The Royal Society is collaborating with JSTOR to digitize, preserve, and extend access to
`Proceedings of the Royal Society of London. Series B, Biological Sciences.
`®
`www.jstor.org
`
`1 of 21
`
`BI Exhibit 1056
`
`

`

`394
`C. Milstein
`what seemed quite clearly basic, with no possible application, became very much
`applied. I do not plan to produce analogous examples of exactly the opposite, of
`which there are many.
`It is not only that I was totally committed to basic immunology before the
`method for the derivation of monoclonal antibodies was developed, but also that
`the method itself evolved from one experiment, among others, performed to
`provide us with a more appropriate cell line with which we could continue our
`studies on the old problem of the nature and origin of antibody diversity.
`I became involved in immunology in 1962, fascinated, as many others, by the
`diversity and specificity of antibodies. Tllls was a problem that had been growing
`in theoretical interest since it was first recognized by Ehrlich at the beginning of the
`century. My involvement was prompted by the developments that were taking
`place at the time and which, in the words of R.R. Porter (1967), offered 'a feasible
`experimental approach to obtaining an answer to the question .... Does amino
`acid sequence alone control antibody specificity and, if so, how is it achieved?'.
`The following period in basic immunology was as fruitful as in our wildest
`dreams. By 1970, our general ideas had settled down to a meaningful picture
`(Milstein & Pink t970) which has not changed in its fundamentals although our
`understanding of the system has been revolutionized by the unfolding of its
`intricacies and complexities. Indeed, it was as a consequence of the advances of
`that period that I became convinced that to further our knowledge of the subject
`we needed a basic change in approach. So my priorities shifted from protein
`chemistry to nucleic acid chemistry and somatic cell genetics. In a short time
`I found myself and my collaborators trying to make mutants of myeloma cells in
`culture and at the same time fusing myeloma cells to alter the stability of their
`expression. The coexistence of those two aims and the need to evolve new ways to
`further them were the essential ingredients from which the research that I will
`describe to you developed.
`
`HYBRID CELL LINES SECRETING PREDEFINED ANTIBODY
`Antibodies are made up of light and heavy chains (as illustrated in figure 1),
`which are usually joined by disulphide bonds, each containing a variable and
`a constant region, usually referred to as the V region and the C region. Each V
`region is a folded polypeptide of about 100-J 20 amino acids and contains one
`intra.chain disulphide bond. The C region contains between. one and four similar
`pseudo-subunits in each cha.in. These define the class of the antibody molecule.
`The C regions are involved in effector functions, such as complement fixation and
`transport across membranes. Within a type or a subclass the C region is highly
`constant. On the other hand the V region is highly variable; with very few excep­
`tions each antibody molecule has a different V region, even when the same anti­
`body specificity is shared by more than one molecule. This dual role of recognition
`and effector functions, although expressed in a single polypeptide chain, is under
`
`2 of 21
`
`BI Exhibit 1056
`
`

`

`The Wellcome Foundation Lecture, 1980
`
`395
`
`the control of independent genetic loci (figure 2). There are key elements in this
`genetic arrangement that make the antibody gene family a unique system : the
`final expression into protein requires a reanangement of the genes and in addition
`there is insufficient coding DNA to account for the diversity of amino acid se­
`quences to which the germ line genes can give rise.
`
`Heavy
`Light
`
`Hinge region
`FIGURE 1. The IgG molecule: light and heavy chains are joined by S-S bridges, which have
`not been drawn as they vary in classes itnd subcll\Sscs of antibodies. They are made up
`of S-S loops of about 100-120 residues each, and tho one at the N-terminus is highly
`variable.
`
`·
`
`The DNA rearrangements occur somatically at some stage during differentiation
`of the stem cells into antibody secreting cells. These changes commit the relevant
`cells to the production of a single antibody structure. But, since the genetic
`changes are independent for each cell, the antibody molecule secreted by each cell
`is different (figure 3). The antibody response is the result of the proliferation of
`some of these ce!Js triggered by the antigenic stimulus.
`Many of the important advances in our present understanding of this system
`have come from studies of myelomas and related lympho-proliferative disorders.
`Myelomas are tumours of antibody-secreting cells that arise spontaneously in
`animals, but that can be induced in mice by injections of mineral oil. They do not
`arise as the result of a specific antigenic stimulation, but they produce and secrete
`an immunoglobulin, myeloma protein, with no defined antibody function.
`Myeloma tumours in experimental animals can be transplanted and adapted to
`grow in tissue culttll'e. On the contrary the natui·ally occuning antibody-pr�ducing
`cells, which proliferate fu the spleen and other lymphoid organs as a result of
`
`3 of 21
`
`BI Exhibit 1056
`
`

`

`V region domains
`
`�
`
`.5
`"
`.c ...
`;:
`�
`
`I
`
`I
`
`K
`
`>.
`
`I Vte1 Vte1
`-- --
`
`JK I
`VK,
`-- -- --
`
`I V>., V>.,
`
`J >.,
`
`h:,
`--
`
`___ ..!. ..... �
`I C>., C>.,
`- --··········
`
`C region domains
`
`heavy chains
`
`VH1 Vtt1
`
`-- - (Ott) JH1
`
`'"'
`
`Cy, Cy1 C-711 Ca
`Cl)
`Cµ
`- - -. .•... ·+----i 1--i 1--i 1--i t--t 1--i
`
`Cc
`1--1
`
`F10URE 2. A schematic representation of the genes coding for antibodies. The three chains are probably in different
`chromosomes. In the mouse the K and heavy chains are probably on chromosomes 6 and 12, respectively
`(Hengartner et al. 1978) and the heavy chains on chromosome 14 in man (Croce et al. 1980) and in rat (Schroder
`et al. 1980). The V regions are coded by V fragments and J fragments of DNA occurring many thousands of bases
`a.part. The number and detailed arrangements of genes in each case vary in different species. For the expression
`of an antibody, individual V, J and C fragments are associated combinatorially within a horizontal array.
`
`<:;:>
`�
`C)
`
`p �
`[
`
`4 of 21
`
`BI Exhibit 1056
`
`

`

`The Wellcome Foundation Lecture, 1980
`
`397
`
`SPLEEN CELLS
`(Die in culture J
`
`< 1+2+3+4 ... n l
`!
`
`�
`
`IMMORTALIZATION BY FUSON TO MYELOMA
`I
`Myeloma line
`G1ows in tulture
`t ------.-
`OIC:S 1n HAr medium
`IFUSIONI � S.C1oonof hyb"cl•
`.nHATmed1um 11----- IAssay ant1bocly I
`���
`� � �
`
`cell hybrids
`Cloning of Somatic
`
`F10URE 3. When an animal is injected with an immunogen the animal responds b y producing
`an enormous diversity of antibody structures directed against different antigens, different
`determinants of a single antigen, and even different antibody structures directed against
`the same determinant. Once these are produced they are released into the circulation and
`it is next to impossible to separate a.11 the individual components present in the serum.
`But each antibody is made by individual cells. The immortali2'ation of specific antibody­
`producing cells by somatic cell fusion ·followed by cloning of the appropriate hybrid
`derivative allows permanent production of each of the antibodies in separate culture
`vessels. The cells can be injected into animals to develop myeloma-like tumours. The
`sertun of the tumour-bearing animals contains large amounts of monoclonal antibody.
`
`5 of 21
`
`BI Exhibit 1056
`
`

`

`C. Milstein
`398
`antigen stimulation, have a very short life span and cannot be cultured in vitro.
`Codominant expression of immortality (from myelomas) and antibody production
`(from lymphoid cells) is achieved in somatic cell hybrids, when myeloma cells are
`fused with such normally occurring antibody-producing cells. The essential
`features of the derivation of hybrids secreting a specific antibody are shown in
`
`clone P3-X63Ag8
`JgGI (K)
`(grows in Le. but
`dies in HAT)
`
`\ I
`
`spleen cells
`SRBC immunized
`
`BALB/c
`(dies in t.c.)
`
`Sp hybrids
`(grow in HAT)
`
`all secrete P3 chains
`most secrete new chains
`some with anti·SRBC activity
`
`isolated clones
`
`Sp 1/7
`
`Sp 2/3
`
`Sp 3/15
`
`anti-SR BC
`
`anli·SRBC
`
`anti-SR BC
`
`IgGI
`lgG2b
`macroglobulin
`FrnunE 4. Fixation of the specific antibody production of a transient spleen cell in a permanent
`tissue culture (t.c.) line. Anti-SRBC hybrids. (Taken from Milstein et al. (1980).)
`
`figure 4. The myeloma parent confers to the hybrid the malignant phenotype and
`its ability to grow permanently in tissue culture. Being prepared as a mutant
`defective in the enzyme hypoxanthine guanine phosphoribosyl transferase, it can­
`not grow under conditions where incorporation of hypoxanthine is essential to
`cell growth (RAT medium; Szyba,lski et al. 196z; Littlefield 1964). The spleen
`parental cell provides the active enzyme and permits the hybrid cells to grow under
`these conditions. In addition it contributes with the reananged V and C genes for
`both h.cavy and light chains which code fo1· a specific antibody. In this way the
`transient prope1·ty of antibody secretion can be fixed as a permanent property of
`an established cell line (figure 3).
`The first permanent lines of cells secreting a predefined antibody were against
`sheep red blood cells and the hapten TNP (Kohler & Milstein 1975, 1976). Since
`
`6 of 21
`
`BI Exhibit 1056
`
`

`

`The Wellcome Found,ation Lecture, 1980
`
`399
`then many other hybrid myelomas have been derived secreting antibodies to small
`ligands (haptens), proteins, carbohydrates, cell surface components, glycolipids,
`viruses and enzymes. The experience accumulated tends to. indicate that the
`method is general. The degree of difficulty in obtaining a specific hybrid myeloma
`seems to be correlated to the response of the immunized animal. When this res­
`ponse is very weak the search for the specific antibody-producing clone among the
`many hybrids secreting non-specific immunoglobulin may require special methods.
`The derivation of hybrids between mouse myelomas and spleen cells from an
`immunized donor represented a departure from previous uses of cell fusion.
`Somatic cell hybrids had been used for two purposes (Ringertz & Savage 1976).
`For gene mapping, interspecific hybrids were made between cells of different
`species to correlate the segregation of chromosomes to the loss of species-specific
`properties (like the electrophoretic mobility of a given enzyme). For studies of
`gene expression, intraspecific hybrids obtained by fusion of two cell types from the
`same species were used. This permitted the study of the loss as well as comple·
`mentation of a variety of cellular functions. In our experiments and for the im­
`mortalization of specific differentiated functions, hybrids were prepared of normal
`and transformed cells genetically and phenotypically as closely related as possible.
`
`THE IMPORTANCE OF COMPATIBLE PHENOTYPES
`There was a very welcome but unexpected feature in the derivation of specific
`antibody-secreting hybrids that we noticed very early on. As, in the spleens we
`were using, well below 1 % of the cells secreted specific antibody, we would have
`been very pleased if 1 % of the hybrids derived secreted the specific antibody. Of
`course, we were delighted but also a bit suspicious when the proportion turned out
`to be one order of magnitude better, around 10%. It soon became quite obvious
`that, together with immortalization, we had enrichment of our selected function.
`This apparent selectivity seems to have several components. One may be related
`to preferential survival of hybrids between actively dividing cells. Since the
`hybrids are prepared with immunized spleen cells, actively dividing cells are en­
`riched with those triggered by antigenic stimulation.
`Another component in the apparent selectivity is probably due to comple­
`mentation. The myeloma parental line is an actively secreting cell and can provide
`to the hybrids the high production and secretion phenotype. This has been shown
`in fusions of myelomas and non-secreting lymphomas (Levy & Diley 1978;
`Laskov et al. 1978; Raschke 1980).
`In a hyperimmune spleen, for each cell actively secreting antibody there are
`perhaps five more cells that synthesize but either do not secrete antibody or
`secrete it only in trace amounts. The fusion of either of them with the high­
`secretor myeloma may result in a high-secreting hybrid.
`There seems to be another most interesting component of the selectivity related
`to phenotypes of the parental cells. If the same population of spleen cells, which
`
`7 of 21
`
`BI Exhibit 1056
`
`

`

`400
`
`C. Milstein
`contain similar numbers of B and T cells, are fused with a. myeloma or a T cell
`lymphoma., the results are quite different. When the fusion parent is the myeloma,
`the growing hybrids express the antibody-secreting phenotype of the pa.rental
`spleen but do not seem to express the <P markers characteristic of T cells. On the
`other hand, when the fusion parent is a T cell lymphoma., the growing hybrids
`preferentially express the T cell characteristics (table 1).
`
`TABLE 1. SELECTIVE FIXATION OF DIFFERENTIATED FUNCTIONS IN
`ESTABLISHED HYBRIDS
`('faken from Milstoin et al. 1980.)
`
`cell phenotype
`
`Ig secreted (%)
`
`Thy-1 surface antigen(%)
`
`parental
`myeloma
`
`other
`lg
`
`> 95
`> 90
`0
`0
`0
`
`0
`ca. 65
`ca. 5
`0
`0
`
`Thy-1, 1
`0
`0
`0
`> 90
`> 95
`
`Thy-1, 2
`
`0
`0
`ca. 40
`ca. 70
`0
`
`parental and hybrid lines
`
`X63 (myeloma)
`(X63 x spleen) hybrids
`spleen
`(BW x spleen) hybrids
`BW (T lymphoma)
`
`The selective effect of fusions with myeloma is more dramatically seen with rat
`myeloma cells than with mouse myelomas. This is illustrated by some recent
`experiments described in table 2. It shows that all or nearly all hybrids derived
`with the rat myeloma line express B cell characteristics of the parental spleen cell.
`It is not clear yet whether the higher apparent selectivity of the rat line when
`compared to the mouse lines is due to phenotypic compatibility, better stability,
`or a combination of both.
`The practical message of these experiments is that 'for the recovery of a dif­
`ferentiated property of a given cell type in an established hybrid it may be best to
`use the parental partners cells of similar type' and that 'for the derivation of
`hybrids with specific T cell functions, thymomas are likely to be better parental
`partners' (Milstein et al. 1976). T cell hybrids prepared in this way constitute now
`a subject of their own.
`The immortalization process therefore does not produce a random sample of all
`the cells from the spleen but it seems to be a random representation of the anti­
`body-producing cells. For this reason it is considered away of dissectingtheimmune
`response of the animal. But the preci<>e meaning of such statements is not as clear
`as it sounds. One of the main reasons is that conela.tion of the antibody present in
`the serum, or even of the antibody-secreting cells, with the products of the isolated
`hybrids must take into consideration fast changes in the differentiated state of the
`cells in question. At present we do not have a definite picture of the relative
`survival advantages of hybrids derived from :8 cells at different states of differenti-
`
`8 of 21
`
`BI Exhibit 1056
`
`

`

`The Wellcome Foundation Lecture, 1980
`
`401
`
`ation. The problem has considerable theoretical interest and could have practical
`implications. But this should not cast a serious shadow over the more general
`statement that roughly speaking the hybridization represents a random im­
`morta.lization of the antibody-producing cells.
`
`TABLE 2. THE EXPRESSION OF THE SPLEEN PARENTAL IMMUNOOLOBULIN
`IS BETTER IN HYBRIDS PREPARED WITH RAT MYELOMAS THAN IN THOSE
`PREPARED WITR MOUSE MYELOMAS
`
`fusion
`
`NNl
`NN2
`NOA!
`XOAI
`
`YS3/5
`YA5
`YA5
`YOLI
`
`NR5/6
`
`parental cells
`
`mouse x mouse
`NSI/1 Ag.4.1 x B10.D2 spleen
`NSI/1 Ag.4.1 x C3H spleen
`NSI/l Ag.4.1/0 x BALB/c spleen
`X63 Ag 8.653 x BALB/c spleen
`rat x rat
`Y3 Ag.1.2.3 x DA spleen
`Y3 Ag.1.2.3 x DA spleen
`Y3 Ag.1.2.3 x DA spleen
`YB2/3.0 Ag x Lou spleen
`
`mouse x rat
`NSI/1 Ag.4.1 x Lou spleen
`
`negative hybrid clones as a
`percentage of growing clones
`
`minimum maximum
`value
`value
`
`best
`estimate
`
`18
`35
`41
`50
`
`0
`0
`0
`10
`
`50
`
`85
`99
`89
`75
`
`37
`50
`50
`51
`
`99
`
`< 44
`< 60
`60
`50
`
`7
`< 30
`< 30
`< 35
`
`< 96
`
`rat x mouse
`< 36
`3
`54
`YN5/6
`Y3 Ag.1.2.3 x C3H spleen
`The results were obtained three to four weeks after fusion, by analysis of the immuno­
`globulin secreted by all successfully growing hybrids followed by a statistical calculation of
`the number of growing clones. (Taken from Milstein et al. (1980).)
`
`THE IMPORTANCE OF COMPATIBLE GENOTYPES
`
`Early fusions were performed between cells of the same inbred strains. We soon
`found that using cells from different mouse strains did not alter the results. We
`then established that fusions between mouse myeloma cells and rat spleen cells
`(mouse x rat) were equally successful (Galfre et al. 1977). However, other com­
`binations were more difficult. Efforts have been made by us and by others to
`immortalize the antibody production of rabbit and human cells by means of
`myeloma cells of mouse or rat origin. Although hybrids can be derived that express
`the antibody of the donor, the expression is quickly lost. It has been a common
`experience of ours and of several other laboratories that the stabilization of the
`expression of interspecific hybrids is possible but rather difficult. So the mouse x
`rat hybrids seem to represent the exception rather than the rule.
`An interesting point concerning the stability of expression of rat antibodies in
`
`9 of 21
`
`BI Exhibit 1056
`
`

`

`402
`
`C. Milstein
`
`the mouse x rat hybrids is that it happens in spite of the preferential loss of rat
`chromosomes. Table 3 summarizes the karyotype analysis of 23 hybrid clones,
`which shows that losses of chromosomes were mostly due to rat chromosomes.
`Furthermore, the small mouse chromosome losses were fairly random, particularly
`among trisomies. On the other hand the losses of rat chromosome were not
`random and some of the chromosomes (i.e. 1, 2, 4, 5, 13, X) were present in all
`cells. We have concluded that the expression of the rat antibodies is stable
`because the immunoglobulin genes themselves are located on stable chromosomes
`(Schroder et al. 1980). Other controlling elements must be also on stable rat
`chromosomes or are not species-specific.
`
`TABLE 3. KARYO'fYPES OF ES'fABLCSHED MOUSE MYELOMA - RAT
`SPLEEN CELL HYBRID CLONES
`(Data. ta.ken from SchrOdor et al. (1980).)
`
`chromosomes
`
`X63 mouse myeloma
`normal rat
`23 hybrid clones
`mouse chromosomes
`rat chromosomes
`
`total
`
`loss
`
`63-64
`42
`74--89
`55-60
`(13) 15-26 (28)
`
`17-·31
`3-8
`14-27
`
`The mouse x human hybrids also show non-random losses of chromosomes,
`and chromosome 14, which is thought to include the H chain gene locus, is preferen­
`tially retained (Croce et al. 1980). Therefore, the preferential loss of human
`chromosome does not seem to be the reason for the preferential loss of the expres­
`sion of the human immunoglobulins, at least as far as the heavy chains are con­
`cerned.
`The disappointing results with interspecific hybrids have reinforced the idea
`that for the production of human (or most other species') monocloMl antibodies
`one needs to use appropriate parental myeloma lines derived from the same species.
`The derivation of such lines is tedious and takes a long time, and for human
`myelomas it has proved difficult. The human myelomas do not grow easily in
`tissue culture. We, and quite a number of other laboratories, have been searching
`for a suitable line. But they are very slow growers and do not give viable hybrids.
`A promising exception may be a line recently derived (N. Kaplan, personal com­
`munication). But a more general solution to the problem would be desirable. We
`still need to define more clearly the reasons for the phylogenetic restrictions and
`see how these can be circumvented. In collaboration with G. Galfre we have been
`experimenting with double fusions". Mouse x human hybrids are first prepared
`and from them hybrids expressing human immunoglobulins are selected. These
`are then tested to see if they can be used as myeloma parents to fuse with human
`
`10 of 21
`
`BI Exhibit 1056
`
`

`

`The Wellcome Foundation Lecture, 1980
`
`403
`
`lymphocytes. The hope is to select a mouse-human line suitable for the derivation
`of (mouse-human) x human hybrids which will express human antibodies in
`a more stable manner.
`
`MONOCLONAL ANTIBODIES AS SPECIFIC REAGENTS
`The potential of the hybrid myeloma technique stems from the fact that the
`antibody-producing cells can be cloned, that the individual clones can be main­
`tained indefinitely and that they are capable of large-scale production of mono­
`clonal antibody. These, therefore, are welJ defined chemical reagents reproduced
`at will. This contrasts with the conventional antisera, which represent an ill
`defined mixture, a mixture that can never be reproduced once the original supply
`is exhausted. For this reason monoclonal antibodies are likely to substitute con­
`ventional antibodies as standard reagents in laboratory practice.
`An interesting example is provided by the well established ABO grouping re­
`agents used for blood transfusions. The reagents used at present are obtained
`from human serum, preferably of hyperimmunized human volunteers. In countries
`like the United Kingdom, where this procedure has not been adopted, the reagents
`tend to be of considerably lower quality. The preparation of ABO typing reagents
`in the U.K. uses 1200 1 of human serum annually from 6000 blood donations. Each
`donation must be carefully scrutinized for the presence of unwanted antibody
`specificities that are likely to be expressed together with the anti-A or anti-B
`specific reactions. From among a variety of hybrid myeloma clones we have
`chosen one (MH2/6D4) that displays the necessary qualities as a provider of
`a standard anti-A reagent (Voak et al. 1981 ). The McAb was used with 1421
`samples in manual tests, including 169 cord samples. They all gave satisfactory
`reaction and no false reactions were detected. In automatic tests, using the Auto
`Grouper 160 machine, 1911 random samples including many A1B and A2B gave
`satisfactory results. The reagent was prepared as the spent medium of stationary
`phase MH2/6D4 cells. We have concluded that mass culture production of this
`monoclonal anti-A provides a cost-effective model for the use of monoclonal anti­
`bodies as a new generation of better-standardized potent reagents for routine
`blood typing. The reagent is at present being tested in the U.K. on a national
`scale.
`The preparation of monoclonal antibodies does not require pure antigens and
`this is the most powerful aspect of the hybrid myeloma method. It therefore
`represents a new strategy for the detection, characterization and purification of
`unknown antigens or minor components of a mixture. Purification of the permanent
`hybrid cell line capable of producing unlimited amounts of specific antibody is
`achieved by cloning the somatic cell hybrids (figure 3). The specific antibody can
`then be attached to Sepharose or other solid supports and used as immunoad­
`sorbants to purify the corresponding antigen by affinity chormatography. A strin-
`
`11 of 21
`
`BI Exhibit 1056
`
`

`

`404
`
`C. Milstein
`gent test of this scheme has been the purification of human leucocyte interferon
`(Secher & Burke 1980). When this project was started, no pure interferon was
`available. Preparations enriched with interferon were used to immunize mice.
`Spleen cells from such animals were fused to myeloma cells and the hybrids tested
`for. the ability of the spent culture medium to neutralize interferon. The biological
`assay is rather lengthy and not very suitable for screenings of this type, and it was
`
`TABLE 4. PuRIFICATION OF CRUDE INTERFERONt (IF) BY SEPHAROSE
`IMMUNOADSORDANT CHR0111ATOGRAPHY WITH A MONOCLONAL ANTIBODY
`(Data taken from Secher & Burko (1980).)
`
`volume/ml
`IF titre/(unit/ml)
`total activity /unit
`total protein/mg
`specific activity /(unit/mg)
`purification factor
`yield(%)
`
`applied to
`column
`
`100
`7.2 x 104
`7.2 x 108
`220
`3.3 x 10•
`
`eluted
`
`0.5
`1.4 x 107
`7.0 x 106
`0.04
`1.8 x 108
`5300
`97
`
`t Culture medium from stimulated Nama!va cells after removal of material that precipi.
`tated at pH 2.
`
`�;:j) �NS1/Ag4.1
`
`Clone etc ...
`
`Frotrn:i;: 5. Monoclonal antibody cascades provide a means fo1· the dissection of all antigens
`of a complex mixture. (Taken from Milstein & Lennox (1980) .)
`
`12 of 21
`
`BI Exhibit 1056
`
`

`

`The Wellcome Foundation Lecture, 1980
`
`405
`
`supplemented with very simple assays to detect the secretion of immunoglobulin
`regardless of antibody activity. Eventually a hybrid clone was isolated and large
`amounts of antibody were prepared and used to make immunoadsorbant columns.
`With these columns it was possible to purify interferon in excellent yields. The
`purification achieved in a single passage of crude material through the columns
`was 5000-fold (table 4). Furthermore, the same antibody is being successfully used
`in a radioimmunoassay of interferon, which will be invaluable in the control of
`interferon production a.ud ptu·ifica.tion as well as for clinical studies.
`The monoclonal antibody strategy for the purification of biological products is
`being taken one step further as a means of isolating all the components of bio­
`logical extracts. In the monoclonal antibody cascade purification (figure 5) an
`unknown mixtme is used to immunize an animal and a random set of monoclonal
`antibodies is prepared. lmmunoadsorbant columns are then prepared and used to
`.remove the corresponding antigens from the original mixture. The rema.ining
`simplified mixture is then used to immunize new animals and derive new hybrid
`myeloma.s. These, in turn, are used to prepare new immunoadsorbant columns to
`simplify further the original mixture before repeating a new cycle. In this way the
`minor antigenic elements of the mixture are slowly enriched and eventually
`monoclonal antibodies to them can also be prepared. So hybrid myelomas provide
`the analytical tool for the characterization of all the individual components and
`at the same time are the permanent source for the preparation of immunoad­
`sorbant columns and successive pmification of all the elements of the mixture.
`
`MONOCLONAL ANTIBODIES AND CELL S URFACE ANTIGENS
`The specificity of the cell surface is a subject of increasing interest in terms of
`functional properties like transport phenomena and cell-cell interactions. But,
`regardless of their functional properties, cell surface antigenic determinants con­
`stitute a simple means of defining cell types and cell lineages. Since the early
`successes with the preparation of monoclonal antibodies to define hitherto un­
`known differentiation antigens (Williams et al. 1977; Springer et al. 1978), the
`application of the method is being extended by many laboratories. This is an area
`in which fa.st and rather dramatic progress is being made. Of particular interest to
`clinical immunologists are monoclonal antibodies defining subsets of lymphoid
`cells in humans (McMichael et al. 1979; Reinherz et al. 1979). Monoclonal anti­
`bodies to cell surface antigens are valuable not only in defining the functional dif­
`ferences of cell subpopulations (White et al. 1978; Reinherz et al. 1980) but also as
`a means to classify leukaemias and to define the anomalous expression of antigenic
`determinants in malignant cells (Bradstock et al. 198oa b,). There are monoclonal
`antibodies, like YD1/23, that react well with certain types of leukaernic cells but
`seem not to react with bone marrow precursors (Janossy et al. 1980). Such reagents
`can be tested for their capacity to remove leukaemic cells from the bone marrow
`of patients. This could increase the effectiveness of autologous marrow transplants
`
`13 of 21
`
`BI Exhibit 1056
`
`

`

`406
`
`C. Milstein
`taken from patients in remission. It has also been suggested that appropriate
`monoclonal antibodies, and YD1/23 is a candidate, may be used in allogenic bone
`marrow transplantations to eliminate cells provoking graft versus host disease
`without removal of the stem and precursor cells.
`However, these and other possible uses are contingent on the more fundamental
`use of monoclonal antibodies to define and characterize the biochemistry and
`antigenicity of cell membranes. 'I'his is true not only for normal and malignant
`cells but also for para.sites and bacteria.
`
`MONOCLONAL ANTIBODIES AND IMMUNOC HEMISTRY
`The nature of antigen-antibody interactions has been one of the fascinating
`subjects of classical immunology. The vast majority of these studies were per­
`formed with mixtures of molecular species. 'l'he use of monoclonal antibodies
`requires a fresh approach to the interpretation of serological reactions. For
`instance, the lattice theory (Marra.ck 1938) predicts that immunoprecipitation
`reactions should not be observed with monoclonal antibodies. Precipitation
`requires the formation of three-dimensional lattices and as monoclonal antibodies
`bind to a single determinant no such lattices can be formed unless the same
`determinant is repeated in the molecule, as in polymeric structures. Figure 6
`illustrates that individual monoclonal antibodies fail to precipitate human IgG
`but mixtures of them do not. This may be the first formal proof, if one is needed,
`of the lattice theory. On the other hand, we have observed that three monoclonal
`antibodies directed against different sites of human C3 complement failed to give
`precipitates (Lachmann et al. 1980). Non-precipitating antibodies are an old
`serological puzzle and monoclonal antibodies may be the way to solve it.
`Complement-mediated cytotoxicity is affected not only by the class of anti­
`bodies but also by their local distribution on the cell surface. Local concentrations
`can be increased by multiple antibodies recognizing neighbouring determinants of
`the same antigen. The importance of such synergistic effects is illustrated in
`figlll'e 7 by monoclonal antibodies that recognize two independent sites of t.he
`same rat histocompatibility antigens. These were called P and S sites. Individual
`monoclonal antibodies to either site were poorly cytotoxic or not cytotoxic but
`the blend was stongly cytotoxic (Howard et al. 1979).
`In a more general way, the interactions between antigen and antibodies have
`been always clouded by the heterogeneity problem. This was true when measuring
`thermodynamic and kinetic parameters as well as following serological reactions.
`The studies of antigen-antibody interactions by crystallography and other modern
`physical methods need no longer depend on a few myeloma proteins for which
`a more or Jess reasonable binding to randomly tested ligands has been detected. It
`will be most instructive to be a.hie to define the multiplicity of antibody structures
`that are capable of interacting with a single antigenic determinant. And here I am
`referring to the general problem of molecttlar recognition and the diversity of
`
`14 of 21
`
`BI Exhibit 1056
`
`

`

`The Wellcome Foundation Lecture, 1980
`
`407
`
`protein-ligand interactions. The use of hybrid myelomas to define the complete
`repertoire of antibodies to single antigens (Reth et al. 1977) is now expanding very
`rapidly. Although its major motivation has been the problem of genetic diversity
`and idiotypic regulation, the purely structural aspect of the diversity is of no less
`interest.
`
`b
`
`a
`
`3
`
`e
`
`b
`c
`a J2+3
`. /
`f ·
`e
`
`FIGURE 6. Double·diffusion analysis with monoclonal antibodies. (Top) The centre well
`contains hurnan serum (diluted ! : 3). The onter wells contain ascites Au id from mice
`bearing the following hybrid myeloma clones secreting anti.human IgC antibodies:
`1, N'H2/17 (anti·y); 2, NH:3/130 (anti-y); 3, Nl-13/41 (anti-1<:); 4, NH3/75 (anti-y).
`Notice that a single McAb does not givo a prccipitin line. However, between two neigh­
`bouring wells containing different j\foAb ( l and 2) a precipitation zone occurs where the
`two McAb mix. (Bottom) The centre well contains NH3/41 (anti-Kon the left), NH3/130
`(anti-y on the right) and a mixt1ire of the two in the centre. The outer wells contain
`myelomaprotcinsof thefollowingsubclasses and types: a, IgC 1