`Vol. 85, pp. 5879-5883, August 1988
`Biochemistry
`
`Protein engineering of antibody binding sites: Recovery of specific
`activity in an anti-digoxin single-chain Fv analogue produced in
`Escherichia coli
`
`(biosynthetic analogue/ anti-digoxin antibodies/variable regions/ gene synthesis/ Escherichia coli expression)
`
`JAMES S. HUSTON*t, DOUGLAS LEVINSON*, MEREDITH MUDGETT-HUNTER*, MEI-SHENG TAt*,
`Jn'H NOVOTNY*, MICHAEL N. MARGOLIES§, RICHARD J. RIDGE*, ROBERT E. BRUCCOLERI*,
`EDGAR HABER*, ROBERTO CREA*, AND HERMANN OPPERMANN*
`
`*Creative Biomolecules, 35 South Street, Hopkinton, MA 01748; and *Molecular and Cellular Research Laboratory. and §Department of Surgery,
`Massachusetts General Hospital and Harvard Medical School, Boston, MA 02114
`
`Communicated by Michael Se/a, May 6, /988 (received for review March 9, /988)
`
`A biosynthetic antibody binding site, which
`ABSTRACT
`incorporated the variable domains of anti-digoxin monoclonal
`antibody 26-10 in a single polypeptide chain (M. = 26,354), was
`produced in Escherichia coli by protein engineering. This
`variable region fragment (Fv) analogue comprised the 26-10
`heavy- and light-chain variable regions (V8 and VL) connected
`by a 15-amino acid linker to form a single-chain Fv (sFv). The
`sFv was designed as a prolyi-V8 -(Iinker)-VL sequence of 248
`amino acids. A 744-base-pair DNA sequence corresponding to
`this sFv protein was derived by using an E. coli codon prefer(cid:173)
`ence, and the sFv gene was assembled starting from synthetic
`oligonucleotides. The sFv polypeptide was expressed as a fusion
`protein in E. coli, using a leader derived from the trp LE
`sequence. The sFv protein was obtained by acid cleavage of the
`unique Asp-Pro peptide bond engineered at the junction of
`leader and sFv in the fusion protein [(leader)-Asp-Pro-V8 -
`(Iinker)-V d- After isolation and renaturation, folded sFv dis(cid:173)
`played specificity for digoxin and related cardiac glycosides
`similar to that of natura126-10 Fab fragments. Binding between
`afl"mity-purified sFv and digoxin exhibited an association con(cid:173)
`stant [K. = (3.2 ± 0.9) x 107 M - 1] that was about a factor of
`6 smaller than that found for 26-10 Fab fragments [K. = (1.9 ±
`0.2) x lOS M- 1] under the same buffer conditions, consisting of
`0.01 M sodium acetate, pH 5.5/0.25 M urea.
`
`It is known that antigen binding fragments of antibodies (1, 2)
`can be refolded from denatured states with recovery of their
`specific binding activity (3-6). The smallest such fragment that
`contains a complete binding site is termed Fv, consisting of an
`M. 25,000 heterodimer of the VH and VL domains (2, 5-11).
`Givol and coworkers were the first to prepare an Fv by peptic
`digestion of murine IgA myeloma MOPC 315 (2). However,
`subsequent development of general cleavage procedures for
`Fv isolation has met with limited success (7-11). As a result,
`the M. 50,000 Fab (1) has remained the only monovalent
`binding fragment used routinely in biomedical applications.
`An Fv analogue was constructed in which both heavy- and
`light -chain variable domains (V H and V L) were part of a
`single polypeptide chain. Synthetic genes for the 26-10
`anti-digoxin V H and V L regions were designed to permit their
`connection through a linker segment, as well as other
`manipulations (12, 13). The synthetic gene for single-chain Fv
`(sFv) was expressed in Escherichia coli as a fusion protein,
`from which the sFv protein was isolated.~ The sFv was
`renatured with recovery of binding specificity and affinity
`similar to those of the parent molecule. Thus, variable
`
`The publication costs of this article were defrayed in part by page charge
`payment. This article must therefore be hereby marked "advertisement"
`in accordance with 18 U .S.C. §1734 solely to indicate this fact.
`
`domains connected artificially to form one polypeptide chain
`can be renatured into properly folded Fv regions.
`
`MATERIALS AND METHODS
`Model Antibody. The digoxin binding site of the lgG2a,K
`monoclonal antibody 26-10 has been analyzed by Mudgett(cid:173)
`Hunter and colleagues (14-16). The 26-10 V region sequences
`were determined from both protein sequencing (17) and DNA
`sequencing of 26-10 H- and L-chain mRNA transcripts (D.
`Panka, J.N., and M.N.M., unpublished data). The 26-10
`antibody exhibits a high digoxin binding affinity (Ka = 5.4 x
`109 M - 1) (14) and has a well-defined specificity profile (15),
`providing a baseline for comparison with the biosynthetic sFv.
`Protein Design. X-ray coordinates for Fab fragments (18-
`20) were obtained from the Brookhaven Data Bank (21) and
`analyzed with the programs CHARMM (22), CONGEN (23),
`and FRODO (24). X-ray data indicated that the Euclidean
`distance between the C terminus of the V H domain and the
`N terminus of the V L domain was =3.5 nm. A 15-residue
`linker should bridge this gap, since the peptide unit length is
`=0.38 nm. The linker should not exhibit a propensity for
`ordered secondary structure or any tendency to interfere with
`domain folding. Thus, the 15-residue sequence (Gly-Gly-Gly(cid:173)
`Gly-Serh was selected to connect the V H carboxyl and V L
`amino termini (Fig. 1).
`Gene Synthesis. Design of the 744-base sequence for the
`synthetic sFv gene was derived from the sFv protein se(cid:173)
`quence by choosing codons preferred by E. coli (25). Syn(cid:173)
`thetic genes encoding the trp promoter-operator, the modi(cid:173)
`fied trp LE leader peptide (MLE), and V H were prepared
`largely as described (26). The gene encoding V H was assem(cid:173)
`bled from 46 overlapping synthetic 15-base oligonucleotides.
`The V L gene was derived from 12 synthetic polynucleotides
`ranging in size from 33 to 88 base pairs, prepared in
`automated DNA synthesizers (model 6500, Biosearch, San
`Rafael, CA; model 380A, Applied Biosystems, Foster City,
`CA). They spanned major restriction sites (Aat II, BstEII,
`Kpn I, Hindiii, Bgl I, and Pst 1), and several fragments were
`flanked by EcoRI and BamHI cloning ends. All segments
`were cloned and assembled in pUC vectors. The linker
`between V H and V L• encoding (Gly-Gly-Gly-Gly-Serh, was
`cloned from two polynucleotides spanning Sac I and Aat II
`sites. The complete sFv gene was assembled from the V H•
`V L• and linker genes to yield a single sFv gene, corresponding
`
`Abbreviations: Fv, variable region fragment (V H V L dimer); MLE,
`modified trp LE leader sequence; sFv, single-chain Fv; V H• heavy(cid:173)
`chain variable region; V L• light-chain variable region.
`tTo whom reprint requests should be addressed.
`~The sequence reported in this paper is being deposited in the
`EMBL/GenBank data base (lntelliGenetics, Mountain View, CA,
`and Eur. Mol. Bioi. Lab., Heidelberg) (accession no. J03850).
`
`5879
`
`BEQ 1030
`Page 1
`
`
`
`5880
`
`Biochemistry: Huston et al.
`
`Proc. Nat/. Acad. Sci. USA 85 (1988)
`
`pBR322 vector, was constructed by a three-part ligation using
`the sites Ssp I, EcoRI, and Pst I (Fig. 2D). Intermediate DNA
`fragments and assembled genes were sequenced by the
`dideoxy chain-termination method (28).
`Fusion Protein Expression. Single-chain Fv was expressed
`as a fusion protein (Fig. 2) with the MLE leader gene (29) and
`under the control of a synthetic trp promoter-operator (29).
`E. coli strain JM83 was transformed with the expression
`plasmid and protein expression was induced in M9 minimal
`medium by addition of indoleacrylic acid (10 ~g/ml) at a cell
`density with A 600 = 1. The high expression levels of the
`fusion protein resulted in its accumulation as insoluble
`protein granules, which were harvested from cell paste (Fig.
`3, lane 1).
`Fusion Protein Cleavage. The MLE leader was removed
`from sFv by acid cleavage of the Asp-Pro peptide bond (32-
`34) engineered at the junction of the MLE and sFv sequences.
`The washed protein granules containing the fusion protein
`were cleaved in 6 M guanidine hydrochloride/10% acetic
`acid, pH 2.5, incubated at 37°C for 96 hr. The reaction was
`stopped by ethanol precipitation, and the precipitate was
`stored at - 20°C (Fig. 3, lane 2).
`Protein Puriftcation. The acid-cleaved sFv was separated
`from remaining intact MLE-sFv species by chromatography
`on DEAE-cellulose. The precipitated cleavage mixture was
`redissolved in 6 M guanidine hydrochloride/0.2 M Tris·HCl,
`pH 8.2/0.1 M 2-mercaptoethanol, and dialyzed exhaustively
`against column buffer (6 M urea/2.5 mM Tris·HCl, pH 7.5/1
`mM EDTA), made 0.1 Min 2-mercaptoethanol, and chro(cid:173)
`matographed on a column (2.5 x 45 em) of Whatman DE 52.
`Elution of the intact fusion protein was retarded relative to
`sFv during the column buffer wash, with leading fractions
`
`FIG. 1. Computer-generated views of an sFv model. These
`pictures display the sFv based on the x-ray structure of the Fab from
`murine IgA myeloma McPC 603 (20). One possible conformation of
`the linker (Giy-Giy-Giy-Giy-Serh is shown connecting the C termi(cid:173)
`nus of the V H domain and theN terminus of the V L domain. (A) View
`showing the linker and binding site of McPC 603. (B) View of
`opposite side showing linker and free terminal residues (Asp-1 of V H
`and Lys-113 of V J; this orientation was obtained by rotating the
`molecule in (A) 180° about a vertical axis. Color coding: gray, V H
`domain; white, V L domain; salmon, linker and free terminal residues;
`other colors, complementarity determining region (CDR) segments
`[lavender (H1), green (H2), orange (H3), blue (Ll), red (L2), yellow
`(L3)]; purple, side chains from CDRs that directly contact the
`phosphorylcholine hapten in McPC 603.
`
`to aspartyl-prolyl-V w(linker)-V L• flanked by EcoRI and Pst
`I restriction sites (Fig. 2). The trp promoter-operator, starting
`from the Ssp I site, and the MLE leader gene, ending in the
`EcoRI site, were assembled from 36 overlapping 15-base
`oligomers. The final expression plasmid, based on the
`
`A
`4 0
`30
`20
`10
`Met Lys Ala Ile Phe Val Leu Lya Gly Ser Leu Asp Arg Asp Leu Asp Ser Arg Leu Asp Leu Asp Val Arg Thr Asp His Lys Asp Leu ser Asp His Leu Val Leu Val Asp Leu Ala
`ATG AAA GCA ATT TTC GTA CTG AAA GGT TCA C:TG GAC AGA GAT CTG GAC TCT CGT CTG GAT CTG GAC GTT CGT ACC GAC CAC AAA GAC CTG TCT GAT CAC CTG GTT CTG GTC GAC CTG GCT
`Sall
`Bglii
`
`~
`Arq As n Asp Leu Ala Arg Ile Val Thr Pro Gly ser Arg Tyr Val Ala Asp Leu Glu Phe Asp
`CGT AAC GAC CTG GCT CGT ATC GTT ACT CCC GGG TCT CGT TAC GTT GCG GAT CTG GAA TTC GAT
`Smai
`EcoRI
`
`"
`
`8
`40
`30
`20
`1 0
`Pro Clu Val Cln Leu Cln Cln Ser Cly Pro Clu Leu Val Lys Pro Cly Ala Ser Val Arg Met ser Cys Lys Ser Ser Cly Tyr Ile Phe Thr Aap Pbe Tyr Met Aaa Trp Val Arg Gln
`CCC CAA CTT CAA. CTC CAA. CAC TCT CCT CCT CAA. TTC CTT AAA CCT GCC CCC TCT CTG CCC ATC TCC TGC AAA TCC TCT GGG TAC ATT TTC ACC GAC TTC TAC ATG AAT TGG CTT CGC CAC
`Narl
`Fspl
`BstXI
`so
`70
`60
`80
`Ser His Cly Lys Ser Leu Asp Tyr Ile Gly 'l'yr Ile aer Pro 'l'yr Ber Gly Val Tbr Gly 'l'yr Aaa GlD Lya Phe Lya Gly Lys Ala Thr Leu Thr Val Asp Lys Ser Ser Ser Thr Ala
`TCT CAT CCT AAC TCT CTA CAC TAC ATC CCC TAC ATT TCC CCA TAC TCT GCG GTT ACC GGC TAC AAC CAG AAG TTT AAA GCT AAG CCC ACC CTT ACT GTC CAC AAA TCT TCC TCA ACT GCT
`Xbai
`PflMI
`BstEII
`Oral
`SalT
`
`120
`110
`100
`90
`Tyr Met Glu Leu Arg ser Leu Thr ser Glu Asp Ser Ala Va l Tyr Tyr Cys Ala Gly ler ler Gly AaD Lya Trp Ala Met Aap Tyr Trp Gly Hie Gly Ala Ser Val Thr Val ser Ser
`TAC ATG GAG CTG CCT TCT TTG ACC TCT GAG GAC TCC GCG GTA TAC TAT TCC GCG GGC TCC TCT CGT AA.C AAA TGG GCC ATG GAT TAT TGG GGT CAT GGT GCT AGC GTT ACT GTG AGC TCT
`sacii
`Ncol
`Nhel
`Sacl
`
`160
`150
`140
`130
`Gl y Gly ely Gly Ser Gly Gly Cly Gly Ser Gly Gly Gly Gly ser Asp Val Val Met Thr Gln Thr Pro Leu Ser Leu Pro Val Ser Leu Gly Aap Gln Ala Ser 11• Ser Cya Ar9 ler
`GGT GGC GCT CGC TCG CCC GGT GGT GGG TCC CGT GGC GGC GGA TCT GAC GTC GTA ATG ACC CAG ACT CCG CTC TCT CTG CCC GTT TCT CTG GGT GAC CAG GCT TCT ATT TCT TGC CGC TCT
`Aatll
`BatEil
`
`200
`190
`180
`170
`aer Gla Bar Leu Val Ria Bar AaD Gly Aan Tbr Tyr Leu Aaa Trp Tyr Leu Gln Lys Ala Gly Gln Ser Pro Lya Leu Leu 11• Tyr Lya Val Bar Aaa Arq Pbe ler Gly Val Pro Aap
`TCC CAG TCT CTG GTC CAT TCT AAT GGT AAC ACT TAC CTC AAC TGG TAC CTG CAA AAG GCT GGT CAG TCT CCC AAG CTT CTG ATC TAC AAA GTC TCT AAC CCC TTC TCT CCT GTC CCG GAT
`PflHl
`BstXI
`Kpnl
`Hindlll
`
`240
`23 0
`220
`210
`Arg Phe Ser Gly Ser Gly Ser Gly Thr Asp Phe Thr Leu Lys 11e Ser Arg Val Glu Ala Glu Aap Leu Gly lle Tyr Phe Cys Bar Glll Tbr Tbr Bia Val Pro Pro Tbr Phe Gly Gly
`CGT TTC TCT GGT TCT GCT TCT GGT ACT GAC TTC ACC CTG AAG ATC TCT CGT GTC GAG CCC GAA GAC CTG GGT ATC TAC TTC TGC TCT CAG ACT ACT CAT CTA CCG CCG ACT TTT GGT GGT
`Bglii
`
`Gly Thr Ly s Leu Glu Ile Lys Arg *OC
`GGC ACC AAG CTC GAG ATT AAA CGT TAA CTGCAC
`Xhol
`Hpai
`Pstl
`
`EcoR I
`Ssor \
`
`c
`
`.
`~------ s Fv - - - - - - - - - - - ,
`.
`'
`'
`[ ~
`t
`t
`Pstl
`MLE
`VL
`VH
`Acid Cleavage
`Site
`
`L inker
`
`0
`
`FIG. 2. Sequence of MLE-sFv fusion protein and sFv protein, with corresponding DNA sequence and some ml!,jor restriction sites, and
`design of expression plasmid. (A) Leader peptide. (B) sFv protein. (C) Schematic drawing of the fusion protein. (D) Schematic drawing of
`expression plasmid. The leader and sFv are shown as they appear after acid cleavage of the fusion protein. During construction of the gene,
`fusion partners were joined at the EcoRI site that is shown as part of the leader sequence. The complementarity determining regions of V H and
`V L are boldface and the linker peptide is underlined. The pBR322 plasmid, opened at the unique Ssp I and Pst I sites, was combined in a three-part
`ligation with an Ssp I/ EcoRI fragment bearing the trp promoter-operator and MLE leader and with an EcoRI/ Pst I fragment carrying the sFv
`gene. The resulting expression vector of 4801 base pairs (bp) (D) confers tetracycline resistance on positive transformants.
`
`Aflm
`
`BEQ 1030
`Page 2
`
`
`
`Biochemistry: Huston et a/.
`
`Proc. Nat/. Acad. Sci. USA 85 (1988)
`
`5881
`
`94-
`67-
`':'
`~ 43-
`29-
`'<
`~ 20.1-
`
`14.4- -
`
`2
`
`3
`
`4
`
`5
`
`FIG. 3. Analysis of protein at progressive stages of purification
`by NaDodS04/PAGE on a 15% polyacrylamide gel (30, 31). The
`same ouabain-Sepharose pool of sFv was run in lanes 4 and 5, but the
`gel sample for lane 5 was not reduced, while all others were reduced
`before electrophoresis. TheM, values calculated for sFv (26,354) and
`MLE-sFv (33,203) polypeptides were less than gel migration indi(cid:173)
`cated, as a result of the low mean residue weight for sFv (106) and
`other sources of error in NaDodS04/PAGE experiments (30).
`
`being devoid of MLE-sFv. Rechromatography of impure
`fractions yielded more purified material that was added to the
`DE 52 pool (Fig. 3, lane 3).
`Refolding. The DE 52 pool of sFv in 6 M urea/2.5 mM
`Tris·HCI/1 mM EDT A was adjusted to pH 8 and reduced
`with 0.1 M 2-mercaptoethanol at 37°C for 90 min. This was
`diluted at least 1:100 with 0.01 M sodium acetate (pH 5.5) to
`a concentration below 10 p.,g/ ml and dialyzed at 4°C for 2 days
`against acetate buffer. Under these conditions, sFv remained
`soluble throughout refolding, whereas substitution of0.15 M
`NaCI/0.05 M potassium phosphate, pH 7.0/0.03% NaN3
`(PBSA) caused precipitation of sFv.
`Aff'mity Chromatography. Active sfv was purified by affin(cid:173)
`ity chromatography at 4°C on a ouabain-amine-Sepharose
`column, as described (14), except that all eluants were dis(cid:173)
`solved in 0.01 M sodium acetate (pH 5.5). Bound sFv was
`eluted from the resin with 20 mM ouabain (Fig. 3,lanes 4 and
`5) and dialyzed against 0.01 M acetate buffer. Protein concen(cid:173)
`trations were quantitated by amino acid analysis (35) (Table 1).
`Sequence Analysis of Gene and Protein. The complete sFv
`gene was sequenced in both directions by the dideoxy method
`of Sanger et a/. (28). Automated Edman degradation was
`conducted on intact sFv protein, as well as on two major CNBr
`fragments (residues 108-129 and 140-159) with a model470A
`gas-phase sequencer equipped with a model 120A on-line
`analyzer (Applied Biosystems) (37). The CNBr fragments of
`gel-purified sFv were separated by NaDodS04/PAGE and
`
`Table 1. Estimated yields during purification, normalized to a
`1-liter fermentation
`
`Wet pellet
`weight, g
`12
`2.3
`
`Protein
`weight, mg
`1440*
`480*t
`144tt
`18.1~
`
`Mol %of
`prior step
`
`100
`38
`12.6§
`
`Step
`Cell paste
`MLE-sFv granules
`DE 52 pool
`Active sFv
`*Determined by Lowry analysis (36).
`tDetermined by absorbance measurements.
`~Determined by amino acid analysis.
`§Calculated from the concentration of sFv protein specifically eluted
`from ouabain-Sepharose, compared to DE 52 pool, both measured
`after dialysis against 0.01 M acetate buffer by amino acid analysis.
`Yield is 4. 7% relative to MLE-sFv.
`
`transferred electrophoretically onto an Immobilon membrane
`(Millipore) from which stained bands were cut out and se(cid:173)
`quenced (38).
`Specificity and Affinity Determinations. Specificities of
`anti-digoxin 26-10 Fab and sFv were assessed by radio(cid:173)
`immunoassay (16) (Fig. 4 and Table 2). Equilibrium binding
`measurements utilized immunoprecipitation techniques to
`separate bound and free [3H]digoxin, with association con(cid:173)
`stants calculated from Sips plots (39) and binding isotherms
`(40) (Fig. 5) as well as Scatchard plots (16).
`
`RESULTS
`DNA sequencing of the complete sFv gene and its major
`oligonucleotide fragments indicated that the sFv gene pos(cid:173)
`sessed the intended sequence (Fig. 2), incorporating the V H
`and V L sequences of monoclonal antibody 26-10. The protein
`sequence expected for this sFv gene product was confirmed
`at the amino terminus of intact sFv (residues 1-15 for
`affinity-purified sFv; residues 1-40 for DE 52 pool) and for
`CNBr fragments over internal regions extending from residue
`107 of V H into the linker (sFv residues 108-129) and over V L
`residues 5-24 (sFv residues 140-159). These results suggest
`that purified sFv protein was a faithful expression product of
`the synthetic sFv gene.
`The yields of protein at various stages of isolation are given
`in Table 1, with purity assessed by NaDodS04 /PAGE (Fig.
`3). Affinity chromatography of the renatured DE 52 pool
`provided the final purification step (Fig. 3, lanes 4 and 5) and
`yielded 12.6% of the renatured protein as active sFv (Table
`1). Successful affinity purification of very dilute renatured
`sFv suggests that monomeric protein was adsorbed to resin.
`Thus, all of the sFv bound to ouabain-Sepharose may be
`considered to have been active, with stoichiometric binding
`capacity for digoxin.
`The sFv specificity profile was reproduced for samples in
`various buffers at different stages of purification, appearing
`
`lOO
`
`26-10sFv
`-
`--- 26-10 Fob
`o Digoxin
`1:>. Acetyl Strophonthidin
`• Ouabain
`
`~ 80 ,
`
`~soL
`~ I
`
`~
`
`0
`
`/ , /
`
`/
`
`,
`
`/ , ,
`,
`' I
`
`I
`I
`I
`I
`I
`I
`I
`I
`I
`I
`
`I '
`
`INHIBITOR CONCENTRAT/ON( M}
`
`FIG. 4. Specificity profiles for sFv and 26-10 Fab species.
`Microliter plates were coated first with affinity-purified goat anti(cid:173)
`mouse Fab antibody, followed by sFv or 26-10 Fab in 1% horse
`serum/0.01 M sodium acetate, pH 5.5. Then 1251-labeled digoxin
`(50,000 cpm) having specific activity of 1800 p.Ci/ p.g (Cambridge
`Diagnostics; 1 Ci = 37 GBq) was added in the presence of a series
`of glycoside concentrations. The inhibition of radioligand binding by
`each of seven cardiac glycosides was plotted and relative affinities
`for each digoxin analogue were calculated (Table 2). The sFv
`inhibition curves have been displaced to lower glycoside concentra(cid:173)
`tions than corresponding 26-10 Fab curves, because the concentra(cid:173)
`tion of active binding sites on the plate was less for sFv than for26-10
`Fab. When 0.25 M urea was added to the sFv in 0.01 M sodium
`acetate (pH 5.5), more active sFv bound to the goat anti-mouse Fab
`on the plate. Hence, the sFv specificity profile shifted toward higher
`glycoside concentrations, closer to the position ofthat for 26-10 Fab.
`
`BEQ 1030
`Page 3
`
`
`
`5882
`
`Biochemistry: Huston et a/.
`
`Proc. Nat/. A cad. Sci. USA 85 ( /988)
`
`Table 2. Specificity analysis
`
`26-10
`antibody Normalizing
`species
`glycoside
`
`Digoxin Digoxigenin Digitoxin Digitoxigenin
`
`Fab
`
`sFv
`
`Digoxin
`Digoxigenin
`Digoxin
`Digoxigenin
`
`1.0
`0.9
`1.0
`0.1
`
`1.2
`1.0
`7.3
`1.0
`
`0.9
`0.8
`2.0
`0.3
`
`1.0
`0.9
`2.6
`0.4
`
`Acetyl
`strophanthidin Gitoxin Ouabain
`1.3
`9.6
`15
`1.1
`8.1
`13
`5.9
`150
`62
`0.8
`8.5
`21
`Results are expressed as normalized concentration of iQhibitor giving 50% inhibition of 1251-labeled digoxin binding.
`Relative affinities for each digoxin analogue were calculated by dividing the concentration of each cardiac glycoside at 50%
`inhibition by the concentration of digoxin or digoxigenin that gave 50% inhibition for each type of 26-10 species.
`from Scatchard analysis (R1 [sFv] = (2.9 ± 0.4) x 10- 8 M;
`R1 [26-10 Fab] = (9.4 ± 0.5) x 10- 9 M) and total protein
`concentration from amino acid analysis ([sFv] = 1.3 x 10- 7
`M; [26-10 Fab] = 2.2 x 10- 8 M).
`The association constant for digoxin binding (Ka) was
`determined from binding data (Fig. 5) by binding isotherm
`analysis (40) [Ka (sFv) = 5.2 X 107 M- 1; Ka (Fab) = 3.3 X
`108 M - 1], by Sips analysis using linear regression analysis to
`calculateKa(39)[Ka(sFv) = 2.6 X 107 M- 1;Ka(Fab) = 1.8
`x 108 M- 1], and by Scatchard analysis (16) [Ka (sFv) = (3,2
`± 0.9) x 107 M-I; Ka (Fab) = (1.9 ± 0.2) x 108 M- 1]. In
`summary, Ka for sFv was 3-5 x 107 M - 1 , while Ka for 26" 10
`Fab ranged from 2 to 3 x 108 M - 1, under the conditions of
`0.25 M urea in 0.01 M sodium acetate (pH 5.5). Since the
`26-10 Fab had a lower Ka in this buffer than in PBSA at pH
`7 (Ka = 3.3 x 109 M- 1) (12), the sFv binding constant may
`have been similarly reduced.
`
`to be independent of refolding conditions (data not shown).
`The comparison of sFv with 26-10 Fab revealed some
`differences between their specificity profiles. Relative to
`acetyl strophanthidin, digoxin bound more tightly to the sFv
`than to 26-10 Fab, ouabain bound less tightly (Fig. 4), and
`other cardiac glycosides exhibited slight shifts In specificity
`(Table 2). When relative affinities are expressed in relation to
`digoxigenin, good agreement between sFv and 26-10 Fab
`values can be found for all analogues except digoxin. These
`data indicate that the sFv is slightly more specific for digoxin
`than the parent 26-10 Fab, but that major features ofthe 26-10
`combining site have been reproduced in the sFv.
`Although the sFv bound to ouabain-Sepharose was fully
`active, upon elution a substantial fraction of protein appeared
`to form inactive aggregates. The extent of sFv self-asso"
`ciation was aggravated further in PBSA at pH 7 but could be
`reduced by keepiqg the sFv in dilute acetate buffer at pH 5.5,
`and further minimized by adding urea to a concentration of
`0.25 M. This low concentration of urea enhanced activity by
`an order of magnitude without any apparent change in the Ka
`observed in acetate buffer alone. Under these conditions,
`active binding species represented 22% of sFv and 43% of
`26-10 Fab in solution, based on active site concentration (R1)
`
`~
`~
`~ 25
`~
`~ " 20
`~
`~ 15
`8
`~
`~
`~ c:s
`~
`~ 0
`ll:l
`
`10
`
`5
`
`A
`
`•
`•
`
`~~~
`
`• .. a ....
`
`-10
`LOG UNBOUND DIGOXIN CONCENTRATION !MJ
`
`-7
`
`-6
`
`-5
`
`-4
`
`DISCUSSION
`A single-chain biosynthetic antibody binding site was shown to
`closely mimic the antigen binding affinity and specificity of the
`parent antibody. Connection of V" and V L by a 15-residue
`linker may substitute for constant region contacts in the Fab
`
`8
`
`.......
`~
`s::::
`.......
`~
`~ 8
`~ f..::
`~ 6
`~
`8
`~ 4
`~ 1;,5
`C5 2
`~
`~ Q:;) 0
`-9
`-7
`-8
`-6
`-10
`-11
`LOG UNBOUND DIGOXIN CONCENTRATION /MJ
`
`FIG. 5. Analysis of digoxin binding affinity. (A) sFv binding isotherm and Sips plot (Inset). (B) 26-10 Fab binding isotherm and Sips plot
`(Inset). Binding isotherms display data plotted as the concentration of digoxin bound versus the log of the unbound digoxin concentration, and
`the dissociation constant corresponds to the ligand concentration at 50% saturation (40). Sips plots (Inset) present the data in linear form, with
`the same a,bscissa as the binding isotherm· but with the ordinate representing log[r/(n -
`r)] (defined below). The average intrinsic association
`r)] = a log C - a log K., where r equals mol of digoxin bound
`constant (KJ was calculated from the m<>Qified Sips equation (39), log[r/(n -
`per mol of antibody at an unbound digoxin concentration equal to C; n is mol of digoxin bound at saturation of the antibody binding site, and
`a is an index of heterogeneity, which describes the distribution of association constants about the average intrinsic association constant, K •.
`Least-squares linear regression analysis of the data indicated correlation coefficients for the lines obtained were 0.96 for sFv and 0.99 for 26-10
`Fab. Equilibrium binding was conducted in solution, as follows. Aliquots (100 ~LI) of [3H]digoxin at a series of concentrations (10- 6-10- 10 M)
`in 0.01 M sodium acetate (pH 5.5) with 1% bovine serum albumin were added to 26-10 Fab or sFv (100 ~LI) at a fixed concentration i1;1 0.01 M
`sodium acetate, pH 5.5/0.5 M urea/1% bovine serum albumin. After 2-3 hr of incubation at room temperature, the protein was precipitated by
`the successive addition of goat anti-mouse Fab serum, the IgG fraction of rabbit anti-goat IgG, and protein A-Sepharose. After 2 hr on ice, bound
`and free [3H]digoxin were separated by vacuum filtration of samples, and radioligand bound to the protein entrapped on glass fiber filters was
`measured by scintillation counting.
`
`BEQ 1030
`Page 4
`
`
`
`Biochemistry: Huston et al.
`
`Proc. Nat/. Acad. Sci. USA 85 (1988)
`
`5883
`
`and thereby aid recovery of native binding properties in the
`sFv. For example, the Fv(GAR) exhibited a reduction by a
`factor of 1000 in the riboflavin binding affinity of Fab(GAR)
`(11), while the MOPC 315 Fv bound dinitrophenol almost as
`well as the parent antibody (2). Construction of a single-chain
`Fv may have minimized the refolding problems of two-chain
`species, such as incorrect domain pairing or aggregation
`during the renaturation process (41). In fact, reconstitution of
`26-10 Fv from separately cloned V 8 and V L domains has thus
`far proven unsuccessful (J.S.H. and M.M.-H., unpublished
`data), while in vitro recombination of 26-10 Hand L chains
`produced a low yield of antibody with significantly reduced
`affinity for digoxin (16). Furthermore, past efforts to produce
`antibodies from cloned H and L chains gave very low
`recoveries (42-44). The present 12.6% yield of active sFv is 9
`times greater than that reported for antibody activity regained
`from H and L chains expressed in E. coli (42).
`The sFv and 26-10 Fab both contain identical V8 and VL
`polypeptide sequences, but other features of sFv covalent
`structure might perturb 26-10 combining site properties in the
`single-chain Fv. Proline has been added to the V 8 N terminus
`and connection of V regions via the linker has eliminated the
`charge on the V L a-amino group. The V region N termini may
`be in sufficient proximity of the combining site to influence
`binding properties. In a mutant of another anti-digoxin
`antibody, deletion of the first two residues ofV 8 dramatically
`changed its binding affinity (45). Introduction of the linker
`terminal carboxyl and may
`has also eliminated the V 8
`introduce constraints on folded V domains.
`Given the feasibility of making the 26-10 single-chain Fv
`biosynthetically, protein engineering can be used to advan(cid:173)
`tage in further studies. Variation in protein association has
`been related to restricted changes in primary sequence (46,
`47), and one may therefore expect that aggregation of 26-10
`sFv can be moderated by alteration of surface residues linked
`to self-association. Binding site variants of 26-10 sFv may
`likewise be constructed (13), or its entire framework re(cid:173)
`placed, while keeping complementarity determining region
`sequences unchanged (12). The immunopharmacology of
`biosynthetic antibody binding sites could prove particularly
`interesting, insofar as their small size may accelerate the
`pharmacokinetics and reduce the immunogenicity observed
`for Fab fragments administered intravenously (48). Further
`research on the single-chain Fv and related immunoconju(cid:173)
`gates may lead to biomedical applications that have been
`heretofore impossible with conventional antibody fragments.
`
`We are grateful for the expertise of Sarah Hardy, Abbie White,
`Denise Maratea, Clare Corbett, Rou-Fun Kwong, Larry Haith,
`Gay-May Wu, and Robert Juffras, and for the encouragement of
`Charles Cohen and Prof. Serge Timasheff. This project was sup(cid:173)
`ported in part by the National Institutes of Health through SBIR
`Grant CA 39870 and by HL 19259.
`
`1. Porter, R. R. (1959) Biochem. J. 73, 119-126.
`2.
`lobar, D., Hochman, J. & Givol, D. (1972) Proc. Nat/. Acad.
`Sci. USA 69, 2659-2662.
`3. Haber, E. (1964) Proc. Nat/. Acad. Sci. USA 52, 1099-1106.
`4. Whitney, P. L. & Tanford, C. (1965) Proc. Nat/. Acad. Sci.
`USA 53, 524-532.
`5. Hochman, J., lobar, D. & Givol, D. (1973) Biochemistry 12,
`1130-1135.
`6. Hochman, J., Gavish, M., lobar, D. & Givol, D. (1976)
`Biochemistry 15, 2706-2710.
`7. Sharon, J. & Givol, D. (1976) Biochemistry 15, 1591-1594.
`8. Kakimoto, K. & Onoue, K. (1974) J. lmmunol. 112, 1373-1382.
`9. Lin, L.-C. & Putnam, F. W. (1978) Proc. Natl. Acad. Sci. USA
`75, 2649-2653.
`10. Reth, M., lmanishi-Kari, T. & Rajewsky, K. (1979) Eur. J.
`lmmunol. 9, 1004-1013.
`
`11. Sen, J. & Beychok, S. (1986) Proteins: Struct. Funct. Genet. 1,
`256-262.
`12. Jones, P. T., Dear, P. H., Foote, J., Neuberger, M.S. &
`Winter, G. (1986) Nature (London) 321, 522-525.
`13. Sharon, J., Gefter, M. L., Manser, T. & Ptashne, M. (1986)
`Proc. Natl. Acad. Sci. USA 83, 2628-2631.
`14. Mudgett-Hunter, M., Margolies, M. N., Ju, A. & Haber, E.
`(1982) J. lmmunol. 129, 1165-1172.
`15. Mudgett-Hunter, M., Anderson, W., Haber, E. & Margolies,
`M. N. (1985) Mol. lmmunol. 22, 477-488.
`16. Hudson, N. W., Mudgett-Hunter, M., Panka, D. J. & Margo(cid:173)
`lies, M. N. (1987) J. lmmunol. 139, 2715-2723.
`17. Novotny, J. & Margolies, M. N. (1983) Biochemistry 77, 1155-
`1158.
`18. Saul, F., Amzel, L. M. & Poljak, R. J. (1978) J. Bioi. Chern.
`253, 585-597.
`19. Marquart, M., Deisenhofer, J., Huber, R. & Palm, W. (1980) J.
`Mol. Bioi. 141, 369-391.
`20. Satow,Y.,Cohen,G. H.,Padlan,E. A.&Davies,D. R.(1986)
`J. Mol. Bioi. 190, 593-604.
`21. Bernstein, F. C., Koetzle, T. F., Williams, G. J. B., Meyer,
`E. F., Brice, M.D., Rodgers, J. R., Kennard, 0., Shimanou(cid:173)
`chi, T. & Tasumi, M. J. (1977) J. Mol. Bioi. 112, 535-542.
`22. Brooks, B. R., Bruccoleri, R. E., Olafson, B. D., States, D. J.,
`Swaminathan, S. & Karplus, M. (1983) J. Comput. Chern. 4,
`187-217.
`23. Bruccoleri, R. E. & Karplus, M. (1987) Biopolymers 26, 137-
`168.
`24. Jones, T. A. (1978) J. Appl. Crystallogr. 11, 268-272.
`25. Grantham, R., Gautier, C. & Govy, M. (1980) Nucleic Acids
`Res. 8, 1893-1912.
`26. Roberts, D. M., Crea, R., Malecha, M., Alvarado-Urbina, G.,
`Chiarello, R. H. & Watterson, D. M. (1985) Biochemistry 24,
`5090-5098.
`27. Yanisch-Perron, C., Vieira, J. & Messing, J. (1985) Gene 3,
`103-119.
`28. Sanger, F., Nicklen, S. & Coulson, A. R. (1977) Proc. Natl.
`Acad. Sci. USA 74, 5463-5467.
`29. Miozzari, G. & Yanofsky, C. J. (1978) J. Bacteriol. 133, 1457-
`1466.
`30. Weber, K., Pringle, J. R. & Osborn, M. (1972) Methods
`Enzymol. 26, 3-27.
`31. Laemmli, U.K. (1970) Nature (London) 227, 680-685.
`32. Piszkiewicz, D., Landon, M. & Smith, E. L. (1970) Biochem.
`Biophys. Res. Commun. 40, 1173-1178.
`33. Fraser, K. J., Poulsen, K. & Haber, E. (1972)Biochemistry 11,
`4974-4977.
`34. Poulsen, K., Fraser, K. J. & Haber, E. (1972) Proc. Natl.
`Acad. Sci. USA 69, 2495-2499.
`35. Bidlingmeyer, B. A., Cohen, S. A. & Tarvin, T. L. (1984) J.
`Chromatogr. 336, 93-104.
`36. Lowry, 0. H., Rosebrough, N.J., Farr, A. L. & Randall, R. J.
`(1951) J. Bioi. Chern. 193, 265-275.
`37. Hewick, R. M., Hunkapiller, M. W., Hood, L. E. & Dreyer,
`W. J. (1981) J. Bioi. Chern. 256, 7990-7997.
`38. Matsudaira, P. (1987) J. Bioi. Chern. 262, 10035-10038.
`39. Smith, T. W., Butler, V. P., Jr., & Haber, E. (1970) Biochem-
`istry 9, 331-337.
`40. Klotz, I. M. (1982) Science 217, 1247-1249.
`41. Jaenicke, R. (1984) Angew. Chern. 23, 395-413.
`42. Cabilly, S., Riggs, A. D., Pande, H., Shively, J. E., Holmes,
`W. M., Rey, M., Perry, L. J., Wetzel, R. & Heyneker, H. L.
`(1984) Proc. Nat/. Acad. Sci. USA 81, 3273-3277.
`43. Boss, M.A., Kenten, J. H., Wood, C. R. & Emtage, J. S