REVIEW
`
`Engineered antibodies now represent over 30% of biopharmaceuticals in clinical trials, as highlighted by recent
`approvals from the US Food and Drug Administration. Recombinant antibodies have been reduced in size, rebuilt
`into multivalent molecules and fused with , for example radionuclides, toxins, enzymes, liposomes and viruses. The
`emergence of recombinant technologies has revolutionized the selection, humanization and production of
`antibodies, superseding hybridoma technology and allowing the design of antibody-based reagents of any
`specificity and for very diverse purposes.
`
`Engineered antibodies
`
`The discovery of hybridoma technology
`by Kohler and Milstein1 in 1975 heralded
`a new era in antibody research and clini-
`cal development. Mouse hybridomas were
`the first reliable source of monoclonal antibodies and were de-
`veloped for a number of in vivo therapeutic applications (Table 1;
`see also FDA product approval information at http://www.
`fda.gov/cber/efoi/approve.htm). Throughout the 1990s, innova-
`tive recombinant DNA technology, including chimerization and
`humanization, enhanced the clinical efficiency of mouse anti-
`bodies and led to the recent wave of approvals by the FDA for
`therapeutic immunoglobulin (Ig) and Fab molecules (monova-
`lent antibody fragment produced by proteolysis) (Table 1 and
`Fig. 1; refs. 2,3). These developments have continued, and in
`2002 the FDA approved the first radiolabeled antibody for cancer
`immunotherapy (Zevalin)4.
`The list of approved antibody therapeutics against cancer
`and against viral and inflammatory diseases is growing
`rapidly (Table 1), with more than 30 antibodies in late-phase
`clinical trials2,3. Most recently, innovative structural designs
`have improved in vivo pharmacokinetics, expanded immune
`repertoires and permitted screening against refractory targets
`and complex proteome arrays, while new molecular evolu-
`tion strategies have enhanced affinity, stability and expres-
`sion levels. This review describes these emerging technologies
`and discusses the creation of a vast range of engineered, anti-
`body-based reagents that specifically target biomarkers of
`human health and disease1–3. We review antibodies designed
`as intact molecules and recombinant fragments and then
`focus on the latest technologies for attaching additional ther-
`apeutic payloads, such as radionuclides, drugs, enzymes and
`vaccine-inducing epitopes.
`
`Intact antibodies, humanization and de-immunization
`Intact antibodies provide high-specificity, high-affinity target-
`ing reagents and are usually multivalent (Fig. 1). Their simulta-
`neous binding to two adjacent antigens increases functional
`affinity and confers high retention times, for example, on cell
`surfaces. In addition, intact antibodies comprise Fc domains,
`which can be important for cancer immunotherapy through
`their abilities both to recruit cytotoxic effector functions2,5 and
`to extend the serum half-life, mediated by the neonatal Fc re-
`ceptor6. Unmodified mouse monoclonal antibodies formed the
`first wave of FDA-approved immunotherapeutic reagents
`(Table 1), although their in vivo applications were limited be-
`cause repeated administrations provoked an anti-mouse im-
`mune response2. Simple strategies have been developed to
`avoid, mask or redirect this human immune surveillance; these
`strategies include fusion of mouse variable regions to human
`constant regions as ‘chimeric’ antibodies, ‘de-immunization’
`by removal of T-cell epitopes and ‘humanization’ by grafting
`
`©2003 Nature Publishing Group http://www.nature.com/naturemedicine
`
`PETER J. HUDSON
`& CHRISTELLE SOURIAU
`
`mouse surface residues onto human ac-
`ceptor antibody frameworks (Table 1; see
`also Supplementary Note online)2.
`Modern alternative strategies now allow
`selection of fully human antibodies directly from natural or
`synthetic repertoires, including live transgenic mice producing
`purely human antibodies7. Human antibody-display libraries
`were used to transform a mouse antibody in vitro into a fully
`human derivative (D2E7), which is likely to be the first FDA-ap-
`proved fully human anti-inflammatory antibody. Many other
`fully human antibodies, including Efalizumab for the treat-
`ment of psoriasis, are in clinical evaluation (see http://www.
`fda.gov/cber/efoi/approve.htm).
`
`Design of ‘antibody fragments’ for unique clinical applications
`For cytokine inactivation, receptor blockade or viral neutraliza-
`tion, the Fc-induced effector functions are often unwanted and
`can be simply removed by proteolysis of intact antibodies to
`yield monovalent Fab fragments (ReoPro, Remicade; Table 1).
`Proteolysis, however, does not easily yield molecules smaller
`than a Fab fragment, and microbial expression of single-chain
`Fv (scFv) is currently the favored method of production (Fig. 1).
`In scFvs, the variable (VH and VL) domains are stably tethered
`together with a flexible polypeptide linker (Fig. 1)2,3. In com-
`parison with whole antibodies, small antibody fragments such
`as Fab or scFv exhibit better pharmacokinetics for tissue pene-
`tration and also provide full binding specificity because the
`antigen-binding surface is unaltered. However, Fab and scFv
`are monovalent and often exhibit fast off-rates and poor reten-
`tion time on the target8,9. Therefore, Fab and scFv fragments
`have been engineered into dimeric, trimeric or tetrameric con-
`jugates to increase functional affinity through the use of either
`chemical or genetic cross-links (Fig. 1)2,3,10,11. Various methods
`have been devised to genetically encode multimeric scFvs, of
`which the most successful design was the simple reduction of
`scFv linker length to direct the formation of bivalent dimers
`(diabodies, 60 kDa), triabodies (90 kDa) or tetrabodies (120
`kDa) (Fig. 1)11. Indeed, the first clinical trials of scFv fragments
`are likely to be as multivalent reagents, because they exhibit
`high functional affinity and have been very successful in pre-
`clinical studies8,12–14.
`
`Pharmacokinetics of intact antibodies versus fragments
`The efficiency of antibodies in vivo, for example in cancer ther-
`apy, lies in their capacity to discriminate among tumor-associ-
`ated antigens at low levels. Immunotherapy has been more
`successful against circulating cancer cells than solid tumors be-
`cause of better cell accessibility. This is illustrated by the FDA
`approval of intact antibodies: Rituxan for the treatment of non-
`Hodgkin lymphoma and Campath and Mylotarg for the treat-
`ments of leukemia (Table 1)15. Only two monoclonal antibodies
`
`NATURE MEDICINE • VOLUME 9 • NUMBER 1 • JANUARY 2003
`
`129
`
`Lassen - Exhibit 1046, p. 1
`
`

`

`REVIEW
`
`Fig. 1
`Schematic representation of an intact
`Ig together with Fab and Fv fragments and
`single V (colored ovals; dots represent anti-
`gen-binding sites) and C domains (uncol-
`ored). Engineered recombinant antibodies are
`shown as scFv monomers, dimers (diabodies),
`trimers (triabodies) and tetramers (tetrabod-
`ies), with linkers represented by a black line.
`Minibodies are shown as two scFv modules
`joined by two C domains. Also shown are Fab
`dimers (conjugates by adhesive polypeptide
`or protein domains) and Fab trimers (chemi-
`cally conjugated). Colors denote different
`specificities for the bispecific scFv dimers (dia-
`bodies) and Fab dimers and trimers.
`
`©2003 Nature Publishing Group http://www.nature.com/naturemedicine
`
`have been approved for the treatment of
`solid tumors: Herceptin for the treatment
`of breast carcinoma and PanoRex for
`colon cancer (in Germany). Although the
`mechanisms of action are still under in-
`vestigation, Herceptin appears to utilize
`Fc receptors and angiogenesis16, whereas
`Rituxan activates apoptosis through re-
`ceptor dimerization17.
`Radiolabeled antibodies are important clinical reagents for
`both tumor imaging and therapy and also provide an effective
`evaluation of pharmacokinetics18. The choice of radionuclide
`dictates the application. For example, Zevalin is approved for
`lymphoma therapy as a rapidly cleared, intact mouse antibody
`to match the clearance rates of yttrium-90 (ref. 4). Therapeutic
`administration requires a balance between long dissociation
`rates at the target site and slow blood clearance, which can lead
`to accumulation in the liver and high radiation exposure of
`other tissues. Biodistribution studies in solid tumors have also
`revealed that whole IgG molecules are too large (150 kDa) for
`rapid tumor penetration. The best tumor-targeting reagents
`comprise an intermediate-sized multivalent molecule, provid-
`ing rapid tissue penetration, high target retention and rapid
`blood clearance8,12–14. For example, diabodies (60 kDa) are effi-
`cacious with short-lived radioisotopes for clinical imaging as a
`result of the fast clearance rates12–14. Larger molecules, such as
`minibodies (90 kDa), are used with long-lived radioisotopes
`and are suitable for tumor therapy because they achieve a
`higher total tumor ‘load’14. Fab dimers (110 kDa) have also
`been effective in preclinical studies19.
`The short half-life of antibody fragments can also be ex-
`tended by ‘pegylation’, that is, a fusion to polyethylene glycol
`(PEG)20. Renal and hepatic localization of intact radiolabeled
`antibody fragments constitutes a major problem. An impor-
`tant study demonstrated that a previously undescribed radio-
`iodination reagent could liberate radionuclide from the anti-
`body fragment before incorporation into renal cells21. The ra-
`dionuclide is excreted rapidly, thus decreasing the total renal
`radiation dose21. The development of new metabolizable
`chelates will further improve the pharmacokinetics of recom-
`binant antibodies for cancer targeting. Modifications to sur-
`face charge designed to alter the isoelectric point (pI), such as
`glycolation, can also reduce the tissue (kidney) uptake22. The
`improved functional affinity, tumor penetration and biodistri-
`bution of these engineered antibody fragments will stimulate
`the development of a new generation of reagents for imaging
`and therapy.
`
`Intact IgG
`
`Monovalent
`fragments
`
`Diabodies
`(~60 kDa)
`
`Fab
`(~55 kDa)
`
`Fv
`
`scFv
`(~30 kDa)
`
`Bivalent
`
`Bispecific
`
`Triabodies (~90 kDa)
`
`V Domain
`(~15 kDa)
`
`C Domain
`(~15 kDa)
`
`Fab conjugates: dimers and trimers
`
`Minibody
`
`Adhesive
`domains
`or
`helices:
`dimeric
`Fab
`
`Chemical
`conjugate:
`trimeric Fab
`
`Trivalent
`
`Tetrabodies (~120 kD)
`
`Tetravalent
`
`Engineering multiple specificity in antibody fragments
`Bispecific antibodies contain two different binding specificities
`fused together and, in the most simple example, bind to two
`adjacent epitopes on a single target antigen, thereby increasing
`the avidity. Alternatively, bispecific antibodies can cross-link
`two different antigens and are powerful therapeutic reagents,
`particularly for recruitment of cytotoxic T cells for cancer treat-
`ment23,24. Bispecific antibodies can be produced by fusion of
`two hybridoma cell lines into a single ‘quadroma’ cell line;
`however, this technique is complex and time-consuming, and
`it produces unwanted pairing of the heavy and light chains. Far
`more effective methods to couple two different Fab modules
`incorporate either chemical or genetic conjugation or fusion to
`adhesive heterodimeric domains, including designed CH3 do-
`mains23,24. Bispecific diabodies provide an innovative alterna-
`tive therapeutic25 (Fig. 1).
`
`Bifunctional antibodies. The original ‘magic bullet’ concept is
`still alive: antibodies have been fused to a vast range of mole-
`cules that provide important ancillary functions after target
`binding. These include radionuclides (discussed earlier) and
`also cytotoxic drugs, toxins, peptides, proteins, enzymes and
`viruses, the latter for targeted gene therapy26–30. For cancer ther-
`apy, bifunctional antibodies are engineered to effectively target
`tumor-associated antigens at low levels and then deliver a cyto-
`toxic payload to tumor cells. The latest antibody-toxin conju-
`gates are stable in vivo and minimally immunogenic28,29.
`Antibodies have also been fused to lipids and PEGs20, both to
`enhance in vivo delivery and pharmacokinetics and to direct
`drug-loaded liposomes31,32. As immunoliposomes, anti-transfer-
`rin receptor antibodies have been used to deliver drugs to the
`brain, passing through the blood-brain barrier32. Antibody-en-
`zyme fusions have also been developed for prodrug activation,
`primarily for cancer therapy33.
`
`Antibody libraries: construction, display and selection. Library
`display has superseded hybridoma technology for the selec-
`tion of human antibodies through the creation of large nat-
`ural and synthetic immune repertoires in vitro34–36. From these
`
`130
`
`NATURE MEDICINE • VOLUME 9 • NUMBER 1 • JANUARY 2003
`
`Lassen - Exhibit 1046, p. 2
`
`

`

`REVIEW
`
`libraries, specific high-affinity antibodies can be selected by
`linking phenotype (binding affinity) to genotype, thereby al-
`lowing simultaneous recovery of the gene encoding the se-
`lected antibody. Antibodies are usually displayed as
`monovalent Fab or scFv fragments and then, as required, re-
`assembled into intact Ig or multivalent variants after selec-
`tion34,37. If the repertoire is sufficiently large, a high-affinity
`Fab or scFv can be selected directly or, more frequently, the
`recovered gene can be subjected to cycles of mutation and
`further selection to enhance affinity (Fig. 2). Furthermore,
`new methods of selection and screening have been designed
`to specifically isolate antibodies with desired characteristics,
`such as enhanced stability, high expression or capacity to ac-
`tivate receptors38,39.
`
`Bacteriophage display. Fd phage and Fd phagemid technolo-
`gies are currently the most widely used in vitro methods for
`the display of large repertoires and for the selection of high-
`affinity recombinant antibodies against a range of clinically
`important target molecules35,37,40,41. Innovative selection
`methods have proved powerful for isolating antibodies
`against previously refractory antigens, such as new tumor-as-
`sociated antigens, cell surface receptors and HLA-A1-pre-
`sented peptides35,42. Important improvements in selection
`technology have included array screening for high-avidity
`antibodies43 and recovery of internalized phage from live cells
`to select against internalizing (human) receptors38. Phage
`technology has been applied to complete proteome analysis
`using membrane-based screening41. The latest methods for
`generating large phage libraries and avoiding the limitations
`imposed by bacterial cell transformation are discussed in the
`Supplementary Note online.
`
`Libraries of mRNA-protein complexes. Ribosome display relies
`on stabilized complexes of antibody, ribosome and mRNA to
`replace bacteriophage as the display platform39,44,45. Ribosome
`complexes are constructed totally in vitro, thereby eliminating
`the need for cell transformation and allowing the production
`of large libraries, ≤1014 members. The system is limited only by
`the requirement of a ribonuclease-free environment for selec-
`tion and buffer compositions suitable for antibody folding.
`Indeed, picomolar affinity antibodies have been selected and
`rapid affinity maturation cycles carried out using this innova-
`
`tive in vitro method39,45. Covalent display using puromycin-sta-
`bilized mRNA-protein complexes is an alternative strategy to ri-
`bosome display46,47.
`
`Cell surface libraries. Before the advent of bacteriophage sys-
`tems, antibodies had been displayed on or in bacterial cells,
`although replica plating had limited screening to libraries of
`<108. The recent development of high-speed flow cytometers
`has re-activated the efforts in cell surface display, and several
`high-affinity antibodies have been
`isolated by
`this
`method48,49.
`
`Transgenic mice. Transgenic mice have been produced that
`lack the native mouse immune repertoire and instead harbor
`most of the human antibody repertoire in the germline.
`Injection of antigens into these mice leads to the development
`of human antibodies that have undergone mouse somatic hy-
`permutation and selection to relatively high affinity7.
`Antibodies can be recovered by classic hybridoma technology
`or, for more efficient affinity enhancement, by in vitro display
`and selection technologies (Fig. 2).
`
`Production, stability and expression levels
`Production of antibodies for preclinical and clinical trials has
`been evaluated in numerous expression systems, including
`bacteria, yeast, plant, insect and mammalian cells. Bacteria
`are favored for expression of small, non-glycosylated Fab and
`scFv fragments, usually with terminal polypeptides such as c-
`Myc, His or FLAG, for affinity purification11. Mammalian or
`plant cells are favored for intact antibodies and, occasionally,
`also for expression of scFvs, diabodies and minibodies14,50
`There is still hope that eukaryotic cell cultures, such as those
`of the yeast Pichia pastoris, will allow efficient production of
`fully processed scFvs, albeit with high-mannose oligosaccha-
`rides51. Additional expression methods are discussed in the
`Supplementary Note online.
`
`Affinity maturation
`Both transgenic mice and display libraries typically produce
`human antibodies with binding affinities (KD) ranging from
`10–7 to 10–9 M. Obtaining higher-affinity antibodies is impor-
`tant for efficient binding to the antigenic target for in vitro di-
`agnosis, viral neutralization, cell targeting and in vivo
`
`Table 1 FDA-approved therapeutic antibodies
`
`Product namea
`
`Orthoclone OKT3
`ReoPro
`Rituxan
`Zenapax
`Remicade
`Simulect
`Synagis
`Herceptin
`Mylotarg
`CroFab
`DigiFab
`Campath
`Zevalin
`
`Specificity
`
`CD3
`GpIIb/gpIIa
`CD20
`CD25
`TNF-α
`CD25
`RSV
`Her-2
`CD33
`Snake venom
`Digoxin
`CD52
`CD20
`
`Product type
`
`Mouse
`Chimeric Fab
`Chimeric
`Humanized
`Chimeric
`Chimeric
`Humanized
`Humanized
`Humanized
`Ovine Fab
`Ovine Fab
`Humanized
`Mouse
`
`Indication
`
`Transplant rejection
`Cardiovascular disease
`Non-Hodgkin lymphoma
`Transplant rejection
`Crohn disease, rheumatoid arthritis
`Transplant rejection
`Respiratory syncytial virus
`Metastatic breast cancer
`Acute myeloid leukemia
`Rattlesnake antidote
`Digoxin overdose
`Chronic lymphocytic leukemia
`Non-Hodgkin lymphoma
`
`Year
`
`1986
`1994
`1997
`1997
`1998, 1999
`1998
`1998
`1998
`2000
`2000
`2001
`2001
`2002
`
`aProduct names are registered trademarks. Recent developments in FDA approvals can be obtained from http://www.fda.gov/cber/efoi/approve.htm. Updates on products
`relevant to lymphoma immunotherapy can be obtained from http://www.lymphomainfo.net/therapy/immunotherapy/. The latest product developments, antibody formula-
`tion and prescribing details can usually be obtained from http://www.productname.com or the manufacturer’s Internet site.
`
`NATURE MEDICINE • VOLUME 9 • NUMBER 1 • JANUARY 2003
`
`131
`
`©2003 Nature Publishing Group http://www.nature.com/naturemedicine
`
`Lassen - Exhibit 1046, p. 3
`
`

`

`REVIEW
`
`a
`
`c
`
`1. Display
`
`4. Mutation
`
`2. Selection
`
`3. Amplification
`
`1. B-cell surface display
`
`In vivo:
`4.
`B cell germline
`Somatic recombination
`Somatic mutation
`Class switching
`
`2. Antigen stimulation:
`T-helper cell binds to
`activated B cell
`
`3. Clonal expansion of B cells
`
`b
`
`d
`
`Fig. 2 Affinity maturation cycles.
`a, General strategy of protein dis-
`play, selection (gene recovery), am-
`plification and gene mutation. For
`the specific affinity maturation cy-
`cles (b–d), alternative technologies
`are listed for each component of the
`cycle. b, Mammalian in vivo process of
`antibody maturation52. c, Bacterio-
`phage display of engineered anti-
`body repertoires (library) using an in
`vitro cycle of mutation and selec-
`tion. d, Ribosome display of engi-
`neered antibody repertoires (library)
`with an in vitro cycle of mutation
`and selection.
`
`1. Phage library
`
`1. Ribosome library
`
`In vitro:
`4.
` Error-prone PCR
`Mutating enzyme
`
`3. PCR
`
`2. Panning
`on antigen;
`recovery
`of binders
`by RT-PCR.
`
`Cloning for high-level expression
`
`Alternative scaffolds
`Intact antibodies, Fab and scFv
`fragments provide an antigen-
`binding surface comprising six
`CDR loops; these can be mu-
`tated, sequentially or collec-
`tively, to bind to a vast array of
`target molecules. Some target
`molecules are refractory to the
`immune repertoire, however,
`particularly those with cavities
`or clefts that require a small penetrating loop for tight binding.
`The natural mammalian antibody repertoire simply does not en-
`code penetrating loops, and only rarely has this type of antibody
`been selected57. Unexpectedly, both camelids (camels, llamas
`and related species) and sharks produce natural, single V-like do-
`main repertoires displaying cavity-penetrating CDR loops that
`complement the repertoire of conventional antibodies58–60. This
`theory has led to a number of attempts to design single-domain
`display libraries in vitro, based on V domains61 and other Ig-like
`scaffolds60,62. These small molecules complement both antibody
`and peptide libraries and are expected to have improved phar-
`macokinetics for several clinical applications, including those
`that require access to buried (immunosilent) sites or clefts in en-
`zymes, receptors and viruses62. Many important diagnostic tar-
`gets, notably prions63, have also been refractory to conventional
`antibodies, and we expect that new molecular libraries and scaf-
`folds will be required to provide the required binding reagents.
`
`4. In vitro
`Error-prone PCR
`DNA shuffling
`Chain shuffling
`Site-directed
`mutagenesis
`
`2. Panning
`on antigen;
`recovery
`of binders
`
`3. Transform
`into E. coli
`
`High-level expression
`
`©2003 Nature Publishing Group http://www.nature.com/naturemedicine
`
`imaging. To improve antibody affinity, various in vitro strate-
`gies have recently been optimized to mimic the mammalian
`in vivo process of somatic hypermutation and selection52 (Fig.
`2). These include site-specific mutagenesis based on structural
`information, combinatorial mutagenesis of complementar-
`ity-determining regions (CDRs), random mutagenesis of the
`entire gene or chain shuffling39,48,53,54.
`After a decade of developing library display strategies, it is
`now obvious that the most successful methods rely on several
`cycles of mutation, display, selection (recovery) and gene am-
`plification (Fig. 2). These cycles of mutation and selection can
`be carried out using either in vitro or in vivo strategies and
`have been far more effective than precisely designed alter-
`ations for affinity enhancement39,48. Even with the most de-
`tailed structural information, the techniques for design of
`precisely complementary surfaces through interface muta-
`tions remain in their infancy.
`Using the cycles depicted in Figure 2, affinity enhance-
`ment can be restricted to mutations in the antigen-binding
`surface (CDR loops). Importantly, mutations in the underly-
`ing framework regions have frequently provided large in-
`creases in affinity, stability and expression39,45,48. Random
`mutations over the entire V-domain genes can be derived
`from Escherichia coli mutator cells, homologous gene re-
`arrangements or error-prone PCR. Sequential ‘chain shuf-
`fling’ of the two V genes in the Fv module is also ‘random’
`but offers the advantage that only one V domain is altered
`at a time, while the other domain is kept constant to pro-
`vide a defined specificity55. Recent advances include the in-
`corporation of highly mutagenic enzymes such as mRNA
`reverse transcriptase and DNA polymerase with no proof-
`reading activity to achieve a high gene mutation rate56. The
`integration of such polymerases into the ribosome display
`and selection process could rapidly generate large libraries
`of mutants (Fig. 2).
`
`Clinical applications
`Aside from radioimmunotherapy (discussed earlier), there are
`a variety of other clinical applications of engineered antibod-
`ies for viral infection, cancer, autoimmune disease, allograft
`rejection, asthma, stroke and glaucoma surgery. Specific clin-
`ical applications are discussed below.
`
`Pathogen neutralization and antiviral therapy. Antibody
`binding can directly and effectively block the activity of many
`pathogens, often without requiring Fc-mediated cytotoxicity.
`Indeed, this has always been the promise of antibody-mediated
`viral neutralization. The first monoclonal antibody for the
`treatment of viral disease, Synagis, was approved by the FDA in
`1998 (Table 1). Synagis is a humanized antibody used for the
`prevention of severe respiratory syncytial virus (RSV) disease64.
`Despite this success, and the wide range of antibodies available
`against human immunodeficiency type 1 (HIV) and herpes sim-
`
`132
`
`NATURE MEDICINE • VOLUME 9 • NUMBER 1 • JANUARY 2003
`
`Lassen - Exhibit 1046, p. 4
`
`

`

`REVIEW
`
`plex virus (HSV), the use of recombinant antibodies as thera-
`peutics for viral infection has been limited65. Only a few rare an-
`tibodies have exhibited potent neutralization in vitro and
`antiviral efficacy in animal models57. This is probably due to
`viral efficiency both in producing escape mutants and in evolv-
`ing immunosilent receptor-binding surfaces65. For neutraliza-
`tion of other pathogenic molecules, monomeric Fab molecules
`have recently been approved by the FDA as antivenenes
`(CroFab; Table 1), and both scFv fragments and oligoclonal an-
`tibody mixtures have been effective against bacterial toxins66,67.
`
`Intracellular antibodies. Antibody fragments can be expressed
`as intracellular proteins, typically as scFvs termed intrabodies,
`and equipped with targeting signals either to neutralize intra-
`cellular gene products or to target cellular pathways. For exam-
`ple, expression of p21ras, erbB2, huntingtin and MHC have all
`been
`individually downregulated using
`antibodies68–71.
`Intrabodies also have important antiviral potential, particularly
`through their targeting of intracellular action to mandatory
`viral proteins such as the Vif, Tat or Rev components of HIV72.
`Antibody frameworks have been adapted that substantially im-
`prove expression levels and solubility in the intracellular reduc-
`ing environment73. Direct in vivo selection from large libraries
`will greatly facilitate the isolation of many previously unknown
`intracellular antibodies or ‘intrabodies’74,75. Obviously, the ex-
`pression of intrabodies in vivo can be encoded into gene therapy
`vectors, and this could ultimately be their most powerful clini-
`cal application.
`
`Cancer therapy and cell recruitment strategies. The promise of
`engineered antibodies for effective cancer therapy, especially ra-
`dioconjugates, has been described earlier. Blocking angiogene-
`sis to prevent the establishment and growth of tumors is
`becoming an important strategy76,77. Cancer cells can be de-
`stroyed by cell recruitment of cytotoxic T cells, natural-killer
`(NK) cells or macrophages that can be targeted by encoding cell
`surface antibodies (usually scFv)78. Alternative cell recruitment
`strategies include bifunctional antibodies, fused to cytokines for
`T-cell stimulation and proliferation at the tumor site79.
`
`Innovative vaccine applications. Troybodies are engineered
`vaccine antibodies containing cryptic T-cell epitopes to enhance
`antigen presentation80. Troybodies effectively target antigen-pre-
`senting cells (APCs) and, after processing, expose cryptic T-cell
`epitopes to direct T-cell activation. In the preferred format, the
`Fv domain provides APC specificity and the C domains encode
`the cryptic T-cell epitopes. These new vaccines can be redesigned
`to target many different APCs and enhance immunity to many
`different T-cell epitopes. Alternative vaccine strategies include
`the use of engineered APC-targeted antibodies that direct aden-
`oviruses to deliver vaccine-inducing epitopes as a gene therapy
`capsule30 and B7-targeted scaffolds (scFv and VL domains) that
`enable antigen-loading of dendritic cells61.
`
`Biosensors and microarrays: the future of diagnosis.
`A likely prediction is that biosensing devices and microarrays
`will dominate the in vitro diagnostic market by 2005. Antibodies
`currently provide high-sensitivity reagents for a huge range of
`diagnostic kits, accounting for approximating 30% of the $20
`billion per year diagnostic industry. It is therefore not surprising
`that antibodies are the paradigm for proof-in-principle of new
`biosensing devices, focused initially on glass-surface microar-
`
`rays81. Already in 2002 we have seen more protein-friendly sur-
`faces being developed as array platforms for antibody-based di-
`agnosis (Triage from Biosite, San Diego, California; Protein
`Profiling Biochip from Zyomyx, Hayward, California; Hydrogel
`from Perkin-Elmer, Boston, Massachusetts). These platforms
`will become increasingly available over the next few years, dri-
`ven by the demand for new reagents to diagnose the vast array
`of biomarkers stemming from proteomics discovery programs.
`These platforms will also be developed for robust ex vivo appli-
`cations, including the detection of microbial contaminants,
`pesticides and biological (warfare) pathogens82.
`
`Conclusion
`During the past few years, there has been growing excitement in
`both scientific and commercial communities in engineered anti-
`bodies. Scientific interest has stemmed from the elucidation of
`the key elements required for antibody fragment design, effi-
`cient expression and desired pharmacokinetics. Commercial in-
`terest is driven by increasing sales (Table 1). Many of the
`antibody fragments and fusion proteins discussed in this review
`are now undergoing scale-up production. It is likely that we will
`soon witness development of display and screening technologies
`incorporating nanoarray robotics. By providing a highly stable,
`protease-resistant scaffold, engineered recombinant antibody
`fragments will continue to be the model for selection of high-
`affinity clinical targeting reagents.
`
`Note: Supplementary information is available on the Nature Medicine website.
`
`Acknowledgments
`We thank many colleagues at CSIRO and the CRC for their helpful contribu-
`tions, and especially B. Mason and K. Wark for their assistance with the
`manuscript.
`
`1. Milstein, C. With the benefit of hindsight. Immunol. Today 21, 359–364 (2000).
`2. Carter, P. Improving the efficacy of antibody-based cancer therapies. Nat. Rev.
`Cancer 1, 118–129 (2001).
`3. Hudson, P.J. & Souriau, C. Recombinant antibodies for cancer diagnosis and
`therapy. Expert Opin. Biol. Ther. 1, 845–855 (2001).
`4. Wiseman, G.A. et al. Radiation dosimetry results for Zevalin radioimmunotherapy
`of rituximab-refractory non-Hodgkin lymphoma. Cancer 94, 1349–1357 (2002).
`5. Clynes, R.A., Towers, T.L., Presta, L.G. & Ravetch, J.V. Inhibitory Fc receptors
`modulate in vivo cytoxicity against tumor targets. Nat. Med. 6, 443–446 (2000).
`6. Ober, R.J., Radu, C.G., Ghetie, V. & Ward, E.S. Differences in promiscuity for an-
`tibody–FcRn interactions across species: implications for therapeutic antibodies.
`Int. Immunol. 13, 1551–1559 (2001).
`7. He, Y. et al. Efficient isolation of novel human monoclonal antibodies with neu-
`tralizing activity against HIV-1 from transgenic mice expressing human Ig loci.
`J. Immunol. 169, 595–605 (2002).
`8. Goel, A. et al. 99mTc-labeled divalent and tetravalent CC49 single-chain Fvs: novel
`imaging agents for rapid in vivo localization of human colon carcinoma. J. Nucl.
`Med. 42, 1519–1527 (2001).
`9. Adams, G.P. et al. High affinity restricts the localization and tumor penetration of
`single-chain Fv antibody molecules. Cancer Res. 61, 4750–4755 (2001).
`10. Tomlinson, I. & Holliger, P. Methods for generating multivalent and bispecific an-
`tibody fragments. Methods Enzymol. 326, 461–479 (2000).
`11. Todorovska, A. et al. Design and application of diabodies, triabodies and tetra-
`bodies for cancer targeting. J. Immunol. Methods 248, 47–66 (2001).
`12. Nielsen, U.B., Adams, G.P., Weiner, L.M. & Marks, J.D. Targeting of bivalent anti-
`ErbB2 diabody antibody fragments to tumor cells is independent of the intrinsic
`antibody affinity. Cancer Res. 60, 6434–6440 (2000).
`13. Tahtis, K. et al. Biodistribution properties of indium-111-labeled C-functionalized
`trans-cyclohexyl diethylenetriaminepentaacetic acid humanized 3S193 diabody
`and F(ab′)2 constructs in a breast carcinoma xenograft model. Clin. Cancer Res. 7,
`1061–1072 (2001).
`14. Yazaki, P.J.et al. Tumor targeting of radiometal-labeled anti-CEA recombinant T84.66
`diabody and t84.66 minibody: comparison to radioiodinated fragments. Bioconjug.
`Chem. 12, 220–228 (2001).
`15. Gura, T. Therapeutic antibodies: magic bullets hit the target. Nature 417, 584–586
`(2002).
`16. Izumi, Y., Xu, L., di Tomaso, E., Fukumura, D. & Jain, R.K. Tumour biology: herceptin
`
`NATURE MEDICINE • VOLUME 9 • NUMBER 1 • JANUARY 2003
`
`133
`
`©2003 Nature Publishing Group http://www.nature.com/naturemedicine
`
`Lassen - Exhibit 1046, p. 5
`
`

`

`REVIEW
`
`acts as an anti-angiogenic cocktail. Nature 416, 279–280 (2002).
`17. Cardarelli, P.M. et al. Binding to CD20 by anti-B1 antibody or F(ab′)2 is sufficient for
`induction of apoptosis in B-cell lines. Cancer Immunol. Immunother. 51, 15–24 (2002)