`
`Published OnlineFirst November 16, 2011; DOI: 10.1158/1535-7163.MCT-11-0519 Published OnlineFirst November 16, 2011; DOI: 10.1158/1535-7163.MCT-11-0519
`
`Single-Chain Antibody-Based Immunotoxins Targeting Her2/neu:
`Design Optimization and Impact of Affinity on Antitumor Efficacy
`and Off-Target Toxicity
` 
`Yu Cao, James D. Marks, Qian Huang, et al.
`Mol Cancer Ther  
`
`2012;11:143-153. Published OnlineFirst November 16, 2011.
`
` 
` 
` 
` 
`
` 
` 
`
`Updated version
` 
`Supplementary
`Material
` 
`
`Access the most recent version of this article at:
`
` 10.1158/1535-7163.MCT-11-0519
`doi:
` 
`Access the most recent supplemental material at:
`http://mct.aacrjournals.org/content/suppl/2011/11/15/1535-7163.MCT-11-0519.DC1.html
` 
`
`
`
`Cited Articles
` 
`Citing articles
` 
`
`This article cites by 45 articles, 18 of which you can access for free at:
`
`http://mct.aacrjournals.org/content/11/1/143.full.html#ref-list-1
` 
`This article has been cited by 3 HighWire-hosted articles. Access the articles at:
`
` http://mct.aacrjournals.org/content/11/1/143.full.html#related-urls
` 
`
`E-mail alerts
` 
`Reprints and
`Subscriptions
` 
`Permissions
` 
`
`
`
` related to this article or journal.Sign up to receive free email-alerts
`
`To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at
`
`.pubs@aacr.org
` 
`To request permission to re-use all or part of this article, contact the AACR Publications Department at
`.
`permissions@aacr.org
` 
`
`Downloaded from
`mct.aacrjournals.org
`mct.aacrjournals.org Downloaded from
`
`
`on October 16, 2014. © 2012 American Association for Cancer Research.
`on October 16, 2014. © 2012 American Association for Cancer Research.
`
`IMMUNOGEN 2112, pg. 1
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`Published OnlineFirst November 16, 2011; DOI: 10.1158/1535-7163.MCT-11-0519
`
`Preclinical Development
`
`Molecular
`Cancer
`Therapeutics
`
`Single-Chain Antibody-Based Immunotoxins Targeting
`Her2/neu: Design Optimization and Impact of Affinity on
`Antitumor Efficacy and Off-Target Toxicity
`
`Yu Cao3, James D. Marks1, Qian Huang2, Stephen I. Rudnick4, Chiyi Xiong2, Walter N. Hittelman3,
`Xiaoxia Wen2, John W. Marks3, Lawrence H. Cheung3, Kim Boland4, Chun Li2, Gregory P. Adams4, and
`Michael G. Rosenblum3
`
`Abstract
`
`Recombinant immunotoxins, consisting of single-chain variable fragments (scFv) genetically fused to
`polypeptide toxins, represent potentially effective candidates for cancer therapeutics. We evaluated the
`affinity of various anti-Her2/neu scFv fused to recombinant gelonin (rGel) and its effect on antitumor efficacy
`and off-target toxicity. A series of rGel-based immunotoxins were created from the human anti-Her2/neu scFv
`8 to
`C6.5 and various affinity mutants (designated ML3-9, MH3-B1, and B1D3) with affinities ranging from 10
`11 mol/L. Against Her2/neu-overexpressing tumor cells, immunotoxins with increasing affinity displayed
`10
`improved internalization and enhanced autophagic cytotoxicity. Targeting indices were highest for the highest
`affinity B1D3/rGel construct. However, the addition of free Her2/neu extracellular domain (ECD) signifi-
`cantly reduced the cytotoxicity of B1D3/rGel because of immune complex formation. In contrast, ECD addition
`had little impact on the lower affinity constructs in vitro. In vivo studies against established BT474 M1 xenografts
`showed growth suppression by all immunotoxins. Surprisingly, therapy with the B1D3-rGel induced
`significant liver toxicity because of immune complex formation with shed Her2/neu antigen in circulation.
`The MH3-B1/rGel construct with intermediate affinity showed effective tumor growth inhibition without
`inducing hepatotoxicity or complex formation. These findings show that while high-affinity constructs can be
`potent antitumor agents, they may also be associated with mistargeting through the facile formation of
`complexes with soluble antigen leading to significant off-target toxicity. Constructs composed of intermediate-
`affinity antibodies are also potent agents that are more resistant to immune complex formation. Therefore,
`affinity is an exceptionally important consideration when evaluating the design and efficacy of targeted
`therapeutics. Mol Cancer Ther; 11(1); 143–53. Ó2011 AACR.
`
`Introduction
`
`Immunotherapeutic approaches using antibodies have
`been widely explored against a variety of tumors, but an
`effective treatment of solid tumors remains a potential
`problem because therapeutic antibodies must diffuse into
`tumors through a disordered vasculature and against a
`hydrostatic pressure gradient (1, 2). Because low–molec-
`
`Authors' Affiliations: 1Department of Anesthesia, University of California,
`San Francisco, California; 2Department of Experimental Diagnostic Imag-
`ing, 3Immunopharmacology and Targeted Therapy Laboratory, Depart-
`ment of Experimental Therapeutics, University of Texas M.D. Anderson
`Cancer Center, Houston, Texas; and 4Developmental Therapeutics Pro-
`gram, Fox Chase Cancer Center, Philadelphia, Pennsylvania
`
`Note: Supplementary data for this article are available at Molecular Cancer
`Therapeutics Online (http://mct.aacrjournals.org/).
`
`Corresponding Author: Michael G. Rosenblum, Department of Experi-
`mental Therapeutics, University of Texas MD Anderson Cancer Center,
`Houston, TX 77054. Phone: 713-792-3554; Fax: 713-794-4261; E-mail:
`mrosenbl@mdanderson.org
`
`doi: 10.1158/1535-7163.MCT-11-0519
`Ó2011 American Association for Cancer Research.
`
`ular weight antibody fragments have been shown to have
`better tumor diffusion properties (3), single-chain variable
`fragments (scFv) were favored to deliver protein-based
`toxins to cancer cells (4, 5).
`A variety of scFv-based immunotoxins have been engi-
`neered that are suitable for diverse therapeutic applica-
`tions. An anti-CD174 scFv designated SGN-10 fused with
`Pseudomonas exotoxin (PE) was developed for optimal
`tumor penetration but clinical studies were limited by renal
`toxicity and gastritis (6, 7). Chaudhary and colleagues (8)
`and Powell and colleagues (9) generated LMB-2, anti-CD25
`scFv-PE immunotoxin and described promising preclinical
`efficacy on malignant cells from patients with adult T-cell
`leukemia. However, common toxicities included transam-
`inase evaluation. The therapeutic window for this class of
`constructs may be optimized by various design changes to
`lower the efficacious dose, improve specificity by reducing
`off-target effects, thereby allowing an increase in the max-
`imal tolerated dose (10–12).
`Tumor–antigen affinity and specificity of scFvs are
`important variables that may impact off-target tissue
`distribution and toxicity in vivo. These attributes have led
`
`www.aacrjournals.org
`
`143
`
`
`
`mct.aacrjournals.org Downloaded from
`
`on October 16, 2014. © 2012 American Association for Cancer Research.
`
`IMMUNOGEN 2112, pg. 2
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`Published OnlineFirst November 16, 2011; DOI: 10.1158/1535-7163.MCT-11-0519
`
`Cao et al.
`
`to the commonly held concept that scFv must have high
`affinity to be therapeutically relevant. However, studies
`by Adams and colleagues (13) and Rudnick and Adams
`(14) have suggested that high-affinity scFv may be sub-
`optimal vehicles and that lower affinity scFv appear to
`diffuse more uniformly throughout the tumor interior. In
`addition, because the presence of shed tumor antigen has
`the potential to misdirect the targeted constructs through
`immune complex formation (10, 15), higher affinity scFv
`could potentially be at risk compared with lower affinity
`constructs.
`Although previous studies primarily focused on the
`in vivo behavior of scFv, few companion studies have
`been conducted to determine whether scFv-based
`immunotoxins display the same behavior with regard
`to the relationship between affinity, tumor penetration,
`tumor residence, and efficacy. Our present knowledge
`of
`the affinity/function relationship of scFv-based
`immunotoxins is insufficient to afford accurate predic-
`tions as to whether a given scFv is appropriate for toxin
`delivery. A comprehensive head-to-head comparison
`of recombinant immunotoxins with different affinities
`targeting the same epitope on an antigen would be
`useful to guide the developmental strategy for future
`immunotoxins.
`We previously reported the construction and charac-
`terization of anti-Her2/neu immunotoxins constructed
`by fusing scFv C6.5 with the recombinant gelonin (rGel).
`These constructs showed highly efficient activity against
`Her2/neu-positive tumor cells (16). In the current study,
`we generated a series of rGel-containing fusion constructs
`composed of C6.5 and its mutants with varying affinities
`to Her2/neu and examined the impact of affinity on
`in vitro cytotoxicity, pharmacodynamics, and antitumor
`efficacy. In addition, we investigated the effect of antibody
`affinity on behavior in the presence of soluble antigen,
`formation of immune complexes, and the coincident
`development of off-target toxicity.
`
`Materials and Methods
`
`Plasmid construction
`The gene encoding human anti-Her2/neu scFv (C6.5
`and its affinity mutants, ML3-9, MH3-B1, and B1D3 cre-
`ated by site-directed amino acid substitutions in the
`CDR3s; ref. 17) were supplied by Dr. James D Marks
`(University of California, San Francisco, San Francisco,
`CA; Fig. 1A). Illustrations of the immunotoxin constructs
`are shown in Fig. 1B. Recombinant immunotoxins con-
`taining each scFv and rGel were constructed by over-
`lapping PCR and were designated C6.5/rGel, ML3-9/
`rGel, MH3-B1/rGel, and B1D3/rGel, respectively.
`
`Protein expression and purification
`The immunotoxins were expressed in Escherichia coli
`strain AD494 (DE3) pLysS and purified by immobilized
`metal affinity chromatography (IMAC) essentially as pre-
`viously described (16).
`
`Binding affinity of immunotoxins
`The binding affinity and specificity of the immunotox-
`ins were tested by ELISA on Her2/neu-positive (SKBR3,
`BT474 M1) and -negative (MCF7) cells. Rabbit anti-rGel
`antibody and horseradish peroxidase–conjugated goat
`anti-rabbit IgG were used as a tracer in this assay as
`described previously (16).
`
`Internalization and competitive inhibition analysis
`Immunofluorescence-based internalization studies
`were conducted on Her2/neu-positive (SKBR3, BT474
`M1) and -negative (MCF7) cells. Immunofluorescence
`staining and competitive inhibition were analyzed as
`described in Supplementary Methods.
`
`Cytotoxicity of scFv/rGel and competitive
`cytotoxicity assays
`The cytotoxicity of immunotoxins on log-phase Her2/
`neu-positive and -negative cell lines were tested with the
`crystal violet staining method, and competitive assays
`were conducted as described in Supplementary Methods
`(18).
`
`Western blot analysis of apoptosis and autophagy
`The detection of apoptosis and autophagy on BT474 M1
`cells treated with immunotoxins was analyzed as
`described in Supplementary Methods.
`
`In vivo efficacy studies
`BALB/c nude mice bearing subcutaneous BT474 M1
`tumors were established and treated (intravenously, tail
`vein) with immunotoxins, as described in Supplementary
`Methods.
`
`Tissue distribution study
`The MH3-B1/rGel and B1D3/rGel was labeled with
`IRDye800CW according to the manufacturer’s protocol.
`The tissue distribution assays and the imaging analysis
`are further described in Supplementary Methods.
`
`Coimmunoprecipitation assay
`Liver samples from mice after treatment with MH3-B1/
`rGel or B1D3/rGel were collected. Samples were exam-
`ined for the presence of antigen:immunotoxin complexes
`as described in Supplementary Methods.
`
`In situ immunofluorescent detection
`Samples of liver tissues from mice were further pre-
`pared for immunofluorescence staining tracing Her2/neu
`antigen and scFv/rGel immunotoxins as described in
`Supplementary Methods.
`
`Liver toxicity study
`Hepatotoxicity was investigated by measuring activi-
`ties of alanine transaminase (ALT), aspartate transami-
`nase (AST), and lactate dehydrogenase (LDH) in collected
`serum from treated mice according to an assay kit (Roche).
`The histologic examination for hepatotoxicity was
`
`144
`
`Mol Cancer Ther; 11(1) January 2012
`
`Molecular Cancer Therapeutics
`
`Downloaded from
`
`mct.aacrjournals.org
`
`on October 16, 2014. © 2012 American Association for Cancer Research.
`
`IMMUNOGEN 2112, pg. 3
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`Published OnlineFirst November 16, 2011; DOI: 10.1158/1535-7163.MCT-11-0519
`
`Affinity Impact of Immunotoxins on Efficacy and Toxicity
`
`Figure 1. Construction and preparation of scFv/rGel immunotoxins. A, amino acid mutations and affinity parameters of the C6.5 and its mutants, ML3-9, MH3-
`B1, and B1D3. The listed amino acids for each scFv indicate mutations to the sequence and the substituting amino acids. Dashes indicate no changes from
`the original sequence. B, schematic diagram of immunotoxin constructs containing scFv (C6.5, ML3-9, MH3-B1, or B1D3) and rGel. C, SDS-PAGE analysis
`of purified immunotoxins.
`
`assessed by hematoxylin and eosin staining. Further
`details are presented in Supplementary Methods.
`
`Statistical analysis
`Statistical analyses were conducted with SPSS version
`17.0.2 software (SPSS Inc.). Data were presented as mean
` SD, and significance was determined using a 2-sided
`Student t test, unless otherwise noted. A value of P < 0.05
`was considered statistically significant.
`
`Results
`
`Preparation of scFv/rGel fusion constructs
`The scFv/rGel constructs were created from human
`anti-Her2/neu scFv C6.5 and various affinity mutants
`
`(designated ML3-9, MH3-B1, and B1D3, in increasing
`affinity order). The affinities of the scFv ranged from
`
`8 to 1011 mol/L (Fig. 1A; refs. 17, 19). The immuno-
`10
`toxin genes were cloned into vector pET-32a(þ) separately
`(Fig. 1B). Sequenced DNA clones were subsequently
`transformed into E. coli AD494 (DE3) pLysS for protein
`expression. As shown in Fig. 1C, after purification, all the
`immunotoxins migrated on SDS-PAGE at the expected
`molecular weight of 55 kDa under both reducing and
`nonreducing conditions.
`
`Binding and cellular internalization of the fusion
`constructs
`To ensure that immunotoxins retained antigen-binding
`ability, the fusion proteins were compared in an ELISA-
`
`www.aacrjournals.org
`
`Mol Cancer Ther; 11(1) January 2012
`
`145
`
`Downloaded from
`
`mct.aacrjournals.org
`
`on October 16, 2014. © 2012 American Association for Cancer Research.
`
`IMMUNOGEN 2112, pg. 4
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`Published OnlineFirst November 16, 2011; DOI: 10.1158/1535-7163.MCT-11-0519
`
`Cao et al.
`
`based–binding assay using Her2/neu-positive (SKBR3,
`BT474 M1) and -negative (MCF7) cells. All the scFv/rGel
`constructs showed specific and significant ELISA binding
`to Her2/neu-positive cells with negligible binding to
`negative cells (Fig. 2A). The equilibrium dissociation
`constants (K
`d) were calculated (GraphPad Prism), and
`the affinities of immunotoxins for BT474 M1 cells were
`found to be 53.13 nmol/L (C6.5/rGel), 1.45 nmol/L (ML3-
`9/rGel), 0.18 nmol/L (MH3-B1/rGel), and 27 pmol/L
`(B1D3/rGel). The correlation between the K
`d values of
`the scFvs and fusion constructs was found to be significant
`with a correlation coefficient of 0.939 (P < 0.01), indicating
`that introduction of the rGel component did not affect the
`binding affinity of the scFv.
`We next examined whether the various affinity scFv/
`rGel fusions could specifically internalize into target cells.
`Immunofluorescence staining was conducted on Her2/
`neu-positive and -negative cells. As quantified by relative
`fluorescence (Fig. 2B), the internalization efficiency exhib-
`ited a moderate increase with increasing binding affinity
`in Her2/neu-positive cells. For BT474 M1 cells, the rela-
`tive fluorescence intensities were 56.30 (C6.5/rGel), 73.69
`(ML3-9/rGel), 86.29 (MH3-B1/rGel), and 90.41 (B1D3/
`rGel). There was a good correlation of between increases
`in apparent affinity and internalization efficiency (r2 ¼
`0.8289; P < 0.01) indicating that efficient binding to the cell
`surface appears to be primarily responsible for rapid
`internalization after cell exposure.
`
`In vitro cytotoxicity of scFv/rGel fusion constructs
`All the scFv/rGel constructs and rGel were tested
`against a number of different tumor cell lines (Table 1).
`As expected, there appeared to be a good correlation (r2 ¼
`0.7812; P < 0.01) between apparent affinity and IC50 values.
`Targeting indices were found to be highest for the highest
`affinity B1D3/rGel construct. This study showed that for
`the scFv/rGel immunotoxins, binding affinity appears to
`mediate internalization efficiency and this appeared to
`directly impact the overall cytotoxic effects observed.
`Furthermore, against Her2/neu-negative cells, there was
`little or no specific cytotoxicity of the constructs compared
`with rGel itself.
`
`Effects of various fusion constructs on cytotoxic
`mechanisms
`The cytotoxic effects mediated by scFv/rGel immuno-
`toxins were analyzed in BT474 M1 cells. As shown in Fig.
`3A, scFv/rGel fusions did not activate caspase-dependent
`apoptosis in target cells, showing no cleavage of the
`caspase substrate PARP. We next assessed LDH release
`and found that exposure of BT474 M1 cells to immuno-
`toxins did not induce necrotic cell death (data not shown).
`Then, we examined whether the cytotoxic effects of
`these immunotoxins activate autophagic signaling. As
`shown in Fig. 3B, the ratio of LC3-II formation to the
`b-actin control was shown to be increased after treatment
`with the fusion constructs, showing that autophagic flux
`
`Figure 2. Characterization and comparison of scFv/rGel immunotoxins. A, evaluation binding activity of the scFv/rGel to Her2/neu-positive (SKBR3, BT474 M1)
`and -negative (MCF7) cells by whole-cell ELISA. B, quantification of internalization rate of the immunotoxins on Her2/neu-positive and -negative cells. Cells
`were subjected to immunofluorescent staining with anti-rGel antibody (fluorescein isothiocyanate–conjugated secondary antibody). The bar graphs were
`calculated from relative fluorescence estimation, and the values are expressed as mean  SD (n > 50).
`
`146
`
`Mol Cancer Ther; 11(1) January 2012
`
`Molecular Cancer Therapeutics
`
`Downloaded from
`
`mct.aacrjournals.org
`
`on October 16, 2014. © 2012 American Association for Cancer Research.
`
`IMMUNOGEN 2112, pg. 5
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`Published OnlineFirst November 16, 2011; DOI: 10.1158/1535-7163.MCT-11-0519
`
`Affinity Impact of Immunotoxins on Efficacy and Toxicity
`
`Table 1. Comparative IC50 values of fusion constructs against various types of tumor cell lines
`IC50 (nmol/L) with T.I.a
`
`Cell line
`
`Type
`
`Her2/neu level
`
`SKBR3
`BT474 M1
`NCI-N87
`Calu3
`MDA-MB-231
`MCF7
`A375m
`Me180
`
`Breast
`Breast
`Gastric
`Lung
`Breast
`Breast
`Melanoma
`Cervical
`
`þþþþ
`þþþþ
`þþþþ
`þþþþ
`þ
`þ
`þ
`þ
`
`C6.5/rGel
`
`ML3-9/rGel
`
`MH3-B1/rGel
`
`B1D3/rGel
`
`rGel
`
`6.4 (165)
`18.9 (12)
`30.1 (32)
`24.3 (22)
`145.8 (2)
`246.9 (1)
`61.4 (3)
`160.8 (1)
`
`5.0 (211)
`10.9 (20)
`20.4 (47)
`18.1 (29)
`149.9 (2)
`246.7 (1)
`126.9 (2)
`185.4 (1)
`
`2.7 (387)
`3.9 (56)
`9.1 (106)
`13.0 (41)
`155 (2)
`260.6 (1)
`153.9 (1)
`194.6 (1)
`
`1.9 (567)
`1.2 (177)
`4.9 (199)
`10.0 (53)
`204.8 (1)
`266.9 (1)
`173.9 (1)
`213.1 (1)
`
`1,061.0
`219.8
`965.6
`531.1
`297.3
`247.4
`207.3
`222.5
`
`Abbreviation: T.I., targeting index.
`aTargeting index represents IC50 of rGel/IC50 of immunotoxin.
`
`was induced by rGel-based immunotoxins in BT474 M1
`cells. In addition, autophagic induction by fusion con-
`structs was further validated by the observed selective
`release of cellular HMGB1 (Fig. 3C; ref. 20). These data
`
`indicated that the observed cytotoxic effects of scFv/rGel
`fusions in BT474 M1 cells appeared to be mediated not
`through an apoptotic or necrotic mechanisms but by the
`efficient induction of autophagic cell death.
`
`Figure 3. Cell-killing mechanism
`analysis of the immunotoxins on
`BT474 M1 cells. A, Western blot
`analysis of PARP cleavage after 24
`and 48 hours of scFv/rGel fusion
`treatment. B, analysis of LC3 after
`treatment with the scFv/rGel fusions
`for 24 and 48 hours. C, analysis of cell
`extract and medium for HMGB1
`protein after immunotoxin treatment
`for 48 hours.
`
`www.aacrjournals.org
`
`Mol Cancer Ther; 11(1) January 2012
`
`147
`
`Downloaded from
`
`
`
`on October 16, 2014. © 2012 American Association for Cancer Research. mct.aacrjournals.org
`
`IMMUNOGEN 2112, pg. 6
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`Published OnlineFirst November 16, 2011; DOI: 10.1158/1535-7163.MCT-11-0519
`
`Cao et al.
`
`Influence of soluble Her2/neu extracellular domain
`on immunotoxin activity
`Shedding of target antigen from the surface of tumor
`cells into circulation may present obstacles for antibodies
`to effectively target tumor cells in vivo (21). To investigate
`the impact of soluble antigen on scFv/rGel immunotox-
`ins, we evaluated the internalization of the immunotoxins
`in the presence of free Her2/neu extracellular domain
`(ECD) in BT474 M1 cells (Fig. 4A and Supplementary Fig.
`S1). The addition of ECD reduced the internalization for
`all the constructs. The highest affinity B1D3/rGel con-
`struct showed a significant reduction (P < 0.01), whereas
`the lower affinity fusions exhibited the lesser impact of
`ECD on internalization.
`We next applied coimmunoprecipitation to examine
`whether the decreased internalization observed was
`because of the immune complexes of the immunotoxins
`with ECD. As shown in Fig. 4B, the highest affinity
`construct (B1D3/rGel) and the lowest affinity construct
`(C6.5/rGel) formed the highest and lowest amount of
`immune complex with ECD, with intermediate-affinity
`molecules (ML3-9/rGel and MH3-B1/rGel) generating
`
`intermediate levels of immune complexes. Therefore, the
`significant reduction in cell internalization observed with
`B1D3/rGel fusion was the result of immune complexes
`formation with soluble ECD, further preventing binding
`of the immunotoxins via cell-associated antigen.
`A competitive cytotoxicity assay was conducted on
`SKBR3 and BT474 M1 cells by adding 20 nmol/L ECD to
`various concentrations of each fusion construct (Fig. 4C).
`All the fusion constructs showed an increase in IC50 value
`in the presence of ECD. Constructs with low and medium
`affinity showed the least impact of ECD on cytotoxic
`effects whereas B1D3/rGel showed the greatest influence.
`Furthermore, the addition of various concentration of
`ECD to a fixed (20 nmol/L) dose of immunotoxins showed
`similar effects (Supplementary Fig. S2). B1D3/rGel was
`impacted to the greatest extent in the presence of ECD,
`whereas the constructs with low or medium affinity
`showed less impact on cytotoxicity.
`
`Her2/neu antigen shed from tumor cells
`Because the Her2/neu antigen can be shed from target
`cells and may impact the cytotoxic effects observed with
`
`Figure 4. Competitive analysis of the scFv/rGel in the presence of free Her2/neu ECD. A, quantification of competitive internalization rate of the fusion proteins
`on BT474 M1 cells. Cells were treated with the mixture of 20 nmol/L immunotoxins and different concentration of ECD. Values are expressed as mean  SD
`(n > 50). B, coimmunoprecipitation of scFv/rGel and ECD complex. The mixture supernatants were subjected to Her2/neu immunoprecipitation (IP), followed
`by immunoblotting (IB) for rGel. C, competitive cytotoxicity of the immunotoxins in the presence of 20 nmol/L ECD. Values are presented as IC50.
`
`148
`
`Mol Cancer Ther; 11(1) January 2012
`
`Molecular Cancer Therapeutics
`
`
`
`mct.aacrjournals.org Downloaded from
`
`on October 16, 2014. © 2012 American Association for Cancer Research.
`
`IMMUNOGEN 2112, pg. 7
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`Published OnlineFirst November 16, 2011; DOI: 10.1158/1535-7163.MCT-11-0519
`
`Affinity Impact of Immunotoxins on Efficacy and Toxicity
`
`scFv/rGel immunotoxins, we measured the endogenous
`Her2/neu antigen levels in cell media and serum from
`mice bearing tumor xenografts by quantitative ELISA.
`The medium of Her2/neu-positive cells (SKBR3, BT474
`M1, NCI-N87, and Calu3) were collected daily for 7 days.
`As shown in Supplementary Fig. S3A, the Her2/neu
`antigen was present in the medium of all Her2/neu-
`positive cells, and the concentrations increased relative
`to cell number. Levels of antigen in culture media were
`0.4 nmol/L from all lines tested during a 72-hour cyto-
`toxicity assay and appeared to be well below levels that
`would impact immunotoxin efficacy.
`Measurements of shed Her2/neu antigen in the blood
`of mice bearing BT474 M1 tumor showed that Her2/neu
`levels increase in parallel with tumor size (Supplementary
`Fig. S3B). Levels of Her2/neu in serum increased from 2 to
`12 nmol/L for mice with 200 mm3 tumors up to 1,800 mm3,
`respectively. The correlation between shed Her2/neu
`levels and tumor volume was found to be significant with
`
`a correlation coefficient of 0.797 (P < 0.01). At sufficiently
`high Her2/neu levels, the efficacy of high-affinity–tar-
`geted therapeutics could be impacted.
`
`Antitumor activity of scFv/rGel fusions in xenograft
`models
`We next evaluated the ability of various scFv/rGel
`immunotoxins to inhibit the growth of established BT474
`M1 tumor xenografts in BALB/c nude mice after systemic
`administration. BT474 M1 cells were implanted into mice
`and tumors were allowed to grow to 200 mm3 in volume.
`Mice were then treated with each fusion protein and rGel
`as control at a total dose of 24 mg/kg. As shown in Fig. 5A,
`treatment with the scFv/rGel fusions all showed great
`antitumor effects, with the intermediate-affinity MH3-B1/
`rGel showing more enhanced and long-lasting tumor
`inhibition effects than lower affinity C6.5/rGel and
`ML3-9/rGel. There was little obvious toxicity observed
`with the administration of the immunotoxins with the
`
`In vivo study of the
`Figure 5.
`immunotoxins against BT474 M1
`tumor xenografts in nude mice. A,
`treatment of BT474 M1 tumors with
`immunotoxins at the dose of
`24 mg/kg. Mean tumor volume was
`calculated by W  L  H as
`measured by digital calipers. B,
`average body weight of the mice
`during the immunotoxin treatment. C
`and D, whole body imaging results of
`the mice intravenously injected with
`either IR-MH3-B1/rGel or IR-B1D3/
`rGel at 4, 24, 48, and 72 hours.
`Arrows point at tumor (T) and liver (L).
`E and F, the comparison of tissue-to-
`muscle ratio of IR-MH3-B1/rGel and
`IR-B1D3/rGel at 24 and 72 hours
`after injection into xenograft mice
`(n ¼ 5).
`
`www.aacrjournals.org
`
`Mol Cancer Ther; 11(1) January 2012
`
`149
`
`Downloaded from
`
`mct.aacrjournals.org
`
`on October 16, 2014. © 2012 American Association for Cancer Research.
`
`IMMUNOGEN 2112, pg. 8
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`Published OnlineFirst November 16, 2011; DOI: 10.1158/1535-7163.MCT-11-0519
`
`Cao et al.
`
`exception of the highest affinity B1D3/rGel. As shown
`in Fig. 5B, mice treated with this agent showed consider-
`able body weight loss (27%), and all the mice in this
`group died after fourth injection. Further studies were
`initiated to examine the reason of the toxicity of B1D3/
`rGel compared with other constructs.
`
`In vivo optical imaging
`To examine the in vivo toxicity observed in mice treated
`with the highest affinity B1D3/rGel, we used a fluorescent
`molecular imaging probe (IRDye 800CW) to label MH3-
`B1/rGel and B1D3/rGel for in vivo biodistribution studies
`(22). Nude mice bearing BT474 M1 tumors were injected
`(intravenously) with 1.5 nmol/L IRDye800-MH3-B1/rGel
`(IR-MH3-B1/rGel) or IRDye800-B1D3/rGel (IR-B1D3/
`rGel). The mice were then imaged at different times (4,
`24, 48, and 72 hours) with the IVIS Optical Imaging System
`(Fig. 5C and D). Both IR-MH3-B1/rGel and IR-B1D3/rGel
`were shown to accumulate in BT474 M1 tumors with the
`first 4 hours of postinjection (tumor to contralateral back-
`ground ratio of 1.85  0.12 and 1.85  0.09 for each) and
`reached maximal concentrations at 48 hours (tumor to
`contralateral background ratio of 2.70  0.31 and 2.58 
`0.15, respectively). In addition, accumulation of IR-B1D3/
`rGel was observed in the liver postinjection from 48 to 72
`hours, compared with IR-MH3-B1/rGel.
`To avoid any measuring errors caused by limited tissue
`penetration of fluorophores, animals were sacrificed;
`tumor and major organs were collected at the 24-hour
`and 72-hour time points and were subjected immediately
`to NIRF imaging (Supplementary Fig. S4). At 24 hours
`postinjection,
`there were no significant differences
`between the tissue distribution of IR-MH3-B1/rGel and
`IR-B1D3/rGel (Fig. 5E and Supplementary Fig. S5A).
`However, a better biodistribution of IR-MH3-B1/rGel
`over IR-B1D3/rGel could be identified 72 hours after
`injection, with the tissue-to-muscle ratio of tumor being
`1.5 times higher than the later one (e.g. 2.84  0.23 vs. 1.89
` 0.23, P < 0.002; Fig. 5F and Supplementary Fig. S5B).
`Correspondingly, a 1.6-fold higher liver retention of IR-
`B1D3/rGel over IR-MH3-B1/rGel was observed (4.65 
`0.61 vs. 2.89  0.56, respectively; P < 0.01). Both IR-MH3-
`B1/rGel and IR-B1D3/rGel were found to accumulate in
`the kidneys likely due to renal clearance of the low–
`molecular weight agents. In the other major tissues we
`collected, there were no significant differences in the
`distribution of these immunotoxins to tissues such as
`heart, lung, spleen, muscle.
`
`Accumulation of Her2/neu antigen and B1D3/rGel in
`the liver
`To identify the cause of the liver distribution with the
`highest affinity B1D3/rGel, we administered B1D3/rGel
`and MH3-B1/rGel to nude mice with or without tumors at
`a dose of 1.5 nmol/L. Mice were sacrificed 72 hours after
`injection. After homogenization, the liver samples were
`subjected to immunoprecipitation using anti-Her2/neu
`antibody and immunoblotted to assess the fusions. As
`
`shown in Fig. 6A, we found immune complexes of Her2/
`neu antigen and the B1D3/rGel in tumor-bearing mice but
`not in tumor-free mice. In contrast, we found lower
`amounts of immune complexes in the livers of mice
`treated with MH3-B1/rGel.
`Immunofluorescence staining confirmed colocalization
`and accumulation of B1D3/rGel with Her2/neu antigen
`in the liver of tumor-bearing mice, but less staining of
`antigen was observed from the livers in the MH3-B1/rGel
`treatment group (Fig. 6B).
`
`In vivo toxicity of immunotoxins in mice
`We then examined the influence of tumor-derived
`shed antigen on the hepatotoxicity of the B1D3 and
`MH3-B1 fusion toxins. Serum samples were collected
`72 hours after the administration of the fusions or rGel
`to mice with or without tumors, and enzymatic activ-
`ities of liver enzymes were determined (Fig. 6C and
`Supplementary Fig. S6). We found slight increases in the
`serum ALT, AST, and LDH levels in tumor-bearing
`versus tumor-free mice treated with rGel or MH3-B1/
`rGel. The greatest increases in all 3 markers were found
`with tumor-bearing mice treated with the high-affinity
`B1D3/rGel construct. This increase was not observed in
`tumor-free mice confirming that the observed hepato-
`toxicity was due to immune complexes of B1D3/rGel
`and Her2/neu antigen localizing in the liver.
`To verifythe hepatotoxicity in animals, mouse liverswere
`harvested 72 hours after immunotoxin injection and exam-
`ined. Compared with animals treated with rGel or MH3-
`B1/rGel, the livers of mice treated with B1D3/rGel showed
`severe liver damage characterized by marked necrosis and
`vacuolar degeneration of hepatocytes and extensive hem-
`orrhage (Fig. 6D). This suggests that the in vivo efficacy of
`high-affinity immunotoxins can be significantly impaired
`by the presence of shed antigen. Furthermore, the immune
`complexes formed by the immunotoxins and the shed
`antigen contribute to significant hepatotoxicity.
`
`Discussion
`
`Using a panel of scFv/rGel fusions specific for the
`same Her2/neu epitope (17) and nearly identical in their
`sequences and structure, we were able to assess the influ-
`ence of affinity. This appears to be one of the first com-
`prehensive examinations of the impact of affinity on the
`in vitro and in vivo behavior of immunotoxins. More impor-
`tantly, the current study clearly described the increased
`potential for high-affinity immunotoxins to form immune
`complexes in vivo resulting in off-target toxicity in liver
`responsible for clearing these complexes. These data
`have potential relevance to a number of anti-Her2/neu
`approaches (23, 24), as well as other targets for which there
`is a level of circulating antigen present (25, 26).
`Previous studies suggested that the binding affinity for
`antigen plays a pivotal role in the total concentration and
`penetration of scFv into tumors (13, 14). On the basis of
`scFv/rGel immunotoxins, we showed that increasing
`
`150
`
`Mol Cancer Ther; 11(1) January 2012
`
`Molecular Cancer Therapeutics
`
`Downloaded from
`
`
`
`on October 16, 2014. © 2012 American Association for Cancer Research. mct.aacrjournals.org
`
`IMMUNOGEN 2112, pg. 9
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`Published OnlineFirst November 16, 2011; DOI: 10.1158/1535-7163.MCT-11-0519
`
`Affinity Impact of Immunotoxins on Efficacy and Toxicity
`
`Figure 6. Analysis of the accumulation of Her2/neu antigen and scFv/rGel (MH3-B1/rGel or B1D3/rGel) complex–driven liver toxicity after 72 hours of injection.
`A, coim