`Printed in Great Britain
`
`I
`
`REVIEW ARTICLE
`
`Some properties and applications of monoclonal antibodies
`
`Paul A. W. EDWARDS
`Ludwig Institutefor Cancer Research (London Branch), Haddow Laboratories, Royal Marsden Hospital,
`Sutton, Surrey SM2 SPX, U.K.
`
`Introduction
`A monoclonal antibody is the antibody produced
`by a single clone of B lymphoid cells:
`all the
`antibody molecules have the same amino acid
`sequence and hence the same binding properties. A
`myeloma is a tumour of antibody-secreting cells,
`apparently derived from a single cell, and is thus a
`clone of cells producing a monoclonal antibody. For
`some years it has been possible to grow myelomas of
`rodents as permanent cell lines in tissue culture or as
`tumours in animals, but in general they produce
`antibody to unknown or uninteresting antigens.
`Kohler & Milstein (1976; see also Milstein, 1980, for
`historical background) found a way to construct
`hybrid myelomas-'hybridomas'-that make anti-
`body to an antigen of choice. In their technique,
`lymphocytes from an animal making useful anti-
`bodies are fused with cells of a stock myeloma cell
`line to give hybrid myeloma cells. Each hybrid
`produces the antibody specified by the lymphocyte
`that took part in the fusion; the myeloma confers on
`the hybrid cell the ability to grow in culture or as a
`tumour in an animal.
`
`Production of monoclonal antibodies
`Outline
`A mouse (or rat) is immunized with antigen, say
`human erythrocytes. Some 3-4 days after a final
`immunization, the animal's spleen contains many
`lymphocytes making antibody to erythrocytes. Cells
`from the spleen are mixed with stock myeloma cells,
`in the presence of poly(ethylene glycol), to make
`hybrid cells. The stock myeloma cell line used is
`usually a selected variant that has lost the ability to
`produce its own immunoglobulin or at least the
`immunoglobulin heavy chain (Table 1). The hybrids
`would otherwise produce this immunoglobulin as
`well as that of the immunized lymphocyte. In order
`that the hybrids can be grown in culture without
`being swamped by unhybridized myeloma cells, the
`myeloma cell line employed is also a mutant that
`Abbreviations used: SDS, sodium dodecyl sulphate; Ig,
`immunoglobulin.
`
`lacks hypoxanthine phosphoribosyltransferase (EC
`2.4.2.8; HPRT or HGPRT), an enzyme of a salvage
`pathway of nucleic acid metabolism. These cells die
`in medium containing hypoxanthine, aminopterin
`and thymidine ('HAT medium'; Littlefield, 1964)
`because the aminopterin blocks nucleic acid bio-
`synthesis and they are unable to utilize the hypoxan-
`thine. Hybrid cells, however, are wild-type and they
`survive and grow. Unhybridized spleen cells die out
`in a few days. Fusions yield hundreds, sometimes
`thousands, of individual hybrid cell lines, typically
`10% of which make antibody to the antigen used to
`immunize the mouse, fewer with poor antigens. After
`10-14 days the growing hybrids are tested for the
`production of antibody to the antigen, in this case
`erythrocytes. Culture medium in which the hybrids
`have been growing and secreting
`antibody
`is
`incubated with erythrocytes, the erythrocytes are
`washed and the antibody that has bound to them is
`then detected, for example by binding of radio-
`labelled rabbit anti-(mouse immunoglobulin) anti-
`body. Interesting hybrids are identified and cloned.
`Antibody may be obtained from the hybrids as
`'culture supernatant' or ascitic fluid (see Table 2
`below). Culture supernatant is medium in which the
`cells have been growing and secreting. It contains
`antibody, usually together with foetal calf serum
`(which may contain bovine immunoglobulin), but no
`irrelevant mouse or rat immunoglobulin. (It is now
`possible to use serum-free culture medium: Iscove &
`Melchers 1978; Chang et al., 1980; Klinman &
`McKearn, 1980.) Ascitic fluid is produced when
`hybrids are grown in the peritoneal cavity of mice or
`rats as appropriate. It contains a variety of serum
`proteins and irrelevant immunoglobulin, but they do
`not interfere with most applications. Some 1-5 ml of
`ascites fluid can be obtained from a mouse, so that
`100mg batches of antibody can be made without
`difficulty. Rats give much larger volumes of ascites
`fluid.
`The hybrid cells can be stored in liquid N2 and
`recultured indefinitely, provided variant cells that no
`longer produce
`antibody
`eliminated when
`are
`necessary by recloning.
`Mouse, rat and human hybridomas can now be
`
`Vol. 200
`
`0306-3283/81/100001-10$01.50/1 (© 1981 The Biochemical Society
`
`Merck Ex. 1074, pg 1595
`
`
`
`2
`
`Species
`Mouse
`
`Rat
`Human
`
`Table 1. Some myeloma cell lines usedfor making hybrids
`Immunoglobulin
`produced
`Light chain only
`None
`None
`None
`Light chain only
`IgE
`IgG
`IgG
`
`Myeloma cell line
`
`NS1
`Sp20
`FO
`X63.Ag8.653
`Y3
`SKO-007
`GM1500 6TG-A1 2
`LICR LON/HMy2
`
`P. A. W. Edwards
`
`Reference
`Kohler & Milstein, 1976
`Schulman et al., 1978
`Fazekas de St. Groth & Scheidegger, 1980
`Kearney et al., 1979
`Galfre et al., 1979
`Olsson & Kaplan, 1980
`Croce et al., 1980
`P. A. W. Edwards & M. J. O'Hare, unpublished work
`
`made, fusing lymphocytes with a myeloma of the
`same species (Table 1). Some stable hybrids were
`made between rat or human lymphocytes and mouse
`myelomas before rat and human myelomas were
`available (Howard et al., 1980; Schlom et al., 1980).
`No successful hybrids have been made with rabbit
`lymphocytes (Yarmush et al., 1980). For large scale
`production, rat hybrids are preferable to mouse, but
`for research purposes, mouse hybrids have the
`advantage that many more reagents and procedures
`for mouse immunoglobulin are available (Hudson &
`Hay, 1980). Human hybrids are desirable for the
`study of human immune responses, and perhaps for
`producing human antibody for therapeutic applica-
`tions if rodent antibody is not available or not
`acceptable.
`The production of a new monoclonal antibody
`requires a lot of time and a wide range of experience,
`materials and equipment, both for making and for
`characterizing the antibody. The antigen or mixture
`of antigens used must be known to give a good
`conventional immune response, and a sufficiently
`reliable, sensitive and rapid assay must be available
`for screening. However, in experienced hands, an
`appropriately designed fusion experiment is usually
`successful.
`
`Some notes on methods
`For an extensive review of methods the reader is
`referred to Goding (1980). Fazekas de St. Groth &
`Scheidegger (1980) give a useful protocol for fusion,
`while Hudson & Hay (1980) is an indispensable
`manual of laboratory procedures.
`Plating strategy and cloning. Cloning a single
`hybrid
`under
`adverse
`conditions
`is
`time-
`as
`consuming as performing a fusion. Also, if one is
`screening for an antibody that discriminates between
`two antigens, a specific clone will be missed if it is
`mixed with a clone that produces non-specific
`protocols
`antibodies.
`Hence fusion
`should
`be
`designed to generate clonal hybrids from the outset,
`usually by plating on feeder cells (non-dividing cells
`such as macrophages or thymocytes that provide a
`favourable environment for the hybrids: Goding,
`
`1980) in a large number of wells in microtest plates
`so that only about 10% of the wells contain hybrids
`(e.g. Fazekas de St. Groth & Scheidegger, 1980). An
`alternative is to plate the hybrid cells in agar so that
`they grow as clones (Sharon et al., 1979; Edwards,
`1980). Individual positive clones can be identified by
`a replica-plating assay (Sharon et al., 1979), or the
`agar can be in 24-well plates so that testing for
`antibody is performed conventionally; clones are
`picked from wells containing antibody and retested
`after isolation (Edwards, 1980).
`Hybrids should be recloned by limiting dilution,
`rather than in agar (for protocols, see Goding, 1980,
`or Hudson & Hay, 1980), since agar gives a lower
`cloning efficiency and so may select for unwanted
`cells that are more vigorous than the desired hybrid.
`In addition, colonies of cells picked from agar are
`not always pure clones.
`Screening. When screening, it
`is essential to
`identify desirable hybrids quickly, as the labour of
`maintaining and characterizing them is greater than
`that needed to perform another fusion. The screening
`assay must be sensitive enough to give a strong
`l,ug of antibody/ml if con-
`response for about
`centration methods are not to be used. As discussed
`below, some assays do not usually work with
`monoclonal antibodies, for example, cytotoxicity
`tests (Howard et al., 1979).
`Antibodies are most often detected by binding
`them to the target antigen, followed by detection
`251I-labelled or enzyme-linked antibody to
`with
`al.,
`mouse immunoglobulin
`(Williams et
`1977;
`Goding, 1980). However, as far as possible, the
`antibody should be tested in the way that it is to be
`used, and a more direct approach may save labour.
`For example, immune precipitation of antigen has
`been used for screening after a preliminary rapid test
`for abundant antibody binding to Protein A of
`al.,
`Staphylococcus
`(Brown
`1980).
`et
`aureus
`Staining of tissue sections or cells by immuno-
`(Goding,
`1980)
`immunocyto-
`fluorescence
`or
`chemical methods (e.g. Schlom et al., 1980) are
`screening methods for
`suitable
`antibodies
`that
`mark particular cells or special structures such as
`
`1981
`
`Merck Ex. 1074, pg 1596
`
`
`
`Properties and applications of monoclonal antibodies
`
`3
`
`intermediate filaments; scarce cells, that may not be
`abundant enough to be a target for a binding assay,
`can be detected this way (McIlhinney et al., 198 1).
`Among tricks suggested for streamlining screen-
`replica-plating methods, using either
`ing
`a
`are
`al.,
`blotting technique (Sharon et
`1979) or a
`commercially available sample-transferring 96-well
`plate (Schneider & Eisenbarth, 1979; Bankert et al.,
`1980b), and autoradiography
`of plates
`as
`an
`alternative to y-counting after binding '251I-labelled
`second antibody (Parkhouse & Guarnotta, 1978).
`Culture fluid can be concentrated by rapidly freezing
`I ml in a narrow tube, then allowing it to thaw while
`being centrifuged at about 10OOg at room tem-
`perature. The concentrate is found at the bottom of
`the tube. Samples from 96-well plates can be pooled
`along rows and columns for preliminary screening
`(Brown et al., 1980).
`Higher vield of useful hvbrids. Several attempts
`have been made to increase the proportion of
`hybrids that make antibody to the antigen of choice,
`but none are yet established methods. An improved
`immunization method for soluble antigens has been
`claimed by Staehli et al., (1980), and a good yield of
`antibodies for staining sections of fixed tissue can be
`obtained by immunizing with fixed cells (R. A. J.
`Mcdlhinney, personal communication). Middleton et
`al. (1980), in a preliminary report, induced tolerance
`to human B cells in mice, then immunized with a T
`cell fraction and obtained a greater proportion of
`antibodies specific to T cells. Yelton et al. (1978)
`fused cells that had been selected for their ability to
`bind the chosen antigen. Particularly interesting is
`the report of Bankert et al. (1980a) that spleen cells
`fuse with hapten-coated myeloma cells to give a high
`proportion of anti-hapten hybrids.
`Problems. It is widely believed that the success of
`the cell fusion depends primarily on the condition of
`the myeloma cells used. If the yield of hybrids is low,
`fresh cells should be obtained from a successful
`laboratory. The method of routine passage of the
`myeloma may be critical: high density and low
`serum or spinner culture are recommended, followed
`by optimal conditions just before fusion (G. Galfre,
`communication).
`personal
`Another
`problem
`is
`contamination
`with Mvcoplasma, which
`causes
`cultures to die out gradually. It appears to be
`controllable by injecting infected cells into mice as
`for the production of ascites (Hudson & Hay, 1980)
`and recovering the apparently Mvcoplasma-free cells
`when the ascites is drained (P. A. W. Edwards,
`unpublished work; see also the methods of Marcus
`et al., 1980 and Schimmelpfeng et al., 1980).
`Apparent instability of hybrids, i.e. a loss of
`antibody production, is thought to be caused by
`non-secreting variants or unrelated hybrids over-
`growing the desired hybrids. It should be possible to
`control this by recloning the hybrid cells (Howard et
`
`Vol. 200
`
`al., 1980; Schlom et al., 1980). Sometimes when a
`sample of hybrid cells is thawed from storage in liquid
`N2 almost all the cells die. Such cultures will usually
`re-establish if routinely, on thawing, part of the
`sample is seeded at low cell density, say 104 cells/ml,
`on feeder cells (P. A. W. Edwards, unpublished
`work; for feeder cell recipes see Goding, 1980).
`
`Special properties of monoclonal antibodies
`Monoclonal antibodies are not just exceptionally
`high quality antibodies. To the research worker, the
`most important benefit of the monoclonal antibody
`technique is that it makes possible the identification,
`assay, marking and purification of antigens that
`have not been purified, and that are in fact usually
`completely unknown at the outset. This is because
`antibodies
`the method generates
`individual
`to
`components of a mixture. For example, in a mouse
`individual
`immunized with human erythrocytes,
`lymphocytes will be making antibody to individual
`erythrocyte. When hybrid
`of the
`components
`myelomas are made with the lymphocytes, the
`hybrids produce these antibodies. From such a
`hybridization experiment we obtained two mono-
`clonal antibodies that bind specifically to be erythro-
`cyte membrane glycoprotein glycophorin A, each
`recognizing a different part of the molecule, as well
`as other antibodies that bind to other molecules on
`the surface of the erythrocyte (Edwards, 1980;
`Anstee & Edwards, 1981). The various antibodies
`can now be used to purify these surface molecules or
`to detect them on other cells. This example also
`illustrates the ability of the method to produce
`antibodies to individual antigenic determinants on a
`single molecule, which is particularly useful when
`some determinants are unique to that molecule while
`other determinants are also present on different
`molecules.
`Of rather specialized interest is the possibility of
`capturing rare or inaccessible immune responses. It
`should be possible by screening large numbers of
`hybrids to obtain good monoclonal antibodies to
`antigens that give weak or widely cross-reactive
`conventional antisera, although the effort required
`should not be underestimated. Antibodies produced
`to tumours may be studied by making hybrid
`myelomas with lymphocytes from lymph nodes
`adjacent to tumours (Schlom et al., 1980).
`Monoclonal antibodies also have some revolution-
`ary practical advantages over conventional antisera,
`but they also have limitations. Table 2 summarizes
`the differences. All the molecules of a monoclonal
`antibody are identical in amino acid sequence and
`hence in binding properties; the antibody can thus
`have exceptional specificity. A monoclonal antibody
`is insignificantly contaminated by other irrelevant
`antibodies, and it will therefore give a low back-
`ground in assays and staining reactions. It can
`
`Merck Ex. 1074, pg 1597
`
`
`
`4
`
`P. A. W. Edwards
`
`Table 2. Summary ofthe properties ofmonoclonal and conventional antibodies
`Monoclonal antibody
`
`Property
`Useful antibody content
`Irrelevant immunoglobulin
`Other serum proteins
`
`Conventional antiserum
`0.1-1.0 mg/ml
`10mg/ml
`Present
`
`Ascites fluid
`0.5-5 nig/mI
`About 0.5-1 mg/mIt
`Some presentt
`
`Culture supernatant
`5-25,g/ml
`In principle, none*
`Normally 10% (v/v) foetal
`bovine serum*
`One antigenic determinant of one component of
`immunizing material
`Invariant
`
`Usually absent but complete if binds to a common
`determinant
`
`May not work
`
`All antigenic determinants
`of all components of
`immunizing material
`Varies between batches
`
`Partial with antigens
`bearing common antigenic
`determinants
`Applicable
`
`Binds to
`
`Reproducibility of specificity
`and affinity
`Cross-reactions with other
`antigens
`
`Applicability of conventional
`immunological procedures
`e.g. precipitation of antigen
`Class and subclass of
`immunoglobulin
`Physical properties of
`immunoglobulin
`Kinetics of binding
`
`Yes
`Typical mixture of all
`
`No
`One only; may be any
`
`Typical spectrum
`
`Individual property of antibody
`
`Typical spectrum though
`average behaviour varies
`
`Wide variation expected: selected by screening method
`
`* See text.
`t Variable: contains a filtrate of serum plus a variable amount of whole serum due to haemorrhage.
`
`usually be purified easily so that radioactive labelling
`and conjugation are easily accomplished and, in
`contrast with conventional antisera, it is easy to
`35S
`incorporate
`14C
`3H,
`into
`the antibody
`or
`molecules biosynthetically, by growing the cells in
`medium containing labelled amino acids. A mono-
`clonal antibody can be reproduced and distributed
`indefinitely and its properties will always be the
`same. Hence radioimmunoassays and related techni-
`ques may be permanently standardized on an
`international basis. The antibody may even replace
`the antigen as a standard. Although monoclonal
`antibodies are expensive to make and characterize in
`the first instance, the cost of continued production is
`comparable with, or probably less than, the cost of
`producing conventional antisera. Quality control
`may also be easier because the presence of the
`monoclonal antibody can be checked, for example,
`by its characteristic isoelectric focusing pattern and
`its activity measured by a convenient titration. So, if
`the initial cost can be covered, it is to be expected
`that monoclonal antibodies will gradually replace
`many conventional commercial antisera.
`Limitations
`Monoclonal antibodies have some important
`limitations. For example they generally do not work
`in that most basic of immunological tests, the
`Ouchterlony double-diffusion assay, in which anti-
`gens and antibody are allowed to diffuse towards
`
`each other in agar to form a precipitate. However,
`their limitations can usually be overcome provided
`they are understood.
`A classical antiserum contains antibodies to a
`number of antigenic determinants on its target
`antigen, whereas a monoclonal antibody will only
`bind to one determinant. An antiserum therefore
`provides quite a precise identification of its target
`antigen: an unknown molecule that can compete for
`all its antibody molecules, as in a classical radio-
`immunoassay, will almost certainly be identical to
`the known target antigen. In contrast, a monoclonal
`antibody is unable to distinguish between a group of
`different molecules that
`antigenic
`all bear the
`determinant it recognizes, or even between deter-
`minants that have a sufficient structural similarity to
`bind the antibody. In a radioimmunoassay for a
`peptide hormone, where part
`of the
`peptide's
`sequence may be common to several hormones, such
`complete cross-reactions might be a serious problem
`(Bundesen et al., 1980). Also, monoclonal anti-
`bodies will not usually precipitate their target antigen
`because they can only cross-link
`antigens into
`dimers rather than form a lattice-hence the failure of
`Ouchterlony assays. Classical antisera fix comple-
`ment more readily than do monoclonal antibodies
`because complement requires at least two bound
`antibody molecules on neighbouring determinants
`(Howard et al., 1979; see below). A solution to these
`problems is to use suitable blends of monoclonal
`
`1981
`
`Merck Ex. 1074, pg 1598
`
`
`
`Properties and applications of monoclonal antibodies
`
`5
`
`antibodies; for both precipitation (Jefferis et al.,
`1980) and cytotoxicity (Howard et al., 1979) it has
`been shown that a mixture of two monoclonal
`antibodies is sufficient.
`A conventional antiserum contains a variety of
`antibody molecules of different immunoglobulin
`class, with different physical properties, different
`affinities for antigen, etc. Many practical procedures
`rely on this-for example, some of the antibodies
`will normally be IgG and a proportion will have high
`affinity. A monoclonal antibody has a unique
`structure and so it may not satisfy some of these
`conditions.
`It may, for example, be unusually
`susceptible to denaturation by freezing or iodination,
`or it may elute from a DEAE-cellulose column at an
`atypical point (Mason & Williams, 1980). In assays
`with monoclonal antibodies it is particulary impor-
`tant to select appropriate concentrations of antibody
`and antigen, and the time and temperature of the
`reaction must be carefully
`chosen. Mason &
`Williams (1980) have studied the binding of three
`monoclonal antibodies to the rat lymphocyte. Under
`typical assay conditions, binding was determined by
`rates of association and dissociation, and did not
`reach equilibrium in 1 h at 4°C. The antibodies had
`association rate constants of about 105-106 M-I. S-1,
`slow enough for the slowest to have a half-time of
`binding of 5 h when the concentration of antigen was
`0.5 nm (107 thymocytes/ml). The antibodies dis-
`sociated about ten times faster at room temperature
`than at 40C, though association rates changed only
`about 2-fold. So for example, if a high concentration
`of antigen is used in a screening assay, antibodies
`may be obtained that will bind too slowly for rapid
`detection of low concentrations of antigen. Mason &
`Williams (1980) also showed that an antibody that
`binds satisfactorily to a cell surface, where it can
`bind bivalently, may nevertheless dissociate rapidly
`from antigen that has been solubilized from the
`membrane, to which it can only bind monovalently.
`Such an antibody may not be suitable for purifi-
`cation of the antigen by affinity chromatography.
`
`Some applications
`I have chosen examples showing the variety of
`uses of monoclonal antibody technology.
`The cell surface
`To date, the fields most affected by monoclonal
`antibody technology have been the cell surface and
`the related area of separating and analysing cell
`populations from tissues by using cell
`surface
`antigens. The ability
`to make antibodies
`that
`recognize only one antigenic determinant in
`a
`mixture has made it possible to obtain antibodies to
`individual components of the cell surface most of
`which have not previously been described. The
`composition of the cell surface may be analysed in
`Vol. 200
`
`this way, and individual cell types within a popu-
`lation can be isolated for functional studies. A good
`example is the work on the rat T lymphoid cell
`surface by Williams and co-workers (Brown et al.,
`1981; Brideau et al., 1980; Williams, 1980).
`When the plasma membrane of the rat thymocyte
`(immature T lymphocyte in the thymus) is analysed
`by SDS/polyacrylamide-gel electrophoresis, three
`main bands that stain for carbohydrate and are of
`apparent molecular weights 25 000, 95000 and
`150000 are seen. They are known respectively as
`Thy-1, W3/13 and the leucocyte-common antigen.
`Only Thy-I
`could be purified by conventional
`methods. Monoclonal antibodies were made by
`immunizing mice with thymocyte plasma membrane,
`or glycoprotein fractions from it. Individual anti-
`bodies were obtained to all the three major glyco-
`proteins and to the rat Ia antigens, but in addition
`antibodies to four previously unknown components
`of the thymocyte surface were obtained. The
`antibodies to the leucocyte-common antigen and
`W3/13 were used with remarkable success to purify
`the respective glycoproteins; W3/13 glycoprotein
`was purified 7000-fold simply by affinity chromato-
`graphy on the monoclonal antibody, followed by gel
`filtration (Brown et al., 1981).
`These antibodies have also been used to deter-
`mine the number of copies present per thymocyte of
`the glycoproteins to which they bind, and to detect
`them, or at least their antigenic determinants, on
`other cell types, i.e. brain, kidney, liver and lympho-
`cytes, myeloid cells and erythrocytes (Williams,
`1980). The three major carbohydrate-bearing glyco-
`proteins of rat thymocytes occur on a quite
`restricted range of cell types: Thy-I and W3/13 are
`not even present on all leucocytes;
`leucocyte-
`common antigen is not a substantial constituent of
`kidney,
`liver or brain.
`All of the other four
`glycoproteins identified are at least as restricted in
`their distribution (Williams, 1980).
`Two of the antibodies obtained to as yet uniden-
`tified antigens, MRC OX8 and W3/25, were found
`to bind to complementary subpopulations of T
`lymphocytes (Brideau et al., 1980). All mature T
`cells bound either W3/25 or MRC OX8, but not
`both. So without any prior knowledge of this
`subdivision of T lymphocytes, two reagents were
`obtained that could be used to identify and separate
`the two types of cell using the fluorescence-acti-
`vated cell sorter*. Their role in the immune response
`* In the fluorescence-activated cell sorter (FACS) a
`stream of cell suspension is broken into minute droplets
`each of which contains a single cell. A laser illuminates
`the stream, and the light-scattering and fluorescence
`characteristics of each illuminated cell are instantly deter-
`mined. A cell with particular characteristics can be
`selected by deflecting the fall of its droplet with an electric
`field (Herzenberg et al., 1976).
`
`Merck Ex. 1074, pg 1599
`
`
`
`6
`
`P. A. W. Edwards
`
`was examined by injecting the separated cells into
`irradiated rats. Cells bearing the antigen recognized
`by W3/25 had helper function and gave a graft-
`versus-host reaction in the assay used. The cells
`labelled by MRC OX8 were required for a parti-
`cular form of suppression of antibody response
`(White et al., 1978; Brideau et al., 1980).
`Similar work is proceeding on the mouse (e.g.
`Ledbetter & Herzenberg,
`1979;
`Coffman &
`Weissmann, 1981) and particularly on the human
`(reviewed
`by Reinherz & Schlossman,
`1981;
`Greaves, 198 1; Janossy & HofThrand, 198 1) haemo-
`poeitic systems.
`At present work on solid tissues such as epithelia
`and brain lags behind; antibodies with interesting
`specificities have been reported but generally not yet
`Clearly there is much to come, e.g.
`exploited.
`biochemical analysis of the surfaces of cells not of
`bone marrow origin, marking of cell types in the
`nervous system (Barnstable,
`1980),
`analysis of
`embryonic development with surface markers (e.g.
`Stern
`1978; Solter & Knowles,
`al.,
`1978;
`et
`Mcllhinney et al., 1981), and the selective culture of
`individual cell types from complex tissues (Edwards
`et al., 1980).
`Functional effects. A serious limitation of current
`work on the structure of the cell surface is that
`molecules can be identified but rarely can any
`function be assigned to them. A powerful way of
`using monoclonal antibodies is to select an antibody
`that has an effect on a cell function or activity in a
`bioassay. For example, Webb et al. (1979) found
`that the W3/25 monoclonal antibody inhibited the
`mixed lymphocyte reaction (MLR) in the rat. In this
`reaction, lymphocytes from one strain of rat, the
`responders, are mixed with inactivated lymphocytes
`from another strain. The responders multiply in
`response to the foreign inactivated lymphocytes and
`generate cytotoxic cells. The W3/25 antibody blocks
`the reaction when it is bound to the responding cells
`but responses to mitogens were unaffected. The
`blocking is reversed by washing off the antibody. All
`responding cells bear the W3/25 antigen. None of
`the three major
`the monoclonal antibodies
`to
`thymocyte glycoproteins (see above) were found to
`have any effect. There is therefore a strong suspicion
`that the surface molecule recognized by the W3/25
`antibody is involved in the interaction that generates
`the response. In the human the monoclonal anti-
`body OKT3 has similar effects (Reinherz et al.,
`1980) but it binds to all peripheral T cells and to
`only some thymocytes; surprisingly, it is reported to
`be mitogenic (Van Wauwe et al.,
`1980). This
`approach is obviously subject to the caveat that the
`W3/25 antigen might not be involved in the mixed
`lymphocyte reaction but may simply be close enough
`to the interacting site for the interaction to be
`physically obstructed when the antibody is bound.
`
`An advantage of monoclonal antibodies in such
`experiments is their purity and specificity for a single
`site on the target molecule.
`The cell surface: conclusions What has emerged
`from the use of monoclonal antibodies to study the
`cell surface? The most striking impression is that
`different types of cell have a quite different surface
`example,
`composition.
`histo-
`For
`from
`apart
`compatibility antigens, the known glycoproteins of
`the rat lymphocyte have a resticted distribution (see
`above). The two major membrane glycoproteins of
`the human erythrocyte, glycophorin A and band 3,
`are confined to erythroid cells in bone marrow, and
`glycophorin at least is absent from other human cells
`tested (Edwards, 1980; Robinson et al., 1981). Data
`for solid tissues are sketchy, but monoclonal anti-
`bodies to three antigens of human milk fat globule
`membrane all react only with epithelial cells in the
`breast and a few other types of epithelial cells, such
`as goblet cells in the colon (C. S. Foster & P. A. W.
`Edwards, unpublished work). Four out of five
`surface molecules identified by monoclonal anti-
`bodies on the surface of 3T3 fibroblasts were absent
`from other cell lines (Hughes & August, 1981). The
`great majority of antibodies raised to melanoma cell
`lines bind to a restricted range of cells (Dippold et
`al., 1980; Herlyn et al., 1980). It seems that it will be
`quite easy to obtain antibodies that distinguish
`between different cell types, though it may be much
`more difficult or impossible to obtain antibodies
`uniquely marking one type of cell. It also seems that
`most major cell surface molecules are present on a
`limited range of cell types.
`It is also clear that to establish the specificity of
`monoclonal antibodies for
`cells
`in tissues
`it
`is
`necessary to stain sections of the tissue with the
`antibody, preferably by enzyme-linked staining such
`as the immunoperoxidase method in which the
`presence of bound antibody is revealed by deposition
`of a coloured reaction product (McMichael et al.,
`1979; Mcllhinney et al., 1981). Permanent cell lines
`have been used as substitutes for the intact tissue,
`but they represent only certain types of cell and their
`origin is not always as certain as their designation
`would suggest. Also cells are known to lose or alter
`(e.g.
`differentiation
`their
`of
`in
`culture
`state
`Greenburg et al., 1980). For example, three mono-
`clonal antibodies raised in this laboratory to human
`breast epithelial cells bind to three distinct but over-
`lapping populations of breast epithelial cells in the
`intact tissue (P. A. W. Edwards & C. S. Foster, un-
`published work). An individual cell line from breast
`generally expresses antigens characteristic of one or
`(the lines can be
`two cell types in the tissue
`heterogeneous in culture), so testing the antibodies
`on a panel of lines presents a confusing picture, each
`antibody binding to a different selection of the cell
`lines. Such complexity is also observed with anti-
`
`1981
`
`Merck Ex. 1074, pg 1600
`
`
`
`Properties and applications of monoclonal antibodies
`
`7
`
`bodies to melanoma cell lines, for example (Herlyn
`et al., 1980; Dippold et al., 1980).
`Resolution ofother mixtures ofantigens
`Intermediate filaments are found in the cyto-
`plasm of cells and are intermediate in sizb between
`microtubules and actin filaments. Unlike micro-
`tubules and actin, different cell types have different
`intermediate filaments-neurofilaments in neurones,
`vimentin filaments in fibroblasts, and tonofilaments
`in epithelial cells (Lazarides, 1980). Intermediate
`filaments appear to contain more than one type of
`polypeptide chain, but are insoluble and so difficult
`to fractionate; their function is unknown. It has
`proved possible to raise monoclonal antibodies that
`1981) and
`detect both common (Pruss et al.,
`distinctive (Lane, 1981; Wood & Anderton, 1981)
`determinants on the filaments. Presumably it will be
`possible to define the location on the filaments of all
`the polypeptide components, and some idea of their
`function may be obtained. Three of the antibodies
`have been microinjected into fibroblast cells (Lin &
`Feramisco, 1981; Klymkowsky, 1981). Two of the
`three caused the filaments to aggregate rapidly
`around the nucleus and to remain there for several
`hours without any disturbance of actin filaments,
`microtubules, cell morphology, membrane ruffling,
`or the normal movements of cytoplasmic vesicles
`and mitochondria. So, surprisingly, none of these are
`organization
`of intermediate
`dependent on the
`filaments.
`Chromosome proteins. Of major interest is the
`specific association of certain non-histone chromo-
`somal proteins with actively transcribed DNA, and
`proteins
`the role of these
`in control of gene
`differentiation.
`expression
`in
`Fractions of non-
`histone chromosomal proteins may be used to
`immunize mice and produce monoclonal antibodies
`recognizing individual proteins. For example, the
`antibodies can be used with immunofluorescence
`the proteins on Drosophila
`methods to
`locate
`chromosomes, where regions active in transcription
`on a chromosome