Human Ig heavy chain constant regions are encoded by a cluster of genes, the IGHC locus, on 14q32.3. Several forms of IGHC deletions and duplications spanning one to five genes have been described in different populations, with frequencies of 1.5–3.5% and 4.5–44%, respectively. Despite the common occurrence of these gene rearrangements, little is known about the breakpoint sites; evidence obtained from deletions in the IGHC locus and in other regions of the human genome suggests that they preferentially occur in highly homologous regions and might be favored by a variety of recombinogenic signals. We present here a detailed study of three homozygotes for the most common type of IGHC multiple gene deletion, spanning the A1-GP-G2-G4-E genes. Using a combination of Southern blotting, long-range PCR, and automated sequencing, the unequal crossover events of all of the six studied haplotypes have been mapped to a region of ∼2 kb with almost complete homology between EP1-A1 and E-A2, flanked by two minisatellites. These results are consistent with the hypothesis that segments of complete homology may be required for efficient homologous recombination in humans. The possible role of minisatellites as recombination signals is inferred, in agreement with current knowledge.

The human Ig heavy chain constant region locus (IGHC)4 is a multigene family composed of 11 genes, in the order M, D/G3, G1, EP1, A1/GP/G2, G4, E, A2 on chromosome 14q32.3 (1, 2, 3, 4). In humans, the locus is very polymorphic both at the protein (Gm and Am allotypes) and DNA (i.e., restriction fragment length polymorphism (RFLP), switch regions) level. Its considerable genetic variation is also highlighted by the high frequency of haplotypes carrying gene deletions or duplications (4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21). An extreme example is the G4 gene duplication, which is present in Caucasians with a gene frequency of ∼45% (22). Seven types of deletions and 10 types of duplications have been reported so far, spanning from 20 to 150 kb. The most frequent deletion/duplication encompasses the A1, GP, G2, G4, and E genes (∼150 kb). Analysis of the overall structure of the deleted/duplicated haplotypes suggests that they probably originated through unequal crossing-over during meiosis, due to mispairing between highly homologous regions, especially of the two large G-G-E-A clusters.

Haplotypes with deletions or duplications are present with a variable frequency in all populations analyzed thus far, with a significantly higher frequency of duplications over deletions (Refs. 20 and 21 and the references therein). According to the unequal crossover model, duplications and deletions should originate at the same rate. Therefore, the higher frequency of duplicated vs deleted haplotypes suggests that selective pressures may be operating on the locus to eliminate deletions and/or favor duplications.

Our understanding of the origins of this remarkable genetic instability and our interpretation of the role that selection may play in determining the distribution of the variant haplotypes in the population depend upon establishing the molecular mechanisms that lead to the individual recombination events. So far, the recombination sites of only two deletions have been mapped. The first study, based on a Southern blot approach followed by library construction and sequencing, succeeded in locating the breakpoint of a G1-G4 gene deletion to a highly homologous region located 3′ of G3 and G4, outside of the switch sequences (23). More limited analyses of an A1-E deletion mapped the breakpoint to a large region between the E genes and the switch A regions (24, 25).

In this report, we use Southern blotting, long-range PCR, and sequencing to define the deletion breakpoint in three individuals homozygous for a deletion from A1 to E, which is the most common type of deletion in the Italian population (20).

Peripheral blood samples were obtained from three unrelated subjects (BON, DEM, and MON) and 10 controls from different peninsular Italian regions. These three subjects are homozygous for a deletion of ∼150 kb encompassing the A1, GP, G2, G4, and E genes. Analysis of IGHC RFLP markers and Gm allotypes suggests a different origin of the deleted haplotypes in MON and DEM (10, 13); BON is homozygous for all tested markers (26). Informed consent was obtained from all subjects.

A total of 5 μg of genomic DNA, extracted using standard phenol/chloroform methods (27), was digested with SacI, PstI, PvuII, and SacI+PstI restriction endonucleases (Roche Diagnostics, Monza, Italy) and run in 0.8% TBE (0.045 M Tris-borate, 0.002 M EDTA)-agarose. DNA was transferred onto Hybond N+ nylon filters (Amersham Pharmacia, Little Chalfont, U.K.) and hybridized to the following genomic probes using standard techniques:

pHSα2-2.6 is a 2.6-kb SacI/PstI fragment (see Fig. 1 A) subcloned from the λ-TOU ε-α phage (kindly provided by M. P. Lefranc, University of Montpellier II, Montpellier, France).

FIGURE 1.

Structure of the two homologous E-A subregions within the human IGHC locus. A, the EP1-A1 region and the E-A2 region. The restriction sites are shown above, and the published sequences are shown beneath the map, with their GenBank accession numbers. The pHSα1-0.9 and pHSα2-2.6 plasmid clones, used as probes, are indicated. Circled restriction sites are subregion-specific. VNTR sequences are indicated by hatched boxes. Switch α and Iα are indicated by gray boxes. B, Rationale for hemi-nested, long-range PCR. The primers are positioned with respect to the maps in A. Primers 886 and 889 are specific for the EP1 and E regions, respectively, whereas primers 399, 1817, 562, and 891 are common for both regions. To amplify the regions downstream of EP1 and E, the forward primers 886 and 889 were used in distinct reactions with the same reverse primer 891, located ∼9 kb downstream of each gene. This PCR product was used in a hemi-nested reaction to obtain regions A, B, and C distinctive of the EP1-A1 or the E-A2 regions.

FIGURE 1.

Structure of the two homologous E-A subregions within the human IGHC locus. A, the EP1-A1 region and the E-A2 region. The restriction sites are shown above, and the published sequences are shown beneath the map, with their GenBank accession numbers. The pHSα1-0.9 and pHSα2-2.6 plasmid clones, used as probes, are indicated. Circled restriction sites are subregion-specific. VNTR sequences are indicated by hatched boxes. Switch α and Iα are indicated by gray boxes. B, Rationale for hemi-nested, long-range PCR. The primers are positioned with respect to the maps in A. Primers 886 and 889 are specific for the EP1 and E regions, respectively, whereas primers 399, 1817, 562, and 891 are common for both regions. To amplify the regions downstream of EP1 and E, the forward primers 886 and 889 were used in distinct reactions with the same reverse primer 891, located ∼9 kb downstream of each gene. This PCR product was used in a hemi-nested reaction to obtain regions A, B, and C distinctive of the EP1-A1 or the E-A2 regions.

Close modal

pHSα1-0.9 is a 0.9-kb PstI-PvuII fragment (see Fig. 1 A) subcloned from pH-Igα11 (kindly provided by T. Honjo, Kyoto University, Kyoto, Japan).

These probes cross-hybridize with two regions between EP1-A1 and E-A2 genes.

The subcloning of λ-TOU ε-α and pH-Igα11 was performed using standard procedures (27). The clones obtained were sequenced using a primer walking strategy and dye-terminator chemistry on a model 377 ABI Prism automatic sequence analyzer (Applied Biosystems, Perkin-Elmer, Monza, Italy). Data were elaborated using the Navigator program (Perkin-Elmer) and compared with GenBank with BLAST.

To amplify the regions downstream of EP1 and E, specific forward primers 886 (5′-acactgggttcagctgacactacactgggttc-3′) and 889 (5′-ggcctcacacagccctccggtgtaccacag-3′) were used with the same reverse primer (891, 5′-tgtgtccacagatgtggctggatgctgagg-3′) located 9 kb downstream of each gene. The reaction was performed using the “Expand Long Template PCR System” (Roche Diagnostics) according to the manufacturer’s instructions with the following cycling parameters: 2 min at 93°C; 10 cycles of 93°C for 10 s, 65°C for 30 s, and 68°C for 12 min; 20 cycles of 93°C for 10 s, 65°C for 30 s, and 68°C for 12 min with an increase of 20 s per cycle; and a final extension of 10 min at 68°C.

Hemi-nested PCRs of regions A (Fig. 1 B; primers 886 or 889 and 399: 5′-agtgaccggagacccaggggcctg-3′) and B (primers 562 (5′-tgggcaaaaccagaagaggcc-3′) and 891) were amplified from one-fiftieth of the first PCR using the same long-range PCR conditions described above and the following parameters: 94°C for 2 min; 10 cycles of 94°C for 10 s, 65°C for 30 s, 68°C for 6 min; 20 cycles of 94°C for 10 s, 65°C for 30 s, 68°C for 6 min with a 20-s increase per cycle; and a final extension of 10 min at 68°C. The two products were digested using frequent-cutting restriction endonucleases and separated on a 3% low melting point agarose or 3:1 Nusieve agarose (FMC, Rockland, ME) in 1× TBE buffer.

Restriction sites were localized on the sequence using the mapping facilities at the Biocomputing Service Group (Heidelberg, Germany). The critical region C was amplified in three overlapping fragments, using one-fiftieth of the 9-kb product as a template. Primer pairs were: 1817 (5′-ctcccaggcccctggtctccgggtcatcc-3′)/1843 (5′-ggggagtcagctttgtctgtgctcagtggg-3′), 562/2265 (5′-cgatggaattttctgtgcgtgcgtgaggacgcctgcc-3′), and E05 (5′-tttctctgcccgcaccctga-3′)/891.

Region C was screened for the following sequences involved in hot spots and chromosomal rearrangements using Lalign 2.0 DNA analysis software (52): the χ consensus sequence GCTGGTGG of Escherichia coli, which stimulates generalized recombination (28); the human hypervariable minisatellite consensus sequences GGGCAGGARG (29) and AGAGGTGGGCAGGTGG (30); the consensus sequences for Translin (ATGCAG and GCCCWCCT), which are found at chromosomal translocation breakpoints in lymphoid malignancies (31); the ATGACGT sequence present in the recombinogenic M26 allele of Schizosaccharomyces pombe ade6 (32); the tetranucleotide repeats TCTG4–6 and CAGG7–9 and the 289-bp consensus sequence identified in two recombinational hot spots of mouse MHC, between the Aβ3 and Aβ2 genes, and in the Eβ gene, respectively (33, 34, 35, 36); and the retrotransposon long terminal repeat sequence TCATACACCACGCAGGGGTAGAGGACT, located at the recombinational hot spot of the E β gene in mouse MHC (37).

The structure of the two homologous EP1-A1 and E-A2 regions as reported in the literature (2) and integrated by present data is shown in Fig. 1 A.

Southern blot analyses of BON, DEM and MON had previously demonstrated a BamHI-RFLP containing EP1 at the 5′ end of the deleted region (i.e., absence of the BamHI site circled in Fig. 1) and a PstI-RFLP containing A2 at its 3′ end (Refs. 10, 13, and 26; and data not shown); therefore, we assumed that the breakpoint was located in the intergenic region between EP1 and A2 in all six deleted chromosomes.

In normal chromosomes, three restriction sites distinguish the EP1-A1 from the E-A2 region (Fig. 1 A): a PvuII site upstream of the A1 switch absent in the corresponding position upstream of A2, a SacI site present only in the middle of the switch A2 sequence, and the above-cited BamHI site present within the E gene but not within EP1.

Thus, genomic DNA of the three probands was further analyzed with probes pHSα2-2.6 and pHSα1-0.9 (see Fig. 2) that, after digestion with the aforementioned enzymes, hybridize to fragments specific for the EP1-A1 or the E-A2 regions (Fig. 3) (38).

FIGURE 2.

Southern blot analysis of the EP1-A1 and E-A2 regions in the three deleted subjects (MON, DEM, and BON) and a control (C). The probes and restriction enzymes used are indicated above and below each panel, respectively. Polymorphisms are present in the fragments containing the A1 and A2 switch regions, as shown by PvuII-cut genomic DNAs hybridized with pHSα2-2.6 as well as by PstI and SacI/PstI hybridized with pHSα1-0.9. On SacI and PvuII blots, pHSα1-0.9 gives the same pattern as pHSα2-2.6 (data not shown). The pHSα2-2.6 probe on PstI filters shows a 3.9-kb band in BON and DEM but two bands of 3.9 and 4.4 kb in MON and in the control (B). This variation is restricted to the SacI/PstI fragment and derives from a 33-mer VNTR that is present in both the EP1-A1 and E-A2 fragments (see text). Southern blots should be interpreted with caution because both homologous regions EP1-A1 and E-A2 can give rise to 2.15/2.7-kb SacI/PstI bands, thereby hampering the genotype interpretation in deleted chromosomes.

FIGURE 2.

Southern blot analysis of the EP1-A1 and E-A2 regions in the three deleted subjects (MON, DEM, and BON) and a control (C). The probes and restriction enzymes used are indicated above and below each panel, respectively. Polymorphisms are present in the fragments containing the A1 and A2 switch regions, as shown by PvuII-cut genomic DNAs hybridized with pHSα2-2.6 as well as by PstI and SacI/PstI hybridized with pHSα1-0.9. On SacI and PvuII blots, pHSα1-0.9 gives the same pattern as pHSα2-2.6 (data not shown). The pHSα2-2.6 probe on PstI filters shows a 3.9-kb band in BON and DEM but two bands of 3.9 and 4.4 kb in MON and in the control (B). This variation is restricted to the SacI/PstI fragment and derives from a 33-mer VNTR that is present in both the EP1-A1 and E-A2 fragments (see text). Southern blots should be interpreted with caution because both homologous regions EP1-A1 and E-A2 can give rise to 2.15/2.7-kb SacI/PstI bands, thereby hampering the genotype interpretation in deleted chromosomes.

Close modal
FIGURE 3.

Observed alleles for A1 and A2 RFLPs. Multiple fragment sizes within the same box indicate allelic variants.

FIGURE 3.

Observed alleles for A1 and A2 RFLPs. Multiple fragment sizes within the same box indicate allelic variants.

Close modal

On SacI-digested DNA, pHSα2-2.6 and pHSα1-0.9 showed a single 4.8-kb band in BON and DEM and a heterozygous 4.8/5.4-kb pattern in MON, with the latter band arising from a polymorphism (Fig. 2,A); the normal EP1-A1 fragment of 7.1 kb was lacking in all three individuals. Similarly, both probes showed on PvuII filters a 7.1-kb band in BON and DEM and 7.1/7.8-kb bands in MON; again, no EP1-A1-specific fragment (5.4 kb) was detected. Lack of A1 and retention of A2 was also supported by the hybridization of probe pHSα1-0.9 to double SacI/PstI digest (Fig. 2,C): BON, DEM, and MON showed bands ∼2.0 kb, confirming the presence of the switch A2 SacI site. Furthermore, the two probes show the specific 2.9- to 3.1-kb fragments on PstI filters, indicating that the region upstream of A2 is not deleted (Fig. 2,C). On the basis of these results, the breakpoints map upstream of the indicated SacI and PvuII sites in all probands (six haplotypes) and can therefore be assigned to an interval of ∼10 kb between these sites and the aforementioned BamHI site (see Fig. 1 A).

The genomic sequences available in GenBank only partially cover the intervals between EP1-A1 and E-A2. Unsequenced regions of ∼2.6 kb and ∼6.5 kb lay between E-A2 and between EP1-A1, respectively (see Fig. 1 A). We tried to fill these gaps by sequencing the entire 2.6-kb SacI-PstI subclone from λ-TOU ε-α phage for E-A2 and the corresponding region from plasmid pH-Igα11 for EP1-A1, as well as the 5′ adjacent SacI-SacI fragment of 2.3 kb.

The two sequences were almost completely identical (identity equaling 98%). Their analysis in deleted haplotypes was exploited to refine breakpoint mapping by restriction analysis and sequencing (see below).

Variable numbers of tandem repeats (VNTRs) were also detected in the corresponding positions of the two regions. A 33-mer VNTR sequence (consensus AAGGATTTCD AGGTGARGGM VCCCTGGATT GTC) spans approximately one-third of the SacI-PstI fragment (see Fig. 1,A) in both the EP1-A1 and the E-A2 regions; because of its large extension (>0.9 kb), we could not overlap sequences from the two strands. PCR with primers flanking the minisatellite allowed us to establish a wide length variability on genomic DNA, with smaller alleles of ∼25–30 repeats and larger alleles of ∼40–45 repeats. The consensus sequence did not reveal any homology with known sequences or hypervariable minisatellites. The VNTR polymorphisms can show identical patterns in both the EP1-A1 and E-A2 regions; this makes them less suitable for defining deletion breakpoints (see also Fig. 2 B).

In addition, within the 0.5-kb SacI-PstI fragment of the SacI-SacI subclone from both the EP1-A1 and E-A2 regions, sequencing revealed a VNTR with repeats of 40–41 bp (consensus TCCCTTCACC CACCCCTCCT GCAYCTATC YDCAGCCCCC A), each possessing a canonical minisatellite core (CNTCCTGCCC) and a sequence (CCCACCCC) perfectly matching Nakamura’s consensus (CCCACNNC) (39). Two repeats also showed a χ-like sequence (CCACAGC) with a single nucleotide missing from the consensus (CCACCAGC) (28).

Most of the critical interval in which the breakpoints are located was amplified by long-range PCR. Two specific forward primers (886 for EP1 or 889 for E) were designed where that the two sequences diverge (i.e., EP1 becomes a truncated pseudogene); a common reverse primer (891) was tailored 9 kb downstream. For each region, two hemi-nested reactions (primers indicated in Fig. 1 B) produced fragment A (∼3.5 kb) and fragment B (∼4 kb).

The two amplification products from four control subjects (eight haplotypes) were then tested with several frequent-cutting enzymes to detect patterns specific for the EP1-A1 and the E-A2 regions. Four enzymes (DdeI, AluI, MaeIII, and SmaIII) in fragment A and four (AluI, DdeI, NlaIII, and MaeIII) in fragment B were found to differentiate the EP1-A1 region from the E-A2 region (Fig. 4 A).

FIGURE 4.

Analysis of regions A and B amplified by hemi-nested, long-range PCR. A, The EP1-A1 or E-A2 discriminatory restriction sites are indicated. B, The results for BON, DEM, and MON are indicated below. Numbers 1 and 2 refer to individual chromosomes. Black dots indicate EP1-A1 region sites, whereas white dots indicate the E-A2-associated restriction sites. In region B, DdeI and MaeIII (data not shown) display the same differences as AluI.

FIGURE 4.

Analysis of regions A and B amplified by hemi-nested, long-range PCR. A, The EP1-A1 or E-A2 discriminatory restriction sites are indicated. B, The results for BON, DEM, and MON are indicated below. Numbers 1 and 2 refer to individual chromosomes. Black dots indicate EP1-A1 region sites, whereas white dots indicate the E-A2-associated restriction sites. In region B, DdeI and MaeIII (data not shown) display the same differences as AluI.

Close modal

In region A, both chromosomes of probands BON and MON showed only EP1-A1-specific sites (Fig. 4,B); proband DEM was shown to be homozygous for EP1-A1 sites with SmaI and MaeIII, but EP1-A1/E-A2 was heterozygous with DdeI and HinfI, which might indicate a gene conversion event on one chromosome (DEM2 in Fig. 4,B). In fragment B, all sites showed an E-A2-typical pattern in the three probands. Thus, the breakpoints could be restricted to a region comprising the two VNTRs and their flanking regions (region C in Fig. 1), which was further analyzed by direct sequencing of overlapping PCR products.

The sequencing of region C in four control subjects in λ-TOU ε-α and in pH-Igα11 permitted us to draw the map shown in Fig. 5. Only positions differing between EP1-A1 and E-A2 in all of the chromosomes studied are reported.

FIGURE 5.

Sequence analysis of region C. Nucleotide differences between EP1-A1 and E-A2 regions and polymorphisms are indicated. VNTR sequences are shown as gray boxes. The double slash in the 33-mer VNTR indicates that it has not been completely sequenced. Numbered arrows indicate the position and orientation of sequences homologous to known recombinogenic signals, reported in detail in Table I. The asterisk indicates a match of 29/32 nucleotides, with a low complexity region in the IGLC (Ig light chain constant region) locus, pinpointed by BLAST searching (ctcattcacagacacacacgcaggcacacactc). Below, the analysis in BON, DEM, and MON is represented as black and white dots; these dots indicate the presence of an EP1-A1 or an E-A2 nucleotide at a given position.

FIGURE 5.

Sequence analysis of region C. Nucleotide differences between EP1-A1 and E-A2 regions and polymorphisms are indicated. VNTR sequences are shown as gray boxes. The double slash in the 33-mer VNTR indicates that it has not been completely sequenced. Numbered arrows indicate the position and orientation of sequences homologous to known recombinogenic signals, reported in detail in Table I. The asterisk indicates a match of 29/32 nucleotides, with a low complexity region in the IGLC (Ig light chain constant region) locus, pinpointed by BLAST searching (ctcattcacagacacacacgcaggcacacactc). Below, the analysis in BON, DEM, and MON is represented as black and white dots; these dots indicate the presence of an EP1-A1 or an E-A2 nucleotide at a given position.

Close modal

The differences used to map the breakpoint sites in deleted subjects are summarized in Fig. 5. The first seven base differences locate the breakpoint 3′ of the 41-mer VNTR. Two informative base differences are present 5′ of the second VNTR, whereas there are 10 single base differences and one deletion at the 3′. Thus, the crossover event could be mapped to the 2-kb region of high homology between the two VNTRs in all subjects. The two nucleotide differences in the middle of the 2-kb region of homology are polymorphic, and were not considered for breakpoint mapping.

Known signals favoring recombination were searched for in region C (see Materials and Methods). One χ sequence and three binding sites for Translin showed complete identity with their consensus; in addition, a number of sites showed a single nucleotide difference from other putative recombinogenic sequences (Fig. 5 and Table I).

Table I.

Homologies to recombinogenic consensus sequences in region Ca

Sequence DesignationConsensusSpeciesSequence Foundb
Translin consensus GCCCWCCT Homo sapiens GCCGACCT 
Ade6 sequence ATGACGT Seccaromyces pombe ATGACC
χ sequence GCTGGTGG Escherichia coli GCTGGTGG in EP1-A1 
    GCTGGCGG in E-A2 
Ade6 sequence ATGACGT S. pombe AGGACGT 
Hypervariable minisatellite GGGCAGGARG H. sapiens GTGCAGGAGG 
Hypervariable minisatellite GGGCAGGARG H. sapiens GGGAAGGAAG 
Translin consensus ATGCAG H. sapiens ATGCAG 
χ sequence GCTGGTGG E. coli GCTGATGG 
Translin consensus GCCCWCCT H. sapiens GCCCACCA 
10 Translin consensus GCCCWCCT  GCCCACCT rev 
11 Translin consensus ATGCAG H. sapiens ATGCAG 
12 Translin consensus GCCCWCCT H. sapiens GCCCAGCT 
13 Ade6 sequence ATGACGT S. pombe ATGCCGT 
Sequence DesignationConsensusSpeciesSequence Foundb
Translin consensus GCCCWCCT Homo sapiens GCCGACCT 
Ade6 sequence ATGACGT Seccaromyces pombe ATGACC
χ sequence GCTGGTGG Escherichia coli GCTGGTGG in EP1-A1 
    GCTGGCGG in E-A2 
Ade6 sequence ATGACGT S. pombe AGGACGT 
Hypervariable minisatellite GGGCAGGARG H. sapiens GTGCAGGAGG 
Hypervariable minisatellite GGGCAGGARG H. sapiens GGGAAGGAAG 
Translin consensus ATGCAG H. sapiens ATGCAG 
χ sequence GCTGGTGG E. coli GCTGATGG 
Translin consensus GCCCWCCT H. sapiens GCCCACCA 
10 Translin consensus GCCCWCCT  GCCCACCT rev 
11 Translin consensus ATGCAG H. sapiens ATGCAG 
12 Translin consensus GCCCWCCT H. sapiens GCCCAGCT 
13 Ade6 sequence ATGACGT S. pombe ATGCCGT 
a

Position numbering refers to Fig. 5.

b

Sequence positions differing from consensus are boldface and underlined.

Unequal crossing-over between mispaired homologous regions is the cause of many genetic disorders and of evolutionary variability in multigene families (reviewed in Ref. 40). One notable example is the IGHC locus, where deletions/duplications of single or multiple genes reach a polymorphic frequency. Because homology is often as elevated in intergenic regions of IGHC as in the coding sequences, intergenic exchanges giving rise to hybrid genes are rare among recombinant haplotypes (41). The first recombinational breakpoint within the IGHC, described by Keyeux et al. (23) in a G1-G4 homozygous deleted individual, occurred in a highly homologous region, outside of the switch sequences, which the authors named “hsg”. These regions were subsequently shown to contain repetitive MER-4 elements (42). Hsg homology regions were also located downstream of each G gene except G2, suggesting that even other deletions (e.g., of EP1-GP and EP1-G4) could derive from intrachromosomal recombinations between these regions. By analogy, the A1-E and GP-A2 deletions might arise through recombinations between homologous regions downstream of the E or A genes but different from hsg. In general, it could be argued that extensive stretches of high sequence homology are necessary for the generation of deletions and duplications in the IGHC locus.

The most common deletion/duplication, involving the genes from A1 to E, depends upon mispairing between two G-G-E-A blocks. Recently, a homozygous A1-E deletion (T17) was independently investigated by Wiebe et al. (24) and Chen et al. (25). Using a Southern blot approach, the deletion breakpoint was located in two different fragments due to the different interpretations of the RFLP pattern. One group had assigned the breakpoint to a 5.2-kb SacI/SacI fragment containing the switch A2. However, this region is highly polymorphic because of both the switch and the 33-mer VNTR; therefore, the fragment was likely a normal polymorphism unrelated to the deletion. Chen et al. (25) restricted the breakpoint region to a 2.0-kb BamHI-SacI fragment just downstream of the EP1 gene. Unfortunately, this conclusion is also incorrect because Southern blotting can only place the T17 breakpoint between the discriminatory sites reported in Fig. 1. It should be noted that the PstI digestion used by these authors cannot distinguish the EP1-A1 from the E-A2 derived fragments. Using long-range PCR, we have evidence that fragments of the same length can originate from either region, likely due to the 33-mer VNTR (see Fig. 3 and the legend to Fig. 2).

Our study localizes the recombination for the A1-E deletion in three unrelated homozygotes within the intergenic region between EP1/E and A1-A2 genes, which show ∼98% sequence identity. More precisely, the breakpoint was pinpointed in all individuals by sequencing to a ∼2-kb region of absolute homology (only two nucleotide differences) between two VNTRs. Furthermore one haplotype (DEM2) also shows an intermixed restriction site pattern for EP1-A1 and E-A2, compatible with a gene conversion event in proximity of the exchange site. Gene conversion events have been reported by several authors in association with meiotic crossing-over and other homologous recombination events. These findings are in agreement with current double-strand break models of recombination (43).

Our data suggest that in humans, long stretches of identity favor recombination between two sequences. Indeed, sequence homology has been demonstrated in recombinational hot spots such as those of the mouse MHC locus (44), pseudoautosomal region of sex chromosomes (45), hereditary neuropathy with liability to pressure palsies, and Charcot-Marie-Tooth 1A disease (46).

In addition to sequence identity, other specific factors or signals may have targeted the recombination event between the two G-G-E-A blocks. In fact, the same mispairing could theoretically also generate other deleted/duplicated haplotypes involving the GP-A2, EP1-G4, and G1-G4 genes: however, these are significantly rarer. Variable selective pressures are one potential explanation for this difference. However, it seems unlikely for a deletion of the A1-GP-G2-G4-E genes to be significantly favored in terms of fitness over one spanning, for instance, GP-G2-G4-E-A2. Alternatively, specific recombination-prone sequences between EP1-A1 and/or between E-A2 may have favored recombination in the interval. The low linkage disequilibrium level observed in particular within the E-A2 region (47, 48, 49) supports this model.

Repetitive DNA sequences have been proposed as preferred sites of recombination from microorganisms to humans. In the IGHC locus, the repetitive “switch” elements upstream of each functional CH gene are good candidates in light of their programmed involvement in somatic switch recombination. In an insertional mutagenesis assay, the Sμ region was shown to represent a recombinogenic hot spot with a 100-fold higher insertion frequency than the average genomic region (50).

In this respect, the two VNTRs found in the surroundings of the recombinational region may be particularly important. The consensus core sequence of human hypervariable minisatellites shows similarity to the E. coli χ sequence (39), a hot spot that promotes generalized recombination catalyzed by RecA-RecBC (51). Direct experimental evidence has also been obtained showing that minisatellite sequences enhance the ability of a DNA molecule to undergo recombination at a distance, perhaps acting as recognition sites for a specific enzyme or macromolecular complex (30).

In our study, the breakpoint critical region in all six investigated haplotypes lay in the same ∼2-kb sequence containing two VNTRs and several consensus sequences that are known to be involved in recombination in humans, yeast, and bacteria. This strongly suggests, although it does not directly prove, that these recombinogenic elements may represent the site of the unequal crossing-over, and may even coincide with an alleged recombination hot spot of the IGHC locus.

We thank Profs. M. P. Lefranc and T. Honjo for providing clones, A. Plebani and M. G. Guerra for patients, Prof. N. Migone for critical discussion, and C. Boccazzi for technical support. We are also very grateful to Dr. Enza Ferrero for critical revision of the manuscript.

1

This work was supported by Ministero dell’Universitá e della Ricerca Scientifica e Tecnologica (60%).

2

Electronic database information: Biocomputing Service Group, http://www.genius.embnet.dkfz-heidelberg.de:8080/menu/; Lalign 2.0 DNA analysis software, http://vega.igh.cnrs.fr/bin/lalign-guess.cgi. EMBL accession numbers are as follows: intergenic EP1-A1 region, AJ238957 and AJ238959; intergenic E-A2 region, AJ238958 and AJ238960; breakpoint region in BON, AJ238961; breakpoint region in DEM, AJ238962; breakpoint region in MON, AJ238963; 3′ of the 33-mer VNTR for all of the subjects, AJ238964.

4

Abbreviations used in this paper: IGHC, Ig heavy chain constant region locus; RFLP, restriction fragment length polymorphism; VNTR, variable number of tandem repeats.

1
Ravetch, J. V., U. Siebenlist, S. Korsmeyer, T. Waldman, P. Leder.
1981
. Structure of the human immunoglobulin μ locus: characterisation of embryonic and rearranged D and J genes.
Cell
27
:
583
2
Flanagan, J. G., T. H. Rabbitts.
1982
. Arrangement of the human immunoglobulin heavy chain constant region genes implies evolutionary duplication of a segment containing γ, ε, and α genes.
Nature
300
:
709
3
Lefranc, M. P., G. Lefranc, T. H. Rabbitts.
1982
. Inherited deletion of immunoglobulin heavy chain constant region genes in normal human individuals.
Nature
300
:
760
4
Hofker, M. H., M. A. Walter, D. W. Cox.
1989
. Complete physical map of the human immunoglobulin heavy chain constant region gene complex.
Proc. Natl. Acad. Sci. USA
86
:
5567
5
Lefranc, M. P., G. Lefranc, G. G. DeLange, T. A. Out, P. J. van den Broek, J. van Nieuwoop, J. Radl, A. N. Helal, H. Chaabani, E. van Loghem, et al
1983
. Instability of the human immunoglobulin heavy chain constant region locus indicated by different inherited chromosomal deletions.
Mol. Biol. Med.
1
:
207
6
Migone, N., S. Oliviero, G. G. DeLange, D. L. Delacroix, D. Boschis, F. Altruda, L. Silengo, M. DeMarchi, A. O. Carbonara.
1984
. Multiple gene deletions in the human immunoglobulin heavy chain cluster.
Proc. Natl. Acad. Sci. USA
81
:
5811
7
Chaabani, H., N. T. Bech-Hansen, D. W. Cox.
1986
. A multigene deletion within the immunoglobulin heavy chain region.
Am. J. Hum. Genet.
37
:
1164
8
Bech-Hansen, N. T., D. W. Cox.
1986
. Duplication of the human immunoglobulin heavy chain γ2 gene.
Am. J. Hum. Genet.
38
:
67
9
Bottaro, A., M. DeMarchi, N. Migone, A. O. Carbonara.
1989
. Pulsed-field gel analysis of human immunoglobulin heavy-chain constant region gene deletions reveals the extent of unmapped regions within the locus.
Genomics
4
:
505
10
Bottaro, A., M. DeMarchi, G. G. DeLange, C. Boccazzi, F. Caldesi, R. Gallina, A. O. Carbonara.
1989
. New types of multiple and single gene deletions in the IGHC locus.
Immunogenetics
29
:
44
11
Bottaro, A., M. DeMarchi, G. G. DeLange, A. O. Carbonara.
1989
. Gene deletions within the human immunoglobulin heavy chain constant region gene cluster.
Exp. Clin. Immunogenet.
6
:
55
12
Bottaro, A., U. Cariota, G. G. DeLange, M. DeMarchi, R. Gallina, A. Vlug, A. O. Carbonara.
1990
. Multiple levels of analysis of an IgG4 gene deletion.
Hum. Genet.
86
:
191
13
Bottaro, A., U. Cariota, M. DeMarchi, A. O. Carbonara.
1991
. Pulsed-field electrophoresis screening for the immunoglobulin heavy chain constant region (IGHC) multigene deletions and duplications.
Am. J. Hum. Genet.
48
:
745
14
Hendriks, R. W., M. van Tol, G. G. DeLange, R. K. Shuurman.
1989
. Inheritance of a large deletion within the human immunoglobulin heavy chain constant region gene complex and immunological implications.
Scand. J. Immunol.
29
:
535
15
Smith, C. I. E., L. Hammarström, J. I. Henter, G. G. DeLange.
1989
. Molecular and serological analysis of IgG1 deficiency caused by new forms of the constant region of the Ig H chain gene deletions.
J. Immunol.
142
:
4514
16
Walter, M. A., C. A. Chambers, B. Zimmerman, D. W. Cox.
1990
. A multigene deletion in the immunoglobulin heavy chain region in a highly atopic individual.
Hum. Genet.
85
:
643
17
Olsson, P. G., M. H. Hofker, M. A. Walter, S. Smith, L. Hammarström, C. I. E. Smith, D. W. Cox.
1991
. Ig heavy chain variable and constant region genes in common variable immunodeficiency.
J. Immunol.
147
:
2540
18
Olsson, P. G., H. Rabbani, L. Hammarström, C. I. E. Smith.
1993
. Novel human immunoglobulin heavy chain constant region gene deletion haplotypes characterised by pulsed-field electrophoresis.
Clin. Exp. Immunol.
94
:
84
19
Brusco, A., U. Cariota, A. Bottaro, C. Boccazzi, A. Plebani, A. G. Ugazio, R. Galanello, A. M. van Leeuwen, G. G. DeLange, S. Depelchin, et al
1994
. Structural and immunologic analysis of gene triplications in the Ig heavy chain constant region locus.
J. Immunol.
152
:
129
20
Brusco, A., U. Cariota, A. Bottaro, C. Boccazzi, A. Plebani, A. G. Ugazio, R. Galanello, M. G. Guerra, A. O. Carbonara.
1995
. Variability of the immunoglobulin heavy chain constant region locus: a population study.
Hum. Genet.
95
:
319
21
Rabbani, H., Q. Pan, N. Kondo, C. I. E. Smith, L. Hammarström.
1996
. Duplications and deletions of the human IGHC locus: evolutionary implications.
Immunogenetics
45
:
136
22
Brusco, A., F. Cinque, S. Saviozzi, C. Boccazzi, M. DeMarchi, A. O. Carbonara.
1997
. The G4 gene is duplicated in 44% of the human immunoglobulin heavy chain constant region haplotypes.
Hum. Genet.
100
:
84
23
Keyeux, G., G. Lefranc, M. P. Lefranc.
1989
. A multigene deletion in the human IGH constant region locus involves highly homologous hot spots of recombination.
Genomics
5
:
431
24
Wiebe, V., A. Helal, M. P. Lefranc, G. Lefranc.
1994
. Molecular analysis of the T17 immunoglobulin CH multigene deletion (del A1-GP-G2–G4-E).
Hum. Genet.
93
:
520
25
Chen, Z.-Q., M. H. Hofker, D. W. Cox.
1995
. Defining the breakpoint of a multigene deletion in the immunoglobulin heavy chain gene cluster.
Immunogenetics
41
:
69
26
Bottaro, A., R. Gallina, A. Brusco, U. Cariota, C. Boccazzi, M. R. Barilaro, A. Plebani, A. G. Ugazio, A. M. van Leeuwen, G. G. DeLange, et al
1993
. Familial clustering of IGHC deletions and duplications: functional and molecular analysis.
Immunogenetics
37
:
356
27
Sambrook, J., E. F. Fritsch, T. Maniatis.
1989
.
Molecular Cloning: A Laboratory Manual
2nd Ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.
28
Smith, G., S. Kunes, D. Schultz, A. Taylor, K. Triman.
1981
. Structure of χ hot-spot of recombination in the murine major histocompatibility complex.
Science
234
:
173
29
Jeffrey, A., V. Wilson, S. Lay Thein.
1985
. Hypervariable “minisatellite” regions in human DNA.
Nature
314
:
67
30
Wahls, W., L. Wallace, P. Moore.
1990
. Hypervariable minisatellite DNA is a hot-spot for homologous recombination in human cells.
Cell
80
:
95
31
Aoki, K., K. Suzuki, T. Sugano, T. Tasaka, K. Nakahara, O. Kuge, A. Omori, M. Kasai.
1995
. A novel gene, Translin, encodes a recombination hot-spot binding protein associated with chromosomal translocations.
Nat. Genet.
10
:
167
32
Schuchert, P., M. Langsford, E. Kaslin, J. Kohli.
1991
. A specific DNA sequence is required for high frequency of recombination in the ade6 gene of fission yeast.
EMBO J.
10
:
2157
33
Kobori, J., E. Strauss, K. Minard, L. Hood.
1986
. Molecular analysis of the hot-spot of recombination in the murine major histocompatibility complex.
Science
234
:
173
34
Steinmetz, M., D. Stephan, K. Fischer-Lindahl.
1986
. Gene organisation and recombinational hot-spot in the murine major histocompatibility complex.
Cell
50
:
719
35
Uematsu, Y., H. Kiefer, R. Schulze, K. Fischer-Lindahl, M. Steinmetz.
1986
. Molecular characterisation of a meiotic recombinational hot-spot enhancing homologous unequal crossing-over.
EMBO J.
5
:
2123
36
Shiroishi, T., N. Hanzawa, T. Sagai, M. Ishiura, T. Gojobori, M. Steinmetz, K. Moriwaki.
1990
. Recombination hot-spot specific to female meiosis in the mouse major histocompatibility complex.
Immunogenetics
31
:
79
37
Zimmerer, E., H. Passmore.
1991
. Structural and genetic properties of the Eb recombinational hot-spot in the mouse.
Immunogenetics
33
:
132
38
Keyeux, G., J. E. Bernal.
1996
. New allele variants of the immunoglobulin switch (Sα) regions.
Hum. Genet.
97
:
695
39
Nakamura, Y., M. Carlson, K. Krapcho, M. Kanamori, R. White.
1988
. New approach for isolation of VNTR markers.
Am. J. Hum. Genet.
43
:
854
40
Lupski, J. R..
1998
. Genomic disorders: structural features of the genome can lead to DNA rearrangements and human disease traits.
Trends Genet.
14
:
417
41
Natvig, J. B., H. G. Kunkel.
1974
. A hybrid IgG4-IgG2 Ig.
J. Immunol.
112
:
1277
42
Jurka, J..
1990
. Novel families of interspersed repetitive elements from the human genome.
Nucleic Acids Res.
18
:
137
43
Lichten, M., A. S. Goldman.
1995
. Meiotic recombination hot-spots.
Annu. Rev. Genet.
29
:
423
44
Steinmetz, M., Y. Uematsu, K. Fischer-Lindahl.
1987
. Hot-spots of homologous recombination in mammalian genomes.
Trends Genet.
3
:
7
45
Rappold, G. A..
1993
. The pseudoautosomal regions of the human sex chromosomes.
Hum. Genet.
92
:
315
46
Lopes, J., N. Ravisé, A. Vandenberghe, F. Palau, V. Ionasescu, M. Mayer, N. Lévy, N. Wood, N. Tachi, P. Bouche, et al
1998
. Fine mapping of de novo CMT1A and HNPP rearrangements within CMT1A-REPs evidences two distinct sex-dependent mechanisms and candidate sequences involved in recombination.
Hum. Mol. Genet.
7
:
141
47
Bech-Hansen, N. T., P. S. Linsley, D. W. Cox.
1983
. Restriction fragment length polymorphisms associated with immunoglobulin Cγ genes reveal linkage disequilibrium and genomic organization.
Proc. Natl. Acad. Sci. USA
80
:
6952
48
Migone, N., G. G. DeLange, A. Piazza, L. L. Cavalli-Sforza.
1985
. Genetic analysis of eight linked polymorphisms within the human immunoglobulin heavy chain region.
Am. J. Hum. Genet.
37
:
1146
49
Bottaro, A., R. Gallina, M. DeMarchi, A. O. Carbonara.
1989
. Genetic analysis of new restriction fragment length polymorphisms (RFLP) in the human IgH constant gene locus.
Eur. J. Immunol.
19
:
2151
50
Baar, J., M. J. Shulman.
1995
. The Ig heavy chain switch region is a hot-spot for insertion of transfected DNA.
J. Immunol.
155
:
1911
51
Smith, G. R..
1983
. χ hot-spots of generalised recombination.
Cell
34
:
709