Isotype switching of an immunoglobulin heavy chain transgene occurs by DNA recombination between different chromosomes

Cell, Vol. 63, 537-546,

November

2, 1990, Copyright

0 1990 by Cell Press

lsotype Switching of an lmmunoglobulin Heavy Chain Transgene Occurs by DNA Recombination between Different Chromosomes Rachel M. Getstein: Wayne N. Frankeel,* Chih-Lin Hsieh,§ Jeannine M. Durdik,tll Satyajit Rath,t# John M. Coffin,* Alfred Nisonoff,t and Erik Selsing’t * Department of Pathology *Department of Molecular Biology Tufts University School of Medicine Boston, Massachusetts 02111 t Rosenstiel Basic Medical Sciences Research and Department of Biofogy Brandeis University Waltham, Massachusetts 02254 5 Department of Pathology Stanford University School of Medicine Stanford, California 94305

Center

Summary Pansgenic mice carrying an immunoglobulin P heavy chain transgene exhibit irotype switching of the transgene. We have now characterized the mechanism of transgene switching in these mice. The site of ~1transgene insertion in one tmnsgenic line has been localized to chromosome 5 using a series of polymorphic endogenous retmviruses as genetic markers in backcross mice. The endogenous immunoglobulin heavy chain locus resides on mouse chromosome 12, which shows that transgene lsotype switching can occur between two different chromosomee even though normal antibody gene switching has genemlly been thought to occur within one chromosome. We find that transgene isotype switching involves interchromosomal DNA recombination, and our date suggest that the same enzymatic mechanisms mediate both normal isotype switch recombination and intefchmmosomal transgene switching. Our findings also support the notion that the isotype switching mechanism can induce chromosomal translocations such as observed for the c-myc gene in some B cell tumors. Introduction After exposure to antigen, IgM-bearing B cells are capable of differentiating to plasma cells, which secrete antibody of the IgG, IgA, or IgE isotype. This process has been designated as isotype switching. In IgM-producing cells, the VDJ gene segment (comprising recombined heavy chain variable [VH], diversity [D], and joining [JH] gene segments) and the closely linked u constant region (Cp) gene segments constitute the transcription unit for heavy chain antibody production. lsotype switching links the VDJ region to one of the several heavy chain constant

region (CH) genes located downstream of QL. Evidence exists for two distinct mechanisms of isotype switching. In the first, isotype switch recombinations are mediated by DNA rearrangements within the CH gene cluster that bring the VDJ into proximity of a new CH region, deleting the Cu and other intervening CH genes (Honjo and Kataoka, 1978). Evidence has shown that many switch recombinations occur by a “looping-out and deletion” mechanism acting within a single chromosome (e.g., Schwedler et al., 1990; Matsuoka et al., 1990; lwasato et al., 1990). Switch (S) regions located 5’of each CH (with the exception of CS) (Rabbits et al., 1980; Cory and Adams, 1980) have been implicated in mediating class switch recombination (Sakano et al., 1980; Davis et al., 1980). Both sitespecific switch recombinase(s) (Davis et al., 1980) and homologous recombination (Kataoka et al., 1981; Nikaido et al., 1981) have been proposed to be involved in class switch DNA recombination. Switch recombinations have been described for plasmacytomas (Davis et al., 1980; Sakano et al., 1980; Eckhardt and Birshtein, 1985) hybridomas (Petrini et al., 1987; Hummel et al., 1987; Katzenberg and Birshtein, 1988), and B cell blasts (Radbruch et al., 1988; Winter et al., 1987). In contrast, it has been reported that the ability of B lymphocytes or B cell lymphomas to express two isotypes, such as IgM and IgE, on the ceil membrane does not require DNA rearrangement (Perlmutter and Gilbert, 1984; Yaoita et al., 1982; Chen et al., 1988). Alternative RNA splicing of a large primary nuclear transcript containing the VDJ region and a number of CH regions has been suggested to explain this result; in this model, RNA splice site choice would select the CH gene that would be linked to VDJ in the mRNA. We have previously described transgenic mice carrying a recombined VDJ Cj.r heavy chain gene (ARSu) that, with appropriate endogenous light chains, confers specificity for the hapten p-azophenylarsonate (ARS) (Durdik et al., 1989). Upon immunization, the product of the ARSu transgene dominates the anti-ARS antibody response of these mice. Unexpectedly, the ARSu transgene can undergo isotype switching; that is, many of the transgene-derived antibodies in the immunized mice are IgG rather than IgM even though the ARSu transgene contains only Cu and no Cy sequences. We have now investigated the mechanism involved in transgene isotype switching. In one transgenie line, we have localized the transgene to mouse chromosome 5, indicating that transgene isotype switching can occur between unrelated chromosomes. Additionally, we have isolated, from an IgG2a-producing transgenic hybridoma, a recombined transgene copy that displays an Su to SrPa join at the DNA level. Thus, transgene isotype switching can occur by an interchromosomal DNA recombination. Results

IPresent address: Allergy and Clinical immunology, University of Colorado Health Sciences Center, Denver, Colorado 60262. X Present address: lmmunobiology Section, Yale School of Medicine, New Haven, Connecticut 06510.

lsotype Switching Can Occur by an Interchromosomal Mechanism We have previously shown that the ARS~I transgene

can

undergo isotype switching after immunization of transgenie mice with ARS-KLH (Durdik et al., 1989). To ascertain whether transgene isotype switching can occur between separate chromosomes, we mated transgenic mice of the ARS5 line with BALB/cJ mice to produce Fl mice carrying both C57BU8J and BALBlcJ chromosomes. These Fl mice were analyzed to determine whether the transgene V region (which is necessarily encoded on a C57BU8J chromosome) could be found linked to a BALB/cJ Cy region within a single polypeptide chain. To distinguish Cyl gene products of BALB/cJ mice, we made use of an antibody recognizing an allotypic marker characteristic of the. BALB/cJ strain (Igh-4a; Oi and Herzenberg, 1979): As a marker of the transgene V region, we used the AD&reactive idiotypic determinant, which is present onall nonmutated (and many mutated) transgenederived h&av)i:chains in ARS5 mice but which has not been found among antiARS antibodies produced by normal C57BU8J or BALBlcJ mice immunized with ARS-KLH (Hornbeck and Lewis, 1983; Durdik et al., 1989). In previous analyses of our transgenic mice, the AD&reactive marker has proven to be a reliable indicator of transgene expression (Durdik et al., 1989; Rath et al., 1989). We analyzed antiARS.igGl antibodies of immunized Fl mice for allotype and AD8, reactivity. In a representative experiment, both transgenic and normal littermate Fl mice produced antiARS antibodies (545 uglml and 3700 uglml, respectively), but AD8 levels were much higher in the transgenic Fl mouse (1400 ug/ml) than in the nontransgenie Fl mouse (48 ug/ml). The transgenic and nontransgenie Fl mice expressed comparable levels of IgGl antiARS antibodies that bear the Igh-4a allotype characteristic of the BALB/cJ parent (120 uglml and 130 uglml, respectively). Depletion of AD&reactive antibodies led to the removal of nearly all antiARS antibody in the transgenic Fl serum (545 kg/ml antiARS before depletion; 20 uglml after depletion),, indicating that most of this antibody is AD8 reactive and therefore transgenic in origin. Depletion of ADB-reactive antibodies had essentially no effect on the content of lgh-4a antiARS in fhe nontransgenic Fl serum (130 uglml of lgh-4a anti-ARS before depletion; 140 j.rg/ml after depletion) but removed all detectable Igh4a antiARS antibodies from the transgene-bearing Fl mouse serum (120 uglml of lgh-4a anti-ARS before depletion; <5 uglml after depletion). These results indicate that, in the transgenic Fl animals, essentially all IgGl antiARS antibodies that have BALB/cJ constant regions also have V regions derived from the ARSu transgene. This indicates, therefore, that the transgene isotype switching mechanism can operate between C57BU8J and BALBlcJ chromosomes. ARSp mansgenes Are Located on Proximal Chromosome 5 in the ARS5 Line Because four ARE+ transgenic mouse lines all displayed isotype switching (Durdik et al., 1989), it seemed unlikely that all had transgene integration into mouse chromosome 12, which carries the endogenous immunoglobulin heavy chain locus. However, to prove whether interchromosomal isotype switching of the ARSu transgene

I

abcdefghijB’C I I ’ ’

I

I

I

I

I

I

I

3 30

5,48 1% 74

i 28 11,12

Transgene

+++-

++++

++

+-

Figure 1. Segregation of Mpmv Proviruses in Backcross Mice A Southern blot of Pvull-digested genomic DNA from C57BU6JARS5 (lane B’), BALB/cJ (lane C), and ten (C57BLWARS5 x BALB/cJ)Fl x BALB/cJ backcross mice (lanes a-j) was hybridized to an oligonucleotide probe specific for endogenous Mpmv proviruses, as previously described (Frankel et al., 1990). C57BU6J proviruses Mpmv-1, -2, -5, -6, -7, -8, -9, -11, and 12 were scored by their presence or absence, and BALB/cJ proviruses Mpmv-18, -19, -28, and -30 were scored as two versus one copy. A second restriction enzyme, EcoRI, was used to read the segregation pattern of Mpmv-5, -11, -12, and 18 as well as to confirm patterns of other proviruses (data not shown). In this figure, Mpmv-7 (indicated by an arrow) has one recombinant (lane c) with the transgene; transgene segregation was determined separately and is indicated below the autoradiogram. The transgene and Mpmv-7 marker were perfectly concordant in 26 additional backcross mice (data not shown). No significant linkage was observed in the 36 mice analyzed for any of the Mpmv markers indicated in the figure. The positions of DNA size standards 9.4, 8.6, 4.3, and 2.3 kb are shown by arrows at the left.

was occurring between homologous chromosomes or between entirely different chromosomes, we determined the chromosomal location and approximate map position of the ARSu transgenes in the ARS5 transgenic line. ARS5 mice were mated with normal BALB/cJ mice and the resultant (C57BU8J x BALB/cJ)Fl transgenic mice were backcrossed to BALB/cJ. Because in these mice, and all.others previously tested (Durdik et al., 1989), the u transgene segregated as a single gene, the transgene segregation pattern was compared with those of proviral cellular DNA junction fragments from the nonecotropic family of endogenous murine leukemia viruses. These genetically stable insertion elements are found interspersed throughout the genome and are extremely useful for mapping genes in crosses where the two parental strains differ in proviral content (Frankel et al., 1990). Three classes of endogenous nonecotropic proviruses can be distinguished on Southern blots by using oligonucleotide probes specific for polytropic, modified polytropic, and xenotropic sequences, respectively (Stoye and Coffin, 1988).

lsotype 539

Switching

of an lmmunoglobulin

Transgene

The strain C57BU6J contains 30 nonecotropic proviruses not present in BALBlcJ, each of which has recently been assigned to a chromosomal region (Frankel et al., 1969a, 1969b, 1990, unpublished data). To test for linkage of the transgene to one or more of these proviruses, backcross DNAs were Southern blotted and hybridized to virus class-specific probes. An example of one such blot is shown in Figure 1. We found that the transgene was closely linked to the modified polytropic provirus Mpmv-7. With only 1 recombinant in 36 gametes, the transgene is within 2.8 CM + 2.7 CM SE of Mpmv-7 (x2 = 41.6, P < 0.0001; Figure 1). To confirm this, 35 of these mice, including the one recombinant with Mpmv-7, were also scored for the polytropic provirus Pmv-5, which is known to be within about 1 CM of Mpmv-7. No recombinants were found between the transgene and Pmvd. The transgene was not significantly linked to nonecotropic proviruses on chromosomes 1, 2, 3, 4, 6, 7, 8, 10, 11, 12, 13, 14, 15, 16, or 19 (Figure 1 and data not shown). Thus, these data demonstrate that the u transgene in ARS5 mice is located near Mpmv-7 and Pmv-5 on proximal mouse chromosome 5. lsotype Switching of the ARSp Transgene Occurs at the DNA Level To determine whether interchromosomal isotype switching of the ARSu transgene was occurring at the DNA level, we first analyzed Sy switch region configurations in several IgG-secreting transgenic hybridomas. Figure 2a shows Southern blot analyses of six transgenic hybridomas that secrete ARS binding IgGl carrying the AD8reactive idiotype characteristic of the transgene. Two of these hybridomas (3G3 and 3D6) have also been found to express the transgene V region by mRNA sequence analysis (data not shown). Of the six IgGl-secreting hybridomas, five exhibit either one or two recombined Sri segments. Thus, in these transgenic hybridomas that express IgGl antibodies with transgene-encoded V regions, the endogenous Cyl genes have undergone DNA recombination. The presence of two recombined Sri segments in some of the transgenic hybridomas suggests that unexpressed endogenous JH alleles may also have switched to Cyl; this would be similar to findings in hybridomas obtained from normal mice where both expressed and unexpressed Igh chromosomes are frequently switched to the same CH gene (Hummel et al., 1987; Winter et al., 1987; Katzenberg and Birshstein, 1988). In Figure 2a, the 5D9 hybridoma does not show recombined Sri segments, thus we cannot draw definite conclusions about the context of the Cyl gene in this cell line. However, there are no germline Sri alleles remaining in 5D9 and it is therefore possible that DNA recombination has occurred between the Syl and Crl regions in this hybridoma. Southern blot analyses of the IgGl-producing transgenie hybridomas also show that variability of transgene copy number is found among these cell lines (data not shown). Hybridomas 3G3,3H12, and 6C12 appear to have about the same copy number as ARS5 kidney DNA (approximately 30-50 copies), whereas hybridomas 5D9, 3D6, and 2D3 have only about one-half to threequarters of this number. Another IgGl-producing transgenic hy-

bridoma (1ClO) has no full-length transgene copies. We do not know the basis for this variation in transgene copy number among transgenic hybridomas. However, this phenomenon may not always be related to isotype switching, because variation in transgene copy number is also observed in IgM-producing hybridomas obtained from nonimmunized ARS5 mice (Rath et al., 1989). Analyses of the Sy2a region in three transgenic lgG2aproducing hybridomas (all expressing the transgene V region as determined by mRNA sequence analyses) showed results entirely analogous to those obtained from the IgGl-producing hybrids (data not shown). In each case, one or two recombined Sy2a or Cy2a regions could be detected and the transgene copy number was variable. One IgG2a-expressing hybridoma, 284, exhibits no full-length copies of the transgene. Thus, for both isotypes, hybridomas that express the transgene V region associated with an endogenous C region display DNA recombinations of the appropriate endogenous C region gene. To determine whether the Sr DNA recombination events that had occurred in the transgenic hybridomas also involved the transgene VDJ segment, we analyzed the hybrids by Southern blots to determine whether DNA fragments that hybridized with both a transgene V region probe and a Sy probe could be detected. For many hybridomas, this was not possible using standard agarose gel electrophoresis, probably due to a number of reasons: the length of the intron DNA separating V and C regions in switched heavy chain genes, the large number of restriction sites within this intron DNA, and the large number of bands in hybridomas that hybridize with some Sy probes. However, in the IgG2a-producing 284 hybridoma we did detect a IO kb Bglll fragment that hybridized with both a transgene V region probe and a Sy2a probe (Figure 2b) as well as with JH and pUC13 probes (data not shown). Also, in an IgGl-producing hybridoma, lC10, comigration of bands hybridizing with pUC13 and Cyl probes was observed with both Kpnl (Figure 2c) and Bglll (data not shown). The sum of these results indicates that, in the 284 and 1ClO hybridomas, transgene-associated sequences (pUC13, VH, or JH) have recombined with an endogenous Cy gene at the DNA level. Taken together with assignment of the transgene to chromosome 5, it seems quite likely that these recombined genes reflect translocation events between chromosomes 5 and 12. lsotype Switching of the AR% Transgene Appears to Involve a Class Switch DNA Recombination Event To characterize the switched transgene in the 284 hybridoma, we screened a genomic library containing 9-11 kb fragments of Bglll-digested 284 DNA and isolated a genomic clone, designated 284.1, that represented the 10 kb Bglll fragment detected in Southern blot analyses. Figure 3a depicts the structure of this clone. We determined the nucleotide sequence of the VDJ region in the 284.1 clone to ascertain whether this gene was expressed in the 284 hybridoma. Comparison of the 284.1 sequence (Figure 3b) with the sequence of 284 polysomal y2a mRNA (data not shown) showed identity at the 416 nucleotide po-

Transgene

v

S, 2a

probe

probe

9.4

6.6

4.4

pUC PROBE Figure

2. Southern

Blot Analyses

Cyl PROBE of Transgenic

Hybridoma

DNAs

(a) The hybridomas analyzed secrete ARS-reactive IgGl antibodies that also react with AD8. In addition, 3G3 (Durdik et al., 1989) and 3D6 produce transgene-derived VDJ linked to Cy as determined by mRNA sequencing (data not shown). Genomic DNAs from ARS5 transgenic kidney, X63.Ag8.653 (Ag8), SP2/0@14 (SP2/0), or hybridoma cells (3G3 was derived from spleen cells of an immunized transgenic C57BU6J mice fused to X83.Ag8.653; all others were derived from spleen cells of an immunized transgenic [C57BUW x BALBlcJ]Fl mouse fused to SP2/0-Ag14) were digested with EcoRI, electrophoresed through a 0.8% agarose gel, transferred to nitrocellulose, and hybridized with radioactive probe derived from the 10 kb Hindlll-EcoRI fragment from PyIIEH1O.O (Mowatt and Dunnick, 1986). The migration positions for germline Syl hybridization bands for C57BU6J and BALB/cJ are indicated by arrows. Hybridization bands corresponding to recombined Syl sequences are indicated by triangles. The lower intensity of the recombined bands relative to germline bands is most likely due to deletion of a large proportion of the region to which the probe hybridizes during switch recombination. From other blots, the hybridization band in hybridoma 3D6 that migrates close to the germline C57EU6 position may actually represent a recombined ST1 region. (b) Genomic DNAs were digested with Bglll, electrophoresed in duplicate through a single 0.8% agarose gel, and transferred to nitrocellulose, and the two parts of the gel hybridized with separate probes. The transgene V region probe used for this experiment was prepared from a 133 nucleotide Avall-Rsal fragment from the phage clone 2B4.1. Similar results were obtained with the same 133 nucleotide fragment derived from VH36-65 (pV133) (obtained from Dr. L. J. Wysoki). The SyPa probe is derived from the 2.7 kb Hindlll fragment of pSy2a.3 (Eckhardt and Birshtein, 1985). The arrow indicates a 10 kb Bglll fragment that hybridizes with both V and SyPa probes. (c) Genomic DNAs were digested with Kpnl, electrophoresed in duplicate through a single 0.8% agarose gel, and transferred to nitocellulose, and the two parts of the gel hybridized with separate probes. The ARS5 kidney lane, hybridized with pUC18, is taken from an autoradiogram of the filter that was exposed for 1 hr; the other lanes are from the same filter exposed overnight. The CT1 probe is derived from the 6.8 kb EcoRl fragment (from Pyl/EHlO.O/E6.8 obtained from Wesley Dunnick) containing the BALBlc Cyl and contains 3.0 kb 5’ to the Cyl coding region. (d) Genomic DNA was digested with EcoRI, electrophoresed through a 0.8% agarose gel, transferred to nitrocellulose, and hybridized with a pUC18 probe. The arrow indicates a transgene band present both in ARS5 kidney DNA and in 284 DNA.

chromosome

3 kb deletion \Bss

5 E

transgene put vector

&I

E

transgens LVDJ

E

E

4

E

cp

E

E

E

CY2a

c A FY I kb

switch

chrOmOSOme 12

site

b

--- --- _-____--- --- --- __- --- --JH2

su

GAATAGAGACCTGCAGT'%'EGCCA+j+AC

2B4.1

GGTGAGGTGTGAGGAACCAG GCCCAACAAC T,,TGAAAhACTTAAGGAAAGTAGTGATGTG

IlII/lI

IIllIIIIII

TAACTCTCCAGCCACAGTAA

IIIIIIIIII

IIIIllIIIi

IllIIIIlII

GAACCACGCCCAACMCTATGAAAAACTTAKGAAAG TAGTGATGTG

Figure

3. Restriction

Map and Sequences

of the Switched

Transgene

Copy

in the 284

Hybridoma

(a) Organization of the switched transgene copy in 284. The microinjected ARSu transgene is depicted at the top, the endogenous CTPa gene is depicted on the bottom, and the recombined transgene copy isolated from the 284 cell line (284.1) is shown in the middle. Black and cross-hatched segments correspond to sequences derived from the microinjected AR@ gene, whereas open segments are from the endogenous C57BU6J Cy2a region. The heavily outlined 264 CyPa switch region is present within the 284.1 clone, whereas other CTPa sequences are not included in this clone. The restriction map of the 10 kb 284 Bglll fragment was derived from the genomic clone. The remainder of the map was determined by Southern blot analyses with various hybridization probes. A triangle and dashed lines indicate the 284 switch recombination site joining the transgene Sp to endogenous STPa. The repetitive switch region sequences within SW and SyPa are indicated by dots. The 5’ Bglll site in 284.1 must be derived from mouse genomic DNA at the transgene integration site because there is no Bglll site in the pUC13 vector used to construct the transgene. Analy sis of clone 284.1 showed that the pUCl3 sequences present in the cloned Bglll fragment contain -200 bp of mouse DNA sequences interspersed with plasmid sequences near the Bglll site. Presumably this is due to multiple recombination events during the chromosomal integration of microinjetted DNA, as has been reported previously (Covarrubias et al., 1966). We have not precisely characterized the structure of this region in the 284 transgene. (b) L-VDJ DNA sequence of genomic clone 284.1. Identity with the R16.7 transgene is indicated by a dash. The leader and intron sequences of R16.7 are unpublished data obtained from the laboratory of Dr. Allan Maxam. The sequence of the C57BU6J VH most homologous to the transgene is also shown; identity to R16.7 is indicated by a dash. (c) Sequence of the 284.1 switch recombination site. DNA sequence from clone 284.1 is compared with Su sequence of the injected transgene as well as to BALB/cJ SyPa sequence. Sequences with at least 5/7 homology to the consensus sequence YAGGTTG (Y = pyrimidine) (Marcu et al., 1962) are noted by boxes, while the sequences TGGG and TGAG (Nikaido et al., 1962) are indicated by dots above the sequence. We do not know the basis for the three nucleotide differences observed between the transgene Su and that of 284.1. It is possible that these differences represent mutations induced during the switch recombination process as noted by Dunnick et al. (1969). The Su sequence determined from the transgene was identical to the published BALB/cJ sequence (Sakano et al., 1960). The SyPa sequence was determined from Ml3 subclone PC3a (generously provided by Dr. Laurel Eckhardt, derived from the Sy2a.3 clone; Eckhardt and Birshtein, 1965). Identity is indicated by vertical lines.

Cdl 542

sitions that could be unambiguously determined in the mRNA sequence. This establishes that the 284.1 clone represents the expressed heavy chain gene in the IgG2aproducing 2B4 cell line. The sequence of the VDJ region in 284.1 also demonstrates that this gene is derived from the ARSu transgene according to the same criteria that we have previously used to determine whether hybridoma mRNAs were transgene derived (Durdik et al., 1989). First, the 284.1 VDJ region is highly homologous to the R18.7 VDJ region present in the microinjected transgene. Second, the V-D joins in the 2B4.1 and R18.7 VDJ regions are identical, including !he TCGA N region insertion. Third, at the four positions where the R18.7 VH sequence differs from the A/J germline VH38-85 sequence (Durdik et al., 1989) 284.1 and R18.7 are identical. Fourth, at the single position where the R18.7 JH2 sequence differs from the C57BL/8J germline JH2 sequence (second position of codon 107) 284.1 and R18-7 are again identical. In addition, only one VH gene in C57BU8J mice hybridizes weakly with aVH38-85specific probe under highly stringent hybridization conditions; we have isolated and sequenced this C57BL18J VH gene. Comparison of sequences shows that 34 positions differ between the microinjected R18.7 VH region and the isolated endogenous C57BL/8J VH gene. The 284.1 gene is identical to the microinjected transgene rather than the endogenous C57BLl8J VH gene at 33 of these 34 nucleotides. Finally, the presence of pUC13 plasmid sequences located upstream of the 284.1 VDJ region also indicates the transgene origin of the switched gene in the 2B4 hybridoma. The 17 nucleotide differences observed between the R18.7 and 284.1 VDJ sequences indicate that the expressed transgene copy in 284 has undergone somatic mutation during antigenic stimulation in the mouse; we have previously found somatic mutation of transgene VDJ sequences in other IgG-producing transgenic hybridomas (Durdik et al., 1989). The 284 hybridoma differs from most transgene-expressing hybridomas in that 284 secretes anti-ARS antibodies that are CRlA+ but that do not react with AD8 (data not shown). The lack of AD8 reactivity for 284 may be due to the numerous somatic mutations present in the transgene expressed in this cell line. Five hybridomas expressing VH38-85 but not recognized by AD8 (Cannon and Woodland, 1983; Manser and Gefter, 1988) have amino acid replacements at either amino acid 58 or 59 or both (Sharon et al., 1989; Wysoki et al., 1988). In 284, both amino acids 58 and 59 differ from VH38-85. To characterize the recombination between the transgene and the endogenous C57BLl8J Sy2a region in the 284 hybridoma, we sequenced the relevant region of the 284.1 clone (Figure 3~). Comparison of the 284.1 sequence with corresponding portions of BALBlcJ Su and SyPa sequences shows that the 2B4 recombination directly juxtaposes SW and Sy2a regions at a single recombinational breakpoint with no apparent loss or gain of any nucleotides. The exact recombination point can only be localized to one of two possible sites due to the presence of a G nucleotide in both SF and SyPa at the breakpoint. Restriction mapping of ten restriction sites in

the DNA between the JH region and the S@y2a breakpoint in the 284.1 clone showed complete agreement with the map of the ARE+ transgene. Thus, the S&SySa joining appears to be the only recombinational event that has occurred within the transgene switch region in the 284 cell line. The presence of only a single transgene copy in the 2B4 hybridoma suggests that the switched 284 transgene represents one of the terminal transgene copies within the ARS5 transgene array. The 5’-terminal copy of the transgene array should exhibit a restriction map of sequences upstream of the VDJ exon that is distinct from the bulk of transgene copies. Thus, if the 284 transgene is a terminal copy, then its restriction map upstream of VDJ should be distinct from other transgene copies and should not be altered by the switch recombination event in 284. As shown in Figure 2c, this prediction was found to be true. When ARSB and 284 DNAs were digested with EcoRl and hybridized with a pUCl8 probe (thus monitoring sequences upstream of the VDJ exon), the 284 transgene corresponded to a single distinct transgene copy seen in ARS5 kidney DNA (presumably one of the terminal copies in the transgene array). Similar results were obtained with HindIll or BamHl digestion (data not shown). These findings indicate that, during the isotype switching process in 284, no recombination events have occurred upstream of the transgene VDJ region for a distance of at least 8 kb. Tandemly repeated sequences are found in both SKI and SyPa regions, and some, but not all, normal class switch DNA recombination events occur within these repetitive sequences. The Su breakpoint in the 284.1 clone is located 5’ of the Su repetitive region, whereas the Sy2a breakpoint is located 3’ of the SyPa repetitive elements. Both of these breakpoints are within 500 bp of previously characterized switch recombination sites (Katzenberg and Birshstein, 1988; Eckhardt and Birshtein, 1985), whereas known switch sites span a range of approximately 10 kb within Su and a range of approximately 8 kb within Sy2a. Several groups have found conserved sequences located near many, but not all, switch site breakpoints. Marcu et al. (1982) noted elements having homology to the consensus sequence, YAGGTTG, upstream of a number of switch recombination breakpoints. In 284.1, two elements similar to this consensus are found upstream of the Su breakpoint, whereas three elements are found upstream of the Sy2a breakpoint. Although the YAGGTTG consensus has not previously been implicated in switch recombination involving Sy2a, it is clear from our sequences, and also those of Nikaido et al. (1982) that SrPa contains sequences related to YAGGTTG. In addi+ tion, Nikaido et al. (1982) have found that the sequence TGGG or TGAG is located close to many switch breakpoints; we find both these sequences located near the 284.1 switch site. Thus, by the criteria of breakpoint location and surrounding sequences, the 284.1 recombination is indistinguishable from any other analyzed class switch DNA recombinations and therefore appears likely to have been mediated by the same enzymatic machinery that is involved in normal DNA class switching.

lsotype

Switching

of an lmmunoglobulin

Transgene

543

d CHROMOSOME

12

CHROMOSOME

t

f

IgH

TRANSGENE

5

1

TRANSLOCATION AND REPLICATION

Figure 4. Biotinylated pARSu Probe Hybridized to Metaphase Spreads Shown are metaphase spreads from (a) transgenic ARS5 blood lymphocytes, (b) lE9 hybridoma ization were detected with avidin-FITC, and chromosomes were counterstained with propidium chromosomes. The diagram (d) represents the predicted types of chromosomes and location of cations between chromosome 5 and chromosome 12 are induced by transgene DNA switch

Chromosomai ll’ansiocations Are Associated with lansgene isotype Switching We were interested in determining the chromosomal location of the AFtQ transgene in IgG-producing hybridomas where we had not been able to detect a transgene recombination by Southern blot analyses. To do this, we performed in situ hybridization; to facilitate detection of the transgene in these experiments, we chose two hybridomas (lE9 and 1GlO) that retained multiple copies of the transgene. The lE9 hybrid has about the same number of transgene copies as ARSB kidney (30-50 copies), whereas 1GlO has about half this copy number (data not

cells, and (c) 1GlO hybridoma cells. Sites of hybridiodide. The arrows denote hybridization signals on transgenic copies, assuming that transgene translorecombinations.

shown). Sequence analyses have shown that both lE9 and lG10 express the transgene VDJ linked to Cy2a (Durdik et al., 1989). Figure 4a shows control in situ hybridization results from ARS5 normal blood lymphocytes. By genetic mapping, we have found that the AR35 transgene is localized proximal to the centromere of chromosome 5. Metaphase spreads from 15 ARSS blood lymphocytes show a single hybridization site that is consistent with the localization of the transgene array toward the centromeric portion of chromosome 5. Figure 4b shows representative in situ hybridization

Cell 544

results from the IgG2a-expressing lE9 hybridoma. Metaphase spreads from 25 hybridoma cells were analyzed. Two sites of transgene hybridization are observed; one of these is at the middle of a large chromosome, whereas the other is at the telomere of a small chromosome. Thus, a portion of the transgene array has undergone a clear chromosomal translocation in lE9. The relative sizes of the hybridizing chromosomes and the location of transgene hybridization on each chromosome are exactly what would be predicted based on reciprocal translocations between chromosome 5 and chromosome 12 (see Figure 4d). Based on the known orientation of the /g/t locus on chromosome 12 (VH telomeric and CH centromeric; Erikson et al., 1985, 1988; Wirschubsky et al., 1985), if the transgene translocations in lE9 do involve chromosome 12, then the large hybridizing chromosome in Figure 4b would presumably contain the switched transgene copy expressed as IgGPa. Figure 4c shows representative in situ hybridization results from the IgG2a-expressing lGl0 hybridoma. All 25 metaphase spreads analyzed from this hybrid display a single hybridization site present at the middle of a large chromosome. This site appears to be analogous to the large hybridizing chromosome in lE9. The lack of hybridization to a “reciprocal” small chromosome in 1GlO could be accounted by a number of explanations; the simplest would be that this small chromosome was lost during the hybridoma fusion process. Thus, both lE9 and 1GlO display transgene chromosomal translocations. The apparent loss of the small reciprocal chromosome in lGl0 could be related to the loss of transgene copies in this cell line. However, we have also found the loss of transgene copies in IgM-producing hybridomas where the transgene has not undergone an isotype switch (Rath et al., 1989). It seems possible that some transgenes could be lost by recombination between copies within the transgene array on chromosome 5; such recombinations might reflect the same mechanisms involved in deleting portions of the Sk region in nontransgenie B cells that are undergoing switch recombination (e.g., Ott et al., 1987). Clearly, further study is required to elucidate the mechanisms of transgene loss in hybridomas. Discussion We have found that isotype switching of an immunoglobulin w chain transgene occurs by an interchromosomal DNA recombination mechanism. In situ hybridization studies suggest that chromosomal translocations are associated with transgene isotype switching. Sequence analyses suggest that the enzymes that mediate normal antibody gene class switching DNA recombination are also involved in interchromosomal isotype switching of a heavy chain transgene. Thus, our results suggest that DNA switch recombinations are not limited to occur within a single chromosome. In addition, our study demonstrates the power of using endogenous MLV proviruses for efficiently mapping transgenes, or any other heritable murine trait.

We have cloned and sequenced the switched transgene copy from one hybridoma, 284. However, Southern blot analyses of a second hybridoma, lC10, also show that transgene DNA becomes linked to Sy sequences during transgene isotype switching. Furthermore, analyses of five additional hybridomas show that Sy DNA recombinations are associated with transgene isotype switching. Finally, in situ hybridization studies of two hybridomas show that chromosomal translocations of the transgene are associated with transgene isotype switching. Thus, the 284 hybridoma does not appear to be a special case in showing a transgene DNA switch recombination, but seems to be representative of the randomly chosen transgeneexpressing IgG hybridomas that we have analyzed. Although our analyses of the recombined transgene in the 284 hybridoma suggest a simple interchromosomal DNA recombination event, we cannot rule out the possibility that two or more sequential DNA recombinations might have occurred to produce the recombined 284 transgene. The presence of pUC13 vector sequences linked upstream of the transgene VDJ region in 284 and 1ClO argues against the involvement of homologous recombination (or gene conversion) between the transgene and endogenous heavy chain genes during interchromosomal switching. It is possible, however, that the P transgene first underwent a switch recombination to endogenous Su sequences and then subsequently switched to Sy2a, because previous reports have shown that sequential switches of this type can eliminate all evidence of the first switch in the final recombined gene (Eckhardt and Birshtein, 1985). Nevertheless, although it would seem that the transgene might be capable of undergoing a class switch recombination to an endogenous C57BU8J Cb gene and give rise to polypeptide chains that express both the AD8reactive idiotype and the C57BU8J pb allotype, we have not yet detected any clear examples of such molecules from either 119 IgM-producing hybridomas produced from nonimmunized transgenic mice or 21 anti-ARS, IgM-producing hybridomas produced from transgenic mice immunized with ARS-KLH. Taken together, these results suggest that the single recombination site detected in the 264 recombined transgene represents a single interchromosomal DNA recombination event. In addition, the apparent lack of switching between the transgene Su region and the endogenous SP region might suggest that DNA sequence homology (which is much greater between two SP regions than between Sk and Sy regions) may not play a major role in class switch DNA recombination. The Sp and SyPa DNA sequences surrounding the 284 transgene recombination site are indistinguishable from sequences surrounding normal antibody gene switch sites. Because normal switch sites range over 5-10 kb of DNA sequence, the location of the 284 recombination is unremarkable even though the 284 site is not exactly identical to any previously determined switch site. Thus, considering the information that is available concerning the sequence recognition requirements for class switching enzymes, it seems likely that these enzymes are involved in isotype switching of the ARS transgene. The interchromosomal DNA switch recombinations that

lsotype 545

Switching

of an lmmunoglobulin

Transgene

are apparently involved in isotype switching of the ARS transgene appear to have some analogy to the c-myc chromosomal translocations that are found in some 6 cell tumors. It has been suggested that “aberrant” switch recombination might be involved in those c-myc translocations that exhibit chromosomal breakpoints near heavy chain switch regions (Piccoli et al., 1984; Taub et al., 1982; Stanton et al., 1984; Dunnick et al., 1983). The interchromosomal isotype switching observed in our transgenie mice suggests that the switch recombination machinery can act between different chromosomes and provides support for the notion that this mechanism might play a role in c-myc oncogene activation. Several other groups have reported results that indicate isotype switching of a p transgene. Weaver et al. (1988) found a single hybridoma derived from a nonimmunized transgenic mouse that appeared to express a somatically mutated, isotype-switched transgene copy. Shimizu et al. (1989), in a separate transgenic mouse system, found that a human TVtransgene could apparently undergo isotype switching with endogenous mouse Cy genes in vitro. It would therefore appear that several distinct transgenic mouse lines can undergo interchromosomal isotype switching. Because the molecular mechanism of transgenie isotype switching has not been defined for other transgenic mice, it seems possible that all use the DNA recombination mechanism that we have found. It has been suggested that 6 cells and plasma cells might use different mechanisms for isotype switching. We have found that transgenic hybridomas, which might be considered representative of Ig-secreting plasma cells, exhibit isotype switches that have involved DNA recombination; however, we do not know at what stage of 6 cell maturation these switches occurred. It will be interesting to determine specifically whether transgene switching occurs in 6 cells and, if so, what switching mechanism is involved. Two questions that are not yet answered concern the frequency of transgene switching in our transgenic mice and the frequency of interchromosomal switching in nontransgenie mice. In immunized ARS5 transgenic mice, 50%90% of the anti-At% response is IgG and is derived from the transgene (Durdik et al., 1989). Because the transgene is located on chromosome 5 in AFiS5 mice, these anti-AM IgG molecules must be derived by interchromosomal isotype switching. However, we do not know the percentage of transgene switching that involves DNA switch recombination or the percentage of those B cells that respond to ARS that eventually undergo an isotype switch. This can best be addressed by analysis of transgene switching in vitro. lnterchromosomal isotype switching has been reported in rabbits and in mouse hybridomas. The molecular mechanisms for these isotype switches have not yet been determined. The transgene switch recombination events that we have found certainly provide a possible molecular basis to explain the results obtained from normal animals. In nonimmunized rabbits, about 3%-5% of serum IgA appears to come from interchromosomal switching in vivo (Knight et al., 1974). Analysis of a mouse hybridoma cell

line passaged in tissue culture indicated only a very low frequency of interchromosomal switching in vitro (Kipps and Herzenberg, 1988). It will be interesting to determine the frequency of in vivo interchromosomal switching in immunized mice. Additionally, the ability of an immunoglobulin heavy chain transgene to undergo isotype switch recombination should prove useful for defining the precise molecular requirements for isotype switching using constructs with switch regions of varying size and sequence. The fact that the transgene isotype switch occurs during an immune response to antigen suggests that this system could also provide information about sequences important for the regulation of isotype switch recombination in vivo. Experimental

Procedures

Animals C57BU6J and BALB/cJ mice were originally obtained from the Jackson Laboratory (Bar Harbor, ME). At the start of immunization mice were 6-12 weeks old. The construction of pARSu and transgenic mouse lines have been described previously (Durdik et al., 1969). All transgenic mice used in this study were derived.from ARSS.

Southern

Blot and Genetlc

Analysis

of Endogenous

Provlruses

The segregation of proviral fragments in (C57BL/6J-ARS5 x BALB/cJ) x BALBlcJ backcross mice was analyzed on Southern blots as previously described, with only minor modifications (Stoye and Coffin, 1966; Frankel et al., 1990). Briefly, 10 ug of genomic DNA was digested with restriction enzyme and electrophoresed in horizontal 0.6% agarose Tris-borate-EDTA gels. Gels were transferred overnight to Zetabind membranes (Cuno Life Sciences) under alkali conditions, and blots were hybridized to =P 5’ end-labeled oligonucleotide probes specific for Mpmv, Pmv, or Xmv provirus, washed, and exposed to X-ray film at -70°C for 2-3 weeks using an intensifying screen. All endogenous Mpmv, Pmv, and Xmv proviruses of C57BV6J and BALE/c-J were readily identifiable as Pvull or EcoRl proviral cellular DNA 3’junc tion fragments, and almost ail proviruses have now been chromosomaliy mapped (Frankel et al., 1969a, 1969b, 1990; Frankel and Coffin, unpublished data). Thirty proviruses were present in C57BU6.l but not BALB/cJ mice, and these were scored by presence (C57BU6J allele) or absence (BALB/cJ allele) in the backcross. While these markers were present on 14 different chromosomes, only Mpmv-7 and Pmv-5 were significantly linked to the transgene. An additional 12 proviruses were present in BALB/cJ but not C57BU6J; these were scored by band intensity (mice with two copies inherited the BALBlcJ allele; those with one copy inherited the C57BU6J allele), but none were associated with the transgene.

Southern

Analyses

of Transgenlc

Mice and tlybrldomas

DNA was analyzed by Southern (1975) blotting according to standard procedures. Approximately 30 ftg of DNA digested to completion with a restriction enzyme (New England Biolabs) was eiectrophoresed through 0.6% agarose (FMC) Tris-acetate-EDTA gels and transferred to nitroceliuiose paper (Sartorius). Filters were hybridized to 32P-labeied DNA probes at 65OC. For the genetic analysis of (C57BU6JARS5 x BALB/cJ) x BALBlcJ backcross mice, genomic DNA from these mice was typed for the presence or absence of the transgene by Southern blot analysis as above. Filters were hybridized with 32P-labeled JH probe or 32P-labeled Cu probe, either of which results in transgene-specific patterns.

Genomlc Cloning Clone 284.1 Genomic DNA from hybridoma 284 was digested to completion with Bglll. DNA fragments of 9-11 kb were isolated and ligated with phage EMBW (BamHI vector kit obtained from Stratagene, La Jolla, CA). Approximately 250,mO phage were screened by hybridization with =P-labeled JH probe on nitrocellulose filters (Schleicher & Schuell Inc.,

Keene, NH). One signal was obtained and purified to homogeneity by standard plaque purification methods. 86 VH A single weakly hybridizing band is observed in Southern blots of EcoRCdigested C576U6J kidney DNA hybridized with =P-labeled pV133 and washed at high stringency (Siekevitz et al., 1962; unpublished data). This band migrates at 5.6 kb, whereas the strongly hybridizing band in A/J kidney DNA migrates at 6.2 kb. To isolate the gene responsible for this hybridization, genomic DNA from C57BU6J kidney was digested to completion with EcoRI. DNA fragments of 5.5-6.2 kb were isolated and ligated with EcoRl-digested phage EMBL12 (Natt and Scherer, 1986). Approximately 120,000 phage were screened using pV133. Six strong signals were found and two were plaque purified to homogeneity. These two clones both contained the 5.8 kb EcoRl fragment expected from Southern blot experiments and had identical restriction maps within the 5.8 kb fragment. One of these clones was sequenced and is similar to, but distinct from, an analogous VH gene present in BALB/cJ mice (Wysoki and Gefter, 1969).

DNA and RNA Sequencing DNA sequences were determined by the dideoxynucleotide chain termination methods (Sanger et al., 1977) with pUC sequencing vectors or phage Ml3 (Messing et al., 1981; Vieiraand Messing, 1962). In some cases internal primers were used: for VH of 284.1 those utilized in Durdik et al. (1989) as well as d(ATGAATGCAATTAT), which corresponds to a sequence just 3’of the first Xba site, which is located 5’of Sp, and d(CAAGGCACCACTCTCACAGTCTCCTCA), which anneals within JH2. RNA sequence analysis was as in Durdik et al. (1989).

Hybridoma

Lines

IgG2a-producing hybridomas were obtained by fusion of transgenic spleen cells with the X63.Ag8.653 myeloma cell line 3 days following an i.p. boost of 100 kg of KLH-ARS in saline. Two ARS5 mice were used for two separate fusions; one had been immunized three times with 100 wg of KLHARS in complete Freund’s adjuvant at 3 months, 2 months, and 1 week prior to the i.p. boost, whereas the other was immunized at 5 months and 4 months prior to the i.p. boost. Hybridoma lGl0 was isolated from the former animal, whereas hybridomas lE9 and 284 were isolated from the latter. IgGl-producing hybridomas were obtained by fusion of spleen cells obtained from a (C57BU6J x BALB/cJ)Fl transgenic mouse with the SP2/0-Ag14 myeloma line 7 days following an i.p. boost of 100 kg of KLHARS in saline; the mouse had been immunized twice previously with 100 vg of KLHARS in complete Freund’s adjuvant at 1 month and 2 weeks prior to the i.p. boost. All hybridoma lines were subcloned by limiting dilution.

Quantltation

of Antibodies

Assays for anti-ARS were carried out in polyvinylchloride microtiter plates coated with bovine serum albumin-ARS as described (Dohi and Nisonoff, 1979); wells were developed with 1251-labeled affinity-purified rabbit anti-mouse Fab. For the AD8 depletion experiment, Fl mice were obtained by mating of ARS5 (ighb) mice with BALBlcJ (Igha) mice. Transgenic and nontransgenic Fl mice were immunized twice with 100 pg of KLHARS in complete Freund’s adjuvant; once at 1 month and once at 2 weeks prior to bleeding. Immune serum was adsorbed on ADESepharose; this treatment reduced total AD8 reactivity from 1400 &ml to 8 pglml in the transgenic serum, and from 48 pglml to 15 @g/ml in the nontransgenic serum. All values reported for AD8depleted samples are corrected for immunoglobulin dilution as estimated by Fab quantitation. To determine the level of IgGl antiARS antibodies of the a allotype, 1251-labeled lg(4.a)lO.9 MAb obtained from the American Type Culture Collection was used. lg(4.a)lO.g specifically recognizes the BALB/cJ Igh4a allotype on mouse IgGl (Oi and Herzenberg, 1979) and does not react with C57BU6J serum. Reactivity with the anti-idiotypic rat MAb AD6 (Hornbeck and Lewis, 1983) as well as CRIA and the R16.7 private idiotype was quantitated as in Durdik et al. (1989).

Pmparation

of Metaphaee

Spreads

Blood lymphocytes were isolated from 100 ~1 of whole blood of an ARS5 transgenic mouse and were cultured in RPM1 1640 medium with 10% fetal calf serum and 60 @/ml phytohemagglutinin for 72 hr. Cells

from hybridomas lGl0 and lE9 and the lymphocyte cultures were harvested by standard methods and spread on microscopy slides.

In Situ Hybridization The whole pARSv plasmid was nick translated with Bio-11-dUTP (Sigma) for in situ hybridization. Procedures for preannealing, hybridization, washing, blocking, detection, and amplification were described previously (Lichter et al., 1988; Pinkel et al., 1986) with minor changes. After denaturation and preannealing for 15 min at 3PC, a mixture of 15 vg/ml biotinylated probe, 0.2 mglml mouse genomic DNA, 0.25 mglml salmon sperm DNA, 50% formamide, 2x SSC, and 10% dextran sulfate was hybridized to chromosome spreads from ARS5 blood lymphocytes and hybridoma subclones for 16 hr. After hy bridization, the slides were washed in 50% formamide, 2x SSC (three times for 5 min, 42OC), followed by two washes in 4x SSC, 0.1% Tween (5 min each, 42OC) and one wash in 0.1x SSC, 0.1% Tween at 42OC (5 min). Sites of hybridization were detected with avidin conjugated with FITC (Vector Lab) followed by one round of amplification. Metaphase spreads were counterstained with 200 rig/ml propidium iodide, mounted with 90% glycerol and 2.3% DAPCO, then analyzed and scanned on a Phoibos 1000 confocal laser-scanning microscopy system.

Acknowledgments We thank Rebecca Harrison, Shari Laprise, Wayne Kotzker, and Matthew Lemer for expert technical assistance. We appreciate the generous contribution of DNA clones by Barbara Birshstein, Wesley Dunnick, and Laurel Eckhardt. We thank Joan Press, Ranjan Sen, Henry Wortis, and Thereza Imanishi-Kari for critical reading of the manuscript. We acknowledge support from NIH grants to E. S., A. N., and J. M. C. W. N. F. is a fellow of the Leukemia Society of America. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “edvertisement” in accordance with 18 USC Section 1734 solely to indicate this fact. Received

May 14, 1990; revised

July 26, 1990

References Cannon, L. E., and Woodland, R. T. (1983). Rapid and sensitive procedure for assigning idiotypic determinants to heavy or light chains: ap plication to idiotopes associated with the major cross-reactive idiotype of A/J anti-phenylarsonate antibodies. Mol. Immunol. 20, 1283-1288. Chen, Y. W., Word, C., Dev, V., Uhr, J. W., Vitetta, E., and Tucker, I? W. (1988). Double isotype production by a neoplastic B cell line. II. Allelitally excluded production of p and yl heavy chains without CH gene rearrangement. J. Exp. Med. 764, 562-579. Cory, S., and Adams, J. M. (1980). Deletions are associated with somatic rearrangement of immunoglobulin heavy chain genes. Cell 79, 37-51. Covarrubias, L., Nishida, Y., and Mintz, S. (1986). Early postimplantation embryo lethality due to DNA rearrangements in a transgenic mouse strain. Proc. Natl. Acad. Sci. USA 83, 6020-6024. Davis, M. M., Calame, K., Early, P. W., Livant, D. L., Joho, R., Weissman, I. L., and Hood, L. (1960). An immunoglobulin heavy chain gene is formed by at least two recombinational events. Nature 283,733-739. Dohi, Y., and Nisonoff, A. (1979). Suppression of idiotype and generation of suppressor T cells with idiotype-conjugated thymocytes. J. Exp. Med. 750, 909-918. Dunnick, W., Shell, 6. E., and Dery, C. (1983). DNA sequences near the site of reciprocal recombination between a c-myc oncogene and an immunoglobulin switch region. Proc. Natl. Acad. Sci. USA 80, 7629-7273. Dunnick, W., Wilson, M., and Stavnezer J. (1989). Mutations, duplication and deletion of recombined switch regions suggest a role for DNA replication in the immunoglobulin heavy chain switch. Mol. Cell. Biol. 9, 1850-1856. Durdik, J., Gerstein, R. M., Rath, S., Robbins, P. F., Nisonoff. A.. and Selsing, E. (1969). lsotype switching by a microinjected v immunoglob-

lsotype 547

Switching

of an lmmunoglobulin

ulin heavy chain gene 68,2346-2350.

in transgenic

Transgene

mice. Proc.

Natl. Acad.

Eckhardt, L. A., and Birshtein, B. K. (1965). Independent ulin class-switch events occurring in a single myeloma Cell. Biol. 5, 856-866.

Sci. USA

immunoglobcell line. Mol.

Erikson, J., Miller, D. A., Miller, 0. J., Abcarian, P W., Skurla, R. M., Mushinski, J. F., and Croce, C. M. (1965). The c-myc oncogene is translocated to the involved chromosome 12 in mouse plasmacytoma. Proc. Natl. Acad. Sci. USA 82, 4212-4216.

Natt, E., and Scherer, G. (1986). EMBL 12, a new lambda replacement vector with sites for Sall, Xbal, BamHl and EcoRl. Nucl. Acids Res. 14, 7128. Nikaido, T., Nakai, S., and Homo, T. (1981). Switch region of immunoglobulin Cu gene is composed of simple tandem repetitive sequences. Nature 292, 845-848. Nikaido, T, Yamawiki-Kataoka, Y., and Honjo, T (1982). Nucleotide quences of switch regions of immunoglobulin CE and Cy genes their comparison. J. Biol. Chem. 257, 7322-7329.

seand

Erikson, J., Mushinski, J. F., and Croce, C. M. (1986). The locus for the serum prealbumin is proximal to the heavy chain locus on mouse chromosome 12gt. J. Immunol. 138, 3137-3139.

Oi, V. T, and Herzenberg, L. A. (1979). Localization of murine la-lband Ig-la (IgGZa) allotypic determinants detected with monoclonal antibodies. Mol. Immunol. 18, 1005-1017.

Frankel, W. N., Stoye, J. P, Taylor, B.A., and Coffin, J. M. (1989a). Genetic analysis of endogenous xenotropic murine leukemia viruses: association with two common mouse mutations and the viral restriction locus /%I. J. Virol 83, 1763-1774.

Ott, D. E., Alt, F. W., and Marcu, K. B. (1987). lmmunoglobulin heavy chain switch region recombination within a retroviral vector in murine pre-B cells. EMBO J. 8, 577-584.

Frankel, W. N.. Stoye, J. P, Taylor, B. A., and Coffin, J. M. (1969b). Genetic identification of endogenous polytropic proviruses using recombinant inbred mice. J. Virol. 83, 3810-3621. Frankel, W. N., Stoye, J. P, Taylor, B. A., and Coffin, J. M. (1990). A linkage map of endogenous murine leukemia proviruses. Genetics 724, 221-236. Homo, T., and Kataoka, T. (1976). Organization of immunoglobulin heavy chain genes and allelic deletion model. Proc. Natl. Acad. Sci. USA 75, 2140-2144. Hornbeck, P V., and Lewis, G. K. (1963). ldiotype connectance in the immune system. I. Expression of a cross-reactive idiotype on induced anti-p-azophenylarsonate antibodies and on endogenous antibodies not specific for arsonate. J. Exp. Med. 157, 1116-1136. Hummel, M., Berry, J. K., and Dunnick, W. (1987). Switch region content of hybridomas: the two spleen cell /gh loci tend to rearrange to the same isotype. J. Immunol. 10, 3539-3548. Iwasato, T., Shim& A., Homo, T, and Yamagishi, H. (1990). Circular DNA is excised by immunoglobulin class switch recombination. Cell 82, 143-149. Kataoka, T., Miyata, T., and Homo, T (1981). Repetitive class-switch recombination regions of immunoglobulin genes. Cell 23, 357-366.

sequences in heavy chain

Katzenberg, D. R., and Birshtein, B. K. (1986). Sites of switch recombination in IgGPb and IgG2a-producing hybridomas. J. Immunol. 140, 3219-3227. Kipps, T. J.. and Herzenberg, L. A. (1966). recombination generating immunoglobulin variants. EMBO J. 5, 283-268. Knight, K. L., Malek, T. R., and Hanly, W. secretory immunoglobulin molecules: (paternal) variable-region allotypes and region allotypes. Proc. Natl. Acad. Sci.

Homologous chromosome allotype and isotype switch

C. (1974). Recombinant rabbit alpha chains with maternal paternal (maternal) constantUSA 77, 1169-1173.

Lichter, P, Cremer, T.. Borden, J., Manuelidis, L., and Ward, D. C. (1988). Delineation of individual human chromosomes in normal metaphase and interphase cells by in situ suppression hybridization using recombinant DNA libraries. Hum. Genet. 80, 224-234. Manser, T., and Gefter, M. L. (1986). The molecular evolution of the immune response: idiotope-specific suppression indicates that B cells express germ-line encoded V genes prior to antigenic stimulation. Eur. J. Immunol. 78, 1439-1444. Marcu, K. B., Lang, R. B., Stanton, model for the molecular requirements class switching. Nature 298, 87-89.

L. W., and Harris, L. J. (1982). A of immunoglobulin heavy chain

Matsuoka, M., Yoshida, K., Maeda, T., Usuda, S., and Sakano, H. (1990). Switch circular DNA formed in cytokine-treated mouse splenocytes: evidence for intramolecular DNA deletion in immunoglobulin class switching. Cell 82, 135-142. Messing, J., Crea, R., and Seeburg, F? (1981). A system for shotgun DNA sequencing. Nucl. Acids Res. 9, 309-321. Mowatt, M. R., and Dunnick, W. A. (1986). DNA sequence of the murine yl switch segment reveals novel structural elements. J. Immunol. 136. 2674-2683.

Perlmutter, A. F!, and Gilbert, W. (1984). Antibodies of the secondary response can be expressed without switch recombination in normal mouse B cells. Proc. Natl. Acad. Sci. USA 87, 7l89-7t93. Petrini, J., Shell, B., Hummel, M., and Dunnick, W. (1987). The immunoglobulin heavy chain switch: structural features of yl recombinant switch regions. J. Immunol. 738, 1940-1948. Piccoli, S. P, Caimi, l? G., and Cole, M. D. (1984). A conserved sequence at c-myc oncogene chromosomal translocation break points in plasmacytomas. Nature 310, 327-330. Pinkel, D., Straume, A. T, and Gray, J. W. (1986). Cytogenetic analysis using quantitative, high sensitivity, fluorescence hybridization. Proc. Natl. Acad. Sci. USA 83, 2934-2938. Rabbits, T. H., Forster, A., Dunnick, W., and Bentley, D. L. (1980). The role of gene deletion in the immunoglobulin heavy chain switch. Nature 283, 351-356. Radbruch, A., Muller, W., and Rajewsky, K. (1986). Classswitch recombination is IgGl specific on active and inactive IgH loci of IgGlsecreting B-cell blasts. Proc. Natl. Acad. Sci. USA 83, 3954-3957. Rath, S., Durdik, J., Gerstein, R. M., Selsing, E., and Nisonoff, A. (1989). Quantitative analysis of idiotypic mimicry and allelic exclusion in mice with a u lg transgene. J. Immunol. 743, 2074-2080. Sakano, H., Maki, R., Kurosawa, Y., Roeder, W., and Tonegawa, S. (1980). Two types of somatic recombination are necessary for the generation of complete immunoglobulin heavy chain genes. Nature 286, 676-683. Sanger, F., Nicklen, with chain-terminating 5483-5467

S., and Coulson, A. R. (1977). DNA sequencing inhibitors. Proc. Natl. Acad. Sci. USA 74,

Schwedler, U., Jack, H.-M., and Wabl, M. (1990). Circular product of the immunoglobulin class swtich rearrangement. 345452-456.

DNA is a Nature

Sharon, J., Gefter, M. L., Wysoki, L. J., and Margolies, M. N. (1989). Recurrent somatic mutations in mouse antibodies to p-azophenylarsonate increase affinity for hapten. J. Immunol. 742, 596-601. Shimizu, A., Nussenzweig, T. (1989). lmmunoglobulin in a human immunoglobulin USA 88, 8020-8023.

M. C., Mizuta, T R., Leder, P, and Honjo, double-isotype expression by trans-mRNA transgenic mouse. Proc. Natl. Acad. Sci.

Siekevitz, M., Gefter, M. L., Brodeur, P, Riblet, R., and MarshakRothstein, A. (1982). The genetic basis of antibody production: the dominant anti-arsonate idiotype response of the strain A mouse. Eur. J. Immunol. 12, 1023-1032. Southern, fragments

E. M. (197’5). Detection of specific separated by gel electrophoresis.

sequences among DNA J. Mol. Biol. 98, 503-517.

Stanton, L. W., Yang, J.-Q., Eckhardt, L. A., Harris, L. J., Birshtein, B. K., and Marcu, K. B. (1984). Products of a reciorocal chromosome translocation involving the c-rnkc gene in a murine plasmacytoma. Proc. Natl. Acad. Sci. USA 877, 829-833. Stoye, J. P, and Coffin, J. M. (1988). Polymorphism of murine endogenous proviruses revealed by using virus class-specific oligonucleotide probes. J. Virol. 62, 168-175. Taub. R., Kirsch, I., Morton, C., Lenoir, G., Swan, D., Tronick, S., Aaronson, 9, and Leder, I? (1982). Translocation of the c-myc gene into the

Cdl 548

immunoglobulin heavy chain locus in human Burkitt lymphoma murine plasmacytoma cells. Proc. Natl. Acad. Sci. USA 79, 7839-7841. Vieira, J., and Messing, J. (1982). The pUC plasmids, an M13mp7derived system for insertion mutagenesis and sequencing with synthetic universal primers. Gene 79, 259-288. Weaver, D., Reis, M. H., Albanese, C., Costantini, F., Baltimore, D., and Imanishi-Kari, T. (1988). Altered repertoire of endogenous immunoglobulin gene expression in transgenic mice containing a rearranged mu heavy chain gene. Cell 45, 247-259. Winter, E., Krawinkel, V., and Radbruch, A. (1987). Directed lg class switch recombination in activated murine B cells. EMBO J. 6, 1883-1871. Wirschubsky, Z., Ingvarsson, S., Carstenssen, A., Weiner, F., Klein, G., and Sumegi, J. (1985). Gene localization on sorted chromosomes: definitive evidence on the relative positioning of genes participating in the mouse plasmacytoma-associated typical translocation. Proc. Natl. Acad. Sci. USA 82, 8975-8979. Wysoki, L. J., and Gefter, M. L. (1989). The molecular basis of a Vu gene polymorphism that determines the expression of a major idiotype. Mol. Immunol. 26, 1143-1150. Wysoki, L. J., Manser, T., and Gefter, M. L. (1988). Somatic evolution of variable region structures during an immune response. Proc. Natl. Acad. Sci. USA 83, 1847-1851. Yaoita, Y., Kumagai, Y., Okumura, K., and Homo, T. (1982). Expression of lymphocyte surface IgE does not require switch recombination. Nature 297, 897-899 GenBank

Accession

The accession M38389.

number

Number for the sequence

reported

in this paper

is