2003; baxter - Supplements - Haematologica

2003; baxter - Supplements - Haematologica 2003; baxter - Supplements - Haematologica

supplements.haematologica.org
from supplements.haematologica.org More from this publisher
09.07.2015 Views

22J. Voorberg et al.hemophilia A. 23,26 Independently, Arai and coworkersisolated 3 human antibodies from aphage library of an unrelated patient. 29 Epstein-Barr immortalization of peripheral B-lymphocyteswas used to isolate a human monoclonalantibody directed against the C2 domain of factorVIII. 30 Sequence analysis of the VH domain ofthese clones reveals some interesting features.Germline gene segment DP-5 (1-24) of the VH1family is used by about half of the anti-C2 antibodiesthat have been isolated so far. 23,26,29,30 TheVH domain of the remainder of human antibodiesdirected against the C2 domain is derivedfrom germline gene segments DP-10 (1-69), DP-14 (1-18) and DP-88 (1-e). 23,26 Phylogeneticanalysis of gene segments belonging to the VH1family shows that DP-10 (1-69), DP-14 (1-18)and DP-88 (1-e) are present within one branchof this family. 26 Consistent with the more extensivedifferences in nucleotide sequence, DP-5 (1-24) is present within a distinct branch of theVH1 family (see Figure 3). The available data suggestthat based on the characteristics of their VHdomains, anti-C2 antibodies can be subdividedin two different classes. In a previous section itwas proposed that anti-A2 and anti-A3 antibodiesare derived from VH gene segments that arecommonly used in the immunoglobulin repertoire.Both DP-10 (1-69) and DP-14 (1-18) areused in approximately 3-5% of peripheral IgG + Bcells whereas DP-5 (1-24) and DP-88 (1-e)occur in less then 1% the total IgG + repertoire(data extracted from ref. #21). These percentagesindicate that these germline gene segmentsare not dominantly expressed in IgG + peripheralB-cells. The restricted usage of VH gene segmentsDP-5 (1-24), DP-10 (1-69), DP-14 (1-18) andDP-88 (1-e) for assembly of anti-C2 antibodiessuggests that B cells expressing human antibodieswith these VH gene segments are positivelyselected following exposure of naïve B-cells tofactor VIII.Structural elements present within the VH genefamilies may facilitate interaction of antibodiescontaining these gene segments with the C2domain. Based on the three-dimensional structureof a large collection of antibodies an overalltopology of antigen binding sites has been proposedby Chothia and co-workers. 22 Six antigenbinding loops derived from both heavy (H1, H2,H3) and light chains (L1, L2, L3) have beendefined that partially overlap with the hypervariablecomplementary determining regions(CDRs) 15,31 (Figure 4A). Both the H1 and H2region are contained within the VH gene segmentof an antibody. For region H1 three socalledcanonical structures have been defined thatdiffer in the number of amino acids containedwithin the H1 region. Similarly, the existence offour different canonical structures for the H2region has been proposed. 22 Interestingly, bothgermline gene segments DP-10 (1-69), DP-14(1-18) and DP-88 (1-e) possess the H1-H2canonical structure 1-2. This combination ofcanonical structures is observed in only 7 out of51 VH gene segments. Four of them [DP-3 (1-f),DP-10 (1-69), DP-14 (1-18) and DP-88 (1-e)]belong to the VH1 family whereas DP-73 (5-51),5-a and DP-21 (7-4.1) belong to the VH5 andVH7 family. 15 Although regions H1 and H2determine only part of the architecture of anantigen binding site, the presence of canonicalstructure 1-2 may facilitate interactions of VHdomains with antigenic determinants present inthe C2 domain. A canonical structure for the H2region of germline gene segment DP-5 (1-24),could not be defined due to the presence of anglutamic acid instead of the more usual threonine,alanine or leucine at position 71. 22,32 Interestingly,in three of out of four human antibodiesthat are encoded by the germline gene segmentDP-5 (1-24) amino acid substitutions havetaken place at this position 26,30 (Figure 4A).Replacement of the glutamic acid for an alanineFigure 3. Phylogenetic tree of the VH1 family. Anti-C2 antibodies are derived from two classes of VHgene segments. The first class comprises VH genesegments DP-10 (1-69), DP-14 (1-18) and DP-88 (1-e). The second class of anti-C2 antibodies incorporatesVH gene segment DP-5 (1-24).haematologica vol. 88(supplement n. 12):september 2003

IV International Workshop on Immune Tolerance in Hemophilia 23at position 71 may affect the canonical structureof the VH domains of the DP-5 (1-24) encodedantibodies which now becomes identical to thatof the VH domains of the other anti-C2 antibodies(i.e. 1-2). The above analysis suggests thatrestricted usage of VH segments may endow anti-C2 antibodies with unique structural featuresthat may play a role in creating an appropriateparatope.In the previous section evidence was presentedfor the existence of two classes of human antibodiesthat bind to the C2 domain of factor VIII.The first class contains VH domains that arederived from the closely related VH gene segmentsDP-10 (1-69), DP-14 (1-18) and DP-88(1-e) whereas the second class of human antibodiesis characterized by the presence ofgermline gene segment DP-5 (1-24). 26 Epitopemapping studies using a series of human/porcineand factor V/VIII hybrids reveals that these twoclasses of antibodies bind to distinct sites withinthe C2 domain of factor VIII. Residues contributingto the epitope of human antibodiesencoded by VH segments DP-10 (1-69), DP-14(1-18) and DP-88 (1-e) resides in the carboxylterminal part of the C2 domain (residues 2243-2332). In contrast, human antibodies containingVH segment DP-5 (1-24) interact with residuesat the amino- and carboxyl terminal part of theC2 domain. Recently, the three-dimensionalstructure of the C2 domain of factor VIII wasdetermined. 33 Based on the structure it was proposedthat three loops containing hydrophobicresidues intersect the phospholipid bilayer whereasa layer of positively charged residues can interactwith negatively charged phosphate moietiesof phosphatidylserine. 33 In a follow-up study thethree-dimensional structure of a complex of thehuman monoclonal antibody BO2C11 and theC2 domain was determined at 2.0Å 34 (Figure 4B).The VH domain of BO2C11 is encoded by VHgene segment DP-5. 30 Figure 4A shows an alignmentof the amino acid sequence of BO2C11with the DP-5 (1-24) encoded antibodies wehave recently isolated. 26 All four DP-5 (1-24)encoded antibodies have a relatively small CDR3which comprises 8-11 amino acids. The patternof somatic hypermutation varies considerablybetween BO2C11 and the clones isolated byphage display, which is not surprising, since theseantibodies were derived from different patients.A common replacement between 3 of the 4 antibodiesis a substitution of glutamic acid at position71 for an alanine. This suggests that clonesharboring this amino acid replacement are preferentiallyselected from the total pool of C2domain reactive B-cells. A possible structuralbasis for this selective enrichment has been discussedin a previous paragraph. Based on thethree-dimensional structure of BO2C11 and theC2 domain, multiple atomic contacts have beendetected between the heavy chain of BO2C11and the C2 domain (Figure 4A/4B). 34 Comparisonof the amino acid sequence of the WR cloneswith that of BO2C11 reveals that the majority ofcontact residues are conserved in the sequence ofthe WR clones (see Figure 4A). This suggests thatthe interactive surface of the complex betweenBO2C11 and the C2 domain is representative forother DP-5 (1-24) encoded anti-C2 antibodies.Small differences between binding of individualDP-5 (1-24) encoded antibodies to the C2domain can, however, occur. The DP-5 (1-24)encoded residue Asp 52 of BO2C11 that interactswith Arg 2215 in the C2 domain has been substitutedfor an asparagine or alanine in the VHdomains of the WR clones (see Figure 4A). Alsoresidue Asp 99 present in the CDR3 of BO2C11(indicated by an arrow; Figure 4A) is not conservedamong the WR clones. Since several additionalnegatively charged amino acids are presentin both CDR2 and CDR3 of WR1, WR16 andWR17, the effect of these substitutions on theoverall architecture of the interactive surface ismost likely limited. Single amino acid substitutionsin the C2 domain have recently been evaluatedfor their reactivity with factor VIIIinhibitors. 35 A reduction in functional inhibitionof human antibodies was observed when residuesMet 2199 /Phe 2200 (Loop I; Figure 4B) and Leu 2252(Loop II; Figure 4B) were substituted for an Ile,Leu and Phe, respectively. 35 Combination of thesesubstitutions resulted in a further decrease in theantigenicity of the C2 domain. The significanceof electrostatic interactions mediated by Arg 2210and Arg 2215 for binding to factor VIII inhibitorshas not yet been assessed. Residues Arg 2210 andArg 2215 are not critical for binding of factor VIIIto phospholipids. 36 Factor VIII variants harboringsubstitutions at these sites may, therefore, displaya reduced antigenicity while maintainingtheir phospholipid binding properties.Conclusions and implicationsOverall, our data show that phage display is auseful technology for isolating large numbers ofhuman monoclonal antibodies from the repertoireof patients with inhibitors. Inspection ofthe primary amino acid sequence of these antibodiessuggest that only a restricted number ofvariable heavy chain segments is used for theassembly of human antibodies that react withthe C2 domain. In contrast, less restriction isobserved for anti-A2 and anti-A3 antibodies, atleast at the level of VH gene segments. The availabilityof a large panel of human antibodiesderived from different patients is likely toincrease our knowledge on the number and complexityof B-cell epitopes on factor VIII. Subsequentmodification of antigenic sites within theA2, A3 and C2 domains may provide a basis forthe reduction of the antigenicity and perhaps alsoimmunogenicity of factor VIII. 4 Alternatively, itmay be possible to develop antibody-basedreagents that interfere with the binding ofinhibitory antibodies to factor VIII. In this respectit is worth mentioning that the majority of variabledomain antibody fragments do not interferehaematologica vol. 88(supplement n. 12):september 2003

IV International Workshop on Immune Tolerance in Hemophilia 23at position 71 may affect the canonical structureof the VH domains of the DP-5 (1-24) encodedantibodies which now becomes identical to thatof the VH domains of the other anti-C2 antibodies(i.e. 1-2). The above analysis suggests thatrestricted usage of VH segments may endow anti-C2 antibodies with unique structural featuresthat may play a role in creating an appropriateparatope.In the previous section evidence was presentedfor the existence of two classes of human antibodiesthat bind to the C2 domain of factor VIII.The first class contains VH domains that arederived from the closely related VH gene segmentsDP-10 (1-69), DP-14 (1-18) and DP-88(1-e) whereas the second class of human antibodiesis characterized by the presence ofgermline gene segment DP-5 (1-24). 26 Epitopemapping studies using a series of human/porcineand factor V/VIII hybrids reveals that these twoclasses of antibodies bind to distinct sites withinthe C2 domain of factor VIII. Residues contributingto the epitope of human antibodiesencoded by VH segments DP-10 (1-69), DP-14(1-18) and DP-88 (1-e) resides in the carboxylterminal part of the C2 domain (residues 2243-2332). In contrast, human antibodies containingVH segment DP-5 (1-24) interact with residuesat the amino- and carboxyl terminal part of theC2 domain. Recently, the three-dimensionalstructure of the C2 domain of factor VIII wasdetermined. 33 Based on the structure it was proposedthat three loops containing hydrophobicresidues intersect the phospholipid bilayer whereasa layer of positively charged residues can interactwith negatively charged phosphate moietiesof phosphatidylserine. 33 In a follow-up study thethree-dimensional structure of a complex of thehuman monoclonal antibody BO2C11 and theC2 domain was determined at 2.0Å 34 (Figure 4B).The VH domain of BO2C11 is encoded by VHgene segment DP-5. 30 Figure 4A shows an alignmentof the amino acid sequence of BO2C11with the DP-5 (1-24) encoded antibodies wehave recently isolated. 26 All four DP-5 (1-24)encoded antibodies have a relatively small CDR3which comprises 8-11 amino acids. The patternof somatic hypermutation varies considerablybetween BO2C11 and the clones isolated byphage display, which is not surprising, since theseantibodies were derived from different patients.A common replacement between 3 of the 4 antibodiesis a substitution of glutamic acid at position71 for an alanine. This suggests that clonesharboring this amino acid replacement are preferentiallyselected from the total pool of C2domain reactive B-cells. A possible structuralbasis for this selective enrichment has been discussedin a previous paragraph. Based on thethree-dimensional structure of BO2C11 and theC2 domain, multiple atomic contacts have beendetected between the heavy chain of BO2C11and the C2 domain (Figure 4A/4B). 34 Comparisonof the amino acid sequence of the WR cloneswith that of BO2C11 reveals that the majority ofcontact residues are conserved in the sequence ofthe WR clones (see Figure 4A). This suggests thatthe interactive surface of the complex betweenBO2C11 and the C2 domain is representative forother DP-5 (1-24) encoded anti-C2 antibodies.Small differences between binding of individualDP-5 (1-24) encoded antibodies to the C2domain can, however, occur. The DP-5 (1-24)encoded residue Asp 52 of BO2C11 that interactswith Arg 2215 in the C2 domain has been substitutedfor an asparagine or alanine in the VHdomains of the WR clones (see Figure 4A). Alsoresidue Asp 99 present in the CDR3 of BO2C11(indicated by an arrow; Figure 4A) is not conservedamong the WR clones. Since several additionalnegatively charged amino acids are presentin both CDR2 and CDR3 of WR1, WR16 andWR17, the effect of these substitutions on theoverall architecture of the interactive surface ismost likely limited. Single amino acid substitutionsin the C2 domain have recently been evaluatedfor their reactivity with factor VIIIinhibitors. 35 A reduction in functional inhibitionof human antibodies was observed when residuesMet 2199 /Phe 2200 (Loop I; Figure 4B) and Leu 2252(Loop II; Figure 4B) were substituted for an Ile,Leu and Phe, respectively. 35 Combination of thesesubstitutions resulted in a further decrease in theantigenicity of the C2 domain. The significanceof electrostatic interactions mediated by Arg 2210and Arg 2215 for binding to factor VIII inhibitorshas not yet been assessed. Residues Arg 2210 andArg 2215 are not critical for binding of factor VIIIto phospholipids. 36 Factor VIII variants harboringsubstitutions at these sites may, therefore, displaya reduced antigenicity while maintainingtheir phospholipid binding properties.Conclusions and implicationsOverall, our data show that phage display is auseful technology for isolating large numbers ofhuman monoclonal antibodies from the repertoireof patients with inhibitors. Inspection ofthe primary amino acid sequence of these antibodiessuggest that only a restricted number ofvariable heavy chain segments is used for theassembly of human antibodies that react withthe C2 domain. In contrast, less restriction isobserved for anti-A2 and anti-A3 antibodies, atleast at the level of VH gene segments. The availabilityof a large panel of human antibodiesderived from different patients is likely toincrease our knowledge on the number and complexityof B-cell epitopes on factor VIII. Subsequentmodification of antigenic sites within theA2, A3 and C2 domains may provide a basis forthe reduction of the antigenicity and perhaps alsoimmunogenicity of factor VIII. 4 Alternatively, itmay be possible to develop antibody-basedreagents that interfere with the binding ofinhibitory antibodies to factor VIII. In this respectit is worth mentioning that the majority of variabledomain antibody fragments do not interferehaematologica vol. 88(supplement n. 12):september <strong>2003</strong>

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!