Selective inhibition of class switching to IgG and IgE by recruitment of the HoxC4 and Oct-1 homeodomain proteins and Ku70/Ku86 to newly identified ATTT cis-elements.

Immunoglobulin (Ig) class switching is central to the maturation of the antibody response as IgG, IgA, and IgE are endowed with more diverse biological effector functions than IgM. It is induced upon engagement of CD40 on B lymphocytes by CD40L expressed by activated CD4+ T cells and exposure of B cells to T cell-secreted cytokines including interleukin-4 and transforming growth factor-beta. It begins with germ line IH-CH transcription and unfolds through class switch DNA recombination (CSR). We show here that the HoxC4 and Oct-1 homeodomain proteins together with the Ku70/Ku86 heterodimer bind as a complex to newly identified switch (S) regulatory ATTT elements (SREs) in the Igamma and Iepsilon promoters and downstream regions to dampen basal germ line Igamma-Cgamma and Iepsilon-Cepsilon transcriptions and repress CSR to Cgamma and Cepsilon. This mechanism is inactive in the Calpha1/Calpha2 loci because of the lack of SREs in the Ialpha1/Ialpha2 promoters. Accordingly, in resting human IgM+IgD+ B cells, HoxC4, Oct-1, and Ku70/Ku86 can be readily identified as bound to the Igamma and Iepsilon promoters but not the Ialpha1/Ialpha2 promoters. CD40 signaling dissociates the HoxC4.Oct-1. Ku complex from the Igamma and Iepsilon promoter SREs, thereby relieving the IH-CH transcriptional repression and allowing CSR to unfold. Dissociation of HoxC4.Oct-1. Ku from DNA is hampered by CD153 engagement, a CD40-signaling inhibitor. Thus, these findings outline a HoxC4.Oct-1. Ku-dependent mechanism of selective regulation of class switching to IgG and IgE and further suggest distinct co-evolution and shared CSR activation pathways in the Cgamma and Cepsilon as opposed to the Calpha1/Calpha2 loci.


Immunoglobulin (Ig) class switching is central to the maturation of the antibody response as IgG, IgA, and IgE are endowed with more diverse biological effector functions than IgM. It is induced upon engagement of CD40 on B lymphocytes by CD40L expressed by activated CD4 ؉ T cells and exposure of B cells to T cell-secreted cytokines including interleukin-4 and transforming growth factor-␤. It begins with germ line I H -C H transcription and unfolds through class switch DNA recombination (CSR). We show here that the HoxC4 and Oct-1 homeodomain proteins together with the Ku70/Ku86 heterodimer bind as a complex to newly identified switch (S) regulatory ATTT elements (SREs) in the I␥ and I⑀ promoters and downstream regions to dampen basal germ line I␥-C␥ and I⑀-C⑀ transcriptions and repress CSR to C␥ and C⑀.
This mechanism is inactive in the C␣1/C␣2 loci because of the lack of SREs in the I␣1/I␣2 promoters. Accordingly, in resting human IgM ؉ IgD ؉ B cells, HoxC4, Oct-1, and Ku70/ Ku86 can be readily identified as bound to the I␥ and I⑀ promoters but not the I␣1/I␣2 promoters. CD40 signaling dissociates the HoxC4⅐Oct-1⅐Ku complex from the I␥ and I⑀ promoter SREs, thereby relieving the I H -C H transcriptional repression and allowing CSR to unfold. Dissociation of HoxC4⅐Oct-1⅐Ku from DNA is hampered by CD153 engagement, a CD40-signaling inhibitor. Thus, these findings outline a HoxC4⅐Oct-1⅐Ku-dependent mechanism of selective regulation of class switching to IgG and IgE and further suggest distinct co-evolution and shared CSR activation pathways in the C␥ and C⑀ as opposed to the C␣1/C␣2 loci.
In class switching, the Ig constant H chain (C) region is substituted with C␥, C␣, or C⑀. Class-switched Igs are a critical feature of the high affinity late and memory antibody responses, because IgG, IgA, and IgE carry out those effector functions including binding to receptors for the Fc portion of ␥, ␣, and ⑀ H chains on phagocytic and proinflammatory cells and passage through "mucosae" that are required for the eradication of microbial pathogens. In general, Ig class switching occurs in germinal center (GC) 1 B cells upon stimulation by CD40L and cytokines expressed by activated CD4 ϩ T cells. It is initiated by transcription of the intervening (I), switch (S), and C regions of the upstream (donor) and downstream (acceptor) C H loci (1)(2)(3). Splicing of the nascent I H -S-C H RNA yields germ line I H -C H transcripts and probably couples transcription with the cleavage step of CSR, possibly through the action of activation-induced cytidine deaminase (AID) (4). CSR is effected through double-strand DNA breaks of the donor and acceptor S regions, excision of intervening DNA as a switch circle (SC) (2), and non-homologous end-joining (NHEJ) of the DNA ends (1)(2)(3) coupled with mismatch repair (5).
Germ line I H -C H transcription is driven by the I H promoter lying upstream of each I H region (6 -8) upon CD40L-, IL-4-, and/or transforming growth factor-␤-induced binding of the NF-B/c-Rel, Stat6, Smad, activating protein-1, or B cell lineage-specific activator protein transcription factors to I H promoter-specific cis-elements (1, 6, 8 -13). In CD40-induced (GC) B cells, germ line I H -C H transcription and CSR can be effectively down-regulated by bidirectional CD30:CD153-dependent signaling, which interferes with the recruitment of tumor necrosis factor receptor-associated factor molecules to CD40 and inhibits NF-B activation (14,15). However, the regulation of germ line I H -C H transcription and CSR in general and in non-CD40-induced (pre-GC) B cells in particular remains to be defined.
We report here a mechanism of selective inhibition of class switching to IgG and IgE. This mechanism relies on the binding of HoxC4, Oct-1, and Ku70/Ku86 to newly identified SREs 1 The abbreviations used are: GC, germinal center; AID, activationinduced cytidine deaminase; C H , heavy chain constant region; ChIP, chromatin immunoprecipitation assay; CHX, cycloheximide; CSR, class switch DNA recombination; HD, homeodomain; HIM, HD interaction motif; I H , intervening region of C H gene; Ig, immunoglobulin; NHEJ, non-homologous end-joining; S, switch (region); SC, switch circle; SRE, S-regulatory ATTT element; V H DJ H , variable/diversity/joining regions of heavy chain gene; IL, interleukin; GFP, green fluorescent protein; mAb, monoclonal antibody; GST, glutathione S-transferase; Ab, antibody; RT, reverse transcriptase; EMSA, electrophoretic mobility shift assay; DBB, DNA binding buffer; DAC, DNA affinity chromatography; mut, mutated; WT, wild type; MALDI-TOF, matrix-assisted laser desorption ionization time-of-flight; RE, responsive element; ht, human trimeric; Stat, signal transducer and activator of transcription. that exists in the I␥ and I⑀ but not in the I␣1/I␣2 promoters. Such a binding dampens basal germ line I␥-C␥ and I⑀-C⑀ transcription and represses CSR to C␥ and C⑀. It is potentiated by CD153 signaling and reversed by CD40 signaling . Thus, by  selectively binding to the I␥ and I⑀ promoters, HoxC4, Oct-1,  and Ku70/Ku86 differentially regulate class switching to IgG,  IgE, and IgA and would minimize S region double-stranded  DNA breaks, thereby contributing to the stability of the Ig H  chain locus. EXPERIMENTAL PROCEDURES Human C␥, C␣, and C⑀ Loci Sequence Analysis-S repeats and ATTT motifs were identified within the I H promoters, S regions, and flanking areas using the Pustell dot-matrix comparison (16) and nucleotide subsequence search (MacVector TM , Accelrys, Burlington, MA).
Transfection and Gene Reporter Assays-CL-01 B cells were transfected with firefly luciferase gene reporter pGL3 vector (10 g) and Renilla luciferase gene control pRL-CMV vector (10 ng) (Promega). 7D7 and 4B6 B cells were transfected with the same reporter and control vectors together with expression vector(s) (2 g) as specified. Electroporation (525 V/cm, 950 microfarads) was performed in duplicates using a Gene Pulser II (Bio-Rad). Firefly and Renilla luciferase activities were measured as reported previously (8) at the specified times using a MLX microtiter plate luminometer (Thermo Labsystems, Chantilly, VA).
Electrophoretic Mobility Shift Assays (EMSAs) and Protein Purification-Cytoplasmic and nuclear protein extraction, probe labeling, EMSA, and supershift reactions were performed as reported previously (8). [␥-32 P]ATP-labeled and cold I H SRE double-stranded oligonucleotides encompassed the sequences depicted in Fig. 1B. The sequences of the S␥3 and S⑀ SRE oligonucleotides were 5Ј-CAGCGGCAGACCA-GAAATGGGG-3Ј and 5Ј-GGGTTGGGGTGATTTAAACTGAGT-3Ј, respectively, encompassing the ATTT motif immediately 5Ј of the S␥3 region and the first ATTT motif of the S⑀ region. The sequence of the Sp1 oligonucleotide was 5Ј-ATTCGATCGGGGCGGGGCGAGC-3Ј. The SRE binding activity of CL-01 nuclear extracts (75 mg) was first enriched by 20 -50% (NH 4 ) 2 SO 4 precipitation, dialyzed against DNA binding buffer (DBB) (10 mM Tris-HCl, pH 7.6, 200 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, 10% glycerol), and then fractionated on a Centricon concentrator using a 100-kDa molecular mass cut-off membrane (Millipore Corp., Bedford, MA). The Ͼ100-kDa fraction was applied to a Superose 6 gel filtration column (Amersham Biosciences), which was eluted at a 100 l/min rate with DBB containing 20% glycerol. Fractions (500 l) were collected and tested for SRE binding activity by EMSA. The SRE binding fractions were pooled and loaded onto a DAC column consisting of an agarose matrix bearing streptavidin (Pierce) and loaded with 5Ј-biotinylated pentamerized double-stranded I␥3 5Ј-SRE or mutSRE oligonucleotides (Qiagen Sciences, Germantown, MD). The column was washed with 200 mM KCl DBB containing 20% glycerol. The DNA-bound proteins were eluted using a 300 -800 mM KCl gradient in DBB 20% glycerol and collected in 500-l fractions, which were monitored for SRE binding activity and protein content. Proteins were visualized in SDS-PAGE using the Rapid Silver Stain Plus kit (Bio-Rad). The fractions eluted at 300 mM KCl contained the strongest SRE binding activity and were pooled, concentrated on Centricon filters, and applied to SDS-PAGE. The resolved protein bands were stained by Coomassie Brilliant Blue G-250 (Bio-Rad), excised, and subjected to in-gel proteolysis by trypsin. Peptides mixtures were separated by a C8 reverse-phase column with a linear 0 -60% CH 3 CN gradient in 0.1% trifluoroacetic acid for 1 h. The molecular masses of the peptide mixtures were determined by MALDI-TOF mass spectrometry using a Voyager-DE short tandem repeat (PerSeptive Biosystems, Inc., Framingham, MA). The complete list of accurately measured masses of the tryptic peptides was used to search for protein candidates in the OWL protein sequence data base with the program ProFound (prowl.rockefeller.edu/cgi-bin/ProFound). Internal sequencing of tryptic peptides was performed as described previously (25).
Detection of Reciprocal Recombination SCs-Genomic DNA was extracted from B cells using the QIAmp DNA mini kit (Qiagen). Specific S␥-S, S⑀-S, and S␣-S reciprocal SCs were amplified from genomic DNA (500 ng) using nested PCRs and S, S␥-(␥1-4), and S␣-(␣1/␣2) region-specific primers (17). PCR kinetics entailed a 1-min denaturation at 94°C, a 1-min annealing at 68°C, and a 4-min extension at 72°C for two rounds of 30 cycles. Before each PCR, DNA was denatured for 5 min at 94°C. The identity of PCR-amplified DNA was confirmed by Southern blot analysis using S-specific SC probes (17).
Phosphatase Treatment of Nuclear Extracts-Extracts prepared from freshly isolated human IgM ϩ IgD ϩ B cells were incubated with 5 and 20 milliunits of acid phosphatase (Grade I, Roche Applied Sciences) in DBB in a final volume of 30 l at 25°C for 10 min. A 10-l aliquot of the phosphatase-treated sample was then tested in each binding assay.
GST Proteins and Pull-down Assays-GST fusion proteins were expressed, purified using GSH-agarose beads according to the manufacturer's protocol (Amersham Biosciences), and analyzed for homogeneity by SDS-PAGE and silver staining. L-[ 35 S]Methionine-labeled proteins were translated using the TnT Quick coupled transcription/translation systems (Promega) method. For pull-down experiments, 5 l of in vitro translated protein was mixed with 50 g of nuclear extract and applied to GSH-agarose beads (20 l) equilibrated in binding buffer B (25 mM Tris-HCl, pH 7.9, 1 mM dithiothreitol, 150 mM NaCl, 0.01% Nonidet P-40). After 2 h at 4°C, the beads were washed with buffer containing 150 mM KCl. Bound proteins were then eluted in SDS sample buffer, separated in SDS-PAGE, fixed, dried, and autoradiographed at Ϫ70°C.
Co-immunoprecipitation, Circular DNA Pull-down, and Immunoblotting Assays-Nuclear extracts (200 g of protein in 500 l of buffer B) were precipitated with indicated mAbs and Protein-G Plus-agarose (Santa Cruz Biotechnology, Inc.) or with CNBr-activated-Sepharose 4B beads (Amersham Biosciences) bearing circular pBSK vector (1 g of DNA/20 l of bead) (Stratagene Corp.). For circular DNA pull-down assay, buffer B was supplemented with 0.05 mg/ml poly(dI-dC)⅐poly(dI-dC) (Sigma) and WT or mutSRE oligonucleotides (50 ng). Pulled-down proteins were fractionated through 12% SDS-PAGE and transferred to nitrocellulose membranes. After blocking, these membranes were blotted with mAbs to HoxC4, Oct-1, Ku70, and Ku86, washed, and then incubated with a horseradish peroxidase-conjugated rabbit Ab to mouse IgG (Santa Cruz Biotechnology, Inc.). After horseradish peroxidase addition, the specific proteins were visualized using an enhanced chemiluminescence detection system (Amersham Biosciences). The whole cell extracts from tonsillar B cells were transferred to polyvinylidene difluoride membranes, blotted with mAbs to HoxC4, Ku70, Ku86, rabbit IgG to Oct-1, and actin (Sigma), washed, and then incubated with horseradish peroxidase-conjugated donkey Ab to mouse IgG or horseradish peroxidase-conjugated goat Ab to rabbit IgG (Santa Cruz Biotechnology, Inc.).

RESULTS
ATTT SREs Critically Modulate I␥ and I⑀ but Not I␣ Promoter Activity-We have suggested the existence of a repression of germ line I␥3-C␥3 transcription in resting human IgM ϩ IgD ϩ B cells (8) and tentatively ascribed it to I␥3 promoter tetrameric ATTT SREs. In the I␥1, I␥2, I␥3, and I␥4 promoters, such SREs exist in two identical copies, one straddling the 5Ј boundary of the IL-4-responsive element Stat6binding site (IL-4 RE-Stat6) and the other straddling the 3Ј boundary of the IL-4 RE-Stat6. Two SREs were also identified in the I⑀ promoter, both of them 3Ј of the IL-4 RE-Stat6 (Fig.  1A). Additional ATTT motifs exist in the downstream I H -S regions and flanking areas: 3, 3, 2, 3, and 13 in the ␥1, ␥2, ␥3, ␥4, and ⑀ loci, respectively. No SRE was found in the I␣1/I␣2 promoters or I␣-S␣ regions.
Transfection of human IgM ϩ IgD ϩ CL-01 B cells, our model of inducible CSR and somatic hypermutation (8,17,19), with I␥ or I⑀ promoter-driven luciferase gene reporter pGL3 vectors containing WT or mutated (ATTT to ggTT) mutSRE(s) was performed to analyze the role of these SREs in I H promoter activity (Fig. 1B). Mutation of both the 5Ј-and 3Ј-SREs (double mutSRE) in the I␥ and I⑀ promoters resulted in a 10-(I␥2) to 15-fold (I⑀) increase of basal reporter gene transcription and a 4-(I␥2) to 6-fold (I⑀) increase of human trimeric (ht) CD40Linduced transcription (Fig. 2). This possibly reflects the suboptimal induction of transfected CL-01 cells by soluble CD40L as underscored by our previous findings that similar culture conditions induce switching in approximately one-third of CL-01 cells (17). The double mutSRE also reverted the ability of CD153 signaling to reduce the basal (no reduction in double mutSRE versus a 25-40% reduction in WT SRE) and htCD40Linduced I␥ and I⑀ promoter-driven reporter gene transcription (no reduction in double mutSRE versus a 79-84% reduction in WT SRE). Mutation of the 5Ј-SRE alone resulted in enhancement of basal and htCD40L-induced I␥ or I⑀ promoter-driven transcription that was 28 -56% lower than that of the double mutSRE (data of supplemental Fig. 2A versus those of Fig. 2). Finally, the absence of the ATTT SRE in I␣1/I␣2 was associated with a significantly higher basal activity of this promoter as compared with the I␥1, I␥2, I␥3, I␥4, and I⑀ promoters (supplemental Fig. 2B). Thus, the ATTT SREs critically mediate basal and CD153-induced inhibition of I␥ and I⑀ promoterdriven transcription.
I␥ and I⑀ SREs Recruit HoxC4, Oct-1, and Ku70/Ku86 -EMSAs were performed using radiolabeled oligonucleotides encompassing the I␥3 or I⑀ 5Ј-SRE, the SRE immediately 5Ј of the S␥3 region (S␥3 SRE) or the most 5Ј-SRE in the S⑀ region (S⑀ SRE), and nuclear extracts from freshly isolated human peripheral blood IgM ϩ IgD ϩ B cells to characterize the transfactors specifically binding to the identified I␥ and I⑀ SREs. Two specific and closely migrating SRE-protein complexes (complexes A and B) were identified by all of the four probes (Fig. 3A). The formation of such complexes was inhibited by cold WT but not mutSRE (ATTT to ggTT) oligonucleotides containing the I␥1, I␥2, I␥3, I␥4, I⑀, S␥3, or S⑀ 5Ј-or 3Ј-SRE (data not shown). To analyze the composition of complexes A and B, nuclear extracts from CL-01 IgM ϩ IgD ϩ B cells were subjected to sequential (NH 4 ) 2 SO 4 precipitation, gel filtration, and SRE DNA affinity chromatography (DAC), which eventually yielded proteins of 100, 89, 72, and 34-kDa apparent molecular masses (Fig. 3, B-C). These proteins were identified as Oct-1, Ku86, Ku70, and HoxC4 by in-gel trypsin digestion, peptide mass fingerprinting, internal sequencing, and specific mAbs in eluates from a DAC column bearing SRE I␥3 oligonucleotides (Fig. 3D, left panel). Additional proteins of 190, 144, and 120-kDa apparent molecular masses were also detected by silver staining, but they accounted for bands of minor intensity and were not sequenced (Fig. 3C). The binding of HoxC4, Oct-1, and Ku70/Ku86 to DNA was mediated specifically by the SRE and was not attributed to "stickiness" of these proteins for free DNA ends as shown by the following: (i) the failure of HoxC4, Oct-1, and Ku70/Ku86 to bind to a mutSRE DAC column (Fig. 3D, left panel); (ii) the efficient pull-down of HoxC4, Oct-1, and Ku70/Ku86 by beads bearing circular pBSKII plasmid DNA containing 29 copies of the ATTT motif; and (iii) the inhibition of this pull-down by 100-fold molar excess of WT but not mutSRE oligonucleotides (Fig. 3D, right  panel). Thus, whether in the I H promoter or S area context, the ATTT SREs recruit the HoxC4 and Oct-1 HD proteins together with the Ku70/Ku86 heterodimer in a sequence-specific and DNA end-independent fashion.
HoxC4, Oct-1, and Ku70/Ku86 Form a DNA-binding Complex in B Cell Nuclei-The nature of the SRE-binding proteins was further verified by EMSAs utilizing nuclear extracts from freshly isolated IgM ϩ IgD ϩ B cells, radiolabeled I␥3 and I⑀ 5Ј-SRE as well as S␥3 and S⑀ SRE probes, and specific anti-HoxC4, anti-Oct-1, anti-Ku70, anti-Ku86, and anti-Ku70/Ku86 mAbs. Anti-Ku70, anti-Ku86, and anti-Ku70/86 mAbs but not control HoxC8-or Oct-2-specific Abs-supershifted complex A, whereas anti-HoxC4 and anti-Oct-1 mAbs inhibited the formation of complex B (Fig. 4A). Comparable results were obtained in EMSAs utilizing I␥3 and I⑀ 3Ј-SRE probes as well as I␥1, I␥2, or I␥4 5Ј-and 3Ј-SRE probes (data not shown). Pull-down experiments using GST-HoxC4, GST-Oct-1, and GST-Ku70 fusion proteins, GSH-agarose beads, and in vitro translated 35 Slabeled HoxC4 and Oct-1 proteins premixed with freshly isolated IgM ϩ IgD ϩ B cell nuclear extracts showed that HoxC4, Oct-1, and Ku effectively interact with one another in B cell nuclei. Both 35 S-labeled HoxC4 and 35 S-labeled Oct-1 bound to GST-HoxC4, GST-Oct-1, and GST-Ku70, indicating a significant self-association among HoxC4 and Oct-1 proteins as well  35 S-labeled HoxC4 and Oct-1 bind to immobilized GST-HoxC4, GST-Oct-1, and GST-Ku70 but not GST or GST-Ku70 mutHIM fusion proteins. C, detection of nuclear HoxC4-Ku70/Ku86, Oct-1-Ku70/86, and HoxC4-Oct-1 complexes in CL-01 B cell nuclear extracts by co-immunoprecipitation and Western blotting. D, in vivo binding of HoxC4, Oct-1, and Ku70/Ku86 to I␥ and I⑀ but not I␣ promoter DNA as assessed by ChIP assay. Cross-linked chromatin was precipitated from freshly isolated IgM ϩ IgD ϩ B cells by Abs specific for human HoxC4, Oct-1, or Ku70/Ku86. The precipitated DNA was specified by PCR using the I␥3, I⑀, and I␣ promoter primers. One of three ChIP assays is shown yielding comparable results. as direct physical interaction between HoxC4 and Oct-1, HoxC4 and Ku70, and Oct-1 and Ku70 (neither 35 S-labeled HoxC4 nor 35 S-labeled Oct-1 bound to GST alone) (Fig. 4B). Consistent with the critical role of the C-terminal HIM (24) in Ku70 binding, GST-Ku70 mutHIM , a GST fusion protein encoding Ku70 lacking its HIM, reacted with neither 35 S-labeled HoxC4 nor 35 S-labeled Oct-1. Also, a mAb that specifically recognizes the Ku70/Ku86 heterodimer interface (26) co-precipitated HoxC4 and Oct-1 from CL-01 nuclear extracts, and an anti-Oct-1 mAb co-precipitated HoxC4, Ku70, and Ku86 (Fig.  4C), indicating that Ku interacts with these HD proteins in B cell nuclei. Finally, ChIP assays in which the I␥ and I⑀ promoter sequences were specified in the DNA that had been precipitated from freshly isolated IgM ϩ IgD ϩ B cells by anti-Oct-1 Ab, anti-HoxC4 mAb, or anti-Ku mAb demonstrated direct binding of HoxC4⅐Oct-1⅐Ku to the I␥ and I⑀ but not I␣ promoters (Fig. 4D). Thus, HoxC4, Oct-1, and Ku70/Ku86 can exist as discrete components of a HD-dependent nuclear complex and specifically bind to the I␥ and I⑀ promoters and S region DNA in vitro and in vivo.

cells (left and central panels) and CL-01 IgM ϩ IgD ϩ B cells (right panel). Arrowheads indicate complexes A and B as well as the supershifted complex A (AЈ). B, in vitro translated
CD40 and IL-4 Signaling Dissociates HoxC4, Oct-1, and Ku70/Ku86 from SRE in a Dephosphorylation-dependent Manner-If binding of HoxC4⅐Oct-1⅐Ku to the I H promoter SREs is responsible for the basal repression of germ line I H -C H transcription, then htCD40L-induced germ line I H -C H transcription and subsequent CSR should entail the dissociation of the HoxC4⅐Oct-1⅐Ku complex from SREs, and this dissociation should be prevented by physiological CD40-signaling inhibitors such as CD153 (15). EMSAs using I␥3 and I⑀ promoter 5Ј-SRE as well as 5Ј-S␥3 and S⑀ SRE probes showed that freshly isolated IgM ϩ IgD ϩ B cells cultured for 2 days with either htCD40L alone or htCD40L and IL-4 but not an agonistic anti-CD153 mAb or IL-4 alone decreased the level of SREbound HoxC4⅐Oct-1⅐Ku complexes A and B by Ͼ95% (Fig. 5A). Comparable results were obtained utilizing I␥3 and I⑀ 3Ј-SRE probes as well as I␥1, I␥2, or I␥4 5Ј-and 3Ј-SRE probes (data not shown). The htCD40L-induced dissociation of HoxC4⅐Oct-1⅐Ku from SREs was efficiently inhibited by cycloheximide, a protein synthesis inhibitor, or CD153 cross-linking, which has been shown to dampen germ line I H -C H transcription and repress CSR (15,27). It was concomitant with increase of nuclear HoxC4, Oct-1, and Ku70/Ku86 proteins (Fig. 5B), suggesting a posttranslational modification in the htCD40L-induced dissociation of HoxC4⅐Oct-1⅐Ku from SRE. Incubation of nuclear extracts from freshly isolated IgM ϩ IgD ϩ B cells with increasing amounts of acid phosphatase prior to the addition of the SRE probes and separation on native gel resulted in decreased SRE-binding by HoxC4⅐Oct-1⅐Ku (Fig. 5C). This was prevented by sodium phosphate, an inhibitor of acid phosphatase. Accordingly, pretreatment of B cells with okadaic acid, a Ser/Thr phosphatase inhibitor, efficiently abrogated the CD40-induced dissociation of HoxC4⅐Oct-1⅐Ku from the SRE (Fig. 5D).
To prove that in vivo activation of germ line I H -C H transcrip- tion and CSR, as that occurring in the GC of peripheral lymphoid organs, is associated with the CD40L-dependent dissociation of the HoxC4⅐Oct-1⅐Ku complex from the I H promoter, we sorted human tonsil B lymphocytes into four fractions representing sequential stages of differentiation as follows: IgD ϩ CD38 Ϫ naïve pre-GC B cells; IgD ϩ CD38 ϩ early centroblasts; IgD Ϫ CD38 ϩ centroblasts/centrocytes; and IgD Ϫ CD38 Ϫ memory B cells (17). Germ line I H -C H transcription is absent in pre-GC B cells, appears in early centroblasts, peaks in centroblasts/centrocytes, and is extinct in memory B cells, whereas mature V H DJ H -C␥ and V H DJ H -C⑀ transcripts appear as a result of downstream CSR in centrocytes and are consistently expressed in memory B cells (17). Accordingly, pre-GC and memory B cells exhibited a strong SRE binding activity, which was consistent with the lack of I␥-C␥ and I⑀-C⑀ transcription in these lymphocytes (Fig. 6). In contrast, early centroblasts and centroblasts/centrocytes, which harbored germ line I␥-C␥ and I⑀-C⑀ as well as mature V H DJ H -C␥ and V H DJ H -C⑀ (centroblasts/centrocytes only) transcripts, were devoid of HoxC4⅐Oct-1⅐Ku SRE binding activity. Consistent with the kinetics of in vitro induction by htCD40L and IL-4, CD38 ϩ B cells up-regulated HoxC4, Oct-1, Ku70, and Ku86 transcripts. Up-regulation of the HoxC4 transcripts was reflected in the up-regulation of the related proteins, whereas the Oct-1, Ku70, and Ku86 proteins were abundant prior to up-regulation of the respective transcripts. Thus, CD40 and IL-4 signaling, which promotes germ line I H -C H transcription and triggers CSR to C␥ and C⑀, dissociates the HoxC4⅐Oct-1⅐Ku inhibitory complex from the SREs of the I␥ and I⑀ promoters and 3Ј-flanking regions in vitro and in vivo.
Overexpression of HoxC4, Oct-1, and Ku70/Ku86 Represses CSR to C␥ and C⑀ but Not C␣-To prove that HoxC4, Oct-1, and Ku70/Ku86 critically repress germ line I H -C H transcription as well as CSR in the C␥ and C⑀ loci, we co-transfected 7D7 and 4B6 IgM ϩ IgD ϩ B cells, both CL-01 cell subclones selected for spontaneous switching to IgG, IgA, and IgE, with a pIRES2 expression vector containing nil or cDNA encoding HoxC4, Oct-1, Ku70, Ku70 mutHIM , and/or Ku86 together with the I␥3, I⑀, or I␣1/␣2 promoter-driven luciferase gene reporter pGL3 vector. Overexpression of HoxC4 or Oct-1 alone reduced only moderately the activity of co-transfected I␥3 or I⑀ promoters, whereas overexpression of both HoxC4 and Oct-1 or Ku70/ Ku86 reduced the I␥3 or I⑀ promoter activity by up to 85% but had no effect on basal I␣1/␣2 promoter activity (Fig. 7A). This was specific, because no inhibition could be measured when double mutSRE I␥3 and I⑀ promoters were used (supplemental Fig. 7A). The C-terminal HIM was critically required, because overexpression of Ku70 mutHIM and Ku86 or Ku70 mutHIM alone failed to inhibit I␥3 and I⑀ promoter-driven gene reporter transcription and ablated the HoxC4-or Oct-1-mediated inhibition of I␥3 and I⑀ promoter activity.
The inhibition of I␥ and I⑀ promoter-driven transcription by HoxC4, Oct-1, and Ku70/Ku80 reflects the ability of these trans-factors to effectively dampen endogenous germ line I H -C H transcription and repress CSR to C␥ and C⑀. 7D7 and 4B6 B cells were transfected with a bicistronic pIRES2 expression vector encoding GFP and HoxC4, Oct-1, Ku70, transcripts. An analysis of germ line and mature transcripts in the C␥1 (supplemental Fig. 7B), C␥2, or C␥4 loci yielded comparable results (data not shown). Thus, HoxC4, Oct-1, and Ku70/Ku86 effectively repress germ line I␥-C␥ and I⑀-C⑀ transcription and CSR to C␥ and C⑀ but not germ line I␣-C␣ transcription and CSR to C␣1/C␣2 in a fashion that is dependent on the HIM of Ku70. DISCUSSION We have defined here a novel mechanism that inhibits class switching to IgG and IgE but not IgA. By identifying the key elements of this inhibitory mechanism, we provide the first evidence for an important role of a Hox protein and a novel function for Oct-1 and Ku70/Ku86 in B cell differentiation. We show that HoxC4 and Oct-1 together with the Ku70/Ku86 heterodimer form a HD-dependent complex, which is recruited to ATTT motifs in the human I␥ and I⑀ promoters to dampen germ line I␥-C␥ and I⑀-C⑀ transcription and repress direct and sequential CSR to C␥ and C⑀. The HoxC4⅐Oct-1⅐Ku-dependent inhibitory mechanism is operational in the presence of AID that plays a critical role in CSR (28) and would provide a threshold of transcriptional and recombinational inertia that must be overcome for effective initiation and unfolding of CSR, thereby contributing to the homeostasis of the H chain locus.
Hox proteins are phylogenetically conserved HD-containing trans-factors that serve principally as transcriptional repressors. They modulate transcription by binding to the HD-specific ATTT/A core-motif (29,30) and critically regulate not only embryonic pattern formation, axis specification, and organogenesis (31) but also adult cellular processes including selective hematopoietic lineage differentiation and stem cell renewal (32). Oct-1 is a ubiquitous member of the POU (Pit, Oct, Une) family of transcription factors that regulates both general and cell type-specific genes (33) including V and C genes in the Ig locus (34). Ku70/Ku86, the ATP-dependent DNA helicase II subunits of the DNA-dependent protein kinase, serves as DNA end-binding and alignment factors in NHEJ DNA repair. NHEJ is critical not only in Ig V(D)J gene recombination and CSR but also in overall genome maintenance (3,35).
The HoxC4⅐Oct-1⅐Ku complex may function as a common effector in the modulation of IgG and IgE class switching at different stages of the B cell natural history. In pre-GC and perhaps memory B cells, the complex would maintain the basal to the IL-4 RE and displace Bcl6 from it, they would also dissociate HoxC4⅐Oct-1⅐Ku from the SREs, thereby lifting the inhibition off of the Ig H chain locus. This, together with the binding of NF-B to the CD40 RE, would allow I H -C H transcription and CSR to unfold. Engagement of CD153 by CD30 as expressed on activated suppressor T cells (15) inhibits not only the CD40-induced activation of NF-B/Rel but also the CD40-dependent dissociation of HoxC4⅐Oct-1⅐Ku from the SREs, thereby exerting a net repressive activity on I H -C H transcription and CSR.
repression of CSR in the C␥ and C⑀ loci. The partial overlap of the 5Ј-SRE with the IL-4 RE-Stat6 (I␥) or the proximity of these cis-elements (I⑀) would entail a complex regulation of the I␥ and I⑀ promoters, allowing for competition and/or interplay among the respective trans-factors. In GC B cells, CD40 engagement and exposure to IL-4 induce the binding of NF-B to the CD40 RE and binding of Stat6 to the IL-4 RE. This would result in Bcl6 displacement (36) and dissociation of HoxC4⅐Oct-1⅐Ku from the SREs, which would in turn lift the inhibition off of the Ig H chain locus and activate germ line I␥-C␥ and I⑀-C⑀ transcription and CSR to C␥ and C⑀ (Fig. 8). These processes are counteracted by a CSR inhibitory signal from B cell CD153 upon engagement by CD30 on suppressor T cells. Here, we demonstrate that CD153 signaling, which inhibits the CD40induced activation of NF-B (14,15), effectively prevents the CD40-dependent dissociation of HoxC4⅐Oct-1⅐Ku from SREs. Thus, in addition to repressing basal germ line I H -C H transcription and CSR in non-CD40-induced (pre-GC) B cells, HoxC4⅐Oct-1⅐Ku mediates the CD153-dependent inhibition of germ line I H -C H transcription and CSR in CD40-induced (GC) B cells.
Our findings suggest a role for a CD40 signaling-induced Ser/Thr phosphatase in the dissociation of the HoxC4⅐Oct-1⅐Ku complex from SREs. In vitro Ser/Thr-specific dephosphorylation of HoxC4⅐Oct-1⅐Ku abrogated their SRE binding, and pretreatment of freshly isolated B cells with okadaic acid prevented CD40-induced dissociation of HoxC4⅐Oct-1⅐Ku from SREs (Fig. 5D). A similar dephosphorylation-dependent regulation of Hox protein activity has been reported previously (37) in Drosophila melanogaster. The overexpression of HoxC4⅐Oct-1 and/or Ku70⅐Ku86 (Fig. 7) would probably overcome the dephosphorylating activity of endogenous phosphatase(s), thereby allowing for the expression of the CSR inhibitory activity by the repressor complex. Upon CD40-induced dissociation from SREs, the amount of HoxC4, Oct-1, and Ku proteins increased in cell nuclei (Fig. 5B), probably as a result of increased transcription and protein synthesis (HoxC4 and Oct-1) or cytoplasmic-to-nuclear translocation (Ku70/Ku86) (38), suggesting that these proteins play a role in later CSR-related events such as NHEJ (39) or regulate transcription of CSRrelated genes by binding other HD recognition sequences.
For efficient inhibition of I␥ and I⑀ promoter activity, HoxC4 and Oct-1 rely on the recruitment of the Ku70/Ku86 heterodimer. Overexpression of Ku70 mutHIM reverted HoxC4-and Oct-1-dependent I H promoter inhibition and enhanced basal promoter activity (Fig. 7, A and B), possibly through displacement of endogenous Ku70 and formation of Ku70 mutHIM /Ku86 heterodimers, which would effectively reduce the availability of the functional endogenous Ku70/Ku86 heterodimer but not associate with HD proteins. Importantly, forced expression of HoxC4 and Oct-1 as well as Ku70/Ku86 effectively repressed both germ line I␥-C␥ and I⑀-C⑀ transcription and CSR to C␥ and C⑀. Again, this repression was dependent on the ability of Ku70 to interact with HD proteins, because overexpression of the Ku70 mutHIM /Ku86 heterodimer failed to repress germ line I H -C H transcription and enhanced CSR to IgG and IgE as detected by the increased level of mature V H DJ H -C␥ and V H -DJ H -C⑀ transcripts and reciprocal SCs. The mechanism of Kudependent transcriptional inhibition was not addressed in this study. It may include DNA-dependent protein kinase-dependent phosphorylation of I H promoter-and/or S region-bound trans-factors (40), inhibition of histone acetyltransferases (41), or recruitment of histone deacetylases including Sir2-related proteins (42).
Inhibition of IgG and IgE class switching by HoxC4⅐Oct-1⅐Ku is consistent with the C␥and C⑀-shared CSR activation path-way, which includes CD40 and IL-4R signaling, and possibly reflects the co-evolution of the C␥ and C⑀ loci arising from the duplication of a single ancestral locus (43,44). The lack of ATTT motifs within the I␣1/I␣2 promoters accounts for the failure of HoxC4⅐Oct-1⅐Ku to inhibit IgA class switching and emphasizes the difference in regulation (transforming growth factor-␤ versus IL-4) and ancestral origin (early versus late) of the C␣1/C␣2 versus the C␥ and C⑀ loci. Also, it probably underlies the T cell CD40L independence and distinct anatomical compartmentalization of IgA-secreting cells, which are mainly CD5 ϩ (B-1 lymphocytes). B-1 lymphocytes accumulate preferentially in the splanchnic district and near external membranes and play a critical role in the first line of defense against microbial pathogens (45). Similar to B-1 cells, IgA appears early in phylogeny, emerging prior to IgG and IgE as the first of the "mature" isotypes in birds. The lack of regulation of IgA class switching by HoxC4⅐Oct-1⅐Ku may be compensated by other mechanisms. In the mouse, germ line I␣-C␣ transcription and S 3 S␣ CSR are repressed by B cell lineage-specific activator protein, which binds to a specific cis-element within the I␣ promoter (46) or by the late SV40 factor, which binds to appropriately spaced CTGG repeats within S and S␣ regions, thereby recruiting histone deacetylases and the co-repressor Sin3A (47).
The inhibitory activity unveiled here is probably part of a broader regulation of the Ig H chain locus by HoxC4⅐Oct-1⅐Ku.
Our preliminary experiments suggest that these trans-factors also regulate the human H chain 3Ј-hs1,2 enhancer element, which probably plays a role in Ig class switching. 2 Because of the wider recurrence of ATTT motifs in the human genome, HoxC4⅐Oct-1⅐Ku or other HD protein-Ku complexes could be involved in general transcriptional inhibition and anti-recombinogenic functions as part of the overall genomic caretaker activity as suggested by the extreme genomic instability of Ku70 Ϫ/Ϫ and Ku86 Ϫ/Ϫ mice (35). Dysregulation of the HoxC4⅐ Oct-1⅐Ku-mediated inhibitory function could cause aberrant CSR and chromosomal translocation and contribute to B cell lymphomagenesis.