Annu. Rev. Pharmacol. Toxicol. 1992. 32:579--621
`Copyright © by Annual Reviews Inc.
`All rights reserved
`
`CYTOTOXIC CONJUGATES
`
`CONTAINING TRANSLATIONAL
`
`INHIBITORY PROTEINS
`
`
`
`S. Ramakrishnan, D. Fryxell, D. Mohanraj, M. Olson, B-Y Li.
`
`
`
`
`
`Department of Pharmacology, 3-249 Millard Hall, University of Minnesota,
`
`
`
`
`Minneapolis. Minnesota 55455
`
`
`
`KEY WORDS: targeting toxins, immunotoxins, chemotherapy, carrier molecules, growth
`
`
`
`
`
`
`
`factor conjugates
`
`INTRODUCTION
`
`Cytotoxic drugs administered during conventional chemotherapy are taken up
`
`
`
`
`
`
`by sensitive nontarget organs besides the tumor cells and this often leads to
`
`
`unwanted side effects. These problems have initiated studies during the past
`
`
`
`decade to develop novel methods in which the therapeutic reagents could be
`
`
`
`to specific cell populations to widen the therapeutic window. The
`directed
`
`
`
`concept of targeting drugs was proposed nearly a hundred years ago by Paul
`
`
`
`
`Erlich, who envisioned the possibility of transporting toxic substances to
`
`
`
`tumor cells by carrier molecules (magic bullets). This novel approach became
`
`
`
`a reality with the advent of tumor selective monoclonal antibodies (Mabs)
`
`
`produced by somatic cell hybridization. Both cytotoxic drugs and toxin
`
`
`polypeptides have been chemically linked to Mabs. During the past decade
`
`
`
`
`various molecules have been produced and tested in a number of model
`
`
`
`systems (reviewed in 1-10). The range of targets includes (a) ex vivo
`
`
`
`treatment of bone marrow grafts to remove contaminating leukemic cells and
`
`
`T cells, (b) administration of conjugates to inhibit tumor growth
`alloreactive
`
`579
`
`Annu. Rev. Pharmacol. Toxicol. 1992.32:579-621. Downloaded from www.annualreviews.org
`
`by Reprints Desk, Inc. on 05/20/14. For personal use only.
`
`IMMUNOGEN 2176, pg. 1
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`
`580 RAMAKRISHNAN ET AL
`
`in restricted anatomical spaces such as the peritoneum, and (c) systemic
`
`
`
`
`
`
`application of conjugates to inhibit diffused tumors and solid tumors. Success
`
`
`of the last two applications depends largely on the pharmacological and
`
`
`
`
`toxicological properties of the chimeric molecules. This article focuses on the
`
`
`most recent developments in targeting toxin polypeptides.
`
`
`Immunotoxins are composed of a cytotoxic molecule linked to a carrier
`
`
`
`molecule by either a reducible or a nonreducible bond. The cytotoxic moiety
`
`
`
`is often a toxin polypeptide that catalytically inhibits a vital biosynthetic
`
`
`
`pathway. In general, the cytotoxic moiety used in targeting is a translational
`
`
`inhibitory protein but some investigators have also used enzymes such as
`
`
`
`phospholipases to destabilize the integrity of the tumor cell plasma mem­
`
`
`brane. Various carrier proteins have been investigated to deliver the toxin
`
`
`
`
`molecules, including tumor selective monoclonal antibodies, growth factors,
`and lymphokines.
`
`CHOICE OF TOXIN MOLECULES
`
`Three types of toxin polypeptides are used in the preparation of immunotoxins
`
`
`
`(IT)l: (a) bacterial toxins (b) plant toxins and (c) fungal toxins. All three
`
`
`
`
`
`groups of molecules catalytically inhibit protein synthesis in eukaryotes but
`
`
`
`each at a distinct step in translation. Most of the work on the bacterial toxins
`
`
`was carried out on two highly toxic molecules, namely diphtheria toxin (DT)
`
`studies on the structure/function and pseudomonas exotoxin A (PE). Detailed
`
`
`
`
`of these two molecules are primarily responsible for the advances made in
`
`
`preparing highly effective chimeric molecules against tumor cells. Based on
`
`
`their biochemical characteristics, plant toxins can be grouped as Type I,
`
`
`
`
`
`single chain polypeptides that enzymatically inhibit translation, and Type II
`
`
`(e. g. ricin and abrin), which are heterodimers. The A chains of Type II toxins
`
`are the true toxic moieties, whereas the B chains contain binding sites for
`
`
`carbohydrates through which the A chain gains access to the interior of the
`
`
`cell. Fungal toxins such as alpha sarcin are also single chain proteins but are
`
`
`
`
`functionally different from Type I polypeptides. Alpha sarcin, for example, is
`
`a phosphodiesterase while the Type I toxins are N-glycosidases.
`
`I Abbreviations used: CDR: complementarity-determining regions; DT: diphtheria toxin;
`
`
`
`
`
`
`
`
`EGF: epidermal growth factor; Fc: fragment crystallizable; GVHD: graft versus host disease;
`
`
`
`
`
`HAMA: human anti-murine antibody; IT: immunotoxins; KDEL: endoplasmic retention signal
`
`
`
`
`
`
`sequence; LDL: low density lipid; PAP: pokeweed antiviral protein; PE: pseudomonas exotoxin
`
`
`
`
`
`A; PEG: polyethylene glycol; RIP: ribosomal inhibitory proteins; SCA: single chain antibodies;
`
`
`
`SPDP: N-Succinimidyl 3-(2-pyridyl)dithiopropionic acid; TeS: trichosanthin; TFR: transferrin
`
`
`
`receptor; tPA: tissue plasminogen activator.
`
`Annu. Rev. Pharmacol. Toxicol. 1992.32:579-621. Downloaded from www.annualreviews.org
`
`by Reprints Desk, Inc. on 05/20/14. For personal use only.
`
`IMMUNOGEN 2176, pg. 2
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`
`CYTOTOXIC CONJUGATES 581
`
`RICIN
`
`
`
`Structural Features
`
`Ricin is a toxic lectin isolated from castor beans. Its precursor is a single
`
`
`
`
`
`
`polypeptide composed of a signal peptide and two chains linked by a 1 2
`
`
`
`
`
`amino acid joining peptide (J peptide) . It is proteolytic ally cleaved to form
`
`the mature ricin of 62 kd containing an A chain (267 amino acids) and a B
`chain (262 amino acids) linked by a disulfide bond ( 11 ) . The toxin is post­
`
`
`translationally glycosylated on both A and B chains with high mannose
`
`
`branched chains containing xylose and fucose. The A chain has two N-linked
`
`
`sugars at positions 1 0 and 236. The B chain has N-linked carbohydrates
`
`
`
`
`attached to residues 95 and 1 35. Heterogeneity in composition of sugars
`
`
`has been observed in all four sites (12) . In the mature ricin molecule the
`
`A chain is linked to the B chain by a reducible bond between Cys 259 of the
`
`A chain and Cys 4 of the B chain. Ricin has been crystallized, and the three­
`
`
`
`
`
`dimensional structure has been elucidated by Robertus and colleagues ( 1 3 ,
`of 2 . 5 A. Interaction
`1 4) at a resolution
`
`between the subunits i s mainly
`
`
`
`due to hydrophobic forces among various aromatic rings and aliphatic side
`chains.
`The carboxy terminal of the B chain seems to be involved in association
`
`
`
`with the A chain. Three domains can be distinguished in theA chain by X-ray
`
`
`
`crystallography ( 14). Domain I consists of amino acids 1-11 7 dominated by a
`
`
`six-stranded B-sheet and two helices. Domain II encompasses residues 1 18-
`
`2 1 0 and is mainly made of alpha helices. The C-terminal end constitutes
`
`
`
`Domain III, and the striking feature of this part of the molecule is a stretch of
`
`
`
`
`hydrophobic residues located between positions 247 and 257. This region of
`
`
`
`the molecule is suggested to play a key role in the translocation of the toxic
`
`
`subunit into the cytoplasm.
`The A chain of ricin is a specific N-glycosidase that catalytically hydro­
`
`
`
`
`
`
`lyzes the glycosidic bond of adenosine residue 4324 in the 28S rRNA (15).
`
`
`
`
`
`Depurination results in irreversible inactivation of ribosomes and inhibition of
`
`
`
`protein synthesis. Ricin is such a potent toxin that a single A chain molecule
`
`
`
`can inactivate 1 500 ribosomes per minute and once in the cytosol can kill the
`
`cell ( 1 6) . The Km for reticulocyte ribosomes is about 0.1 -0.2 f.tM and the
`
`
`Kcat about 1 ,500 min-i. The depurination mechanism of ricin is predicted to
`
`
`
`
`
`be similar to that of adenosine monophosphate nucleosidase involving the
`
`
`
`formation of an oxycarbonium ion. Based on site-directed mutagenesis and
`
`
`
`
`
`X-ray diffraction studies, important residues of the A chain either in direct
`
`
`
`
`contact with the substrate or those involved in the actual catalysis have been
`
`
`
`identified. The catalytic site is present in a cleft; Tyr 80 and Tyr 123 are
`
`
`
`
`
`present at the top of the cleft. Neither is bclieved to be important for catalysis
`
`
`
`but both may contribute to the tight binding of the substrate. Substitution of
`
`Annu. Rev. Pharmacol. Toxicol. 1992.32:579-621. Downloaded from www.annualreviews.org
`
`by Reprints Desk, Inc. on 05/20/14. For personal use only.
`
`IMMUNOGEN 2176, pg. 3
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`
`582 RAMAKRISHNAN ET AL
`
`other residues present around the cleft such as Asn 209 and Trp 211 does not
`
`
`
`
`
`significantly change the enzymatic activity (17, 1 8) .
`Tight binding o f the substrate (28S rRNA) is responsible for the partial
`
`
`
`
`
`
`
`
`breakage of the N-C bond (18). Two important changes involving either Glu
`
`
`
`177 or Arg 1 80 have consistently been shown to dramatically decrease the
`
`
`
`biological activity of the A chain. The positive charge on Arg 1 80 is supposed
`
`
`
`to help in the partial breaking of the bond by ion pairing to the phosphate
`
`
`backbone. The destabilized N-C bond is then vulnerable to nucleophilic attack
`
`
`
`
`by a water molecule. Interestingly, intact ribosomes are better substrates than
`
`
`
`
`
`the synthetic oligomers homologous to the conserved region covering the
`
`
`
`
`
`depurination site. It is therefore likely that the ribosomal proteins may con­
`
`
`
`tribute to the proper presentation of the sensitive site in a manner facilitating
`
`
`
`
`catalysis or that some important regions of ricin A chain may interact with
`
`
`
`
`
`ribosomal proteins in stabilizing the enzyme-substrate complex. This issue is
`
`
`
`
`
`currently being investigated by genetic deletion analyses in many labora­
`tories.
`The B chain of ricin has two functions: (a) binding reversibly to galactose
`
`
`
`
`
`presented by cell surface glycopeptides and glycolipids, and (b) facilitating
`
`
`
`
`uptake of whole ricin by endocytosis and transport through the Golgi appara­
`
`
`tus before translocation of the A chain into the cytosol (19). It is still
`
`
`contentious whether the two functions of the B chain are separable. Some
`
`
`
`
`studies indicate that toxin entry could be mediated by mechanisms in­
`
`
`
`
`
`dependent of galactose recognition, but other evidence supports the galactose
`
`
`
`binding activity on the B chain as an important step in the entry to the cytosol
`(20-24).
`
`Conjugates Prepared with Ricin and Ricin A Chain (RTA)
`
`
`
`
`
`Many immunotoxins using intact ricin and monoclonal antibodies or growth
`
`
`
`
`factors have been developed (25, 26). They are extremely cytotoxic to
`
`
`
`relevant target cells, but the specificities are suboptimal since B chains may
`
`
`
`
`
`bind to surfaces on nontarget cells. Another problem is the rapid clearance of
`
`
`ricin IT from the bloodstream (27). This is due at least partially to the
`
`
`
`clearance of native ricin by oligosaccharide-mediated uptake by the liver cells
`
`
`(28, 29). To circumvent this problem, immunotoxins have been made with
`
`
`
`chemically/enzymatically deglycosylated A chain (30, 3 1 ) or recombinant A
`
`
`
`
`
`chain. Chemical treatment destroys the mannose and other sugar residues on
`
`
`
`
`native A chain while recombinant A chain expressed in bacteria is free of B
`
`
`chain and completely lacks oligosaccharide side chains. These constructs
`
`
`
`
`have better pharmacological properties than conjugates made with native ricin
`A chain.
`Several groups have obtained ricin A chain (rRT A) by genetic engineering
`
`
`
`
`
`techniques. Shire et al (32) expressed a synthetic 842 bp RT A gene in
`
`Annu. Rev. Pharmacol. Toxicol. 1992.32:579-621. Downloaded from www.annualreviews.org
`
`by Reprints Desk, Inc. on 05/20/14. For personal use only.
`
`IMMUNOGEN 2176, pg. 4
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`
`CYTOTOXIC CONJUGATES 583
`
`coli at the level of 1 .5 mg/liter of culture. O'Hare et al (33)
`
`
`Escherichia
`
`
`cloned ricin A chain cDNA using vector pDs/3 and expressed the recombinant
`
`A chain with ten extra amino acids at the N-terminus in E. coli at the level of
`
`2-3 mglliter of culture. Piatak et al (34) obtained a correct 267 amino acid A
`
`
`
`
`
`
`chain from E. coli using several different host/vector systems at the level of
`
`
`1-2% to 6-8% of total cell proteins. Frankel et al (35) used an expression
`
`
`system to produce a fusion protein containing ricin A chain linked to the
`
`
`
`enzyme f3-galactosidase via a collagen fragment. Treatment of the fusion
`
`
`
`protein with collagenase released free ricin A chain that could be recovered
`
`
`
`from the enzyme by selectively absorbing the latter on an affinity column. In
`
`
`
`
`our laboratory, the coding region of the ricin A chain gene was cloned into the
`
`
`pET3b vector. The cloned fragment is flanked by the T7 promoter and
`
`
`
`terminator in this construction and expressed in E. coli, BL21(DE3). The
`
`
`
`production of purified A chain protein with 1 1 extra amino acids at the
`
`
`N-terminus was 80-90 mg/liter of culture or 24-26% of total cell proteins.
`
`
`
`due to The specificities of the IT made from nRTA or rRTA are excellent
`
`
`
`the absence of nonspecific binding of B chain (27, 36-39). However, the
`
`
`
`cytotoxicities of these conjugates were relatively less (40) than intact ricin IT
`
`
`
`due to the loss of translocation functions asssociated with the B chain (40-
`
`
`43). Some of the immunoconjugates prepared with ricin A chain were not
`
`cytotoxic, whereas the same antibody linked to ricin was effective against
`
`
`
`
`target cells in the presence of lactose. Translocation signals associated with
`
`
`the B chain of ricin have been effectively used in potentiating ricin A chain
`
`
`
`
`conjugates. The B chain can be either delivered directly to cell bound A chain
`
`
`conjugates or indirectly by a second antibody homing onto the same cell. In
`
`
`
`
`
`both approaches however, lactose has to be included to prevent nonspecific
`toxicity
`
`
`
`. Alternatively, intact ricin could be modified (blocked ricin) by
`
`
`
`
`
`chemical modifications (41-43) that retain the translocation properties with­
`
`
`
`
`
`out the nonspecific toxicity mediated by the sugar binding sites. The cell
`
`
`
`binding, enzymatic specificity, and mechanism of cytosol entry between
`
`bacterial toxins and plant toxins are very different.
`
`SINGLE CHAIN RIBOSOMAL INHIBITORY PROTEINS
`
`
`(TYPE I RIP)
`
`preparations This class of proteins is used as an alternative in immunotoxin
`
`
`
`
`
`
`
`and has many advantages in the treatment of several clinical diseases. Perhaps
`
`
`the best-studied single chain toxins are pokeweed antiviral proteins (PAP)
`
`isolated from the plant Phytolacca
`
`
`saporin 6, isolated from the
`americana,
`
`
`
`plant Saponaria and gelonin purified from Gelonum multiflorum.
`officinalis
`
`
`
`Similar proteins have been isolated from other plants (reviewed in 44) and are
`
`Annu. Rev. Pharmacol. Toxicol. 1992.32:579-621. Downloaded from www.annualreviews.org
`
`by Reprints Desk, Inc. on 05/20/14. For personal use only.
`
`IMMUNOGEN 2176, pg. 5
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`
`584 RAMAKRISHNAN ET AL
`
`generally basic proteins of about 30 kd. Most are nonglycosylated although
`
`
`
`recent evidence suggests the presence of simple sugars. The mechanism of
`
`
`action of Type I RIP seems to be similar to ricin in depurinating a single
`
`
`
`adenine residue at position 4324 of 28S rRNA.
`A major problem in using dual chain toxins (Type II) is that they also bind
`
`
`
`to saccharide residues present on normal cells. Before they can be used in IT
`
`
`
`preparations, these toxins must either be modified to prevent binding or must
`have their B chain removed. Single chain toxins, on the other hand, do not
`bind to any cell surface components.
`and evolutionarily, Type I
`Structurally
`
`
`
`RIP are related to Type II heterodimeric toxins. They usually do not contain
`
`free sulfhydryl groups (SH); these are often introduced to facilitate conjuga­
`
`
`
`tion to antibodies. SH groups can be introduced by modifying the epsilon
`
`
`amino group of lysine that, in certain instances, has led to a reduction in
`
`
`
`biological activity. Modification of lysine residues with reagents such as
`
`
`SPDP results in the loss of a positive charge on the molecule. Alternative
`
`
`
`
`
`thiolation procedures using 2-iminothiolane retain the positive charge and do
`
`
`
`not significantly reduce the biological activity of the toxins.
`
`
`
`
`
`Pokeweed Antiviral Protein (PAP)
`
`Three different forms (PAP-I, PAP-II, and PAP-S) of RIP have been isolated
`
`
`from the plant P. americana (45, 46). PAP-I and II were found in the leaves,
`
`
`
`
`
`and PAP-S was purified from the seeds. Pokeweed plants express the various
`
`
`
`forms of PAP at different seasons (46). Amino terminal sequences of these
`
`
`
`proteins are homologous with minor variations and there is limited immuno­
`
`
`
`logical crossreactivity between these proteins. Recent studies have shown that
`
`
`
`PAP inactivates ribosomes in a manner similar to ricin A chain (47). IT
`
`
`
`prepared with PAP have been highly specific in cytotoxicity to target cells
`
`
`
`(48-50). These conjugates proved useful in bone marrow purging and showed
`
`
`
`minimal toxicity to pluripotent progenitor cells.
`
`PAP-linked conjugates could be potentiated severalfold
`by ammonium
`and monensin (51), which indicates
`chloride
`
`that the conjugates were routed
`
`
`
`
`through acidic intracellular compartments. Since PAP does not have complex
`
`
`
`sugars, it is relatively more stable than native ricin A chain conjugates (52).
`
`
`
`Free PAP, however, is rapidly excreted from the circulation. PAP-IT directed
`
`
`to CD40 proved to be effective against cionogenic cells from acute
`
`
`
`lymphoblastic leukemia and non-Hodgkin's lymphoma in the absence of
`
`
`
`potentiators. PAP-IT specific toward CD7 were only effective against a T cell
`
`
`
`
`acute lymphoblastic leukemia lineage (53). A recent study showed that Type I
`
`
`
`
`RIP including PAP are useful in inhibiting phagocytic parasites when targeted
`
`via antibodies (54).
`
`Annu. Rev. Pharmacol. Toxicol. 1992.32:579-621. Downloaded from www.annualreviews.org
`
`by Reprints Desk, Inc. on 05/20/14. For personal use only.
`
`IMMUNOGEN 2176, pg. 6
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`
`CYTOTOXIC CONJUGATES 585
`
`Saporin
`Saporin is resistant to proteolysis, has a pI greater than 10, and has no
`
`
`
`
`
`
`
`
`carbohydrate chains. Yoshikawa et al (55) have described the separation of
`
`
`two distinct saporins from the seeds, N(l) and P(2). When conjugated to an
`
`
`
`
`
`anti-CD4 antibody, saporin kills CD4 cells effectively. Neither CD4 nor CD8
`
`
`
`
`
`cells internalize RTA-IT made with anti-T4 or anti-T8 specificities; the cells
`
`are not killed by either of these IT (56). Another study included three saporin
`
`IT that recognized CDS, CD2, or CD3. All three IT bound T cells and were
`
`
`
`
`
`
`effective in inhibiting protein synthesis in cell-free systems. The anti-CD2 IT,
`
`
`
`however, had weak toxicity to whole cells, whereas the other two were quite
`
`
`
`
`
`potent. The anti-CDS IT was marginally potentiated by amantadine (57).
`
`
`
`
`Saporin conjugated to the transferrin receptor (TFR) suppresses leukemic
`
`
`
`
`stem cell generation. Progenitors of the erythroid lineage were more sensitive
`
`than those of the myeloid lineage. There was no effect in vitro on normal or
`
`
`
`
`leukemic human hematopoietic progenitors. Since TFR expression is associ­
`
`
`
`ated with proliferation, it would be expected that the most susceptible cells
`
`
`
`
`would be those that are actively dividing. The antibody itself did not block
`
`
`
`
`TFR function (58). Similarly. Barbieri et al (59) found that saporin IT with
`
`
`antibodies directed toward plasma cell antigens bound and killed RAJI and
`
`
`U266 cells. Both had a high toxicity toward bone marrow, but only one
`
`
`
`
`
`inhibited differentiation of myeloid precursors. Bifunctional antibodies used
`
`
`
`
`to deliver saporin to specific tissues were tenfold more potent if the portion
`
`
`binding the toxin was obtained from a polyclonal mixture rather than a
`
`
`
`monoclonal source. This was shown for guinea pig leukemic cells (60).
`
`Trichosanthin
`Trichosanthin (TCS) is characterized by high content of Asn, Asp, GIu, Gin,
`
`
`
`
`
`
`
`and no Cys residues or carbohydrates. Its sequence is homologous to that of
`
`
`
`
`RTA, as is the tertiary structure. TCS loses over 90% of its activity when
`
`
`
`
`modified with SPDP; treatment with 2-iminothiolane resulted in the addition
`
`
`
`
`of 1.5 sulfhydryl groups per molecule on average. When conjugated to an
`
`
`
`
`anti-hepatoma antibody, the IT had a 50-fold higher toxicity than native TCS.
`
`
`
`Cytotoxicity of the IT was time dependent. Trichokirin, isolated from a
`
`
`
`
`related plant, is homologous but not completely identical to TCS (61).
`
`
`
`
`
`Trichokirin linked to an anti-Thy1.2 antibody was cytotoxic against T2 cells
`
`
`
`
`(62). Conjugation with dimethyl 3, 3'-dithiobis propionimidate (DTBP) re­
`
`
`sulted in only a small loss of activity; derivatization with SPDP and 2-
`
`
`
`
`
`iminothiolane reduced the biological activity significantly. The ability to
`
`
`
`
`inhibit cell-free translation was comparable to that of RTA and had the same
`
`
`
`
`enzymatic target. NH4CI had a negative dose-related effect on toxicity,
`
`
`
`
`
`characteristic of a requirement for passage through an acidic cellular compart-
`
`Annu. Rev. Pharmacol. Toxicol. 1992.32:579-621. Downloaded from www.annualreviews.org
`
`by Reprints Desk, Inc. on 05/20/14. For personal use only.
`
`IMMUNOGEN 2176, pg. 7
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`
`586 RAMAKRISHNAN ET AL
`
`ment. Monensin potentiated the trichikirin-IT fivefold. Clearance of the IT
`
`
`
`
`
`
`was slower than RTA-IT and had a biphasic pattern with half-lives of 0.5 and
`8 hours.
`A closely related toxin, J3-trichosanthin, can be effective against cells in the
`
`
`
`
`
`
`
`
`
`lymphocytic and monocytic lineages and can also block HIV replication in
`
`
`
`
`acutely and chronically infected cells. A rank order potency for inhibition
`of
`
`
`
`
`protein synthesis is j3-TCS, j3-momorcharin, luffaculin > a-momorcharin
`>
`
`
`TCS > momorcochin. These toxins are homologous in amino acid composi­
`
`
`
`tion, tertiary structure, have molecular weights between 28 and 32 kd, and
`
`have N-terminal Asp residues
`(63).
`
`Other Single Chain RIP
`Anti-Thy 1.1 antibodies conjugated to momordin or bryodin with SPDP were
`
`
`
`Thy 1.1 positive
`
`effective against
`
`
`target cells in the low nanomolar range.
`
`
`Bryodin IT were slightly more toxic. A two-to fivefold
`loss of activity was
`
`
`
`
`seen following SPDP treatment. This loss can possibly be prevented by
`
`
`
`
`treatment with 2-iminothiolane instead. The positive charge in vivo (high pI)
`(64).
`
`
`
`may shield the disulfide bond and confer higher stability in circulation
`
`Alpha Sarcin
`The toxin a-sarcin is also a basic nonglycosylated protein, but has a molecular
`
`
`
`
`
`weight of 1 7 kd. The smaller size of this peptide could prove useful in
`
`
`
`
`penetration of solid tumors and be easier to use in clinical settings. It has no
`
`
`
`sequence similarity to plant RIP but is homologous to restrictocin and
`
`
`
`
`mitogellin. a-sarcin is an example of single chain toxins that hydrolyze the
`
`
`
`phosphodiester bond in the 28S rRNA. An a-sarcin-IT had similar pharmaco­
`and was stable for 4 days at
`
`kinetics as other RIP with the same antibody
`
`
`
`
`37°C, much better than RTA-IT. The activity of the a-sarcin IT was 800-fold
`
`better than for a-sardn alone and approached the potency of RTA-IT. The
`
`
`
`a-sarcin IT had a longer biological in serum, with clearance follow­
`half-life
`
`
`ing the biphasic pattern common to RTA-IT (0.8 and 6 hr half-lives). IT with
`
`gelonin or momordin behaved similarly (65).
`
`
`The gene for a-sarcin has recently been sequenced
`(66) and shown to
`
`
`encode a stretch of 27 amino acids at the amino terminus that could potentially
`
`be cleaved. This could afford the cell protection
`
`against its own product.
`
`Restrictocin is homologous to a-sarcin and its gene has also been cloned and
`
`
`
`
`
`expressed, showing a single 52-base intron, possible regions for transcription
`
`
`
`signals, and polyadenylation sites flanking the open reading frame. An im­
`
`
`mature form of restrictocin may be produced that could undergo activation
`Both genes show homology to the U2 ribonuclease.
`during secretion.
`IT made
`have been used in IT
`
`with restrictocin were toxic in vitro (67). Other proteins
`(68-72).
`
`Annu. Rev. Pharmacol. Toxicol. 1992.32:579-621. Downloaded from www.annualreviews.org
`
`by Reprints Desk, Inc. on 05/20/14. For personal use only.
`
`IMMUNOGEN 2176, pg. 8
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`
`CYTOTOXIC CONJUGATES 587
`
`PSEUDOMONAS EXOTOXIN A
`
`Structural Features
`
`Pseudomonas exotoxin A (PE) is a single chain protein toxin produced by
`
`
`
`
`
`The molecular weight of PE is 66,000 daltons, and
`
`Pseudomonas aeruginosa.
`
`
`in its native form PE has four intrachain disulfide bridges, no titratable
`
`
`sulfhydryl groups, and is devoid of sugars and other posttranslational mod­
`
`
`
`ification (73). Structural analysis of PE has indicated that it is composed of
`
`
`
`
`three domains when interacting with mammalian cells. Domain Ia located at
`
`
`
`the N-terminal end of the molecule (aa 1-252) is responsible for cell binding.
`
`
`
`
`Domain II located in the middle of the molecule (aa 253-364) is suggested to
`
`
`be associated with translocation of the toxin moiety across cell membranes,
`
`
`
`
`and Domain III located at the C-terminal end (aa 405-613) possesses the ADP
`
`
`
`
`ribosylation activity. ADP ribosylation of elongation factor 2 irreversibly
`
`
`
`inhibits the synthesis of proteins and leads to cell death. The role of Domain
`
`Ib (aa 365-404) has not yet been determined (74). Domain II contains
`
`
`
`
`important information for protein secretion (75); the absence of Domain I
`
`
`
`shortens the half-life of OVB3-PE immunotoxin in mice (76). Once inside the
`
`
`cell, native PE undergoes proteolytic cleavage at a site rich in arginine
`
`
`
`
`
`residues. Primary sequence analysis indicates the presence of a consensus
`
`
`
`
`
`sequence homologous to the endoplasmic retention signal sequence (KDEL)
`
`
`at the C-terminal end. Amino acid substitutions at this region affect transloca­
`
`tion of PE to the cytoplasm (77).
`
`
`Pseudomonas exotoxin is a lethal toxin with an ID50 for a 20-gram mouse
`
`in the range of 0.1-0.2 JLg. Liver toxicity is a major cause of the death in
`
`
`
`
`
`mice (78-80). Intact PE has been coupled to a variety of tumor-binding mono­
`
`clonal antibodies (for example 260F9, 454Cl l, 280D ll, JLl, 106AIO,
`245E7, 520C9) to produce potent IT, with ID50 in the 0.01-0.1nM range
`
`
`
`after a 24 hr exposure of cells to IT (8). In addition to killing cancer cells
`
`
`
`
`in culture, PE-containing IT are also effective in inhibiting tumor growth
`
`in mice. However, PE conjugates cannot be given in large amounts to
`
`
`animals or patients because of residual binding of PE-IT to normal cells
`(74).
`To minimize this undesirable side effect, Chaudhary et al (75, 81) used
`
`
`
`
`
`
`
`genetic engineering to construct a plasmid containing codons for amino acids
`
`
`252-613 of PE and changed the glutamic acid at position 252 to lysine by
`
`
`
`
`site-directed mutagenesis. This construct was expressed in E. coli and pro­
`
`
`
`duced a protein of 40,000 daltons (Lys PE40) that was 100-fold less toxic to
`mice than was intact PE and lacked Domain Ia of PE (74). An extra lysine
`
`
`
`
`residue at the N-terminus of PE40 also facilitated derivatization with cross­
`
`
`
`linking reagents (75, 81). Lys57 (in Domain la) was shown to be involved in
`
`
`
`
`the cellular binding of PE (82) and, when mutated to arginine, decreased the
`
`Annu. Rev. Pharmacol. Toxicol. 1992.32:579-621. Downloaded from www.annualreviews.org
`
`by Reprints Desk, Inc. on 05/20/14. For personal use only.
`
`IMMUNOGEN 2176, pg. 9
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`
`588 RAMAKRISHNAN ET AL
`
`
`
`
`toxicity to murine cells in vitro and also was less toxic to mice. Use of PEarg57
`as the toxic moiety in IT also reduced liver damage.
`
`Cytotoxic Conjugates Made with Native PE and Genetically
`
`
`
`
`Engineered P E
`Pai et al (76) coupled a monoclonal antibody OVB3, which reacts with many
`
`
`
`
`
`
`human carcinomas, to native PE, lys PE40, and PEarg57, and compared the
`
`
`characteristics of the three IT. Each was cytotoxic to human tumor cell lines
`
`
`
`expressing the OVB3 antigen on their surfaces. All three IT caused complete
`
`
`regression of 50mm3 tumors with no toxic effects to the animals at therapeutic
`
`
`doses. The half-life in blood of OVB3-PE and OVB3-PEarg57 was 20 hr,
`
`
`whereas the half-life of OVB3-LysPE40 was 4 hr. But the single dose LD50
`
`value for OVB3-lys PE40 (200lLg) was much larger than for OVB3-PE or
`
`OVB3_PEarg57 (2.5 J-Lg and 3.5 J-Lg). This means OVB3-lys PE40 can be
`
`
`
`
`
`administered in higher doses to treat cancer more efficiently (76). In a recent
`
`
`
`
`study, PE linked to an antibody directed to human cervical carcinoma was
`
`
`
`
`evaluated for tumor growth inhibition in vivo. The F(ab)z fragment of the
`
`
`
`
`or nonreducible antibody was chemically conjugated to PE via a reducible
`
`
`
`
`thioether bond. Both conjugates were tumor suppressive when injected 4 to 14
`
`days after tumor transplantation (83).
`Immunotoxins made by chemically coupling protein toxins to specific
`
`
`
`
`
`
`
`
`
`antibodies are faced with low yields, formation of heterogenous products, and
`
`
`
`
`difficulties in large-scale production. To circumvent s()me of these problems,
`
`
`
`
`various chimeric toxins have been prepared by genetically fusing the coding
`
`
`
`
`
`sequences of toxin moieties to carrier proteins (84). Chaudhary et al (85)
`
`
`
`
`constructed a single chain antibody toxin fusion protein by ligating the DNA
`
`
`
`fragment of anti-Tac (Fv) with the gene of PE40. Anti-Tac is a monoclonal
`
`
`
`
`antibody to the P55 subunit of the human interleukin-2 receptor. A variable
`
`
`domain is the smallest binding unit of an antibody. The fusion protein
`
`anti-Tac (Fv)-PE40 was highly cytotoxic
`
`to two IL-2 receptor-bearing human
`
`
`cell lines but was not cytotoxic to receptor-negative cells (85). PE-40 was also
`
`
`to create a linked to the HIV -binding portion of the human CD4 molecule
`
`
`
`
`
`hybrid protein, CD4-PE40. CD4-PE40 displayed selective toxicity toward
`
`
`
`
`cells expressing the HIV envelope glycoprotein gp120 and could therefore be
`
`
`
`used as a therapeutic agent for the treatment of Acquired Immune Deficiency
`Syndrome (86).
`In addition, many receptor-specific chimeric toxins have been constructed
`
`
`
`
`
`by fusing cDNAs encoding cytokines (TGFa, IL2, IL4, IL6) to the gene of
`
`
`
`
`PE40. Each chimeric protein was specifically cytotoxic to the appropriate
`
`
`
`
`receptor-bearing cells (87-90). Interestingly, the position of the ligand moi­
`
`
`
`ety, i.e. cell recognition element, in the chimeric molecule might be impor­
`
`
`tant. When the N-terminus of PE40 was fused to the C-terminus of TGF-a
`
`Annu. Rev. Pharmacol. Toxicol. 1992.32:579-621. Downloaded from www.annualreviews.org
`
`by Reprints Desk, Inc. on 05/20/14. For personal use only.
`
`IMMUNOGEN 2176, pg. 10
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`
`CYTOTOXIC CONJUGATES
`589
`
`(TGF-a-PE40), the conjugate was 30-fold more potent than a fusion between
`
`
`
`
`
`the C-tenninus of PE40 to the N-tenninus of TGF-a (PE40-TGF-(a). The
`
`
`fonner construct recognized and bound with about the same affinity as EGF.
`
`
`
`
`In native PE the cell recognition domain (Ia) is attached directly to the
`
`
`
`translocation domain. The close association of the receptor recognition and
`
`
`
`translocation domains may possibly facilitate movement of toxin into the
`
`cytosol (88).
`
`DIPHTHERIA TOXIN
`
`Structural Features
`
`Diphtheria toxin (DT) is a single polypeptide exotoxin secreted by
`
`
`
`
`
`
`
`and is lysogenic for bacteriophage Betatox+ that
`
`Corynebacterium diphtheriae
`
`
`
`
`carries the gene for DT (91, 92). Intact DT has two disulfide bridges. After
`
`
`
`
`mild trypsin digestion ("nicking") and reduction of the disulfide bonds, DT
`
`
`can be separated into an N-terminal 2 1 . 1 7-kd fragment A of 1 93 amino acids
`
`( 1-193) and a C-terminal 37.20-kd fragment B of 342 amino acids from
`
`position 1 94 to 535 (8, 93, 94) .
`The A chain enzymatically ADP ribosylates a unique amino acid (diptha­
`
`
`
`
`
`mide) of elongation factor-2 (EF-2), resulting in irreversible inactivation and
`
`tennination of cellular protein synthesis (95). The B chain has two functions:
`
`
`
`
`
`binding cell surface receptors and translocation of the A chain to the cytosol.
`
`
`
`
`B inding activity is located in the 8-kd C-terminal fragment of the B chain, and
`
`
`the last 50 amino acids are confirmed to be required to form the receptor
`
`
`
`
`binding domain (94, 96). The translocation activity is in the N-terminal three
`
`fourths of fragment B . From amino acids 270 to 370 and amino acids 42