Engineering interchain disulfide bonds into conserved ...

Protein Engineering vol.7 no.5 pp.697-704. 1994

Engineering interchain disulfide bonds into conserved framework regions of Fv fragments: improved biochemical characteristics of recombinant immunotoxins containing disulfide-stabiiized Fv

Yoram Reiter, Ulrich Brinkmann, Keith O.Webber, Sun-Hee Jung, Byungkook Lee and Ira Pastan Laboratory of Molecular Biology, Division of Cancer Biology, Diagnosis and Centers, National Cancer Institute. National Institutes of Health, Building 37, Room 4E16, Bethesda. MD 20892, USA

Introduction Fv fragments of antibodies are heterodimers consisting of the variable domains of heavy (VH) and light (VL) immunoglobulin chains. They are the smallest functional moieties of antibodies required for binding of the antigen (Bird et al, 1988; Huston et al, 1988; Chaudhary et al, 1989; Brinkmann et ai, 1991; Milenic et ai, 1991; Pastan and FitzGerald, 1991; Batra et al., 1992; Yokota et al., 1992). Their small size makes them useful agents in the development of immunotherapeutic and immunodiagnostic applications, like the therapy of cancer when coupled to toxins (immunotoxins) or for tumor imaging, since they should have better tissue and tumor penetration than whole antibodies (Yokota et al., 1992). The heterodimers of whole IgG or Fab fragments are connected by a disulfide bond. However, Fv

Here we describe the construction and properties of an Fv immunotoxin in which the Fv is stabilized by a disulfide bond at alternative position VH111-VL48. An immunotoxin composed of such a disulfide-stabiiized Fv derived from mAb B3, which recognizes a carcinoma related carbohydrate antigen (LeY) and a truncated form of PE is very active towards B3 antigenexpressing carcinoma cells and not active towards antigennegative cells. However it is somewhat less active than the VH44-VL105-stabilized molecule. We also compared the stability of disulfide-stabiiized (ds) Fv and a single chain (sc) Fv immunotoxins; we found the dsFv immunotoxins to be 697

Downloaded from http://peds.oxfordjournals.org/ at National Institutes of Health Library on August 19, 2014

Using molecular modeling technology, we have recently identified two positions in conserved framework regions of antibody Fv fragments (Fvs) that are distant from CDRs, and potentially can be used to make recombinant Fv fragments in which the unstable VH and VL heterodimer is stabilized by an interchain disulfide bond inserted between structurally conserved framework positions. A disulfide bond has been introduced at one of these positions, VH44-V L 105, and shown to stabilize various Fvs that retain full binding and specificity. Recombinant immunotoxins, e.g. B3(dsFv)-PE38KDEL in which this disulfide-stabiiized Fv moiety is connected to a truncated form of Pseudomonas exotoxin (PE; PE38KDEL) which contains the translocation and ADP ribosylation domains, are indistinguishable in binding and specificity from its single-chain immunotoxin counterparts. We have now analyzed the alternative position, (VH111-VL48), predicted by the modeling methodology, for disulfide stabilization of mAb B3(Fv) by producing a recombinant immunotoxin with such disulfide-stabiiized (ds) Fv. This immunotoxin was also very active and retained full specificity to B3 antigen-positive cells. However, it was 2- to 3-fold less active than the VH44-VL105 dsFv-molecule. We also tested various biochemical features of VH44-VL105 and VH111-VL48 dsFv immunotoxins and compared them with the corresponding single-chain immunotoxin. We found the dsFv immunotoxins were more stable in human serum and more resistant to thermal and chemical denaturation than the single chain (sc) Fv immunotoxin. Because dsFv immunotoxins and dsFvs have full activity and specificity and improved stability, they may be more useful than scFv immunotoxins as therapeutic and diagnostic agents. Key words: antibody/dsFv/mAb B3/'Pseudomonas exotoxin

fragments are not connected by a disulfide bridge and thus by themselves are unstable (Glockshuber et al., 1990). Stable Fvs can be produced by making recombinant molecules in which the VH and VL domains are connected by a peptide linker so that the antigen combining site is regenerated in a single protein (Bird etal., 1988; Huston et al., 1988). In many cases, such singlechain Fvs retain the specificity and affinity of the antibody, and have been successfully used for tumor imaging (Milenic et al., 1991) and to make scFv-toxin fusion proteins (recombinant immunotoxins). These are being evaluated as antitumor agents (Chaudhary et al., 1989; Brinkmann et al., 1991; Pastan and FitzGerald, 1991; Batra et al, 1992). Another possible method to stabilize Fv fragments is by introducing disulfide bonds between the VH and VL domains at positions that confer heterodimer stabilization but do not interfere with antigen binding. Many have introduced disulfide bonds in proteins in an attempt to stabilize them. For example, intramolecular disulfide bonds were introduced in dihydrofolate reductase (Villafranca et al, 1983, 1987), T4 lysozyme (Matsumura et al., 1989), subtilisin (Mitchinson and Wells, 1986; Wells and Powers, 1986; Pantoliano et al., 1987; Takagi et al., 1990), ribonuclease H (Kanaya et al., 1991), a j3/a barrel protein (Eder and Wilmanns, 1992), intermolecular disulfide bonds in X repressor (Pabo and Suchanek, 1986; Sauer et al, 1986) and an a-helical coiled-coil (Zhou et al., 1993). Using molecular modeling techniques, we have identified two positions for disulfide bridges in between the framework regions of Fvs that can potentially be used for such disulfide stabilization of the heterodimer. In contrast to a previous description of Fv-stabilizing interchain disulfide bonds (Glockshuber et al., 1990), these positions are in the conserved framework regions, distant from CDRs, and therefore should be generally applicable to many Fvs without affecting antigen binding. The pairs of positions are related to one another by the pseudo-two-fold symmetry that approximately relates the VH and VL domains. We have recently shown that disulfides introduced at one of these positions, VH44-VL105 of B3(Fv), can stabilize three different Fvs that retain full binding and specificity (Brinkmann et al., 1993b; Reiter et al, 1994a,b). These disulfide-stabiiized Fvs were analyzed by themselves as well as in the form of recombinant immunotoxins in which the disulfide-stabiiized Fv moiety is connected to a truncated form of Pseudomonas exotoxin (PE; PE38KDEL) which contains the translocation and ADP ribosylation domains of the toxin (Brinkmann et al., 1993b).

Y.Reiter et al.

significantly more stable towards chemical and thermal denaturation than the scFv immunotoxin. Materials and methods Design of disulfi.de bonds between VH and VL of mAb B3 The positions of disulfides for stabilization of B3(Fv) were identified using a computer-modeled structure of B3(Fv), generated by mutating and energy minimizing the amino acid sequence and structure of McPC603 as described previously (Jung et al., 1994). The assignment of framework regions and CDRs is according to Kabat et al. (1991). The two different positions of B3(dsFv) are called B3(dsFv)[H44-L105], as described previously by Brinkmann et al. (1993b) and B3(dsFv)[Hl 11-L48] whose characterization is described herein.

Production ofdsFv immunotoxins The components of the disulfide-stabilized immunotoxins VJCyslO5]-PE38KDEL, VJCys48]-PE38KDEL, VH[Cys44], V H [Cyslll] or single-chain immunotoxins, were produced in separate E.coli BL21 (XDE3) cultures containing the corresponding expression plasmid (see Figure 2). All recombinant proteins accumulated in cytoplasmic inclusion bodies (IBs) which were purified, solubilized, reduced and diluted into redox shuffling refolding buffer, as described previously (Buchner et al., 1992; Brinkmann et al., 1993b). Properly folded immunotoxins were purified by ion exchange (Q-Sepharose and Mono Q) and size exclusion chromatography (Brinkmann et al., 1991). Cytotoxicity assays Activity of immunotoxins was determined by inhibition of protein synthesis as described previously (Brinkmann et al., 1991). For competition experiments, mAb B3 or mAb HB21 was added 15 min before the addition of the immunotoxins. Stability assays The stability of Fv immunotoxins was determined by incubating them at 10 /ig/ml at 37°C in human serum, PBS or various concentrations of urea. Thermal stability was determined by incubating them at 10 /tg/ml in PBS for 3 h at various 698

Results Construction of plasmids for expression of recombinant disulfide-stabilized immunotoxins The positions for cysteine replacements in the framework region of B3(Fv) that can potentially be used for disulfide stabilization have been identified previously using molecular modeling techniques (Jung etal., 1994). The two possible positions of the interchain engineered disulfides in B3(Fv), H44-L105 and H111-L48 are shown in Figure 1. The parent plasmid for the generation of plasmids for expression of B3 disulfide-stabilized immunotoxins encodes a single-chain immunotoxiji B3(Fv)—PE38KDEL. In this molecule the VH and VL domains of mAb B3 are held together and stabilized by a peptide linker, (Gly4-Ser)3 (see Figure 1A) and fused to the PE38KDEL gene encoding the translocatibn and ADP-ribosylation domains of PE with a KDEL (endoplasmic reticulum retention) sequence at the C-terminus (Chaudhary etal., 1990; Brinkmann et al., 1991). The plasmids for expression of the components of B3(dsFv) immunotoxins were made by site-directed mutagenesis and subcloning (see Materials and methods). An outline of the plasmid construction is shown in Figure 2. B3(dsFv)[H44-L105] immunotoxin is composed from B3(VH)[Cys44] and B3(V L )[Cysl05]-PE38KDEL encoded by separate plasmids pYR38-2 and pULI39 respectively as described (Brinkmann et al., 1993b). The B3(dsFv)[HlllL48] immunotoxin is composed from B3(V H )[Cyslll] and B3(VL)[Cys48]-PE38KDEL encoded by pYR38-3 and pYR38^ respectively. Expression, refolding and purification of disulfide-stabilized immunotoxins Recombinant disulfide-stabilized immunotoxins were made from separate E.coli BL21 (XDE3) cultures containing the corresponding plasmid encoding the components [Cys44]B3(VH) or [Cysl05]B3(VL)-PE38KDEL for B3(dsFv)[H44-L105]PE38KDEL and [Cysl 11]B3(VH) or [Cys48]B3(V L )~ PE38KDEL for B3(dsFv)[Hlll-L48]-PE38KDEL. These cultures were induced by isopropyl-/3-D-thiogalactoside (IPTG), upon which the recombinant proteins accumulated in large amounts within the cell in insoluble inclusion bodies. After cell disruption, the IBs were isolated separately, solubilized, reduced and refolded in a 2:1 (VH:VL—toxin) molar ratio in renaturation buffer that contained redox-shuffling and aggregation-preventing additives. The final total protein concentration in the refolding mixture was 100 pg/ml. Refolding was performed for 24 h and then a final oxidation step was done in which an excess of oxidized glutathione was added to the refolding solution to increase the yield of properly folded functional immunotoxin (Buchner et al, 1992; Brinkmann et al., 1993b). Highly purified dsFv immunotoxins were recovered after refolding by a purification scheme previously established for the dsFv[H44-L105] immunotoxins (Brinkmann et al., 1993b), which includes ion exchange (Q-Sepharose and Mono Q) and size exclusion chromatography (Figure 3). The yield of properly folded B3(dsFv)[H44-L105]PE38KDEL from the renaturation system is considerably higher than the corresponding single-chain immunotoxins (Reiter et al., 1994a). The yield of B3(dsFv)[Hlll-L48]-PE38KDEL was similar to that obtained with the corresponding single-chain immunotoxin (Table I).


Construction of expression plasmids Uracil-containing single-stranded DNA from the F + origin present in our expression plasmids was obtained by cotransfection of Escherichia coli CJ236 (Bio-Rad, Richmond, CA) with M13 helper phage and was used as a template for sitedirected mutagenesis as described (Kunkel, 1985). The sequence of B3(Fv) and the construction of plasmids in which Arg44 of B3(VH) and SerlO5 of B3(VL) are replaced by cysteines, and in which a stop codon followed by an EcoRI site is at the 3' end of the B3(VH) gene has been previously described (Brinkmann et al., 1991, 1993b). To make plasmids for expression of the alternative dsFv, B3(dsFv)[Hlll-L48] the oligonucleotide 5'-TGAGGAGACAGTGACCAGGGTACCACAGCCCCAGTAAGCAAACCA-3' was used to change Glnl 11 of B3(VH) to cysteine and the oligonucleotide 5'-TAGATCAGGAGCTTTGGACACTGGCCAGGTTTCTGCAGGTACCAT-3' to change Ser48 of B3(VL) to cysteine, and 5'-CCGCCACCACCGGATCCGCGAATTCATTAGG AGACAGTGACCAGAGTC-3' to introduce stop codons followed by an EcoKl site at the 3' end of the B3(VH) gene. Restriction sites (,4sp718, Ball and EcoRI) introduced into these oligonucleotides to facilitate identification of mutated clones or subcloning are in bold. Expression plasmids for the components of the B3(dsFv)[Hlll-L48] immunotoxin were made by subcloning as described in Figure 2.

temperatures. Active immunotoxin remaining after incubation was determined by cytotoxicity assays on A431 cells.

Disulfide-stabilized Fv fragment

Comparison of the specific activity ofscFv and dsFv immunotoxins Analysis of the cytotoxic activity of an immunotoxin provides a convenient assessment of the function of the Fv moiety, because the cytotoxicity of PE-derived immunotoxins is mediated by specific binding of the Fv portion to antigen on the target cells. Therefore specificity as well as binding of Fvs can be determined by measuring the specific activity of immunotoxins towards antigen positive and negative cell lines. Specificity is demonstrated by selectively killing only antigen-positive cells; affinity correlates with the concentration of immunotoxin required to kill antigen bearing cells. Thus, analysis of the specific Fv-mediated toxicity of scFv and dsFv molecules enables us to compare directly these Fv derivatives (Brinkmann et al., 1993a). A comparison of the specific activity of the scFv and the two dsFv immunotoxin versions shows that all three proteins bind to and kill the same spectrum of cells. They are highly cytotoxic

Table I. Yield of properly folded active monomeric immunotoxins B3 immunotoxin (PE38KDEL fusion)

Yield (% of input protein)

scFv dsFv[H44-L105] dsFv[HUl-L48]

3 15 3

One liter of fermentation bacterial culture was induced at ODggo „„, = 9 with IPTG for 2 h and reached a final OD of 18. This yields for the components of B3(dsFv) immunotoxins — 4 g of inclusion bodies which contain 350 mg recombinant protein. Inclusion body proteins (100 mg) were refolded in 1 1 of renaturation solution for 24 h as described in Materials and methods. Properly folded immunotoxins were purified by ion exchange (Q-Sepharose and MonoQ) and size exclusion chromatography. Protein concentrations were determined according to Bradford (1976). The yields are percentage of active purified immunotoxin from total protein input in the renaturation solution.

699


Fig. 1. Structure models of the variable region of B3(Fv), B3(scFv), B3(dsFv)[H44-L105] and B3(dsFv)[Hlll-L48]. (A) Putative structure of B3(Fv) obtained by mutating the McPC6O3 sequence to that of B3 and energy minimization (Brinkmann et al., 1993b; Jung et al., 1994). (B) B3(scFv) containing a (Gly4Ser)3 peptide linker between the C-terminus of B3(VH) and the N-terminus of B3(VL). (C) B3(dsFv)[H44-L105] in which Arg44 and SerlO5 were changed to cysteine to form a disulfide bond between VH and V L . (D) B3(dsFv)[Hlll-L48] in which Glnlll and Ser48 were changed to cysteine to form a second disulfide-stabilized Fv. Framework regions are green (VH) and yellow (VL); the CDRs are blue (VH) and violet (VL). The peptide linker (Gly4Ser)3 in B3(scFv) (panel B) is brown. Arg44, SerlO5, Glnlll and Ser48 in B3(Fv) (panel A) and B3(scFv) (panel B) and the disulfide bond between Cys-H44 and Cys-L105 in B3(dsFv)[H44-L105] (panel C) and between Cys-Hlll and Cys-L48 in B3(dsFv)[Hlll-L48] (panel D) are red.

Y.Reiter et al.

£1

Toxin

603 (VL)

I :: I I I I I I I I I I :

: II : III : : : I I

: I

pULI9

| B3 (VH)

GASTRES GVPDRFTGSGSGTDFTLTISSVQAEDLAVYYC QNDHSYPLT FGAGTK

B3 (VL)

KVSNRFS GVPDRFSGSGSGTDFTLKISRVEAEDLGVYYC FQGSHVPFT FGSG --CDR2FR3 ---CDR3-- --FR4-

:

I :l

pULI39

B3 (VL)

pYR38-2

II I

IIII I: IIIII : I

I

DVKLVESGGGLVQPGGSLKLSCATSGFTFS DYYMY WVRQTPEKRLEWVA YISN. .

603 (VH)

NKYTTEYSASVKG RFIVSRDTSQSILYLQMNALRAEDTAIYYCAR NYYGSTWYFDV

IIIII

I::I I I I I I I I I I

B3 (VL)

pYR38-4

PE38KDEL

B3 (VH)

Ndel

R4*4C II III II : III

PE38KDEL

48

B3 (VH)

:

PE38KDEL

B3 (VH)

pYR38-5

PR! -CDR1 FR2 -CDR2EVKLVESGGGLVQPGGSLRLSCATSGFTFS DFYME WVRQPPGKRLEWIA ASRNKG

:IIIIIIIIIIIII IIII : IIIIIIIIIII I: II

[4

B3 (VL)

:

:I

B3(VH) DDSSAAYSDTVKG RFTISRDNARNTLYLQMSRLKSEDTAIYYCAR G.LAWGAWFAY - F R3 CDR3--603

(VH) WGAGTTVTVS

II I I I I I I B3(VH) WGQGTLVTVS FR4--A Q111C

Amp

Fig. 2. Disulfide connections between VH and V L of B3(Fv) and plasmids for the expression of dsFv immunotoxins. (A) Comparison of the variable regions amino acid sequence of mAb B3 and mAb McPC603. *, positions of cysteine replacement in framework region of B3(Fv), Arg44-Cys in VH and SerlO5-Cys in VL to stabilize the Fv at the H44-L105 combination and ", positions Glnlll-Cys in VH and Ser48-Cys in VL to stabilize the B3(Fv) at the H111-L48 combination. The assignment of framework regions 1-4 (FRl-4) and CDR1-3 is according to Kabat et al. (1991). Our data suggest that many other Fvs can be stabilized by these interchain disulfide bonds and the corresponding positions can be identified by simple sequence alignment to this figure. (B) Plasmids for expression of dsFv immunotoxins. pULI9 codes for the scFv immunotoxin B3(Fv)-PE38KDEL (Brinkmann et al., 1991). pYR38-2 and pULI39 encoding B3(VH Cys44) and B3(VL CyslO5)-PE38KDEL are derived from pULI9 by site-directed mutagenesis and subcloning as described (Brinkmann et al., 1993b). The expression plasmid pYR38-4 for [Cyslll]B3(V H ) was made by introducing the Cys VH 111 mutation and subsequent deletion of an Eco91 fragment coding for B 3 ( V L ) - P E 3 8 K D E L . pYR38-5 encoding [Cys 48]B3(VL)-PE38KDEL was constructed by subcloning a [Cys48]VLcontaining Pstl-HMTLl fragment into pULI21 (Brinkmann et al., 1993a), which encodes B3(VL)-PE38KDEL.

to B3 antigen-expressing cells but not to cells that do not bind mAb B3 (Figure 4A and B; Table 2). B3(dsFv)[H44-L105]PE38KDEL had the same activity as the corresponding singlechain immunotoxin B3(Fv)—PE38KDEL. The activity of the new dsFv immunotoxin, B3(dsFv)[Hl 11-L48]-PE38KDEL was consistently 3- to 4-fold lower than the activity of B3(dsFv)[H44-L105]-PE38KDEL or single-chain B3(Fv)-PE38KDEL. The cytotoxicity of both dsFv immunotoxins is competed by mAb B3 but not HB21, a control mAb that binds to the human transferrin receptor present on the same cells (Figure 4C). These findings indicate that the binding region of the two disulfidestabilized Fvs as well as the linker-stabilized molecule retains specificity. Improved stability of dsFv immunotoxins Previous studies indicated that the B3(dsFv)[H44-L105] immunotoxin is more stable than its scFv immunotoxins counterpart in human serum and various buffers (Brinkmann et al., 1993b; Reiter et al., 1994a). Analysis of the newly made dsFv immunotoxin, B3(dsFv)[Hlll-L48]-PE38KDEL also shows an increased stability compared with the linker-stabilized scFv immunotoxin counterpart upon prolonged incubation in PBS and human serum (Figure 6A and B). To test whether the increased stability of dsFv immunotoxins results from reduced aggregation of properly folded purified dsFv molecules after refolding compared with single-chain immunotoxin, we analyzed the molecular form of these molecules by their mobility in size 700

Reduced

Non-reduced

M 1 2 3

M

1 2 3

Fig. 3. Recombinant B3(Fv) immunotoxins. SDS-PAGE of purified B3(dsFv) immunotoxins. M, molecular weight standards from top to bottom: 106, 80, 49, 32, 7 and 18 kDa. Non-reduced: lane 1, B3(scFv)PE38KDEL: lane 2, B3(dsFv)[H44-L105]-PE38KDEL; lane 3, B3(dsFv)[Hlll-L48]-PE38KDEL. Reduced: lane 1, B3(scFv)PE38KDEL; lane 2, B3(dsFv)[H44-L105]-PE38KDEL: lane 3. B3(dsFv)[Hl 11-L48J-PE38KDEL.

exclusion chromatography. Figure 5 shows the chromatographic profiles of scFv and dsFv immunotoxins after incubation in PBS for 1 or 4 h at 37°C. After 1 h, large amounts of scFv immuno-


603 (VL)

EcpRI

Hindlll 105

S48C |105C

603 (VH)

Pstl

II I I I II II III I

B3(VL) DVLMTQSPLSLPVSLGDQASISC RSSQIIVHS.NGNTYLE WYLQKPGQSPKLLIY

I I I I I I I I I I I I I II I I I I I I l : l I I l : l I I I

immunotoxin

J

Ndel FRl CDR1 FR2 DIVMTQSPSSLSVSAGERVTMSC KSSQSLLNSGNQKNFLA WYQQKPGQ.PPKLLIY

dsFv

scFv immunotoxin


B3(Fv)-PE38KDEL B3(dsFv)[H44-L105]-PE38KDEL B3(dsFv)[H111-L48]-PE38KDEL

1000

100-1 8060-

S

g.

[H44-L105] 4020-

.01

1

.1

10

ng/ml

tig/ml

Fig. 4. Specific cytotoxicity of dsFv immunotoxins toward different cell lines. (A) Comparison of cytotoxicity toward A431 cells of B3(Fv) —PE38KDEL and B3(dsFv)[H44-L105]-PE38KDEL or B3(dsFv)[Hlll-L48]-PE38KDEL (B) Cytotoxicity of B3(dsFv)[Hlll-L48]-PE38KDEL toward various cell lines. (C) Competition of cytotoxicity of B3(dsFv)[H44-L105]-PE38KDEL and B3(dsFv)[Hlll-L48]-PE38KDEL on A431 cells by addition of excess mAb B3. Note that addition of equal amounts of control mAb HB21 which bind to a different antigen on the same cells does not compete.

Table II. Cytotoxicity of B3(Fv) immunotoxins towards different cell lines Cell line

A431 MCF-7 LNCaP HUT102

B3 antigen

Cytotoxicity (IClJ0; ng/ml) (PE38KDEL fusion) B3(scFv)

B3(dsFv)[H44-L105]

B3(dsFv)[Hlll-L48]

0.5 0.4 8.0 >1000

0.4 0.4 7.0 >l000

1.2. 0.9 20

Cytotoxicity assays were performed by measuring the incorperation of [3H]leucine into cell proteins as described by Brinkmann et al. (1991). Data are given as IDso values; the concentrations of immunotoxin that cause a 50% inhibition of protein synthesis after a 24 h incubation with immunotoxin. Immunotoxins tested were the single-chain Fv immunotoxin B3(Fv)-PE38KDEL and the disulfide-stabilized Fv immunotoxins B3(dsFv)[H44-L105]-PE38KDEL or B3(dsFv)[Hlll-L48]-PE38KDEL. The level of B3 antigen expression determined by immunofluorescence is marked + + +, + , - , for high, low and no detectable expression, respectively.

toxin are in the form of large aggregates and after 4 h, two aggregate peaks could be detected and only a small amount of monomer was left. As shown in Figure 6A, no activity was detected after 4 h incubation of scFv immunotoxin in PBS at 37°C. In contrast to this, the dsFv immunotoxin analysis demonstrates no aggregates after 1 h and only a small amount of aggregates after 4 h. Figure 6A shows that the dsFv immunotoxin retains almost full activity after 4 h in PBS. These results demonstrate that properly folded dsFv immunotoxin molecules have a reduced tendency to aggregate after refolding compared with single-chain immunotoxins. This accounts to a large extent for the enhanced

stability of dsFv immunotoxin in various buffers and probably also in human serum. In addition, we compared the stability of the scFv and both dsFv immunotoxins towards thermal denaturation and denaturation with urea. As shown in Figure 6C, both dsFv immunotoxins are more stable towards denaturation than the scFv immunotoxin. Irreversible thermal inactivation of scFv immunotoxin is observed at 37°C, and at 42°C almost no activity remains. This denaturation is due to inactivation of the Fv portion of the protein because the toxin itself is not irreversibly denatured even at 50°C. In contrast, the dsFv immunotoxins retain full activity at temperatures up to 42 °C and retain still significant activity at 45 °C. When analyzing for denaturation by urea, the 701


o

1000

Y.Reiter et al.

dsFv-immunotoxin

scFv-immunotoxin 8 -

8 -

4 •

•

i

•

i

•

r « i • i • i • i

i

ffi

o c

1hr

Io

8 •

m

JD CO

4 •

4hr 8

4•

8

12

16

20

Elution volume (ml)

8

12

16

20

24

Elution volume (ml)

Fig. 5. Reduced aggregation of B3(dsFv) immunotoxin. B3(Fv)-PE38KDEL and B3(dsFv)[H44-L105]-PE38KDEL were incubated at 100 /ig/ml at 37°C in PBS for 1 or 4 h. The molecular form of these immunotoxins was then analyzed by size exclusion chromatography on a TSK 3000 column. The monomer peak elutes at 18-19 ml. The large aggregates elute at 6 ml which corresponds to the void volume of the column. Smaller aggregates of scFv immunotoxin, B3(Fv)-PE38KDEL, elute at 13 ml.

scFv immunotoxin is irreversibly denatured by exposure to 0.5 M urea, while the dsFv immunotoxins tolerate up to 6 M urea (Figure 6D). These results demonstrate the improved stability of disulfide-stabilized immunotoxins compared with the linkerstabilized molecule. Discussion Using molecular modeling techniques, we have recently identified two possible positions in the conserved framework regions of Fvs that potentially can be used to make disulfide-stabilized Fv immunotoxins. One potential site, [ V J ^ - V L I O S ] was recently used by us to produce a dsFv immunotoxin that retained full activity in vitro and in vivo (Brinkmann et al., 1993b, Reiter etal., 1994b). We have demonstrated in this study that the second possible position identified by the computer analysis in the framework region of the B3(Fv) can also be used for disulfide stabilization of the Fv. The newly made dsFv immunotoxin, B3(dsFv)[Hl 11L48J-PE38KDEL still retained high cytotoxic activity toward antigen bearing cells but was somewhat less active than the original dsFv immunotoxin. We also demonstrate that both forms of dsFv immunotoxins are more stable in human serum and more resistant to thermal and chemical denaturation compared with the corresponding scFv immunotoxin. Thus, dsFv immunotoxins have improved biochemical characteristics that make them better agents than scFv immunotoxins. 702

Disulfide-stabilized Fv immunotoxins Until recently, the only general method to produce stable recombinant Fv fragments was to use a peptide linker to connect covalently the VH and VL chains (Bird et al., 1988; Huston et al., 1988). This stabilization method often results in Fvs with good or reasonable antigen binding, however, sometimes such Fvs fail to bind properly, perhaps because the linker peptide interferes with binding or is unable to stabilize the Fv sufficiently. Furthermore, many active scFv molecules are unstable, aggregate easily and their production is difficult with relatively low yields. An alternative way to generally stabilize Fv molecules is to introduce by site-directed mutagenesis interchain disulfide bonds that connect VH and VL. Such disulfide bonds need to meet three criteria, (i) The disulfide bond should connect amino acids in structurally conserved framework regions of VH and VL so that the disulfide stabilization will be applicable for Fvs with unknown 3-D structure, (ii) The distance between VH and VL must be small enough to enable the formation of a disulfide bond without generating strains on the Fv structure, (iii) The disulfide bond should be at a sufficient distance from the CDRs to avoid interfering with antigen binding. Using molecular modeling techniques, two potential disulfide bridge sites were identified in B3(Fv), which meet these criteria. One potential site was to replace Arg44 in VH and SerlO5 in VL by cysteines (Jung et al., 1994). We have shown recently that by using these positions, disulfide-stabilized Fv immunotoxins that retain full cytotoxic and binding activity could be obtained (Brinkmann et al., 1993b;


2 v


20

10

o n n

DC

30

20

40

40

60

80

Time(hrs) in human serum

8060"

dsFv[H44-L105] dsFv[H111-L48] scFv PE

40 20-

25

PE scFv dsFv[H44-L10S] dsFv[H111-L48]

30

35

40

45

Temperature (°C)

50

0

Molar Urea

Fig. 6. Stability of B3(dsFv) immunotoxins. Immunotoxins were incubated at 10 ^g/ml and lmmunotoxin remaining after different lengths of incubation was determined by cytotoxicity assays on A431 cells. (A) Stability of B3(Fv)-PE38KDEL, B3(dsFv)[H44-L105]-PE38KDEL and B3(dsFv)[Hlll-L48]PE38KDEL at 37°C in PBS. (B) Stability of B3(Fv)-PE38KDEL, B3(dsFv)[H44-L105]-PE38KDEL and B3(dsFv)[Hlll-L48]-PE38KDEL in human serum at 37°C. (C) ThermostabUity of B3(Fv)-PE38KDEL, B3(dsFv)[H44-L105]-PE38KDEL, B3(dsFv)[Hlll-L48]-PE38KDEL and PE in PBS for 3 h at various temperatures. (D) Urea stability of B3(Fv) and B3(dsFv) immunotoxins and PE incubated for 3 h at 10°C in PBS containing various concentrations of urea.

Reiter et al., 1994a). To determine whether stabilization of Fvs by disulfides at these positions is generally applicable, we made and analyzed three different dsFv-containing immunotoxins (Reiter et al., 1994a). All three dsFv immunotoxins were active in vitro and in vivo and in one case the dsFv immunotoxin was more active in vitro and in vivo than its single-chain counterpart (Y.Reiter, U.Brinkmann, S.-h.Jung, B.K.Lee, P.G.Kusprzyk, C.R.King and I.Pastan, J. Biol. Oiem., submitted). Furthermore, all three disulfide-stabilized Fv immunotoxins were more stable and produced in better yields than the corresponding single-chain Fv immunotoxins. We also have produced a disulfide-stabilized Fv fragment of the anti-Tac antibody and found it binds to antigen with the same affinity as the scFv (K.O.Webber and I.Pastan, manuscript in preparation). Alternative dsFv immunotoxin, B3(dsFv)[Hl 11-L48]— PE38KDEL Our computer-assisted analysis suggested a second possible position in the framework region of the Fv that might be used for disulfide stabilization of Fvs. This position was at the 'opposite side' of the molecule and involves changing Glnl 11 in VH and Ser48 in VL to cysteines. These two alternative positions are related to each other by the pseudo-symmetry axis that approximately relates the VH and VL. In each case, one of the residues involved in the putative disulfide bond (VH position 111 and VL position 105) is flanked on each side by a highly conserved glycine residue that can absorb local distortions to the structure potentially introduced by the engineered disulfide bond.

This work shows that disulfide stabilization of B3 Fv using the position VH111-VL48 resulted in a functional dsFv immunotoxin that retained high specific activity toward antigen bearing cells. Compared with B3(dsFv)[H44-L105], however, this new dsFv immunotoxin was 2- to 3-fold less active. This difference in activity appears to reflect differences in binding affinity. The binding of B3(dsFv)[Hlll-L48]-PE38KDEL was reduced compared with B3(dsFv)[H44-L105]-PE38KDEL (data not shown). Nevertheless, the specificity of both dsFvs was retained. Also the yield of the dsFv[Hlll-L48] was lower than the B3(dsFv)[H44-L105] immunotoxin. Thus it appears that although disulfide stabilization of B3(Fv) is possible at both positions identified by computer modeling, the composition V J ^ - V L I O S is favored over the VH111-VL48 combination. Further analysis using other Fvs will need to be done to determine whether preference for VH44-V L 105 disulfides in B3(dsFv) is specific for B3 and differs in other Fvs, or is a general finding. Improved stability of dsFv immunotoxins Characterization of both dsFv immunotoxins showed enhanced stability upon prolonged incubation in buffers and in human serum compared with the single-chain immunotoxin. The enhanced stability of dsFv immunotoxins results from reduced aggregation of properly folded dsFv molecules after refolding compared with single-chain immunotoxins. In addition, comparing the stability of the two dsFv and scFv immunotoxins with thermal and chemical denaturation we found that the thermostability of dsFv immunotoxins is increased compared with the scFv immunotoxin 703


• 0)

20

Time(hrs) in PBS

Y.Reiter et al.

Acknowledgements We thank E.Lovelace and A.Harris for cell culture assistance, Joe Cammisa for support with computer graphics, and A.Jackson and J.Evans for editorial assistance. Yoram Reiter is supported by a grant from the Rothschild Foundation.

References Batra,J.K., Kasprzyk,P.G., Bird,R.E., PastanJ. and King,R.C. (1992) Proc. NatlAcad. Sci. USA, 89, 5867-5871. Bird.R.E., Hardman,K.D., JacobsonJ.W., Johnson,S., Kaufrnan,B.M., Lee,S.M., Lee,T., Pope.S.H., Riordan,G.S. and Withlow.M. (1988) Science, 242,

423-426. Bradford,M.M. (1976) Anal. Biochem., 72, 848-854. Brinkmann.U., Pai.L.H., FitzGerald,D.J., Willingham,M. and PastanJ. (1991) Proc. NatlAcad. Sci. USA, 88, 8616-8620. Brinkmann.U., Lee,B. and PastanJ. (1993a) J. Immunol., 150, 2774-2782. Brinkmann,U., Reiter,Y., Jung,S.-H., Lee,B. and Pastan,I. (1993b) Proc. Nail Acad. Sci. USA, 90, 7538-7545. Buchner.J., Pastan,I. andBrinkmann,U. (1992) Anal. Biochem., 205, 263-270. Chaudhary.V.K., Queen.C, Junghans,R.P., Waldmann,T.A., FitzGerald,D.J. and Pastan.I. (1989) Nature, 339, 394-397. Chaudhary,V.K., Jinno.Y., FitzGerald.D.J. and Pastan.I. (1990) Proc. NatlAcad. Sci. USA, 87, 308-312. Eder,J. and WUmanns.M. (1992) Biochemistry, 31, 4437-4444. Glockshuber.R., Malia.M., Pfitzinger.I. and Pluckthun,A. (1990) Biochemistry. 29, 1362-1367. Huston,J.S., Levinson,D., Mudgett-Hunter,M., Tai,M.S., Novotny,J., Margolies,M.N., Ridge,R.J., Bruccoleri,E.H., Crea,R. and Oppermann.H. (1988) Proc. NatlAcad. Sci. USA, 85, 5879-5883. Jung,S.-H., Lee,B. and PastanJ. (1994) Proteins: Struct. Fund. Genet., in press. Kabat.E.A., Wu.T.T., Perry.H.M., Gottesman.K.S. and Foeller,C. (1991) US Department of Health and Human Services, Public Health Service, National Institutes of Health. NIH publication No. 91-3242. Kanaya,S., Katsuda,C, Kimura.S., Nakai,T., Kitakuni,E., Nakamura,H., Katayanagi.K., Morikawa,K. and Ikehara,M. (1991) J. Biol. Chem.. 266, 6038-6044. Kunkel,T.A. (1985) Proc. NatlAcad. Sci. USA, 82, 488-492. Matsumura,M., Becktel.W.J., Levitt,M. and Matthews,B.W. (1989) Proc. Natl Acad. Sci. USA, 86, 6562-6566. Milenic,D.E., Yokota,T., Filpula,D.R., Finkelman.M.A.J.. Dodd,S.W..

704

Wood,J.F., Withlow,M., Snoy.P. and Schlom.J. (1991) Cancer Res., 51, 6363-6371. Mitchinson,C and Wells.J A. (1986) Biochemistry, 28, 4807-4815. Pabo.C.O. and Suchanek,E.G. (1986) Biochemistry, 25. 5987-5991. Pantoliano,M.M., Ladner,R.C, Bryan,P.N., RoIlence.M.L., Wood,J.E. and Powlos.T.L. (1987) Biochemistry, 26, 2077-2082. PastanJ. and FitzGerald.D. (1991) Science, 254, 1173-1177. Reiter.Y., Brinkmann.U., Kreitman.R.J., Jung,S.-H., Lee.B and Pastan,I. (1994a) Biochemistry, in press. Reiter.Y., Pai,L., Brinkmann,U., Wang,Q.-C. and Pastan.I. (1994b) Cancer Res., in press. Sauer,R.T., Hehir.K., Stearman.R.S., Weiss,M.A., Jeitler-Nilsson,A.. Suchanek,E , G. and Pabo.C.O. (1986) Biochemistry, 25, 5992-5998. Takagi,H., Takahashi,T., Momose.H., lnonye,M., Maeda,Y., Matsuzawa,H. and OhtaJ. (1990) J. Biol. Chem., 265, 6874-6878. Villafranca.J.E., Howell,E.E., Voet.D.H., Strobel.M.S., Ogden,R.C, AbelsonJ., N. and Kraut.J. (1983) Science, 222, 782-788. VillafrancaJ.E , Howell.E.E., Oatley,S.J., Xuang,N.-H. and Kraut.J. (1987) Biochemistry, 26, 2182-2189. Wells,J.A. and Powers,D.B. (1986) J. Biol. Chem., 261, 6564-6570. Yokota.T., Milenic.D.E., Withlow.M. and Schlom.J. (1992) Cancer Res., 52, 3402-3408. Zhou.N.E., Kay.C.M. and Hodges,R.S. (1993) Biochemistry, 32, 3178-3187. Received November 11, 1993; revised February 14, 1994; accepted February 17, 1994


counterpart. Its stability towards denaturation by urea is also increased. dsFv immunotoxins are resistant to 6 M urea, while the scFv immunotoxin denatures irreversibly at 0.5 M urea. It is likely that the dsFv immunotoxins unfold in 6 M urea but this is reversible. After incubation in urea, the immunotoxins are diluted into incubation medium during the cytotoxicity assay. Under these conditions, the dsFv immunotoxins probably refold efficiently, since the disulfides are still in place and thereby regain full specific activity. The increased stability to thermal and chemical denaturation is due to the improved characteristics of the disulfide-stabilized Fv compared with the linker-stabilized Fv fragment because the toxin moiety of the immunotoxin itself is stable at high temperatures (50 °C) and also resistant to irreversible denaturation at high concentrations of urea. The thermostability of dsFv compared with scFv immunotoxins is an important factor in clinical applications, since inactivation of scFv immunotoxin can occur at a physiological temperature (37 °C), while dsFv immunotoxins are completely stable and active at 37°C. An important question is whether disulfide-stabilized immunotoxins are stable at or near the target tumor cells, in vivo, in a potentially reducing environment (oxygen limitation). All three disulfide-stabilized Fv immunotoxins produced so far are active in vivo in antitumor experiments. Their antitumor activity was either the same as the activity of the corresponding singlechain immunotoxin (Reiter et al., 1994b; Y.Reiter, R.J.Kreitman, U.Brinkmann and I.Pastan, Int. J. Cancer, submitted) or in one case, more active than the single-chain immunotoxin (Y.Reiter, U.Brinkmann, S.-h.Jung, B.K.Lee, P.G.Kasprzyk, C.R.King and I.Pastan, J. Biol. Chem., submitted).