04/03/2001
`
`04:18 PM Cancer Research-V.81
`
`PS039-03.tex
`
`PS039-03.xml
`
`LaTeX2e(2000/11/15)
`
`Textures 2.0
`
`Recombinant Immunotoxins in
`Targeted Cancer Cell Therapy
`
`Yoram Reiter
`Faculty of Biology
`Technion-Israel Institute of Technology
`Haifa 3200, Israel
`
`I. Introduction
`II. Design of Recombinant Immunotoxins
`A. The Toxin Moiety
`B. The Targeting Moiety—Recombinant Antibody Fragments
`III. Construction and Production of Recombinant Immunotoxins
`IV. Preclinical Development of Recombinant Immunotoxins
`V. Application of Recombinant Immunotoxins
`A. Recombinant Immunotoxins against Solid Tumors
`B. Recombinant Immunotoxins against Leukemias and Lymphomas
`VI. Other Applications of Recombinant Antibody Fragments
`A. Radioimaging and Radioimmunotherapy
`B. Prodrug Therapy with Fv-Enzyme Fusion Proteins
`VII. Challenges and Future Directions of Recombinant Immunotoxins
`A. Immune Responses and Dose-Limiting Toxicity
`B. Specificity
`References
`
`Targeted cancer therapy in general and immunotherapy in particular combines ratio-
`nal drug design with the progress in understanding cancer biology. This approach takes
`advantage of our recent knowledge of the mechanisms by which normal cells are trans-
`formed into cancer cells, thus using the special properties of cancer cells to device novel
`therapeutic strategies.
`Recombinant immunotoxins are excellent examples of such processes, combining the
`knowledge of antigen expression by cancer cells with the enormous developments in
`recombinant DNA technology and antibody engineering.
`Recombinant immunotoxins are composed of a very potent protein toxin fused to
`a targeting moiety such as a recombinant antibody fragment or growth factor. These
`molecules bind to surface antigens specific for cancer cells and kill the target cells by
`catalytic inhibition of protein synthesis.
`Recombinant immunotoxins are developed for solid tumors and hematological ma-
`lignancies and have been characterized intensively for their biological activity in vitro on
`cultured tumor cell lines as well as in vivo in animal models of human tumor xenografts.
`The excellent in vitro and in vivo activities of recombinant immunotoxins have lead to
`their preclinical development and to the initiation of clinical trail protocols. Recent trail
`
`Advances in CANCER RESEARCH
`0065-230X/01 $35.00
`
`93
`
`Copyright (cid:2)C 2001 by Academic Press.
`All rights of reproduction in any form reserved.
`
`IMMUNOGEN 2152, pg. 1
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`04/03/2001
`
`04:18 PM Cancer Research-V.81
`
`PS039-03.tex
`
`PS039-03.xml
`
`LaTeX2e(2000/11/15)
`
`Textures 2.0
`
`94
`
`Yoram Reiter
`
`results have demonstrated potent clinical efficacy in patients with malignant diseases that
`are refractory to traditional modalities of cancer treatment: surgery, radiation therapy,
`and chemotherapy.
`The results demonstrate that such strategies can be developed into a separate modality
`of cancer treatment with the basic rationale of specifically targeting cancer cells on the
`basis of their unique surface markers. Efforts are now being made to improve the current
`molecules and to develop new agents with better clinical efficacy. This can be achieved
`by development of novel targeting moieties with improved specificity that will reduce
`toxicity to normal tissues.
`In this review, the design, construction, characterization, and applications of recom-
`binant immunotoxins are described. Results of recent clinical trails are presented, and
`future directions for development of recombinant immunotoxins as a new modality for
`cancer treatment are discussed. (cid:2)C 2001 Academic Press.
`
`I. INTRODUCTION
`
`The rapid progress in understanding the molecular biology of cancer cells
`has made a large impact on the design and development of novel therapeutic
`strategies. These are developed because treatment of cancer by chemotherapy
`is limited by a number of factors and usually fails in patients whose malig-
`nant cells are not sufficiently different from normal cells in their growth and
`metabolism. Other limiting factors are the low therapeutic index of most
`chemotherapeutic agents, the emergence of drug-resistant populations, tu-
`mor heterogeneity, and the presence of metastatic disease. The concept of
`targeted cancer therapy is thus an important means to improve the thera-
`peutic potential of anticancer agents and lead to the development of novel
`approaches such as immunotherapy.
`The approach of cancer immunotherapy and targeted cancer therapy com-
`bines rational drug design with the progress in understanding cancer biology
`(1–4). This approach takes advantage of some special properties of cancer
`cells: many of them contain mutant or overexpressed oncogenes on their
`surface, and these proteins are attractive antigens for targeted therapy. The
`first cell-surface receptor to be linked to cancer was the EGF receptor, which
`is present in lung, brain, kidney, bladder, breast, and ovarian cancer (5, 6).
`Several other members of the EGF family of receptors, the erbB2, erbB3, and
`erbB4 receptors, appear to be abundant on tumors of breast and ovary and
`erbB2, for example, is the target for Phase I and II immunotherapy clinical
`trails (7, 8).
`Other promising candidates for targeted therapy are differentiation anti-
`gens that are expressed on the surface of mature cells but not on the immature
`stem cells.
`The most widely studied examples of differentiation antigens which are
`currently being used for targeted therapy are expressed by hematopoietic
`
`IMMUNOGEN 2152, pg. 2
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`04/03/2001
`
`04:18 PM Cancer Research-V.81
`
`PS039-03.tex
`
`PS039-03.xml
`
`LaTeX2e(2000/11/15)
`
`Textures 2.0
`
`Recombinant Immunotoxins
`
`95
`
`malignancies and include CD19, CD20, and CD22 on B-cell lymphomas
`and leukemias and the IL-2 receptor on T-cell leukemias (9–11). Differenti-
`ation antigens have also been found on ovarian, breast, and prostate cancer
`(12–14).
`Another class of antigens, termed tumor-associated antigens (TAA), are
`molecules which are tightly bound to the surface of cancer cells and are as-
`sociated with the transformed cancer cells. An example is the carbohydrate
`antigen Lewis Y, which is found in many types of solid tumors (15). An-
`other class of TAAs are cancer peptides that are presented by class I MHC
`molecules on the surface of tumor cells (16, 17).
`It should be possible to use these molecular cell-surface markers as targets
`to eliminate the cancer cells while sparing the normal cells. For this approach
`to be successful, we must generate a targeting moiety which will bind very
`specifically the antigen or receptor expressed on the cancer cell surface and
`arm this targeting moiety with an effector cytotoxic moiety. The targeting
`moiety can be a specific antibody directed toward the cancer antigen or
`a ligand for specific overexpressed receptor. The cytotoxic arm can be a
`radioisotope, a cytotoxic drug, or a toxin. One strategy to achieve this is
`to arm antibodies that target cancer cells with powerful toxins which can
`originate from both plants and bacteria. The molecules generated are termed
`recombinant immunotoxins.
`The goal of immunotoxin therapy is to target a very potent cytotoxic agent
`to cell surface molecules which will internalize the cytotoxic agent and result
`in cell death. Developing this type of therapy has attracted much interest in
`the past years. Since immunotoxins differ greatly from chemotherapy in their
`mode of action and toxicity profile, it is hoped that immunotoxins will have
`the potential to improve the systemic treatment of tumors incurable with
`existing modes of therapy.
`As shown in Fig. 1, immunotoxins can be divided into two groups: chemi-
`cal conjugates (or first-generation immunotoxins) and second-generation (or
`recombinant immunotoxins). They both contain toxins that have their cell-
`binding domains either mutated or deleted to prevent them from binding to
`normal cells, and that are either fused or chemically conjugated to a ligand
`or an antibody specific for cancer cells (Table I).
`First-generation immunotoxins, composed of whole antibodies chemi-
`cally conjugated to toxins, demonstrated the feasibility of this concept.
`Cancer cells cultured in vitro could be killed under conditions in
`which the immunotoxin demonstrated low toxicity toward cultured nor-
`mal cells. Clinical trials with these agents had some success; however, they
`also revealed several problems, such as nonspecific toxicity toward
`some normal cells, difficulties in production, and, particularly for the
`treatment of solid tumors, poor tumor penetration owing to their large
`size.
`
`IMMUNOGEN 2152, pg. 3
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`04/03/2001
`
`04:18 PM Cancer Research-V.81
`
`PS039-03.tex
`
`PS039-03.xml
`
`LaTeX2e(2000/11/15)
`
`Textures 2.0
`
`96
`
`Yoram Reiter
`
`Fig. 1 Immunotoxins for targeted cancer therapy. First-generation immunotoxins are whole
`monoclonal antibodies to which the toxin is chemically conjugated. Second-generation immuno-
`toxins made by recombinant DNA technology by fusing recombinant antibody fragments to
`the toxin (usually a truncated or mutated form of the toxin).
`Three types of recombinant antibody fragments are used as the targeting moiety in recom-
`binant immunotoxins. Fabs are composed of the light-chain and the heavy-chain Fd fragments
`(VH and CH1), connected to each other via the interchain disulfide bond between CL and CH1.
`ScFv fragments are stabilized by a peptide linker which connects the carboxyl terminus of VH
`or VL with the amino terminus of the other domain. The VH and VL heterodimer in dsFv
`is stabilized by engineering a disulfide bond between the two domains. The biochemical and
`biological properties described in the figure are depicted for B3-lysPE38 (LMB-1) (89) (a first-
`generation antibody–PE chemical conjugate), B3(Fv)-PE38 (LMB-7) (67) (second-generation
`recombinant scFv-immunotoxin for a scFv-immunotoxin), and B3(dsFv)-PE38 (LMB-9) (74)
`(for a second-generation recombinant dsFv-immunotoxin).
`
`Second-generation immunotoxins have overcome many of these problems.
`Progress in the elucidation of the toxins’ structure and function, combined
`with the techniques of protein engineering, facilitated the design and con-
`struction of recombinant molecules with a higher specificity for cancer cells
`and reduced toxicity to normal cells. At the same time, advances in recom-
`binant DNA technology and antibody engineering enabled the generation
`of small antibody fragments. Thus, it was possible to decrease the size of
`immunotoxins significantly and to improve their tumor-penetration poten-
`tial in vivo. The development of advanced methods of recombinant-protein
`
`IMMUNOGEN 2152, pg. 4
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`04/03/2001
`
`04:18 PM Cancer Research-V.81
`
`PS039-03.tex
`
`PS039-03.xml
`
`LaTeX2e(2000/11/15)
`
`Textures 2.0
`
`Recombinant Immunotoxins
`
`97
`
`Table I Examples of Recombinant Immunotoxins against Cancer
`
`Immunotoxin
`
`Antigen
`
`Toxin
`
`Cancer
`
`Clinical trail References
`
`Anti CD7-dgA
`
`CD7
`
`DAB389-IL2
`
`IL-2R
`
`Anti-Tac (Fv)-PE38 CD25
`(LMB-2)
`DT-Anti-Tac(Fv)
`
`CD25
`
`RFB4(dsFv)-PE38
`Di-dgA-RFB4
`
`CD22
`CD22
`
`B3-lysPE38
`(LMB-1)
`B3(Fv)-PE38
`(LMB-7)
`B3(dsFv)-PE38
`(LMB-9)
`BR96(sFv)-PE40
`e23(Fv)-PE38
`FRP5(scFv)ETA
`Tf-CRM107
`HB21(Fv)-PE40
`MR1(Fv)-PE38
`SSI(Fv)-PE38
`
`DT
`
`PE
`
`DT
`
`Ricin Non-Hodgkin’s
`lymphoma
`T-cell lymphoma,
`Hodgkin’s disease
`B and T lymphoma,
`leukemias
`Leukemias,
`lymphoma
`B leukemias
`PE
`Ricin Leukemias,
`non-Hodgkin’s
`lymphoma
`Carcinomas
`
`Lewis Y
`
`Lewis Y
`
`Lewis Y
`
`PE
`
`PE
`
`PE
`
`Carcinomas
`
`Carcinoma
`
`PE
`Lewis Y
`PE
`erbB2/HER2
`PE
`erbB2/HER2
`DT
`Transferrin-R
`PE
`Transferrin-R
`Mutant EGF-R PE
`Mesothelin
`PE
`
`Carcinoma
`Breast cancer
`Breast cancer
`Glioma
`Various
`Liver, brain tumors
`Ovarian cancer
`
`Phase I
`
`156
`
`Phase III
`
`122, 125
`
`Phase I
`
`66,106
`
`—
`
`Phase I
`—
`
`Phase I
`
`Phase I
`
`Phase I
`
`—
`Phase I
`—
`Phase I
`—
`—
`—
`
`78
`
`76
`157
`
`89
`
`67
`
`74
`
`95
`68
`96
`103
`55
`99
`100
`
`production enabled the large-scale production of recombinant immunotox-
`ins of high purity and quality for clinical use in sufficient quantities to per-
`form clinical trials.
`Another strategy to target cancer cells is to construct chimeric toxins in
`which the engineered truncated portion of the toxin (PE or DT) gene is fused
`to cDNA encoding growth factors or cytokines. These include transforming
`growth factor (TGF)-␣ (18), insuline-like growth factor (IGF)-1 (19), acidic
`and basic fibroblast growth factor (FGF) (20), IL2 (21), IL4 (22), and IL6
`(23). These recombinant toxins (oncotoxins) are designed to target specific
`tumor cells that overexpress these receptors.
`This review will summarize knowledge of the design and application of
`second-generation recombinant Fv immunotoxins, which utilize recombi-
`nant antibody fragments as the targeting moiety, in the treatment of cancer,
`
`IMMUNOGEN 2152, pg. 5
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`04/03/2001
`
`04:18 PM Cancer Research-V.81
`
`PS039-03.tex
`
`PS039-03.xml
`
`LaTeX2e(2000/11/15)
`
`Textures 2.0
`
`98
`
`Yoram Reiter
`
`and will also discuss briefly the use of recombinant antibody fragments for
`other modes of cancer therapy and diagnosis.
`
`II. DESIGN OF RECOMBINANT IMMUNOTOXINS
`
`A. The Toxin Moiety
`
`The toxins that are most commonly used to make immunotoxins are ricin,
`diphtheria toxin (DT), and Pseudomonas exotoxin (PE). The genes for these
`toxins have been cloned and expressed in Eschesichia coli, and the crystal
`structures of all three proteins have been solved (36, 41). This information,
`in combination with mutational studies, has elucidated which toxin subunits
`are involved in their biological activity and, most important, the different
`steps of the cytocidal process. DT, PE, and ricin, and their derivatives, have
`all been successfully used to prepare immunotoxin conjugates (3, 24), but
`only PE- and DT-containing fusion proteins generate active recombinant
`immunotoxins (1, 25). This is because the toxic moiety must be separated
`from the binding moiety after internalization (26, 27). PE and DT fusion
`proteins generate their free toxic moieties by proteolytic processing. Ricin
`does not possess such a proteolytic processing site, and therefore cannot be
`attached to the targeting moiety with a peptide bond without losing cytotoxic
`activity. Recently, proteolytic processing sites were introduced into ricin by
`recombinant DNA techniques to try to overcome this problem (28).
`
`1. DIPHTHERIA TOXIN AND DT DERIVATIVES
`
`DT is a 58-kDa protein, secreted by pathogenic Corynebacterium diph-
`theria, which contain a lysogenic beta phage (29). DT ADP-ribosylates eu-
`karyotic elongation factor 2 (EF2) at a “diphthamide” residue located at
`+
`His 415, using NAD
`as a cofactor (30). This modification arrests protein
`synthesis and subsequently leads to cell death (31). Only a few, and perhaps
`only one, DT molecule needs to reach the cytosol in order to kill a cell. When
`DT is isolated from the culture medium of C. diphtheria, it is composed of
`an N-terminal 21-kDa A subunit and a C-terminal 37-kDa B subunit held
`together by a disulfide bond. DT is the expression product of a single gene
`(29), which, when secreted into the medium, is processed into two fragments
`by extracellular proteases. When DT is produced as a recombinant single-
`chain protein in E. coli, it is not cleaved by the bacteria but is instead cleaved
`by a protease in the target cells (32).
`The A domain of DT contains its enzymatic activity. The N terminus of
`the B subunit of DT (or the region between A and B in single-chain DT)
`
`IMMUNOGEN 2152, pg. 6
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`04/03/2001
`
`04:18 PM Cancer Research-V.81
`
`PS039-03.tex
`
`PS039-03.xml
`
`LaTeX2e(2000/11/15)
`
`Textures 2.0
`
`Recombinant Immunotoxins
`
`99
`
`mediates translocation of the A subunit into the cytoplasm. The B domain,
`especially its C terminus, is responsible for the binding of DT to target cells.
`Deletions or mutations in this part of the molecule abolish or greatly di-
`minish the binding and toxicity of DT (33–35). DT enters cells via coated
`pits and is proteolytically cleaved within the endocytic compartment if it is
`not already in the two-chain form, and reduced. It also undergoes a con-
`formational change at the acidic pH present in endosomes, which probably
`assists translocation of the A chain into the cytosol perhaps via a porelike
`structure mediated by the B chain (36–38). Derivatives of DT that are used
`to make immunotoxins have the C terminus altered by mutations or par-
`(cid:4)
`tially deleted (DAB486
`DAB389, DT388) but retain the translocation and
`ADP-ribosylation activity of DT (39). Recombinant antibody-fusion pro-
`teins with such derivatives target only cells that bind the antibody moiety of
`the immunotoxin.
`
`2. PSEUDOMONAS EXOTOXIN (PE) AND PE DERIVATIVES
`
`Two major research studies have enabled the use and genetic manipu-
`lation of PE for the design of immunotoxins (1–3). The first is the eluci-
`dation of the crystal structure of PE, showing the toxin to be composed
`of three major structural domains; the second is the finding that these do-
`mains are different functional modules of the toxin. PE is a single-chain
`66-kDa molecule secreted by Pseudomonas aeruginosa that, like DT, ir-
`+
`reversibly ADP-ribosylates the diphthamide residue of EF2, using NAD
`as cofactor (40). As a consequence, protein synthesis is inhibited and cell
`death ensues. PE is composed of three major domains (41). Different func-
`tions have been assigned to each domain by mutational analysis (42). The
`N-terminal domain 1a mediates binding to the a2 macroglobulin receptor
`(43). Domain lb is a small domain that lies between domain II and domain
`III and has no known function (44). Domain II mediates translocation of do-
`main III, the carboxyl-terminal ADP-ribosylating domain, into the cytosol
`of target cells (45) (see Fig. 2). Translocation occurs after internalization of
`the toxin and after a variety of other steps, including a pH-induced confor-
`mational change (46–48), proteolytic cleavage at a specific site in domain II
`(27), and a reductive step that separates the amino and carboxyl fragments.
`Ultimately, the carboxyl-terminal portion of PE is translocated from the
`endoplasmic reticulum into the cytosol. Despite a similar mode of action
`of PE and DT, which is ADP-ribosylation, and a similar initial pathway
`of cell entry (internalization via coated pits and endocytic vesicles) and of
`processing (proteolytic cleavage and a reductive step), they share almost no
`sequence homology. The only similarity is the spatial arrangement of key
`residues in their active sites, which are arranged around residue Glu553 in
`PE and Glu145 in DT (49–52).
`
`IMMUNOGEN 2152, pg. 7
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`04/03/2001
`
`04:18 PM Cancer Research-V.81
`
`PS039-03.tex
`
`PS039-03.xml
`
`LaTeX2e(2000/11/15)
`
`Textures 2.0
`
`100
`
`Yoram Reiter
`
`Fig. 2 The biological activity of Pseudomonas exotoxin A. The Fv portion of the immuno-
`toxin targets domains II and III of PE to a cell surface receptor or other target molecule on the
`tumor cell (A). The immunotoxin enters the cell by internalization and is transferred into the
`endosome (B). Within the endosome, the molecule unfolds due to a fall in pH. The conforma-
`tional change exposes a proteolytic site, and a proteolytic cleavage occurs in the translocation
`domain between amino acid 279 and 280 (C). A disulfide bond is then broken, thus creating
`two fragments: the Fv moiety and a small part of domain II, and the rest of domain II connected
`to domain III (D). The carboxyl-terminal fragment containing the ADP-ribosylation domain
`(domain III) and most of the translocation domain (domain II) are carried into the endoplasmic
`reticulum (E), and translocation occurs from the endoplasmic reticulum into the cytosol (F).
`The enzymatically active domain ADP-ribosylates elongation factor 2 at a diphtamide residue
`+
`located at His 415, using NAD
`as a cofactor. This modification arrests protein synthesis and
`subsequently leads to cell death by apoptosis. In DT the poteolytic processing occurs between
`residues 193 and 194. The catalytic A-chain (amino acids 1–193) then translocates to the cytosol
`through the endosome with the help of translocation domain residues 326–347 and 358–376,
`which form an ion channel.
`
`When the whole toxin is used to make an immunotoxin, nonspecific tox-
`icity occurs mainly due to binding of the toxin portion to cells, mediated by
`the binding domain. Consequently, the goal of making improved derivatives
`of PE-based immunotoxins has been to inactivate or remove the binding
`domain. Molecules in which the binding domain has been retained but in-
`activated by mutations were made (53); however, a better alternative to
`
`IMMUNOGEN 2152, pg. 8
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`04/03/2001
`
`04:18 PM Cancer Research-V.81
`
`PS039-03.tex
`
`PS039-03.xml
`
`LaTeX2e(2000/11/15)
`
`Textures 2.0
`
`Recombinant Immunotoxins
`
`101
`
`inactivating the cell-binding domain by mutations is to remove it from PE.
`The prototype molecule with this sort of deletion is PE40 (aa 253–613, MW
`40 kDa). Because PE40 and its derivatives described below lack the bind-
`ing domain (aa 1–252), they have very low nonspecific toxicity, but make
`very active and specific immunotoxins when fused to recombinant antibod-
`ies (54, 55). Currently, almost all PE-derived recombinant immunotoxins are
`constructed with PE38 (MW 38 kDa), a PE40 derivative that has, in addi-
`tion to the deletion of domain 1a, a second deletion encompassing a portion
`of domain Ib (aa 365–379) (44). Another useful mutation is to change the
`carboxyl-terminal sequence of PE from REDLK to KDEL. This improves the
`cytotoxicity of PE and its derivatives, presumably by increasing their delivery
`to the endoplasmic reticulum, where translocation takes place (56, 57).
`
`B. The Targeting Moiety—Recombinant
`Antibody Fragments
`
`The antibody moiety of the recombinant immunotoxin is responsible for
`specifically directing the immunotoxin to the target cell, meaning that the
`usefulness of the immunotoxin depends on the specificity of the antibody
`or antibody fragment that is connected to the toxin. Consequently, for the
`construction of recombinant immunotoxins, the only antibodies that should
`be used are those that recognize antigens that are expressed on target cancer
`cells and are not present on normal cells, present at very low levels, or are
`only present on less essential cells (Table I).
`Receptors for growth factors such as the epidermal growth factor (EGF),
`interleukin 2 (IL-2), interleukin 4 (IL-4), interleukin 6(IL-6), or erbB2 are
`common targets for targeted cancer therapy because they are highly ex-
`pressed on many cancer cells. Other carcinoma-related antigens include de-
`velopmental antigens such as complex carbohydrates, which are often highly
`abundant on the surface of cancer cells. The use of antibodies for immuno-
`toxin production also requires that the antibody–antigen complex be inter-
`nalized, because the mechanism of PE-toxin killing requires endocytosis as
`a first step in the entry of the toxin into the cell.
`Recombinant immunotoxins contain antibody fragments as the targeting
`moiety. These fragments can be produced in E. coli and are the result of
`intensive research and development in recombinant-antibody technologies
`(58–60). Several antibody fragments have been used to construct recombi-
`nant immunotoxins (Fig. 1). One type contains Fab fragments in which the
`light-chain and the heavy-chain Fd fragments (VH and CH1) are connected
`to each other via an interchain disulfide bond between CL and CH1. The
`toxin moiety can be fused to the carboxyl end of either CL or CH1.
`
`IMMUNOGEN 2152, pg. 9
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`04/03/2001
`
`04:18 PM Cancer Research-V.81
`
`PS039-03.tex
`
`PS039-03.xml
`
`LaTeX2e(2000/11/15)
`
`Textures 2.0
`
`102
`
`Yoram Reiter
`
`IMMUNOGEN 2152, pg. 10
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`04/03/2001
`
`04:18 PM Cancer Research-V.81
`
`PS039-03.tex
`
`PS039-03.xml
`
`LaTeX2e(2000/11/15)
`
`Textures 2.0
`
`Recombinant Immunotoxins
`
`103
`
`Fabs can be produced in E. coli either by secretion, with coexpression of
`light chains and Fd fragments, or by expression of the chains in intracellular
`inclusion bodies in separate cultures; in the latter case, they are reconsti-
`tuted by a refolding reaction using a redox-shuffling buffer system. Several
`immunotoxins with Fab fragments have been constructed and produced in
`this way (1–4, 61).
`The smallest functional modules of antibodies required for antigen binding
`are Fv fragments. This makes them especially useful for clinical applications,
`not only for generating recombinant immunotoxins but also for tumor imag-
`ing, because their small size improves tumor penetration. Fv fragments are
`heterodimers of the variable heavy-chain (VH) and the variable light-chain
`(VL) domains. Unlike whole IgG or Fab, in which the heterodimers are held
`together and stabilized by interchain disulfide bonds, the VH and VL of Fvs
`are not covalently connected and are consequently unstable; this instabil-
`ity can be overcome by making recombinant Fvs that have the VH and VL
`←−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−
`Fig. 3 Cloning, construction, and composition of scFv- and dsFv-immunotoxins. (A) cloning
`and construction of recombinant scFv and dsFv immunotoxins. The genes encoding the VH
`and VL variable domains are cloned usually from hybridoma mRNA by reverse transcription,
`cDNA synthesis, and subsequent PCR amplification using degenerate primers that are comple-
`(cid:4)
`(cid:4)
`mentary to the 5
`or 3
`end of the VH and VL genes or by primers which are designed according
`to the amino-terminal amino acid sequence of the MAb to be cloned and conserved sequences
`at the N terminal of the heavy and light constant regions. The variable genes can be also cloned
`by constant-domain primers and using the RACE method (rapid amplification of cDNA ends).
`Restriction sites for assembling of the peptide linker sequence which connects the VH and VL
`domains, and for cloning into the expression vector, are also introduced by PCR. Construction
`of dsFv involves the generation of two expression plasmids which encode the two components
`of the dsFv VH-cys and VL-cys. The cysteines are introduced in position 44 in FR2 of VH and
`position 100 of FR4 of VL or position 105 of FR4 in VH and position 43 of FR2 in VL (num-
`bering system of Kabat et al.) by site-directed mutagenesis using as template a uracil-containing
`single-stranded DNA of the scFv construct from the F+ origin present in the expression plas-
`mid and cotransfection with M13 helper phage. In addition to the cysteines, cloning sites, ATG
`(cid:4)
`(cid:4)
`translation-initiation codons, and stop codons are introduced at the 5
`end and 5
`end of the
`VH and VL genes, as shown by site-directed mutagenesis or PCR. The antibody variable genes
`are subcloned into an expression vector which contains the gene for a truncated form of Pseu-
`domonas exotoxin. This expression vector is controlled by the T7 promoter and upon induction
`of the T7 RNA polymerase, which is under the control of the lacUV5 promoter, in E. coli BL21
`␭DE3 by IPTG, large amounts of recombinant protein are produced. (B) composition of re-
`combinant immunotoxins. In PE-derived recombinant Fv immunotoxins the Fv region of the
`targeting antibody is fused to the N terminus of a truncated form of PE which contains the
`translocation domain (domain II) and enzymatically active ADP-ribosylation domain (domain
`III). The cell-binding domain of whole PE (domain I) is replaced by the Fv targeting moiety, thus
`preserving the relative position of the binding-domain function to the other functional domains
`of PE. In the dsFv immunotoxins there are two components. In one the VH or VL domains are
`fused to the amino terminus of the truncated PE, and the other variable domain is covalently
`linked by the engineered disulfide bond. DT-derived immunotoxins are fused to the carboxyl
`terminus due to the inverse arrangement of the functional modules of PE and DT.
`
`IMMUNOGEN 2152, pg. 11
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`04/03/2001
`
`04:18 PM Cancer Research-V.81
`
`PS039-03.tex
`
`PS039-03.xml
`
`LaTeX2e(2000/11/15)
`
`Textures 2.0
`
`104
`
`Yoram Reiter
`
`covalently connected by a peptide linker that fuses the C terminus of the VL
`or VH to the N terminus of the other domain (Fig. 1). These molecules are
`termed single-chain Fvs (scFvs) (62, 63), and many retain the specificity and
`affinity of the original antibody. The cloning, construction, and composition
`of recombinant Fv fragments of antibodies and of Fv-immunotoxins are de-
`scribed in Fig. 3. Many recombinant immunotoxins have been constructed
`using scFvs, in which molecules the scFv gene is fused to PE38 to generate a
`potent cytotoxic agent with targeted specificity (1–4, 64–70) (Figs. 1 and 3).
`Until recently, the construction of scFvs was the only general method avail-
`able to make stable Fvs. However, many scFvs are unstable or have reduced
`affinity for the antigen compared with the parent antibody or Fab fragment.
`This is because the linker interferes with binding or because the linker does
`not sufficiently stabilize the Fv structure, leading to aggregation and loss of
`◦
`activity. This is particularly true at physiological temperatures (37
`C). To
`overcome these problems, an alternative strategy has been developed that
`involves generating stable Fvs by connecting the VH and VL domains by an
`interchain disulfide bond engineered between structually conserved frame-
`work residues of the Fv; these molecules are termed disulfide-stabilized Fvs
`(de Fvs) (60, 71–73). The positions at which the cysteine residues were to
`be placed were identified by computer-based molecular modeling; as they
`are located in the framework of each VH and VL, this location can be used
`as a general method to stabilize almost all Fvs, without the need for any
`structural information. Many dsFvs have been constructed in recent years
`(mainly as dsFv immunotoxins, in which the dsFv is fused to PE38), and
`they show several advantages over scFvs (60, 74–76). In addition to their in-
`creased stability (owing to a decreased tendency to aggregate), they are often
`produced in higher yields than scFvs; in several cases, the binding affinity of
`the dsFv was significantly improved over that of the scFv.
`
`III. CONSTRUCTION AND PRODUCTION OF
`RECOMBINANT IMMUNOTOXINS
`
`In the recombinant immunotoxins derived from PE, the recombinant an-
`tibody fragments are fused to the amino terminus of the truncated deriva-
`tive of PE (with the cell-binding domain deleted), e.g., PE40 or PE38. This
`restores the original domain arrangement of PE, which consists of an
`N-terminal binding domain followed by the translocation domain and the
`C-terminal ADP-ribosylation domain (Fig. 1b). Only fusions of an antigen-
`binding domain (Fv) to the amino terminus of truncated PE are active;
`carboxyl-terminal fusions are not active because the bulky antigen-binding
`domain blocks translocation of the C-terminal fragment into the cytoplasm
`(1, 25).
`
`IMMUNOGEN 2152, pg. 12
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`04/03/2001
`
`04:18 PM Cancer Research-V.81
`
`PS039-03.tex
`
`PS039-03.xml
`
`LaTeX2e(2000/11/15)
`
`Textures 2.0
`
`Recombinant Immunotoxins
`
`105
`
`DT immunotoxins are fusions of mutated DT with antigen-binding regions
`of a recombinant antibody. However, in this case the antigen-binding domain
`must be fused to the C terminus of DT (77, 78). This corresponds to the
`inverse arrangement of the functional modules of PE and DT (see Fig. 3).
`DT immunotoxins are active only when the enzymatically active N-terminal
`domain is free to translocate into the cytosol.
`The expression vectors used for DT immunotoxins are very similar to
`those used with PE, with the exception that the DNA fragments encoding
`(cid:4)
`the binding moiety are ligated to the 3
`-end of the DT coding region.
`The cloning of the antibody variable regions is performed using cloning
`techniques that are now well established (59) (Fig. 3). The plasmid vec-
`tor for the expression of scFv immunotoxins or the components of dsFv
`immunotoxins is a high-copy-number plasmid derived from vectors made
`and described by Studier and Moffatt (79). These contain the T7 promoter,
`translation-initiation signals, and a transcription terminator, as well as an
`F+ phage-replication origin to generate single-stranded DNA to be used for
`site-directed mutagenesis.
`When these plasmids are transformed into E. coli BL21/DE3 (which con-
`tain the T7 RNA-polymerase gene under the control of the lac UV5 pro-
`moter), they generate large amounts of recombinant protein upon IPTG
`induction. The recombinant scFv immunotoxin or the components of the
`dsFv immunotoxin accumulate in insoluble intracellular inclusion bodies.
`[dsFv immunotoxins require two cultures, one expressing the VH and one
`expressing the VL; the toxin moiety (PE38)