THE JOURNAL OF BIOLOGICAL CHEMISTRY 8 : ) 1991 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 266, No. 21, Issue of July 25, pp. 14095-14103,1991 Printed in U.S.A.

Single-chain Site-specific Mutations of Fluorescein-Amino Acid Contact Residues in High Affinity Monoclonal Antibody 4-4-20"

(Received for publication, February 20, 1991)

Lisa K. DenzinS, Marc Whitlows, and Edward W. Voss, Jr.SP From the $Department of Microbiology, University of Illinois, Urbana, Illinois 61801 and the SGenex Corporation, Gaithersburg, Maryland 20877

Previous crystallographic studies of high affinity anti-fluorescein monoclonal antibody 4-4-20 (KO = 1.7 X 10'' M-') complexed with fluorescyl ligand resolved active site contact residues involved in binding. For better definition of the relative roles of three light chain antigen contact residues (L27dhi", L32tYr and L34erg), four site-specific mutations (L27dhi" to L27dlYS, L32ty' to L32Phe, and L34a'g to L34lY" and L34hi") were generated and expressed in single-chain antigen bind- ing derivatives of monoclonal antibody 4-4-20 con- taining two different polypeptide linkers (SCA 4-4-201 205c, 25 amino acids and SCA 4-4-201212, 14 amino acids). Results showed that L27dhi" and L32ty' were necessary for wild type binding affinities, however, were not required for near-wild type Qmax values (where Qmax is the maximum fluoroscein fluorescence quenching expressed as percent). Tyrosine L32 which hydrogen bonds with ligand was also characterized at the haptenic level through the use of 9-hydroxyphen- ylfluoron which lacks the carboxyl group to which L32 tyrosine forms a hydrogen bond. Results demonstrated that wild type SCA and mutant L32phe possessed simi- lar HPF binding characteristics. Active site contact residue L34"'g was important for fluorescein quench- ing maxima and binding affinity (L34hi" mutant), how- ever, substitution of lysine for arginine at L34 did not have a significant effect on observed Qmax value. In addition, substitutions had no effect on structural and topological characteristics, since all mutants retained similar idiotypic and metatypic properties. Finally, two linkers were comparatively examined to deter- mine relative contributions to mutant binding proper- ties and stability. No linker effects were observed. Collectively, these results verified the importance of these light chain fluorescein contact residues in the binding pocket of monoclonal antibody 4-4-20.

Despite recent advances in deciphering the genetic basis of Ab' diversity (Tonegawa, 1983; Eisen and Siskind, 1964), the

*This work was supported by a grant from the Biotechnology Research Development Corporation, Peoria, IL. The costs of publi- cation of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertise- ment" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

(I To whom correspondence should be addressed Dept. of Micro- biology, University of Illinois, 131 Burrill Hall, 407 s. Goodwin Ave., Urbana, IL 61801. Tel.: 217-333-1738; Fax: 217-244-6697.

' The abbreviations used are: Ab, antibody molecule; Ag, antigen; Fab, 50-kDa antigen binding fragment derived by papain digestion of Ig molecules; FDS, disodium salt of fluorescein; F1, fluorescein; VH and V,,, variable regions of immunoglobulin heavy (H) and light (L) chains; Fv, heterodimer of VH and VL; HPF, 9-hydroxyphenylfluoron; HRP, horseradish peroxidase; Id, idiotype; mAb, monoclonal anti- body; Met, metatype; PCR, polymerase chain reaction; SCA, single- chain antigen binding protein.

identification and relative roles of molecular interactions that occur between an antibody and a specific antigen remain unclear. Three-dimensional structural analyses of Fab-anti- gen complexes have defined specific active site contact resi- dues in many mAbs (for review see Mian et al., 1991); however, the strategic engineering of antibody active sites (to alter binding, affinity, and specificity) remains ill-defined with no established set of rules specifying structure-function relation- ships. Detailed and comprehensive analysis of antibody active site structures is necessary to generate a data base in order to formulate these rules. Any such rules, however, will apply to a restricted class of antigens based on intrinsic physical- chemical properties.

The immune response to dianionic fluorescein in BALB/ CV mice has evolved into an attractive model system to study antibody active site structure since the anti-F1 response has been shown to be both structurally and functionally diverse as demonstrated by extensive affinity maturation and lack of dominant idiotypes (Kranz and Voss, 1981; Kranz et al., 1983; Reinitz and Voss, 1985). Using polyaromatic F1 conjugated to a T-cell-dependent carrier as the immunogen, a panel of idiotypically related monoclonal Abs were obtained which spaned an affinity range of 104-10'" M" (Reinitz and Voss, 1984; Bates et al., 1985; Bedzyk et al., 1986). Recently, VH and VL nucleotide and deduced amino acid sequences were ob- tained for two anti-fluorescein idiotypic families (Bedzyk et al., 1989; Bedzyk et al., 1990a). Results suggested that rela- tively small differences in primary structures between idi- otypically related mAbs were still too complex to allow precise interpretations, although the differences between idiotypi- cally related antibodies, which varied in binding properties, strongly suggested which amino acid residues were important in ligand binding. The idiotype family approach also lacks site specific focusing. Therefore, to isolate specific molecular interactions between a defined Ag and Ab, high affinity pro- totype mAb 4-4-20 was chosen for site-specific mutagenesis studies. In addition, the three-dimensional structure of the Fab fragments complexed with F1 was determined based on x-ray diffraction studies and active site contact residues were defined (Herron et al., 1989).

TO facilitate site specific mutagenesis studies, a single-chain derivative of mAb 4-4-20 was synthesized (Bird et al., 1988; Bedzyk et al., 1990b). Single-chain antigen binding proteins (SCA'") involve only the variable immunoglobulin domains, VL and V H , associated by a polypeptide linker between the carboxyl terminus of VL and the amino terminus of VH. Following protein expression in Escherichia coli, refolding, and purification, SCA 4-4-20 containing a 14-amino acid linker exhibited a binding affinity within 2- to 3-fold of mAb 4-4-20 and a similar Q,,,,, (maximum F1 fluorescence quench- ing expressed as percent), idiotype, and metatype (Bedzyk et al., 1990b). These results validate SCA 4-4-20 as an ideal

14095

14096 Anti-F1 Binding Site Mutations

model system to study ligand-antibody interactions using site- specific mutagenesis.

It has been shown for the anti-phosphocholine myeloma protein McPC603 (Poljak, 1984) and anti-hen lysozyme Ab DL3 (Amit et al., 1986) that the majority of ligand contact residues are heavy chain encoded however, for anti-F1 mAb 4-4-20 direct contributions to ligand binding were provided primarily by the light chain and include L27dIyB, L32tY', and L 3 P K (numbering scheme of Kabat et al., 1987) (Herron et al., 1989). Contact residue L32ty' participated in a hydrogen bond with the benzyl-carboxyl group and also in a stacking interaction with the xanthenone-conjugated ring system. Haptenic contact residue, L27dh'", although within hydrogen bonding distance to the 3' enolic oxygen, does not appear to be directly involved in a hydrogen bond, and L34"Ig partici- pated in a salt bridge with the 6' carbon xanthenone ring enolic group. It has been reported previously that two idiotyp- ically related anti-F1 mAbs, high affinity mAb 4-4-20 and mAb 9-40, possessed approximately a 400-fold and a 2-fold difference in binding affinity and Qm,,, respectively (Bates et al., 1985). Primary structural analysis suggested that 4-4-20 and 9-40 used similar VH and VL genes and differed at nine heavy and two light chain residues. One of the VL residues, L34, was an arginine in high affinity 4-4-20 and a histidine in 9-40 (Bedzyk et al., 1990a). Therefore, these three VL antigen binding residues were chosen for initial mutagenesis studies. The following changes were made: L32tYr to L32ph", L34"Ig to L34lYs and L34h'", and L27dhi" to L27dlY" and expressed in 4-4-20 single-chain antigen binding proteins containing two different linkers (14 and 25 amino acids, respectively) to ascertain linker contributions to stability of active site mu- tants. Following expression, denaturation, refolding, and pu- rification, mutant SCAs were characterized in terms of bind- ing affinity, Q,,,,, (maximum quenching of bound fluorescein), X,,, (absorption profiles of bound Fl), idiotype, and metatype.

MATERIALS AND METHODS

Monoclonal Antibodies-mAb 4-4-20 was generated through chem- ically mediated cell fusion of hyperimmune splenocytes from an adult BALB/cV mouse immunized with Fl(1)-keyhole limpet hemocyanin emulsified in Freund's complete adjuvant (Difco) to nonsecreting Sp2/0-Ag14 myeloma cells (Galfre et al., 1977). mAb 4-4-20 (IgGZ., c) possessed an affinity of 1.7 X 10'' M-' for the fluorescyl ligand and has been extensively characterized (Kranz and Voss, 1981; Bates et al., 1985; Bedzyk et al., 1990a).

Strains, Plasmids, and Media-Expression vectors pGX8772 and pGX8773 encode SCA 4-4-20 fused to the OmpA signal sequence and containing the 212 (GSTSGSGKSSEGKG) and 2 0 5 ~ (SSAD- DAKKDAAKKDDAKKDDAKKDG) linkers, respectively. The expression vectors utilize a hybrid OL/PR X promotor, and expression was acheived by temperature shift from 30 to 42 "C in E. coli strain GX6712 (F galK2 rpsL clBh7) (Scandella et al., 1985). All E. coli cultures were grown in Luria broth or 2 X YT containing 100 pg/ml ampicillin.

Oligonucleotide-directed Mutagenesis-Template for mutagenesis was prepared by ligating the 900-base pair ClaI-EamHI fragment from pGX8772 into pMTL23 (Chambers et al., 1988) digested with the same enzymes to form pLKD4. Polymerase chain reaction meth- odology, shown in Fig. 1, was used for SCA 4-4-20 mutant construc- tion. Oligonucleotides used in PCR reactions were synthesized by the phosphoramidite method (Beaucage and Caruthers, 1981) at the University of Illinois Genetic Engineering Facility and sequences are shown in Fig. 1C. Reaction conditions for amplification of DNA fragments were as follows: 10 mM Tris-HCI, pH 8.3, 2.5 mM MgCl,, 50 mM KCI, 200 pg/ml gelatin, 200 mM each of 4 dNTPs, 0.1% Triton X-100, 5.0 units of Thermus aquaticus DNA polymerase (Promega), 10 ng of template DNA (pLKD4), and 30 pmol of each primer. The reactions were incubated in a thermal cycler (MJ Research, Inc.) using the following profile: 92 "C, 5 min; 45 "C, 3 min; 72 "C, 2 min followed by 30 cycles of 92 "C, 1 min; 45 "C, 1 min; 72 "C, 1 min. A mixed oligonucleotide pool (L32L34), designed to generate a number

of different base changes in the resulting PCR product, and M13 reverse primer were used to amplify a mutagenized 200-base pair fragment (Fig. 1A). The L32phe, L34h'", and L34ly8 mutants were constructed using this amplification procedure. The L27dlY" mutant was constructed from a 700-base pair fragment which was amplified using M13 universal and a mutagenic primer (L27d) (Fig. 1E). Fol- lowing amplification, PCR products were purified in low melting temperature agarose (Seaplaque, FMC), cloned into SmaI-digested pTZ18U (Mead et al., 1986), and screened for mutations by dideoxy sequencing. Desired mutants were digested with ClaI-KpnI (L32 and L34 mutants) or BglII-Hind111 (L27d mutant) and ligated into pGX8772 and pGX8773 digested with the same enzymes. Correct clones were identifed by restriction analysis and dideoxy sequencing followed by small scale protein expression in E. coli strain GX6712. Mutant clones were designated as follows: L32phe, pLKD72; L34his, pLKD80; L34IYs, pLKD84; L27dlY*,pLKD27.

Sequence Determination-Following cloning, PCR products were

stranded plasmid DNA template (Kraft et al., 1988) and Sequenase'" sequenced by the dideoxy chain termination procedure using double

(U. S. Biochemical Corp.). Expression and Purification of Proteins-For large scale protein

production, 500 ml of an overnight culture in 2 X YT plus ampicillin was used to inoculate 20 or 30 liters of media in a B. Braun Fermentor (B. Braun Biotechology). Cells were grown a t 30 "C until they reached a OD,,, of 1.0 at which time expression was acheived by temperature shift to 42 "C. After 1 h at 42 "C, cells were harvested and stored at -20 "C.

SCA protein was isolated and purified from aggregates in E. coli as follows. Cell paste was weighed, suspended in 1 ml of lysis buffer (50 mM Tris, pH 8.0, 1 mM EDTA, 0.1 mM phenylmethylsulfonyl fluo- ride)/3 g of cell paste and passed through a French pressure cell at 20,000 p. s. i. twice to lyse cells. Following centrifugation a t 20,000 X g for 45 min, pellets were suspended in l/z the above volume of lysis buffer, passed through the French press, and centrifuged at 20,000 X g for 30 min. Resultant pellets were washed twice with lysis buffer plus 0.5% Triton X-100 followed by a final wash in lysis buffer. The SCA pellet was solubilized in 6 M guanidine HCI in 50 mM Tris-HC1 pH 8.0, 10 mM CaCl,, 100 mM KCl, 0.1 mM phenylmethylsulfonyl fluoride (10 ml/g of SCA) and centrifuged a t 25,000 X g for 30 min. The SCA protein was renatured by rapid dilution (1:200-1:lOOO) into 50 mM Tris, pH 8.0, 10 mM CaC12, 50 mM KCl, 0.1 mM phenylmeth- ylsulfonyl fluoride. Following filtration and concentration, the re- folded SCA proteins were purified by affinity or cation exchange chromatography.

Affinity Chromatography-A F1-Sepharose immunoadsorbent was prepared using fluorescein-I-amine linked to epoxy-activated Sepha- rose 4B (Pharmacia LKB Biotechnology Inc.) as described previously (Sundberg and Porath, 1974). For affinity purification of SCA 4-4-20 proteins, refolded SCA was loaded onto a 2-ml F1-Sepharose column in 50 mM Tris, pH 8.0, 0.15 M NaCI. Following extensive washing, SCA proteins were eluted with 50 mM FDS in the same buffer. Eluted fractions were dialyzed extensively to remove free FDS.

Fluorescence Quenching Assay-Fluorescence quenching assay measurements were performed as described by Watt and Voss (1978) using an Aminco-Bowman spectrophotofluorometer. Cation exhange purified SCA 4-4-20/212 proteins and nonpurified SCA 4-4-20/205c proteins were used for Qm.. assays. Purified protein was not required since assays used for characterization are routinely performed with y globulin fractions of mouse ascites (Bates et al., 1985).

Dissociation Rate Assay-Dissociation rates of SCA 4-4-20 proteins (90% liganded with fluorescein or 9-hydroxyphenylfluoron) were de- termined a t 4 "C as described (Herron, 1984). Affinity constants were determined from dissociation rates by computer program OFFRATES version 1.2 (Interactive Software; Urbana, IL) using previously deter- mined association rate of 5 X lo6 M" s-l for anti-fluorescyl antibodies to calculate intrinsic affinities (K. = k l / k , ) (Kranz et al., 1982).

Metatypic Analysis-Xenogenic anti-Met reagent specific for the liganded state of mAb 4-4-20 was generated and purified as described previously (Weidner and Voss, 1991). Dissociation rates of affinity purified SCA 4-4-20/212 proteins were determined in the presence of 1.28 mg/ml xenogenic anti-Met y globulin as described by Weidner and Voss (1991).

Zdiotypic Analysk-Polyclonal rabbit anti-4-4-20 Id generation and characterization have been described (Bates et al., 1985, Bedzyk et al., 1986). Monoclonal Ab 4-4-20 was labeled with horseradish per- oxidase as described by Smith and Ulrich (1983). Polystyrene wells (Nunc-Immuno Plate Maxisorb) coated (50 pl/well) with absorbed and fractionated polyclonal anti-Id in 10 mM Tris, pH 8.0, were

Anti-F1 Binding Site Mutations 14097

incubated a t 37 "C for 2 h. Coated wells were washed three times with TBS-T (10 mM Tris, pH 8.0. 0.15 M NaC1, 0.05% Tween 20) and masked with TEBT (TBS-T plus 1% bovine serum albumin, 1 mM EDTA) for 30 min a t 37 "C. Horseradish peroxidase-labeled mAb 4- 4-20 was added to wells in the presence of SCA inhibitors (50 pl) diluted 1: lO in TBS. Inhibitor concentrations ranged from 1 X lo-' t o 1 X mol of active sites/well. Monoclonal Ab 4-4-20 was included as a control. Following incubation a 37 "C for 2 h, wells were washed three times with TBS-T, and 3,3',5,5'-tetramethylbenzidene (Organon Teknika) substrate (50 p l ) was added and incubated for 30 min at room temperature. Enzyme activity was terminated by the addition of 2 N H2S04 and optical density determined using a Dyna- tech MR600 automatic 96-well microtiter reader with a 450-nm cut- off filter.

Spectral Properties-Visible absorption spectra of liganded SCA 4- 4-20/212 mutants were obtained using a Beckman DU-64 spectro- photometer from which the absorption wavelength maxima of bound fluorescein was determined.

RESULTS

Construction and Expression of SCA 4-4-20 Mutants-Pre- vious studies have shown that SCA 4-4-201212 is a suitable model system to study the antibody active site (Bedzyk et al., 1990b). Therefore, three active site contact residues were chosen for initial mutagenesis studies, L32"', L34'Ig, and L27d"" based upon crystallographic results (-2.7 A). Muta- genesis was accomplished using PCR methodology (Fig. 1). Following amplification, PCR products were cloned, primary structures determined to verify mutations and subcloned into SCAs 4-4-20/212 and 4-4-201205~ genes. All mutant SCAs were expressed in E. coli from a hybrid OL/PR X promotor (Fig. 2), extracted from inclusion bodies with guanidine hy- drochloride, and refolded by rapid dilution. Percent refolded protein depended upon dilution employed and ranged from -8% for a 1:200 dilution to -20% for a 1 : lOOO (data not shown). The SCA 4-4-201212 mutants were purified by affin- ity or cation exchange chromatography, and SCA 4-4-201205~ proteins were affinity-purified. SDS-polyacrylamide gel elec- trophoresis analysis (Laemmli, 1970) of cation exchange and affinity-purified 4-4-20 SCAs showed that proteins were >90% pure (Fig. 2). Three to six mg of purified SCA was obtained per liter of E. coli cells.

Spectral Properties of SCA Mutants-Anti-fluorescein an-

tibodies can be characterized on the basis of percent fluores- cence quenching (QmBx) of excited bound fluorescein ligand. Therefore, the ability of SCA 4-4-20 mutants to quench fluorescein (4 = 0.92 for free F1, where 4 denotes quantum yield) fluorescence was determined (Table I). Qmsx values for wild type SCAs (83.6 h 0.1% and 85.9 f 0.4% for 4-4-20/205c and 4-4-20/212, respectively) compared well with Qmsx values obtained with affinity purified mAb 4-4-20 (83.6 f 1.6%), however, were slightly below the previously reported value for mAb 4-4-20 (Bates et al., 1985). Substitutions L32tYr to L32phe (77.9 f 0.3%, 205c; 83.1 f 0.8%, 212) and L27dhi" to L27dIY6 (76.4 f 0.3%, 205c; 84.2 f 0.8%, 212) had little effect on quenching properties, independent of linker examined, and these mutants retained greater than 90% of wild type activity (Table I). For SCA 4-4-20 L34IYs mutants, observed Qms. values were 57.3 f 1.2% (205c) and 74.9 f 0.5% (212). These values correspond to 65.8 and 87.2% of wild type activity. The significant difference in these two values can probably be attributed to the use of a partially pure 4-4-201205~ protein fraction and not to linker effects. The only mutation to have a significant effect on quenching maxima was the L348'g to L34hi" substitution. Values for L34hi" mutants (28.0 f 0.1%, 205c; 24.1 f 0.9,212) represented 33.4 and 28.1% of wild type activity, respectively.

Maximum quenchingvalues (4-4-201212 mutants) were also determined for the F1 structural analogue 9-hydroxyphenly- fluoron. Active site contact residue L32tyr participated in a hydrogen bond with the benzoate ring carboxyl group and also in a stacking interaction with the xanthenone conjugated ring structure. HPF is devoid of the 5' phenylcarboxyl group to which residue L32tYris hydrogen-bonded (Fig. 3). Therefore, HPF (4 = 0.21 for free HPF; Martin and Lindqvist, 1975) was included in these studies to examine the role of the hydrogen bond not only at the protein level (L32phe mutant) but also at the haptenic level (Table I). Qmax value for SCA 4- 4-201212 (59.0 f 0.3%) was identical (within error) to value observed for mAb 4-4-20 (58.9 f 0.3%). SCA 4-4-20/212 mutants L32ph" (59.2 f 0.4%) and L27dIY" (57.3 f 2.1%) exhibited quenching maxima similar to wild type, as expected. SCA 4-4-201212 L34lYs quenched HPF fluorescence 13.4 f

FIG. 1. Amplification and cloning strategy for construction of SCA 4- 4-20 active site mutants. A, scheme for amplification and cloning of L32 and L34 SCA mutants. B, amplification and cloning of L27d SCA mutant. C, primers used for construction of mutants. For the L32L34 primer, underlined nucleo- tide bases were altered on a 50:50 molar ratio with the wild type base in order to obtain specific mutations in L32 and L34. For construction of the L27d mu- tant, mutated bases are underlined. Ab- breviations: B, BamHI; Bg, BglII; C, ClaI; H, HindIII; K, KpnI.

A internal mixed primers

R&ke

+A- W2W4

+ - Y U I ;v" ' , pLKD4

K B Amplification

by PCR

1 1 1 1 1 1 C K C K C K

.1 Clone and Sequence

Subclone deaired mutants into pGX8772 and pGX8773

(Clal/Kpnl)

Y wl v'; ' , pU(D72

C K B v; 1 pWD8O

C K

U I B

I . U I Y V I pWD84

C K B

B M13 E 7 d Universal - r; YI p m 4

wl C B g H B

Amplification by PCR

BP H B 1 Clone and Sequence

Subclone deaired mutant into pGX8772 and pGX8773

(Bglll/Hindlll)

- Y ,& U I

"I( ry pLKD27

BP H B

14098 Anti-Fl Binding

0.1% (22.7% of wild type activity) and no quenching was observed for the L34"'$ mutant.

In addition to quenching the fluorescence of bound F1, anti- F1 antibodies produce a bathochromic shift of 10-20 nm in ligand absorption maximum. Table I1 summarizes X,,, anal- yses for SCA 4-4-201212 active site mutants. SCA 4-4-201212 exhibited a X,,., of 505 nm which was identical to the reported X,,,, for mAb 4-4-20 (Kranz and Voss, 1981; Bates et al., 1985). Except for the L34h'" mutant, all SCAs expressed near wild type red shifted absorption maxima (504-507 nm). The SCA 4-4-201212 L34"'" mutant displayed a X,,, value of 497 nm.

Determination of Dissociation Rates-Liganded SCAs re- sulting from affinity chromatographic purification were ana- lyzed using off-rate fluorescence analysis as described previ- ously (Herron, 1984). Table I11 summarizes binding affinities of SCA 4-4-20 proteins. Wild type SCAs possessed affinities for fluorescyl ligand of 4.9 X 10'' M" (205c) and 4.2 X lo9 M-l

(212) which were approximately 4-fold lower than previously reported affinity for mAb 4-4-20 (1.7 X 10'" M-'; Kranz and Voss, 1981; Bates et al., 1985). Substitution of phenylalanine for tyrosine at residue L32 resulted in a -30-fold decrease in affinity (1.6 X 10" M-'; 205c and 212) for both linkers exam- ined. SCA 4-4-20 L27dI."" mutants exhibited binding affinities

kd 1 2 3 4 5 6

45.4-

29.1 -

18.1-

14.4-

5.a-

3.0-

FIG. 2. SDS-polyacrylamide gel electrophoresis (13%) analysis of overexpressed SCA 4-4-201205~ and purified SCA 4-4-20 proteins. Lane 1, protein molecular weight markers; lane 2, uninduced SCA 4-4-20/205c in E. coli strain GX6712; lane 3, over- expressed SCA 4-4-20/205c from denatured inclusion bodies before purification; lane 4 , affinity-purified SCA 4-4-20/205c; lane 5, cation ion exchange purified SCA 4-4-20/212; lane 6, affinity-purified SCA 4-4-20/212. All samples were reduced with dithiothreitol and boiled before loading. Other SCA 4-4-20 mutants were analyzed by SDS- polyacrylamide gel electrophoresis yielding similar results (data not shown).

Site Mutations

of 4.4 X 10' M" (205c and 212 linkers) which corresponded to approximately a 10-fold decrease in affinity. For SCA active site mutant, L34Iy", affinities of 1.2 X 10' M" (205c) and 1.1 X 10' M" (212) were observed and for L34h'S, affinities of 6.6 X lo6 M" and 4.5 X lo6 W'. These affinities represent approximately a 400-fold decrease in binding affinity for the fluorescyl ligand for the L34Iy" mutants and a 742-fold (205c) and a 933-fold (212) decrease in affinity for the L34h'" mu- tants. No major linker effects were observed for any of the mutant proteins examined, although affinities for all 4-4-201 205c proteins were slightly higher than those obtained for the 4-4-201212 proteins.

To examine L32tsr hydrogen bond contributions to active site structure, binding affinities for the F1 analogue, HPF, were also determined for 4-4-201212 wild type and the L32ph' mutant. SCA 4-4-201212 possessed a binding affinity of 3.7 X 10' M" (Table 111) which was slightly below that obtained for mAb 4-4-20 (7.1 X 10' "I). SCA L32phe mutant exhibited a binding affinity for HPF of 1.4 X 10' M-', which was approx- imately 3-fold lower than wild type SCA.

Metatypic Analysis-Metatypic analyses are based on the elicitation of antibodies which preferentially bind the liganded state of an antibody active site (Voss et al., 1988). Binding of allogenic (Voss et al., 1988) and xenogenic (Weidner and Voss, 1991) anti-Met reagents elicited a measurable delay of the dissociation rate of the fluorescyl ligand bound to mAb 4-4- 20. Weidner and Voss (1991) reported a 23.1- and 61.5-fold increase in affinity for liganded mAb 4-4-20 and SCA 4-4-201 212, respectively, in the presence of xenogenic anti-Met re- agent. Therefore, relative affinities were calculated for SCA 4-4-201212 mutants in the presence of anti-Met and compared with affinities in the absence of anti-Met to determine if SCA mutants had retained metatype. Binding affinities obtained for SCA 4-4-201212 active site mutants are presented in Table IV. Dissociation rates were measured at 4 "C in the presence of 1.28 mg/ml anti-Met y globulin, and affinities were calcu- lated using previously determined k, = 5.0 X 10' M' s-' (Kranz et al., 1982). All SCA 4-4-201212 mutants displayed a delayed dissociation rate in the presence of anti-Met (Table IV). SCA 4-4-201212 exhibited a 8-fold affinity enhancement, which differed from the value reported by Weidner and Voss (1991) because a different anti-Met y globulin was used. SCA mu- tants L34lYs and L34h'" expressed the largest affinity enhance- ment (2636 x and 1911 x, respectively), L27dIys displayed an enhancment of 32 x and L321'h' the smallest affinity enhance- ment of 9 x. For the L34IyS (2.9 X 10'" "') and L27dty' (1.4 X 10" M-') mutants, the anti-Met reagent was able to enhance the affinity to near wild type (3.3 X 10'" M-') values (Table IV) .

Idiotypic Analysis-In addition to assessing antibody active site structure by probing with the anti-Met reagent, native

TABLE I Quenching properties of anti-Fl SCA 4-4-20 mutants

SCA Qmllr FDS AQ,,,. FDS Qmn. FDS & L o x FDS Qmnx HPF ABmn. HPF 205c" 205ch 212" 212h 212" 212h

96 Wild type' 83.6 f 0.1 85.9 f 0.4 59.0 -C 0.3 L27dh" to L27d"* 76.4 -C 0.3 91.4 84.2 -C 0.8 98.0 57.3 -C 2.1 97.1 ~ 3 2 1 ~ 1 to ~3211l1~. 77.9 f 0.3 93.2 83.1 -C 0.8 96.7 59.2 f 0.4 100.0 ~ 3 4 m to ~ 3 4 1 ~ 5 57.3 f 1.2 68.5 74.9 f 0.5 87.2 13.4 -C 0.1 22.7 L34"'C to L34h" 28.0 -C 0.1 33.4 24.1 -C 0.9 28.1 0.0 0.0

" Fluorescence of single chain antigen binding protein (205c or 212 linker) bound relative to free fluorescein

* Value represents percent quenching relative to wild type. ' Qmn, values for affinity purified mAb 4-4-20 were 83.6 f 1.6% and 58.9 f 0.3% for FDS and HPF, respectively.

(FDS) or HPF at protein saturation. Value represents mean of triplicate trials -C S.D.

Anti-F1 Binding

active structures were analyzed with an anti-idiotype reagent to determine the degree of idiotypic relatedness between SCA 4-4-20/212 wild type and mutant proteins. Soluble SCA in- hibitors were used to inhibit binding of HRP-4-4-20 with xenogenic anti-4-4-20 Id. Results of anti-Id experiments are shown in Fig. 4. All SCA proteins expressed nearly identical anti-Id inhibition curves and were also virtually identical to the inhibition curve obtained for mAb 4-4-20.

DISCUSSION

Previous x-ray diffraction studies delineated active site contact residues for mAb 4-4-20 Fab fragments complexed with ligand (Herron et al., 1989). This report describes con- struction and characterization of active site mutants for three ligand contact residues, L27dh'", L32ty', and L34aTg (other active site contact residues were L91"", L96t'P, and H33t'P). Interactions between these contact residues and fluorescyl ligand are schematically represented in Fig. 5 and include the close proximity of one enolic group on the xanthonyl ring to the imidazolium nitrogen of L27dhis, the electrostatic inter- action between the guanidino group of L34arg and the other enolic oxygen, and a hydrogen bond between the 5' phenyl- carboxyl group of L32tYr (see also Fig. 6). The phenolic ring of tyrosine L32 was also involved in a stacking interaction with the xanthonyl moiety of fluorescein. To ascertain anti- fluorescein active site contact residue contributions to F1 binding and structural characteristics, the following amino acid substitutions were constructed L27dhi" to L27dlYs, L32tyr

A c o o - A

FLUORESCEIN 9-HYDROXY- PHENYLFLUORON

FIG. 3. Comparative structures of fluorescein and struc- tural analogue 9-hydroxyphenylfluoron.

TABLE I1 Spectral properties of S C A 4-4-201212 mutants

SCA A,.,"

nm Wild type 505 L27dh'"to L27dlYS 505 L32'" to L32ph' 504 L34"" to L34"" 507 L34"'K to L34h" 497

" Absorption wavelength maxima of bound fluorescein.

Site Mutations 14099

to L32Phe, and L34"" to L34'"" and L34his. The arginine to histidine substitution a t residue L34 was based on sequence differences between idiotypically related mAbs 4-4-20 and 9- 40. Monoclonal Ab 9-40 exhibits about a 400-fold decrease in binding affinity and Q,,,,, approximately 50% lower when compared with mAb 4-4-20 (Bates et al., 1985). Primary structural analysis (Bedzyk et al., 1990a) coupled with x-ray crystallographic data (Herron et al., 1989) revealed that mAb 9-40 L34hi" was germ line encoded and 4-4-20 L34ar9 correlated with increased 4-4-20 binding and enhanced Qmax.

Mutations were expressed in two single-chain antigen bind- ing derivatives of mAb 4-4-20, one of which (SCA 4-4-20/212) was previously shown to have a binding affinity approximately 4-fold lower than mAb 4-4-20 and identical Qmsx, idiotype, and metatype (Bedzyk et al., 1990b). To investigate linker contributions to binding and stability of SCAs, active site mutants were expressed in SCAs containing two different polypeptide linkers (205c and 212). Anti-fluorescein mutant SCAs were characterized by their F1 binding properties (bind- ing affinity, Qmax, and X,,,) and also by their structural characteristics (metatype and idiotype).

In the fluorescein-anti-fluorescein system fluorescein is both the antigen and a fluorescent probe. Therefore, the sensitive fluorescent properties of fluorescein can be used to derive information regarding the microenvironment of the antibody active site (Voss, 1990). To assess contributions to F1 quenching by active site contact residues examined, quenching maxima were determined for each SCA mutant. Fluorescein quenching results (Table I) showed that the only amino acid substitution to have a significant effect on F1 quenching maxima was the L34arg to L34h'" conversion (28.0%, 205c; 24.1%, 212). All other modified residues examined, except for the L34'" subsitution (205c linker), exhibited greater than 87% of wild-type quenching activity, independent of linker examined. These data suggest that changes made at these residues (L27dIYs, L32phe, and L341y") may not be impor- tant in the quenching mechanism. SCA 4-4-20/212 mutants were also characterized by a red shift in the absorption spectrum when fluorescein is bound by Ab. As expected, these data (Table 11) are very similar to the Qma. data since the only mutant to exhibit a large bathochromic shift in X,,, was the L34his mutant. There are four possible explanations for ob- served Qmax and X,,, values obtained for the L34hi" mutants. First, substitution of histidine for arginine in the 4-4-20 active site could lead to greater ligand flexibility resulting, perhaps, in a reduced Q,,, value. Second, unlike arginine and lysine, the histidine at L34 may not be able to form a salt bridge with the 6' carbon xanthenone ring enolic group of fluorescein thereby changing the electronic configuration of the ligand,

TABLE 111 Binding constants of anti -FL SCA 4-4-20 mutants

SCA KO FDS A K o FDS 205c" 205cb

KO FDS AK" FDS 2 12"

K , HPF 212h 212"

X lo6 M" M" X 10" M" M" X 10" M"

Wild type' L27dh" to L27dlY" L32'" to L321'h'' L34"'X to L34IYS L34"'X to L34h'"

4900 rt 370 4200 k 560 37 k 4.0

31 X 160 k 6.1 26 X 14 f 2.0 440 k 43 11 x 440 f 1.5 10 x ND 160 k 65

12 k 1.8 408 X 11 k 1.2 382 X NDd 6.6 f 1.7 742 X 4.5 t 1.6 933 x ND

" Reactions monitored at 4 "C with 90% of SCA (205c or 212 linker) active site containing either free fluorescein

* Decrease in affinity relative to wild type (205c or 212). ' Binding affinities for mAb 4-4-20 are 1.7 X 10'' M-' (Kranz and Voss, 1981) and 7.1 X 10' M" for FDS and

(FDS) or HPF. Values represent triplicate trials rt S.D.

HPF, respectively. Not determined.

14100 Anti-F1 Binding Site Mutations

TABLE IV Effect of anti-Met reagent on binding kinetics of liganded SCA 4-4-

20/212 mutants SCA K,.M".~ AK,,M'

x 109 ~ - 1 "' 33 f 4.4 a x 14 f 0.5 32 X

L34" to L34IY" 1.4 f 0.1 9 x 29 f 4.7 2636 X

L348'X to L34h'" 8.6 f 2.2 1911 x

Wild type L27dh'* to L27dIYs L32"' to L32phe

" Reactions monitored at 4 "C. Affinity of SCA 4-4-20/212 mutants in the presence of 1.28 mg/

ml anti-Met y globulin in reaction mixture. ' Increase in affinity due to presence of anti-Met reagent.

100

80

6 60

s 40

2od 0 -0 -7 -6 -6 -4 -3 -2 -1 0 1

Log (rrnoiee active sitedwell)

FIG. 4. Inhibition of HRP-labeled mAb 4-4-20 Id anti-Id interaction by SCA 4-4-201212 and mutant SCAs. Polystryene plates were coated with xenogenic polyclonal anti-4-4-20 Id and co- incubated with HRP-4-4-20 and the following inhibitors: mAb 4-4- 20 (open triangles), SCA 4-4-20/212 wild type (closed diamonds), SCA 4-4-20/212 L32ph' (open circles), SCA 4-4-20/212 L34lYs (open dia- monds) , SCA 4-4-20/212 L34h'" (closed triangles), SCA 4-4-20/212 L27dIYn (closed circles). Inhibition of HRP-4-4-20 binding was quan- titated spectrophotometrically after the addition of 3,3',5,5'-tetra- methylbenzidene substrate. Data points represent mean of triplicate trials.

resulting in reduced quenching. Third, it has been suggested that tryptophan residues in the active sites of anti-F1 Abs are responsible for observed quenching (Watt and Voss, 1977; Templeton and Ware, 1985). Therefore, a histidine at residue L34 could alter other active site contact residue (like H33t'P and L96"p) interactions with ligand, reducing contact, or destabilizing interactions necessary for maximal quenching. Site-specific mutagenesis studies are currently in progress to examine H33'Ip and L96t'P contributions to quenching. Finally, observed Qmax values could correlate with reduced affinity for the fluorescyl ligand; however, in anti-F1 antibodies, there appears to be no direct relationship between affinity and Qmax (Voss, 1990).

In terms of binding affinity, all SCA 4-4-20 mutants con- structed exhibited a decrease in binding affinity, regardless of linker utilized. Disruption of L32tyr hydrogen bond with the

A

8 FI

FIG. 5. Diagram of anti-fluorescein mAb 4-4-20 active site amino acid residues and specific interactions with fluorescyl ligand was based on x-ray diffraction data of 4-4-20 Fab fragments liganded with fluorescein (Herron et al., 1989). A , interactions of residues L34"Ig and L91"" with 6' carbon enolic group of F1 and L27dhiS with 3' carbon enolic group. Residues H100atY', H1OObtYr, and H32tY' form an aromatic active site shell into which the fluorescyl ligand fits tightly. B, aromatic interactions involved in mAb 4-4-20 binding. L32'yr, L96t'P, and H33t'P participate in stacking interactions with fluorescyl ligand. In addition, L3ZtYr was involved in a hydrogen bond with the phenylcarboxyl group of fluorescein.

phenylcarboxyl group of fluorescein (Fig. 6) by substitution of phenylalanine at this residue resulted in approximately a 30-fold decrease in binding affinity. Substitution of phenyl- alanine at this residue disrupted the hydrogen bond; however, the phenyl ring still maintained the stacking interaction with the xanthonyl moiety of fluorescein resulting in only a 30- fold decrease in affinity. Rationale for the construction L27dlYS mutant was based on x-ray diffraction data, which showed that there may be a hydrogen bond between L27dhi" and the F1 enolic oxygen near the mouth of the site; however, exact interactions could not be extracted. Therefore, histidine was changed to a lysine to increase side chain length and, hope- fully, enhance interaction with the fluorescyl ligand. Results of fluorescence-based kinetic studies revealed that this sub- stitution decreased binding affinity about 10-fold, indicating destabilizing effects. Increased side chain length may have been responsible for the observed decrease in binding affinity. The final residue examined, L34arg, has been implicated in mAb 4-4-20 increased binding affinity, based on x-ray diffrac- tion studies (Herron et al., 1989), primary structural analysis (Bedzyk et al., 1990a) and in vitro heavy and light chain reassociation between mAb 4-4-20 and lower affinity 9-40 (L34hi8) (Bedzyk and Voss, 1991). As expected, construction of L34 mutants resulted in decreased binding affinity. First, conversion of arginine (L34) to lysine resulted in a 400-fold decrease in affinity. Second, changing L34 to histidine (as in mAb 9-40) resulted in approximately an 800-fold decrease in affinity. Active site contact residue L34"Ig is involved in a salt bridge with an enolic oxygen on the fluorescyl ligand at the deepest part of the site (Fig. 5), and, therefore, stepwise substitutions of the smaller primary amine and imidazole ring at this residue probably cannot span the required distance to

Anti-Fl Binding Site Mutations 14101

FIG. 6. Resolution of the interac- tions of L32 tyrosine with bound fluorescein ligand in mAb 4-4-20. The hydrogen bond and stacking inter- actions are based oc x-ray crystallo- graphic results at 1.8 A (M. Whitlow, A. J. Howard, J. F. Wood, K. D. Hardman, and E. W. Voss, Jr., manuscript in prep- aration).

reform a salt bridge, resulting in a lower binding affinity for F1.

Structural characteristics of SCA 4-4-20/212 mutant pro- teins were determined by monitoring idiotypic and metatypic properties to determine if native active site structures had been reformed during mutant refolding. Initially, this was accomplished by determining if mutant SCAs had retained parental idiotype. By using an inhibition assay with polyclonal anti-idiotype sera specific for mAb 4-4-20, idiotype inhibition curves were superimposible (Fig. 4). This implied that SCA 4-4-20/212 mutants exhibited identical idiotypic properties when compared with both wild type SCA and mAb 4-4-20. These results were expected since heavy chain complemen- tarity determining region 3 residues are usually implicated in idiotype expression (Perlmutter, 1984; Nishinarita et al., 1985; Schilling et al., 1980; Clevinger et al., 1980; Rudikoff et al., 1983).

To further evaluate active site structure-function properties of mutant SCAs, metatypic characteristics were determined. Recently, Weidner and Voss (1991) generated a xenogenic polyclonal anti-Met reagent specific for liganded mAb 4-4-20. The xenogenic reagent reacted inefficiently with the nonli- ganded state of mAb 4-4-20 (idiotype); however, it was able to react with other liganded idiotypically related mAbs to varying degrees. The hallmark of anti-Met activity has be- come a delay in the ligand's dissociation rate, resulting in an artifkally enhanced affinity. Therefore, to determine if mu- tant SCAs had retained metatypic properties, dissociation rates in the presence of anti-Met reagent were determined. The metatype reagent artificially enhanced the affinity of all SCA 4-4-20/212 mutants showing that metatype was retained since anti-Met reagent was able to bind to all mutants; how- ever, the change in affinity varied significantly (Table IV). Single-chain antigen binding protein 4-4-20/212 displayed an enhanced affinity of 3.3 X 10" M" (8 X increase). The anti- Met reagent was able to enhance the affinities of the L34lYs and L27diys mutants to near that level (2.9 X 10" M-'; 2636 X and 1.4 X 10" K 1 ; 32 X, respectively). However, the enhancement of affinity was not as great for the L32ph' (1.4 X lo9 M-'; 9 X) and L34hi" (8.6 x lo9 M-I; 1911 X).

The L32ty' hydrogen bond to the fluorescyl ligand (Fig. 6) was further examined in studies involving the use of a struc- tural analogue of fluorescein, 9-hydroxyphenylfluoron (Fig. 3) which lacks the 5' phenylcarboxyl group to which residue

L32 is hydrogen bonded. These studies were conducted to further examine the role of the hydrogen bond at both the protein (L32phe mutant) and haptenic levels. Since HPF lacks the carboxyl group to which L32ty' hydrogen bonds, quenching values and binding affinities obtained for wild type SCA and the L32phe mutant should be similar. As displayed in Tables I and 111, Qm., values were identical (within error) and the L32Ph" mutant possessed a binding affinity approximately 3- fold lower than wild type. This difference in affinity may be attributed to slightly different stacking interactions with the xanthonyl moiety of fluorescein between tyrosine and phen- ylalanine since the phenylalanine active site structure may not be identical to wild type. Tyrosine L32 is also important for antigen binding in other antibody active sites suggesting that this residue is not only important for anti-F1 binding. Of the six active site structures reviewed by Mian et al. (1991), four Abs (HyHEL-10, D1.3, B1312, and 4-4-20) use L32tm as an antigen contact residue.

Additionally, HPF quenching values were determined for the other SCA 4-4-20/212 mutants. Results of these studies showed that the L34Iy" mutant, which was able to quench F1 fluorescence 87.2% (212 linker), only quenched HPF fluores- cence 13.4% indicating that the disruption of L34"Ig salt bridge coupled with loss of hydrogen bond by L32tm significantly reduced quenching ability. This could be due to the low affinity for ligand (HPF) that is caused by this structural analogue induced double mutation (affinity for HPF possesed by this mutant was not determined because it was less lo6 M-', a required value for the dissociation rate assay). These results indicate that a L32ty' stacking interaction with the fluorescyl ligand is not sufficient for HPF quenching in the L34Iy" mutant, and the hydrogen bond must also be present. The two remaining SCA mutants quenched HPF fluorescence like wild type (L27d1yB) or not at all (L34hi").

Recently, Batra et al. (1990) constructed anti-Tac-PE40 chimeric single-chain immunotoxins containing three differ- ent polypeptide linkers (14, 15, and 16 amino acids) and showed that all three had similar binding properties, suggest- ing that linker effects were unimportant. Therefore, to further assess linker contributions to stability and activity of SCAs, fluorescent based binding assays (Q,,,,, and binding affinity) were determined for mutants expressed in SCAs containing two different polypeptide linkers (14 and 25 amino acids, respectively). The linkers employed were designed to span the

14102 Anti-F1 Binding Site Mutations

distance between the VL and V H (35 A). The 25-amino acid linker (205c) was designed to form an a helix and therefore is longer that the other linker (212). As results indicate (Tables I and 111) there is no significant difference in QmaX and binding affinities for either linker examined; however, protein refolding studies comparing the 212 and 205c linkers have shown that the 205c linker allows SCA 4-4-20 to refold more quickly.'

These data, coupled with a recent Ab active site structure- function review by Mian et al. (1991) led to a model for the active site structure of anti-fluorescein Abs. Mian and co- workers (1991) suggested that active site tryptophan and tryosine residues, two of the most common residues in antigen binding pockets, produce a binding site that behaves as a template responsible for general binding and that various forces have resulted in a fine tuning of the binding site by the addition of other antigen contact residues (like serine, argi- nine, histidine, and asparagine). The 4-4-20 binding pocket can be viewed in these terms, since 3 of 6 antigen contact residues are tyrosine and tryptophans. These aromatic resi- dues, in addition to heavy chain residues H32tY', H1OOatY' and H1OObtY', whose sidegroups do not contact Ag directly, form an aromatic pocket or shell into which fluorescein tightly binds (Fig. 5). Other contact residues (L27dhiR, L34'I9, and L91"") could be considered the fine tuning and in the case of L34arg, required for high affinity and Qmsx characteristics.

In conclusion, through the use of constructed active site mutants of an anti-fluorescein SCA, it has been shown that three light chain residues are important in binding and/or quenching properties of mAb 4-4-20. Although L32tYr and L27dhis are necessary for wild type binding affinities, Qmax values were not significantly affected by these modifications. Residue L34arE, which had been previously implicated in in- creased binding affinity and characteristic Qmax values (Bed- zyk et al., 1990a; Bedzyk and Voss, 1991), was confirmed by site-specific mutagenesis studies. Metatypic and idiotypic properties appeared to be unaffected by the mutations exam- ined. Effects of L32tYr hydrogen bond were further examined using a structural analogue of F1 (HPF) and results demon- strated that L32phe mutant behaves similarly to wild type when HPF was in the active site of this SCA. Additionally, two linkers were examined to determine their contributions to ligand binding and stability and no linker effects were observed.

The single-chain expression system employed in these stud- ies facilitated convenient and rapid manipulation of the an- tibody active site. Other expression systems have been devel- oped that also allow for such manipulations. Rees and co- workers (1990) have developed an expression system that allows for the accumulation of Ab heavy and light chains in the cytoplasm of protease deficient strains of E. coli. Addi- tionally, other groups have recently reported expression sys- tems that allow for the secretion of Fv fragments from E. coli (Skerra and Pluckthun, 1988; Ward et al., 1989; Skerra et al., 1991). Such secretion systems should facilitate engineering and random mutagenesis of the Ab active site. In the future, these technological advances will undoubtly assist active site defining studies and serve to establish rules designed to make the engineering of antibody active sites a rational and efficient process.

Acknowledgments-We wish to thank Karla M. Weidner for her expert technical assistance in the metatype assays. We also thank

' M. W. Pantoliano, R. E. Bird, L. S. Johnson, E. D. Asel, S. W. Dodd, J . F. Wood, and K. D. Hardman, manuscript in preparation.

Karl D. Hardman and Leslie S. Johnson for their interest, sugges- tions, and sharing of results in related studies.

REFERENCES Amit, A. G., Mariuzza, R. A., Phillips, S. E. V., and Poljak, R. J .

Batra, J. K., FitzGerald, D., Gately, M., Chaudhary, V. K., and

Bates, R. M., Ballard, D. W., and Voss, E. W., Jr. (1985) Mol.

Beaucage, S. L., and Caruthers, M. H. (1981) Tetrahedron Lett. 22,

Bedzyk, W. D., and Voss, E. W., Jr. (1991) Mol. Zmmunol. 28, 27-34 Bedzyk, W. D., Reinitz, D. M., and Voss, E. W., Jr. (1986) Mol.

Immunol. 23, 1319-1328 Bedzyk, W. D., Johnson, L. S., Riordan, G. S., and E. W. Voss, Jr.

(1989) J. Biol. Chem. 264, 1565-1569 Bedzyk, W. D., Herron, J. N., Edmundson, A., and Voss, E. W., Jr.

(1990a) J. Biol. Chem. 265 , 133-138 Bedzyk, W. D., Weidner, K. M., Denzin, L. K., Johnson, L. S.,

Hardman, K. D., Pantoliano, M. W., Asel, E. K., and Voss, E. W., Jr. (1990b) J. Biol. Chem. 265,18615-18620

Bird, R. E., Hardman, K. D., Jacobson, J. W., Johnson, S., Kaufman, B. M., Lee, S. M., Lee, T., Pope, S. H., Riordan, G. S., and Whitlow, M. (1988) Science 242, 423-426

Chambers, S. P., Prior, S. E., Barstow, D. A,, and Minton, N. P. (1988) Gene (Amst.) 68, 139-149

Clevinger, B., Schilling, J., Hood, L., and Davie, J. M. (1980) J. Ezp. Med. 151, 1059-1070

Eisen, H. N., and Siskind, G. W. (1964) Biochemistry 3,996-1008 Field, H., Yarranton, G. T., and Rees, A. R. (1990) Protein Eng. 3,

Galfre, G., Howe, S. C., Milstein, C., Butcher, G. W., and Howard, J. C. (1977) Nature 266, 550-552

Herron, J. N. (1984)in Fluorescein Hapten: An Immunological Probe (Voss, E. W., Jr., ed) pp. 49-76, CRC Press, Boca Raton, FL

Herron, J. N., He, X., Mason, M. L., Voss, E. W., Jr., and Edmundson, A. B. (1989) Proteins 5, 271-280

Kabat, E. A,, Wu, T. T., Reid-Miller, M., Perry, H. M., and Gottes- man, K. S. (1987) Sequences of Proteins of Immunological Interest, United States Department of Health and Human Services, Na- tional Institutes of Health, Bethesda, MD

Kraft, R., Tardiff, J., Krauter, K. S., and Leinwand, L. A. (1988) Biotechniques 6, 544

Kranz, D. M., and Voss, E. W., Jr. (1981) Mol. Zmmunol. 18, 889- 898

Kranz, D. M., Herron, J. N., and Voss, E. V., Jr. (1982) J. Biol. Chem.

Kranz, D. M., Ballard, D. W., and Voss, E. V., Jr. (1983) Mol.

Laemmli, U. K. (1970) Nature 227,680-685 Martin, M. M., and Lindqvist L. (1975) J. Lumin. 10, 381-390 Mead, D. M., Szczesna-Skorupa, E., and Kemper, B. (1986) Protein

Mian, I. S., Bradwell, A. R., and Olson, A. J. (1991) J. Mol. Biol.

Nishinarita, S., Claflin, J. L., and Lieberman, R. (1985) J. Immunol.

Perlmutter, R. M. (1984) in The Biology of Idiotypes (Greene, M. I., and Nisonoff, A., eds) pp. 59-74, Plenum Publishing Corp., New York

Poljak, R. J. (1984)in The Biology of Idiotypes (Greene, M. I., and Nisonoff, A., eds) pp. 131-140, Plenum Publishing Corp., New York

Reinitz, D. M., and Voss, E. W., Jr. (1984) Mol. Immunol. 21, 775- 784

Reinitz, D. M., and Voss, E. W., Jr. (1985) J. Immunol. 135 , 3365- 3371

Rudikoff, S., Pawlita, M., Pumphrey, J., Mushinski, E., and Potter, M. (1983) J. Exp Med. 158 , 1385-1400

Scandella, D., Arther, P., Mattingly, M., and Neuhold, L. (1985) J. Cell. Biochem. 9B, 203

Schilling, J., Clevinger, B., Davie, J. M., and Hood, L. (1980) Nature 283,35-40

Skerra, A., and Pluckthun, A. (1988) Science 240, 1038-1041

(1986) Science 233, 747-753

Pastan, I. (1990) J. Biol. Chem. 265, 15198-15202

Immunol. 22,811-877

1859-1862

641-647

257,6987-6995

Immunol. 20,1313-1322

Eng. 1 , 67-74

217,133-151

134,436-442

Tonegawa, S. (1983) Nature 3 0 2 , 575-581 Weidner, K. M., and Voss, E. W., Jr . (1991) J. Biol. Chem. 2 6 6 , Voss, E. W., Jr. (1990) Comments Mol. Cell. Biophys. 6, 197-221 2513-2519