`non-Glm(g)
`non-Glm(g)
`non-GJm(g)
`GJm(ul
`
`Cyl
`
`TM15
`HP6044
`HP6046
`TM10 - non-IgG1
`TM14
`Glm(fl
`HP6014
`IgG2
`
`Cy3
`
`8a4
`Jl0d
`X3a8
`lal
`Je10
`49/2C3
`RJ4
`
`non-IgG4
`non-IgG4
`non-IgG4
`non-IgG4
`IgG4
`
`. non-IgGJ
`
`FIGURE 3. Molecular localization of epitopes recognized by mAbs to human IgG. For the
`origin and specificity of these antibodies see references 24 and 25. The anti-IgA antibodies
`are described in reference 26.
`
`range of serological markers. The chimeric antibodies were therefore typed
`using 37 different monoclonal anti-human antibodies (Fig. 3). On the basis of
`this typing (Table I), it will be seen that the two IgG3s are allotypically distinct.
`The ')'3 gene of the XEZZ')'3 clone (23) yields an IgG3 of the nG3m(g) iso(cid:173)
`allotype; in other words, it is not of the G3m(g) allotype. Indeed, the DNA
`library from which the XEZZ')'3 clone was isolated was made from a Tunisian
`individual known to be homozy"gous for the G3m(b) allotype (23). However, the
`')'3 gene from the coslgl clone (22) yields an IgG.G3m(g) antibody. The chimeric
`IgA2 types as nA2m(2), a result that is predicted from the sequence of the a2
`gene used here which shows it to be of the A2m(l) allotype (27). Thus, all the
`chimeric antibodies type exactly as expected and the chimeric antibodies are not
`therefore serologically distinguished in these assays from authentic human anti(cid:173)
`bodies as regards the C" region determinants.
`Binding to Protein A. The chimeric antibodies were further tested in their
`binding of radio-iodinated Staphylococcal protein A over a pH range from 3 to
`10. The results (Fig. 4) indicate that the chimeric IgGl, IgG2, and IgG4 bind
`well to protein A and show a very similar pH dependence, binding occurring
`down to pH 4.5. This similarity of pH dependence could be due to titration of
`one of the conserved residues in the antibody C"2/C"3 domain binding site for
`protein A (see review in reference 9) or could be due to titration of one of the
`protein A side chains themselves. No protein A binding was detected with IgM,
`IgE, IgA2, or either of the IgG3s.
`Binding of Human Clq. The binding of Clq to the chimeric antibodies was
`
`

`

`BRUGGEMANN ET AL.
`
`1357
`
`TABLE I
`Serological Typing of Chimeric Antibodies
`
`Chimeric antibody
`
`IgG2
`+
`
`IgG3*
`+
`
`+
`+
`+
`
`+
`
`+
`ND
`ND
`ND
`ND
`ND
`
`+
`
`+
`+
`
`+
`ND
`ND
`ND
`ND
`ND
`
`IgG3•
`+
`
`IgG4
`+
`
`+
`
`+
`+
`+
`
`+
`ND
`ND
`ND
`ND
`ND
`
`+
`+
`
`+
`+
`
`ND
`ND
`ND
`ND
`ND
`
`IgA2
`
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`+
`
`+
`+
`ND
`
`IgM
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`+
`
`Specificity
`
`Number
`used
`
`7
`
`I
`I
`5
`
`IgGI
`+
`+
`
`+
`
`+
`
`+
`ND
`ND
`ND
`ND
`ND
`
`IgG
`IgGI
`Glm(f)
`IgG2
`IgG2,3,4
`IgGI,2,4
`IgG3
`G3m(u)
`4
`nG3m(g)
`I
`IgG4
`IgGI,2,3
`2
`I
`IgA
`3
`IgAI
`I
`lgA2
`I
`nA2m(2)
`6
`IgM
`* The -y3 C" region derives from clone coslgl (22).
`+ The -y3 C" region derives from phage >.EZZ-y3 (23).
`
`lg61
`
`lg62
`
`lg64
`
`_20,000
`E
`a.
`.!!.
`-., 5,000
`§
`_g
`c( 1,000
`C
`~
`ii
`
`200
`
`3 4 5 6 7 8 9 10
`3 4 5 6 7 8 9 10
`3 4 5 6 7 8 9 10
`pH
`pH
`pH
`FIGURE 4. Binding of protein A to chimeric antibodies. The amount of radio-iodinated
`protein A bound to immobilized antibodies is plotted as a function of pH. No binding was
`detected to the IgM, IgE, IgA2, or either of the IgG3s over the entire pH range tested.
`
`assayed using various concentrations ofradiolabeled Clq and hapten-derivatized
`red cells that were coated with amounts of antibody that were shown to be
`saturating. The results (Fig. 5, left) show that Clq binding is detected with the
`IgM, IgGI, and both IgG3 antibodies but not with IgG2, IgG4, or IgE. The
`binding to IgM is weaker than to IgG 1 and IgG3 and, of these two IgG subclasses,
`the IgG3s bound more Clq than did the IgG 1.
`Complement-mediated Hemolysis. Antibodies were tested over a wide concen(cid:173)
`tration range for their ability to lyse hapten-coupled human red cells in the
`presence of human complement. The results (Fig. 5, right) show that IgG2,
`IgG4, and IgE did not mediate hemolysis, whereas IgM, IgG 1, and both IgG3s
`were effective. Indeed, it is notable that the IgG 1 was considerably more effective
`in this hemolytic assay than the IgG3s. We have previously demonstrated (17) that
`
`

`

`EFFECTOR FUNCTIONS OF HUMAN CHIMERIC ANTIBODIES
`
`1358
`
`0.06
`
`a 0.04
`~
`,er
`6
`.g 0.02
`
`,:i
`C
`:I
`
`lgG.G3m(g)
`
`lgG.G3m(b)
`
`100
`
`41
`Ill
`jg 80
`~ a:
`$5 60
`"'
`u 40
`;;::: ·u
`
`41
`CL 20
`UJ
`
`lgG1
`lgM
`
`lgG.G3m(b)
`
`~ 0
`
`0
`
`lgG.G3m(g)
`lgE
`~lgG2
`0 .f.-..11:i·lf_;.....,.,~~•~;;::~..:.,_
`25 lgG4
`1.5
`0.4
`0.1
`6
`0.02
`15
`10
`5
`0
`(µg/ml)
`Antibody Concentration
`Input C1q Concentration (µg/ml)
`FIGURE 5. Binding of Clq and complement-dependent hemolysis by chimeric antibodies.
`(Right) ·Binding of various concentrations of radiolabeled Clq to NIP-human red cells that
`have been coated with saturating amounts of chimeric antibody. (Left) Lysis of [51Cr)NIP(cid:173)
`human red cells by various concentrations of chimeric antibody in the presence of human
`complement. The allotypes of the two IgG3 antibodies [IgG.G3m(b) and IgG.G3m(g)] are
`indicated.
`
`60
`
`I
`&!
`0
`;;; 30
`
`(cid:127)
`
`lgG1
`
`lgG.G3m(b)
`•
`Y lgG.G3m(g)
`
`0 lgM
`•lgG2
`+ lgA
`I lgG4
`• lgE
`
`FIGURE 6. ADCC by chimeric antibodies. The
`lysis of [51 Cr)NIP-conjugated human target cells
`(HPB-ALL T cell line) by human PBMC was meas(cid:173)
`ured as a function of the concentration of anti-NP
`chimeric antibody.
`
`-4
`·5
`-1
`-6
`-2
`-3
`10 10 10 10 10 10
`Antibody Concentration (mg/ml)
`
`the efficacy of anti-NP antibodies in this hemolytic assay is critically dependent
`on their affinity for the hapten. However, this cannot explain the results obtained
`here. Binding-inhibition assays (not shown) indicate that the IgG 1 and IgG3
`antibodies have indistinguishable affinities for NIP-caproate. Furthermore, the
`IgG 1 and IgG3 samples gave very similar titers in hemagglutination assays using
`hapten-derivatized red cells. Thus, although IgG3 is the best subclass as regards
`C 1 q binding it is considerably less potent than IgG 1 in hemolytic assays. Inter(cid:173)
`estingly, a similar but less marked reversal of hemolytic efficiency as compared
`with Clq binding is observed with the two IgG3 allotypes. Whereas the
`IgG.G3m(g) binds Clq slightly better, the IgG.G3m(b) is somewhat more effective
`in hemolysis.
`Antibody-dependent Cell-mediated Cytotoxicity. The efficacy of the antibodies in
`mediating ADCC was tested using a hapten-derivatized human T cell line (HPB(cid:173)
`ALL) as target and mononuclear cells from human volunteers as effector cells.
`It was found (Fig. 6) that only IgG 1 and IgG3 were effective in mediating ADCC
`with the IgG 1 showing greater potency than either of the IgG3s.
`
`

`

`BRUGGEMANN ET AL.
`
`1359
`
`Discussion
`The cell lines established during the course of this work provide a source of
`chimeric antibodies that can be easily purified to homogeneity in a one-step
`purification on hapten sorbents. To confirm that these chimeric human antibod(cid:173)
`ies made in mouse plasmacytoma cells exhibit the features expected of antibodies
`possessing human C" regions, we have characterized the antibodies as regards
`their serological properties, the mobilities in SDS-PAGE of both the native and
`the unglycosylated forms as well as their binding to protein A. In all these
`respects, the chimeric antibodies behave exactly as expected from their authentic
`human counterparts.
`The known antigen specificity of these chimeric antibodies has allowed us to
`assay their efficacy in complement-mediated lysis and in ADCC. This has given
`rise to some novel and unexpected findings. The hierarchy of the binding of the
`aggregated chimeric IgG subclasses to human Clq described here agrees well
`with the previous results on the hierarchy in Clq binding of monomer IgG
`myeloma subclasses (28); however, aggregated IgG will bind with a higher avidity
`than monomeric IgG. Nevertheless, when we measured the efficacy of the
`antibodies in complement-mediated hemolysis, a very different result was ob(cid:173)
`tained. It is evident that the IgG I is very much more effective than the IgG3 in
`this assay. At present it is not clear why there is this lack of correlation between
`Clq binding and hemolytic efficacy; it will clearly be of interest to compare the
`IgG 1 and IgG3 chimeric antibodies in intermediate stages of the pathway such
`as Cl activation and C4 and C3 binding. It will also be of interest to determine
`how the comparative hemolytic efficacy of IgG 1 and IgG3 depends upon the
`density and nature of the antigen on the target cell.
`The results presented here also indicate that the IgG 1 and IgG3 are the only
`antibodies that are really effective in ADCC. The greater effectiveness of IgG 1
`as compared with IgG3 is a novel and interesting finding. The relative inactivity
`of the IgG2 and IgG4 antibodies was, however, predictable. Although the exact
`nature of the effector cell responsible for mediating ADCC has not been unam(cid:173)
`biguously identified, the Fe receptors found on the possible effector cell types
`(lymphocytes and monocytes) bind IgG 1 and IgG3 much better than IgG2 and
`IgG4 (reviewed in reference 9).
`Finally, it is worth noting that the results obtained in this work suggest that
`for many therapeutic purposes an IgG 1 antibody might be greatly preferred to
`the other IgG subclasses as it appears to be considerably more effective in
`mediating both complement-dependent Iysis and ADCC.
`
`Summary
`Cell lines have been established that secrete a matched set of human chimeric
`IgM, IgG 1, IgG2, IgG3, IgG4, IgE, and IgA2 antibodies that are directed against
`the hapten 4-hydroxy-3-nitrophenacetyl. These chimeric antibodies secreted
`from mouse plasmacytoma cells behave exactly like their authentic human
`counterparts in SDS-PAGE analysis, binding to protein A and in a wide range of
`serological assays. The antibodies have been compared in their ability to bind
`human Clq as well as in their efficacy in mediating lysis of human erythrocytes
`in the presence of human complement. A major conclusion to emerge is that
`
`

`

`1360
`
`EFFECTOR FUNCTIONS OF HUMAN CHIMERIC ANTIBODIES
`
`whereas IgG3 bound Clq better than did IgGl, the chimeric IgGl was much
`more effective than all the other IgG subclasses in complement-dependent
`hemolysis. The IgG 1 antibody was also the most effective in mediating antibody(cid:173)
`dependent cell-mediated cytotoxicity using both human effector and human
`target cells. These results suggest that IgG 1 might be the favoured IgG subclass
`for therapeutic applications.
`
`We are indebted to Catherine Teale and Mark Frewin for invaluable technical assistance
`and to B. Gorick, T. Honjo, N. Hughes-Jones, M.-P. Lefranc, C. P. Milstein, T. H.
`Rabbitts, and U. Weltzien for gifts of DNA clones or reagents.
`
`Received for publication 27 July 1987.
`
`References
`I. Bo1,1lianne, G. L., N. Hozumi, and M. J. Shulman. 1984. Production of functional
`chimaeric mouse/human antibody. Nature (Lond.). 312:643.
`2. Morrison, S. L., M. J. Johnson, L. A. Herzenberg, and V. T. Oi. 1984. Chimeric
`human antibody molecules: mouse antigen-binding domains with human constant
`region domains. Proc. Natl. Acad. Sci. USA. 81 :6851.
`3. Neuberger, M. S., G. T. Williams, E. B. Mitchell, S. S. Jouhal, J. G. Flanagan, and
`T. H. Rabbitts. 1985. A hapten-specific chimaeric IgE antibody with human physio(cid:173)
`logical effector function. Nature (Lond. ). 314:268.
`4. Sahagan, B. G., H. Dorai, J. Saltzgaber-Muller, F. Toneguzzo, C. A. Guindon, S. P.
`Lilly, K. W. McDonald, D. V. Morrissey, B. A. Stone, G. L. Davis, P. K. McIntosh,
`and G. P. Moore. 1986. A genetically-engineered murine/human chimeric antibody
`retains specificity for human tumor-associated antigen.]. Immunol. 13 7: 1066.
`5. Sun, L. K., P. Curtis, E. Rakowicz-Szulczynska,J. Ghrayeb, S. L. Morrison, N. Chang,
`and H. Koprowski. 1986. Chimeric antibodies with 17-lA-derived variable regions
`and human constant regions. Hybridoma. 5:Sl 7.
`6. Liu, A. Y., R. A. Robinson, K. E. Hellstrom, E. D. Murray, C. P. Chang, and I.
`Hellstrom. 1987. Chimeric mouse-human IgG 1 antibody that can mediate lysis of
`cancer cells. Proc. Natl. Acad. Sci. USA. 84:3439.
`7. Spiegelberg, H. L. 1974. Biological activities of immunoglobulins of different classes
`and subclasses. Adv. Immunol. 19:259.
`8. Winkelhake, J. L. 1978. Immunoglobulin structure and effector functions. Immuno(cid:173)
`chemistry. 15:695.
`9. Burton, D.R. 1985. Immunoglobulin G: functional sites. Mol. Immunol. 22:161.
`10. Oi, V. T., S. L. Morrison, L. A. Herzenberg, and P. Berg. 1983. Immunoglobulin
`gene expression in transformed lymphoid cells. Proc. Natl. Acad. Sci. USA. 80:825.
`11. Neuberger, M. S., and G. T. Williams. 1986. Protein engineering of antibody. In
`Protein Engineering: Applications in Science, Medicine and Industry. M. Inouye and
`R. Sarma, editors. Academic Press, New York. 311-317.
`12. Potter, H., L. Weir, and P. Leder. 1984. Enhancer-dependent expression of human
`K immunoglobulin genes introduced into mouse pre-B lymphocytes by electropora(cid:173)
`tion. Proc. Natl. Acad. Sci. USA. 81:7161.
`13. Reth, M., T. Imanishi-Kari, and K. Rajewsky. 1979. Analysis of the repertoire of
`anti-NP antibodies in C57BL/6 mice by cell fusion. Eur.]. Immunol. 9: 1004.
`14. Laemmli, U. K. 1970. Cleavage of structural proteins during the assembly of the
`head of bacteriophage T4. Nature (Lond.). 2278:680.
`15. Galfre, G., and C. Milstein. 1981. Preparation of monoclonal antibodies: strategies
`and procedures. Methods Enz.ymol. 73:3.
`
`

`

`BRUGGEMANN ET AL.
`
`1361
`
`16. Briiggemann, M., and K. Rajewsky. 1982. Regulation of the antibody response
`against hapten-coupled cells by monoclonal anti-hapten antibodies of various isotypes.
`Cell. Immunol. 71 :365.
`17. Neuberger, M. S., and K. Rajewsky. 1981. Activation of mouse complement by
`monoclonal mouse antibodies. Eur. ]. Immunol. 11: 1012.
`18. Hale, G., M. Clark, and H. Waldmann. 1985. Therapeutic potential ofrat monoclonal
`antibodi