IMMUNOGEN 2101, pg.1
`Phigenix v. Immunogen
`|PR2014-00676
`
`IMMUNOGEN 2101, pg. 1
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`
`V _
`
`This material may be protected by Copyright law (Title 17 U.S. Code)
`
`
`
`_.1.‘,__._,_—-._,
`
`1,11, J. C£i.’l‘('£’l'.' 88, 267—273 (2000)
`2(‘ W0 Wiley—Liss, Inc.
`
`Publication of the international Union Against Cancer
`
`
`
`A NOVEL RECOMBINANT FUSION TOXIN TARGETING
`HE‘°.—2/NEU—OVER—EXPRESSING CELLS AND CONTAINING HUMAN TUMOR
`NECROSIS FACTOR
`
`Michael G. ROSENBLUW‘, Shirley A. HORN and Lawrence H. CHEUNG
`
`Iiizmimap.'nri'imic0l0gy and Targeted Therapy Secrirm, Depru'rm.em‘ ofBioiJ1tn1tt7I0fl'ier(tp)7, Uiir'v.v3rsir_v of Texas M.D. Anderson Cancer
`Cgfif
`3 Houston, Te.r(zs, USA
`I‘
`
`i.
`
`/ -
`
`
`
`
`
`_,_j______j.,_____,,_____4_:—_, ,e~_.,—e-——
`
`"“ “” ""'“*“"‘“ "ffects of TNF.
`it appear to be
`ry membrane-
`_RmENB,_UM, MICHAEL G., HORN, SHIRLEY A., and Ci-rsuNo, LAWRENCE H., A novel recombinant fusion toxin targeting HER-2/neu—over~ re surveillance
`lfipressmg cells and containing human tumor necrosis factor. Int. J. Cancer, 88, 267-273 (2000).
`3 its cytotoxic
`I
`breast cancer
`we to the authors’ error, a mistake appeared in the legend to Figure 6. The figure and correct legend are printed below.
`escaping host
`studies aimed
`)1-neflufl'lOl'5 regret this error.
`.R2/net: signal
`I cancer cells.
`ial to serve as
`: agents and to
`agents such as
`:r and Pastan,
`2 demonstrated
`tumor antigen
`sulting in poor
`: antibody size;
`;, leading to re-
`.e concerns have
`antibodies while
`:tions within the
`., 1994). Studies
`ed the utility of
`. such as 0L~IFN
`. Reisfeld at al.
`gth antibodies in
`in cytokines (de-
`iese recombinant
`‘9l; Reisfeld and
`
`SKBRRU-lPJ
`
`5K.BR'i(LF;
`
`
`
`<—- HER zmcu
`
`+ SKBR3 LP
`—I— SKBR3 HP
`
`100
`
`In
`
`4.oraa
`
`2|!
`
`"A.ofContra‘
`
`
`
`TNF concentration
`(Unitsirnl)
`
`FIGURE 6 — Western analysis of 2 variants of SKBR-3 cells (insert)
`demonstrates that SKBR-3—LP cells express approximately 5~fold
`higher I-[ER2 protein than SKBR—3»HP cells. Direct comparison of the
`cytotoxic effects of continuous exposure of various concentrations of
`TNF demonstrates that the cell line expressing higher levels of HER2
`was effectively resistant to TNF while the cells expressing low levels
`of HER2 were sensitive to the cytotoxic effects of TNF.
`
`roraring the bind—
`.n, consist of the
`ant)
`linked by a
`itact IgGs, scFvs
`Da) and structural
`iparable antigen-
`an the analogous
`It has been pro-
`:1‘ penetration into
`1 reduction in the
`immunogeniciry ouservcu Wuu J..\'. .ruuumst.,...d Fabs (Savage er
`al., 1993) compared to that of intact nturine antibodies. Numerous
`recombinant antibodies fused to plant or bacterial toxins such as
`P.rcudoirrorros exotoxin, ricin, and gelonin have been reported
`
`Grant sponsor: National Institutes of l-lealth: Grant number: CA 16672;
`research conducted, in part, by the Clayton Foundation for Research.
`
`“"Corres aondence to: Michael G. Rosenblum. lmmunopharmacology and
`Targeted "herapy Section, Box 044-, Department of Bioimrnunotherapy,
`University of Texas M.D. Anderson Cancer Center. 1515 Holcoinbe Blvd,
`Houston, TX 77030. USA. Fax: +7l3~794-426].
`E—mail: mrosenbl@nntes.mdacc.tmc,edu
`
`Received 17' January 2000; Revised 13 March 2000; Accepted 17 March
`2000.
`
`and colon and salivary gland actenocarcinonias tsemba at al.,
`1985- Yokota er ni., I988). Slamon er til. (1987) found HER2/}1€L£
`over-expressed in approximately 30% of 189 primary breast car-
`cinomas examined. Their study demonstrated that over-expression
`Of HER2/neir was correlated with poor disease prognosis. Fol-
`low»: in studies have also suggested that HER2/net: cellular expres—
`sion is associated with a shortened disease-free survival (DFS)
`(Sesludri er nl., 1993; Mansour er al., 1994). Thus, the clinical
`Observations of the importance of HER2/net: as a negative prog-
`HOSII». indicator in vivo have been repeatedly confirmed by rnolec—
`U131‘ studies demonstrating the central role of this oncogene in
`Promotion of the growth of transformed cells and in increasing
`their metastatic potential.
`One of the key roles this oncogene appears to play is in mod-
`U13-tirtii of the cellular response to cytotoxic cytokines such as
`Tumor necrosis factor (TNF) (Tang at 511., I994; Lichtenstein er al.,
`1991‘. A variety of research groups have demonstrated that HER2/'
`
`j<
`IMMUNOGEN 2101, pg. 2
`Phigenix v. Immunogen
`|PR2014-00676
`
`—
`
`IMMUNOGEN 2101, pg. 2
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`
`int. J. Cancer.‘ 88, 267—273 (2000)
`@ "000 Wiley-Liss. Inc.
`
`
`
`Publication of the International Union Against Cancer
`
`A NOVEL RECOMBINANT FUSION TOXIN
`TARGETING
`HER—2/NEU—OVER-EXPRESSING CELLS AND CONTAINING HUMAN TUMOR
`NECROSIS FACTOR
`
`Michael G. RoseNBLuM’*‘, Shirley A. HORN and Lawrence H. CHEUNG
`Imn.tuinp/rar‘12iuc0fogy rmd Ttirgefed Therapy Section, Depar‘I1n'em ofBioimmuriofherap)‘, University
`Cgnirr‘, Houston, Texas, USA
`.19‘
`
`0_fTe.m: M.D. Anderson Cancer
`
`the proto-oncogene HER2lneu in
`Cver-expression‘ of
`breast cancer and certain other tumors appears to be a
`¢en.ral mechanism that may be partly responsible for cellu-
`lar progression of the neoplastic phenotype. Transfection of
`mar ‘imalian cells and over-expression of HER21‘neu appears
`to result in reduced sensitivity to the cytotoxic effects of
`tun or necrosis factor (TNF) and reduced sensitivity to im-
`mune effector killing. The single-chain recombinant antibody
`sFv73 recognizes the cell-surface domain of HER2Ineu. The
`cDNA for this antibody was fused to the cDNA encoding
`human TNF, and this sFv23I'TNF fusion construct was cloned
`into a plasmid for expression in Escherichia colt‘. The fusion
`protein was expressed and purified by ion-exchange chroma-
`mgr aphy. SDS-PAGE demonstrated a single band at the ex-
`pected m.w. (43 kDa). Western analysis confirmed the pres-
`ent . of both the antibody component and the TNF
`component in the final fusion product. The fusion construct
`was tested for TNF activity against L-929 cells and found to
`have biological activity similar to that of authentic TNF (SA
`420 nM). The scFv23lTNF construct bound to SKBR-3
`(Hl:R2-positive) but not to A-375 human melanoma (HERZ-
`negative) cells. Cytotoxicity studies against log-phase human
`breast carcinoma cells (SKBR-3-HP) over-expressing HER2l
`neu demonstrate that the sFv23I'TNF fusion construct was
`l,Dt 0-fold more active than free TNF. Tumor cells express-
`ing higher levels of HER2lneu (SKBR-3-LP) were relatively
`resf :tant to both the fusion construct and native TNF. These
`studies suggest that fusion constructs targeting the HER2Ineu
`sur'1ce domain and containing TNF are more effective cyto-
`toxic agents in vitro than native TNF and may be effective
`aga"n5t tumor cells expressing intermediate, but not high,
`levels of HER2Ineu. int. ]. Cancer 88:267-273., 2000.
`© 9900 Wiley-Lisa‘, Inc.
`
`net:-transformed cells are resistant to the cytotoxic effects of TNF.
`In addition, NIH 3T3 cells lransfected with HER2/hen appear to be
`resistant to immune effector cell killing mediated by membrane-
`bound TNF. Since TNF plays a central role in immune surveillance
`functions (Saks and Rosenblum, I992), resistance to its cytotoxic
`effects mediated by HER2lneu over—expression in breast cancer
`may allow transformed cells a growth advantage by escaping host
`defense mechanisms. To date, there have been few studies aimed
`at elucidating the biochemical events linking HER2/new signal
`transduction with TNF signal transduction in breast cancer cells.
`Monoclonal antibodies (MAbs) have the potential to serve as
`targeting vehicles for various classes of therapeutic agents and to
`be utilized for targeting protcinaceous therapeutic agents such as
`interleulcins,
`lymphokines, and cytokines (Theuer and Pastan,
`I993; Reisfeld er (11,,
`l997). However,
`there are demonstrated
`pitfalls of this approach,
`including hetcrogenous tumor antigen
`expression in viva; pharmacological barriers resulting in poor
`tumor penetration by the antibody, due in part to antibody size;
`antigenicity of antibodies and antibody constructs, leading to re-
`duced utility. Molecular approachcs to address these concerns have
`provided numerous options to reconfigure natural antibodies while
`simultaneously incorporating effector or toxin functions within the
`same molecule (Colnaghi er ai., 1993; Hand :31 al., 1994). Studies
`by our group and by others initially demonstrated the utility of
`chemical constructs of antibodies and cytokines such as or-IFN
`(Ozzello er a1'.,
`I994; Zuckerman er al.,
`I987). Rcisfeld er (If.
`(i997) have studied fusion constructs of full-length antibodies in
`which framework domains are replaced by human cytokines (de-
`scribed as immunocytokiues). Clinical trials of these recombinant
`constructs are now in progress (Gillies er mi. 1991; Reisfeld and
`Gillies, l996a,b).
`
`"lire HER2/marl proto-oncogene encodes a 185 kDu transmem-
`brace glycoprotein kinase (gp 185) with extensive homology to the
`epiuermal growth factor (EGF) receptor (Shepard er a.’., 199];
`Jardincs er ol., 1993: King at (n’.,
`I985; Schechter or 5:1,, 1985;
`Yamamoto er of, 1986). Transfection studies suggest that HER2l
`new over—expression may play a direct role in cellular transforma-
`tion to a neoplastic phenotype (DiFiore er ul., 1987; Hudziak at 171..
`i987). Amplification of the gene and overexpression of the gp 185
`protein product have been described in a number of human can-
`cer . including mammary and ovarian carcinomas. gastric tumors,
`and colon and salivary gland adcnocarcinomas (Semba cr 111.,
`1995; Yokota er a1., I988), Slamon er al. (1987) found HER2/new
`Over-expressed in approximately 30% of I89 primary breast car-
`cin smas examined. Their study demonstrated that over—expression
`Of HER2/ncu was correlated with poor disease prognosis. Fol-
`lov —up studies have also suggested that HER2/nen cellular expres-
`Sion is associated with a shortened disease—free survival (DFS)
`(SC .hadri er 0].,
`I993; Mansour er 171., 1994). Thus, the clinical
`observations of the importance of HER2/ueu as a negative prog-
`no: ‘ic indicator in viva have been repeatedly confirmed by molec-
`ular studies demonstrating the central role of this oncogene in
`PH notion of the growth of transformed cells and in increasing
`their metastatic potential.
`line of the key roles this oncogene appears to play is in mod-
`ulation of the cellular response to cytotoxic cytokincs such as
`1'L1I.tor necrosis factor (TNF) (Tang er m'., 1994; Lichtenstein er ctl.,
`1991). A variety of research groups have demonstrated that HERZI
`
`Single-chain antibodies (scFvs or sFvs), incorporating the bind-
`ing characteristics of the parent immunoglobulin, consist of the
`antibody VL and VH domains (the Fv fragment) linked by a
`designed flexible peptide tether. Compared to intact IgGs, scFvs
`have the advantages of smaller size (approx. 30 l<Da) and structural
`simplicity (single-chain vs. 4 chains) with comparable antigen-
`binding affinities, and they are more stable than the analogous
`2-chain Fab fragments (Pantolino er nl., 1991). It has been pro-
`posed that the smaller size of scFvs provides better penetration into
`tumor tissue, improved pharmacoltinetics, and a reduction in the
`immunogenicity observed with i.v. administered Fabs (Savage er
`01., 1993) compared to that of intact murine antibodies. Numerous
`recombinant antibodies fused to plant or bacterial toxins such as
`Pseircloniancis cxotoxin. ricin, and gelonin have been reported
`
`
`Grant sponsor: National Institutes of Health; Grant number: CA l6672;
`research conducted, in part, by the Clayton Foundation for Research.
`
`
`“Correspondence to: Michael G. Rosenbluni, lmmuuophiumacology and
`Targeted Therapy Section, Box 044. Department of Bioimmunothcrapy,
`University of Texas M.D. Anderson Cancer Center, 15 l 5 Holcombc Blvd.,
`Houston. TX 77030, USA. Fax: +713-794-4261.
`E—mail: mrosenbl@notes.mdacc.tmc.edu
`
`2000.
`Received 17 January 2000; Revised l3 March 2000‘ Accepted 17 March
`
`
`
`IMMUNOGEN 2101, pg. 3
`
`Phigenix v. Immunogen
`|PR2014-00676
`
`IMMUNOGEN 2101, pg. 3
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`
`268
`
`ROSENBLUM 51‘ AL.
`
`(Preijers, 1993) and are currently Llndergoittg pre-Clinical and
`phase I trials (Theuer and Pastan, 1993: Uckun and Frankel, 1993).
`TNF is a cytotoxic polypeptide secreted primarily by activated
`macropiiages (Carswell er ((1., 1975: Aggarwal er al.. 1985). which
`shares some sequence homology (30%) with another peptide hor-
`mone. lymphotoxin (LT or TNF-B) secreted by activated lympho-
`cytes (Rosenblum and Donato. 1989). Purified recombinant human
`TNF is a single-chain, non-glycosylated polypeptide of m.w. 17.1
`kDa. In vitro, TNF is cytostatic or cytotoxic to a number of human
`tumor cells, including SKBR—3 breast carcinoma and A—375 hu-
`man melanoma. Several grottps have demonstrated that human
`tumor cells can display between 100 and 5,000 TNF receptor sites
`per cell (Tsujimoto er m'., 1985: Sugarman et at. 1985). However,
`no apparent correlations have been observed between receptor
`number (or affinity) and the cellular response to the cytotoxic
`effects of TNF. suggesting that post-receptor signaling events
`primarily modulate TNF biochemical effects (Nitsu at (11,. I985).
`Clotting and sequencing of the genes encoding the TNF receptor
`have shown that 2 separate gene products encode TNF—binding
`activity and appear to be homologous in portions of their extra-
`cellular binding domains (Stauber er al., 1988; Hohmann at al'.,
`1989; Loetscher er al., 1990; Schail et aI., 1990). This family of
`molecules appears distinct when compared with sequences defin-
`ing characte1'istics of other cytokine or growth factor receptors.
`Previous studies in our laboratory have detnonstrated that chem-
`ical conjugates of human TNF and MAbs display significant
`targeted cytotoxic properties against tumor cells in culture that
`appear to be far superior to those of native TNF (Rosenblum er al.,
`1989, 1991, 1995). In addition, studies in xenograft models sug-
`gest that these immunocytokines readily accumulate specifically in
`tumor tissues and demonstrate superior in viva anti-tumor activity
`compared with native TNF. The purpose of the current study was
`to extend these original observations by developing a second-
`generation molecular construct of a recombinant single—chain an-
`tibody fused to the TNF molecule,
`thereby incorporating both
`antibody and TNF functions within the same molecule. We there-
`fore designed and constructed a recombinant fusion toxin com-
`posed of a single-chain antibody targeting the HERE/lieu proto-
`oncogene and containing human TNF as a cytotoxic e1Tector
`molecule. Further studies were designed to examine the biological
`properties of the fusion <:0nstruCt and its potential for overcoming
`HER2-mediated resistance to the cytotoxic effects of TNF.
`
`MATERIAL AND METHODS
`
`lines SKBR—3 and L—929 were obtained
`The mainnialian cell
`from the ATCC (Rockville, MD). T1'yptone and yeast extract were
`purchased from Difco (Detroit, MI). L-Arabinose was purchased
`from Sigma (St. Louis, MO). Rabbit polyclonal anti-scFv23 anti-
`body was a generous gift from Oncologix (Gaithersburg, MD).
`Goat anti-rabbit
`IgG peroxidase conjugate was obtained from
`Bochringer—Mannheim (Indianapolis. IN). DTE (dithioerythritol)
`was purchased from Sigma. Tween-20 came from Fisher Scientific
`(Pittsburgh, PA). FBS was purchased from Atlanta Biologicals
`(Norcross. GA).
`
`sFv23fl"NF gene c0nsmtct'r'0n
`The CDNA encoding the single-chain anti-H'ER2/net: antibody
`designated sFv23 was obtained from Oncologix. and the CDNA
`encoding mature human TNP was a generous gift from Dr. J.
`Klostergaard (MD. Anderson Cancer Center. Pouston, TX). The
`sFv23/'TNF CDNA was constructed by 2-step PCR. The first step
`consisted of separate PCR amplification of the antibody and TNF
`coding sequences. utilizing forward and reverse primers for each
`sequence. The final step consisted of PCR of the sequences.
`utilizing overlap primers additionally incorporating a
`flexible
`tether (G4-S) between the antibody and TNF (Fig. 1).
`E\‘))rc.r.rfr:n rJ_f'fttst'mi ])J‘0fE‘lIl sFv23/TNF in Esr‘l‘m'ic'111'a calf
`Bacterial colonies transfected with the plasmid carrying the
`sFv23fTNF insert were agitated in a bacterial shaker (lnnova 4330',
`
`111 Linker
`
`(Gly Set Th: Scr Gly Sc: Gly Lys 5:‘ Ser Glu Gly Ly: Ely)
`‘l'
`
`pdfl
`
`v,
`
`v,,
`
`114?
`
`""""""""""""""" __
`
`linker G,S
`{Gly Gt)-rGlyGly Sn’)
`
`Thmmhin
`canvage 5.”
`V
`
`(Hit;
`
`:
`
`
`
`FIGURE 1 — Design of the sFv23/TNF fusion construct and assembly
`oi’ the expression vector.
`
`1 of TYE medium
`New Brunswick Scientific, Edison. NJ) in I
`(15 g tryptone, 10 g yeast extract, 5 g NaCl) containing 50 ,.Vg:/nil
`tetracycline (Sigma) overnight at 37°C at 250 rpm. Bacterial cells
`were harvested by centrifugation, and the pellet was disperstd into
`2 1 of fresh TYE. Expression of the target protein was induced by
`addition of arabinose to a final concentration of 0.1%. The culture
`was further incubated at 37°C for 2 to 3 h. Cells were collected by
`centrifugation. resuspended in 80 ml extraction bLtffe1' [So mM
`Tris-HCl ("pH 8), 20 mM EDTA, 0.25 mg;/ml lysozyine (Siun1a)],
`and incubated with shaking for 1 hr at room temperature. Triton
`X-100 and sodium chloride were added to the sample at :1 final
`concentration of 2% and 0.5 M. respectively, then incubateu for Elll
`additional 30 min. After centrifugation, the insoluble fraction of
`the inclusion bodies was resuspended in 160 ml of 50 mi\i Tris-
`HCl (pH 8.0), 20 n1M EDTA and sonicated (6 X 20 sec) using it
`Vir Sonic 300 sonicator (Virtis, Gardiner. NY). Inclusion bodifis
`were then harvested by centrifugation, washed 3 times in the same
`buffer, and stored at —80°C.
`
`Protein .r0{nbi'li;(tti0r1, refolding, and pu."t'fFcatt'0n
`Insoluble inclusion bodies were denatured by addition -if 6 M
`guanidine. I00 mlVl Tris-HCl (pH 7.5), 2 mM EDTA. and 50 HIM
`DTE to make a final concentration of 3 mg/ml protein (as assessed
`by Bradford protein determination). After a minimum of 2 W
`agitation at room temperature. solubilized proteins were diluted
`l00—fold in refolding buffer [100 mM Tris-HCI (pH 7.5). 2 1“-M
`EDTA] and incubated at l2"C for 48 hr. The protein wa subse-'
`quentiy bound to a small column containing SP Sepharosfi E35‘
`Flow Resin (Pharmacia, Gaithersburg. MD).,The bound tractiolt
`was eluted from the column by addition of’! M NaCl in 100 IUM
`Tris—HCl (pH 7.5), 2 mM EDTA. Eluted protein fractit 15 W9”
`pooled and dialyzed against TBS, and the final product was 1115“
`further characteri2'.ed as described below.
`
`lfl/(?.i"f(’}‘I1 l'l"1€fn"l{)(f.F
`
`Protein samples from the crude extracts of E. cnli, the purified
`inclusion bodies. sFv23. and r-TNF were analyzed by SD”-PAGE
`under reducing conditions. The gel was electrophoretically l1'fl_“5'
`ferred overnight onto a nitrocellulose transfer and immol llilflmn
`membrane (Protran: VWR, Sugar Land, TX). The membrane “'3;
`incubated in 5% BSA/TBS [20 mM Tris—HCl, 137 mM l" tclv fin]
`0.5% Tween 20 (pH 7.6)] and then incubated for
`1
`ill‘ ‘"1?
`,
`anti-sFv23 rabbit polyclonal antibody (l:l0,000 dilutior.
`in TB
`
`.
`
`IMMUNOGEN 2101, pg. 4
`Phigenix v. Immunogen
`|PR2014-00676
`
`IMMUNOGEN 2101, pg. 4
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`
`Tween).AftersuccessivewashingwithTBS/Tween—20,themem-
`
`
`
`ANTI-I-IERE/NEU FUSION TOXIN CONTAINING TNF
`
`269
`
`brare was incubated with goat anti—rabbit IgG horseradish perox-
`idase (l:5.000 dilution in TBS). The membrane was developed
`ustrg ECL reagents (Amersham. Arlington Heights, IL) and ex—
`Posed to X-ray film (Kodak, Rochester, NY).
`1,9,9 TNF biocrsroy
`
`The cytotoxic activity of TNF was determined based on cyto-
`[oxltjlly to the transformed murine fibroblast cell line L-929. Log—
`plume Cells in culture media (RPMI—l640 with l.5 mM glutamine
`and 10% FBS) were plated in a 96-well
`tissue culture plates
`(Fal ‘on, Lincolr}’Park. NJ) at a density of 2 X 10“ cells/well and
`incubated overnight at 37°C in a 5% CO1 atmosphere. Then, 200
`IL] CFTNF in PBS starting at 100 units/rnl and serial dilutions were
`added in the presence of actinomycin D (0.5 ug/‘nil, Sigma).
`Ljkrwise, serial dilutions of sFv23-TNF were added, and the plate
`was incubated for 24 hr. Surviving adherent cells were then stained
`by zclding 100 it] of crystal violet 10.5% (wzv) in ethanol]. The
`stain was incubated on the plates for 0.5 ht, excess stain was
`renrwvcd,
`the plates were washed with water and allowed to
`air—dry, and the remaining dye was solubilized by addition of I50
`nl L "S0renson‘s buffer (0.1 M sodium citrate, pH 4.2). Plates were
`read on a microplate ELISA reader at 540 nm.
`Binuing sriidier 0fsFv23/TNF
`
`Ending of sFv23/TNF to SKBR-3 cells was also assessed by
`ELISA. L0g—phase SKBR—3 cells were washed in PBS, and 50,000
`cell
`/50 pl PBS were added to each well of a 96—well tissue culture
`plate a11d dried overnight in a 37°C oven. Plates were blocked by
`addi‘ion of 100 pl 5% BSA in PBS. A 50 ul aliquot of sFv23/TNF
`fusion protein and serial 2—fold dilutions were then added to
`app opriate wells and incubated for l hr at room temperature. After
`3 washes (PBS/Tween—20), anti—sFv23 rabbit polyclonal antibody
`(1 t. g/ml in PBSA/Tween—20) was added and incubated for J hr.
`Wells were tapped dry, and I00 ul of horseradish peroxidase-
`conf igated goat anti~rabbit antibody (Boehringer-Mannheim) were
`added. Plates were developed by addition of ABTS substrate in 0.]
`M citrate buffer (pH 4.2, Sigma) and incubated for 1 hr. Optical
`density was measured at 405 nm on a Bio—Tek Autoreader.
`
`Cym1o.ricffy of TNF mid .s‘Fr23/TNF ugrtinsr liziirrarr brerisr
`i‘ltllt"I‘ SKBRJ CGHX
`
`Log—phase SKBR-3 cells were diluted to 8,000 cells/100 f..t.l
`rnefiiurn. Aliquots (100 pl) were added to 96-well, Plat-bottomed
`
`tissue culture plates and incubated for 24hr at 37°C with 5% C0,.
`Purified sFv 23/TNF or recombinant human TNF was diluted 1:B
`in culture medium (McCoy's 5a with 1.5 mM glutam_ine and 10%
`FBS). Aliquots of each sample (200 tel) were added to the plate in
`Z—fold serial dilutions. Plates were further incubated at 37°C, 5%
`CO3 for 72 hr. Remaining adherent cells were stained by adding
`150 nl crystal violet (Fisher, 0.5% in 20% methanol). Dye~stained
`cells were solubilized by addition of 150 it] of Sorenson‘s buffer
`[0.l M sodium citrate (pH 4.2) in 50% ethanol], and the plates
`were read at 540 nm in an ELISA plate reader (BioTei<,
`lnc.,
`Winooskj, VT).
`
`R ESULTS
`
`PCR products were cloned into a Vector for bacterial expression
`of the recombinant insert. The complete insert was submitted for
`dideoxynucleotide sequencing (M.D. Anderson Cancer Center
`Core Sequencing Facility), and the final gene product sequence
`was confirmed. We utilized a flexible l4—amino acid linker tojoin
`the VH and VL regions and a smaller tether (G48) to link the TNF
`to the antibody (Fig. 1).
`is
`Bacterial expression of the scFv23/TNF fusion construct
`shown in Figure 2. After growth and induction with arabinose at
`37°C, production of the construct was approximately 5% to 10%
`of total protein, as assessed by SDS—PAGE. Production of the
`target fusion construct was estimated to be 25 to 50 mgll, as
`assessed by Western analysis. Purification of the soluble protein
`utilizing ion—exchange chromatography resulted in essentially ho-
`mogeneous material. as assessed by SDS-PAGE after elution from
`the exchange resin. Yield of final purified material was approxi-
`mately l00 ptg/l of bacterial culture. Western analysis of the
`product utilizing antibodies to either TNF or scFv23 (Fig. 3)
`demonstrated an immunoreactive species with both antibodies at
`the expected m.w. (43 kDa).
`Binding of thc sFv23/TNF fusion toxin to adherent SKBR—3
`cells was assessed by ELISA using an anti-TNF antibody. Binding
`of both the native sFv23 single—chain antibody and the sFv23/TNF
`fusion construct is shown in Figure 4. The binding of both agents
`was similar and appeared to be dose—dependant. A slightly higher
`binding of the fusion construct compared to the antibody was
`noted at
`the highest concentrations tested. Optimal binding to
`target cells occurred after incubation with 0.75 p.M of the fusion
`construct (Fig. 5). There was no apparent binding of the construct
`
`LAj§
`
`SAMPLE
`
`- A 'J‘NF(t7kD) Standard
`' B
`Uninduccd sFv23 bacterial lysate
`. C
`Induced st-‘V23 soluble lysate
`-
`I")
`/\t“t'mity (IMAC) resin prior to elution
`' U
`sl-‘V23 cluatc from affinity resin
`
`Uninduced sFv23-TNF bacterial lysate
`F
`'
`- G Induced sl-‘V23-TNF soluble lysate
`
`- H Affinity (IMAC) resin prior to elution
`
`'
`
`I
`
`SFV23-TNF conjugate from affinity resin
`
`Molecular weight markers
`
`A B c D E F o u I
`— M I
`
`.1
`_i
`
`
`
`;—
`
`‘.5;
`:
`?
`i d
`
`'
`
`i
`
`t t I t
`
`t
`t
`
`_ FIGURE 2—Expression and purification of both
`Immobilized metal affinity column (IMAC).
`
`the sFv23 antibody as well as the sFv23/TNF fusion construct fmin E. coti. utilizing an
`
`_
`
`IMMUNOGEN 2101, 5
`
`Phigenix v. Immunogen
`|PR2014-00676
`
`IMMUNOGEN 2101, pg. 5
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`
`ROSENBLUM ET AL.
`
`C 00m assie
`
`Anti-sfv23 Ab
`
`Anti-TNF Ah
`
`
`
`
`
`FIGURE 3- SDS—l’AGE and Coornassie staining profiles of bacteria prior to induction (left lane) and following induction (right lane by
`addition of ITPG. The position of the scFv’-23/"NF fusion construct is indicated by the arrow. Western analysis using anti~sFv23 antibody
`demonstrates reactivity against purified scFv23 antibody (left lane) and the scFv23/TNF fusion construct (right lane); probing the same We: em
`blot with an anti-TNF antibody demonstrates reactivity only with the scFv23/TNF construct.
`
`OD.590
`
`
`
`—v— .
`
`125
`
`i
`
`CONCENTRATIO N(uM)
`
`FIGURE 5— ELISA binding of the scFv23/TNF construct to either
`antigeirpositive SKBR—3 or antigen-negative A-373 cells.
`
`0.0
`
`i?
`
`025
`
`
`also
`
`
`0'75
`
`
`100
`
`
`l
`
`I I
`
`l 1
`
`activity could be due to stearic hindrance of the antibody
`interfering with interaction of TNF with its receptor sitr Al-
`ternatively, since TNF operates optimally as a compact triiner
`in solution (Rosenbluin er 0]., 1991),
`the antibody comp inenl
`could also interfere with optimal aggregation of the TNF com-
`ponent in solution.
`Two breast carcinoma cell lines, SKBR—3—LP and SKBR—3—HP-
`were assessed for their relative expression of HER2/net: and 1'61‘
`ative sensitivity to TNF-induced cytotoxicity. The LP variant cell
`line contained relatively higher levels of H5132/1:21: compared [0
`the HP line (Fig. 6). In addition, the LP line demonstraterl niuch
`greater resistance to TNF»incluced cytotoxic effects. The Cy[0l0X'
`icity of TNF and scFv23fTNF was assessed against
`log—‘)l1f|5“’
`SKBR—3—HP cells; as demonstrated in Figure 7. the ICSU for TNF
`exceeded 40.000 units/ml. while the scFv23fTNF fusion toxin
`demonstrated an [C50 value 2,0UO—foid lower, at approximate1)' 20
`units/ml. In contrast, the activity of both TNF and scFv2."TNF
`was assessed against TNF—resistant.cells expressing higher leVfl_l-‘
`of HER2/neu. These cells were relatively resistant to the cyt
`f07“°
`effects of both the sr:Fv23/TNF construct and TNF (Fig. 8). HOW‘
`ever. the N35,, of TNF against these cells was 5 X 107 un. s/nllw
`while that of the sFv23/"TNF construct was approximately 20~f01d
`
`lower(2 X I0“units/nil).
`
`‘
`
`i
`
`IMMUNOGEN 2101 , pg. 6
`Phigenix v. Immunogen
`IPR2014-00676
`
`.3U
`
`.25
`
`.20
`
`.15
`
`.10
`
`.05
`
`
`
`Absorbonce(4U5nm)
`
`El sFv23 (His) 5
`
`O sFv23-TNF (His)5
`
`D
`
`250
`
`500
`
`750
`
`IEIUU
`
`Concentrciti on (pM)
`
`FIGURE 4 — ELISA binding of either sFv23 or sFv23lTNF to SKBR—
`3—in1mobilized cells.
`
`to antigen—negative (A375) cells tested under identical conditions
`(Fig. 5).
`The cytotoxicity of TNF and scFv23/TNF fusion toxin was
`assessed against marine L—929 cells (data not shown).
`lC50
`values were I and I00 pM for native TNF and the ant1'bocly/
`TNF fusion toxin, respectively. This denionstrated approxi-
`mately a l00~fold decrease in the apparent specific activity for
`the construct compared to native TNF. This decrease in specific
`
`IMMUNOGEN 2101, pg. 6
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`
`ANTl-l-IER2fN.':'U FUSION TOXIN CONTAINING TNF
`
`27]
`
`‘.5lISKBIUIHP)
`
`
`
`ll-:1I-5~r<*=U=» 4% HER 2iNEU
`
`%ofControl
`
`+SKBR3 LP
`-I-SKBR3 HF‘
`
`10U
`
`80’
`
`_._ -I-NF
`—-l— 5fV23.lTNF
`
`
`
`103
`
`it)“
`
`105
`I
`105
`Concentration
`(Unitslml)
`
`107
`
`10°
`
`109
`
`100
`
`an
`
`an
`
`40
`21)
`
`Ea
`is
`
`G g
`
`o
`
`lfl‘
`
`—.—_
`
`,
`1D’
`
`-—,—.—.—.-. -.,
`10'
`10’
`TNF Concentration
`(UnitsImI)
`
`.—.—
`
`rm
`1|)‘
`
`FIGURE. 8- Cytotoxicity of TNF and sFv23/TNF against log-phase
`SKBR-3-LP cells in culture. The N35“ of TNF against these resistant
`cells was 5 X 107 unitslnil. while that of the sFv23/TNF construct was
`approximately 20-fold lower (2 X 10" units/ml).
`
`
`
`
`
`.—j_...__....__j_._m,?jrt__?..?m_—j.._,.jg..au.::——~—:—-:~. u #:"—‘f*‘--T:‘—'‘1‘
`
`
`
`I’ IGURE 6 —Western analysis of 2 variants of SKBR—3 cells (insert)
`demonstrates that SKBR-3-LP cells express approximately 5~fold
`l()\\ :r HER2 protein than Sl{BR—3—I-IF cells. Direct comparison of the
`cytotoxic effects of continuous exposure of various concentrations of
`TNT‘ demonstrates that the cell line expressing higher levels of HER2
`was effectively resistant to TNF while the cells expressing low levels
`of ‘IERZ were sensitive to the cytotoxic effects of TNF.
`
`100
`
`90
`
`80
`
`70
`
`60
`
`50
`
`40
`
`30
`
`20
`
`10
`
`%ofControl
`
` 9
`
`I
`
`sFv23-TNF (His)5
`TNF
`
`
`
`mi
`
`102
`
`103
`
`104
`
`105
`
`ms
`
`Concentration (u/ml)
`
`FIGURE 7 — Cytotoxicity of either TNF or purified sFv23/TNF
`flganist log-phase SKBR—3-HP cells in culture. The IC50 of the sFv23/'
`TNF construct (20 units/ml) was approximately 2,000~fold lower than
`that of free TNF (4.000 units/ml).
`
`DISCUSSION
`
`l\-umerous groups have examined the biological activity of
`antibodies chemically conjugated with or genetically fused to
`numerous human cytokines (Ozzello er al, 1993'. K_im er (ll'l..
`1993). We initially described a novel construct containing recom-
`
`binant human TNF—ot chemically conjugated to the murine anti-
`melanoma MAb ME-018 (Rosenblum er ci.l., 1989, I991, Rosen-
`blum er cil'.,
`I995). Against TNF-sensitive melanoma cells,
`the
`construct was several—fold more cytotoxic than free TNF. In ad-
`dition, this ME/TNF constrtict was extremely cytotoxic to human
`melanoma target cells virtually resistant to TNF itself. Follow—up
`studies demonstrated that the augmented cytotoxicity of the ME’
`TNF construct was primarily due to the antibody component of the
`construct binding to target cells. Studies of the MEWTNF construct
`in xenograft
`tumor models demonstrated efficient
`localization
`within solid tumors after systemic administration, similar to that of
`the native antibody. Finally, this construct demonstrated impres-
`sive cytotoxicity against s.c. xeriografts in nude mice. These stud-
`ies clearly demonstrated that antibody-mediated delivery of agents
`such as cytotoxic cytokines is feasible and has the potential
`to
`modify and enhance the in vivo and in vfrro biological properties
`of the original cytokine. Since our initial report of antibody/TNF
`conjugates. other groups have also reported antibody/cytokine
`constructs with agents such as GM-CSF, IL-2, and lymphotoxin
`(Reisfeld and Gillies. l996a,h; Rcisfeld er al., 1996; Dreier er nl.,
`1998).
`Over-expression ofthe HER2lueti oncogene appears to suppress
`the cellular cytotoxic response to TNF (Hudziak at all, 1990;
`Lichtenstein er a[_, 1990, 1992). However, the cellular mechanisms
`that could account for this effect have not been elucidated. In a
`study by Hudzialt at (if. (1990), transformation of NIH 3T3 cells
`with HER2/neat provided cellular resistance to the cytotoxic effects
`of TNF without appreciably affecting TNF receptor function in the
`transformants. The Current study. examining 2 SKBR-3-selected
`variant cell lines expressing high and intermediate levels of HER2,
`demonstrates that the ICED values for TNF were >10 and 0.2 nM,
`respectively (Fig. 5). thus confirming previous observations. Stud-
`ies by Lichtenstein cf (ii. (I990, I992), utilizing ovarian and breast
`tumor cell
`lines over-expressing HER2 to various degrees, also
`found that resistance to TNF was associated with H3R2/mzu ex-
`pression. The TNF resistance was not due to reduction in TNF
`receptor number or signal-transduction events such as Ml-IC ex-
`pression or ADP-ribose polymerase activity.
`In another study,
`these authors demonstrated that increased HER2/lieu expression
`also correlated with resistance to killing by lymphokine-activated
`killer cells (Fady er ul.. 1993). Therefore. 0Ver—expression of
`HER2 in breast cancer cells appears to provide a distinct growth
`advantage since the transformed phenotype can effectively escape
`growth control by soluble cytokine mediators and by immune
`effector cells empowered to maintain hon] oslasis.
`
`IMMUNOGEN 2101, pg. 7
`Phigenix v. Immunogen
`|PR2014-00676
`
`IMMUNOGEN 2101, pg. 7
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`
`272
`
`ROSENBLUM ET AL.
`
`The scFv23i”TNF construct characterized in otrr sttrdy represents
`a design improvement over our initial chemical construct of a
`frr1l—Eeugth antibody conjugated to recombinant TNF. C