The Ets factor Spi-B is a direct critical targetof the coactivator OBF-1Boris Bartholdy*†, Camille Du Roure*, Alain Bordon*, Dianne Emslie‡, Lynn M. Corcoran‡, and Patrick Matthias*§

*Friedrich Miescher Institute for Biomedical Research, Novartis Research Foundation, Maulbeerstrasse 66, CH-4058 Basel, Switzerland; and‡The Walter and Eliza Hall Institute of Medical Research, Parkville, Victoria 3050, Australia

Edited by Richard A. Flavell, Yale University School of Medicine, New Haven, CT, and approved June 15, 2006 (received for review December 6, 2005)

OBF-1 (Bob.1, OCA-B) is a lymphoid-specific transcriptional coacti-vator that associates with the transcription factors Oct-1 or Oct-2on the conserved octamer element present in the promoters ofseveral ubiquitous and lymphoid-specific genes. OBF-1-deficientmice have B cell-intrinsic defects, lack germinal centers, and haveseverely impaired immune responses to T cell-dependent antigens.Crucial genes that are regulated by OBF-1 and that might explainthe observed phenotype of OBF-1 deficiency have remained elusiveto date. Here we have generated transgenic mice expressing OBF-1specifically in T cells and examined these together with micelacking OBF-1 to discover transcriptional targets of this coactivator.Using microarray analysis, we have identified the Ets transcriptionfactor Spi-B as a direct target gene critically regulated by OBF-1 thatcan help explain the phenotype of OBF-1-deficient mice. Spi-B hasbeen implicated in signaling pathways downstream of the B cellreceptor and is essential for germinal center formation and main-tenance. The present findings establish a hierarchy between thesetwo factors and provide a molecular link between OBF-1 and B cellreceptor signaling.

B cell development � gene regulation � transcription factors �transgenic mice

The proline-rich, lymphoid-specific protein OBF-1 (Bob.1,OCA-B; see refs 1 and 2) forms a ternary complex with the

POU proteins Oct-1 or Oct-2 and DNA on the conserved octamerelement (5�-ATGCAAAT-3�). This motif is present in the promot-ers of a number of eukaryotic genes, including ubiquitously activegenes, such as H2B and U snRNA genes, but also genes that arerestricted to distinct cell types (reviewed in ref. 3). Notably, theoctamer element is present in all Ig heavy and light chain genevariable (V) region promoters as well as in the heavy chain intronicenhancer (4) and is critical for Ig gene transcription (3, 5–9). OBF-1strongly potentiates transcription from octamer-containing pro-moters such as the Ig � light chain promoter in transfection assaysand in vitro (1, 2, 10). In B cells containing an inducible OBF-1allele, it was found that the activity of octamer-dependent reportersdepended on OBF-1 (11). Yet, in vivo data from OBF-1�/� miceindicated that OBF-1 is dispensable for early antigen-independentB cell development and that the level of unswitched Ig mu genetranscription in mature B cells is unaffected in the absence of thiscoactivator (10, 12–14). OBF-1 appears to play a role in cell survivalat early B cell stages, and, in the absence of OBF-1, a reduction oftransitional B cells in the spleen was observed. This finding suggeststhat OBF-1 is important for production of transitional B cells in thebone marrow (15) or B cell homing from the bone marrow to thespleen (16).

Importantly, histochemical studies have shown that OBF-1function is essential for the formation of germinal centers (GC)in secondary lymphoid organs. As a consequence, OBF-1-deficient mice have defects in antigen-dependent B cell devel-opment and show a dramatically impaired immune response toT cell-dependent (TD) antigens (10, 13). In addition, it wasrecently found in an in vitro culture system that OBF-1 is criticalfor the final stages of antibody-secreting cell differentiation; inthe absence of OBF-1, the repressor protein Blimp1�PRDM1

fails to be induced, and downstream targets, such as Pax-5 orBcl6, are not down-regulated (17).

OBF-1 expression is largely restricted to B cells (10), but itcan also be induced in T lymphocytes by costimulation in vitro(18, 19). The phenotype observed in OBF-1-deficient miceclearly coincides with the expression pattern of the OBF-1 genein B cells, which peaks at two distinct time points in B celldevelopment: at the immature B cell stage in the bone marrowand with the highest expression levels in GCs and GC-derivedB cell lymphomas (20, 21). The increased amount of OBF-1protein in GC B cells is partly due to posttranscriptionalregulation (22, 23).

To date, known direct target genes depending on OBF-1 arethe BLR1 gene, which is regulated by NF-�B, OBF-1, and Oct-2in B cells (24), the CCR-5 gene in T cells (25), and the Bcell-specific B29 and mb1 genes (26). Recently, the Kcnn4promoter, the Lck distal promoter (27), and the Adh2-likepromoter (28) have been identified to be directly bound andactivated by OBF-1. However, the physiological relevance ofthese different target genes for GC formation and TD immuneresponses remains unclear. Here we present evidence that theEts factor Spi-B is a crucial target of OBF-1.

Spi-B is a member of the Ets family of transcription factors thatare characterized by the presence of a conserved DNA-bindingdomain (the Ets domain), a motif of 85 aa that forms a uniquehelix–loop–helix structure binding purine-rich DNA sequences (29,30). Spi-B is closely related to another Ets factor crucial forhematopoiesis, PU.1�Spi-1, through high structural homology andby its ability to transactivate in vitro target genes identical to thosetransactivated by PU.1 (31, 32). Spi-B also interacts with thecoactivator PU.1 interacting protein (Pip�IRF-4) (33). In vivo datashow, however, that the target genes for PU.1 and Spi-B overlaponly partially (34). Both PU.1 and Spi-B are required for normal Bcell receptor (BCR) signaling (34), but whereas Spi-B can replacePU.1 in myeloid development, it cannot replace PU.1 in lymphoiddevelopment (35). Spi-B-deficient mice exhibit severe abnormali-ties in B cell function and selective TD humoral immune responsesaccompanied by a defect in GC formation and maintenance (36).Spi-B-deficient mice have a BCR signaling defect and appear toinitiate the production of GCs within splenic primary B cell follicles,but these structures decay prematurely due to BCR-mediatedapoptosis. The Peyer’s patches in Spi-B-deficient mice, just as inOBF-1 deficient mice, are normal in appearance and cell numbers.

Here we show conclusively that OBF-1-deficient mice havestrongly reduced levels of Spi-B and that OBF-1 acts directly on

Conflict of interest statement: No conflicts declared.

This paper was submitted directly (Track II) to the PNAS office.

Freely available online through the PNAS open access option.

Abbreviations: BCR, B cell receptor; GC, germinal center; PMA, phorbol myristic acetate(phorbol 12-tetradecanoate 13-acetate12-O-tetradecanoylphorbol-13-acetate); Pn, pro-moter n; PNA, peanut agglutinin; TD, T cell-dependent.

†Present address: Division of Hematology�Oncology, Beth Israel Hospital, Harvard MedicalSchool, Boston, MA 02215.

§To whom correspondence should be addressed. E-mail: patrick.matthias@fmi.ch.

© 2006 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0509430103 PNAS � August 1, 2006 � vol. 103 � no. 31 � 11665–11670

IMM

UN

OLO

GY

Dow

nloa

ded

by g

uest

on

Nov

embe

r 30

, 202

0

the Spi-B promoter to enhance transcription of this Ets factor.These results establish a molecular hierarchy between these twotranscription factors, which are both crucial for the TD immuneresponse and GC formation.

ResultsGeneration and Phenotype of lck-OBF-1 Mice. OBF-1 expression waspreviously found to be mostly B cell-restricted but could beinduced in T cell lines and primary murine thymocytes after invitro stimulation with phorbol myristic acetate (PMA) andionomycin (18, 19). However, the physiological role of OBF-1 inT cells has remained unclear and in OBF-1�/� mice T celldevelopment and function appear normal (10, 20). To furtherdefine the potential function of OBF-1 in T cells, we generatedtransgenic mice expressing an HA epitope-tagged OBF-1 cDNAunder the control of the murine proximal lck promoter (see Fig.6A, which is published as supporting information on the PNASweb site). This promoter is highly active in thymocytes through-out their development and then shows low activity in peripheralT cells (37). In agreement with this activity pattern, high levelsof transgenic OBF-1 mRNA and protein were found in thethymus, and lower levels were found in the spleen (Fig. 7, whichis published as supporting information on the PNAS web site);by contrast, transgenic OBF-1 mRNA or protein were notdetectable in nonlymphoid organs of these transgenic lines (datanot shown). Several independent transgenic mouse lines weregenerated, which all showed the phenotype described below.

We first examined overall T cell development by flow cytom-etry by using antibodies against the T lineage surface markersCD4 and CD8. Total thymocyte numbers, as well as the numbersof double positive, double negative, or single positive CD4 orCD8 T cells were all found to be normal in these transgenic mice(Fig. 6B). Upon further characterization by FACS analysis, somespecific alterations in distinct T cell populations were observedin the thymus or in the periphery. For example, a subset ofCD4�CD8� thymocytes also coexpressed the IL-2 receptor�-chain (CD25; Fig. 8, which is published as supporting infor-mation on the PNAS web site). These CD4�CD8�CD25� cellswere found to express a higher level of OBF-1 than the CD25�

cells, thus establishing a link between CD25 expression andOBF-1 levels (Fig. 9, which is published as supporting informa-tion on the PNAS web site). These cells, however, do not appearto correspond to CD25� CD4� regulatory helper T cells basedon the absence of Foxp3 expression, and no elevated number ofCD25� cells was observed in the spleen (data not shown). Inaddition, other specific T cell populations such as �� anddendritic epidermal T cells were also influenced by OBF-1expression (Figs. 10 and 11, which are published as supportinginformation on the PNAS web site). Together, these resultsindicate that OBF-1 overexpression in the T cell compartmentleads to very specific effects on a number of minor T cell subsetsbut does not overtly perturb normal T cell development.

Because overall T cell development and total T cell numberswere largely unchanged by the ectopic expression of OBF-1, wedecided to use thymocytes RNA for microarray experimentsto identify potential OBF-1 target genes. To this end, twoindependent comparisons were performed: (i) transgenic vs.WT total thymocytes and (ii) transgenic thymocytes,CD4�CD8�CD25� vs. CD4�CD8�CD25� (Fig. 9). The resultsof these independent microarray experiments were combinedand allowed to identify overlapping subsets of genes that aremodulated by OBF-1 expression.

The Expression of Spi-B Is Proportional to OBF-1 Levels in MouseThymocytes and Reflects the Activity of the Octamer-ContainingPromoter 2 (P2). The lymphoid-specific Ets family transcriptionfactor Spi-B was identified as one interesting potential OBF-1target gene in the microarray analysis. Its elevated expression in

the transgenic mice was confirmed by Northern blot analysis andwas compared with the expression in OBF-1�/� thymocytes.Whereas the WT mice express low levels of Spi-B mRNA, theexpression of this gene is severely reduced in OBF-1�/� thymo-cytes (Fig. 1A). Thus, expression of Spi-B and OBF-1 mRNA aredirectly correlated in thymocytes in vivo.

The Spi-B gene is transcribed from two distinct promoters (38);the downstream promoter (P2) contains a perfect octamer site andis therefore a good potential target of coactivation by OBF-1. Theuse of an RNase protection probe that distinguishes between the

Fig. 1. Expression of Spi-B, from the octamer site containing P2, is propor-tional to OBF-1 levels. (A) The Spi-B mRNA level is proportional to the amountof OBF-1 expression in thymocytes. Thymocytes were isolated from OBF-1-deficient WT or lck-OBF-1 transgenic mice, and total RNA was analyzed byNorthern blotting with a Spi-B-specific probe (Left). The membrane wasrehybridized with a probe for �-actin as a control for loading. Quantificationby PhosphorImaging and normalization to the expression level of �-actin areshown (Right). (B) Schematic representation of the mouse Spi-B gene in thepromoter region. The structure of the mRNAs originating from P1 (upstream)or P2 (downstream) is indicated. Exon sequences are depicted as boxes. Thesequence surrounding the octamer site in P2 is shown at the bottom of B. (C)Spi-B P2 is under the control of OBF-1. An RNase protection assay with RNAfrom thymocytes of the indicated genotypes is shown (Left). A probe for S16was included as a control in the hybridization reaction, and its signal was usedfor the normalized expression presented (Right).

11666 � www.pnas.org�cgi�doi�10.1073�pnas.0509430103 Bartholdy et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 30

, 202

0

mRNA isoforms transcribed from the two different promotersrevealed that promoter 1 is equally active in cells of the differentOBF-1 genotypes, whereas P2 activity is strongly modulated inthymocytes lacking or overexpressing OBF-1 (Fig. 1C).

Transcription from Spi-B P2 Is Controlled by OBF-1. To verify theimportance of the octamer site in P2, we performed transienttransfection assays using luciferase reporter genes driven by theSpi-B gene P1 (P1-luc) or P2 (P2-luc); we also used variants of P2with different mutations in the octamer site. As shown in Fig. 2A,upon cotransfection of an OBF-1 expression vector in 293T cells,the activity of P1 remained unaffected, but the activity of P2 wasstrongly enhanced. In addition, a point mutation in the octamersite that abolishes binding of OBF-1 to the Oct-1�2–DNAcomplex (P2-S-luc) (39) strongly reduced promoter activation tothe same extent as a multiple mutation of the octamer site(P2-M-luc), which completely abrogates binding of Oct factors(Fig. 2 A). In addition, when the same reporters were transfectedin K46 B cells, which express OBF-1 and Spi-B, a stronglyreduced activity was observed for the mutated reporter, indi-cating that endogenous OBF-1 is able to bind to and activate theP2 promoter (Fig. 2B). EMSAs (Fig. 2C) provided additional invitro evidence for the formation of a ternary complex between

Oct-1 or the Oct-1 POU domain and OBF-1 on the octamer siteof P2 (lanes 9 and 15), which is visible as a supershifted complex,compared with the complex obtained with Oct-1 or POU-1 alone(lanes 6 and 12). These results demonstrate that the octamer sitein the Spi-B P2 is necessary for allowing OBF-1 binding in aternary complex with Oct-1 (or Oct-2) and that it mediatescoactivation by OBF-1.

OBF-1 Binds in Vivo on Spi-B P2. We next performed ChIP exper-iments. For these experiments, primary thymocytes from WT ortransgenic mice were isolated, chromatin-bound proteins werecross-linked with formaldehyde, and OBF-1 was immunopre-cipitated with an anti-HA antibody. As shown in Fig. 3A, theSpi-B P2 could be amplified by PCR in the precipitate fromtransgenic thymocytes but not from WT samples. The BLR-1promoter, which has been shown to be regulated by OBF-1 (24),was used as a positive control and was robustly amplified in theimmunoprecipitated transgenic sample. To establish a link be-tween expression of OBF-1 and Spi-B in B cells, ChIP experi-ments were first repeated with primary splenic B cells isolatedfrom WT mice. However, because OBF-1 is expressed only atlow levels in these cells (20) and the available anti OBF-1antibodies are not sensitive enough, no conclusive results wereobtained (data not shown). We next used Abelson virus-transformed pro-B cell lines established from WT or OBF-1�/�

mice. Unlike primary mouse splenic B cells, v-Abl transformedpro-B cells are homogeneous and express high levels of OBF-1(our unpublished data). In this case, ChIP analysis with ananti-OBF-1 antibody showed that the Spi-B P2 could be amplifiedin immunoprecipitates from WT but not from OBF-1�/� cells(Fig. 3B). In contrast, control primers in the coding region of theSpi-B gene failed to amplify a fragment (data not shown). Theseresults thus demonstrate unambiguously the presence of OBF-1on the Spi-B P2 in vivo.

Fig. 2. Critical role of the octamer motif for activation of the Spi-B P2. (A)Luciferase reporter constructs under the control of the Spi-B P1 or P2 weretransfected into 293T cells together with an empty expression vector (filledbars) or an OBF-1 expression vector (open bars). The histograms represent themean � SD of three independent experiments. (B) The reporter constructswere transfected into the K46 B cell line expressing endogenous OBF-1. Thehistograms represent the mean � SD of three independent experiments. (C)EMSA with a WT (P2) or mutated (P2-S, P2-M) octamer site DNA probe derivedfrom the Spi-B promoter and in vitro translated proteins. Either full-lengthOct-1 (lanes 4–9) or its isolated POU domain (lanes 10–15) was used forcomplex formation.

Fig. 3. In vivo OBF-1 is bound on the Spi-B P2. (A) Cross-linked chromatin wasisolated from WT or lck-OBF-1 thymocytes and immunoprecipitated with ananti-HA antibody. The precipitated material was used as a template for a PCRwith primers derived from Spi-B P2 or from the BLR1 promoter. (Top andMiddle) The precipitated material was used as a template for a PCR withprimers derived from Spi-B P2 or from the BLR1 promoter, as indicated.(Bottom) The panel labeled ‘‘input’’ shows PCR amplification with the chro-matin before immunoprecipitation. (B) Cross-linked chromatin was isolatedfrom WT or OBF-1�/� Abelson virus-transformed pro-B cells and immunopre-cipitated with antibodies against OBF-1. A PCR with primers derived fromSpi-B P2 is shown for the precipitate (Upper) and the input (Lower).

Bartholdy et al. PNAS � August 1, 2006 � vol. 103 � no. 31 � 11667

IMM

UN

OLO

GY

Dow

nloa

ded

by g

uest

on

Nov

embe

r 30

, 202

0

OBF-1 Is Essential for Spi-B Transcription in B Cells. It was shownrecently that transcription of Spi-B (and of its coactivator IRF-4)is induced upon treatment of v-Abl-transformed pro-B cells withthe Abelson kinase inhibitor STI571 (Gleevec); this treatmentelicits further differentiation of the cells, as characterized byincreased transcription of Rag-1 and Rag-2, Ig � gene rearrange-ment, and enhanced expression of CD25 at the cell surface (40).We therefore used v-Abl transformed pro-B cells to test whetherOBF-1 is essential for STI571-induced expression of the Spi-Bgene. For this, WT or OBF-1�/� pro-B cells were treated for 10 hwith STI571, and expression of Spi-B was measured by real-timePCR with primer pairs that discriminate mRNA originating frompromoters P1 or P2. As shown in Fig. 4A, Spi-B transcription fromP2 was efficiently induced in WT cells but not in cells lackingOBF-1. In addition, P1 was marginally induced by STI571 in WTcells but not in mutant cells, suggesting that OBF-1 may alsoinfluence moderately the activity of P1. This suggestion is notunreasonable, because the start sites of the two transcripts areonly �600 bp apart. Moreover, the expression of endogenousSpi-B from promoter P2 was reduced by 2-fold in mature splenicB cells lacking OBF-1, whereas no significant change in tran-scription from promoter P1 was found (Fig. 4B).

To further demonstrate the role of OBF-1 in induction of theSpi-B gene, a retroviral construct encoding an OBF-1�estrogenreceptor fusion protein was stably introduced into the OBF-1-deficient B cell lymphoma line BM4 (41). As shown in Fig. 4C,parental BM4 cells fail to express Spi-B mRNA; however, in thepresence of hormone, cells expressing the OBF-1�estrogen recep-tor fusion protein show strong Spi-B expression (lanes 1 and 2).

Spi-B Expression in Spleens of Immunized WT and OBF-1�/� Mice. Itwas shown previously that OBF-1�/� mice fail to form GCs uponchallenge with a TD antigen such as 2,4-dinitrophenyl-conjugated keyhole limpet hemocyanin (DNP-KLH) (10). Sim-

ilarly, Spi-B�/� mice have a severe defect in formation of GCs,which are only small and short-lived (36). To investigate whetherdown-regulation of Spi-B could explain the lack of GCs inOBF-1�/� mice, we examined, by in situ hybridization, theexpression of Spi-B in spleens of immunized mice. As can be seenin Fig. 5A, peanut agglutinin (PNA)-positive GCs could bedetected at day 10 after immunization in WT but not OBF-1�/�

mice. Accordingly, in these immunized samples, expression ofSpi-B mRNA was clearly detected in WT but not OBF-1-deficient spleens (Fig. 5A). As transient GC form in SpiB�/�

mice, we tested whether OBF-1-deficient mice might be able toform transient GCs and examined their formation at an earlytime point; as shown in Fig. 12, which is published as supportinginformation on the PNAS web site, small GCs could be detectedat day 6 in control mice but not in mice lacking OBF-1.

Because the in situ hybridization did not allow us to discrim-inate between GC and follicular B cells for Spi-B expression,these two cell fractions were isolated by FACS, and RNA wasused to determine the levels of Spi-B transcripts from eachpromoter. Restricted expression of AID (activation-inducedcytidine deaminase protein gene) to the GC B cell fractionconfirmed the identity of each sorted population (Fig. 5BLower). Spi-B expression from promoter P2 was observed in bothpopulations, albeit to a slightly reduced level in GC B cells (Fig.5B Upper). In contrast, expression from promoter P1 was dra-matically reduced in the GC fraction, indicating that OBF-1-dependent expression of Spi-B is critical in GC B cells (Fig. 5BUpper). Together, these data suggest that OBF1 is required forthe formation and maintenance of GCs, possibly by way of Spi-B.

DiscussionHere we present compelling evidence that the coactivator OBF-1is necessary for efficient transcription of the Spi-B gene in mouse

Fig. 4. OBF-1 is required for Spi-B expression in B cells. (A) OBF-1 is necessaryfor STI571-induced expression of Spi-B in Abelson pro-B cells. WT or OBF-1�/�

Abl pro-B cells were cultured in the presence (open bars) or absence (solid bars)of STI571 (10 nM) for 10 h, and total RNA was extracted. After cDNA synthesis,expression of the Spi-B gene was measured by real-time PCR, with primersspecific for RNA originating from P1 or P2. The histograms show the mean �SD of three independent experiments. (B) Mature B cells from OBF-1�/� micehave reduced levels of Spi-B mRNA originating from promoter P2. Real-timePCR was performed on sorted CD43�-depleted mature splenocytes of OBF-1�/� and OBF-1�/� mice. (C) Rescue of Spi-B transcription by expression ofOBF-1 in a mature OBF-1�/� B cell line. (Upper) The Northern blot shows theOBF-1-deficient BM4 parental cell line and two clones thereof expressing aninducible OBF-1�estrogen receptor fusion protein (lanes 1 and 2); all sampleswere treated with hormone before RNA isolation. (Lower) The blot wasreprobed with �-actin as a loading control.

Fig. 5. Absence of Spi-B expression in OBF-1�/� follicular B cells. (A) Ten daysafter immunization with 2,4-dinitrophenyl-conjugated keyhole limpet hemo-cyanin (DNP-KLH) splenic cryosections were prepared. Formation of GCs wasmonitored by staining of B cells with an anti-B220 antibody (green) togetherwith PNA (red). Spi-B expression was detected by immunohistochemistry witha specific riboprobe (antisense, blue), and nuclei were counterstained withNuclear Fast red (red). A control with a sense Spi-B probe is shown. (B) (Upper)Spi-B expression in GC and follicular B cells of immunized WT mice. Splenic GC(GL7� PNA� B220�) and non-GC (GL7� PNA� B220�) B cells of WT mice wereFACS-sorted 10 days after immunization, and Spi-B mRNA expression wasmeasured by semiquantitative RT-PCR with primers specific for RNA originat-ing from P1 or P2. GAPDH expression was used for normalization. Threeexperiments were performed with identical results. (Lower) RT-PCR analysis ofAID expression, a GC-specific enzyme, demonstrated the purity of the sortedGC B cell population (lane 1) compared with the non-GC B cells (lane 3).Nontemplate controls (without reverse transcription) are shown in lanes 2 and4, respectively.

11668 � www.pnas.org�cgi�doi�10.1073�pnas.0509430103 Bartholdy et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 30

, 202

0

T and B cells. Expression of OBF-1 from the lck promoter intransgenic mice did not lead to major changes in T cell devel-opment or function but modulated expression of specific genes(data not shown). One of them, encoding the transcription factorSpi-B, was identified as a physiologically relevant target.

Although ectopic overexpression might represent an artifi-cial situation, the comparison with OBF-1-deficient thymo-cytes strongly enhances the relevance of our findings. In theabsence of OBF-1, a general reduction of Spi-B mRNA levelswas observed in thymocytes, mainly ref lecting a reduction inthe use of P2 (Fig. 1). The normal physiological significance ofthese findings still remains to be elucidated, because OBF-1protein is undetectable in total thymic extracts, and theabsence of OBF-1 (and of Spi-B) does not notably affect T celldevelopment. It is possible, however, that low levels of endog-enous OBF-1 protein exist in a subset of thymic populations,be it T or thymic B cells. It was recently shown that overex-pression of Spi-B in DN3 thymocytes can reverse T lineagecommitment and block �-selection due to impaired inductionof Egr proteins (42), which, however, was not observed in ourmice.

A number of in vitro experiments showed that OBF-1 can bindto Spi-B P2 and transactivate transcription from that promoter(Fig. 2). For this phenomenon, the perfect octamer site presentin Spi-B P2 is essential. Furthermore, ChIP assays clearly showeda specific association of the ectopic OBF-1 with Spi-B promoterP2 in vivo (Fig. 3).

Although it is still possible that OBF-1 also regulates anadditional gene(s) required for proper expression of Spi-B, ourdata demonstrate that OBF-1 directly controls Spi-B promoteractivity through a consensus octamer site important for itsexpression. Other putative low-affinity OBF-1 sites in the Spi-Bpromoter may explain why our mutated reporter constructs(P2-S-luc and P2-M-luc) do not completely abolish transcriptionactivation in transient assays.

Because OBF-1 expression is known to be critical for the functionof B lymphocytes, we directed our focus to the physiologicallyrelevant B cell compartment. It was recently shown that Spi-B isinduced upon treatment of v-Abl-transformed pro-B cells with thev-Abl kinase inhibitor STI571 (40). We show here that OBF-1 isabsolutely crucial for this induction, which is caused mainly by astrong increase in transcription from Spi-B P2 (Fig. 4A). Althoughnecessary, the presence of OBF-1 is in itself not sufficient forinduction of Spi-B transcription in these cells, and the OBF-1protein or RNA levels in v-Abl-transformed B cells are not notice-ably altered by STI571 treatment (data not shown). This findingindicates that inhibition of the v-Abl kinase does not influenceOBF-1 expression but, rather, promotes Spi-B transcription througha different mechanism. For example, v-Abl activity could repressSpi-B transcription by directly phosphorylating OBF-1, therebyleading to inhibition of ternary complex formation or coactivation,or by phosphorylating other targets, such as Oct-1 or Oct-2. In thecase of Oct-2, it has been proposed that phosphorylation of aresidue in the POU domain interferes with binding of Oct-2 to thenoncanonical octamer sequence in the murine BLR-1 promoter(43). Although v-Abl-transformed pro-B cells are useful to establishthe requirement for OBF-1 in Spi-B transcription in B cells, thesecells do not correspond to the developmental stage, namely matureB cells, at which the main phenotype has been described in micelacking either Spi-B or OBF-1. Neither of these two factors appearsto be essential at this early stage of B cell development, at which theymight be redundant. We hence investigated expression of SpiB inmature B cells and showed that it is reduced �2-fold in maturesplenic B cells of OBF-1�/� mice as compared with WT (Fig. 4B),demonstrating that Spi-B levels critically depend on OBF-1 in Bcells as well. This decrease in Spi-B levels is primarily caused by areduction in transcripts initiated from the octamer-dependentpromoter P2.

It is interesting to note that Spi-B levels are also reduced in thebone marrow of OBF-1 knockout mice (data not shown) and thatthe loss of OBF-1 expression has effects on early B lymphopoi-esis that have been recognized earlier. There is a reduction oftransitional B cells in the spleen of OBF-1-deficient mice (15,16), possibly caused by apoptosis during the negative selection ofself-reactive B cells occurring at this stage (44, 45). Interestingly,OBF-1 deficiency also has an impact on the Ig � repertoire (46)and leads to a reduction of a subset of Ig � genes (47), suggestingthe involvement of OBF-1 in the selection of transitional B cells.This effect of OBF-1 deficiency might be independent of Spi-Band could be caused by direct binding of OBF-1�Oct-1 or -2complexes to specific octamer sites in some Ig � promoters. Incontrast, constitutive overexpression of OBF-1 in B cells from a� enhancer-V region promoter results in a block of B celldevelopment in the bone marrow of transgenic mice (ourunpublished data), further supporting the notion that tightcontrol of the OBF-1 level is critical for early B cell development.

There are several lines of evidence for the involvement ofOBF-1 in modulating BCR signaling strength. First, OBF-1�/�

B cells are impaired in BCR-triggered Ca2� mobilization; thisdefect was restored in CD22�/� OBF-1�/� double-deficientanimals (48). CD22 is a negative regulator of BCR signaling, andCD22-deficient mice display a lowered activation threshold forBCR cross-linking and increased Ca2� mobilization upon BCRstimulation (49). Second, OBF-1�/� B cells also showed areduced proliferation after anti-IgM stimulation (50). It is thustempting to assume that the B cell differentiation defect in thebone marrow of OBF-1�/� mice is BCR signal-dependent. Third,Spi-B itself, together with PU.1, was implicated in signalingdownstream of the BCR, possibly by controlling the transcriptionof a component coupling Syk to downstream targets, such asPLC� and BLNK (34). One direct Spi-B and PU.1 target genein this cascade was recently identified as being Grap2, a hema-topoietic adaptor protein that, when overexpressed in A20 Bcells, inhibited inductive BLNK phosphorylation and its recruit-ment to Ig� (51). In addition, increased apoptosis in the bonemarrow, possibly caused by the increased sensitivity of OBF-1�/�

transitional B cells to BCR cross-linking observed in vitro (46)could explain the receptor-specific block at the T1 stage of B celldevelopment in OBF-1 mutant mice. These effects, but also thereduced numbers of MZ B cells found in one OBF-1-deficientmouse strain (52), can be explained by a model in which OBF-1is required for setting the normal threshold for immature B celldevelopment and selection in the bone marrow.

The crucial role of OBF-1 for the development of GCs in thespleen reflects a different mechanism, because OBF-1�/� CD22�/�

double-deficient mice were still unable to mount humoral immuneresponses or to form GCs, although their BCR activation thresholdwas similar to the WT situation (48). The data presented hereindicate that the expression of Spi-B, controlled by OBF-1 proteinin the GC B cells may be a key element for the GC formation.Importantly, expression of Spi-B from the OBF-1-independentpromoter P1 is normally down-modulated in GC B cells, so thatSpi-B mRNA expression becomes highly dependent on the OBF-1-driven P2 (Fig. 5B); this explains mechanistically how OBF-1 canbe crucial for Spi-B expression in GC B cells.

To conclude, we have identified Spi-B as a physiologicallyrelevant in vivo target of OBF-1. These findings greatly help toextend our understanding of how OBF-1 is involved in B cellactivation and GC formation, and place it directly upstream ofSpi-B in a transcriptional hierarchy.

Materials and MethodsMouse Strains and Cell Lines. Mouse strains and cell lines used aredescribed in Supporting Methods, which is published as support-ing information on the PNAS web site.

Bartholdy et al. PNAS � August 1, 2006 � vol. 103 � no. 31 � 11669

IMM

UN

OLO

GY

Dow

nloa

ded

by g

uest

on

Nov

embe

r 30

, 202

0

PCR. The sequence of all primers used is presented in SupportingMethods.

Immunohistochemistry and in Situ Hybridization. Paraformalde-hyde-fixated thymi were paraffin-embedded, and 6-�m sectionswere prepared for histochemistry. For in situ hybridization, a450-bp probe from the 3� UTR of the murine Spi-B cDNA andthe corresponding antisense probe were digoxigenin-labeled(Roche).

ChIP. ChIP was performed as described in ref. 53 with 10 �g ofchromatin from murine thymocytes or 3 �g of chromatin fromSTI571-treated Abelson B cells. Immunoprecipitation wasperformed with an anti-HA antibody (12CA5) and with the

monoclonal OBF-1 antibody C-20 (SC-955 X; Santa CruzBiotechnology).

We thank Roger Perlmutter (Amgen, Thousand Oaks, CA) for providingus with the lck promoter construct; Daniel Tenen (Harvard University,Boston, MA) for the Spi-B reporter constructs and RPA probe; EdwardOakeley and Herbert Angliker for microarray help; Jean-Francois Spetz(Friedrich Mieschler Institute, Basel, Switzerland) for generation oftransgenic mice; Celeste Simon (University of Pennsylvania, Philadel-phia, PA) for reagents; Melanie Sticker for help with histology; MarieKosco-Vilbois for immunohistochemistry protocols; the Matthias labo-ratory members for stimulating discussion; Ralph Tiedt, Edgar Serfling,and Steffen Junker for critical comments on the manuscript; andNovartis for the generous gift of STI571. This work was funded by theNovartis Research Foundation.

1. Luo, Y. & Roeder, R. G. (1995) Mol. Cell. Biol. 15, 4115–4124.2. Strubin, M., Newell, J. W. & Matthias, P. (1995) Cell 80, 497–506.3. Matthias, P. (1998) Semin. Immunol. 10, 155–163.4. Staudt, L. M. & Lenardo, M. J. (1991) Annu. Rev. Immunol. 9, 373–398.5. Bergman, Y., Rice, D., Grosschedl, R. & Baltimore, D. (1984) Proc. Natl. Acad.

Sci. USA 81, 7041–7045.6. Wirth, T., Staudt, L. & Baltimore, D. (1987) Nature 329, 174–178.7. Jenuwein, T. & Grosschedl, R. (1991) Genes Dev. 5, 932–943.8. Dreyfus, M., Doyen, N. & Rougeon, F. (1987) EMBO J. 6, 1685–1690.9. Mason, J. O., Williams, G. T. & Neuberger, M. S. (1985) Cell 41, 479–487.

10. Schubart, D. B., Rolink, A., Kosco-Vilbois, M. H., Botteri, F. & Matthias, P.(1996) Nature 383, 538–542.

11. Laumen, H., Nielsen, P. J. & Wirth, T. (2000) Eur. J. Immunol. 30, 458–469.12. Nielsen, P. J., Georgiev, O., Lorenz, B. & Schaffner, W. (1996) Eur. J. Immunol.

26, 3214–3218.13. Kim, U., Qin, X. F., Gong, S., Stevens, S., Luo, Y., Nussenzweig, M. & Roeder,

R. G. (1996) Nature 383, 542–547.14. Schubart, K., Massa, S., Schubart, D., Corcoran, L. M., Rolink, A. G. &

Matthias, P. (2001) Nat. Immunol. 2, 69–74.15. Hess, J., Nielsen, P. J., Fischer, K. D., Bujard, H. & Wirth, T. (2001) Mol. Cell.

Biol. 21, 1531–1539.16. Schubart, D. B., Rolink, A., Schubart, K. & Matthias, P. (2000) J. Immunol. 164,

18–22.17. Corcoran, L. M., Hasbold, J., Dietrich, W., Hawkins, E., Kallies, A., Nutt, S. L.,

Tarlinton, D. M., Matthias, P. & Hodgkin, P. D. (2005) J. Exp. Med. 201,1385–1396.

18. Zwilling, S., Dieckmann, A., Pfisterer, P., Angel, P. & Wirth, T. (1997) Science277, 221–225.

19. Sauter, P. & Matthias, P. (1997) Immunobiology 198, 207–216.20. Qin, X. F., Reichlin, A., Luo, Y., Roeder, R. G. & Nussenzweig, M. C. (1998)

EMBO J. 17, 5066–5075.21. Greiner, A., Muller, K. B., Hess, J., Pfeffer, K., Muller-Hermelink, H. K. &

Wirth, T. (2000) Am. J. Pathol. 156, 501–507.22. Boehm, J., He, Y., Greiner, A., Staudt, L. & Wirth, T. (2001) EMBO J. 20,

4153–4162.23. Tiedt, R., Bartholdy, B. A., Matthias, G., Newell, J. W. & Matthias, P. (2001)

EMBO J. 20, 4143–4152.24. Wolf, I., Pevzner, V., Kaiser, E., Bernhardt, G., Claudio, E., Siebenlist, U.,

Forster, R. & Lipp, M. (1998) J. Biol. Chem. 273, 28831–28836.25. Moriuchi, M. & Moriuchi, H. (2001) J. Biol. Chem. 276, 8639–8642.26. Malone, C. S. & Wall, R. (2002) J. Immunol. 168, 3369–3375.27. Kim, U., Siegel, R., Ren, X., Gunther, C. S., Gaasterland, T. & Roeder, R. G.

(2003) Proc. Natl. Acad. Sci. USA 100, 8868–8873.28. Brunner, C., Laumen, H., Nielsen, P. J., Kraut, N. & Wirth, T. (2003) J. Biol.

Chem. 278, 45231–45239.29. Crepieux, P., Coll, J. & Stehelin, D. (1994) Crit. Rev. Oncog. 5, 615–638.

30. Sharrocks, A. D. (2001) Nat. Rev. Mol. Cell. Biol. 2, 827–837.31. Ray, D., Bosselut, R., Ghysdael, J., Mattei, M. G., Tavitian, A. & Moreau-

Gachelin, F. (1992) Mol. Cell. Biol. 12, 4297–4304.32. Ray-Gallet, D., Mao, C., Tavitian, A. & Moreau-Gachelin, F. (1995) Oncogene

11, 303–313.33. Eisenbeis, C. F., Singh, H. & Storb, U. (1995) Genes Dev. 9, 1377–1387.34. Garrett-Sinha, L. A., Su, G. H., Rao, S., Kabak, S., Hao, Z., Clark, M. R. &

Simon, M. C. (1999) Immunity 10, 399–408.35. Dahl, R., Ramirez-Bergeron, D. L., Rao, S. & Simon, M. C. (2002) EMBO J.

21, 2220–2230.36. Su, G. H., Chen, H. M., Muthusamy, N., Garrett-Sinha, L. A., Baunoch, D.,

Tenen, D. G. & Simon, M. C. (1997) EMBO J. 16, 7118–7129.37. Shimizu, C., Kawamoto, H., Yamashita, M., Kimura, M., Kondou, E., Kaneko,

Y., Okada, S., Tokuhisa, T., Yokoyama, M., Taniguchi, M., et al. (2001) Int.Immunol. 13, 105–117.

38. Chen, H. M., Gonzalez, D. A., Radomska, H. S., Voso, M. T., Sun, Z., Zhang,P., Zhang, D. E. & Tenen, D. G. (1998) Gene 207, 209–218.

39. Babb, R., Cleary, M. A. & Herr, W. (1997) Mol. Cell. Biol. 17, 7295–7305.40. Muljo, S. A. & Schlissel, M. S. (2003) Nat. Immunol. 4, 31–37.41. Veale, M. F., Dietrich, W. M. & Corcoran, L. M. (2003) Immunol. Cell Biol.

81, 297–304.42. Lefebvre, J. M., Haks, M. C., Carleton, M. O., Rhodes, M., Sinnathamby, G.,

Simon, M. C., Eisenlohr, L. C., Garrett-Sinha, L. A. & Wiest, D. L. (2005)J. Immunol. 174, 6184–6194.

43. Pevzner, V., Kraft, R., Kostka, S. & Lipp, M. (2000) Biochem. J. 347,29–35.

44. Carsetti, R., Kohler, G. & Lamers, M. C. (1995) J. Exp. Med. 181, 2129–2140.

45. Sater, R. A., Sandel, P. C. & Monroe, J. G. (1998) Int. Immunol. 10,1673–1682.

46. Jankovic, M. & Nussenzweig, M. C. (2003) Int. Immunol. 15, 1099–1104.47. Casellas, R., Jankovic, M., Meyer, G., Gazumyan, A., Luo, Y., Roeder, R. &

Nussenzweig, M. (2002) Cell 110, 575–585.48. Samardzic, T., Gerlach, J., Muller, K., Marinkovic, D., Hess, J., Nitschke, L. &

Wirth, T. (2002) Eur. J. Immunol. 32, 2481–2489.49. Nitschke, L., Carsetti, R., Ocker, B., Kohler, G. & Lamers, M. C. (1997) Curr.

Biol. 7, 133–143.50. Sun, J., Matthias, G., Mihatsch, M. J., Georgopoulos, K. & Matthias, P. (2003)

J. Immunol. 170, 1699–1706.51. Garrett-Sinha, L. A., Hou, P., Wang, D., Grabiner, B., Araujo, E., Rao, S., Yun,

T. J., Clark, E. A., Simon, M. C. & Clark, M. R. (2005) Gene 353, 134–146.52. Samardzic, T., Marinkovic, D., Nielsen, P. J., Nitschke, L. & Wirth, T. (2002)

Mol. Cell. Biol. 22, 8320–8331.53. Masternak, K., Peyraud, N., Krawczyk, M., Barras, E. & Reith, W. (2003) Nat.

Immunol. 4, 132–137.

11670 � www.pnas.org�cgi�doi�10.1073�pnas.0509430103 Bartholdy et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 30

, 202

0