A predicted three-dimensional structure for the human immunodeficiency virus binding domains of CD4 antigen

A predicted three-dimensional structure of the two N-terminalextracellular domains of human CD4 antigen, a cell surface glycoprotein,is reported. This region of CD4, particularly the first domain,has been identified as containing the binding region for theenvelope gp120 protein of the human immuno-deficiency virus.The model was predicted based on the sequence homology of eachdomain with the variable light chain of immunoglobulins. Theframework ß-sheet regions were taken from the crystalcoordinates of REI. For one region in the first domain of CD4there was an ambiguity in the alignment with REI and two alternatemodels are presented. Loops connecting the framework were modeledfrom fragments selected from a database of main chain coordinatesfrom all known protein structures. Residues identified as involvedin binding gp120 have been located in several other studieswithin the first domain of CD4. Epitopes from eight monoclonalantibodies have been mapped onto residues in both domains. Competitionof these antibodies with each other and with gp120 can be interpretedfrom the structural model.     

Expression of immunoglobulin and scavenger receptor superfamily domains as chimeric proteins with domains 3 and 4 of CD4 for ligand analysis

One approach to the analysis of leucocyte cell surface proteinsis to express their domains with part of another protein asa carrier. We report the use of two immunoglobulin superfamily(IgSF) domains from rat CD4 (CD4d3+4) in producing domains fromvarious superfamilies as chimeric proteins in Chinese hamsterovary cell lines. Four types of construct were successfullyexpressed containing: (i) the two IgSF domains of CD48; (ii)the IgSF domain of mb-1 which is part of the B cell antigenrecognition complex; (iii) a T cell receptor V domain; and (iv)the N-terminal domain of CD5 which belongs to the scavengerreceptor superfamily. This CD5 chimeric protein was antigenkfor a panel of CD5 mAbs showing that mAbs with functional effectsreacted with the N-terminal domain of CD5. The CD48 chimericprotein has been used both as multivalent complexes producedby crosslinking with mAbs recognizing CD4 and in a monomericform to analyse the kinetics of the interaction between CD48and CD2 [van der Merwe et al. (1993) EMBO J., 12, 4945–4954].     

Large-scale expression, purification and characterization of small fragments of thrombomodulin: the roles of the sixth domain and of methionine 388

Fragments of human thrombomodulin (TM) have been expressed inlarge quantities in the Pichia pastoris yeast expression systemand purified to homogeneity. Fermentation of P.pastoris resultedin yields of 170 mg/1 TM. Purification to homogeneity resultedin an overall 10% yield, so that quantities of –20 mgpurified fragments can be readily obtained. Smaller fragmentsof TM, such as the individual fourth or fifth domains, werenot active, nor were equimolar mixtures of the two domains.These results demonstrate that the fourth and fifth epidermalgrowth factor (EGF)–like domains together comprise thesmallest active fragment of TM. The fragment containing thefourth and fifth EGFlike domains (TMEGF(4–5)] had 10%the specific activity of rabbit TM. Comparison of the M388Lmutant TMEGF(4–5) fragment with the same mutant TMEGF(4–5–6)fragment showed that the fragment with the sixth domain hada 10–fold better Km value for thrombin than the fragmentthat did not contain the sixth domain; this factor completelyaccounts for the higher specific activity of the fragments containingthe sixth domain. Comparison of the wild–type and M388Lmutants showed that the M388L mutation resulted in a 2–foldincrease in k_cat for the activation of protein C by the thrombin–TMfragment complex, completely accounting for the 2–foldincrease in specific activity of these mutant fragments.     

Interaction between a Fab fragment against gp41 of human immunodeficiency virus 1 and its peptide epitope: characterization using a peptide epitope library and molecular modeling

The molecular interaction of the Fab fragment of the human monoclonalantibody 3D6, directed against the transmembrane protein gp41of human immunodeficiency virus (HTV) 1, with its peptide epitopeis characterized by a panel of overlapping peptides, a peptideepitope library and molecular modeling techniques. The sequenceCSGKLICTTAVPW, corresponding to amino acids 605–617 ofgp41, was identified as the best binding peptide (K_D = 1x10^-8mol/1). This peptide served as a starting point to prepare acellulose-bound peptide epitope library in which each residueof the epitope is substituted by all L- and D-amino acids, resultingin 494 epitope peptide variants which were subsequently analyzedfor binding 3D6. The library was synthesized to identify residuescritical for binding and to obtain information about the molecularenvironment of the epitope peptide bound to 3D6. Both cysteineresidues, as well as isoleucine 6, threonine 8 and proline 12,of the epitope were highly sensitive to substitution. Usingthe data obtained from the epitope characterization, as wellas a low-resolution electron density map of a 3D6 Fab-peptidecomplex, a 3-D model of the Fab-peptide complex was generatedby molecular modeling. The modeling experiments predict bindingof the peptide, which is cyclized via the two cysteine residues,to a pocket formed dominantly by the hypervariable loops complementaritydetermining regions CDR3L, CDR2H and CDR3H.     

Site-directed and deletion mutational analysis of the receptor binding domain of the interleukin-6 receptor targeted fusion toxin DAB389-IL-6

We have used site-directed and in-frame deletion mutationalanalysis in order to explore the structural features of theIL–6 portion of the diphtheria toxin-related interleukin–6(IL–6) fusion toxin DAB₃₈₉-IL–6 that are essentialfor receptorbinding and subsequent inhibition of protein synthesisin target cells. Deletion of the first 14 amino acids of theIL–6 component of the fusion toxin did not alter eitherreceptor binding affinity or cytotoxk potency. In contrast,both receptor binding and cytotoxic activity were abolishedwhen the C–terminal 30 amino acids of the fusion toxinwere deleted. In addition, we explored the relative role ofthe disulfide bridges within the IL–6 portion of DAB₃₈₉-IL–6in the stabilization of structure required for receptor-binding.The analysis of mutants in which the substitution of eitherCys⁴⁴⁰, Cys⁴⁴⁶, Cys⁴⁶⁹ or Cys⁴⁷⁹ to Ser respectively, demonstratesthat only the disulfide bridge between Cys⁴⁶⁹ and Cys⁴⁷⁹ isrequired to maintain a functional receptor binding domain. Inaddition, the internal in-frame deletion of residues 435–451,which includes Cys⁴⁴⁰ and Cys⁴⁴⁶, was found to reduce, but notabolish receptor binding affinity. These results further demonstratethat the disulfide bridge between Cys⁴⁴⁰ and Cys⁴⁴⁶ is not essentialfor receptor-binding. However, the reduced cytotoxic potencyof DAB₃₈₉-IL6(435–451) suggests that the conformationand/or receptor binding sites associated with this region ofthe fusion toxin is/are important for maintaining the wild typereceptor binding affinity and cytotoxic potency.     

Display of {beta}-lactamase on the Escherichia coli surface: outer membrane phenotypes conferred by Lpp'-OmpA'-{beta}-lactamase fusions

Bacterial cell-surface exposure of foreign peptides and solubleproteins has been achieved recently by employing a fusion proteinmethodology. An Lpp'–OmpA(46–159)–Bla fusionprotein has been shown previously to display the normally periplasmicenzyme ß-lactamase (Bla) on the cell surface of theGram-negative bacterium Escherichia coli. Here, we have investigatedthe role of the OmpA domain of the tripartite fusion proteinin the surface display of the passenger domain (Bla) and havecharacterized the effects of the fusion proteins on the integrityand permeability of the outer membrane. We show that in additionto OmpA(46–159), a second OmpA segment, consisting ofamino acids 46–66, can also mediate the display of Blaon the cell surface. Other OmpA domains of various lengths (aminoacids 46–84, 46–109, 46–128, 46–141and 46–145) either anchored the Bla domain on the periplasmicface of the outer membrane or caused a major disruption of theouter membrane, allowing the penetration of antibodies intothe cell. Detergent and antibiotic sensitivity and periplasmicleakage assays showed that changes in the permeability of theouter membrane are an unavoidable consequence of displayinga large periplasmic protein on the surface of E.coli. This isthe first systematic report on the effects that cell surfaceengineering may have on the integrity and permeability propertiesof bacterial outer membranes.     

Monovalent antibody scFv fragments selected to modulate T-cell activation by inhibition of CD86-CD28 interaction

Beside the interaction of the antigen-presenting major histocompatibility complex with the T-cell receptor, a co-stimulatory signal is required for T-cell activation in an immune response. To reduce immune-mediated graft rejection in corneal transplantation, where topical application of drugs in ointments or eye-drops may be possible, we selected single-chain antibody fragments (scFv) with binding affinity to rat CD86 (B7.2) that inhibit the co-stimulatory signal. We produced the IgV-like domain of rat CD86 as a fusion protein in Escherichia coli by refolding from inclusion bodies. This protein was used as a target for phage display selection of scFv from HuCAL-1, a fully artificial human antibody library. Selected binding molecules were shown to specifically bind to rat CD86 and inhibit the interaction of CD86 with CD28 and CTLA4 (CD152) in flow cytometry experiments. In an assay for CD86-dependent co-stimulation, the selected scFv fragment successfully inhibited the proliferation of T-cells induced by CD86-expressing P815 cells.     

Expression of active domains of a human folate-dependent trifunctional enzyme in Escherichia coli

The cDNA encoding the human trifunctional enzyme methylenetetrahydrofolatedehydrogenase-methenyltetrahydrofolate cyclohydrolase-formyltetrahydrofolatesynthetase was engineered to contain a prokaryotic ribosomebinding site and was expressed under the bacteriophage T7 RNApolymerase promoter in Escherichia coli. Site-directed mutagenesiswas used to prepare constructs that encode separately the dehydrogenase/cyclohydrolase(D/C) domain as amino acid residues 1–301, and the synthetase(Syn) domain as residues 304–935. Both domains formedactive enzymes thereby demonstrating their ability to fold independently.The full-length enzyme, D/C and Syn domains were expressed atlevels 4-, 55- and 3-fold higher than the specific activitiesfound in liver. Additional mutagenesis and independent expressionof domains further defined the interdomain region to includeamino acids 292–310. The D/C domain was purified to homogeneityby a single affinity chromatographic step, and the full-lengthprotein in a twostep procedure. The kinetic properties of theD/C domain appear unaltered from those of the trifunctionalenzyme.     

A mutational analysis of receptor binding sites of interleukin-1{beta}:differences in binding of human interleukin-1{beta} muteins to human and mouse receptors

The 3-D crystal structure of interleukin-1ß(IL-1ß)has been used to define its receptor binding surface by mutationalanalysis. The surface of IL-1ß was probed by site-directedmutagenesis. A total of 27 different IL-1ß muteinswere constructed, purified and analyzed. Receptor binding measurementson mouse and human cell lines were performed to identify receptoraffinities. IL-1ß muteins with modified receptor affinitywere evaluated for structural integrity by CD spectroscopy orX-ray crystallography. Changes in six surface loops, as wellas in the C- and N-termini, yielded muteins with lower bindingaffinities. Two muteins with intact binding affinities showed10- to 100-fold reduced biological activity. The surface regioninvolved in receptor binding constitutes a discontinuous areaof 1000 Å2 formed by discontinuous polypeptide chain stretches.Based on these results, a subdivision into two distinct localareas is proposed. Differences in receptor binding affinitiesfor human and mouse receptors have been observed for some muteins,but not for wild-type IL-1ß. This is the first timea difference in binding affinity of IL-1ß muteinsto human and mouse receptors has been demonstrated     

Modelling of the disulphide-swapped isomer of human insulin-like growth factor-1: implications for receptor binding

Insulin-like growth factor-1 (IGF-1) is a serum protein whichunexpectedly folds to yield two stable tertiary structures withdifferent disulphide connectivities; native IGF-1 [18–61,6–48,47–52]and IGF-1 swap [18–61,6–47, 48–52]. Here we demonstratein detail the biological properties of recombinant human nativeIGF-1 and IGF-1 swap secreted from Saccharomyces cerevisiae.IGF-1 swap had a ~30 fold loss in affinity for the IGF-1 receptoroverexpressed on BHK cells compared with native IGF-1.The parallelincrease in dose required to induce negative cooperativity togetherwith the parallel loss in mitogenicity in NIH 3T3 cells impliesthat disruption of the IGF-1 receptor binding interaction ratherthan restriction of a post-binding conformational change isresponsible for the reduction in biological activity of IGF-1swap. Interestingly, the affinity of IGF-1 swap for the insulinreceptor was ~200 fold lower than that of native IGF-1 indicatingthat the binding surface complementary to the insulin receptor(or the ability to attain it) is disturbed to a greater extentthan that to the IGF-1 receptor. A 1.0 ns high-temperature moleculardynamics study of the local energy landscape of IGF-1 swap resultedin uncoiling of the first A-region -helix and a rearrangementin the relative orientation of the A- and B-regions. The modelof IGF-1 swap is structurally homologous to the NMR structureof insulin swap and CD spectra consistent with the model arepresented. However, in the model of IGF-1 swap the C-regionhas filled the space where the first A-region -helix has uncoiledand this may be hindering interaction of Val44 with the secondinsulin receptor binding pocket.     

Improved design of an antigen with enhanced specificity for the broadly HIV-neutralizing antibody b12

In an attempt to design immunogens that elicit broadly HIV-neutralizing antibodies, we recently engineered monomeric HIV-1 gp120 to bind preferentially b12, a broadly neutralizing antibody to the CD4-binding site (CD4bs) on gp120, by mutating four central residues in the CD4bs to alanine and introducing extra N-glycosylation sites potentially to mask unwanted B-cell epitopes. Despite the favorable antigenicity of this mutant, it harbors two potential caveats that may limit its effectiveness to elicit b12-like antibodies: (i) b12-binding affinity is reduced relative to wild-type gp120 and (ii) binding of some non-neutralizing antibodies to the N-terminal C1 region of gp120 is still observed. Here, we sought to correct these potential limitations. By reverting one of the added N-glycosylation sites on the gp120 core, b12 binding was improved without affecting the epitope-masking properties of the original mutant. Furthermore, truncation of the gp120 N-terminus eliminated binding of the anti-C1 antibodies. Finally, based on the binding profiles of additional non-neutralizing antibodies tested here, further N-glycosylation sites were incorporated to mask their corresponding epitopes. The resulting hyperglycosylated gp120 variants bind b12 and another broadly neutralizing antibody, 2G12, with apparent affinities approaching that of wild-type gp120, but do not bind 21 non- or weakly neutralizing antibodies to seven different epitopes on gp120. These hyperglycosylated variants expand our panel of glycoengineered gp120s that are currently being evaluated for their ability to elicit broadly neutralizing antibodies.     

The position of the structurally autonomous kringle 2 domain influences the functional features of tissue-type plasminogen activator

Tissue-type plasminogen activator (t-PA) is composed of structurallyautonomous domains. From the N-terminus of t-PA, a finger-likedomain (F), an epidermal growth factor-like domain (G), twokringle domains (Kl and K2) and a serine protease domain (P)can be discerned. The K2 domain of t-PA is known to be involvedin lysine binding, fibrin binding and fibrin-dependent plasminogenactivation. To study the functional autonomy of the K2 domainin t-PA we constructed, with the aid of a cassette t-PA gene[Rehberg et al. (1989) Protein Engng, 2,371–377], mutantt-PA genes coding for four molecules (FGK1K2P, FGK2K1P, GK1K2Pand GK2K1P) in which the K2 domain was placed in two differentpositions in t-PA. The DNAs of wild-type t-PA and the t-PA variantswere expressed in Chinese hamster ovary cells and the recombinantproteins were purified by affinity chromatography.All moleculeswere expressed in their single-chain form and could be convertedto their two-chain form. With these molecules, lysine binding,fibrin binding and fibrin-dependent plasminogen activation werestudied. All variants showed affinity for lysyl-Sepharose andaminohexyl-Sepharose. Reversal of the K domains (FGK2K1P versusFGK2K1P and GK1K2P versus GK2K1P) resulted in a 23–47%weaker interaction to both lysyl-Sepharose and aminohexyl-Sepharose.Deleting the F domain (FGK1K2P versus GK1K2P and FGK2K1P versusGK2K1P) resulted in a 20–70% improvement of the interactionslysyl-Sepharose and aminohexyl-Sepharose. All variants boundto a forming fibrin clot. Reversal of the K domains (FGK1K2Pversus FGK2K1P) reduced fibrin binding. In the presence of thelysine analogue -amino caproic acid, only FGK1K2P bound to fibrin.All variants activated plasminogen. In the absence of fibrinogenCNBr fragments (mimic of fibrin), the reversal of the K domain(FGK2K1P) resulted in a 2-fold improved plasminogen activation.In the presence of a fibrin mimic, the plasminogen activationsof the F domain deletion analogues GK1K2P and GK2K1P were foundto be decreased 2- to 4-fold. From these results we concludedthat the function of t-PA in lysine binding, fibrin bindingand fibrin-dependent plasminogen activation is dependent onthe correct spatial orientation of the K2 domain within thet-PA molecule     

Molecular mechanics analysis of inhibitor binding to HIV-1 protease

Crystallographic structures of HIV protease with three differentpeptide-mimetic inhibitors were subjected to energy minimizationusing molecular mechanics, the minimized structures analyzedand the inhibitor binding energies calculated. Partial chargeassignment for the hydrogen bonded catalytic aspartk acids,Asp25 and -25', was in good agreement with charge calculationsusing semi-empirical molecular orbital methods. Root mean squaredeviations on minimization were small and similar for both subunitsin the protease dimer. The surface loops, which had the largestB factors, changed most on minimization; the hydrophobic coreand the inhibitor binding site showed little change. The distance-dependentdielectric of D(r) = 4r was found to be preferable to D(r) =r. Distance restraints were applied for the intermolecular hydrogenbonds to maintain the conformation of the inhibitor bindingsite. Using the dielectric of D(r) = 4r, the calculated interactionenergy of the three inhibitors with the protease ranged from–53 to –56 kcal/mol. The groups of the inhibitorswere changed to add or remove a ‘transition state analogue’hydroxyl group, and the loss in energy on the removal of thisgroup was calculated to be 0.9–1.7 kcal/mol. This wouldrepresent 19–36% of the total measured difference in bindingenergy between the inhibitors JG365 and MVT-101.     

Structural analysis of the CD2 T lymphocyte antigen by site-directed mutagenesis to introduce a disulphide bond into domain 1

Many proteins have been predicted to contain domains with immunoglobulin-Iikefolds and hence to be members of the immunoglobulin superfamily(IgSF). However, several members lack the Cys residues capableof forming the disulphide bond that forms a characteristic bridgebetween the ß sheets in the Ig fold, e.g. domain 1of the lymphocyte antigen CD2. The assignment of ßstrands in CD2 by sequence analysis was tested by attemptingto introduce a disulphide bridge between the ß sheetsby mutating two residues in the relevant positions to Cys. Mutant,soluble forms of CD2 were expressed in Chinese hamster ovarycell lines and amino add sequencing showed that a disulphidebond was formed as predicted, but not in the control where oneCys residue was misplaced by four residues. Evidence that bothmutated molecules folded correctly is given by the indistinguishablebinding of three monoclonal antibodies recognising differentepftopes on CD2. The 3-D structure of rat CD2 domain 1 has beendetermined by NMR spectroscopy and X-ray crystallography, confirmingthe predictions from the sequence. Applications of this methodof insertion of disulphide residues for probing protein structuresare discussed, together with other structures of IgSF domainslacking the typical inter-sheet disulphide bond.     

Domain organization of the extracellular region of CD45

CD45 is a large, heavily glycosylated, transmembrane proteinphosphotyrosine phosphatase found on all nucleated cells ofhaematopoietic origin. In lymphocytes, the cytoplasmic phosphataseis necessary for efficient signalling through the antigen receptorbut in contrast little is known about the interactions of theextracellular region of the molecule. This consists of a mucin-likeregion, a novel cysteine-containing region and a region containingthree putative fibronectin type III domains. To confirm thisorganization and to identify parts potentially important forfunction, we have expressed fragments of the extracellular domainof rat CD45 as recombinant soluble proteins. Proteins correspondingto two, three and four domains of CD45 were expressed in secretedforms. Single domains and constructs for proteins with truncationsof the predicted domains were not expressed. This is consistentwith the proposed structural organization. Determination ofthe positions of the disulphide bonds in the N-terminal cysteine-containingregion and the first fibronectin type III domain identifiednovel disulphide bonds within the fibronectin type III domainand an unusual inter-domain disulphide linkage. Circular dichroismspectroscopy indicated that this region of rat CD45 has mainlyß-strand secondary structure and no -helical content.These studies support the proposed domain organization of CD45.     

Synthesis, cloning and expression of the single-chain Fv gene of the HPr-specific monoclonal antibody, Jel42. Determination of binding constants with wild-type and mutant HPrs

The monoclonal antibody Jel42 is specific for the Escherichiacoli histidine-containing protein, HPr, which is an 85 aminoacid phosphocarrier protein of the phosphoenolpyruvate:sugarphosphotransferase system. The binding domain (Fv) has beenproduced as a single chain Fv (scFv). The scFv gene was synthesizedin vitro and coded for pelB leader peptide–heavy chain–linker–lightchain–(His)₅ tail. The linker is three repeats from theC-terminal repetitive sequence of eukaryotic RNA polymeraseII. This linker acts as a tag; it is the antigen for the monoclonalantibody Jel352. The codon usage was maximized for E.coli expression,and many unique restriction endonuclease sites were incorporated.The scFv gene incorporated into pT7-7 was highly expressed,yielding 10–30% of the cell protein as the scFv, whichwas found in inclusion bodies with the leader peptide cleaved.Jel42 scFv was purified by denaturation/renaturation yieldingpreparations with K_d values from 20 to 175 nM. However, basedupon an assessment of the amount of active refolded scFv, thebinding dissociation constant was estimated to be 2.7 ±2.0 nM compared with 2.8 ± 1.6 and 3.7 ± 0.3 nMpreviously determined for the Jel42 antibody and Fab fragmentrespectively. The effect of mutation of the antigen HPr on thebinding constant of the scFv was very similar to the propertiesdetermined for the antibody and the Fab fragment. It was concludedthat the small percentage (~6%) of refolded scFv is a true mimicof the Jel42 binding domain and that the incorrectly foldedscFv cannot be detected in the binding assay.     

Engineering the C-region of human insulin-like growth factor-1: implications for receptor binding

Recombinant wild-type human IGF-1 and a C-region mutant in whichresidues 28–37 have been replaced by a 4-glycine bridge(4-Gly IGF-1) were secreted and purified from yeast. An IGF-1analogue in which residues 29–41 of the C-region havebeen deleted (mini IGF-1) was created by site-directed mutagenesisand also expressed. All three proteins adopted the insulin-foldas determined by circular dichroism. The significantly raisedexpression levels of mini IGF-1 allowed the recording of two-dimensionalNMR spectra. The affinity of 4-Gly IGF-1 for the IGF-1 receptorwas {small tilde}100-fold lower than that of wild-type IGF-1and the affinity for the insulin receptor was {small tilde}10-foldlower. Mini IGF-1 showed no affinity for either receptor. Notonly does the C-region of IGF-1 contribute directly to thefree energy of binding to the IGF-1 receptor, but also the absenceof flexibility in this region eliminates binding altogether.As postulated for the binding of insulin to its own receptor,it is proposed that binding of IGF-1 to the IGF-1 receptor alsoinvolves a conformational change in which the C-terminal B-regionresidues detach from the body of the molecule to expose theunderlying A-region residues.     

A comparison of two murine monoclonal antibodies humanized by CDR-grafting and variable domain resurfacing

The variable domain resurfacing and CDR-grafting approachesto antibody humanization were compared directly on the two murinemonoclonal antibodies N901 (anti-CD56) and anti-B4 (anti-CD19).Resurfacing replaces the set of surface residues of a rodentvariable region with a human set of surface residues. The methodof CDR-grafting conceptually consists of transferring the CDRsfrom a rodent antibody onto the Fv framework of a human antibody.Computer-aided molecular modeling was used to design the initialCDR-grafted and resurfaced versions of these two antibodies.The initial versions of resurfaced N901 and resurfaced anti-B4maintained the full binding affinity of the original murineparent antibodies and further refinements to these versionsdescribed herein generated five new resurfaced antibodies thatcontain fewer murine residues at surface positions, four ofwhich also have the full parental binding affinity. A mutationalstudy of three surface positions within 5 Å of the CDRsof resurfaced anti-B4 revealed a remarkable ability of the resurfacedantibodies to maintain binding affinity despite dramatic changesof charges near their antigen recognition surfaces, suggestingthat the resurfacing approach can be used with a high degreeof confidence to design humanized antibodies that maintain thefull parental binding affinity. By comparison CDR-grafted anti-B4antibodies with parental affinity were produced only after seventeenversions were attempted using two different strategies for selectingthe human acceptor frameworks. For both the CDR-grafted anti-B4and N901 antibodies, full restoration of antigen binding affinitywas achieved when the most identical human acceptor frameworkswere selected. The CDR-grafted anti-B4 antibodies that maintainedhigh affinity binding for CD19 had more murine residues at surfacepositions than any of the three versions of the resurfaced anti-B4antibody. This observation suggests that the resurfacing approachcan be used to produce humanized antibodies with reduced antigenicpotential relative to their corresponding CDR-grafted versions.