featurearticles
`Reviewsand
`
`Molecular mechanisms in allergy and clinical immunology
`
`Series editors: William T. Shearer, MD, PhD, Lanny J. Rosenwasser, MD, and Bruce S. Bochner, MD
`
`Selection, design, and engineering of
`therapeutic antibodies
`
`Leonard G. Presta, PhD Palo Alto, Calif
`
`This activity is available for CME credit. See page 32A for important information.
`
`mAbs account for an increasing portion of marketed human
`biological therapeutics. As a consequence, the importance of
`optimal selection, design, and engineering of these not only has
`expanded in the past 2 decades but also is now coming into play
`as a competitive factor. This review delineates the 4 basic areas
`for optimal therapeutic antibody selection and provides
`examples of the increasing number of considerations necessary
`for, and options available for, antibody design. Though some of
`the advances in antibody technology (eg, antibodies derived
`from phage-display libraries) have already made it to market,
`other more recent advances, such as engineering antibodies for
`enhanced effector functions, may not be far behind, especially
`given the increasing competition for therapeutic antibodies to
`the same target (eg, anti-CD20 and anti–TNF-a). (J Allergy
`Clin Immunol 2005;116:731-6.)
`
`Key words: Antibody engineering, target selection, biologic thera-
`peutics
`
`With the number of marketed therapeutic antibodies
`increasing regularly (Table I), the selection and engineer-
`ing of future therapeutic antibodies has become more
`important, both from the perspective of enhanced utility
`for patients and from the perspective of competitive
`advantage. Reviews are available that cover therapeutic
`antibodies on market,1 design of specific therapeutic anti-
`bodies,2,3 or antibody engineering.4 This review covers
`more general aspects of therapeutic antibody selection,
`design, and engineering. Choice of a therapeutic antibody
`can be divided into 4 basic steps: target selection, antibody
`generation, epitope selection, and engineering/optimiza-
`tion (Table II). Given the remarkable strides in the last
`3 areas over the past 15 years, now the most difficult step
`is often target selection.
`
`From Schering-Plough Biopharma.
`Disclosure of potential conflict of interest: Dr Presta is an employee of Schering
`Plough BioPharma (formerly DNAX).
`Received for publication May 28, 2005; revised August 2, 2005; accepted for
`publication August 2, 2005.
`Available online September 12, 2005.
`Reprint requests: Leonard G. Presta, PhD, Schering-Plough Biopharma, 901
`California Avenue, Palo Alto, CA 94304. E-mail: leonard.presta@spcorp.
`com.
`0091-6749/$30.00
`Ó 2005 American Academy of Allergy, Asthma and Immunology
`doi:10.1016/j.jaci.2005.08.003
`
`Abbreviations used
`ADCC: Antibody-dependent cellular cytotoxicity
`CDR: Complementarity-determining region
`FcgR: IgG Fc g receptor
`HER-2: Human epidermal growth factor receptor 2
`
`TARGET SELECTION
`
`Target selection basically involves knowing the disease
`targets, the biochemical pathways that are problematic,
`and the molecule within the biochemical pathway that
`would make the optimal target for intervention. Although
`this seems intuitively straightforward and simple,
`the
`process of target selection can be difficult. Often the
`biochemical pathway(s) involved in a disease state is not
`fully understood, or only 1 or 2 of the constituent pathway
`molecules have been studied; in these cases, the target that
`you know is the one you go after. Even if the entire
`biochemical pathway involved in a disease state is known,
`all constituents may not be fully considered in that one
`molecule in the pathway has been studied by a group and
`by default selected as the target; the other molecules in the
`pathway are not considered. In this case, the optimal target
`for intervention may not be the one selected. The relative
`importance of each molecule in the pathway must also
`be taken into account. Molecules that are branch points
`leading to 2 or more distinct pathways may or may not be
`an optimal target depending on the involvement of those
`branch pathways in the disease state.
`Regardless of the state of knowledge of a pathway and
`its constituent members, one of the most important aspects
`of target choice is the expression pattern of the target in
`human beings. For example, in the optimal scenario for an
`oncology target,5 one wants the target molecule expressed
`only on tumor cells. Less optimally, one may have to use
`a target that is expressed only on a specific cell type but
`is expressed on both nontumor as well as tumor, such as
`the CD20 target of rituximab (Rituxan; Biogen IDEC,
`San Diego, Calif) on B cells6; in this case, the good may
`be destroyed along with the bad, but the therapeutic
`
`731
`
`

`

`732 Presta
`
`J ALLERGY CLIN IMMUNOL
`OCTOBER 2005
`
`TABLE I. US Food and Drug Administration (FDA)–approved therapeutic mAbs
`
`Trade Name
`
`Proper Name
`
`Antibody Format
`
`Molecular Target
`
`Disease Target
`
`US FDA Approval
`
`OKT3
`ReoPro
`Rituxan
`Simulect
`Remicade
`Zenapax
`Herceptin
`Synagis
`
`Mylotarg
`
`Campath
`Zevalin
`
`Xolair
`Raptiva
`Humira
`Bexxar
`Avastin
`Erbitux
`
`Gemtuzumab
`Ozogamicin
`Alemtuzumab
`Ibritumomab
`Tiuxetan
`Omalizumab
`Efalizumab
`Adalimumab
`Tositumomab
`Bevacizumab
`Cetuximab
`
`Reviews and
`feature articles
`
`Muromonab-CD3 Murine IgG2a
`Abciximab
`Chimeric F(ab)
`Chimeric IgG1/k
`Rituximab
`Chimeric IgG1/k
`Basiliximab
`Chimeric IgG1/k
`Infliximab
`Humanized IgG1/k
`Daclizumab
`Humanized IgG1/k
`Trastuzumab
`Humanized IgG1/k
`Palivizumab
`
`CD3
`Integrin aIIb3
`CD20
`IL-2Ra
`TNF-a
`CD25 (IL-2R)
`HER-2
`RSV F-protein
`
`Humanized IgG4/k with
`toxin conjugate
`Humanized IgG1/k
`Murine-90Y IgG1/k
`
`CD33
`
`CD52
`CD20
`
`Renal transplant rejection
`Cardiac ischemic complications
`Non-Hodgkin lymphoma
`Renal transplant rejection
`Rheumatoid arthritis/Crohn disease
`Renal transplant rejection
`Breast cancer
`Respiratory syncytial
`virus infection
`Acute myeloid leukemia
`
`Chronic lymphocytic leukemia
`Non-Hodgkin lymphoma
`
`Humanized IgG1/k
`Humanized IgG1/k
`Human IgG1/k
`Mouse-131I
`Humanized IgG1/k
`Chimeric IgG1/k
`
`IgE
`CD11a
`TNF-a
`CD20
`VEGF
`Epidermal growth
`factor receptor
`
`Asthma
`Psoriasis
`Rheumatoid arthritis
`Non-Hodgkin lymphoma
`Colorectal cancer
`Colorectal cancer
`
`1986
`1994
`1997
`1998
`1998
`1998
`1998
`1998
`
`2000
`
`2001
`2002
`
`2003
`2003
`2003
`2003
`2004
`2004
`
`VEGF, Vascular endothelial growth factor.
`
`TABLE II. Four basic areas for optimal therapeutic
`antibody selection
`
`Target selection
`d Disease target(s)
`d Biochemical pathway(s) involved in disease
`d Molecular target in pathway
`d Expression level and distribution
`Antibody generation
`d Rodent immunization
`d Phage-display library
`Screening and epitope selection
`d Block interaction between ligand/receptor—target ligand
`or receptor?
`d Mediate effect after binding to cell-bound target
`d Cross-link to elicit intracellular signal (eg, apoptosis)
`d Deliver toxic payload after internalization into cell
`d Use effector function to kill cell
`Format selection/engineering/optimization
`d Long or short half-life?
`d Effector functions—yes or no?
`d If yes, enhanced over normal?
`d Binding affinity
`d Potential problems in sequence (eg, isoaspartate formation,
`glycosylation)
`
`effect outweighs the drawback. More commonly, one looks
`for a target that is significantly overexpressed on tumor
`cells, and hence the therapeutic antibody (at appropriate
`dose) can target the tumor cells with minimal effect on
`nontumor cells. One example is human epidermal growth
`factor receptor 2 (HER-2) targeted by the marketed
`therapeutic antibody trastuzumab (Herceptin; Genentech,
`San Francisco, Calif). Although HER-2 is present in a
`limited number of cell types and tissues,7 approximately
`
`30% of breast cancers overexpress HER-2 to some
`degree.8 Another uncommon, but potentially important,
`consideration is the existence of genetic variants, including
`alternatively spliced forms
`(eg,
`epidermal growth
`receptor9), posttranslationally modified forms
`factor
`(eg, CD4410 and mucins11), or proteolytically processed
`soluble forms of membrane-bound target.10
`
`mAb GENERATION
`
`Once the target has been chosen, antibodies against it
`must be generated. Currently there are 3 major strategies
`for antibody generation: normal rodents, transgenic mice,
`and phage-display. The first 2 involve immunization of
`the rodent with purified target protein, peptide constitu-
`ents of the protein, or DNA encoding the target protein
`(eg, gene delivery for in situ protein expression).12 Once
`the rodent mounts an immune response and makes
`antibodies, hybridomas are generated in which each
`hybridoma makes a single (monoclonal) antibody.13 The
`antibodies secreted from the hybridomas are screened for
`the desired effect. Phage-display is an ex vivo method in
`which large libraries of cloned14 or semisynthetic15,16
`antibody fragments are screened against the purified target
`protein. Antibodies that bind to the target can subse-
`quently be screened for desired effect. Phage-display
`libraries are especially useful when the target is a human
`protein that exhibits high homology to its rodent coun-
`terparts; in these cases, immunization of rodents may elicit
`no response or a paltry one. However, if a mouse knockout
`for the target protein is available for immunization, this
`circumscribes the problem of high homology between
`humans and rodents.
`
`

`

`featurearticles
`Reviewsand
`
`J ALLERGY CLIN IMMUNOL
`VOLUME 116, NUMBER 4
`
`Presta 733
`
`When a therapeutic antibody is the goal, use of several
`of the aforementioned technologies may be prudent,
`though this seems to be uncommon. Different technologies
`may provide antibodies with different characteristics and
`may have varying degrees of robustness in the response.
`If several technologies are used, then the repertoire of
`antibodies may be greater, and selection for the most
`efficacious antibody is easier. Given the expansion of
`antibody therapeutics, there are of course newer antibody
`generation technologies being investigated,
`including
`ribosome display17 (in which a ribosome is used to trans-
`late the mRNA encoding the antibody but without a
`stop codon; the expressed antibody is hence tethered to
`the ribosome, thereby physically connecting the antibody
`[phenotype] to its mRNA [genotype]; after selection for
`binding, the antibody mRNA can be sequenced), and
`generation of mAbs from rabbits,18 chickens,19 and other
`species.20 However, none of these has yet achieved the
`widespread, mainstream use of rodent hybridoma and
`phage-display.
`
`EPITOPE SELECTION
`
`Although it seems obvious that screening properly is
`key to selection of an optimal antibody, as with target
`selection, it may not be straightforward. First one must
`decide what the antibody needs to do, such as block a
`receptor-ligand interaction with an antagonist antibody (do
`you target receptor or ligand? do you want a competitive or
`noncompetitive inhibitor?), crosslink a membrane-bound
`target for an intracellular effect, or deliver a toxic load to a
`cell (in which case internalization of the antibody-toxin
`conjugate is beneficial). Each of these requires unique sets
`of screening methods to find the best antibody.
`Perhaps the most common system is blocking of a
`receptor-ligand pair. The choice between targeting the
`receptor or ligand depends on the biology of the system. In
`complex systems in which one receptor can bind several
`ligands, the receptor would be the target. Conversely, in
`systems in which one ligand binds to several receptors
`and all receptors need to be blocked, the ligand should be
`the target (eg, some TNF family members21). Even in
`the simplest case, 1 ligand and 1 receptor, the choice
`may depend on tissue distribution of ligand and receptor,
`which of the 2 is in limiting supply, or whether attaching
`an antibody to a cell-bound receptor might result
`in
`complications (eg, activation of complement and/or anti-
`body-dependent cellular cytotoxicity [ADCC] and sub-
`sequent destruction of the cell).
`Once the target receptor or ligand is chosen, then
`appropriate screening methods need to be used to select
`the best antibody. Optimally, several screening methods
`are used. If possible (ie, purified receptor and ligand
`proteins are available), protein-protein ELISA-based
`assays are most often used as a first screen; in these, one
`looks for efficient blocking of receptor-ligand binding
`by titration of the candidate antibodies. After this, it is
`often wise to use at least 1 (and the more the better) cell-
`
`based assay in which natural or, less optimally, transfected
`cells with the receptor are screened for an effect, such
`as apoptosis, secretion of downstream molecules in
`the pathway, phosphorylation of receptor, and so forth.
`From a drug development perspective, it is also beneficial
`to screen the candidate antibodies against the target from
`other species, most notably mice and monkeys; being able
`to use the same antibody for in vivo models in rodents or
`other species and for pharmacokinetics/pharmacodynam-
`ics and toxicology evaluation in higher primates is easier
`than having to deal with surrogate antibodies specific for
`these other species. Another selection that is often delayed
`until candidate antibodies have been winnowed to a few
`is testing for cross-reactivity on human tissues; if a candi-
`date antibody has strong or widespread cross-reactivity
`on human tissues on which the target molecule is not
`supposed to be expressed, then that antibody is an unlikely
`candidate from a safety standpoint.
`With regard to antibody affinity, it is not always the case
`that antibodies with the strongest binding are the optimal
`selection. One antibody may bind strongly to a ligand but
`cover only part of the binding site on the ligand for its
`receptor, whereas another antibody may bind slightly less
`strongly but hit the receptor-binding site on the ligand
`head-on. The latter may be the more efficacious choice. In
`some cases, the highest affinity antibody may not exhibit
`optimal penetrance if it has to permeate a tissue or a solid
`tumor. An elegant example was reported by Adams et al,22
`who tested a series of anti–HER-2 single-chain variable
`fragment (Fv) antibodies and showed that antibodies with
`extremely high affinity had impaired tumor penetration
`properties.
`Sometimes a noncompetitive antibody might be just
`as efficacious (or more so) than a competitive antibody.
`Noncompetitive antibodies may, for example, lock the
`ligand (or receptor) in a conformation that precludes
`binding of the partner; in these cases, the antibody may
`have a weaker binding affinity but function as well as a
`more strongly binding direct competition antibody. For
`example, a crystal structure of antihuman IL-10 antibody
`9D7 complexed with IL-10 showed that 9D7 bound away
`from the receptor-binding site on IL-10.23 Likewise,
`mapping of the epitope of Xolair (Genentech) on human
`IgE24 and comparison with the crystal structure of IgE
`bound to FceRI25 and the open/closed forms of IgE26
`suggest that Xolair may lock IgE in the closed conforma-
`tion and thereby prevent IgE from binding to FceRI.
`If the antibody is directed to a cell membrane–bound
`protein and will be carrying a toxic payload,27 internali-
`zation of the antibody-toxin conjugate is optimal; other-
`wise, the toxin may affect nontarget cells. Usually the
`toxin is targeted to an intracellular molecule (eg, calichea-
`micin,28 maytansine,29 protein toxins30). Methods for
`screening that optimize selection of antibodies enhanced
`for internalization have been developed31; an antibody
`may bind strongly to its target molecule but not at an
`epitope that is optimal for internalization. Another special
`case is antibody-cytokine fusions32 in which the conjugate
`should remain on the cell surface and not be internalized
`
`

`

`734 Presta
`
`J ALLERGY CLIN IMMUNOL
`OCTOBER 2005
`
`chain; the carbohydrate was not required for binding or
`biological activity and was removed in the humanized
`antibody.37
`Inspection of the sequence of the CDRs may reveal
`potential problems that need to be investigated. When
`present in a mobile loop, such as a large CDR, an Asn-Gly
`or Asp-Gly sequence may undergo spontaneous isomer-
`ization to form isoaspartic acid38; formation of isoaspar-
`tate may debilitate or completely abrogate the binding of
`the antibody. Substitution of the offending Asn or Asp
`with Ala, Gln, or Glu needs to be evaluated to determine
`whether these substitutions can maintain the antibody
`binding and efficacy; if not, then production and formu-
`lation should be optimized to minimize isoaspartate
`formation. The presence of methionine in a CDR, espe-
`cially if exposed to solvent, can create a problem if the
`methionine is oxidized and this interferes with binding. As
`with the case of isoaspartate, one can investigate substitut-
`ing other amino acids for the methionine or optimization
`of production/formulation to reduce oxidation.
`Engineering of the Fc portion of the antibody (other
`than removing it in the case of therapeutic F[ab] frag-
`ments) centers on either removing or enhancing the Fc
`effector functions. If effector functions are not warranted,
`one could use human IgG4 (as mentioned earlier), human
`IgG1 in which specific amino acids have been altered, or
`human IgG1 in which the conserved Fc carbohydrate (at
`Asn297 in each heavy chain) has been removed (usually
`by alteration of Asn297 to Ala or Gln). Several reports of
`human IgG1 variants with reduced interaction for FcgR39
`or complement40 have been reported, although to date
`none of these has been incorporated into marketed anti-
`bodies. Deglycosylated antibodies have been evaluated in
`rodent models,41 but there are no reports of complete
`evaluation of these in human beings. Although human
`IgG4 is currently being used for some potential therapeutic
`antibodies in development, IgG4 is not without its own
`problem. In human IgG1, the hinge region (ie, the region
`connecting the F[ab] and Fc portions of the antibody) has
`a -Cys-Pro-Pro-Cys- sequence and forms 2 interchain di-
`sulfide bonds between the 2 heavy chains in the antibody.
`However, human IgG4 has a -Cys-Pro-Ser-Cys- sequence.
`Although IgG4 can form the 2 interchain disulfide bonds
`(as in IgG1), the presence of the Ser also allows formation
`of an intrachain disulfide within each heavy chain; con-
`sequently, the 2 heavy chains are noncovalently linked
`and can dissociate from one another.42 From a biological
`perspective, it has been hypothesized that in vivo IgG4
`could form bispecific antibodies via exchange of heavy
`chains between 2 IgG4 antibodies with differing targets.43
`Although this may or may not occur, the presence of
`IgG4 half-molecules (ie, nondisulfide bonded heavy-light
`chains) may complicate production and characterization
`of the therapeutic antibody.
`Enhancing the effector functions has come into fruition
`only recently. In cases in which a therapeutic antibody can
`use ADCC or where cross-linking of target on a cell
`surface is part of an antibody’s mechanism of action,
`improving the binding of the Fc to FcgR could enhance
`
`so that the conjugated cytokine can interact with its cell-
`surface receptor (either on the same cell to which the
`antibody is bound or on another cell). In this case,
`internalization may not be desirable, and antibodies
`should be screened appropriately.
`
`Reviews and
`feature articles
`
`ENGINEERING AND OPTIMIZATION
`
`Before addressing formal engineering, which we can
`define as alteration of specific residues in the antibody to
`improve its function, the question of the antibody format
`must be considered. Most marketed antibodies are com-
`prised of a full-length human IgG1 that provides for long
`half-life and effector functions (including complement
`activation and ADCC via IgG Fc g receptor [FcgR]
`binding); if the presence of effector functions would be
`deleterious, then the choice may be to use a human IgG4
`rather than a human IgG1. However, dependent on the
`mechanism of action of the antibody, other formats may be
`more desirable. For example, if a short half-life is required,
`an F(ab) fragment may be optimal, such as ReoPro
`(Centocor, Malvern, Pa)33; in addition, the F(ab) fragment
`could be manufactured in bacteria, thereby reducing the
`cost compared with manufacture in mammalian cell cul-
`ture. Use of an F(ab) may also be beneficial when a cell-
`bound receptor is the target but cross-linking of receptors
`by a bivalent, full-length antibody would be deleterious.
`We can separate formal engineering and optimization of
`an antibody into 2 sections on the basis of the structure of
`the antibody itself: the F(ab) and the Fc. With the advent
`of humanized antibodies,34 the most common engineering
`of marketed antibodies has been transfer of binding loops
`(ie, the portion of the variable light and heavy chains that
`interact with the target molecule, also referred to as
`complementarity-determining regions [CDRs]) from a
`mouse (or other nonhuman) antibody to a human antibody.
`Although currently the majority of marketed antibodies are
`humanized, this may change as more antibodies derived
`from transgenic mice and phage-display libraries come to
`market. Other aspects of F(ab) engineering used are
`affinity maturation and stability. In the former, residues
`in the CDRs are varied using mutagenesis, and the
`resulting compendium of mutated antibodies are screened
`for improved binding and efficacy. Less prevalent, but still
`important, is engineering to make the antibody more
`stable—important from perspectives of manufacturing
`(ie, enhanced production) as well as formulation (en-
`hanced shelf-life).35
`Other problems specific to a given antibody may occur.
`In approximately 30%36 of mouse and human antibodies,
`Asn-linked glycosylation may be present in the antibody
`variable domain. In these cases, it is necessary to deter-
`mine whether the glycosylation is required for the activity
`of the antibody. If not required, then the glycosylation
`could be removed by altering the Asn to Ala or Gln. For
`example,
`the parental 4D5 mouse antibody used to
`design Herceptin had Asn-linked glycosylation in the
`framework (ie, non-CDR) region of the variable light
`
`

`

`featurearticles
`Reviewsand
`
`J ALLERGY CLIN IMMUNOL
`VOLUME 116, NUMBER 4
`
`Presta 735
`
`the efficacy of the therapeutic antibody. To date, 3
`methods have been reported: alteration of residues in the
`Fc,39 removal of the fucose moiety from the conserved
`carbohydrate in the Fc,44,45 and presence of multiple Fc.46
`The most practical of
`these is removal of
`fucose.
`Normally, human IgG has predominantly fucosylated
`carbohydrate with only a small percentage of defucosy-
`lated.47 Since the discovery of the enhanced FcgR binding
`and consequent enhanced ADCC, a Chinese hamster
`ovary cell line has been engineered in which the a-1,
`6-fucosyltransferase that attaches fucose to the carbohy-
`drate has been knocked out.48 Hence, antibodies produced
`with this cell line lack fucose and have enhanced ADCC.49
`The technology is so new that no antibodies lacking fucose
`have yet been marketed, but it is likely that eventually,
`defucosylated antibodies will be used.
`Finally, the half-life (or clearance rate) of therapeutic
`antibodies can be manipulated via engineering of the Fc.
`By alteration of specific residues in the Fc, binding to the
`FcRn (neonatal Fc receptor or Brambell receptor50) can be
`abrogated or enhanced. Though not yet tested in human
`beings, improved half-life of Fc variants in primates has
`been reported.51 Longer half-life of a therapeutic anti-
`body could lead to less frequent dosing—a boon for the
`patient.
`
`CONCLUSION
`
`Design of a therapeutic antibody involves multiple
`considerations:
`target selection, antibody generation,
`epitope selection, and engineering for optimal efficacy.
`When antibodies first came into the limelight as therapeu-
`tics (1980s), the choices for the latter 3 were limited. Since
`that time, as more antibodies have been developed and
`marketed, progress in these areas has been driven pri-
`marily by competition. For example, once an antibody is
`successfully marketed, its target molecule becomes vali-
`dated—that is, it has been shown that one can indeed make
`a therapeutic antibody against the target and have an effect
`on the disease (in addition to making a profit). Companies
`will then often develop their own antibody to that target;
`however, to beat out the existing (and any other new)
`competition or merely to gain a significant market share,
`the new antibodies might need improved efficacy, safety,
`and/or dosing. In some cases, the new antibody may need
`to be engineered for improved ADCC, longer half-life,
`increased potency (potentially leading to lower dose), or
`it may need to be conjugated to a toxin. Bispecific
`antibodies, in which each arm of the antibody targets a
`different molecule, are also contenders in that they may
`more efficiently destroy or otherwise shut down cells.52
`The increased choices in design of therapeutic antibodies
`will, of course, take some time to sort out and to make it to
`market, but the ability to improve affinity, select the
`effector functions of the antibody, and use nonnative
`antibody forms ideally will improve the utility of thera-
`peutic antibodies for the most
`important reason:
`the
`patient.
`
`REFERENCES
`
`1. Pavlou AK, Belsey MJ. The therapeutic antibodies market to 2008.
`Eur J Pharm Biopharm 2005;59:389-96.
`2. Qu Z, Griffiths GL, Wegener WA, Chang C-H, Govindan SV, Horak ID,
`et al. Development of humanized antibodies as cancer therapeutics.
`Methods 2005;36:84-95.
`3. Tsurushita N, Hinton PR, Kumar S. Design of humanized antibodies:
`from anti-Tac to Zenapax. Methods 2005;36:69-83.
`4. Presta LG. Engineering antibodies for therapy. Curr Pharm Biotechnol
`2002;3:237-56.
`5. Ross JS, Schenkein DP, Pietrusko R, Rolfe M, Linette GP, Stec J,
`et al. Targeted therapies for cancer 2004. Am J Clin Pathol 2004;122:
`598-609.
`6. Rastetter W, Molina A, White CA. Rituximab: expanding role in therapy
`for lymphomas and autoimmune diseases. Annu Rev Med 2004;55:
`477-503.
`7. Press MF, Cordon-Cardo C, Slamon DJ. Expression of the HER-2/neu
`proto-oncogene in normal human adult and fetal tissues. Oncogene 1990;
`5:953-62.
`8. Slamon DJ, Clark GM, Wong SG, Levin WJ, Ullrich A, McGuire WL.
`Human breast cancer: correlation of relapse and survival with amplifi-
`cation of the HER-2/neu oncogene. Science 1987;235:177-82.
`9. Frederick L, Wang X-Y, Eley G, James CD. Diversity and frequency of
`epidermal growth factor receptor mutations in human glioblastomas.
`Cancer Res 2002;60:1383-7.
`10. Cichy J, Pure E. The liberation of CD44. J Cell Biol 2003;161:839-43.
`11. Magro G, Schiappacassi M, Perissinotto D, Corsaro A, Borghese C,
`Belfiore A, et al. Differential expression of mucins 1-6 in papillary
`thyroid carcinoma: evidence for transformation-dependent post-transla-
`tional modifications of MUC1 in situ. J Pathol 2003;200:357-69.
`12. Tang DC, DeVit M, Johnston SA. Genetic immunization is a simple
`method for eliciting an immune response. Nature 1992;356:152-4.
`13. Kohler G, Milstein C. Continuous cultures of fused cells secreting
`antibody of predefined specificity. Nature 1975;256:495-7.
`14. Vaughan TJ, Williams AJ, Pritchard K, Osbourn JK, Pope AR, Earnshaw
`JC, et al. Human antibodies with sub-nanomolar affinities isolated from a
`large non-immunized phage display library. Nat Biotechnol 1996;14:
`309-14.
`15. Knappik A, Ge L, Honegger A, Pack P, Fisher M, Wellnhofer G, et al.
`Fully synthetic human combinatorial antibody libraries (HuCAL) based
`on modular consensus frameworks and CDRs randomized with trinucle-
`otides. J Mol Biol 2000;296:57-86.
`16. Hoet RM, Cohen EH, Kent RB, Rookey K, Schoonbroodt S, Hogan S,
`et al. Generation of high-affinity human antibodies by combining donor-
`derived and synthetic complementarity-determining-region diversity.
`Nat Biotechnol 2005;23:344-8.
`17. Hanes J, Schaffitzel C, Knappik A, Plu¨ckthun A. Picomolar affinity
`antibodies from a fully synthetic library selected and evolved by
`ribosome display. Nat Biotechnol 2000;18:1287-92.
`18. Spieker-Polet H, Sethupathi P, Yam P-C, Knight KL. Rabbit monoclonal
`antibodies: generating a fusion partner to produce rabbit-rabbit hybrid-
`omas. Proc Nat Acad Sci U S A 1995;92:9348-52.
`19. Tsurushita N, Park M, Pakabunto K, Ong K, Avdalovic A, Fu H, et al.
`Humanization of a chicken anti-IL-12 monoclonal antibody. J Immunol
`Methods 2004;295:9-19.
`20. Reichmann L, Muyldermans S. Single domain antibodies: comparison of
`camel VH and camelised human HV domains. J Immunol Methods 1999;
`231:25-38.
`21. Mackay F, Ambrose C. The TNF family members BAFF and APRIL: the
`growing complexity. Cytokine Growth Factor Rev 2003;14:311-24.
`22. Adams GP, Schier R, McCall AM, Simmons HH, Horak EM,
`Alpaugh RK, et al. High affinity restricts the localization and tumor
`penetration of single-chain Fv antibody molecules. Cancer Res 2001;
`61:4750-5.
`23. Josephson K, Jones BC, Walter LJ, DiGiacomo R, Indelicato SR, Walter
`MR. Noncompetitive antibody neutralization of
`IL-10 revealed by
`protein engineering and X-ray crystallography. Structure (Camb) 2002;
`10:981-7.
`24. Presta LG, Lahr SJ, Shields RL, Porter JP, Gorman CM, Fendly BM,
`et al. Humanization of an antibody directed against IgE. J Immunol 1993;
`151:2623-32.
`
`

`

`736 Presta
`
`J ALLERGY CLIN IMMUNOL
`OCTOBER 2005
`
`25. Garman SC, Wurzburg BA, Tarchevskaya SS, Kinet J-P, Jardetsky TS.
`Structure of the Fc fragment of human IgE bound to its high-affinity
`receptor FceRIa. Nature 2000;406:259-66.
`26. Wurzburg BA, Garman SC, Jardetsky TS. Structure of the human IgE-Fc
`Ce3-Ce4 reveals conformational flexibility in the antibody effector
`domains. Immunity 2000;13:375-85.
`27. Niv R, Cohen CJ, Denkberg G, Segal D, Reiter Y. Antibody engineering
`for targeted therapy of cancer: recombinant Fv-immunotoxins. Curr
`Pharm Biotech 2001;2:19-46.
`28. Linenberger M. CD33-directed therapy with gemtuzumab ozogamicin in
`acute myeloid leukemia: progress in understanding cytotoxicity and
`potential mechanisms of drug resistance. Leukemia 2005;19:176-82.
`29. Tolcher AW, Ochoa L, Hammond LA, Patnaik A, Edwards T, Takimoto C,
`et al. Cantuzumab mertansine, a maytansinoid immunoconjugate directed
`to the CanAg antigen: a phase I, pharmacokinetic, and biologic correlative
`study. J Clin Oncol 2003;21:211-22.
`30. Smallshaw JE, Ghetie V, Rizo J, Fulmer JR, Trahan LL, Ghetie MA,
`et al. Genetic engineering of an immunotoxin to eliminate pulmonary
`vascular leak in mice. Nat Biotechnol 2003;21:387-91.
`31. Marks JD. Selection of
`internalizing antibodies for drug delivery.
`Methods Mol Biol 2004;248:201-8.
`32. Helguera G, Morrison SL, Penichet ML. Antibody-cytokine fusion
`proteins: harnessing the combined power of cytokines and antibodies
`for cancer therapy. Clin Immunol 2002;103:233-46.
`33. Reverter JC, Beguin S, Kessels H, Kumar R, Hemmer HC, Coller BS.
`Inhibition of platelet-mediated, tissue-factor-induced thrombin genera-
`tion by the mouse/human chimeric 7E3 antibody; potential implications
`for the effect of c7E3 Fab treatment on acute thrombosis and ‘‘clinical
`restenosis.’’ J Clin Invest 1996;98:863-74.
`34. Jones PT, Dear PH, Foote J, Neuberger MS, Winter G. Replacing the
`complementarity-determining regions in a human antibody with those
`from a mouse. Nature 1986;321:522-5.
`35. Ewert S, Honegger A, Plu¨ckthun A. Stability improvement of antibodies
`for extracellular and intracellular applications: CDR grafting to stable
`frameworks and structure-based framework engineering. Methods 2004;
`34:184-99.
`36. Wright A, Morrison SL. Antibody variable region glycosylation: bio-
`chemical and clinical effects. Springer Semin Immunopathol 1993;15:
`259-73.
`37. Carter P, Presta L, Gorman CM, Ridgway JBB, Henner D, Wong WLT,
`et al. Humanization of an anti-p185HER2 antibody for human cancer
`therapy. Proc Nat Acad Sci U S A 1992;89:4285-9.
`38. Cacia J, Keck R, Presta LG, Frenz J. Isomerization of an aspartic acid
`residue in the complementarity-determining regions of a recombinant
`antibody to human IgE: identification and effect on binding affinity.
`Biochemistry 1996;35:1897-903.
`39. Shields RL, Namenuk AK, Hong K, Meng YG, Rae J, Briggs J, et al.
`High resolution mapping of the binding site on human IgG1 for FcgRI,
`
`FcgRII, FcgRIII, and FcRn and design of IgG1 variants with improved
`binding to the FcgR. J Biol Chem 2001;276:6591-604.
`40. Idusogie EE, Presta LG, Gazzano-Santoro H, Totpal K, Wong PY,
`Ultsch M, et al. Mapping of the C1q binding site on rituxan, a chimeric
`antibody with a human IgG1 Fc. J Immunol 2000;164:4178-84.
`41. Tao M-H, Morrison SL. Studies of aglycosylated chimeric mouse-human
`IgG: role of carbohydrate in the structure and effector functions mediated
`by the human IgG constant region. J Immunol 1989;143:2595-601.
`42. Angal S, King DJ, Bodmer MW, Turner A, Lawson ADG, Roberts G,
`et al. A single amino acid substitution abolishes the heterogeneity of
`shimeric mouse/human (IgG4) antibody. Mol Immunol 1993;30:105-8.
`43. Schuurman J, Perdok GJ, Gorter AD, Aalberse RC. The inter-heavy
`ch