High-Level Rearrangement and Transcription of Yeast Artificial Chromosome-Based Mouse Igκ Transgenes Containing Distal Regions of the Contig¹

The mouse Igκ gene locus has provided a paradigm to study many interesting biologically relevant problems, including: DNA sequence organization and evolution (1), tissue-specific transcriptional regulation (2, 3), site-specific recombination (4, 5, 6), somatic hypermutation (7, 8, 9), various aspects of DNA methylation (10, 11, 12), and the relationship between chromatin structure and function (13, 14, 15, 16, 17).

The mouse κ locus is the largest Ig gene locus thus far identified, spanning more than 3.5 Mb (1, 18, 19). The locus contains 93 potentially functional Vκ regions that have been grouped into 18 families based on sequence homologies, four functional and one nonfunctional Jκ regions, and a single Cκ exon (20, 21, 22, 23, 24, 25). The most 5′ V region family was originally thought to be Vκ2, 3.5 Mb away from the Jκ-Cκ region (18, 19). However, recently a single Vκ24 family member has been found further upstream (23, 24, 25). The most 3′ Vκ region family is Vκ21, and the closest Vκ21 gene member is Vκ21G, 18 Kb away from Jκ1 gene segment (26).³

Several cis-acting regulatory elements have already been identified in the mouse Igκ locus. Except for Vκ region promoter elements, all of these elements reside near or within the Jκ-Cκ region at the 3′ end of the locus. These include: two germline promoters (27), KI-KII sequences (28), a nuclear matrix association region (MAR)⁴ (29), an intronic enhancer (30), and a 3′ enhancer (31). Targeted deletions of these elements in cells or mice have been performed, allowing their functional significance to be addressed in the native locus. Targeted deletion of the KI-KII sequences, a germline promoter, or both reveals a phenotype of suppressed recombination (28, 32, 33). Targeted deletion of the MAR in cultured pre-B cells results in hyperrecombination (11), whereas deletion of the same element from the mouse germline only mildly advances the timing of Vκ-Jκ joining during development (34). Deletion of either the intronic or 3′ enhancers severely affects Igκ gene rearrangement but does not abolish it (35, 36), suggesting either that each enhancer can partially compensate for the loss of the other or that other elements also contribute to Igκ gene expression.

Despite identification of all of the above regulatory elements, the results of transgenic mice studies strongly suggest that additional crucial regulatory elements within the Igκ locus remain to be discovered. Human Igκ germline minilocus transgenes containing all the corresponding known regulatory elements described above have been ectopically introduced into the mouse germline. Only poor and erratic expression was noted relative to that obtained by the endogenous mouse Igκ locus (37, 38, 39, 40, 41). However, because more recent studies have been successful in achieving high-level expression of human Abs after introducing either entire chromosomes or 300-1300 kb yeast artificial chromosomes (YACs) bearing the human Igκ locus into mice (42, 43), it is likely that other important regulatory elements present in the native human locus but not present on the earlier shorter transgene constructs remain to be discovered. The missing sequences may reside in the distal 5′ and 3′ regions of the locus, because the above poorly expressed transgenes did not possess both of these regions on one construct. We propose a similar case for the mouse locus. While sequences centrally located in the locus are frequently deleted upon Vκ-Jκ joining, the upstream and downstream regions are always maintained in a recombined native locus.

YAC technology has provided an ideal tool to manipulate and engineer large DNA fragments. YAC clones have been introduced into transgenic mice and mammalian cells to study gene expression (44, 45). In this report, we have used YAC-based technology to fuse 5′ and 3′ distal regions of the mouse Igκ locus to create 225–290 kb Igκ miniloci with differing lengths of sequence 5′ of the Vκ2 gene cluster. We have generated transgenic mice carrying these miniloci and functionally analyzed their behavior in germline and rearranged gene transcription as well as Vκ-Jκ joining. This approach has for the first time achieved high levels of both rearrangement and transcription of rearranged mouse Igκ transgenes, suggesting that most regulatory elements in the locus reside within our Igκ miniloci. These miniloci should be valuable as reagents for future functional analyses of cis-acting elements.

The selectable markers TRP1 and URA3 on the vector arms of the YAC clone FAR.E8, which resides in the 5′ region of the contig (18), were replaced with LYS2 and HIS3, respectively, by one-step gene replacement (46). The TRP1-targeting vector contained a new selectable marker, LYS2, flanked by fragments of the gene TRP1. It was constructed as follows: an EcoRI and ClaI fragment containing LYS2 was excised from pDA6200 (47) and subcloned into pBluescript (Stratagene, La Jolla, CA) as pSK/LYS2. The 5′ fragment of TRP1 was amplified by primers (5′ to 3′) GCT CTG CAG TGG AAA ACG TTC TTC GGG GCG and CTT GAA TTC TAT TGA AAA AGG AAG AGT ATG using pRS304 (Stratagene) as template, and it was subcloned into the PstI and EcoRI sites of pSK/LYS2. The 3′ fragment of the TRP1 was PCR amplified by primers (5′ to 3′) CTG CAT CGA TAT GAG TCG TGG CAA and AAC CCA GTC GAC AAT CGA GTT CCA ATC CAA from pRS304 and subsequently cloned into the ClaI and SalI sites of pSK/LYS2. The PCR amplifications were conducted for 1 min at 94°C, 2 min at 55°C, and 1 min at 72°C for 30 cycles. The resulting construct was digested with PstI and SalI to release targeting sequences. Then, 1 μg of targeting DNA fragments was transformed into FAR.E8-bearing yeast cells by the lithium acetate method (48). After selecting for growth on medium lacking lysine, targeted transformants were confirmed by PCR and Southern analyses.

Similarly, the URA3-targeting vector contained a new selectable marker, HIS3, flanked by fragments of the gene URA3. The 5′ and 3′ fragments of URA3 were PCR amplified from pRS306 (Stratagene). The primers (5′ to 3′) for the 5′ fragment were AGG AGC TCG AGT CGA AAG CTA CAT ATA AGG AAC GTG and GAT TTT TCC ATG GAG GGC ACA GTG CGG CCG CTA; primers for the 3′ fragment were GCA GAA TTC TCA TGC AAG GGC TCC CTA GC and GTG GAT GAT GTG GTC TCT ACA GGA CTC GAG ATT. To construct the HIS3::ura3 vector, a BamHI fragment containing the HIS3 gene from pRS303 (Stratagene) was subcloned into pBluescript, followed by subcloning of the 5′ and 3′ URA3 fragments into SacI/NotI and EcoRI/XhoI sites, respectively. The targeting fragment was released by digestion with XhoI and was transformed into FAR.E8-bearing yeast cells. The transformants were selected for growth on medium lacking histidine and confirmed for targeting by PCR and Southern analyses. Retrofitted FAR.E8 was analyzed by pulsed-field gel electrophoresis (PFGE) and Southern hybridization to confirm that no obvious deletion or aberrant recombination occurred during these processes.

Chimeric material in the YAC clone FAW.A3, which resides within the 3′ region of the contig (18), was eliminated by chromosome fragmentation (46). The fragmentation vector was constructed as follows. First, a SalI/ClaI yeast telomere fragment was subcloned into pRS315 (Stratagene) containing the LEU2 selectable marker, CEN and ARS sequences. Then a BamHI fragment from plasmid pRSB (49), which contains the cryptic recombination sequence (RS), was subsequently subcloned into pRS315. The targeting vector was linearized by SmaI and transformed into yeast cells bearing FAW.A3. Transformants were selected by growth on synthetic medium lacking leucine. Those transformants resulted from self-ligation and propagation of the targeting vector were eliminated by replica plating on medium lacking both leucine and tryptophan. Final candidates were analyzed by PFGE.

We purified retrofitted FAR.E8 and FAW.A3d after PFGE under conditions of mild shearing for cotransformation. This resulted in DNA fragments in the range of 200–300 kb that were highly recombinagenic. The PFGE run conditions were 1% SeaPlaque agarose (FMC BioProducts, Chicago, IL), 0.25× TBE (45 mM Tris-borate, 1 mM EDTA, pH 8.3), auto algorithm separating DNA with a size range of 400–800 kb, using a Bio-Rad CHEF mapper apparatus (Bio-Rad, Richmond, CA). The gel slices corresponding to the YACs were excised and equilibrated in 1× TAE (40 mM Tris-acetate, 1 mM EDTA, pH 7.8), 100 mM NaCl, 30 mM spermine, and 70 μM spermidine for 2 h at room temperature. Gel slices were melted at 68°C for 10 min and treated with Gelase (Epicentre Technologies, Madison, WI) at 1 U/100 mg of gel at 42°C for 2 h. The DNA of both YACs (1 μg each) was cotransformed into the host strain YPH857 by spheroplast transformation using 10⁹ cells (50). Yeast cells bearing potential recombinant YACs were selected for by the presence of the two external markers, HIS3 and LEU2, for growth in the corresponding dropout medium, and for the absence of the internal markers, URA3 and LYS2, by negative selection with 5-fluoroorotic acid and α-aminoadipic acid. We screened 11 clones of recombinant YACs by PCR for the presence of Vκ2 and Vκ21, the most 5′ and 3′ Vκ gene families, and identified 7 YACs positive for both Vκ families. Further analysis by PFGE as above revealed a 240-kb candidate Igκ minilocus (κML) that had an ideal size for future transgenic and cell culture experiments.

Total DNA from yeast cells carrying the κML, or one of the parent YACs, FAR.E8 or FAW.A3, was partially digested with SfiI and separated by PFGE. DNA was nicked by exposure to UV light (180 mJ) and transferred to Nytran-Plus membrane, pore size 0.22 μM (Schleicher & Schuell, Keene, NH). The DNA was hybridized to either of the vector arm probes, L or R, labeled with ³²P using a Random Primer Labeling kit (Amersham, Arlington Heights, IL). Hybridization was conducted as previously described (51). The hybridization was at 65°C for 18 h. Filters were then washed with 0.2× SSC, 2% SDS for 30 min and 0.1× SSC, 2% SDS for 30 min at 65°C. Membranes were exposed to PhosphorImaging screens, and images were analyzed using ImageQuant software (Molecular Dynamics, Sunnyvale, CA).

The procedure used to create single-base changes has been described previously (52). The Kluvermyces lactis URA3 (kURA3) gene was first amplified from genomic DNA using primers with 50-bp tails that result in a direct repeat spanning the desired mutation. The primers (5′ to 3′) for the Jκ-targeting construct were AAC TAG GGG AAG AGG GAT AAT TGT CTA CCA TGG GAG GGT TTT GTG GAG GTA GCT CTT CAA TTC ATC TTT TTT TTT TTT GTT CTT T and ACC TCC ACA AAA CCC TCC CAT GGT AGA CAA TTA TCC CTC TTC CCC TAG TTG GGT AAT AAC TGA TAT AAT TAA ATT GAA GCT. Each primer was used for single-strand extension using K. lactis genomic DNA as template for 15 cycles at 94°C for 1 min, 55°C for 2 min, and 72°C for 2 min. Both amplification reactions were mixed, and 20 cycles of amplification were conducted under the same conditions. To increase the efficiency of targeting, the resulting PCR product was reamplified by adding 70 bp of flanking sequence to each direct repeat. The primers were (5′ to 3′) GAA AAC TGT CCC ACA AGA GGT TGG AAT GAT TTT CAG GCT AAA TTT TAG GCT TTC TAA ACC AAA GTA ACT AAA CTA GGG GAA GAG GGA TAA TTG TC and ATG TAC TTA GGT TTT ATT TCC AGT CTG GTC CCA TCA CTG AAT GTG ATT TAC AGT GAT TTA TTT TAA CTT TAC CTC CAC AAA ACC CTC CCA TGG TA. The amplification was performed as described above.

The primers (5′ to 3′) for the Cκ-targeting construct were ACG ACA AAA TGG CGT CCT GAA CAG TTG GAC AGA TCA GGA CAG CAA AGA CAG CAC CTA CAG CAT GAG CAG GTG ATT CTG GGT AGA AGA TCG GTC and CTG CTC ATG CTG TAG GTG CTG TCT TTG CTG TCC TGA TCT GTC CAA CTG TTC AGG ACG CCA TTT TGT CGT TGT GTG CTT GCT TCT TTT CTT ATC CG. Similarly, the PCR product was reamplified by adding 70 bp flanking sequence to each direct repeat. The extension primers (5′ to 3′) were AGT CGT GTG CTT CTT GAA CAA CTT CTA CCC CAA AGA CAT CAA TGT CAA GTG GAA GAT TGA TGG CAG TGA ACG ACA AAA TGG CGT CCT GAA CAG and GTC TTG TGA GTG GCC TCA CAG GTA TAG CTG TTA TGT CGT TCA TAC TCG TCC TTG GTC AAC GTG AGG GTG CTG CTC ATG CTG TAG GTG CTG TCT. The conditions for PCR reactions were as described above. Then, 1 μg each of targeting constructs were transformed into κML-bearing yeast cells, and transformants were selected for the kURA3 gene and then screened for targeting by PCR. Next, negative selection with 5-fluoroorotic acid was conducted to isolate cells that had spontaneously looped out the kURA3 gene. PCR amplification and digestion by AvrII/NcoI or by BclI were used to diagnose whether the introduced mutations were retained after recombination.

5′DκML was created by eliminating the 15-kb upstream sequence from κML by chromosome fragmentation. The fragmentation vector was constructed as follows. A ClaI/SalI telomere fragment from pBP103 (American Type Culture Collection, Manassas, VA) was cloned into pRS304 (Stratagene). Targeting sequences were amplified by PCR using a Vκ2(70/3)-specific primer pair engineered to have a PmeI site at its 3′ end followed by SalI sites at both ends. The primers were (5′ to 3′) ACG CGT CGA CAG GAA GCC CAC ATA ACT GCC CCT, and ACG CGT CGA CGT TTA AAC TCC AGA GTC CAG TTT AGA CAC CAG A. The PCR conditions were 1 min at 94°C, 2 min at 55°C, and 1 min at 72°C for 30 cycles. The SalI cleaved products were gel purified and cloned into pRS304. The final vector was linearized by PmeI and transformed into κML-bearing yeast cells. Transformants were selected for the presence of the TRP1 marker and screened for lack of the HIS3 marker by replica plating.

A 684-bp single copy DNA fragment designated P5 was amplified from a region located 3.5 kb upstream of Vκ2(70/3). The primers (5′ to 3′) were GTC TCT TCT AGC CCA CTG ACC ATA G and CTT GAA TGG CCA AGA AGC ACT TAG AGA. The PCR amplifications were conducted 1 min at 94°C, 2 min at 55°C, and 1 min at 72°C for 30 cycles. The P5 fragment was used as a probe in screening the mouse genomic BAC library from Research Genetics (Huntsville, AL). A 100-kb BAC clone (designated 5′κBAC) was identified that hybridized to the P5 probe. 5′κBAC DNA was digested with NotI, and a 90-kb fragment spanning the majority of the BAC insert plus 200 bp of the BAC vector sequence was gel purified after PFGE. The NotI fragment was ligated to a linker plasmid containing a telomere seed fragment, a neomycin resistance gene, and the TRP1 selectable marker, followed by digestion of 500 ng DNA with 6 U of exonuclease III for 10 min at 37°C. The resulting fragment was transformed into κML-bearing yeast cells by spheroplast transformation, and transformants were selected for the TRP1 marker. After replica plating, those transformants that had lost the HIS3 marker were analyzed by PFGE.

First, 20 ng of BAC DNA and 3 μg of total yeast DNA isolated from the YPH857 yeast strain, or the same strain bearing either κML or 5′EκML, was digested by SphI or KspI and resolved by PFGE. DNA transfer, hybridization, and image analysis were performed as described above. A repetitive sequence fingerprinting assay was performed as previously described (53, 54) with minor modifications. BAC or yeast DNA were digested by the 4-bp recognition restriction enzymes, HaeIII or HinfI, and electrophoretically resolved on 1% SeaKem agarose gels (FMC BioProducts) in 1× TAE running buffer. DNA was transferred to Nytran-Plus membranes (Schleicher & Schuell) by neutral transfer in 20× SSC. The probes were an equal-molar mixture of the oligonucleotides (GAA)₆, (GACA)₄, (GGGCA)₃, (GATA)₄, and (CT)₄(CA)₅. The probes were labeled by T4 polynucleotide kinase (NEB, Beverly, MA) following the manufacturer’s instructions. Membranes were prehybridized for 3 h and hybridized for 18 h at 42°C in 25 mM Tris, pH 7.5, 1 M NaCl, 50% formamide, 10% dextran sulfate, 1% SDS, 5× Denhart’s solution, with sonicated denatured salmon sperm DNA at 100 ng/ml. Wash conditions were 0.2× SSC, 0.1% SDS for 20 min at room temperature and twice for 30 min at 45°C. Membranes were exposed to PhosphorImage screens, and the images were analyzed using ImageQuant Software (Molecular Dynamics) as described above.

Because κML and 5′DκML nearly comigrated with yeast chromosome I during PFGE, we bisected this yeast chromosome into 150-kb and 90-kb minichromosomes using a chromosome I fragmentation vector (55), creating a new yeast strain for propagation of these miniloci before their isolation. The bisection vector pFLC273 (a gift from Dr. David Kaback) was linearized with BamHI and transformed into yeast cells. Transformants obtained on selective medium for the URA3 marker were subsequently analyzed by PFGE.

Igκ miniloci DNA was purified by PFGE according to standard protocols (56) with minor modifications. The YACs were separated by PFGE in 1% SeaPlaque agarose gels (FMC BioProducts), 0.5× TAE running buffer, with an auto algorithm set for a size range of 20–300 kb, using a Bio-Rad CHEF mapper apparatus. The gel slices corresponding to the YACs were excised. After equilibration in 1× TAE, slices were positioned on a gel tray and embedded in 4% Nusieve GTG agarose (FMC BioProducts). A second gel run was performed at a 90° angle to the PFGE for 3 h at 4 V/cm. The YAC-containing gel slices were excised and equilibrated in 1× TAE, 100 mM NaCl, 30 μM spermine, and 70 μM spermidine for 2 h at room temperature. The gel slices were melted at 68°C for 10 min and then treated with Gelase (Epicentre Technologies) at 1 U/100 mg of gel at 42°C for 2 h. The agarized solution was concentrated by centrifuging 30 min at 6000 × g in a Millipore filter unit, Ultrafree MC 30,000 NMWL (Millipore, Bedford, MA). The concentrated DNA was dialyzed on a Millipore type VM filter, pore size 0.05 μM, on the surface of microinjection buffer (10 mM Tris, 0.1 mM EDTA, 100 mM NaCl, 30 μM spermine, 70 μM spermidine) for 4 h at room temperature (56). The DNA was microinjected into C57BL/6 × SJL mouse zygotes, using ∼2 pl/cell at 0.5–3 ng/μl. Transgenic founders were identified by assay of tail DNA for the presence of left and right vector arms of the YAC constructs and were bred to C57BL/6 mice to create F₁ progeny for further analyses.

Genomic DNA was isolated from mouse cells or tissues after incubation overnight at 55°C in 0.1 M EDTA, 0.5% N-lauroylsarcosine, 100 μg/ml proteinase K. Lysates were extracted once with equilibrated phenol:chloroform:isopropanol (25:24:1) and once with chloroform. DNA was precipitated by adding 2 vol of ethanol, briefly dried and resuspended in TE (10 mM Tris, 1 mM EDTA, pH 8.0). Samples were digested with restriction enzymes following the manufacturers’ recommendations and resolved in 1% agarose gels. DNA was transferred to Nytran-Plus membrane (Schleicher & Schuell) by neutral transfer in 20× SSC. Probes were labeled with [α³²P]dCTP using the Rediprime II labeling system (Amersham) following the manufacturer’s instructions. Membranes were prehybridized for 2 h and hybridized to labeled probes for 18 h at 65°C as described above. Probed membranes were washed twice for 30 min in 0.2× SSC, 2% SDS and once for 30 min in 0.1× SSC, 2% SDS at 65°C. The images were analyzed using ImageQuant Software (Molecular Dynamics) as described above.

Copy numbers of both the 5′ and 3′ regions of Igκ miniloci were determined by Southern analyses of mouse tail genomic DNA as described above. Because a point mutation was introduced into the miniloci that eliminated a BclI site in the Cκ region, digestion of tail DNA with BglII/BclI and hybridization with a BglII/BclI probe isolated from plasmid pJCκ6.8 (34) yields a 2.4-kb band derived from the two endogenous alleles and a 2.8-kb band derived from the transgenes. Thus, the following equation was used to estimate copy number of the 3′ region: 3′ copy number = (signal of 2.8 kb/signal of 2.4 kb) × 2. The same filters were hybridized with the P5 probe amplified from the 5′ region upstream of the most distal Vκ2 gene segment as described above. The hybridization gives rise to a 1.1-kb band that is derived from both the transgenes and endogenous locus. Using the 1.1-kb band from a nontransgenic mouse tail DNA as the reference (copy number = 2) and the 2.4-kb Cκ band as an internal loading normalization control, the 1.1-kb band from the transgenic mice tail DNA was quantified as that which represented the copy number of the 5′ region as the sum of the transgenes and endogenous alleles. Subtraction of 2 gave rise to the estimated 5′ copy number. The following equation was used for calculation: 5′ copy number = (signal of transgenic 1.1 kb/signal of nontransgenic 1.1 kb) × 2 − 2. However, the 5′ upstream region where the P5 probe was derived from had been deleted in the 5′DκML. To circumvent this problem, the filter was hybridized to a neomycin resistance gene probe. Because all three miniloci contain a single copy of the neomycin resistance gene in the 5′ vector arm, the 5′ copy number of the transgene in the 5′DκML mice was estimated by quantifying signals of hybridization with the neomycin probe using DNA with known 5′ copy numbers from κML transgenic mice lines as the reference. For instance, hybridization of tail DNA isolated from transgenic mice 57-2 and 14-1 with the neomycin probe gave rise to bands with the same intensity after normalization. Thus, the 5′ copy number in transgenic line 57-2 is estimated to be 3, the same as that of 14-1.

The RT-PCR for germline transcription was performed as described previously (24). The primers were as follows (5′ to 3′): κ⁰, ACA GCC AGA CAG TGG AGT ACT ACC; Cκ, TTA GTG GCT CTG TTC CTA TCA CTG TGT CCT CAG G. Reactions in 100 μl were performed with cDNA reverse transcribed from 100 ng total RNA isolated from mouse tissues using the Superscript System (Boehringer Mannheim, Mannheim, Germany) and 5 pmol of each primer at 3 mM Mg²⁺ using the Expand High Fidelity PCR System (Boehringer Mannheim) for 1 min at 94°C, 2 min at 60°C, 1 min at 72°C for 30 cycles. One picomole of the κ⁰ primer labeled with [γ-³²P]ATP using T4 polynucleotide kinase (Boehringer Mannheim) was added to the PCR products after 30 cycles. The final cycle was 1 min at 94°C, 2 min at 60°C, 10 min at 72°C. Then, 10 μl of the PCR products was digested with BclI and electrophoretically separated on 1.5% agarose gels in 1× TAE running buffer. Dried gels were exposed to PhosphorImaging screens, and images were analyzed as above.

PCR assay for gene rearrangement was performed based on the strategy developed by Schlissel and Baltimore (57). The PCR primers for the rearrangement assay were as follows (5′ to 3′): Vκ2 primer, GTC AAG TCA GAG CCT CTT AGA TAG TGG AAA GAC ATA TTT; Vκ9 primer, GCA AGG CGA GTC AGG ACA TTA ATA GCT ATT TAA GCT GG; Vκ20 primer, GTC ACT ATC AGA TGC ATA ACC AGC ACT GAT ATT GAT GAT GAT; Vκ21 primer, GTG AAA GTG TTG AAT ATT ATG GCA CAA GTT TAA TGC AGT; Jκ5 primer, GAG CCC TCT CCA TTT TCT CAA GAT TTT CTG AAC TG. Reactions in 100 μl were performed with 500 ng splenic DNA template and 5 pmol of each primer at 3 mM Mg²⁺ using the Expand High Fidelity PCR System (Boehringer Mannheim) for 1 min at 94°C, 2 min at 65°C, and 1 min 20 s at 72°C for 30 cycles. One picomole of the Vκ2 primer labeled with [γ-³²P]ATP using T4 polynucleotide kinase (Boehringer Mannheim) was added to the PCR products after 30 cycles. The final cycle was 1 min at 94°C, 2 min at 65°C, and 10 min at 72°C. Then, 10 μl of the PCR products were digested with the restriction endonucleases NcoI or AvrII and electrophoretically separated on 1% agarose gels in 1× TAE running buffer. Dried gels were analyzed as above.

Vκ2-Jκ1 and Vκ2-Jκ2 rearrangements were amplified by PCR. The PCR primers used were the Vκ2 primer described above and a Jκ3 primer, CCT TTC TCA TTT CTC CCA CAA ATC TGA. The PCR conditions were 1 min at 94°C, 2 min at 60°C, 1 min at 72°C for 30 cycles using Tag DNA polymerase (Boehringer Mannheim). The PCR products were resolved on 1% agarose gel. The 1-kb Vκ2-Jκ1 and 700-bp Vκ2-Jκ2 rearrangement amplification products were gel-purified using Gel Extraction Kit (Qiagen, Chatsworth, CA) and subsequently cloned into the pGEM-T vector (Promega, Madison, WI). After sequencing, the rearrangement observed in individual clones could be deduced to correspond to either that of the endogenous alleles or transgenes, because of the presence or absence of AvrII/NcoI sites in the inserts.

The primers for rearranged gene transcription were the Vκ and Cκ primers described above. Reactions in 100 μl were performed with cDNA reverse transcribed from 100 ng total RNA isolated from mouse tissues and 5 pmol of each primer at 3 mM Mg²⁺ using the Expand High Fidelity PCR System for 1 min at 94°C, 2 min at 65°C, 1 min at 72°C for 30 cycles. One picomole of the κ⁰ primer or a Vκ primer that was labeled with [γ-³²P]ATP using T4 polynucleotide kinase (Boehringer Mannheim) was added to the PCR products after 30 cycles. The final cycle was 1 min at 94°C, 2 min at 65°C, 10 min at 72°C. Then, 10 μl of the PCR products were digested with BclI and electrophoretically separated on 1.5% agarose gels in 1× TAE running buffer. Gels were dried and analyzed as described above.

Activated B cells from 4-day LPS-cultured splenocytes were fused to the non-κ-producing myeloma cell line P3X63Ag8 (58) by standard techniques. Supernatants from clones growing in hypoxanthine/aminopterin/thymidine-supplemented RPMI 1640 medium were screened for Igκ production by ELISA. Ninety-six-well plates were coated with 3 μg/ml salmon sperm DNA before ELISA. Cells in positive wells were then subcloned by limiting dilution. Total RNA was isolated from expanded clones, and Northern hybridization analysis was performed using 10 μg/lane as described elsewhere (59).

Single-cell suspensions were prepared from spleen and 4 × 10⁷ cells were incubated with an optimal concentration of biotinylated Abs in 100 μl PBS/0.1% BSA for 10 min on ice. After washing with PBS/0.1% BSA, cells were incubated with 20 ml of streptavidin microbeads (Miltenyi Biotech, Auburn, CA) for 30 min at 4°C. After removal of unbound microbeads, the cells were loaded into a mini-MACS unit (Miltenyi Biotech). Columns were washed three times with 200 μl PBS/0.1% BSA before the positive fraction was eluted following the manufacturer’s instructions. During the procedure, aliquots of fractions were stained with streptavidin-PE for FACS analysis to monitor the quality of the fraction. The fractionated cells were lysed with 0.5% N-lauroylsarcosine, 100 mM EDTA, 100 μg/ml proteinase K for genomic DNA isolation.

Using state-of-the-art techniques for manipulating YACs, we created a parent 240-kb Igκ minilocus (κML) by recombining a 5′ YAC (FAR.E8) with a 3′ YAC (FAW.A3), separated by some 3 Mb in the Igκ locus (1, 46, 52, 60 ; see Materials and Methods). As shown in Fig. 1, based on previous physical mapping studies (1, 18, 19, 23, 24, 25, 26), κML is predicted to contain all three functional members of the Vκ2 gene family and one functional member of the Vκ21 gene family, along with uninterrupted sequences spanning from Vκ21G to the 3′ RS. This predicted structure was verified by numerous assays (see below). To distinguish the gene rearrangement and expression of κML from those of the endogenous alleles by PCR and RT-PCR assays (see below), κML also contained engineered single base changes, converting an AvrII site between the Jκ2 and Jκ3 regions to an NcoI site and eliminating a BclI site in the Cκ region by introducing a silent mutation. To test the hypothesis that the sequences 5′ of the most distal Vκ2 gene family member, Vκ2(70/3), may provide regulatory functions to downstream Vκ2 genes, we also created deleted and extended forms of κML, termed 5′DκML and 5′EκML, respectively (Fig. 1). In summary, the YAC-based engineering lead to the creation of three miniloci with differing lengths of sequences upstream of the Vκ2 gene cluster (Fig. 1).

To confirm that κML originated from recombination between YACs FAR.E8 and FAW.A3d and roughly map the break points, we conducted an indirect end-labeling assay (Fig. 2,A). DNA from the two parent YACs and κML was partially digested with SfiI; after PFGE, Southern transfers were hybridized to vector arm probes from each parent YAC. Comparison of the resulting banding patterns between FAR.E8 and κML revealed a break point located between ∼165 and 215 kb downstream of the 5′ end of FAR.E8 (Fig. 2,A, left, broken arrow). Similarly, the junction was identified between 70 and 90 kb upstream of the 3′ end of FAW.A3d (Fig. 2,A, right, broken arrow). The sum of these breakpoint values is within the range of the length of κML as analyzed by PFGE, indicating that no significant internal deletion occurred during recombination. Southern analyses and PCR assays has revealed that in addition to the Jκ-Cκ region, the κML contains the expected Vκ2 and Vκ21 gene family members (Ref. 61 and data not shown). Junctions between the κML and the 5′EκML were also investigated by Southern analyses (Fig. 2,B). A 40-kb SphI fragment in 5′κBAC was retained in 5′EκML as well as an 85-kb KspI fragment (Fig. 2,B, 1–3, arrows). The 8-kb vector at the 5′ end of 5′EκML accounts for the length discrepancy between the KspI fragments of the 5′κBAC and 5′EκML, and the top band in the KspI-digested BAC resulted from partial cleavage due to inhibition by bacterial specific methylation (Fig. 2,B, 2). Furthermore, the extension of the 5′ region was confirmed by genomic fingerprinting assays. κML, 5′EκML, and 5′κBAC DNA digests were hybridized to mouse genome simple repeat probes. 5′EκML digests contained all bands present in the κML digests and those bands that were unique in the BAC digests (Fig. 2,B, 4 and 5, arrows), consistent with the lack of deletions or aberrant rearrangements. Finally, Southern hybridization analysis of the products of chromosome fragmentation at of κML at Vκ2 gene segments indicated that κML contains ∼15 kb of upstream sequence (Fig. 2 C, compare the length of κML in lane 1 to those 5′DκML fragmentation products in lanes 3–4, 6, and 8, arrow). We conclude that the Igκ miniloci thus generated behave in all ways tested predictable for their anticipated structures.

We isolated the three Igκ miniloci constructs described above by PFGE for microinjection into fertilized mouse eggs to create transgenic mice. A total of 16 different transgenic mice were identified by dot blot hybridization using miniloci vector arm probes, 14 of which transmitted the transgenes to F₁ progeny (Table I). The copy numbers of transgenes was determined by genomic Southern hybridizations using 5′ and 3′ end probes. A representative Southern assay for the 3′ copy number estimation in founder animals is shown in Fig. 3. Because a point mutation was introduced into the miniloci that eliminated a BclI site in the Cκ region, after digestion of tail DNA with BglII/BclI, the corresponding endogenous and transgene fragments are 2.4 and 2.8 kb, respectively (Fig. 3), the relative signal intensities of which were used for quantitation. Estimate of 5′ copy number used band signal intensities relative to internal and external standards (Ref. 61 and data not shown). The copy numbers of transgenes was found to range from 1 to 14 (Table I). Among the16 transgenic mouse lines, 11 had the same copy numbers for both 5′ and 3′ ends of the construct, suggesting that most integrated miniloci were intact, possibly organized as tandem arrays (62). It can also be deduced that at least one copy of these constructs must be intact in every transgenic line created from the fact that all such lines exhibited rearrangement of Vκ2 and Vκ21 genes (see below).

To study Igκ miniloci expression, we first evaluated germline transcription in F₁ progeny. By performing RT-PCR with spleen and bone marrow RNA samples, we assayed for the level of spliced germline transcripts produced from the 3′ germline promoter (27). Taking advantage of the eliminated BclI site in the Cκ region of the miniloci constructs, germline transcripts arising from the endogenous and transgene alleles could be distinguished by gel electrophoresis after BclI digestion of the RT-PCR products (Fig. 4, A and B). Fig. 4,B illustrates a representative assay performed on wild-type and two F₁ transgenic mice progeny. Our quantitated results demonstrate that the levels of germline transcription in transgenes are approximately the same as those of the endogenous locus and exhibit position-independent and copy number-dependent expression (Fig. 4 C), except for those transgenic lines with very high copy numbers.

To investigate the level of rearrangement of Igκ miniloci, we assayed for Vκ-Jκ joining using a PCR assay (57). Specific primers for Vκ2 and Vκ21 families present in the miniloci were used to assay for their corresponding rearrangement. Four PCR products are expected that represent VκJκ1, VκJκ2, VκJκ4, and VκJκ5 rearrangements (Fig. 5,A). Taking advantage of the converted AvrII site in the Jκ region of the miniloci constructs, rearrangements arising from the endogenous and transgene alleles could be distinguished by gel electrophoresis after AvrII or NcoI digestion of the PCR products (Fig. 5, A and B). Fig. 5,B illustrates a representative gel in which Vκ2 rearrangement was analyzed from spleen and bone marrow total DNA samples of wild-type and two F₁ transgenic mice progeny. These and other signals from the transgenic mice were quantified and normalized to the level of rearrangement of the corresponding Vκ region of the endogenous Igκ locus (Fig. 5,C). This analysis revealed that both Vκ2 and Vκ21 family members in transgenes were rearranged quite efficiently, exhibiting levels often equal to or greater than those exhibited by their endogenous counterparts (Fig. 5 C). In addition, Vκ2 but not Vκ21 genes in 5′EκML rearranged at a greater level than those in 5′DκML and κML constructs in most transgenic animals (p < 0.005).

Several investigations have revealed that germline Igκ transgenes often rearrange prematurely during B cell development and exhibit nongermline nucleotides (N regions) at Vκ-Jκ joints (34, 40, 63, 64). TdT is expressed during the pro-B stage of development when H chain gene segments undergo rearrangement, and the enzyme is involved in inserting N regions between V-D-J junctions (65). When Igκ genes undergo rearrangement later in B cell development, TdT activity is low and N regions at Vκ-Jκ junctions are rare (66). We cloned and sequenced Vκ2-Jκ1 and Vκ2-Jκ2 PCR products amplified from bone marrow DNA samples of two F₁ progeny of transgenic mice lines 57-2 and 32-2, which contained the 5′DκML and 5′EκML constructs, respectively. In 76 and 63 sequenced junctions amplified from rearranged miniloci in these two transgenic lines, N region sequences were only detected at two junctions, and one N region of 54 junctions was observed in the endogenous allele’s rearranged counterparts (Table II and data not shown). In addition, we found that about 40% of the junctions generated noninterrupted coding sequences from both transgenic and endogenous loci (data not shown). We conclude that the frequency of occurrence of N regions in our transgenes is similar to that exhibited by endogenous loci and thus transgenes do not appear to rearrange prematurely during B cell development.

The Igκ locus is not rearranged in T cells in normal mice, although sometimes corresponding transgenes disobey this tissue specificity rule (66, 67, 68). To assay for tissue specificity of rearrangement, we compared the ratios of transgene/endogenous gene rearrangement observed in thymus tissue with those seen in bone marrow and spleen using the above PCR assay. Fig. 6 shows representative results for nine transgenic animals. The trace of rearrangement seen in thymus for the transgenes and endogenous alleles exhibits the same ratio as that seen in B cells. This result was obtained for all transgenic lines (data not shown). Thus, the trace of rearrangement seen in thymus tissue can be attributed exclusively to B cell infiltration and contamination by hilar lymph nodes, and Igκ miniloci exhibit tissue-specific gene rearrangement.

We also determined the level of transcription from rearranged transgenes by a RT-PCR assay using Vκ2 and Vκ21 primers and a Cκ primer to amplify spliced transcripts (Fig. 7,A). RT-PCR products resisting digestion by BclI represent transcripts derived from rearranged miniloci. Fig. 7,B illustrates a representative gel in which cDNAs arising from rearranged Vκ2 genes were analyzed from spleen and bone marrow total RNA samples of wild-type and two F₁ transgenic mice progeny. These and other transgenic mice signals were quantified and normalized to the level of transcription of the corresponding rearranged Vκ region of the endogenous Igκ locus, as well as for their corresponding level of relative rearrangement (Fig. 7,C). In most F₁ transgenic mice, the level of transcription of the rearranged miniloci approached those of endogenous loci, with rearranged Vκ2 and Vκ21 of the miniloci transcribed at an average of 0.5- and 0.8-fold the level of corresponding endogenous counterparts. In addition, the variation in Igκ miniloci transcription levels between different transgenic lines indicated position effects. Interestingly, Vκ2 and Vκ21 of the miniloci in the same transgenic animals exhibited different sensitivities to position effects, for both the level of rearrangement (Fig. 5) and the level of transcription of rearranged genes (Fig. 7), suggesting that Vκ2 and Vκ21 chromatin accessibility for rearrangement and promoters for rearranged gene transcription are intrinsically different in responding to repressive effects of neighboring chromosomal environments. Despite the difference in the level of rearrangement, transcription of rearranged Vκ2 transgenes exhibited little difference between the three miniloci (p > 0.1). Similarly, the level of mRNA transcribed from rearranged Vκ21 transgenes was not distinguishable between the three constructs (p > 0.1).

To investigate further the transcription levels of transgenes, we generated hybridomas from two transgenic lines, 19-5-1 and 33-1, by fusing LPS-stimulated splenocytes to a non-Igκ chain producing myeloma cell line P3X63Ag8.653 (58). Positive clones in hypoxanthine/aminopterin/thymidine selection medium were screened by ELISA for Igκ producers using salmon sperm DNA as Ag because functionally rearranged Vκ2 family members are known to encode anti-DNA Abs (69). Those positive clones were further screened by a RT-PCR assay for Vκ2 gene expression, and a total of 25 and 14 clones were obtained (Table III). Further analyses revealed that either the transgene or the endogenous locus, but not both, had functionally rearranged Vκ2 genes in these samples (data not shown). The observed higher frequency of hybridomas that expressed the transgenes (Table III) is consistent with the observed higher levels of transgene rearrangement relative to their endogenous counterparts for the transgenic lines used in PCR analysis (Fig. 5). Northern analysis of RNA isolated from these hybridomas revealed that Igκ mRNA levels in clones expressing miniloci (Fig. 8, lanes 3–8) were similar to those expressing endogenous Igκ alleles (Fig. 8, lanes 9–14). These results are in agreement with the RT-PCR data and indicate that the transcription level of rearranged miniloci approaches that of rearranged endogenous alleles.

Previous transgenic experiments using germline human Igκ constructs containing a single Vκ gene segment showed very high levels of rearrangement with essentially every multiple copy of the transgenes exhibiting rearrangement in every B cell (40). Thus, these transgenes did not obey allelic exclusion. In contrast, the lower level of rearrangement exhibited by our Igκ miniloci is consistent with the possibility that their rearrangement may be excluded by the endogenous Igκ locus or other copies of the transgenes. To more directly evaluate allele usage for productive Vκ-Jκ joining in transgenic mice, we isolated Igκ-producing B lymphocytes from two selected transgenic lines using anti-Igκ biotinylated Abs and streptavidin-coated magnetic beads. FACS analysis revealed that the Igκ⁺ cell population was >95% pure (Fig. 9,A). To evaluate allele usage in the Igκ⁺ cell population, we performed Southern analysis to specifically detect the unrearranged 1.2-kb DraI/AvrII or 2.6-kb DraI/NcoI fragments of the endogenous allele (E) and the corresponding 3.6-kb DraI or 1.2-kb DraI/NcoI fragments of the transgene (T) (Fig. 9,B). Miniloci were less used for rearrangement than the endogenous alleles, because a higher percentage of transgenes existed in germline configuration in Igκ⁺ cell DNA compared with tail DNA (Fig. 9 C, e.g., compare lanes 5 and 6 with 7 and 8). These results provide evidence that our Igκ miniloci are responsive to allelic exclusion and prove that not all copies of our transgenes become rearranged in B cells.

Owing to the availability of a variety of forward- and reverse-genetic tools in yeast, YACs are a powerful resource for engineering large DNA constructs. Modifications can be as dramatic as large deletions or as subtle as single base changes. For example, we modified two parent YACs by means of gene replacement, chromosome fragmentation, and introduced single base changes in the minilocus. Given the observation that chimerism can approach 70% during the generation of YAC libraries due to a high frequency of recombination between cotransformed DNA fragments (70), we developed a novel strategy to create recombinant YACs (see Materials and Methods). In comparison to meiotic and mitotic recombination approaches, this method is less time-consuming and more efficient, especially for fusing YACs through heterologous sequences. Finally, we reasoned that an ideal Igκ minilocus should contain sequences always preserved in the native locus after Vκ-Jκ joining and should not exceed several hundred kilobases for ease of introduction into transgenic mice or cultured cells. As summarized in Table IV and discussed in detail below, our analysis of various parameters of transgene expression is the most complete yet reported in the literature, and the results are quite rewarding with respect to the levels of expression achieved.

Germline transcription from the miniloci was nearly proportional to transgene copy number in both bone marrow and spleen tissue from transgenic animals, being similar in level per locus copy to that of the endogenous alleles, except for mice carrying high copy (8 or 14) transgenes, which exhibited reduced levels of germline transcription (Fig. 4 C). Repeat-induced gene silencing has been observed in previous studies, which is possibly caused by a preference for integration into, or formation of, heterochromatin (71). Nevertheless, for the lower copy number transgenes, we can conclude that Igκ miniloci contain all the necessary regulatory elements to specify position-independent and copy number-dependent germline transcription at a level equivalent to that of the endogenous locus. In contrast, germline transcription of much shorter transgenes has been reported to be sensitive to position effects (72). Our miniloci contained 200–250 kb of sequence 5′ of the germline promoters and 25 kb of downstream sequence, including the two known enhancers and a MAR. It is possible that the lack of position effect is simply due to presence of these long stretches of DNA sequences that protect the germline promoters from the flanking chromosomal environment. However, transcription of rearranged Vκ21 genes exhibited a substantial variation between transgenic lines despite the fact that both upstream and downstream sequences remain in rearranged Vκ21 transgenes.

Transgene rearrangement exhibited significant position effects, suggesting that the miniloci lack insulator or chromosomal domain boundary sequences that would override adverse flanking chromosomal environments (73, 74). The observations that rearrangement but not germline transcription exhibited position effects indicates that these processes are not necessarily coupled. This is consistent with other results that have disconnected these events (11), suggesting that germline transcription per se is not sufficient to direct efficient rearrangement. Alternatively, chromatin accessibility may be regulated by mechanisms in addition to or other than germline transcription, and transcription directed from germline promoters may simply be a consequence of an open chromatin structure. In addition, transgene rearrangement was not proportional to locus copy number, which may be in part due to allelic exclusion.

Our results reveal that the 5′ and 3′ Vκ region genes, Vκ2 and Vκ21, are rearranged quite efficiently, up to several fold higher than their endogenous counterparts (Fig. 5,C). This rearrangement was tissue specific as it did not occur in T cells (Fig. 6). In addition, lack of N regions at Vκ-Jκ junctions suggests that transgenes were rearranged at the proper stage in B cell development (Table II). Interestingly, Vκ2 but not Vκ21 genes in 5′EκML rearrange more efficiently than those in 5′DκML and κML, suggesting that rearrangement enhancing sequences may reside within the additional 50-kb upstream region in 5′EκML (Fig. 5 C).

At least two models can be postulated to explain the regulation of Igκ locus rearrangement, assuming that allelic exclusion is equally obeyed for endogenous and transgene rearrangements. One model hypothesizes that a master regulator is present in the Igκ locus that specifies locus commitment to rearrangement. If the probability for commitment to rearrangement of a given locus is determined by a master regulator and that regulator is present in our miniloci, then the miniloci would have the same probability per locus copy for rearrangement as that of the endogenous alleles. There are only four functional Vκ genes in κML in comparison to 93 functional Vκ genes in an endogenous allele (23, 24, 25). Thus, the master regulator model would roughly predict per allele that a certain Vκ gene segment in the miniloci would have 93/4 or a 23-fold greater chance for rearrangement than the endogenous counterpart, assuming all Vκ regions exhibit an equal probability for rearrangement (for discussion purposes only). Alternatively, another model proposes that each Vκ gene determines its own rearrangement potential. In this case, the more Vκ regions that an Igκ locus contains, the greater the probability that this locus would exhibit rearrangement during B cell differentiation. This “V gene number” theory is consistent with the observed κ/λ ratio in the mouse (20:1) and human (6:4). If Vκ gene number determines the probability for locus rearrangement, then each endogenous locus would have 93/4 or a 23-fold greater chance for rearrangement relative to our transgene, assuming all Vκ regions exhibit an equal probability for rearrangement (for discussion purposes only). A certain Vκ gene segment in the miniloci would have the same chance for rearrangement as compared with the endogenous counterpart. PCR rearrangement data strongly support the “V gene number” model because the same Vκ genes in both transgenes and the endogenous loci exhibited very similar levels of rearrangement in most cases, except Vκ2 in 5′EκML (Fig. 5,C). In addition, genomic Southern analyses of Igκ alleles in κ-producing splenic cells reveals that <10% of the transgenes had undergone rearrangement (limits of sensitivity) (Fig. 9), consistent with their responsiveness to allelic exclusion. Previous transgenic experiments using germline human Igκ constructs containing a single Vκ gene segment showed very high levels of rearrangement, supporting the master regulator model. However, these transgenes did not obey allelic exclusion and therefore were aberrantly expressed (40). It is also possible that locus activation is determined not only by the number of Vκ gene segments but also by the strength of their promoters for specifying germline transcription (75, 76, 77, 78).

Previous studies employing much shorter rabbit or human germline Igκ constructs containing the intronic enhancer have observed a dramatic disproportionality between the levels of rearrangement and transcription (40, 67, 72). In our studies, rearranged Vκ2 and Vκ21 regions of the miniloci were transcribed at levels approaching those of their endogenous gene counterparts, although they still exhibited strong position effects. Northern analysis of rearranged Vκ2-expressing hybridomas generated from transgenic mice also revealed high-level transcription of rearranged miniloci. It is significant that this relatively efficient level of transcription has never been observed in previous transgenic studies employing shorter germline Igκ gene constructs (Table IV), indicating that three Igκ miniloci analyzed in our studies contain essentially all regulatory elements to confer copy number-dependent, position-independent germline transcription, tissue and developmental stage-specific efficient Vκ-Jκ rearrangement, and rearranged Igκ gene transcription. These features make these miniloci useful reagents for future functional analyses of regulatory elements in the locus. Although 5′EκML and 5′DκML differ by having either 65 kb or only 0.1 kb of sequence upstream of the 5′ Vκ2 gene, it is surprising that the only difference observed between these constructs was the level of recombination of the Vκ2 gene segments. Finally, because the Igκ miniloci still exhibit position effects, it is possible that they are missing domain boundary elements and a locus control region (73, 74, 79, 80). Nevertheless, our analysis places limits on where these hypothetical elements might be found, within a 15-kb region between the RS and the downstream ribose-5-phosphate isomerase gene (49), and/or >65 kb upstream of Vκ2 (70/3) gene.