`
`CONSTRUCTION, BACTERIAL EXPRESSION AND
`CHARACTERIZATION OF A BIFUNCTIONAL SINGLE ,
`CHAIN ANTI BODY-PHOSPHATASE FUSION PROTEIN
`TARGETED TO THE HUMAN ERBB-2 RECEPTOR
`Winfried Wels*, ma-Maria Harwerth, Marcus Zwickl Norman Hardman
`Bernd Groner and Nancy E. Hynes
`Friedrich Miescher Institute, P.O. Box 2543, CH-4002 Basel, Switzerland. Department of Molecular Biology, Biotechnology
`Section, CIBAGEIGY AG, CH-4000 Basel, Switzerland. *corresponding author.
`
`We have constructed genes expressing
`single-chain antigen binding proteins
`(scFv) which recognize the human erbB-2
`receptor. These genes encode the heavy
`and light chain variable domains of an
`erbB-2 receptor specific monoclonal anti-
`body, MAb FRP5, connected by a peptide
`linker. In order to express a bifunctional
`molecule, a bacterial alkaline phosphatase
`gene was fused 3’ to the scFv gene. The
`scFv(FRP5) and scFv(FRP5)-alkaline phos.
`phatase fusion protein (scFv(FRP5)-PhoA)
`expressed in E. coli specifically recognize
`the human erbB-2 protein and compete
`
`with MAb FRP5 for binding to the recep-
`tor. The bound scFv(FRP5)-PboA protein
`can be detected directly on tumor cells
`using a substrate for alkaline phosphatase,
`showing that the chimeric protein retains
`both binding and enzymatic activity.
`
`T he erbB2 protein, a member of the receptor
`
`tyrosine kinase family’ 2 , is over-expressed in
`approximately 30% of primary human breast
`and ovarian tumors’"’. Patients whose tumors
`display elevated erbB.2 levels appear to have a worse
`. Therefore the routine immunohistochemi(cid:149)
`prognosis7
`cal detection of erbB.2 in breast and ovarian tumors may
`become desirable in the clinic. We have isolated mono-
`
`FIGURE 1 (A) Scheme of the scFv(FRP5) and scFv(FRP5)-phoA
`expression plasmids. Plasmid pWW522-5 contains the gene for
`the expression of scFv(FRP5) inserted into the plasmid pFLAG.
`1.. The fusion gene consists of the IPTG inducible tac promoter,
`the OmpA signal sequence (OmpA SP), 24 nucleotides encod-
`ing the FLAG epitope (FLAG), the PCR amplified eDNA frag.
`ment of the MAb FRP5 V0, a sequence encoding a 15 amino
`acid connecting linker (linker), and the PCR amplified eDNA
`fragment of the MAb FRP5 V 1 Plasmid pWW617.5 contains the
`fusion gene for the expression of scFv(FRP5).PhoA. The 5’ end
`of the gene is identical with that described for the scFv(FRP5).
`At the 3’ cud of the V, domain an open reading frame was
`created and fused to a gene encoding a modified PhoA protein
`(phoA). (B) Nucleotide sequence and deduced amino acid
`sequence of the scFv(FRP5) gene. The sequence shows: the
`OmpA signal peptide (OmpA SP) from plasmid pFLAG-1 (bp
`1-63); the FLAG epitope (bp 64-87); the PstI/BstEII fragment
`encoding the VH domain (bp 106-444); the synthetic 15 amino
`acid linker (bp 454-498); and the PvuIlIBglII fragment encod-
`ing the V 1 domain (bp 505-817). The complementarity deter.
`mining regions (CDR) in the deduced amino acid sequence of
`the FRP5 heavy and light chain variable domains are under -
`lined. The sequence positions bp 87-105, bp 445-504, and bp
`818-836 are from a synthetic linker sequence designed for the
`suhcloning of variable domain cDNAs. (C) Partial nucleotide
`sequence and deduced amino acid sequence of the scFv(FRP5)-
`PhoA gene. The 5’ sequence of the gene from bp 1-780 is
`identical with that shown in (11). An open reading frame was
`created by BglII/BclI cleavage and religation (bp 812-817). The
`inodifiedE. coliphoA gene was cloned 3’ of the V,. domain at the
`XbaI site. The Pro at amino acid position 256 corresponds to
`position 6 of the processed wild type PhoA. The nucleotides at
`position 895 to 2121, not shown in this sequence, correspond to
`bp 427 to 1653 of thepublished phoA sequence". The XbaI site.
` hp 822-827 and the Sad site at bp 2236-2241 were intro-
`duced for sub-cloning by site directed mutagenesis.
`
`1128 BIQECHNOWGY VOL 10 OCTOBER 1992
`
`A (cid:9)
`
`B (cid:9)
`
`OjvH ..
`
`VH (cid:9)
`
`VL
`
`VL (cid:9)
`
`pwW5402
`
`phi (cid:9)
`
`-
`
`pOMMOl7.6
`
`offipA
`
`at’ (cid:9)
`
`04001144 40
`9449.296004 909,0004490012 604090991 IC)’ (cid:9)
`k6T4040AV6..4C.44i000 (cid:9)-?
`
`.. A0744 (cid:9)
`22022499607, (cid:9)&46000600000040
`6000,40
`’0 0 0 0 0 (cid:9)
`A 0 2 6 0 0 0 0 9 9
`’(AL
`,940090001240001’06.0940070o1’09&66009000000004001 (02
`000 (cid:9)(AAA
`0990 K 000009100642 0900 (cid:9)
`31
`
`(’0(0094 ((0
`0 0 0 0 (cid:9)
`14
`
`($CWC(6IAA’0429000226000)900010901040240X22(’ 912200042(0 040
`60
`O 0979460069090009969 (cid:9)
`
`V
`01400906ACAL0000600094900’CMO.00006009000160(( (96A’04904C0040(( (cid:9)000
`H Lj
`5’ (cid:9) 9$9999Ot49 (cid:9)
`00k.)
`0)600(9191 (’090 40 (cid:9)
`006004A090090’A’ 0100202400600IA0000MA600 (cid:9)
`0200 (cid:9)011461009Q(6Ntk6 (cid:9)
`
`09
`
`061
`95
`
`(OMO.,OA940009404000012 0 0174404 94000440110A(40A904000460110t004C (cid:9)
`o (cid:9)
`9 (cid:9)
`4 (cid:9)1
`(cid:9)4
`0 (cid:9)
`Y (cid:9)
`2 (cid:9) 6 (cid:9)
`.0 (cid:9)
`4 (cid:9)00
`6 (cid:9)
`0 (cid:9) 67 (cid:9)
`0.
`9 (cid:9)
`.2406020
`04009004049090006004009000070’010010’o,0000000110109002909006004 (cid:9)
`01 (cid:9)
`1 (cid:9)
`0 (cid:9)
`6 (cid:9)
`0. (cid:9)
`0 (cid:9)
`0 (cid:9)
`0 (cid:9)
`00 (cid:9)
`9 (cid:9)
`4 (cid:9)
`4 (cid:9)
`0 (cid:9) 0 (cid:9)
`0 (cid:9)
`0 (cid:9) 0_0_22_60... (cid:9)
`
`000000440690’040091000.6109092249,404040094922A100204000401100 (cid:9)
`0 (cid:9)
`09 (cid:9)
`0 (cid:9)
`0 (cid:9) 0 (cid:9) 0 (cid:9)
`1 (cid:9)
`. (cid:9)
`0 (cid:9)
`5 (cid:9)
`0 (cid:9)
`9 (cid:9)
`S
`’0 (cid:9)
`.60290,01 .60.090 (cid:9)
`
`420
`0(6
`
`6020
`039
`
`560
`009
`
`00420,494000010)90006090(0001064020A0002000015101014,60001011007 (cid:9)600
`0006050
`
`
`
`V
`
`00040
`10010t0’A&01402A0004’Y.604610t’M601161904114001001<A000’000010
`4V406O0020P6o.0(00004 (cid:9)093
`00)6.)
`(000000.7602600,0 00,0009119007,0000064060101004000.41’ 0700’190200 (cid:9)
`0 (cid:9)
`1 (cid:9)
`0 (cid:9)
`(0 (cid:9) 000 (cid:9)
`4 (cid:9)
`4 (cid:9)
`6 (cid:9)
`0 (cid:9)
`3 (cid:9)
`3 (cid:9)
`4 (cid:9)
`4 (cid:9) 55 (cid:9)
`0 (cid:9)
`
`061
`
`000
`206
`
`01000601004: 40 (cid:9) 4004 (cid:9) 40002040611061109101050020110(090*00 (cid:9)060
`0 (cid:9)
`4 (cid:9)
`9 (cid:9)
`0 (cid:9)
`0 (cid:9)
`0 (cid:9)
`00 (cid:9) A (cid:9)
`0 (cid:9)
`5 (cid:9)
`0 (cid:9) % (cid:9)
`F90 T
`0 (cid:9)
`1 (cid:9)
`6 (cid:9)
`0504L (cid:9)
`200L
`094100000001900)’ 101020400002,41004060.200 40040001:9,zk0o.)0940.o (cid:9)
`04100.3016021 (cid:9)
`001000
`
`’06
`2009
`
`C
`
`4002
`69600j6C( (cid:9)
`001004070190000.0,.000&049012600090096000025019440509 00090944 (cid:9)
`240490590264909090 (cid:9)
`
`0040
`254
`
`pWA M00,52900501000005000141 1 0909490404o40o,0400947990 0002090,....651
`66000007010POO9446 (cid:9) 1.. (cid:9)
`2/0
`phm
`
`4X500000010’49400.4994020041102J0.0044000410 2(60
`299
`
`590’’1494100L (cid:9)
`60141W4o07960990004#519011( 0’ ( q; 40C94000;0 ’0000102091184041 2200
`
`0200
`
`IMMUNOGEN 2061, pg. 1
`Phigenix v. Immunogen
`IPR2014-00676
`
`(cid:9)
`(cid:9)
`(cid:9)
`(cid:9)
`
`
`' 1992 Nature Publishing Group http://www.nature.com/naturebiotechnology
`
`
`
`A (cid:9)
`
`1 (cid:9)
`
`2 (cid:9)
`
`B (cid:9)
`
`2 (cid:9)
`
`1 (cid:9)
`
`-
`
`C
`
`1 (cid:9)
`
`2
`
`74 kDa
`
`- (cid:9)
`
`2lkDa
`
`HOURE 2 SDS-PAGE analysis of scFv(FRP5) and scFv(FRP5).
`PhoA proteins purified from E. coli lysates. The scFv proteins
`were purified by affinity chromotography using the MAb Ml
`which hinds to the FLAG epitope. The purified proteins were
`analysed on 9% SDS-PAGE. (A) A Coomassie stained gel of the
`(lane 1) soluble proteins of a lysate of E. ccli cells expressing
`plasmidpWW6I7-5; (lane 2) the 74 kD scFv(FRP5)-PhoA pro-
`tein. purified =affinity chromatography. The identity of the
`purfied
`scFv proteins was confirmed by a blotting analysis
`using antiPhoA serum (Panel B) or MI anti-FLAG MAb (Panel
`Q. In both panels the 74 kD scFv(FRI’5)-PhoA protein is shown
`in lane 1 and the 27 kD scFv(FRP5) protein is shown in lane 2.
`The filters were incubated with AP-coupled, goat anti-
`rabbit IgC (B) or AP-coupled, sheep anti-mouse lgG (C) and
`bound antibody was detected with a substrate for alkaline
`phosphatase.
`
`A
`
`123456
`
`(cid:151) _
`
`8
`
`(cid:151) i185
`
`1234567
`
`ii
`
`74 ktla
`
`- 27 IsDa
`
`FIGURE 3 The scFv(FRP5) and scFv(FRPS).PhoA proteins bind
`the erbB.2 protein. Aliquots of SKBR3 cell extract were incu-
`bated with scFv(FRP5) (A and B, lanes 1, 6, 7) or scFv(FRP5)-
`PhoA (A and B, lanes 1, 4, 5, 8) followed by additional
`incubation of the mixtures with either pre-immune serum (A
`and B, lane 1), scFv(FRP5) specific antiserum (A and B, lanes 4
`and 6), PhoA specific antiserum (A and B, lane 8), or 21N erbB-
`2 specific antiserum (A and B, lanes 5 and 7). As controls two
`aliquots of the extract were incubated with 21N antiserum (A
`and B, lane 3) or anti.scFv(FRP5) serum (A and B, lane 2). The
`complexes formed were collected with Protein A sepharose
`and the released proteins were separated by 7.5% SDS-PAGE,
`blotted onto PVDF membranes, and an immunoblotting analy-
`sis was done using the 21N antiserum (A, lanes I to 8), anti
`scFv(FRP5) serum (B, lanes I to 7) or the anti.PhoA serum (B,
`12 I-Protein A.
`lane 8). Bound antibodies were detected with
`The arrows indicate the positions of the 185 kD erhB-2 protein,
`the 27 kD scFv(FRP5) and the 74 kD scFv(FRP5).PhoA.
`
`clonal antibodies (MAbs) which specifically bind the
`extracellular domain of the erbB-2 receptor". cDNA
`encoding one of these MAbs, FRP5, has been used to
`construct single chain antigen binding protein (scFv)"
`genes consisting of the variable domains of the heavy
`chain (VH ) and light chain (V 1 .) joined by a short peptide
`linker. Two genes, one encoding an erbB-2 specific scFv
`protein, and the other an scFv-alkaline phosphatase
`(PhoA) fusion protein, were constructed. Both scFv pro-
`teins, which are produced in bacteria, retain specific
`erbB.2 binding activity, and the scFv(FRP5)-PhoA protein
`also displays alkaline phosphatase (PhoA) enzymatic
`activity. A simple immunohistochemical reaction can be
`used to detect erbB-2 molecules on the surface of tumor
`cells after binding of the chimeric scFv(FRP5)-PhoA
`protein.
`
`RESULTS
`Construction of scFv(FRP5) and scFv(FRP5)-PhoA
`expression plasmids. The FRP5 light chain (V,.) and
`heavy chain (V,) variable region cDNAs were isolated by
`reverse transcription of poly(A) containing RNA and
`subsequent PCR amplification as described". The ampli-
`fied DNA sequences were joined into one open reading
`frame by an oligonucleotide encoding 15 amino acids
`(Gly,Gly,Gly,Gly,Ser),. To construct a gene encoding a pro-
`tein with erbB-2 binding activity and with enzymatic
`activity, the alkaline phosphatase (phoA) gene of E. ccli was
`fused to the 3’ end of the scFv(FRP5) gene. Figure IA
`schematically shows the pWW522-5 plasmid containing
`the scFv(FRP5) gene, and the pWW617-5 plasmid con-
`taining the scFv(FRPS)-PIioA gene. For expression in E.
`ccli both the scFv(FRP5) and the scFv(FRP5).phoA genes
`were inserted into the plasmid pFIAG-1, which contains
`an 1FTG inducible toe promoter". The 5’ coding
`sequences specify the ompA signal peptide, which pro-
`motes secretion of recombinant proteins into the pen-
`plasmic space, followed by a sequence encoding the
`FLAG epitope, which allows the isolation of recombinant
`protein via antibody affinity chromatography. The
`sequence of the scFv(FRP5) coding region is shown in
`Figure lB and, in Figure 1C, a portion of the scFv(FRP5)-
`phoA gene is shown. The amino acid at position 256
`(Pro) corresponds to the sixth amino acid of the proc-
`essed PhoA protein".
`Expression and purification of scFv(FRP5) and
`scFv(FRP5)-PhoA from E. coli. The expression plasmids
`pWW522-5 and pWW6I7.5 were transformed into the
`phoA negativeE. coli strain CCI18". Single colonies were
`grown, scFv expression induced with IFI’G, and bacterial
`lysates were prepared. scFv proteins were purified from
`the soluble fraction of bacterial lysates by immunoaffin-
`ity chromatography using the Ml FLAG affinity column.
`Figure 2A shows the Coomassie blue stained gel of the
`cleared lysate of pWW6175 expressing cells. Following
`passage over the Ml FLAG affinity column the predomi-
`nant protein which bound to the column and was eluted
`by treatment with EIJFA was the 74 kD scFv(FRP5).PhoA
`protein. When the same purification protocol was per-
`formed using pWW522.5 expressing cells, the predomi-
`nant protein visible on the Coomassie stained gel was the
`27 kD scFv(FRPS) protein (data not shown). The identi-
`ties of the 74 kD scFv(FRP5).PhoA protein and the 27 kD
`scFv(FRP5) protein were further confirmed by
`immunoblot analysis with antiserum specific for PhoA
`(Fig. 2B) or the FLAG epitope (Fig. 2C).
`The scFv(FRP5) and scFv(FRP5)-PhoA proteins bind
`the erbB-2 protein. Affinity purified scFv(FRP5) and
`scFv(FRP5).PhoA were tested for their ability to bind the
`erbB-2 protein in solution. SKBR3 cells, a human breast
`
`BIqECl1NOLOGY VOL 10 OCTOBER 1992 1129
`
`IMMUNOGEN 2061, pg. 2
`Phigenix v. Immunogen
`IPR2014-00676
`
`
`
`' 1992 Nature Publishing Group http://www.nature.com/naturebiotechnology
`
`
`
`tumor cell line which expresses approximately 1 x 10’
`molecules of the erbB-2 protein per cell", was used as a
`source of antigen. Aliquots of SKBR3 extract were incu-
`bated with scFv(FRP5) or scFv(FRP5)-PhoA to allow the
`formation of erbB-2-scFv complexes. The mixtures were
`then incubated with either preimmune serum, anti-
`scFv(FRP5) serum, anti-PhoA serum, or 21N anti-erbB-2
`serum". As a control, two aliquots of the SKBR3 extract
`were incubated with 21N serum or anti-scFv(FRP5)
`serum. Immune complexes were collected by treatment
`with Protein A.Sepharose and the precipitated proteins
`were analyzed by SDS-PAGE and immunoblotting. The
`filters were treated with 21N serum, anti-scFv(FRP5)
`serum, or anti-PhoA serum. Bound antibodies were
`detected by incubation with (cid:176)
`51-Protein A. The results are
`shown in Figure 3. Preimmune serum does not precipi-
`tate the erbB-2 protein or the scFv proteins (Fig. 3A and
`B, lanes 1). The anti-scFv(FRP5) serum does not precipi-
`tate the erhB-2 protein from the SKBR3 extract (Fig. 3A,
`lane 2).
`The scFv proteins present in the mixtures were precip-
`itated with either anti-scFv(FRP5) serum (Fig. 3B, lanes 4
`and 6), or as part of a complex with 21N anti-erbB-2
`serum (Fig. 3B, lanes 5 and 7). The precipitation of the
`p74 scFv(FRP5)-PhoA protein from the mixture with
`anti-PhoA serum is shown in Figure 3B lane 8. The p185
`erbB-2 protein was precipitated from the SKBR3 cell
`lysate either with 21N antiserum (Fig. 3A, lanes 3, 5 and
`7) or as part of a complex with scFv(FRP5)-PhoA with
`anti-PhoA serum (Fig. 3A, lane 8). The results show
`that both scFv(FRP5) and scFv(FRP5)-PhoA form
`complexes with the erbB-2 protein which can be
`immunoprecipitated with the 21N serum (Fig. 3B) or
`anti-PhoA serum (Fig. 3A).
`Although the anti-scFv(FRP5) serum efficiently precip-
`itates scFv(FRP5) and scFv(FRP5)-PhoA from the mix-
`tures (Fig. 3B), it fails to bring down complexes formed
`between the scFv proteins and the erbB-2 receptor (Fig.
`3A). This suggests that either the binding of scFv proteins
`to the erbB-2 protein is blocked by anti-scFv(FRP5) serum
`or scFv(FRP5) proteins bound to the erbB-2 protein are
`no longer recognized by the antiserum.
`The scFv(FRP5)-PhoA protein is bifunctional. The
`bifunctional nature of the scFv(FRP5)-PhoA protein was
`demonstrated as follows. Varying amounts of affinity
`purified erbB-2 protein were blotted onto membrane
`strips and subsequently incubated with 0.1 juginal of
`scFv(FRP5)-PhoA (Fig. 4A, Panel 1), MAb FRP5 (Fig. 4A,
`Panel 2) or scFv(FRP5) (Fig. 4A, Panel 3). The membranes
`in Panels 2 and 3 were treated with a second antibody,
`AP-coupled, sheep anti-mouse IgG. The bound MAb and
`scFv proteins were detected using Fast Red as a phospha.
`tase substrate’. Incubation of the scFv(FRP5) treated filt-
`ers did not result in detectable reaction product, whereas
`the bound scFv(FRP5)-PhoA protein retained enzymatic
`activity. Since equal amounts of the monovalent
`scFv(FRP5)-PhoA and the bivalent MAb FRP5 were
`included in each reaction and the intensities of the bands
`in Panels 1 and 2 are equivalent, the scFv(FRP5)-PhoA
`and MAb FRP5 appear, in this assay, to be equally
`sensitive.
`Binding properties of MAb FRP5 and scFv(FRP5)-.
`PhoA. A competition experiment was performed to
`determine if MAb FRP5 and the two scFv proteins bind
`the same domain of the erbB-2 protein. An EUSA assay
`using fixed SKBR3 breast tumor cells was carried out by
`incubating cells with 50 nM scFv(FRP5)-PhoA in the pres-
`ence of increasing amounts of either MAb FRP5 or
`scFv(FRP5). The relative activity of the bound
`scFv(FRP5)-PhoA was determined at 405 nM (Fig. 4B).
`
`3
`
`125 (cid:9)
`
`25
`
`100
`
`80
`
`60
`
`at.
`CL
`o cc
`.40
`(u
`E 00
`
`CD (cid:9)
`
`0
`
`1000
`
`10 (cid:9)
`100 (cid:9)
`concentration competitor nM
`FIGURE 4(A) The scFv(FRP5).PhoA protein detects erhB-2 in an
`immuno-slot-blot analysis. Five, 25 and 125ng of affinity puri-
`fied erbB-2 protein were blotted onto a PVDF membrane using
`a slot blot apparatus. Strips were incubated with 0.1 g/m1: (1)
`scFv(FRP5).PhoA; (2) MAb FRP5; (3) scFv(FRP5). The strips in
`lanes 2 and 3 were then treated with AP-coupled, sheep anti-
`mouse IgG. Bound phosphatase activity was detected with the
`substrate Fast Red. (B) Competition analysis. SKBR3 cells were
`fixed with 2% formaldehyde and incubated with 50nM
`scFv(FRP5)-PhoA and the following competitors: MAb FRP5
`(solid circles) at concentrations ranging from 300 to 4.7nM in
`two-fold dilution steps, or scFv(FRP5) (open circles) at concen-
`trations ranging from 600 to 18.8nM. The relative activity of
`bound scFv(FRP5)-PhoA was determined as the absorbance
`at 405mm and is shown as percentage of controls without
`competitor.
`
`The concentration of MAb FRP5 and scFv(FRP5) needed
`to compete binding of the scFv(FRP5)-PhoA by 50% is,
`respectively, 20 nM and 200 nM. In another ELISA exper-
`iment the apparent affinities of both MAb FRP5 and
`scFv(FRP5)-PhoA were found to be, respectively, 0.82 nM
`and 7.2 nM (data not shown). These results suggest that
`the genetically engineered proteins bind to the same
`domain on the erbB-2 protein as does the MAb. The
`apparent binding affinity of the bacterially produced
`scFv(FRP5).PhoA protein is 9-fold lower than that of the
`original MAb FRP5.
`Immunocytocheniical detection of erbB-2 with
`scFv(F.RP5)-PhoA. SKBR3 cells were fixed in 3.7% for-
`maldehyde and incubated with scFv(FRP5)-PhoA (Fig. 5A
`and B) or with scFv(FRP5) (Fig. 5C and D). Bound
`scFv(FRP5)-PhoA was detected using Fast Red as a sub-
`strate. SKBR3 cells incubated with scFv(FRP5)-PhoA
`showed surface staining of erbB-2 in normal light and by
`immunofluorescence. Cells treated with the control
`scFv(FRP5) protein showed no surface staining.
`
`1130 BIQ4tCHNOLQGY VOL 10 OCTOBER 1992
`
`IMMUNOGEN 2061, pg. 3
`Phigenix v. Immunogen
`IPR2014-00676
`
`
`
`A (cid:9)
`
`' 1992 Nature Publishing Group http://www.nature.com/naturebiotechnology
`B
`
`connecting the two variable domains or to the order of
`these domains in the protein. Nevertheless, the affinity of
`the recombinant scFv(FRP5)-PhoA is sufficiently high to
`allow rapid, specific binding to cells expressing the
`erbB-2 receptor. The bacterially produced scFv(FRP5)-
`PhoA is an example of a new type of immunohistochemi-
`cal reagent that can detect an antigen on cell or tissue
`samples in a one step reaction.
`
`EXPERIMENTAL PROTOCOL
`Cloning of FRP5 V, and VL cDNAs. Poly(A) containing
`RNA isolated from the FRP5 hybridoma cell tine was used for
`first strand cDNA synthesis following standard conditions 22
`with MCK2 primer 5’ TCACTGGATGGTGGGAAGATGGA 3’,
`specific for the kappa constant region, or MCH2 primer 5’
`AGATCCAGGGGCCAGTGGAFAGA 3’, specific for the IgGi
`constant region. VH and V L domain cDNAs were amplified
`using PCR as described". For amplification of the V, domain,
`oligos VHIFOR and VHIBACK were used". Amplification of
`the V L domain was accomplished using oligos MCK2 and
`VKIBACK". The PCR fragments were cloned into Bluescript
`KS+ (Stratagene) and sequenced. This revealed that the V,
`sequence had been amplified due to oligo MCK2 priming in
`both directions. P-suiT and Bglli restriction enzyme sites were
`introduced into the FRP5 V, sequence in a separate PCR reac-
`tion using the oligonucleotides 5’ GACATTCAGCTGACCCAG
`3’ and 5’ GCCCGTTAGATCTCCAATTTTGT[CCCCGAG 3’.
`Assembly of the FRP5 scFv genes. A linker sequence for
`joining the V and V domains was constructed using oligonu-
`cleotides and inserted into a modified Bluescript KS-i-, in
`which the two internal PvuII sites had been destroyed. The
`linker encodes a peptide of 15 amino acids (GGGGS) 320 . The
`FRP5 VL fragment was inserted into the resulting plasmid
`pWWI5 as a PvuII/BglII fragment. The V H fragment was
`inserted 5’ of the VL fragment as a PstIIBstEfl fragment. To
`obtain a scFv(FRP5)-PhoA fusion gene, an open reading frame
`was created at the 3’ end of the V, domain by BgllllBclI cleav-
`age and re-ligation. A Xbal restriction enzyme site was intro-
`duced near the 5’ end and a Sacl restriction enzyme site into
`the 3’ non-coding region of the phoA gene by PCR using a
`pBR322 derivative carrying transposon Tn.phoA 14 as a tem-
`plate and the oligoriucleotides 5’ CCC’ICTAGAGCCTGTTCTG-
`GAAAAC 3’ and 5’ CCCGAGG1TGCCATTAAG 3’. After
`digestion with Xhal and Sad, the 1419 bp phoA fragment was
`introduced 3’ of the V L domain using XbaI and Sacl sites
`present in the cloning linker. Finally, the scFv(FRP5)-PhoA
`gene was isolated as a Hindlll/SacI fragment, and sub-cloned
`into pUCI9 yielding plasmid pWW6I5-5. For the expression in
`E. co/i the scFv(FRP5) and scFv(FRP5) PhoA genes were intro-
`duced into the plasmid pFLAG-1 (IBI Biochemicals) 12 . Inser-
`tion of the scFv(FRPS) gene into pFLAG-1 yielded the
`expression plasmid pWW522-5. The plasmid pWW6I7-5 was
`obtained by insertion of the scFv(FRP5)-PhoA gene into
`pFLAG-1 - In both constructs the scFv genes were fused in frame
`to the OmpA signal peptide sequence and a 24 nucleotide
`sequence encoding the FLAG epitope, a synthetic amino acid
`sequence that allows Ca 2 dependent affinity purification of
`recombinant proteins with the Ml anti-FLAG MAb affinity
`matrix (IBI Biochemicals) 12.
`Expression and purification of scFv(FRP5) and scFv(FRP5)-
`PhoA in E. coil. Plasmids pWW522-5 and pWW6I 7-5 were trans-
`formed into the phoA negative E. CO/i strain CCI18 12 , Single
`colonies were grown overnight at 37(cid:176)C in LB medium contain-
`ing 0.4% glucose and ampicillin (100 ig/ml). The cultures were
`diluted 30-fold in 500m1 of the same medium and grown at
`28 (cid:176) C. At an OD,,, of 0,5, the cultures were induced for 45min at
`28(cid:176)C with 1mM IPTG. Cells were harvested by centrifugation
`and lysed by sonication in 20ml PBS containing 5mM CHAPS
`(Boehringer Mannheim), 1mM CaCl 2, 10% glycerol. The lysates
`were cleared by ultracentrifugation at 45,000g for 45min at
`4(cid:176)C. The supernatants containing soluble scFv proteins were
`collected and applied to a Ml FLAG affinity column" , The
`column was washed three times with 10 bed volumes of PBS
`containing 1mM CHAPS, 1mM CaCI,, 10% glycerol. Bound
`proteins were eluted in lml fractions of PBS containing 1mM
`CHAPS, 5mM EDIA, 10% glycerol. The fractions containing
`scFv(FRP5) and scFv(FRP5)-PhoA were pooled and concen-
`trated by ultrafiltration through a YMIO membrane (Amicon).
`The yield of purified scFv proteins was approximately 400isgI1
`E. coli culture,
`
`C , (cid:9)
`D
`FIGURE 5 The scFv(FRP5)-PhoA protein detects the erbB(cid:149)2 pro-
`tein in an immunocytochemical analysis. SKBR3 cells were
`fixed with 3.7% formaldehyde ajtiricubated with (A and B)
`lsgIml scFv(FRP5)-PhoA; (C and D) 1 lA gIml scFv(FRP5). Bound
`PhoA was detected with the substrate Fast Red, In (A) and (C)
`the cells are shown in phase contrast, in (B) and (D) the
`immunoflourescent image of the respective fields is shown.
`
`DISCUSSION
`Overexpression of the erbB-2 receptor protein in
`breast and ovarian tumors appears to correlate with poor
`patient prognosis 9 . Thus, the erbB-2 receptor has poten-
`tial importance, both as a diagnostic marker and as a
`tumor
`target for cancer therapy. MAbs directed against
`cell proteins such as erbB-2 have numerous applica-
`tions‘-‘ but in some instances the size of the antibody
`and its subunit structure make its manipulation for spe-
`cific purposes difficult. The smallest domain of an anti-
`body necessary for specific binding is the Fv domain.
`Recent technological advances have made the cloning
`and production of scFv molecules routine 1, 19,20. This
`approach also allows the direct fusion of the Fv domain
`of an antibody to genes encoding effector molecules
`and the high level expression of these bifunctional mole-
`cules in bacteria. Recombinant scFv molecules have
`already demonstrated some utility in tumor therapy as
`single chain immunotoxins directed towards cell surface
`expressed proteins" ,".
`Both of the scFv proteins we have produced retain
`their affinities for erbB-2, and the scFv(FRP5)-PhoA
`fusion retains enzymatic activity following binding, The
`scFv proteins were expressed in E. coli using a system
`which employs an amino terminal OmpA signal peptide
`to facilitate the secretion of soluble scFv protein into the
`periplasmic space However, the yield of soluble scFv
`proteins was higher in total bacterial lysates containing
`CHAPS than in periplasmic extracts without the addi-
`tion of a detergent (W W, unpublished results). Approxi-
`mately 30% of the scFv protein could be purified from
`total bacterial lysates in a single affinity purification step
`The remaining recombinant protein, which was insolu-
`ble, appeared to be correctly processed at the signal
`peptide cleavage site, but could only be recovered by
`solubilization under denaturing conditions (W W,
`unpublished results).
`The recombinant scFv(FRP5)-PhoA had an approxi-
`mately 9-fold reduced affinity for the erbB-2 receptor
`when compared with the parental MAb. A reduced affin-
`ity for antigen has been described for other scFv pro-
`teins"-". This may be due to the peptide linker
`
`BIQ’ECHNOIDGY VOL 10 OCTOBER 1992 1131
`
`IMMUNOGEN 2061, pg. 4
`Phigenix v. Immunogen
`IPR2014-00676
`
`
`
`' 1992 Nature Publishing Group http://www.nature.com/naturebiotechnology
`
`Bindinactivity of the scFv and scFv(FRP5)PboA proteins.
`The binding of scFv and scFv(FRP5)-PhoA to the erbB-2 pro.
`tein was tested in extracts of SKBR3 breast tumor cells’-’. Four
`million SKBR3 cells were suspended in Imi of 50mM Tris-HC1
`(pH 7.5), 5mM EGTA, 1% Triton X-100, 0.15M NaCI, and 3mM
`PMSF and particulate matter was removed by centrifugation.
`Fifty jil aliquots of the cell extract were brought to a volume of
`350il with PBS and incubated for I on ice with scFv(FRP5) or
`scFv(FRPS).PhoA at a concentration of 31kIml followed by
`additional incubation of the mixtures with either preimmune
`serum, anti-scFv(FRP5) serum, anti-PhoA serum, or 21N anti-
`erbB-2 s-rum’5. Complexes were collected with Protein A-
`sepharose, proteins were released by boiling in sample buffer,
`separated by SDS.PAGE and electroblotted onto PVDF mem-
`branes (Millipore) 26 . Membranes were incubated with 21N anti-
`serum, anti-PhoA serum, or anti.scFv(FRP5) serum, followed by
`treatment of the filters with 1251-Protein A (Amersham).
`Enzymatic activity of the scFv(FRP5)-PhoA protein. The
`bifunctional nature of the scFv(FRP5)-PhoA protein was tested
`as follows. Five-125ng of immunoaffinity purified erbB-2 pro-
`tein were blotted onto a PVDF membrane using a slot-bloc
`apparatus (BioRad). The membrane was cut into three strips
`which were incubated lh at room temperature with 0.1Mg/mi
`of MAb FRP5, scFv(FRP5) or scFv(FRP5)-PhoA. The first two
`strips were incubated an additional hour with AP-coupled,
`sheep anti-mouse IgG. Bound antibody or scFv fusion protein
`was detected by measuring phosphatase activity for 30mm at
`37(cid:176)C in 100mM Tris-HCI, pH 8.2, 1mM ZnCl 2, 1mg1mi Fast Red
`TR salt (Sigma), and 0.4mg/mI AS-MX phosphate (Sigma).
`Binding affinities. The binding affinities of MAb FRP5 and
`the scFv(FRP5)-PhoA protein to erbB-2 on SKBRS human
`breast tumor cells were measured by ELISA. SKBR3 cells, in 96
`well microtiter plates, were fixed for 30mm with 2% formalde-
`hyde/PBS and non-specific binding was blocked by incubation
`with 3% BSAIPBS. One hundred z1 of MAb FRP5 was added in
`three-fold dilutions at concentrations ranging from 50 to
`0.00814g/mI in 3% BSAIPBS for 2h at 4(cid:176)C. Unbound MAb FRP5
`was removed and the wells incubated lh at room temperature
`with 100sl of AP-coupled, sheep anti-mouse 1gG. Specifically
`bound MAb was detected with an AP substrate by incubation in
`200d of IM Tris-HCL (pH 80), 0.4mg/ml p.nitrophenylphos.
`phate disodium (Sigma). The binding of MAb FRP5 to erbB-2
`was determined by measuring the absorbance at 405nm. For
`the determination of the binding of scFv(FRP5)-PhoA to
`SKBRS cells, 100M1 of scFv(FRP5)-PhoA was added in twofold
`dilution steps at concentrations ranging from 12.8 to O.2Mg!ml
`in 3% BSA/PBS. After 2h incubation at 4(cid:176)C unbound
`scFv(FRP5)-PhoA was removed. The activity of bound and free
`scFv(FRP5)-PhoA was determined as the absorbance at 405nm
`after incubation of supernatants, or washed cells, in a total
`volume of 200tl of 1M Tris-HC1 (pH 8.0), 2mM ZnC1 2 ,
`0.4mg/ml p-nitrophenylphosphate disodium for 30min at
`37(cid:176)C. The amount of bound and free scFv(FRP5)-PhoA was
`determined by comparison of the activities to that of
`scFv(FRP5)-PhoA standards. The binding affinity was deter-
`mined using the LIGAND program 27.
`Competition experiments. The enzymatic activity of
`scFv(FRP5).PhoA was determined by ELISA as described above.
`SKBRS cells were incubated with 50nM scFv(FRP5)-PhoA in the
`presence of MAb FRP5 or scFv(FRP5) as competitor. The con-
`centrations of MAb FRP5 or scFv(FRP5) ranged from, respec-
`tively, 300 to 4.7nM, or 600 to 18.SnM. The relative activity of
`bound scFv(FRP5).PhoA was determined as the absorbance at
`405nm as a percentage of controls without competitor.
`Isnmonocytochemistry. For detection of the erbli-2 protein
`by immunocytochemistry, SKBRS cells were fixed with 3.7%
`formaldehyde in PBS, unspecific binding was blocked with 3%
`IA in
`BSA/PBS, and cells were incubated Ih at 37(cid:176)C with
`of scFv(FRP5)-PhoA or scFv(FRP5). Bound scFv(FRP5)-PhoA
`was detected using the Fast Red substrate solution described
`above. In this case 0.8mglml levamisole was included in the
`buffer to inhibit endogenous cellular phosphatase activity.
`The cells were illuminated under fluorescence and bright field
`conditions"’.
`
`Acknowledgments
`We thank Trinad Chakraborty for providing us with the
`PhoA specific serum and Dirk Moritz for the scFv(FRP5) spe-
`cific antiserum. We also thank Roland Ball, Barbara Marte and
`Giorgio Merlo for their helpful comments on the manuscript.
`
`Received 29 May 1992; accepted 13 July 1992.
`
`References
`1. Coussens, L., Yang-Feng, T. L., Liao Y.-C., Chen, F, Gray, A., McGrath,j.,
`Seeburg, P.11., Libermann, T. A., Schlessinger,J., Franke, U., Levinson, A.
`and Ulirich, A. 1985. Tyrosine kinase receptor with extensive homology
`to EGF receptor shares chromosomal location with nen oncogene
`Science 230:1132-1159,
`2. Yamamoto, T., Ikawa, S., Akiyama, T., Semba, K., Nomura, N., Miyajima,
`N, Saito, T. and Toyoshima, K. 1986. Similarity of protein encoded by the
`human c-erbW2 gene to epidermal growth factor receptor. Nature
`819230-234.
`j., Ulirich, A. and
`S. Slamon, B. J, Clark, G. M., Wong, S. G., Levin, W
`McGuire, W. L. 1987. Human breast cancer: correlation of relapse and
`survival with amplification of the HER-21neu oncogene. Science
`235:177-182.
`4. Van de Vijver, M., van de Bersselaar, P.., Devilee, P., Cornelisse, C., Peterse,
`J. and Nusse, R. 1987. Amplification of the neu(c-erbB-2) oncogene in
`human mammary tumors is relatively frequent and is often accompa-
`nied by amplification of the linked c-erbA oncogene. Mol. Cell. Biol.
`7:2019-2023.
`3. Varley,J. M., Swailow,J. It., Brammer, WJ., Whittaker,J. L. and Walker; R.
`1987. Alterations in either c-erbB-2(nen) or c-myc proto.oncogenes in
`breast carcinomas correlate with poor short-term prognosis. Oncogene
`1:423-430.
`6. Berger, M. S., Locher, C. W., Saurer, S., Gullick, W. J., Waterfield, M. D,
`Groner, B. and Hynes, N. E. 1988. Correlation of c-erbB-2 gene amplifica-
`tion and protein expression in human breast carcinoma with nodal
`status and nuclear grading. Cancer Res. 48:1238-1245.
`7. Slamc,n,