HYBRIDOMAVolume 20 Number 4 2001Mary Ann Liebert Inc

Immunogenetic Analysis of Variable Regions Encoding AB1and g-Type AB2 Antibodies from the NeuGc-Containing

Ganglioside Family

ALEXIS PEREZ1 JOSEFA LOMBARDERO1 CRISTINA MATEO1 GERAUDIS MUSTELIER1

MAURO ALFONSO1 ANA M VAZQUEZ1 and ROLANDO PEREZ1

ABSTRACT

The variable regions from P3 a murine monoclonal antibody (MAb) against NeuGc-containing gangliosidesand two anti-idiotype MAbs directed to P3 MAb were cloned and sequenced Comparisons with previouslyreported sequences showed that P3 is a germline antibody encoded by genes from the VHQ52 and Vk19 fam-ilies Analysis of nucleotides at the heavy chain CDR3 (H-CDR3) showed the presence of an extensive 39 Nregion that contains almost 50 of the nucleotides of this CDR In addition amino acid sequence analysis ofthe H-CDRs of this MAb revealed the presence of three arginines two of which are present in the H-CDR3that could be involved in the interaction of P3 MAb with its electronegative epitope on gangliosides Anti-id-iotype 1E10 which seems to define a ldquoregulatoryrdquo idiotope on P3 MAb (it induces Id1 Ab3) represents agermline Ab2 that belongs to the VHJ558 and Vk10 gene families By contrary the anti-idiotype 3B11 is anextensively mutated antibody that belongs to the VH3660 and Vk45 gene families defining a ldquoprivaterdquo id-iotope on P3 MAb Even when different V genes contribute to the variable regions of 1E10 and 3B11 MAbsthey share an acidic motif ED-D-YD-Y-D in H-CDR3 suggesting that both Ab2s recognize paratope posi-tive residues on the Ab1 Therefore complementary electrostatic interactions involving H-CDR3 from bothAb1 and Ab2 might provide a clue to understand the molecular basis for the generation of g-type anti-idio-type antibodies to V regions recognizing glycolylated ganglioside antigens

211

INTRODUCTION

GANGLIOSIDES are sialic acid-containing glycolipids that arenormal components of mammalian tissues They are con-

sidered as attractive targets for cancer immunotherapy in hu-mans due to changes that occur in their expression patterns dur-ing oncogenic transformation Whereas N-glycolyl neuraminicacid (NeuGc) is absent in normal human tissues(1) the expres-sion of NeuGc-containing gangliosides in some human tumorshas been demonstrated(2ndash4) Murine monoclonal antibody(MAb) P3 (IgM k) generated by our group in BALBc micereacts specifically with a broad battery of N-glycolyl-contain-ing gangliosides sulfated glycolipids and with antigens ex-pressed in human breast tumors(5) Immunochemical experi-ments and molecular modeling have led to the delineation of

the epitope recognized by this MAb which includes the car-boxyl and glycolic hydroxyl groups together with the nitrogenfunction of sialic acid(6) Previous work in syngeneic model hasshown the capacity of this MAb to induce an IgG auto-anti-id-iotypic response even when injected alone results that were notobtained when other murine IgM MAbs were used(7)

Looking for internal image anti-idiotypic MAbs (Ab2) weimmunized BALBc mice with P3 MAb coupled to KLH in thepresence of Freundrsquos adjuvant Different MAbs were selectedand cloned for further studies All these antibodies inhibit thebinding of P3 MAb to GM3(NeuGc) and induced anti-anti-id-iotypic antibodies (Ab3) in syngeneic animals but were unableto generate Ab3 with the same specificity of P3 MAb(7) Theseresults suggested that these antibodies were not internal imageAb2 and they were classified as g-type Ab2 These g-type Ab2s

1Department of Antibody Engineering Center of Molecular Immunology PO Box 16040 Havana 11600 Cuba

could be distinguished by their different abilities to induce Ab3sharing idiotopes with P3 MAb in syngeneic animals (Id1

Ab3)In this paper we try to relate those immunochemical prop-

erties with the immunogenetic analysis of the primary struc-tures of P3 MAb and two of the anti-P3 Ab2s previously re-ported 1E10 Ab2 which was able to induce Id1 Ab3 responsein syngeneic model and 3B11 Ab2 that showed the highestbinding affinity for P3 MAb but was unable to induce Id1

Ab3(7) We determined the nucleotide sequences of the variableregions of the heavy and light chains (VH VL) of these anti-bodies by cloning cDNAs derived from the RNA extracted fromhybridoma cells The frequency of complementary chargedresidues on both Ab1 and Ab2 suggests that electrostatic in-teractions contribute to the specificity and high affinity of theg-type idiotypic recognition in the V-region repertoire recog-nizing gangliosides

MATERIALS AND METHODS

Ab1 and Ab2 MAbs

P3 MAb (IgM k) (Ab1) was generated by immunizationof BALBc mice with lyposomes containing NeuGc-GM3 andtetanus toxoid as previously described(5) Syngeneic Ab2MAbs 1E10 (IgG1 k) and 3B11 (IgG1 k) were obtained from BALBc mice immunized with purified P3 MAb cou-pled to KLH (Keyhole Limpet Haemocyanin) in Freundrsquos ad-juvant(7)

Primers

The following primers were used

(1) CM1FOR (39 primer for the heavy chain IgM)59-GGAAGCTTAAGACATTTGGGAAGGACTGAC-TCTC-39

(2) CG1FOR (39 primer for the heavy chain IgG1)59-GGAAGCTTAGACAGATGGGG GTGTCGTTTTG-39

(3) CK2FOR (39 primer for the kappa light chain)59-GGAAGCTTGAAGATGGATACAGTTGGTG -CAGC-39

(4) McKAsXBA (39 primer for the kappa light chain)59-GCGTCTAGAACTGGATGGTGGGAAGATGG-3 9

(5) VH1BACK (59 primer for the heavy chain framework re-gion 1 (FR1))59-AGGT(GC)(AC)A(AG)CTGCAG(GC)AGTC(AT)-GG-39

(6) VK8BACK (59 primer for the VK8 light chain FR1)59-C(AT)GAGAAATTCAGCTGACCCAGTCTC-3 9

(7) VK1BACK (59 primer for the VK1 light chain FR1)59-GACATTCAGCTGACCCAGTCTCCA-3 9

(8) MLALT4RV (59 primer for the light chain signal se-quence)59-GGGGATATCCACCATGAGG(GT)CCCC(AT)-(GA)CTCAG(CT)T(CT)C(TG)GT-3 9

Primers 1-3 contain a Hind III restriction site primers 5ndash7contain a Pst I restriction site primer 4 contains a Xba I re-striction site primer 8 contains an Eco RV restriction site

cDNA synthesis

Total cellular RNA was extracted by Trizol reagent (Gibco-BRL Paisley Scotland) First strand cDNA was synthesizedfrom 5 mg of total RNA using a first-strand cDNA synthesiskit (Gibco BRL) essentially as described by the manufacturerSpecific primers complementary to the coding strand of the 59

end of the constant regions (CH1 heavy and kappa light chains)were used Primers 1 and 2 were used in the VH cDNA syn-thesis of P3 and Ab2 MAbs respectively primer 3 was usedin the VL cDNA synthesis of P3 and 3B11 MAbs primer 4was used in the VL cDNA synthesis of 1E10 MAb Before poly-merase chain reaction (PCR) samples were heat-treated to in-activate reverse transcriptase

PCR amplification of V-genes

cDNAs were amplified by PCR as described by Orlandi etal(8) using one unit of Vent DNA polymerase 200 mM dNTPand 05 mM of each primer in a final volume of 50 mL of re-action buffer essentially as described by the manufacturer (NewEngland Biolabs Inc Beverly MA) Primers 1 and 5 wereused for the amplification of VH P3 Primers 3 and 6 were usedfor the amplification of VL P3 Primers 2 and 5 were used forthe amplification of VH of Ab2 Primers 3 and 7 were used forthe amplification of VL 3B11 Primers 3 and 8 were used forthe amplification of VL 1E10

Each reaction mixture was subjected to 30 cycles of ampli-fication using a thermal minicycler PTC-150 (MJ Research IncWatertown MA) Each cycle consisted of the following steps94degC for 1 min (except first cycle 4 min) 55degC for 1 min and72degC for 1 min with a final extension of 10 min at 72degC Am-plified variable regions were analyzed on a 18 low meltingpoint agaroseTris acetate-EDTA (TAE) gel and visualized withethidium bromide The band of the expected size was excisedand purified by phenol extraction and ethanol precipitation

Cloning and sequencing of PCR products

For cloning purified products were ligated into theM13mp19 vector (Pharmacia Biotech Uppsala Sweden) Theamplified cDNAs encoding the heavy chain variable regions ofP3 and Ab2 MAbs were digested with HindIII and PstI and lig-ated into the M13mp19 vector which was previously digestedwith Hind III and Pst I The amplified cDNA encoding the lightchain variable regions of P3 and 3B11 were digested with HindIII and Pvu II and ligated into M13mp19 vector previously di-gested with Hind III and Hinc II The amplified cDNA encod-ing the light chain variable region of 1E10 was digested withHind III and Eco RV and ligated into M13mp19 vector previ-ously digested with Hind III and Hinc II A standard ligationprotocol(9) was performed using T4 DNA Ligase (Gibco-BRLGaithersburg MD) XL1-Blue cells were transformed andplated for blue-white selection on X-Gal-IPTG plates Whitecolonies where picked and single strand M13mp19 DNA waspurified by phenol extraction after precipitation of phage par-ticles from cultures in liquid media

Sequencing of the variable regions was performed by thedideoxy method using the T7 sequencing kit (PharmaciaBiotech) The M13 universal primer was used for sequencingIn addition each variable region was separately amplified

PEREZ ET AL212

cloned and sequenced at least twice to exclude possible errorsintroduced by Vent DNA polymerase

Assignment of the origin of the variable gene segmentsutilized

Individual VH and VL sequences were searched against theEMBLGenbank database for sequence homology with knownmurine genes The germline gene with the closest homologywas analyzed as a possible donor sequence of the rearrangedgene However in some cases it was impossible to find agermline gene close enough to the analyzed sequence In thosecases we searched for the closest variable region sequence re-ported in EMBLGenbank The JK and JH regions sequencedwere compared with the five known murine JK genes and withthe four known murine JH genes(10) The analysis of the junc-tional sequences in VH sequences was performed as Bangs(11)

D region nucleotides were identified after sequence alignmentof heavy chain CDR3 (H-CDR3) nucleotide sequences withknown murine D minigenes(10) When junctional nucleotidescould have been contributed by either D or the adjacent VH orJH segment they were assigned to VH or JH In cases in which

no VH or JH nucleotides were deleted during recombination withthe D gene possible P additions(12) were identified

RESULTS AND DISCUSSION

Analysis of VH gene usage

The nucleotide sequences and the deduced amino acid se-quences of the heavy chain variable region of P3 MAb (Ab1)and Ab2 antibodies to P3 1E10 and 3B11 are shown in Figs1 and 2 respectively and their gene segment usage is summa-rized in Table 1

VH P3 (sequence shown in Fig 1) is a member of the Q52(VH II) gene family and is 96 identical in nucleotide sequenceto the related germline PJ14 gene from the same family(13) Themost related sequence to VH P3 is the cDNA sequence of asws1(IgG2a k) a mouse antibody generated in ASWSnJ (H-2S)strain specific to U3 and U8 ribonucleoproteic particles(14)

There are only two nucleotide differences between P3 andasws1 (VH from codons 2 to 94) leading to only one aminoacid replacement (56 Ser R Asn) This suggests that both

GENETIC BASIS OF g-TYPE Ab2 FROM GLYCOLYLATED GANGLIOSIDES 213

FIG 1 Nucleotide and deduced amino acid sequences of the mRNA encoding the heavy chain variable region of P3 MAb (VHP3) compared with the most closely related germline and cDNA sequences (see text) Amino acids are numbered according to Ka-bat et al(10) Spaces have been introduced to maximize alignment Dashed lines indicate nucleotide sequence identities N signifiesundetermined bases The amino acid residue encoded by each codon is given above the nucleotide sequence CDR amino acids arerepresented according to the frequency of appearance in each position in a compilation of 1474 murine VH sequences extracted fromKabat database Frequent amino acids (ie with frequencies 20) are underlined amino acids with frequencies between 10 and20 are represented in italics infrequent amino acid (ie frequencies 10) are represented in bold italics

PEREZ ET AL214

FIG 2 Nucleotide and deduced amino acid sequences of the mRNA encoding the heavy chain variable regions of anti-idio-types 1E10 (VH 1E10) and 3B11 (VH 3B11) compared with the most closely related germline and cDNA sequences (see text)(For details see legend to Fig 1 X means unknown amino acid)

VHP3 and VHasws1 may be encoded by a germline elementother than PJ14 The amino acid Ser56 encoded by VH P3 is aconserved amino acid that appears in all the germline VH se-quences from Q52 family reported until 1997(15) Furthermoredivergence nucleotides in codons 45 and 56 of VH P3 are alsopresent in a high percent (956 and 82 respectively) of theclosest 23 sequences (94 or higher identity) extracted fromGenbank database with the only exception of VH asws1 Thisresult strongly suggests that VH P3 could represent a new un-described gene from the Q52 family in BALBc mice

VHQ52 gene family has been involved in the antibody re-sponse against some gangliosides Zenita(16) in a Northern blotanalysis of the VH segments used by a large panel of anti-gan-glioside antibodies found that VH Q52 genes were preferablyused by antibodies directed to gangliosides with sialic acid ina(2 R 6) linkage However in addition to the present reportthe usage of genes from the VH Q52 family have been previ-ously observed in other autoantibodies directed to a(2 R 3)sialic acid-containing gangliosides like GM1(17) and GM2 (GMustelier manuscript in preparation) In BALBc mice the VH

Q52 genes are highly represented in the fetal B-cell repertoiredue to the preferential usage of these genes by CD51 B lym-phocytes (B-1 cells) the predominant B-cell subpopulation atthis period(1819) Furthermore it is known that this preferencefor D-proximal gene families like Q52 and 7183 is also ob-

served in adultrsquos B-1 cells(1920) The frequent appearance of VH

genes from this family in anti-ganglioside antibodies might bedue to the involvement of B-1 cells in their production

VH 1E10 (Fig 2) is a member of the J558 (VH I) familythe most JH-distal group of VH genes in BALBc mice Al-though it has 93 identity from codons 4 to 94 to V1021 agermline VH from the V NP group of genes in the same fam-ily(21) its VH segment is almost identical to the one present inH35-C7 a hybridoma generated in BALBc that produces anantibody to the influenza virus hemagglutinin(22) There is onlyone amino acid difference at the position 94 (Arg R Asp) thatpossibly was originated during the recombination process Thisstrongly suggests that 1E10 and H35-C7 use the same germlineVH gene from J558 gene family but different to the V1021gene defining most likely a new VH gene segment

When compared with germline VH gene segments the cDNAsequence of VH 3B11 showed 89 nucleotide identity fromcodons 1 to 94 with VH EA7 (Fig 2) a member of the 3660(VH III) gene family(23) There are 21 nucleotide positions thatdiffer between both sequences that appear all along the VH seg-ments The most homologous sequence to VH 3B11 is thecDNA sequence from MRB9 a polyrreactive murine MAb thatbinds to purified total histones(24) Both sequences share a 95homology having differences in 11 nucleotide positions(codons 1 to 94) According to this analysis VH 3B11 belongs

GENETIC BASIS OF g-TYPE Ab2 FROM GLYCOLYLATED GANGLIOSIDES 215

TABLE 1 GENE SEGMENT USAGE IN P3 MAb AND Ab2 ANTIBODIES

MAb Isotype VH D JH Vk Jk

P3 IgM k Q52 (II) DSP2 JH3 Vk19 Jk11E10 IgG1 k J558 (I) DSP221 JH2 Vk10 Jk1

DSP23B11 IgG1 k 36ndash60 (III) DSP22 JH3 Vk45 Jk5

FIG 3 Nucleotide and deduced amino acid sequences of the mRNA encoding the light chain variable region of P3 MAb (VkP3)compared with the most closely related cDNA sequence (VkH250-6) (For details see legend to Fig 1)

could be distinguished by their different abilities to induce Ab3sharing idiotopes with P3 MAb in syngeneic animals (Id1

Ab3)In this paper we try to relate those immunochemical prop-

erties with the immunogenetic analysis of the primary struc-tures of P3 MAb and two of the anti-P3 Ab2s previously re-ported 1E10 Ab2 which was able to induce Id1 Ab3 responsein syngeneic model and 3B11 Ab2 that showed the highestbinding affinity for P3 MAb but was unable to induce Id1

Ab3(7) We determined the nucleotide sequences of the variableregions of the heavy and light chains (VH VL) of these anti-bodies by cloning cDNAs derived from the RNA extracted fromhybridoma cells The frequency of complementary chargedresidues on both Ab1 and Ab2 suggests that electrostatic in-teractions contribute to the specificity and high affinity of theg-type idiotypic recognition in the V-region repertoire recog-nizing gangliosides

MATERIALS AND METHODS

Ab1 and Ab2 MAbs

P3 MAb (IgM k) (Ab1) was generated by immunizationof BALBc mice with lyposomes containing NeuGc-GM3 andtetanus toxoid as previously described(5) Syngeneic Ab2MAbs 1E10 (IgG1 k) and 3B11 (IgG1 k) were obtained from BALBc mice immunized with purified P3 MAb cou-pled to KLH (Keyhole Limpet Haemocyanin) in Freundrsquos ad-juvant(7)

Primers

The following primers were used

(1) CM1FOR (39 primer for the heavy chain IgM)59-GGAAGCTTAAGACATTTGGGAAGGACTGAC-TCTC-39

(2) CG1FOR (39 primer for the heavy chain IgG1)59-GGAAGCTTAGACAGATGGGG GTGTCGTTTTG-39

(3) CK2FOR (39 primer for the kappa light chain)59-GGAAGCTTGAAGATGGATACAGTTGGTG -CAGC-39

(4) McKAsXBA (39 primer for the kappa light chain)59-GCGTCTAGAACTGGATGGTGGGAAGATGG-3 9

(5) VH1BACK (59 primer for the heavy chain framework re-gion 1 (FR1))59-AGGT(GC)(AC)A(AG)CTGCAG(GC)AGTC(AT)-GG-39

(6) VK8BACK (59 primer for the VK8 light chain FR1)59-C(AT)GAGAAATTCAGCTGACCCAGTCTC-3 9

(7) VK1BACK (59 primer for the VK1 light chain FR1)59-GACATTCAGCTGACCCAGTCTCCA-3 9

(8) MLALT4RV (59 primer for the light chain signal se-quence)59-GGGGATATCCACCATGAGG(GT)CCCC(AT)-(GA)CTCAG(CT)T(CT)C(TG)GT-3 9

Primers 1-3 contain a Hind III restriction site primers 5ndash7contain a Pst I restriction site primer 4 contains a Xba I re-striction site primer 8 contains an Eco RV restriction site

cDNA synthesis

Total cellular RNA was extracted by Trizol reagent (Gibco-BRL Paisley Scotland) First strand cDNA was synthesizedfrom 5 mg of total RNA using a first-strand cDNA synthesiskit (Gibco BRL) essentially as described by the manufacturerSpecific primers complementary to the coding strand of the 59

end of the constant regions (CH1 heavy and kappa light chains)were used Primers 1 and 2 were used in the VH cDNA syn-thesis of P3 and Ab2 MAbs respectively primer 3 was usedin the VL cDNA synthesis of P3 and 3B11 MAbs primer 4was used in the VL cDNA synthesis of 1E10 MAb Before poly-merase chain reaction (PCR) samples were heat-treated to in-activate reverse transcriptase

PCR amplification of V-genes

cDNAs were amplified by PCR as described by Orlandi etal(8) using one unit of Vent DNA polymerase 200 mM dNTPand 05 mM of each primer in a final volume of 50 mL of re-action buffer essentially as described by the manufacturer (NewEngland Biolabs Inc Beverly MA) Primers 1 and 5 wereused for the amplification of VH P3 Primers 3 and 6 were usedfor the amplification of VL P3 Primers 2 and 5 were used forthe amplification of VH of Ab2 Primers 3 and 7 were used forthe amplification of VL 3B11 Primers 3 and 8 were used forthe amplification of VL 1E10

Each reaction mixture was subjected to 30 cycles of ampli-fication using a thermal minicycler PTC-150 (MJ Research IncWatertown MA) Each cycle consisted of the following steps94degC for 1 min (except first cycle 4 min) 55degC for 1 min and72degC for 1 min with a final extension of 10 min at 72degC Am-plified variable regions were analyzed on a 18 low meltingpoint agaroseTris acetate-EDTA (TAE) gel and visualized withethidium bromide The band of the expected size was excisedand purified by phenol extraction and ethanol precipitation

Cloning and sequencing of PCR products

For cloning purified products were ligated into theM13mp19 vector (Pharmacia Biotech Uppsala Sweden) Theamplified cDNAs encoding the heavy chain variable regions ofP3 and Ab2 MAbs were digested with HindIII and PstI and lig-ated into the M13mp19 vector which was previously digestedwith Hind III and Pst I The amplified cDNA encoding the lightchain variable regions of P3 and 3B11 were digested with HindIII and Pvu II and ligated into M13mp19 vector previously di-gested with Hind III and Hinc II The amplified cDNA encod-ing the light chain variable region of 1E10 was digested withHind III and Eco RV and ligated into M13mp19 vector previ-ously digested with Hind III and Hinc II A standard ligationprotocol(9) was performed using T4 DNA Ligase (Gibco-BRLGaithersburg MD) XL1-Blue cells were transformed andplated for blue-white selection on X-Gal-IPTG plates Whitecolonies where picked and single strand M13mp19 DNA waspurified by phenol extraction after precipitation of phage par-ticles from cultures in liquid media

Sequencing of the variable regions was performed by thedideoxy method using the T7 sequencing kit (PharmaciaBiotech) The M13 universal primer was used for sequencingIn addition each variable region was separately amplified

PEREZ ET AL212

cloned and sequenced at least twice to exclude possible errorsintroduced by Vent DNA polymerase

Assignment of the origin of the variable gene segmentsutilized

Individual VH and VL sequences were searched against theEMBLGenbank database for sequence homology with knownmurine genes The germline gene with the closest homologywas analyzed as a possible donor sequence of the rearrangedgene However in some cases it was impossible to find agermline gene close enough to the analyzed sequence In thosecases we searched for the closest variable region sequence re-ported in EMBLGenbank The JK and JH regions sequencedwere compared with the five known murine JK genes and withthe four known murine JH genes(10) The analysis of the junc-tional sequences in VH sequences was performed as Bangs(11)

D region nucleotides were identified after sequence alignmentof heavy chain CDR3 (H-CDR3) nucleotide sequences withknown murine D minigenes(10) When junctional nucleotidescould have been contributed by either D or the adjacent VH orJH segment they were assigned to VH or JH In cases in which

no VH or JH nucleotides were deleted during recombination withthe D gene possible P additions(12) were identified

RESULTS AND DISCUSSION

Analysis of VH gene usage

The nucleotide sequences and the deduced amino acid se-quences of the heavy chain variable region of P3 MAb (Ab1)and Ab2 antibodies to P3 1E10 and 3B11 are shown in Figs1 and 2 respectively and their gene segment usage is summa-rized in Table 1

VH P3 (sequence shown in Fig 1) is a member of the Q52(VH II) gene family and is 96 identical in nucleotide sequenceto the related germline PJ14 gene from the same family(13) Themost related sequence to VH P3 is the cDNA sequence of asws1(IgG2a k) a mouse antibody generated in ASWSnJ (H-2S)strain specific to U3 and U8 ribonucleoproteic particles(14)

There are only two nucleotide differences between P3 andasws1 (VH from codons 2 to 94) leading to only one aminoacid replacement (56 Ser R Asn) This suggests that both

GENETIC BASIS OF g-TYPE Ab2 FROM GLYCOLYLATED GANGLIOSIDES 213

FIG 1 Nucleotide and deduced amino acid sequences of the mRNA encoding the heavy chain variable region of P3 MAb (VHP3) compared with the most closely related germline and cDNA sequences (see text) Amino acids are numbered according to Ka-bat et al(10) Spaces have been introduced to maximize alignment Dashed lines indicate nucleotide sequence identities N signifiesundetermined bases The amino acid residue encoded by each codon is given above the nucleotide sequence CDR amino acids arerepresented according to the frequency of appearance in each position in a compilation of 1474 murine VH sequences extracted fromKabat database Frequent amino acids (ie with frequencies 20) are underlined amino acids with frequencies between 10 and20 are represented in italics infrequent amino acid (ie frequencies 10) are represented in bold italics

PEREZ ET AL214

FIG 2 Nucleotide and deduced amino acid sequences of the mRNA encoding the heavy chain variable regions of anti-idio-types 1E10 (VH 1E10) and 3B11 (VH 3B11) compared with the most closely related germline and cDNA sequences (see text)(For details see legend to Fig 1 X means unknown amino acid)

VHP3 and VHasws1 may be encoded by a germline elementother than PJ14 The amino acid Ser56 encoded by VH P3 is aconserved amino acid that appears in all the germline VH se-quences from Q52 family reported until 1997(15) Furthermoredivergence nucleotides in codons 45 and 56 of VH P3 are alsopresent in a high percent (956 and 82 respectively) of theclosest 23 sequences (94 or higher identity) extracted fromGenbank database with the only exception of VH asws1 Thisresult strongly suggests that VH P3 could represent a new un-described gene from the Q52 family in BALBc mice

VHQ52 gene family has been involved in the antibody re-sponse against some gangliosides Zenita(16) in a Northern blotanalysis of the VH segments used by a large panel of anti-gan-glioside antibodies found that VH Q52 genes were preferablyused by antibodies directed to gangliosides with sialic acid ina(2 R 6) linkage However in addition to the present reportthe usage of genes from the VH Q52 family have been previ-ously observed in other autoantibodies directed to a(2 R 3)sialic acid-containing gangliosides like GM1(17) and GM2 (GMustelier manuscript in preparation) In BALBc mice the VH

Q52 genes are highly represented in the fetal B-cell repertoiredue to the preferential usage of these genes by CD51 B lym-phocytes (B-1 cells) the predominant B-cell subpopulation atthis period(1819) Furthermore it is known that this preferencefor D-proximal gene families like Q52 and 7183 is also ob-

served in adultrsquos B-1 cells(1920) The frequent appearance of VH

genes from this family in anti-ganglioside antibodies might bedue to the involvement of B-1 cells in their production

VH 1E10 (Fig 2) is a member of the J558 (VH I) familythe most JH-distal group of VH genes in BALBc mice Al-though it has 93 identity from codons 4 to 94 to V1021 agermline VH from the V NP group of genes in the same fam-ily(21) its VH segment is almost identical to the one present inH35-C7 a hybridoma generated in BALBc that produces anantibody to the influenza virus hemagglutinin(22) There is onlyone amino acid difference at the position 94 (Arg R Asp) thatpossibly was originated during the recombination process Thisstrongly suggests that 1E10 and H35-C7 use the same germlineVH gene from J558 gene family but different to the V1021gene defining most likely a new VH gene segment

When compared with germline VH gene segments the cDNAsequence of VH 3B11 showed 89 nucleotide identity fromcodons 1 to 94 with VH EA7 (Fig 2) a member of the 3660(VH III) gene family(23) There are 21 nucleotide positions thatdiffer between both sequences that appear all along the VH seg-ments The most homologous sequence to VH 3B11 is thecDNA sequence from MRB9 a polyrreactive murine MAb thatbinds to purified total histones(24) Both sequences share a 95homology having differences in 11 nucleotide positions(codons 1 to 94) According to this analysis VH 3B11 belongs

GENETIC BASIS OF g-TYPE Ab2 FROM GLYCOLYLATED GANGLIOSIDES 215

TABLE 1 GENE SEGMENT USAGE IN P3 MAb AND Ab2 ANTIBODIES

MAb Isotype VH D JH Vk Jk

P3 IgM k Q52 (II) DSP2 JH3 Vk19 Jk11E10 IgG1 k J558 (I) DSP221 JH2 Vk10 Jk1

DSP23B11 IgG1 k 36ndash60 (III) DSP22 JH3 Vk45 Jk5

FIG 3 Nucleotide and deduced amino acid sequences of the mRNA encoding the light chain variable region of P3 MAb (VkP3)compared with the most closely related cDNA sequence (VkH250-6) (For details see legend to Fig 1)

cloned and sequenced at least twice to exclude possible errorsintroduced by Vent DNA polymerase

Assignment of the origin of the variable gene segmentsutilized

Individual VH and VL sequences were searched against theEMBLGenbank database for sequence homology with knownmurine genes The germline gene with the closest homologywas analyzed as a possible donor sequence of the rearrangedgene However in some cases it was impossible to find agermline gene close enough to the analyzed sequence In thosecases we searched for the closest variable region sequence re-ported in EMBLGenbank The JK and JH regions sequencedwere compared with the five known murine JK genes and withthe four known murine JH genes(10) The analysis of the junc-tional sequences in VH sequences was performed as Bangs(11)

D region nucleotides were identified after sequence alignmentof heavy chain CDR3 (H-CDR3) nucleotide sequences withknown murine D minigenes(10) When junctional nucleotidescould have been contributed by either D or the adjacent VH orJH segment they were assigned to VH or JH In cases in which

no VH or JH nucleotides were deleted during recombination withthe D gene possible P additions(12) were identified

RESULTS AND DISCUSSION

Analysis of VH gene usage

The nucleotide sequences and the deduced amino acid se-quences of the heavy chain variable region of P3 MAb (Ab1)and Ab2 antibodies to P3 1E10 and 3B11 are shown in Figs1 and 2 respectively and their gene segment usage is summa-rized in Table 1

VH P3 (sequence shown in Fig 1) is a member of the Q52(VH II) gene family and is 96 identical in nucleotide sequenceto the related germline PJ14 gene from the same family(13) Themost related sequence to VH P3 is the cDNA sequence of asws1(IgG2a k) a mouse antibody generated in ASWSnJ (H-2S)strain specific to U3 and U8 ribonucleoproteic particles(14)

There are only two nucleotide differences between P3 andasws1 (VH from codons 2 to 94) leading to only one aminoacid replacement (56 Ser R Asn) This suggests that both

GENETIC BASIS OF g-TYPE Ab2 FROM GLYCOLYLATED GANGLIOSIDES 213

FIG 1 Nucleotide and deduced amino acid sequences of the mRNA encoding the heavy chain variable region of P3 MAb (VHP3) compared with the most closely related germline and cDNA sequences (see text) Amino acids are numbered according to Ka-bat et al(10) Spaces have been introduced to maximize alignment Dashed lines indicate nucleotide sequence identities N signifiesundetermined bases The amino acid residue encoded by each codon is given above the nucleotide sequence CDR amino acids arerepresented according to the frequency of appearance in each position in a compilation of 1474 murine VH sequences extracted fromKabat database Frequent amino acids (ie with frequencies 20) are underlined amino acids with frequencies between 10 and20 are represented in italics infrequent amino acid (ie frequencies 10) are represented in bold italics

PEREZ ET AL214

FIG 2 Nucleotide and deduced amino acid sequences of the mRNA encoding the heavy chain variable regions of anti-idio-types 1E10 (VH 1E10) and 3B11 (VH 3B11) compared with the most closely related germline and cDNA sequences (see text)(For details see legend to Fig 1 X means unknown amino acid)

VHP3 and VHasws1 may be encoded by a germline elementother than PJ14 The amino acid Ser56 encoded by VH P3 is aconserved amino acid that appears in all the germline VH se-quences from Q52 family reported until 1997(15) Furthermoredivergence nucleotides in codons 45 and 56 of VH P3 are alsopresent in a high percent (956 and 82 respectively) of theclosest 23 sequences (94 or higher identity) extracted fromGenbank database with the only exception of VH asws1 Thisresult strongly suggests that VH P3 could represent a new un-described gene from the Q52 family in BALBc mice

VHQ52 gene family has been involved in the antibody re-sponse against some gangliosides Zenita(16) in a Northern blotanalysis of the VH segments used by a large panel of anti-gan-glioside antibodies found that VH Q52 genes were preferablyused by antibodies directed to gangliosides with sialic acid ina(2 R 6) linkage However in addition to the present reportthe usage of genes from the VH Q52 family have been previ-ously observed in other autoantibodies directed to a(2 R 3)sialic acid-containing gangliosides like GM1(17) and GM2 (GMustelier manuscript in preparation) In BALBc mice the VH

Q52 genes are highly represented in the fetal B-cell repertoiredue to the preferential usage of these genes by CD51 B lym-phocytes (B-1 cells) the predominant B-cell subpopulation atthis period(1819) Furthermore it is known that this preferencefor D-proximal gene families like Q52 and 7183 is also ob-

served in adultrsquos B-1 cells(1920) The frequent appearance of VH

genes from this family in anti-ganglioside antibodies might bedue to the involvement of B-1 cells in their production

VH 1E10 (Fig 2) is a member of the J558 (VH I) familythe most JH-distal group of VH genes in BALBc mice Al-though it has 93 identity from codons 4 to 94 to V1021 agermline VH from the V NP group of genes in the same fam-ily(21) its VH segment is almost identical to the one present inH35-C7 a hybridoma generated in BALBc that produces anantibody to the influenza virus hemagglutinin(22) There is onlyone amino acid difference at the position 94 (Arg R Asp) thatpossibly was originated during the recombination process Thisstrongly suggests that 1E10 and H35-C7 use the same germlineVH gene from J558 gene family but different to the V1021gene defining most likely a new VH gene segment

When compared with germline VH gene segments the cDNAsequence of VH 3B11 showed 89 nucleotide identity fromcodons 1 to 94 with VH EA7 (Fig 2) a member of the 3660(VH III) gene family(23) There are 21 nucleotide positions thatdiffer between both sequences that appear all along the VH seg-ments The most homologous sequence to VH 3B11 is thecDNA sequence from MRB9 a polyrreactive murine MAb thatbinds to purified total histones(24) Both sequences share a 95homology having differences in 11 nucleotide positions(codons 1 to 94) According to this analysis VH 3B11 belongs

GENETIC BASIS OF g-TYPE Ab2 FROM GLYCOLYLATED GANGLIOSIDES 215

TABLE 1 GENE SEGMENT USAGE IN P3 MAb AND Ab2 ANTIBODIES

MAb Isotype VH D JH Vk Jk

P3 IgM k Q52 (II) DSP2 JH3 Vk19 Jk11E10 IgG1 k J558 (I) DSP221 JH2 Vk10 Jk1

DSP23B11 IgG1 k 36ndash60 (III) DSP22 JH3 Vk45 Jk5

FIG 3 Nucleotide and deduced amino acid sequences of the mRNA encoding the light chain variable region of P3 MAb (VkP3)compared with the most closely related cDNA sequence (VkH250-6) (For details see legend to Fig 1)

PEREZ ET AL214

FIG 2 Nucleotide and deduced amino acid sequences of the mRNA encoding the heavy chain variable regions of anti-idio-types 1E10 (VH 1E10) and 3B11 (VH 3B11) compared with the most closely related germline and cDNA sequences (see text)(For details see legend to Fig 1 X means unknown amino acid)

VHP3 and VHasws1 may be encoded by a germline elementother than PJ14 The amino acid Ser56 encoded by VH P3 is aconserved amino acid that appears in all the germline VH se-quences from Q52 family reported until 1997(15) Furthermoredivergence nucleotides in codons 45 and 56 of VH P3 are alsopresent in a high percent (956 and 82 respectively) of theclosest 23 sequences (94 or higher identity) extracted fromGenbank database with the only exception of VH asws1 Thisresult strongly suggests that VH P3 could represent a new un-described gene from the Q52 family in BALBc mice

VHQ52 gene family has been involved in the antibody re-sponse against some gangliosides Zenita(16) in a Northern blotanalysis of the VH segments used by a large panel of anti-gan-glioside antibodies found that VH Q52 genes were preferablyused by antibodies directed to gangliosides with sialic acid ina(2 R 6) linkage However in addition to the present reportthe usage of genes from the VH Q52 family have been previ-ously observed in other autoantibodies directed to a(2 R 3)sialic acid-containing gangliosides like GM1(17) and GM2 (GMustelier manuscript in preparation) In BALBc mice the VH

Q52 genes are highly represented in the fetal B-cell repertoiredue to the preferential usage of these genes by CD51 B lym-phocytes (B-1 cells) the predominant B-cell subpopulation atthis period(1819) Furthermore it is known that this preferencefor D-proximal gene families like Q52 and 7183 is also ob-

served in adultrsquos B-1 cells(1920) The frequent appearance of VH

genes from this family in anti-ganglioside antibodies might bedue to the involvement of B-1 cells in their production

VH 1E10 (Fig 2) is a member of the J558 (VH I) familythe most JH-distal group of VH genes in BALBc mice Al-though it has 93 identity from codons 4 to 94 to V1021 agermline VH from the V NP group of genes in the same fam-ily(21) its VH segment is almost identical to the one present inH35-C7 a hybridoma generated in BALBc that produces anantibody to the influenza virus hemagglutinin(22) There is onlyone amino acid difference at the position 94 (Arg R Asp) thatpossibly was originated during the recombination process Thisstrongly suggests that 1E10 and H35-C7 use the same germlineVH gene from J558 gene family but different to the V1021gene defining most likely a new VH gene segment

When compared with germline VH gene segments the cDNAsequence of VH 3B11 showed 89 nucleotide identity fromcodons 1 to 94 with VH EA7 (Fig 2) a member of the 3660(VH III) gene family(23) There are 21 nucleotide positions thatdiffer between both sequences that appear all along the VH seg-ments The most homologous sequence to VH 3B11 is thecDNA sequence from MRB9 a polyrreactive murine MAb thatbinds to purified total histones(24) Both sequences share a 95homology having differences in 11 nucleotide positions(codons 1 to 94) According to this analysis VH 3B11 belongs

GENETIC BASIS OF g-TYPE Ab2 FROM GLYCOLYLATED GANGLIOSIDES 215

TABLE 1 GENE SEGMENT USAGE IN P3 MAb AND Ab2 ANTIBODIES

MAb Isotype VH D JH Vk Jk

P3 IgM k Q52 (II) DSP2 JH3 Vk19 Jk11E10 IgG1 k J558 (I) DSP221 JH2 Vk10 Jk1

DSP23B11 IgG1 k 36ndash60 (III) DSP22 JH3 Vk45 Jk5

FIG 3 Nucleotide and deduced amino acid sequences of the mRNA encoding the light chain variable region of P3 MAb (VkP3)compared with the most closely related cDNA sequence (VkH250-6) (For details see legend to Fig 1)

VHP3 and VHasws1 may be encoded by a germline elementother than PJ14 The amino acid Ser56 encoded by VH P3 is aconserved amino acid that appears in all the germline VH se-quences from Q52 family reported until 1997(15) Furthermoredivergence nucleotides in codons 45 and 56 of VH P3 are alsopresent in a high percent (956 and 82 respectively) of theclosest 23 sequences (94 or higher identity) extracted fromGenbank database with the only exception of VH asws1 Thisresult strongly suggests that VH P3 could represent a new un-described gene from the Q52 family in BALBc mice

VHQ52 gene family has been involved in the antibody re-sponse against some gangliosides Zenita(16) in a Northern blotanalysis of the VH segments used by a large panel of anti-gan-glioside antibodies found that VH Q52 genes were preferablyused by antibodies directed to gangliosides with sialic acid ina(2 R 6) linkage However in addition to the present reportthe usage of genes from the VH Q52 family have been previ-ously observed in other autoantibodies directed to a(2 R 3)sialic acid-containing gangliosides like GM1(17) and GM2 (GMustelier manuscript in preparation) In BALBc mice the VH

Q52 genes are highly represented in the fetal B-cell repertoiredue to the preferential usage of these genes by CD51 B lym-phocytes (B-1 cells) the predominant B-cell subpopulation atthis period(1819) Furthermore it is known that this preferencefor D-proximal gene families like Q52 and 7183 is also ob-

served in adultrsquos B-1 cells(1920) The frequent appearance of VH

genes from this family in anti-ganglioside antibodies might bedue to the involvement of B-1 cells in their production

VH 1E10 (Fig 2) is a member of the J558 (VH I) familythe most JH-distal group of VH genes in BALBc mice Al-though it has 93 identity from codons 4 to 94 to V1021 agermline VH from the V NP group of genes in the same fam-ily(21) its VH segment is almost identical to the one present inH35-C7 a hybridoma generated in BALBc that produces anantibody to the influenza virus hemagglutinin(22) There is onlyone amino acid difference at the position 94 (Arg R Asp) thatpossibly was originated during the recombination process Thisstrongly suggests that 1E10 and H35-C7 use the same germlineVH gene from J558 gene family but different to the V1021gene defining most likely a new VH gene segment

When compared with germline VH gene segments the cDNAsequence of VH 3B11 showed 89 nucleotide identity fromcodons 1 to 94 with VH EA7 (Fig 2) a member of the 3660(VH III) gene family(23) There are 21 nucleotide positions thatdiffer between both sequences that appear all along the VH seg-ments The most homologous sequence to VH 3B11 is thecDNA sequence from MRB9 a polyrreactive murine MAb thatbinds to purified total histones(24) Both sequences share a 95homology having differences in 11 nucleotide positions(codons 1 to 94) According to this analysis VH 3B11 belongs

GENETIC BASIS OF g-TYPE Ab2 FROM GLYCOLYLATED GANGLIOSIDES 215

TABLE 1 GENE SEGMENT USAGE IN P3 MAb AND Ab2 ANTIBODIES

MAb Isotype VH D JH Vk Jk

P3 IgM k Q52 (II) DSP2 JH3 Vk19 Jk11E10 IgG1 k J558 (I) DSP221 JH2 Vk10 Jk1

DSP23B11 IgG1 k 36ndash60 (III) DSP22 JH3 Vk45 Jk5

FIG 3 Nucleotide and deduced amino acid sequences of the mRNA encoding the light chain variable region of P3 MAb (VkP3)compared with the most closely related cDNA sequence (VkH250-6) (For details see legend to Fig 1)

to the 3660 (VHIII) family Therefore g-type anti-idiotypic an-tibodies to the same Ab1 (P3 MAb) can be obtained from dif-ferent VH gene families

Analysis of VL gene usage

The light chain variable region cDNA sequences of the threeMAbs presented here are derived from different Vk gene fam-ilies (Table 1) The VL sequence of P3 belongs to the Vk19family wand is 100 identical to one of its members H250-6an anti-influenza virus hemagglutinin MAb originated from aBALBc mouse(22) (Fig 3) This identity involves both the useof the same Vk gene segment (codons 1ndash95) and of the sameJk gene segment (Jk1 codons 96ndash107 Fig 5A) The fact thatthe light chain variable region of P3 MAb is fully identical tothe one corresponding to H250-6 MAb suggests that P3 H chainprobably contributes more than the L chain to the combiningsite to gangliosides

The VL 1E10 sequence is a member of the Vk10 family anduses the V-IdCR gene segment VL 1E10 sequence is identicalto that present in KL218 KL4C11 and KL4B10 MAbs(M63608 M63617 and M63615 respectively in Genbankdatabase) This VL sequence is associated with the antibody re-

sponse against p-azophenylarsonate in A mice (CAL-20xAJ)F1(25) VL 1E10 is 100 identical to VL KL218 incodons 1 to 107 implying the use of the same Vk and Jk (Jk1)gene segments

The VL sequence of 3B11 belongs to Vk-Ox1 superfam-ily(26) It is 993 identical in nucleotides to the VL of 38C13an antibody with unknown specificity produced by a murine B-cell lymphoma of the same name(27) The Vk segment sequences(codons 1ndash95) of both antibodies differ only in two nucleotidesthat do not lead to amino acid replacements In VL 3B11 Jk5is used (Fig 5) and FR4 sequence differs from the sequenceof this gene by one nucleotide leading to one amino acid re-placement (Glu105 R Gln105) In 38C13 Jk4 is used

In conclusion based on the identity of the light chain vari-able regions either from 1E10 or 3B11 MAbs to previously re-ported sequences one might expect a greater contribution ofVHs than VLs to the Ab2 binding sites to P3 MAb

Quantitative and qualitative analysis of somatichypermutations

Sequence analysis presented above of the heavy and lightchain variable regions demonstrated the germline nature of P3

PEREZ ET AL216

FIG 4 Nucleotide and deduced amino acid sequences of the mRNA encoding the light chain variable regions of anti-idiotypes1E10 (Vk1E10) and 3B11 (Vk3B11) compared with the most closely related germline and cDNA sequences (see text) [For de-tails see legend to Fig 1)

and 1E10 MAbs On the other hand 3B11 MAb showed fea-tures of extensive somatic mutations As a replacement to silentmutation ratio (RS) greater than 29 has been proposed to in-dicate a positive selection of nucleotide changes that affectamino acid sequence(28) this ratio was calculated for the CDRand framework regions of VH 3B11 A quantitative and quali-tative analysis of R and S mutations in VH 3B11 showed a pat-tern which is characteristic for hypermutated antibodies Thereis a clear tendency to cluster replacement (and nonconserva-tive) mutations in the H-CDRs whereas all mutations in theframeworks tend to be silent or conservative Whereas FR re-gions of VH 3B11 showed an RS ratio of 2 (42 29) thisratio was 4 (41 29) for the CDRs Furthermore replacementmutations in FRs involved conservative amino acid changeswith the only exception of Ser94 R Arg94 most likely origi-nated during VHndashDndashJH recombination process By contrary all

the replacement mutations that appear in the H-CDRs 1 and 2involved nonconservative amino acid changes

Other data supporting the evidence of somatic mutations inVH 3B11 came from comparisons with the closest homologoussequences reported in Genbank Nucleotides in eight out of tenputative mutated positions of VH 3B11 were found exclusivelyin this sequence and only two nucleotides those in codons 54and 88 were found to be shared by most VH sequences (90and 80 of the sequences with 94 or higher identity re-spectively) suggesting their germline origin

In addition a replacement mutation in FR4 of VL 3B11 con-firms the evidence of somatic mutations given above Howeverit is striking to find such a great difference between the levelsof mutations in both chains Previous studies have shown thatin general somatic mutations in H- and L-chain variable re-gions tend to occur concurrently(29) Nevertheless in some re-

GENETIC BASIS OF g-TYPE Ab2 FROM GLYCOLYLATED GANGLIOSIDES 217

FIG 5 Possible origins of heavy and light chain CDR3 of P3 (A) and anti-idiotypic MAbs 1E10 and 3B11 (B) P and N nu-cleotides are represented with asterisks Dashed lines indicate homology with known D and J genes The germline sequences ofDSP22 DSP26 and DSP27 together with all the germline J sequences are from Kabat et al(10) The germline sequences DSP210and DSP211 are from the International Immunogenetics Database (IMGT httpwwwebiacukimgt ) H-CDR amino acids arerepresented according to their frequencies in each position in a compilation of murine VH from Kabat database (httpim-munobmenwuedu) (For details see legend to Fig 1)

and 1E10 MAbs On the other hand 3B11 MAb showed fea-tures of extensive somatic mutations As a replacement to silentmutation ratio (RS) greater than 29 has been proposed to in-dicate a positive selection of nucleotide changes that affectamino acid sequence(28) this ratio was calculated for the CDRand framework regions of VH 3B11 A quantitative and quali-tative analysis of R and S mutations in VH 3B11 showed a pat-tern which is characteristic for hypermutated antibodies Thereis a clear tendency to cluster replacement (and nonconserva-tive) mutations in the H-CDRs whereas all mutations in theframeworks tend to be silent or conservative Whereas FR re-gions of VH 3B11 showed an RS ratio of 2 (42 29) thisratio was 4 (41 29) for the CDRs Furthermore replacementmutations in FRs involved conservative amino acid changeswith the only exception of Ser94 R Arg94 most likely origi-nated during VHndashDndashJH recombination process By contrary all

the replacement mutations that appear in the H-CDRs 1 and 2involved nonconservative amino acid changes

Other data supporting the evidence of somatic mutations inVH 3B11 came from comparisons with the closest homologoussequences reported in Genbank Nucleotides in eight out of tenputative mutated positions of VH 3B11 were found exclusivelyin this sequence and only two nucleotides those in codons 54and 88 were found to be shared by most VH sequences (90and 80 of the sequences with 94 or higher identity re-spectively) suggesting their germline origin

In addition a replacement mutation in FR4 of VL 3B11 con-firms the evidence of somatic mutations given above Howeverit is striking to find such a great difference between the levelsof mutations in both chains Previous studies have shown thatin general somatic mutations in H- and L-chain variable re-gions tend to occur concurrently(29) Nevertheless in some re-

GENETIC BASIS OF g-TYPE Ab2 FROM GLYCOLYLATED GANGLIOSIDES 217

FIG 5 Possible origins of heavy and light chain CDR3 of P3 (A) and anti-idiotypic MAbs 1E10 and 3B11 (B) P and N nu-cleotides are represented with asterisks Dashed lines indicate homology with known D and J genes The germline sequences ofDSP22 DSP26 and DSP27 together with all the germline J sequences are from Kabat et al(10) The germline sequences DSP210and DSP211 are from the International Immunogenetics Database (IMGT httpwwwebiacukimgt ) H-CDR amino acids arerepresented according to their frequencies in each position in a compilation of murine VH from Kabat database (httpim-munobmenwuedu) (For details see legend to Fig 1)

ports it is possible to find differences in the levels of mutationsbetween the heavy and light chains(3031) This has been inter-preted as an evidence of independent activity of the hypermu-tation machinery in each chain(31)

Although 3B11 and 1E10 MAbs were generated in the samefusion experiment and both are able to inhibit P3 MAb bind-ing to gangliosides immunochemical studies have shown thatboth Ab2s can be differentiated by their binding affinities toAb1 3B11 MAb the one that showed the higher affinity prob-ably as a consequence of an affinity muturation process seemsto define a ldquoprivaterdquo idiotope In fact 3B11 was unable to in-duce in syngeneic model anti-anti-idiotype antibodies thatshared idiotopes with P3 MAb in contrast to the germline-en-coded 1E10 MAb that seems to define a ldquoregulatoryrdquo id-iotope(7)

Analysis of junctional regions

Only 7 nucleotides out of 42 that codify the H-CDR3 of P3MAb belong to a D region (Fig 5A) Either DSP210 orDSP211 gene segments used in reading frame 3 could havecontributed to this region N nucleotides constitute almost 50of the sequence of this H-CDR3 and they appear at both sidesof the D segment In particular the 39N region which is morefrequently absent or shorter than 59N region(1136) is unusuallyextensive in this H-CDR3 with 15 nucleotides Even when thisnumber is higher than average additions catalyzed by TdT en-zyme (4ndash5)(32) the observed (G1C)(A1T) ratio (3) is char-acteristic of the activity of this enzyme(33) The last five aminoacids in H-CDR3 of P3 MAb are codified by P nucleotides(12)

and JH3The D segment of 1E10 has 21 nucleotides and is more likely

to be derived from fusion of thirteen nucleotides of DSP22 andseven nucleotides from another member of DSP2 family (eitherDSP26 DSP27 or DSP211 could have contributed) Bothminigenes are used in reading frame 1 The H-CDR3 of 1E10has a five nucleotides long 59N region whereas 39N region isabsent

The D segment of 3B11 can be readily assigned to the DSP22germline gene (Fig 5B) which also contributed to the D regionin 1E10 Nine nucleotides in H-CDR3 of 3B11 belong to this Dregion for which reading frame 1 is used As it happens in H-CDR3 of 1E10 the H-CDR3 of 3B11 lacks 39N nucleotidesThere are 4 nucleotide additions at 59 of the D segment and Dand JH segments overlap two nucleotides at junction

Analysis of CDR amino acid sequences of P3 MAb

To characterize the residues found in the CDRs of our anti-bodies two approaches were used First we classified residuesin each position in the H-CDRs according to their frequenciesof appearance in a compilation of murine VH sequences extracted from Kabat database (httpimmunobmenwuedu)These frequencies can be retrieved from web site httpbioinfocimsldcukabathtml In the second approach we cal-culated and compared the frequency of certain residues in H-and L-CDRs between our antibodies and a compilation of se-quences obtained and processed by Brard(34) from Kabat data-base(10) The major observation is that several key positionsmainly distributed among the H-chain CDRs present unusualcharged residues

Previously published studies suggested the leading role ofelectronegative groups from gangliosides (carboxyl and hy-droxyl groups in N-glycolyl neuraminic acid) and sulfated gly-colipids (SO3

2 group) in the binding of P3 MAb to these anti-gens(6) These findings first suggested the existence ofcomplementary positive-charged residues in the CDRs of P3MAb able to interact with those electronegative determinantsIn fact VH P3 has a slightly higher content of basic (R3 K1)than acidic (D1 E1) amino acid residues (Fig 1) Analysis ofpositional frequencies of residues in H-CDRs reveals that argi-nine is infrequently found in positions 31(48) 98(42) and100a(28) The presence of lysine at position 64 (CDR2) doesnot constitute an unusual feature of P3 MAb because it is ob-served in 81 of murine VH sequences in Kabatrsquos compila-tion and this amino acid could possibly play a structural role(35)

The presence of arginines at unusual positions and with higherfrequencies than those observed in Kabatrsquos compilation (Table3) has a great relevance in the comprehension of the interac-tion of P3 MAb and its ligands Arginines in H-CDR1 and 3and L-CDR2 of P3 MAb able to establish ionic interactions aswell as hydrogen bonds could be possibly involved in the recog-nition of these structures

Another interesting feature of P3 MAb amino acid sequenceis its H-CDR3 This region contains a high number of positionswith infrequent amino acids (57 of the sequence) given inpart by random N nucleotides In addition whereas several stud-ies have shown the preferred use of D segments in the readingframe 1 which generates tyrosine-rich sequences(3637) theshort D segment used in H-CDR3 P3 is translated in readingframe 3 and codifies two of the infrequent amino acids foundat this region The infrequent amino acid-rich sequence of theH-CDR3 in a germline antibody like P3 MAb could be possi-bly related to the ability of this MAb to induce in syngeneicmodel an IgG anti-idiotypic response in the absence of carrierprotein or adjuvant(7)

Analysis of CDR amino acid sequences of Ab2 MAbs

The major finding obtained from 1E10 and 3B11 sequenceanalysis was the presence of homologous acidic amino acid-rich sequences in their H-CDR3 regions (Fig 2) in contrastwith the low homology observed in their variable region genesegments The acidic motif EDndashDndashYDndashYndashD is codified byN nucleotides between VH and D and also by the first D re-gion nucleotides Both Ab2 share the DSP22 minigene whichcodifies the last residues in the motif This is the only murineD minigene that could contribute with two aspartic residues tothe CDR3 sequence During assembly of the heavy chain vari-able region a selected D gene element is usually trimmed byan exonuclease to give rise to a core sequence that representsonly part of the initial minigene and random nucleotides areadded to one or both ends of this core sequence The presenceof two codons for aspartic acid in DSP22 could therefore in-crease the probability of at least one codon remaining after mu-tilation This could explain the appearance of this minigene inboth H-CDR3 regions in view of the importance that acidicresidues seem to have in the binding sites of 3B11 and 1E10MAbs

As it is shown in Table 2 acidic residues in H-CDR3 of Ab2antibodies constitute the 227 percentage higher than the av-

PEREZ ET AL218

erage representation in Kabatrsquos H-CDR3s whereas basicresidues are slightly under-represented In addition acidicresidues in this CDR are not usually found in their positions inKabatrsquos compilation (with the only exception of Asp101 in VH1E10 which is codified by the JH gene segment) Acidic andbasic amino acid residues are found in CDRs 1 and 2 of Ab2antibodies with levels close to those in Kabatrsquos compilationHowever the presence of unusual acidic residues like Asp31

(CDR1) and Asp58 (CDR2) in VH 1E10 and VH 3B11 re-spectively and Asp54 which appears in a shared motif DGS(Fig 3) is noticeable

In contrast the light chain CDRs (L-CDRs) of Ab2 presentfew unusual features There are no differences in chargedresidue levels between these L-CDRs and Kabatrsquos compilation(Table 2) supporting the previous suggestion of the major con-tribution of the Ab2 heavy chains to the binding site to P3 MAbThe only striking deviation from Kabatrsquos compilation appearsin the levels of asparagine and glutamine residues in L-chainCDR3s (24 higher than L-chain CDR3 frequency in Kabatrsquoscompilation)

Early studies in the field of antigenndashantibody interactionshave shown that cationic antigens develop anionic antibody re-

GENETIC BASIS OF g-TYPE Ab2 FROM GLYCOLYLATED GANGLIOSIDES 219

TABLE 2 FREQUENCIES OF SELECTED AMINO ACIDS IN KABAT CDRS Ab2a

CDR1H CDR2H CDR3H

Ab2 Kabat D Ab2 Kabat D Ab2 Kabat D

RK 00 17 217 151 106 45 00 50 250DE 91 81 10 121 83 38 381 154 227Y 273 209 64 61 116 255 333 237 96FW 91 84 13 91 46 45 143 45 98QN 182 78 104 121 123 202 48 36 12ST 182 178 04 151 189 238 48 130 282

CDR1L CDR2L CDR3L

Ab2 Kabat D Ab2 Kabat D Ab2 Kabat D

RK 43 69 226 71 140 269 00 27 227DE 43 41 02 00 70 270 00 45 245Y 130 74 56 71 30 39 00 81 281FW 00 08 208 00 26 226 71 74 203QN 130 123 07 71 93 222 428 188 24ST 348 366 218 429 334 95 214 250 236

aAb2 stands for 1E10 and 3B11 MAbs Data in Kabat were taken from Brard et al(38) Kabat stands for 1450 sequences com-piled by Kabat et al(10) The third column (D) corresponds to the difference between the first two columns

TABLE 3 FREQUENCIES OF SELECTED AMINO ACIDS IN KABAT CDRS P3a

CDR1H CDR2H CDR3H

P3 Kabat D P3 Kabat D P3 Kabat D

RK 200 17 183 625 106 243 143 50 93DE 00 81 281 625 83 220 71 154 283Y 200 209 209 625 116 253 71 237 217FW 00 84 284 625 46 17 143 45 98QN 00 78 278 00 123 212 71 36 35ST 200 178 22 250 189 61 71 130 259

CDR1L CDR2L CDR3L

P3 Kabat D P3 Kabat D P3 Kabat D

RK 91 69 22 143 140 03 00 27 227DE 91 41 50 00 70 270 00 45 245Y 00 74 274 286 30 253 143 81 62FW 00 08 208 00 26 226 00 74 274QN 91 123 232 00 93 293 286 188 96ST 273 366 293 429 334 95 286 250 36

aP3 stands for P3 MAb Data in Kabat were taken from Brard et al(38) Kabat stands for 1450 sequences compiled by Kabatet al(10) The third column (D) corresponds to the difference between the first two columns

sponses whereas the immunization with anionic antigens in-duces cationic antibodies(38) Variable region sequences analy-sis of anti-histones(3439) and anti-DNA antibodies(40) havestrongly supported these initial observations Therefore thefinding of an acidic motif in H-CDR3 of 1E10 and 3B11 Ab2sand the general excess in acidic amino acids in their H-CDRssuggest that electrostatic interactions with complementary pos-itive residues in the hypervariable loops of P3 MAb are im-portant in g-type Ab2ndashAb1 interactions

Our results prompt us to suggest that H-CDR3 residues from1E10 and 3B11 Ab2 antibodies bind to residues in the H-CDR3of P3 MAb This could seem contradictory in view of the lackof mimicry shown by both Ab2 in syngeneic model(7) How-ever the study of the crystal structure of the complex betweenthe Fv fragments of the anti-lysozyme antibody D13 and theanti-D13 antibody E52 (Ab2b)(41) has shown that the mim-icking is functional and involves similar binding interactionsrather than exact topological replicas This concept is particu-larly relevant to explain the ability of certain antibodies tomimic gangliosides and other carbohydrate-bearing moleculesThere are several plausible explanations to be considered forthe failure of our anti-idiotypic antibodies to mimic N-glycolyl-containing gangliosides As Fields pointed out(41) molecularmimicking depends on several factors like the structure of theantigen the choice of the antibody (Ab1) and of anti-idiotypicantibody in addition to other numerous complexities of im-mune responses Finally it could be also the case that evenwhen our Ab2 carried the internal image of these antigens theycould not break the tolerance established against these wide-spread self-structures in mouse Interestingly in cancer patientswhere N-glycolyl-containing gangliosides are nonself antigensthe immunizations with 1E10 MAb induced a specific responseagainst these antigens (M Alfonso manuscript in preparation)

REFERENCES

1 Higashi H Naiki M Matuo S and Okouchi K Antigen of ldquoserumsicknessrdquo type of heterophile antibodies in human sera Identifica-tion as ganglioside with N-glycolylneuraminic acid Biochem Bio-phys Res Commun 197779388 ndash395

2 Hirabagashi Y Kasabura H Matsumoto M Higashi H Kato S Ka-sai N and Naiki M Specific expression of unusual GM2 ganglio-side with Hanganutziu-Deicher antigen activity on human coloncancer Jpn J Cancer Res (Gann) 198778251 ndash260

3 Hirabagashi Y Higashi H Kato S Taniguchi M and MatsumotoM Occurrence of tumor associated ganglioside antigens withHanganutziu-Deicher antigenic activity on human melanomas JpnJ Cancer Res (Gann) 198778614ndash620

4 Marquina G Waki U Fernaacutendez LE Kon K Carr A Valiente OPeacuterez R and Ando S Gangliosides expressed in human breast can-cer Cancer Res 1996565165ndash5170

5 Vaacutezquez AM Alfonso M Lanne B Karlsson KA Carr A BarrosoO Fernaacutendez LE Rengifo E Lanio ME Alvarez C Zeuthen J andPeacuterez R Generation of a murine monoclonal antibody specific forN-glycolylneuraminic acid-containing gangliosides that also rec-ognizes sulfated glycolipids Hybridoma 199514551 ndash556

6 Moreno E Lanne B Vaacutezquez AM Kawashima I Tai T FernaacutendezLE Karlsson K-A Aringngstroumlm J and Peacuterez R Delineation of the epi-tope recognized by an antibody specific for N-glycolylneuramini cacid-containing gangliosides Glycobiology 19988695ndash705

7 Vaacutezquez AM Peacuterez A Hernaacutendez AM Maciacuteas A Alfonso M

Bombino G and Peacuterez R Syngeneic anti-idiotypic monoclonal an-tibodies to an anti-NeuGc-containing ganglioside monoclonal an-tibody Hybridoma 199817527 ndash534

8 Orlandi R Gussow DH Jones PT and Winter G Cloning im-munoglobulin variable domains for expression by the polymerasechain reaction Proc Natl Acad Sci U S A 1989863833ndash3837

9 Sambrook J Fritsch EF and Maniatis T Molecular cloning A lab-oratory manual 2nd ed Cold Spring Harbor Laboratory PressCold Spring Harbor New York 1989

10 Kabat EA Wu TT Peery HM Gottesmann KS and Foeller C Se-quences of proteins of immunological interest 5th ed US De-partment of Health and Human Services US Government Print-ing Office Washington 1991

11 Bangs LA Sanz IE and Teale JM Comparison of D JH and junc-tional diversity in the fetal adult and aged B cell repertoires J Im-munol 19911461996 ndash2004

12 Lafaille JJ DeCloux A Bonneville M Takagaki Y and TonegawaS Junctional sequences of T cell receptor gd genes implicationsfor gd T cell lineages and for a novel intermediate of V-(D)-J join-ing Cell 198959859ndash870

13 Sakano H Maki R Kurosawa Y Roeder W and Tonegawa S Twotypes of somatic recombination are necessary for the generation ofcomplete immunoglobulin heavy-chain genes Nature 1980286676ndash683

14 Monestier M Losman MJ Novick KE and Aris JP Molecularanalysis of Mercury-induced antinucleolar antibodies in H-2S miceJ Immunol 1994152667 ndash675

15 Almagro JC Hernaacutendez I Ramiacuterez MC and Vargas-Madrazo EThe differences between the structural repertoires in germline genesegments of mice and humans implications for the molecularmechanism of the immune response Mol Immunol 1997341199ndash1214

16 Zenita K Hirashima K Shigeta K Hiraiwa N Takada AHashimoto K Fujimoto E Yago K and Kannagi R Northern hy-bridization analysis of VH gene expression in murine monoclonalantibodies directed to cancer-associated ganglioside antigens hav-ing various sialic acid linkages J Immunol 19901444442 ndash4451

17 Weng NP Ritter E Yucel E Zhang D Ritter G and Marcus DMJSpecificity and structure of murine monoclonal antibodies againstGM1 ganglioside Neuroimmunol 19945561ndash68

18 Jeong HD and Teale JM Contribution of the CD51 B cell to D-proximal VH family expression early in ontogeny J Immunol19901452725 ndash2729

19 Casali P Kasaian MT and Haughton G B-1 (CD5 B) cells InAutoimmunity Physiology and Disease Coutinho A andKazatchkine MD (Eds) Wiley-Liss Inc New York 1994 p 57

20 Kantor AB Merrill CE Herzenberg LA and Hillson JL An un-biased analysis ov V(H)-D-J(H) sequences from B-1a B-1b andconventional B cells J Immunol 19971581175 ndash1186

21 Blankenstein T Zoebelein G and Krawinkel U Analysis of im-munoglobulin heavy chain V-region genes belonging to the V NP-gene family Nucleic Acid Res 1984126887ndash5900

22 Kavaler J Caton AJ Staudt LM Schwartz D and Gerhard W Aset of closely related antibodies dominates the primary antibodyresponse to the antigenic site CB of the APR834 influenza virushemagglutinin J Immunol 19901452312 ndash2321

23 Robbins PF and Nisonoff A Characteristics of two idiotypic sub-families of murine anti-p-azobenzenearsonate antibodies J Im-munol 19871393868 ndash3877

24 Novick KE Fasy TM Losman MJ and Monestier M Polyreac-tive IgM antibodies generated from autoimmune mice and selectedfor histone-binding activity Int Immunol 199241103 ndash1111

25 Wysocki LJ Gridley T Huang S Grandea AG and Gefter MLSingle germline VH and V kappa genes encode predominating an-tibody variable regions elicited in strain A mice by immunizationwith p-azophenylarsonate J Exp Med 19871661ndash11

PEREZ ET AL220

sponses whereas the immunization with anionic antigens in-duces cationic antibodies(38) Variable region sequences analy-sis of anti-histones(3439) and anti-DNA antibodies(40) havestrongly supported these initial observations Therefore thefinding of an acidic motif in H-CDR3 of 1E10 and 3B11 Ab2sand the general excess in acidic amino acids in their H-CDRssuggest that electrostatic interactions with complementary pos-itive residues in the hypervariable loops of P3 MAb are im-portant in g-type Ab2ndashAb1 interactions

Our results prompt us to suggest that H-CDR3 residues from1E10 and 3B11 Ab2 antibodies bind to residues in the H-CDR3of P3 MAb This could seem contradictory in view of the lackof mimicry shown by both Ab2 in syngeneic model(7) How-ever the study of the crystal structure of the complex betweenthe Fv fragments of the anti-lysozyme antibody D13 and theanti-D13 antibody E52 (Ab2b)(41) has shown that the mim-icking is functional and involves similar binding interactionsrather than exact topological replicas This concept is particu-larly relevant to explain the ability of certain antibodies tomimic gangliosides and other carbohydrate-bearing moleculesThere are several plausible explanations to be considered forthe failure of our anti-idiotypic antibodies to mimic N-glycolyl-containing gangliosides As Fields pointed out(41) molecularmimicking depends on several factors like the structure of theantigen the choice of the antibody (Ab1) and of anti-idiotypicantibody in addition to other numerous complexities of im-mune responses Finally it could be also the case that evenwhen our Ab2 carried the internal image of these antigens theycould not break the tolerance established against these wide-spread self-structures in mouse Interestingly in cancer patientswhere N-glycolyl-containing gangliosides are nonself antigensthe immunizations with 1E10 MAb induced a specific responseagainst these antigens (M Alfonso manuscript in preparation)

REFERENCES

1 Higashi H Naiki M Matuo S and Okouchi K Antigen of ldquoserumsicknessrdquo type of heterophile antibodies in human sera Identifica-tion as ganglioside with N-glycolylneuraminic acid Biochem Bio-phys Res Commun 197779388 ndash395

2 Hirabagashi Y Kasabura H Matsumoto M Higashi H Kato S Ka-sai N and Naiki M Specific expression of unusual GM2 ganglio-side with Hanganutziu-Deicher antigen activity on human coloncancer Jpn J Cancer Res (Gann) 198778251 ndash260

3 Hirabagashi Y Higashi H Kato S Taniguchi M and MatsumotoM Occurrence of tumor associated ganglioside antigens withHanganutziu-Deicher antigenic activity on human melanomas JpnJ Cancer Res (Gann) 198778614ndash620

4 Marquina G Waki U Fernaacutendez LE Kon K Carr A Valiente OPeacuterez R and Ando S Gangliosides expressed in human breast can-cer Cancer Res 1996565165ndash5170

5 Vaacutezquez AM Alfonso M Lanne B Karlsson KA Carr A BarrosoO Fernaacutendez LE Rengifo E Lanio ME Alvarez C Zeuthen J andPeacuterez R Generation of a murine monoclonal antibody specific forN-glycolylneuraminic acid-containing gangliosides that also rec-ognizes sulfated glycolipids Hybridoma 199514551 ndash556

6 Moreno E Lanne B Vaacutezquez AM Kawashima I Tai T FernaacutendezLE Karlsson K-A Aringngstroumlm J and Peacuterez R Delineation of the epi-tope recognized by an antibody specific for N-glycolylneuramini cacid-containing gangliosides Glycobiology 19988695ndash705

7 Vaacutezquez AM Peacuterez A Hernaacutendez AM Maciacuteas A Alfonso M

Bombino G and Peacuterez R Syngeneic anti-idiotypic monoclonal an-tibodies to an anti-NeuGc-containing ganglioside monoclonal an-tibody Hybridoma 199817527 ndash534

8 Orlandi R Gussow DH Jones PT and Winter G Cloning im-munoglobulin variable domains for expression by the polymerasechain reaction Proc Natl Acad Sci U S A 1989863833ndash3837

9 Sambrook J Fritsch EF and Maniatis T Molecular cloning A lab-oratory manual 2nd ed Cold Spring Harbor Laboratory PressCold Spring Harbor New York 1989

10 Kabat EA Wu TT Peery HM Gottesmann KS and Foeller C Se-quences of proteins of immunological interest 5th ed US De-partment of Health and Human Services US Government Print-ing Office Washington 1991

11 Bangs LA Sanz IE and Teale JM Comparison of D JH and junc-tional diversity in the fetal adult and aged B cell repertoires J Im-munol 19911461996 ndash2004

12 Lafaille JJ DeCloux A Bonneville M Takagaki Y and TonegawaS Junctional sequences of T cell receptor gd genes implicationsfor gd T cell lineages and for a novel intermediate of V-(D)-J join-ing Cell 198959859ndash870

13 Sakano H Maki R Kurosawa Y Roeder W and Tonegawa S Twotypes of somatic recombination are necessary for the generation ofcomplete immunoglobulin heavy-chain genes Nature 1980286676ndash683

14 Monestier M Losman MJ Novick KE and Aris JP Molecularanalysis of Mercury-induced antinucleolar antibodies in H-2S miceJ Immunol 1994152667 ndash675

15 Almagro JC Hernaacutendez I Ramiacuterez MC and Vargas-Madrazo EThe differences between the structural repertoires in germline genesegments of mice and humans implications for the molecularmechanism of the immune response Mol Immunol 1997341199ndash1214

16 Zenita K Hirashima K Shigeta K Hiraiwa N Takada AHashimoto K Fujimoto E Yago K and Kannagi R Northern hy-bridization analysis of VH gene expression in murine monoclonalantibodies directed to cancer-associated ganglioside antigens hav-ing various sialic acid linkages J Immunol 19901444442 ndash4451

17 Weng NP Ritter E Yucel E Zhang D Ritter G and Marcus DMJSpecificity and structure of murine monoclonal antibodies againstGM1 ganglioside Neuroimmunol 19945561ndash68

18 Jeong HD and Teale JM Contribution of the CD51 B cell to D-proximal VH family expression early in ontogeny J Immunol19901452725 ndash2729

19 Casali P Kasaian MT and Haughton G B-1 (CD5 B) cells InAutoimmunity Physiology and Disease Coutinho A andKazatchkine MD (Eds) Wiley-Liss Inc New York 1994 p 57

20 Kantor AB Merrill CE Herzenberg LA and Hillson JL An un-biased analysis ov V(H)-D-J(H) sequences from B-1a B-1b andconventional B cells J Immunol 19971581175 ndash1186

21 Blankenstein T Zoebelein G and Krawinkel U Analysis of im-munoglobulin heavy chain V-region genes belonging to the V NP-gene family Nucleic Acid Res 1984126887ndash5900

22 Kavaler J Caton AJ Staudt LM Schwartz D and Gerhard W Aset of closely related antibodies dominates the primary antibodyresponse to the antigenic site CB of the APR834 influenza virushemagglutinin J Immunol 19901452312 ndash2321

23 Robbins PF and Nisonoff A Characteristics of two idiotypic sub-families of murine anti-p-azobenzenearsonate antibodies J Im-munol 19871393868 ndash3877

24 Novick KE Fasy TM Losman MJ and Monestier M Polyreac-tive IgM antibodies generated from autoimmune mice and selectedfor histone-binding activity Int Immunol 199241103 ndash1111

25 Wysocki LJ Gridley T Huang S Grandea AG and Gefter MLSingle germline VH and V kappa genes encode predominating an-tibody variable regions elicited in strain A mice by immunizationwith p-azophenylarsonate J Exp Med 19871661ndash11

PEREZ ET AL220

26 Kaartinen M Pelkonen E Even J and Makela O V genes of theprimary antibody response of C57BL10 mice to the hapten pheny-loxazolone Eur J Immunol 1988181095 ndash1100

27 Roth MS Weiner GJ Allen EA Terry VH Harnden CE BoehnkeM Kaminsk MS and Ginsburg D Molecular characterization ofanti-idiotype antibody resistant variants of a murine B cell lym-phoma J Immunol 1990145768 ndash777

28 Shlomchik MJ Marshak-Rothstein A Wolfowicz CB RothsteinTL and Weigert MG The role of clonal selection and somatic mu-tation in autoimmunity Nature 1987328805 ndash811

29 Milstein C From the structure of antibodies to the diversificationof the immune response Scand J Immun 199337386 ndash398

30 Chen J Borden P Liao J and Kabat EA Variable region cDNAsequences of three mouse monoclonal anti-idiotypic antibodies spe-cific for anti-a(1 R 6) dextrans with groove- or cavity-type com-bining sites Mol Immunol 1992291121 ndash1129

31 David D and Zouali M Variable region light chain genes encod-ing human antibodies to HIV-1 Mol Immunol 19953277ndash88

32 Mak TW and Simard JJL B-Lymphocyte genes In Handbook ofImmune Response Genes Plenum Press New York 1998 pp109ndash70

33 Meek K Hasemann C Pollock B Alkan SS Brait M Slaoui MUrbain J and Capra JD Structural characterization of antiidiotypicantibodies Evidence that Ab2s are derived from the germline dif-ferently than Ab2s J Exp Med 1989169519ndash533

34 Brard F Petit S Coquerel G Gilbert D Koutouzov S Peacuterez Gand Tron F Modeling of anti-nucleosome immunoglobulin Fv do-mains Analysis of electrostatic interactions Mol Immunol 199734793ndash807

35 Padlan EA Anatomy of the antibody molecule Mol Immunol199431169 ndash217

36 Ichihara Y Hayashida H Miyazawa S and Kurosawa Y OnlyDFL16 DSP2 and DQ52 gene families exist in mouse im-munoglobulin heavy chain diversity gene loci or which DFL16

and DSP2 originate from the same primoridal DH gene Eur J Im-munol 1989191849 ndash1854

37 Gu H Foster I and Rajewsky K Sequence homologies N sequenceinsertion and JH gene utilization in VHDJH joining implicationsfor the joining mechanism and the ontogenetic timing of Lyl B celland B-CLL progenitor generation EMBO J 199092133 ndash2140

38 Sela M and Mozes E Dependence of the chemical nature of an-tibodies on the net electrical charge of antigens Proc Natl AcadSci USA 196755445

39 Monestier M Fasy JM Losman MJ Novick KE and Muller SStructure and binding properties of monoclonal antibodies to corehistones from autoimmune mice Mol Immunol 1993301069ndash1075

40 Eilat D and Anderson WF Structure-function correlates of au-toantibodies to nucleic acids Lessons from immunochemical ge-netic and structural studies Mol Immunol 1994311377ndash1390

41 Fields BA Goldbaum FA Ysern X Poljak RJ and Mariuzza RAMolecular basis of antigen mimicry by an anti-idiotope Nature1995374739 ndash742

Address reprint requests toAlexis Peacuterez

Department of Antibody EngineeringCenter of Molecular Immunology

PO Box 16040Havana 11600

Cuba

E-mail alexisictcimsldcu

Received for publication October 23 2000 Accepted after re-visions April 21 2001

GENETIC BASIS OF g-TYPE Ab2 FROM GLYCOLYLATED GANGLIOSIDES 221