Phagedisplayofricin Bchainandits single bindingdomains ...SBD1 Pst1i-j SBD2 46 N'1 T143-/DFXhoI N2655 F262 1-XhoI T143 100 bp F Xhol 26 262 255 F N FIG. 1. Structure ofgene 3 (A) and

Proc. Nati. Acad. Sci. USAVol. 89, pp. 3756-3760, May 1992Biochemistry

Phage display of ricin B chain and its single binding domains:System for screening galactose-binding mutantsCANDACE SWIMMER*, SOPHIE M. LEHAR*, JOHN MCCAFFERTYt, DAVID J. CHISWELLt,WALTER A. BLATTLER*, AND BRAYDON C. GUILD**ImmunoGen Inc., 148 Sidney Street, Cambridge, MA 02139; and tCambridge Antibody Technology Ltd., Daly Research Laboratories, Babraham Hall,Cambridge CR2 4AT, United Kingdom

Communicated by Fotis C. Kafatos, December 26, 1991

ABSTRACT We demonstrate that the B chain of ricintoxin preserves its lectin activity when expressed as a fusionprotein on the surface of fd phage. Moreover, B chain, whichfolds into two topologically similar globular domains, can bedissected into amino-terminal and carboxyl-terminal domainsto form single binding domains (SBDs) of B chain, each ofwhich displays specificity for complex galactosides. The specificbinding exhibited by the fusion protein of these SBDs waseliminated when amino acid substitutions Gly-46 in SBD1 orGly-255 in SBD2 for native asparagine were introduced to alterkey residues implicated in hydrogen bonding with substrate.These data demonstrate that it is possible to use a prokaryoticexpression system to stably express and screen ricin B chainand its SBDs for sugar-binding mutants. Expression of ricin Bchain on the surface of fd phage provides a method that can beused to efficiently select mutants with altered binding activitiesfrom a randomly generated library.

The application of methods that use bacteriophages as vehi-cles for exposing peptides (1-5) or whole functional foreignproteins (6-9) shows great promise for rapidly screeninglibraries ofrecombinant gene products. Single-chain antibod-ies (7), Fab fragments (8, 9), and human growth hormone (6)are displayed in native conformation and bind to their specificligand when fused to the adsorption protein, gene 3 protein,of fd bacteriophage. To explore whether a plant lectin wouldexhibit sugar-binding activity when similarly displayed on aphage, DNA encoding the B chain of ricin, the toxic lectinfrom castor beans, was introduced into the fd phage expres-sion system.

Ricin is a heterodimeric toxin comprised of a galactose-binding B chain (Mr = 32,000) and a cytotoxic A chain (Mr =30,500). The B chain of ricin attaches holotoxin to cellsurfaces via complex carbohydrates with terminal galacto-sides and facilitates penetration of cell membranes by thetoxic A chain (10, 11). The x-ray crystallographic structure ofricin shows that the B chain folds into two topologicallysimilar globular domains, each capable of binding one mol-ecule ofthe disaccharide lactose at sites 75 A apart at oppositeends of the molecule (12). Thus, the sugar-binding activity ofB chain appears to be predominantly localized to two bindingsites.Recombinant ricin B chain has been expressed in eukary-

otic expression systems, which allowed study of its lectinactivity by site-directed mutagenesis. Amino acid residuesidentified by the crystallographic analysis as key contactresidues in the two sugar-binding pockets of ricin B chainwere changed. In one report, lectin activity of ricin B chainwas nearly completely abrogated by altering the contactresidue Asn-255 to alanine in one of the two sugar-bindingsites (13), while a second report states that amino acid

mutations must be made in each of the two sugar-bindingpockets to effectively eliminate lectin activity of recombinantB chain (14).

Ricin has been used extensively in the production ofimmunotoxins (15). Immunotoxins consisting of a monoclo-nal antibody conjugated to unmodified ricin are potent cyto-toxic agents; however, they lack cell specificity because ofthe binding of the toxin moiety of the conjugate to cellsthrough its B chain. Immunotoxins comprised of an antibodyconjugated to ricin A chain alone have been shown in mostcases to be less cytotoxic than conjugates containing ricinholotoxin (16). This was explained by the absence ofB chain,which plays an important role in the translocation of A chainacross a cellular membrane to the cytosol (11, 17). Modifi-cation of the galactose-binding domains of ricin B chainthrough the chemical linking of affinity ligands reduces thebinding activity of the B chain to cells by a factor of 10,000while preserving the translocation function. Immunotoxinsmade with this blocked ricin are selective agents that ap-proach the potency of native ricin (11).The goal in producing a "genetically blocked" recombi-

nant ricin B chain is to reduce or eliminate sugar-bindingactivity by mutational changes without interfering with thetranslocation function of the B chain. To this end, we aim togenerate and screen a library ofB chain binding mutants witha range of galactose-binding affinities. In this paper wedescribe a system for the expression and selection of suchmutants on phage.

MATERIALS AND METHODSConstruction of Recombinant Bacteriophage. The fd phage

vector, fdCAT1 (7), was digested with Pst I and Xho I andligated with a Pst I-Sal I double-stranded oligonucleotidecontaining an internal Xho I restriction site followed by asegment of DNA encoding the Gly-Ser repeat [(Ser-Gly3)4].Ligation ofXho I (fdCAT1 vector) and Sal I (synthetic DNAsegment) ends results in the loss ofboth restriction sites in thefinal construction and introduces an additional sequence ofDNA encoding serine, valine, and glutamic acid after theGly-Ser (gs) motif to form fdCATgs. The internal Xho Icloning site is retained.A DNA segment encoding ricin B chain was obtained from

genomic DNA prepared from 4-day-old sprouts of Ricinuscommunis beans and was cloned into pUC19. The DNAsequence and the deduced protein sequence match publishedsequences of ricin B chain (18, 19). The 5' end of the B-chainclone was altered (a TGT codon was changed to AGT) toreplace Cys-4 in the protein with serine. The adapted full-length recombinant B-chain clone encoding 262 amino acidswas introduced as a Pst I-Xho I fragment into the Pst I/XhoI cloning sites of either (i) fdCAT1 to make fdCAT1-B or (ii)fdCATgs to make fdCATgs-B.

Abbreviation: SBD, single binding domain.

3756

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

May

15,

202

1

Proc. NatL. Acad. Sci. USA 89 (1992) 3757

To construct fdCATgs-SBD1, fdCATgs-B was digestedwith Kpn I and Xho I, thereby removing the DNA sequenceencoding amino acids 116-262 of ricin B chain, and a KpnI-Xho I double-stranded oligonucleotide encoding aminoacids 116-143 of B chain was ligated to the vector.To alter the sugar-binding affinity of SBD1, the codon

encoding Asn-46 was changed to one encoding glycine in theB-chain clone by site-directed mutagenesis. A 332-base pair(bp) BamHI-Kpn I fragment of the B-chain clone (includingthe codon for Asn-46) was subjected to oligonucleotide-directed mutagenesis by using Amersham's oligonucleotidein vitro mutagenesis system II. Point mutations in thissegment of DNA were determined by DNA sequence anal-ysis (dideoxy sequencing), and a clone carrying a codon forGly-46 was isolated. A 346-bp Pst I-Kpn I fragment includingthe codon for Gly-46 was substituted for the native sequencein clone fdCATgs-SBD1 to make vector fdCATgs-SBD1-G46.To construct fdCATgs-SBD2, the 5' end of a 520-bp

PflMI-Xho I fragment encoding amino acids 89-262 of Bchain was adapted with a Pst I-P lMI double-stranded oli-gonucleotide encoding amino acids 81-88 ofB chain and thenligated into the Pst I/Xho I cloning site of fdCATgs. To alterthe binding affinity of SBD2 of B chain, the codon encodingAsn-255 was changed to one encoding glycine in clonefdCATgs-SBD2 (an Nco I-Xho I double-stranded oligonu-cleotide encoding the last 12 amino acids ofB chain, includinga codon for Gly-255 was substituted for the native sequence).Growth and Preparation of Bacteriophage. Cultures of

bacteria (JM109 strain; Pharmacia) transformed with recom-binant phage constructs were grown in 400 ml of 2 x YTmedium (16 g of Bactotryptone, 10 g of yeast extract, and 5g of NaCl per liter containing tetracycline at 15 ,g/l) at 37°Cwith shaking for 24 hr. Phage were precipitated twice withpolyethylene glycol (7) and then resuspended in 1 ml (finalvolume) of TE (10 mM Tris HCl/1 mM EDTA, pH 8.0).Concentrations of phage in the final stocks were determinedby extracting phage DNA in 1% SDS and subjecting samplesto electrophoresis on 1% agarose gels, where single-strandedphage M13 DNA (400 ng = 0.167 pmol) was included as thequantitative standard. Gels were stained with ethidium bro-mide, photographed (Polaroid type 55 negative film), and theintensities of the bands were measured with a laser densi-

A

tometer (LKB). The DNA concentration (ng/,ul of phagesolution) was determined by comparison to the M13 stan-dard, and the molar concentration of phage was then calcu-lated by using a M, = 3.3 x 106 for the genomic DNA of fdphage. Phage stocks were then normalized to 0.3 ,uM.ELISA for Lectin Activity of Recombinant Phage. Lectin

activity of recombinant ricin B chain expressed on phage wasmeasured by ELISA as follows. Microtiter plates (96-wellImmulon II, Dynatech) were coated with 50,ul of asialofetuinat 2 ,ug/ml in 0.5 M sodium carbonate buffer (pH 9.9) at 4°Covernight and then blocked with 2% (vol/vol) fetal calfserumin TBS (50 mM Tris HCI, pH 7.5/150 mM NaCl) at roomtemperature for 1 hr. Fifty microliters of solution containingrecombinant phage or purified plant ricin B chain (InlandLaboratories, Austin, TX) at different concentrations inblocking buffer was added to each well, and the plates werethen incubated at room temperature for 1 hr. For competitionbinding assays, phage stocks were diluted into blockingbuffer or buffer containing either 10 A.M asialofetuin or 100mM lactose and incubated at room temperature for 1 hr priorto addition. Plates were washed five times with TBS con-taining 0.1% Tween 20 (Sigma) followed by incubation with50 1,u of either (i) rabbit anti-ricin antibody (Sigma) to detectplant ricin B-chain standards or (ii) sheep anti-M13 antibodyto detect recombinant phage particles. Plates were washed asabove, and secondary antibodies conjugated to alkaline phos-phatase-either (i) goat anti-rabbit immunoglobulin (JacksonImmunoResearch) or (ii) rabbit anti-sheep immunoglobulin(Fisher)-were added. After 1 hr at room temperature, plateswere washed consecutively with TBS containing 0.1% Tweenand TBS, and color was developed by addition of p-nitro-phenyl phosphate (1 mg/ml in 0.1 M sodium carbonatebuffer, pH 9.8/10 mM MgCl2). Absorbance at 405 nm wasdetected with a Titertek Multiscan plate reader.

RESULTSConstruction of Ricin B Chain DNA in Phage Expression

Vectors. The design of the recombinant phage expressionvectors fdCAT1 and fdCATgs used in this study is outlinedin Fig. 1A. Vector fdCAT1 is identical to that used inprevious studies (7). Vector fdCATgs is a derivative offdCAT1 in which the cloning site was modified to introduce

Ss GS GS 100 bp

fd gene 3 P.l...1.

cldeavage etafdCAT1 Pst 1 Xho 1

TCT CAC TCC GCT CAG GTC CM CTG CAG GAC CTC GAG ATC AAA CGGO v a L 0 L E K R

I ~ ~~~~~~~~IIIfdCATgs I Pst 1 Xho 1 |

TCT CAC TCC GCT CAG GTC CM CTG CAG GAC CTC GAG TCG GGA GGC GGT1TCG GTC GAG ATC AAA CGGa vV L a L ELS G G GJ4S V E K R

Bncin B chain Pstl

N'

SBD1 Pst1i-j

SBD2

46 N'1 T143-/DFXhoI

N2655 F2621-XhoIT143

100 bp

F Xhol

26262

255 FN

FIG. 1. Structure of gene 3 (A) andricin B-chain gene (B) constructions inrecombinant fd phage expression vec-tors. (A) Gene 3 of recombinant fd phageexpression vector fdCAT1. Sequences ofthe cloning sites for fdCAT1 and fd-CATgs are presented. Each cloning siteis preceded by the gene 3 signal sequence(SS) indicated (black box). Regions ofgene 3 encoding the Gly-Ser (GS)-richrepeats (stippled boxes) found in maturegene 3 protein are shown. The sequenceofDNA encoding a Gly-Ser repeat intro-duced into the fdCATgs cloning site isindicated by brackets. (B) Structure ofrecombinant ricin B-chain DNA con-structions that are displayed as fusionproteins with the product ofgene 3 on thesurface of fd phage. DNA segments en-coding full-length B chain, SBD1 (aminoacids 1-143), and SBD2 (amino acids81-226) of ricin B chain are indicated.Regions encoding lactose-binding pock-ets in mature B chain, as assessed fromthe crystal structure, are shown (hatchedboxes). Locations of codons encodingAsn-46 (N4) and Asn-255 (N255) aremarked.

146

Pst1r

N

Biochemistry: Swimmer et al.

_-

Dow

nloa

ded

by g

uest

on

May

15,

202

1

3758 Biochemistry: Swimmer et al.

DNA encoding a (Ser-Gly3)4 repeat (a flexible amino acidlinker). We speculated that the Gly-Ser-rich motifs present inthe gene 3 protein separate functional domains and hopedthat the introduction of codons encoding this motif wouldfacilitate the presentation of recombinant ricin B chain as aseparately folded domain in the fusion protein formed withthe gene 3 protein on the phage surface. A segment ofDNAencoding full-length ricin B chain (262 amino acids) wascloned into these vectors to make recombinant phage con-structions fdCAT1-B and fdCATgs-B. In each case, in-framefusions create an open reading frame that extends from thesignal sequence of gene 3 into the ricin B-chain gene and thenthe entire gene 3 (Fig. 1B).

It has been shown that both binding pockets of B chainneed to be mutated to eliminate lectin binding in full-length Bchain (14). A library of mutants in full-length B chain repre-senting all possible permutations of amino acid substitutionsin both binding pockets requires a diversity of approximately1 x 1012 individual members, which is the upper limit offeasibility in the fd phage expression system. Therefore, weset out to test if it would be possible to screen mutants in onebinding pocket at a time, which would reduce the diversity ofthe library to about 1 x 106 clones per binding domain. X-raycrystallographic analysis of ricin has shown that the B chainfolds into two topologically similar globular domains thatbind lactose in similar ways (12), and we explored thepossibility of expressing either one of these globular domainsas a single binding domain (SBD) ofB chain. DNAs encodingthe amino-terminal half of B chain (amino acids 1-143;referred to as SBD1) and the carboxyl-terminal half of Bchain (amino acids 81-262; referred to as SBD2) were intro-duced (as precise in-frame fusions with gene 3) into fdCATgsto make fdCATgs-SBD1 and fdCATgs-SBD2, respectively(Fig. 1B).To test the ability of the phage display system to distinguish

between the binding affinity of SBDs and of binding-deficientSBDs, two binding mutants were constructed that encodeGly-46 in SBD1 and Gly-255 in SBD2. The native asparagineresidues at these positions had been identified in ricin B chainas key binding residues that form hydrogen bonds withgalactose and have been shown by mutational analysis to becritical for binding (12-14). The two vectors were constructedas described in Materials and Methods and are called fd-CATgs-SBD1-G46 and fdCATgs-SBD2-G255.

Expression of Recombinant Ricin B Chain-Gene 3 Proteinon fd Phage. Expression of full-length ricin B chain fused togene 3 protein was detected on recombinant phage by sub-jecting phage samples to electrophoresis on SDS/polyacryl-amide gradient gels and analyzing them by Western blottingwith rabbit antibodies against gene 3 protein (Fig. 2). Aprotein of Mr = 110,000, the size predicted for the B chain-gene 3 fusion protein, was expressed by both recombinantphage fdCAT1-B and fdCATgs-B, while this protein wasabsent from control phage fdCATgs (Fig. 2A). A protein ofsimilar Mr was present on phage fdCAT1-D1.3, representinga single-chain antibody-gene 3 fusion protein. A second bandat Mr = 80,000 was present in all lanes, representing maturegene 3 protein (encoded by fdCATgs) or a proteolysis productof the recombinant B chain-gene 3 fusion protein, as hadbeen reported previously for antibody fused to gene 3 protein(7). Additional bands caused by proteolysis of products werepresent in samples of fdCATgs-B. Only the band at Mr =110,000 observed in samples fdCAT1-B and fdCATgs-B wascross-reactive with the anti-ricin B-chain monoclonal anti-body, TFTB1 (data not shown). It was estimated by visualinspection of the blot that approximately one copy of the Bchain-gene 3 fusion protein was present for every threecopies of proteolyzed fusion protein.The expression ofSBDs of ricin B chain on recombinant fd

phage is shown in Fig. 2B. A protein that migrates slightly

A111)11

4) 4Q1 1

N, 4z,A.

(T T)

,.! b?

110,000 - __

80,000 -lo -

B00

i. i '

u- _ _

F_ :;.__I

FIG. 2. Western blot analysis of recombinant phage expressing Bchain-gene 3 fusion proteins. Equivalent numbers of phage weresubjected to electrophoresis on 4-12%6 polyacrylamide/SDS gradientgels. Resolved proteins were electroblotted onto nitrocellulose andprobed by using rabbit anti-gene 3 protein antibodies [gift of I.Rasched (20)] orTFTB1 (21), a murine anti-ricin B-chain monoclonalantibody. Goat anti-rabbit or goat anti-mouse secondary antibodiesconjugated to alkaline phosphatase were used for staining. Twomajor bands are present at Mr = 110,000 and Mr = 80,000 inrecombinant phage samples. (A) Expression on phage of full-lengthB chain-gene 3 fusion protein in fdCAT1-B and fdCATgs-B (lanegs-B). Controls include fdCATgs without the inserted B chain (lanegs) and fdCAT1-D1.3, which expresses anti-lysozyme single-chainantibody fused to gene 3 protein. (B) Expression on phage of SBDsof B chain fused to gene 3 protein on the surface of phage fdCATgs(abbreviated as gs). gs-SBD1 expresses the amino-terminal domainof B chain; gs-SBD2, the carboxyl-terminal domain of B-chain;gs-SBD1-G46, the Gly-46 mutant of SBD1; and gs-SBD2-G255, theGly-255 mutant of SBD2.

faster than the full-length B chain-gene 3 fusion protein (Mr= 110,000) was present on recombinant phage fdCATgs-SBD1 and on recombinant phage fdCATgs-SBD2. SBD1 ofricin B chain comprises 143 amino acids, while SBD2 com-prises 181 amino acids, which was reflected as a slightdifference in mobility observed for their respective fusionswith gene 3 protein. Bands with equal mobility are present inlanes containing fdCATgs-SBD1-G46 and fdCATgs-SBD2-G255, demonstrating that the presence of Gly-46 in theprotein product of fdCATgs-SBD1-G46 and Gly-255 in theprotein product of fdCATgs-SBD2-G255 does not affect theintegrity of the SBD-gene 3 fusion proteins. In all samples,a band was present at Mr = 80,000, again suggesting partialproteolysis of the SBD-gene 3 fusion proteins. In summary,these data demonstrate that full-length ricin B chain, SBDs ofricin B chain, and their mutants can be expressed as recom-binant fusion proteins on the surface of fd phage.

Lectin Binding Activity of Recombinant Ricin Phage. Lectinactivity of recombinant phage fdCAT1-B and fdCATgs-Bexpressing full-length ricin B chain was assayed by usingasialofetuin in an ELISA (Fig. 3). To permit a comparison ofthe activities of the different constructions in a quantitativeELISA, the concentrations of recombinant phage stockswere determined by quantitating the DNA content of thesolutions as described. Native ricin B chain was includedwith the ELISA as a positive control.Half-maximal binding of recombinant phage expressing

full-length B-chain-gene 3 fusion protein was reached at aphage concentration of 1-2 nM and saturation at about a10-fold higher concentration of 10 nM (Fig. 3). The bindingprofile of these curves matches the profile seen for native Bchain (Fig. 3 Inset). No detectable difference in binding bythe B chain-gene 3 fusion protein was noted in experimentswith fdCAT1 and the derivative vector fdCATgs. No binding

Proc. NatL Acad Sci. USA 89 (1992)

Dow

nloa

ded

by g

uest

on

May

15,

202

1

Proc. Natl. Acad. Sci. USA 89 (1992) 3759

0.6

0.5

(a)0akV.0

0

CA

0.4

0.3

0.2

a)0

cacoenE.0Cl)

1t010-t9 1 -8

Concentration

FIG. 3. Binding curves of recombinant phage expressing full-length ricin B chain-gene 3 fusion proteins. The binding to asialofe-tuin of standardized stocks of phage was measured in an ELISA, andthe molar concentrations of phage tested were plotted versus theabsorbance at 405 nm. The binding activity of native ricin B chain forasialofetuin in the ELISA is plotted in the inserted graph. Eachconcentration of recombinant phage or native B chain was assayedin triplicate, and the average of three independent experiments withthe calculated standard deviations is plotted. o, fdCATgs-B; *,

fdCAT1-B; *, fdCAT1-D1.3; o, fdCATgs.

was detected in experiments with control phage fdCATgs(without inserted B chain). At the highest concentration ofphage particles tested, 30 nM, binding activity started toappear for the phage expressing the anti-lysozyme single-chain antibody, fdCAT1-D1.3. This activity may be due to aweak cross-reactivity by the anti-lysozyme antibody withasialofetuin or, more probably, to nonspecific interactions.The stability of the lectin activity of the B chain-gene 3 fusionprotein was evaluated by retesting fdCAT1-B and fdCATgs-Bafter storage at 40C for several months. In each case, thelectin binding of ricin B chain remained unchanged. It ispossible that a fraction of the recombinant phage exhibitsmultivalent binding of B chain when more than one Bchain-gene 3 fusion protein is present per phage. However,Western blots of recombinant phage expressing B-chain-gene 3 fusion protein show an average ofone fusion per phageparticle.The binding activities of the phage expressing SBDs of

ricin B chain and their mutants were analyzed similarly andcompared with the results observed for phage with full-lengthB chain (Fig. 4). The binding curves for SBD1 and SBD2show similar binding profiles, although shifted to a 3- to 6-foldhigher concentration. Mutation of the SBDs in fdCATgs-SBD1-G46 and in fdCATgs-SBD2-G255 essentially elimi-nated binding activity. The residual binding by these phage at30 nM is similar to that observed for binding offdCAT1-D1.3at this concentration and suggests that at very high concen-trations of phage, nonspecific interactions increase.

Binding Specificity of Ricin B-chain Recombinant Phage.The binding specificity of recombinant phage expressingfull-length B chain or SBDs of B chain was determined byusing a competition binding assay (Fig. 5). An ELISA asdescribed above was performed in which solutions of fd-CAT1-B, fdCATgs-B, fdCATgs-SDB1, and fdCATgs-SBD2were pretreated with either 100 mM lactose or 10 tLMasialofetuin and then added to the asialofetuin-coated micro-titer plates to perform the ELISA. Phage concentrations of 6nM were used, which is the range for maximal binding inphage expressing full-length B-chain and SBDs of B chain.Binding of phage fdCAT1-B and fdCATgs-B was inhibited byincubation with asialofetuin or with lactose. Similarly, while

0.6

0.5-

0.4

0.3

0.2-

0.1

0.0--

-O.i.I... . .

o0-1 1 1 0 10 1 0 9Concentration

1 0 8

FIG. 4. Binding curves of recombinant phage expressing SBDsand SBD mutants fused to gene 3 protein. Binding to asialofetuin wasmeasured in an ELISA as described in the legend to Fig. 3. *,

fdCATgs-SBD1; o, fdCATgs-SBD1-G46; o, fdCATgs-SBD2; *, fd-CATgs-SBD2-G255. The binding curve for fdCATgs-B (A) wasincluded as a standard for comparison with Fig. 3.

the level of binding by SBD1 and SBD2 is lower than forphage expressing full-length B chain, the binding of eachsingle domain was further lowered by the presence of asia-lofetuin or lactose.To analyze the degree to which the mutants fdCATgs-

SBD1-G46 and fdCATgs-SBD2-G255 have lost the specificbinding capabilities of their respective SBDs, we tested iftheir binding activities could still be lowered in a competitivebinding assay (Fig. 6). At the highest phage concentrationmeasured (30 nM), where the binding of fdCATgs-SBD1 andfdCATgs-SBD2 was specifically inhibited by both lactose andasialofetuin, binding of the mutants fdCATgs-SBD1-G46 andfdCATgs-SBD2-G255 was not affected. In addition, the bind-ing of fdCAT-D1.3 was not inhibited. Therefore, the residualbinding observed is due to nonspecific interactions in theELISA and this assay is not able to detect any specificbinding activity for the mutants.

DISCUSSIONThe results presented in this paper show that it is possible tobisect ricin B chain to produce functional SBDs that can beused for a mutational analysis of lectin-binding activity. Such

C.) co_3

0.5 _C)

0.4

.0.

FIG. S. Binding specificity of recombinant phage expressingfull-length B chain-phages fdCAT1-B and fdCATgs-B--or SBDs ofB chain-phages fdCATgs-SBD1 and fdCATgs-SBD2-fused togene 3 protein as measured in a competition binding assay. The phageexpression vector used in this analysis, fdCATgs, is abbreviated gs.An ELISA was performed as described in which samples of recom-binant phage (6 nM) were left untreated (solid bars) or were pre-treated with either lactose (stippled bars) or asialofetuin (hatchedbars) prior to performing the ELISA. Bar graphs are used to indicatethe absorbance (405 nm) measured for each sample of recombinantphage tested. Each bar represents the average of two independentexperiments.

2

0

Biochemistry: Swimmer et al.

0. 1 -

0.0 -

-0. 1i 0 -1 1 1 0o - 7

Dow

nloa

ded

by g

uest

on

May

15,

202

1

3760 Biochemistry: Swimmer et al.

c o LOLOAm~~~~0.5 n

N C,ai

0.4 M C) CD)

0.3~~~~~~~~~~~~~0COnoo 0.1

0.0

FIG. 6. Binding specificity of recombinant phage expressingSBDs-phages fdCATgs-SBD1 and fdCATgs-SBD2-and their re-spective mutants-fdCATgs-SBD1-G46 and fdCATgs-SBD2-G255-fused with gene 3 protein as measured in a competitionbinding assay. The phage expression vector, fdCATgs, is abbrevi-ated gs. An ELISA was performed as described in Fig. 5 except thatthe concentration of recombinant phage assayed was 30 nM. Un-treated samples (solid bars) and samples pretreated with lactose(stippled bars) or asialofetuin (hatched bars) are shown. Controlsrepresented in this assay include phage fdCAT1-D1.3 expressing theanti-lysozyme single-chain antibody D1.3 fused to gene 3 protein andphage fdCATgs expressing gene 3 without an inserted sequence.

an analysis is hampered in full-length ricin B chain by thepresence oftwo independent high-affinity binding sites. As inthis report, attempts to engineer mutations that affect bindinginto recombinant B chain have relied upon interpretation ofkey binding residues identified in the crystal structure (13,14). Displaying libraries of random B-chain mutants of SBDson phage is a powerful alternative means of selecting mutantswith a broad spectrum of binding affinities expressed inSBDs.The data reported here show that when fused with gene 3

protein of fd bacteriophage, full-length B chain and SBDs ofB chain display stable lectin-binding activity similar to theactivity observed for native B chain. Thus nonglycosylated Bchain appears to display full lectin activity in a stable manner.The role of glycosylation has been investigated to determinewhether it contributes to the stability and lectin activity ofexpressed recombinant B chain. When B chain is expressedin the cytoplasm of yeast (22) or in the periplasm of Esche-richia coli (23), it is not glycosylated, it forms insolubleaggregates, and the recombinant products are unstable. How-ever, when ricin B chain is expressed in the endoplasmicreticulum of yeast, it is glycosylated and active (22). B chainexpressed from mRNA transcripts in microinjected Xenopusoocytes is glycosylated and stably expressed (14). In thatreport, changing the asparagine residues of the two N-gly-cosylation sites to glutamine produced a nonglycosylated Bchain completely devoid of lectin activity. E. coli do notglycosylate proteins; therefore, our data show that the pres-ence of carbohydrate side chains is not necessarily a prereq-uisite for proper folding and stability of B chain. Rather, itappears that when expressed as a fusion protein, recombinantnonglycosylated B chain retains an active conformation.On phage, both the amino-terminal domain (SBD1) and the

carboxyl-terminal domain (SBD2) of B chain display lectinactivity. Binding affinity in both domains was strongly re-duced by substituting Gly for Asn-46 in SBD1 and Asn-255 inSBD2, respectively. The effects of the same amino acidchanges on lectin affinity in full-length B chain expressed ineukaryotic cells has been reported previously. In one report,changing simultaneously Lys-40 to methionine and Asn-46 toglycine in domain 1 and changing Asn-255 to glycine indomain 2 completely abolished binding of B chain. Changesin just one domain alone did not completely disrupt binding

in full-length B chain (14). A second report describes a singleamino acid change, Asn-255 to alanine in binding domain 2 offull-length B chain, which abrogates >99%o of the lectinactivity (13). It may be that Ala-255 has an unusual effect onB chain by reducing lectin activity in binding domain 1.However, SBD1 expressed alone on phage displays lectinactivity, and that binding activity is abrogated by mutation,which is consistent with observations made by Wales et al.(14).

Expressing libraries ofrandomly generated B chain bindingmutants on phage presents the opportunity to select pools ofgalactose-binding mutants that encompass broad ranges ofbinding affinities. Each binding pocket can be screenedindependently of the other as a SBD. Pools of galactose-binding mutants in SBD1 and SBD2 will be reunited inrandom combinatorial libraries to reconstitute full-lengthB-chain libraries of galactose-binding mutants displayed onphage in an analogous manner to combinatorial libraries ofantibody variable regions that have been expressed on fdphage (24). These libraries will be screened by affinityselection to identify mutants deficient in lectin activity. Thisis the first step in selecting genetically blocked ricin. Expres-sion of B-chain variants on phage and analysis of theirproperties is a promising avenue toward a better definition ofthe role of lectin binding in the biology of ricin cytotoxicityin eukaryotic cells.

We thank Hedy Adari, Victor Goldmacher, and John Lambert foradvice.

1. Parmley, S. F. & Smith, G. P. (1988) Gene 73, 305-318.2. Cwirla, S. E., Peters, E. A., Barrett, R. W. & Dower, W. J. (1990)

Proc. Nat!. Acad. Sci. USA 87, 6378-6382.3. Devlin, J. J., Panganiban, L. C. & Devlin, P. E. (1990) Science 249,

404-406.4. Scott, J. K. & Smith, G. P. (1990) Science 249, 386-390.5. Greenwood, J., Willis, A. E. & Perham, R. N. (1991) J. Mol. Biol.

220, 821-827.6. Bass, S., Greene, R. & Wells, J. A. (1990) Proteins 8, 309-314.7. McCafferty, J., Griffiths, A. D., Winter, G. & Chiswell, D. J. (1990)

Nature (London) 348, 552-554.8. Hoogenboom, H. R., Griffiths, A. D., Johnson, K. S., Chiswell,

D. J., Hudson, P. & Winter, G. (1991) Nucleic Acids Res. 19,4133-4137.

9. Kang, A. S., Barbas, C. F., Janda, K. D., Benkovic, S. J. &Lerner, R. A. (1991) Proc. Natl. Acad. Sci. USA 88, 4363-4366.

10. Youle, R. J., Murray, G. J. & Neville, D. M., Jr. (1981) Cell 23,551-559.

11. Lambert, J., Goldmacher, V. S., Collinson, A. R., Nadler, L. M. &Blattler, W. A. (1991) Cancer Res. 51, 6236-6242.

12. Montfort, W., Villafranca, J. E., Monzingo, A. F., Ernst, S. R.,Katzin, B., Rutenber, E., Xuong, N. H., Hamil, R. & Robertus,J. D. (1987) J. Biol. Chem. 262, 5398-5403.

13. Vitetta, E. & Yen, N. (1990) Biochim. Biophys. Acta 1049, 151-157.14. Wales, R., Richardson, P. T., Roberts, L. M., Woodland, H. R. &

Lord, J. M. (1991) J. Biol. Chem. 266, 19172-19179.15. Frankel, A. E., ed. (1988) Immunotoxins (Kluwer, Dordrecht, The

Netherlands).16. Vittetta, E. S., Fulton, R. J., May, R. D., Till, M. & Uhr, J. W.

(1987) Science 238, 1098-1104.17. Blattler, W. A., Lambert, J. M. & Goldmacher, V. S. (1989) Cancer

Cells 1, 50-55.18. Halling, K. C., Halling, A. C., Murray, E. E., Ladin, B. F., Hous-

ton, L. L. & Weaver, R. F. (1985) Nucleic Acids Res. 13, 8019-8033.

19. Lamb, I., Roberts, L. M. & Lord, J. M. (1985) Eur. J. Biochem.148, 265-270.

20. Stengele, I., Bross, P., Garces, X., Giray, J. & Rasched, I. (1990)J. Mol. Biol. 212, 143-149.

21. Fulton, R. J., Blakey, D. C., Knowles, P. P., Uhr, I. W., Thorpe,P. E. & Vitetta, E. S. (1986) J. Biol. Chem. 261, 5314-5319.

22. Richardson, P. T., Roberts, L. M., Gould, J. H. & Lord, J. M.(1988) Biochim. Biophys. Acta 950, 385-394.

23. Hussain, K., Bowler, C., Roberts, L. M. & Lord, J. M. (1989)FEBS Lett. 244, 383-387.

24. Clackson, T., Hoogenboom, H. R., Griffiths, A. D. & Winter, G.(1991) Nature (London) 352, 624-628.

Proc. Natl. Acad. Sci. USA 89 (1992)

Dow

nloa

ded

by g

uest

on

May

15,

202

1