Stage-Specific Modulation of IFN-Regulatory Factor 4 Function by Krüppel-Type Finger

This information is current as Sanjay Gupta, Alissa Anthony and Alessandra B. Pernis of September 29, 2021. J Immunol 2001; 166:6104-6111; ; doi: 10.4049/jimmunol.166.10.6104 http://www.jimmunol.org/content/166/10/6104 Downloaded from

References This article cites 43 articles, 22 of which you can access for free at: http://www.jimmunol.org/content/166/10/6104.full#ref-list-1

Why The JI? Submit online. http://www.jimmunol.org/

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

*average by guest on September 29, 2021

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2001 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Stage-Specific Modulation of IFN-Regulatory Factor 4 Function by Kru¨ppel-Type Zinc Finger Proteins1

Sanjay Gupta, Alissa Anthony, and Alessandra B. Pernis2

Optimal humoral responses depend on the activation of Ag-specific B cells, followed by their progression toward a fully differ- entiated phenotype. Acquisition of stage-appropriate patterns of expression is crucial to this differentiation program. How- ever, the molecular mechanisms used by B cells to modulate gene expression as they complete their maturation program are poorly understood. IFN-regulatory factor 4 (IRF-4) plays a critical role in mature B cell function. Using the transcriptional regulation of the human B cell activation marker CD23 as a model system, we have previously demonstrated that IRF-4 is induced in response to B cell-activating stimuli and that it acts as a transactivator of CD23 gene expression. We have furthermore found that IRF-4 function can be blocked by B cell lymphomas 6 (BCL-6) , a Kru¨ppel-type zinc finger repressor normally expressed in

germinal center B cells. However, CD23 expression is known to be down-regulated in plasma cells despite high level expression Downloaded from of IRF-4 and the lack of BCL-6, suggesting that in plasma cells the IRF-4-mediated induction of CD23 is prevented by its interaction with a distinct repressor. In this set of studies, we demonstrate that IRF-4 interacts with B lymphocyte-induced maturation protein/positive regulatory domain I-binding factor 1 (Blimp1/PRD1-BF1), a Kru¨ppel-type zinc finger protein whose expression correlates with terminal B cell differentiation. Functional studies indicate that Blimp1, like BCL-6, can block IRF-4- transactivating ability. These findings thus support a model whereby IRF-4 function is modulated in a stage-specific manner by

its interaction with developmentally restricted sets of Kru¨ppel-type zinc finger proteins. The Journal of Immunology, 2001, 166: http://www.jimmunol.org/ 6104–6111.

ffective humoral responses involve the activation of Ag- tion of mature B cells and are unable to mount Ab responses to specific B cells, followed by their differentiation toward either T-dependent or T-independent Ags. Functional studies have E memory or plasma cells. Progression of an activated B shown that IRF-4 can subserve a dual role in lymphoid cells. By cell along its terminal differentiation pathway requires the acqui- itself, IRF-4 can bind to IFN-stimulated response elements and sition of stage-appropriate patterns of gene expression. Therefore, repress the expression of IFN-inducible (3). In the presence once the molecular machinery responsible for the activation pro- of a cofactor, the protooncogene PU.1, IRF-4 can act as a trans- gram of a B cell has been implemented, it must be selectively of the Ig ␬- and ␭-light chain enhancers and the CD20 by guest on September 29, 2021 modulated to ensure successful completion of terminal differenti- promoter (1, 8, 9). ation. The molecular mechanisms used by B cells to selectively Our studies have used the regulation of the B cell activation regulate gene expression as they proceed along their maturation marker CD23 as a model system. CD23 is a type II integral mem- program are largely unknown. brane protein that belongs to the C-type lectin superfamily (10– IRF-4 is a novel member of the IFN-regulatory factor (IRF)3 12). This family includes a subset of killer-inhibitory receptors that family of transcriptional regulators, whose expression is primarily exert inhibitory functions on NK cells (10–14). Two differentially restricted to the lymphoid compartment (1–3). IRF-4 expression in spliced forms of CD23 (termed CD23a and CD23b) exist in hu- B cells is induced in response to stimuli known to drive B cell mans, whereas only one form, which more closely resembles activation (2, 4–6). Genetic evidence indicates that IRF-4 plays a CD23a, has been consistently found in mice. The two human crucial role in controlling the activation and homeostasis of im- CD23 isoforms are the result of use of alternative transcriptional mune responses (7). Despite exhibiting a normal B cell develop- start sites and are regulated by two distinct promoters (15). CD23a ment, IRF-4-deficient mice display profound defects in the func- and CD23b only differ by a 6/7-aa substitution, leading to the removal of an immunoreceptor tyrosine-based inhibition motif (ITIM) present in CD23a, but not in CD23b. ITIMs have been Department of Medicine, Columbia University, New York, NY 10032 shown to play a critical role in the ability of the killer-inhibitory Received for publication September 20, 2000. Accepted for publication February receptors to inhibit NK cell activation (16, 17). Although the exact 16, 2001. functional role of the CD23a ITIM has not been elucidated, the The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance differential expression of an ITIM motif in the two CD23 isoforms with 18 U.S.C. Section 1734 solely to indicate this fact. suggests that CD23a and CD23b may mediate distinct signaling 1 This work was supported by the New York City Council Speaker’s Fund for Bio- pathways. medical Research: Toward the Science of Patient Care, and by an Arthritis Foundation We have focused our attention on the regulation of CD23b, the Science Grant. ITIM-less isoform of CD23. We have previously found that IRF-4 2 Address correspondence and reprint requests to Dr. Alessandra Pernis, Department of Medicine, Columbia University, 630 West 168th Street, New York, NY 10032. participates in a multiprotein or “enhanceosome-like” complex E-mail address: [email protected] that targets the CD23b IFN-␥ activation site (GAS), a critical reg- 3 Abbreviations used in this paper: IRF, IFN-regulatory factor; Blimp1, B lympho- ulatory element in the promoter of CD23b (4). The IRF-4-medi- cyte-induced maturation protein; CIITA, MHC class II transactivator; GAS, IFN-␥ ated transactivation of CD23b is blocked by B cell, lymphomas 6 activation site; HA, hemagglutinin; ITIM, immunoreceptor tyrosine-based inhibition motif; PRDI-BF1, positive regulatory domain I-binding factor 1; wt, wild type; - protein (BCL-6), a Kru¨ppel-type zinc finger transcriptional repres- PRF site, myc-plasmacytoma repressor factor site. sor present in germinal center B cells (18, 19). Because BCL-6

Copyright © 2001 by The American Association of Immunologists 0022-1767/01/$02.00 The Journal of Immunology 6105 expression is down-regulated upon B cell activation (4, 18, 20), we tants of IRF-4 were prepared by Pfu PCR from pBSKS-IRF-4 plasmid have proposed that induction of high levels of IRF-4 coupled with template using appropriate primers. The cDNA segments encoding these the loss of BCL-6 plays a key role in the ability of activated B cells deletion mutants were then subcloned, in frame, into the filled EcoRI site of pGEX-3X vector to generate GST-IRF-4 deletion mutants. The in frame to up-regulate CD23b expression. Interestingly, upon terminal dif- junction in the GST-IRF-4 fusion constructs was confirmed by DNA se- ferentiation of a B cell into a plasma cell, CD23 expression is quencing in an automated cycle sequencer (Perkin-Elmer, Norwalk, CT). normally down-regulated (21), although plasma cells continue to The full-length human BCL-6 expression vector was a kind gift of R. express high levels of IRF-4 (1) and lack BCL-6 (19). This obser- Dalla-Favera (Columbia University, New York, NY) (27). HA epitope- tagged murine Blimp1 cDNA cloned into pBluescript vector (pBSKS-HA- vation led us to hypothesize that, in plasma cells, the IRF-4-me- Blimp1) was a kind gift of K. Calame (23). The Blimp1 expression con- diated induction of CD23b might be repressed by a different struct (pCEP4-HA-Blimp1) was generated by cloning the coding region of mechanism. the HA-Blimp1 cDNA into pCEP4 expression vector. The full-length hu- Previous work has demonstrated that another Kru¨ppel-type zinc man Blimp1/PRDI-BF1 cDNA cloned into pXM mammalian expression finger protein, Blimp-1 (B lymphocyte-induced maturation pro- vector was a kind gift of T. Maniatis (Harvard University, Cambridge, MA) (25). The CD23b promoter firefly luciferase reporter construct in the pGL3- tein), is mainly detected in late B and plasma cells and is a critical enhancer vector was previously described (4). regulator of terminal B cell differentiation (22). Although the full extent of B lymphocyte-induced maturation protein (Blimp1) func- DNA-binding assays and cell extracts tions is not known, Blimp1 is a critical transcriptional repressor of The preparation and employment of DNA oligonucleotide probes for EM- the c-myc oncogene (23) and the MHC class II transactivator (CI- SAs have been described previously (4). The double-stranded oligonucleo- ITA) (24), both of which are down-regulated upon terminal B cell tides used as probes or cold competitors in these studies were as follows: CD23b GAS (wild-type, wt), 5Ј-gatcGGGTGAATTTCTAAGAAAGGGAC- differentiation. Although Blimp1 is believed to play a key role in 3Ј; CD23b GAS M1, 5Ј-gatcGGGTGAATTTCTAAGGTCGGGAC-3Ј; Downloaded from terminal B cell differentiation, its expression is not confined to CD23b GAS M2, 5Ј-gatcGGGTGGTCTTCTAAGAAAGGGAC-3Ј; CD23b lymphoid cells. Indeed, the human homologue of Blimp1, PRDI- GAS M3, 5Ј-gatcGGGTGAATGCTGAAGAAAGGGAC-3Ј; CD20, 5Ј-gatc BF1 (positive regulatory domain I-binding factor 1), is induced GGGGTCTTTTTCAAGAAGTGAAACCT-3Ј (9); ␬3Ј enhancer, 5Ј-gatcCC TTTGAGGAACTGAAAACAGAACCT-3Ј (28); CIITA, 5Ј-gatcACAGTAA upon viral infection of fibroblasts and was cloned because of its Ј Ј ␤ GGAAGTGAAATTAATTTCAG-3 ; MYC-PRF site, 5 -gatcCGCGTACA ability to target the IRF binding site within the -IFN enhanceo- GAAAGGGAAAGGACTAG-3Ј (23); I⑀ GAS, 5Ј-gatcAACTTCCCAA some, and to repress ␤-IFN gene expression (25). Interestingly, GAACA-3Ј (29). Oligonucleotide competition and Ab interference assays http://www.jimmunol.org/ recent immunohistochemical analysis has revealed that plasma were performed as previously described (4). Nuclear extracts were prepared as cells as well as a subset of germinal center B cells with a partial previously described (4). plasma cell phenotype strongly express both Blimp1 and IRF-4 (26). Immunoprecipitations and Western blot analysis In this study, we report that PRDI-BF1/Blimp1 can bind to the Cell extracts were immunoprecipitated with an anti-IRF-4, or anti-HA Ab, same functional element in the human CD23b promoter to which as previously described (4). The immunoprecipitates were resolved by 7% BCL-6 and IRF-4 had previously been shown to bind, and that, SDS-PAGE. The gel was transferred to a nitrocellulose membrane, and like BCL-6, Blimp1 can repress IRF-4-transactivating ability. then immunoblotted with an IRF-4 or HA epitope Ab. The bands were Thus, IRF-4 function can be modulated in a stage-specific manner visualized by ECL (Amersham Pharmacia Biotech, Piscataway, NJ). by guest on September 29, 2021 by its interaction with developmentally restricted sets of Kru¨ppel- GST pull-down assays and transient transfections type zinc finger proteins. GST pull-down assays were conducted as previously described (4). The bound proteins were eluted from the beads by boiling them in SDS-PAGE Materials and Methods sample buffer, fractionated on a 7% SDS-polyacrylamide gel, and then Cell lines and cultures blotted onto a nitrocellulose membrane. The blot was probed with either a BCL-6, HA epitope, or Stat6 Ab. The human B cell line Ramos (obtained from Dr. S. Lederman, Columbia For expression of recombinant proteins, 293T cells were transfected University, New York, NY) is an EBV-negative Burkitt’s lymphoma. with expression plasmids by calcium phosphate precipitation method. After U266 (American Type Culture Collection (ATCC), Manassas, VA) is de- 24 h of incubation, the transfected cells were harvested for nuclear extract rived from a multiple myeloma cell line. U937 (obtained from Dr. K. preparation. Calame, Columbia University) is a monocytic cell line. Human embryo Transient transfection assays for reporter experiments were performed kidney 293T cells (obtained from Dr. S. Lederman) were cultured in as previously described (4). DMEM supplemented with 10% FCS (Atlanta Biologicals, Norcross, GA). All other cells were grown in IMDM supplemented with 10% FCS (Atlanta Biologicals), as previously described (4). U266 cells (10–20 ϫ 106) were Results stimulated with 1 ␮g/ml of either anti-CD40 or an isotype-matched control Blimp1 binds to the CD23b GAS Ab in a final volume of 10 ml at 37°C for 24 h. The human homologue of Blimp1, PRDI-BF1, has previously been Antibodies shown to target the IRF binding site in the IFN-␤ enhancer (25). Furthermore, the known DNA-binding elements for murine IRF-4 The rabbit polyclonal antiserum against IRF-4 used in EMSA experiments was a generous gift of Hisamaru Hirai (University of Tokyo, Tokyo, Japan) (1, 8) and Blimp1 (23, 24) are strikingly similar (Table I). We thus (3). We subsequently generated our own rabbit polyclonal anti-IRF-4 an- decided to investigate whether Blimp1/PRDI-BF1 can target the tiserum using a similar GST-IRF-4 (nucleotides 441–924) fusion protein as IRF-4 binding site in the human CD23b promoter (4). Therefore, the immunogen (Berkeley Antibody, Richmond, CA). This antiserum was we performed EMSAs with a CD23b GAS probe on extracts from used for immunoprecipitations and immunoblot analyses (4). Blots were U266. This is a myeloma cell line, representative of a terminally also reprobed with a commercially available anti-IRF-4 antiserum (Santa Cruz Biotechnology, Santa Cruz, CA), which gave identical results. Rabbit differentiated plasma cell (30), which lacks BCL-6 and expresses polyclonal antisera against human Stat6, or BCL-6, were purchased from high levels of Blimp1/PRDI-BF1 (data not shown). As shown in Santa Cruz Biotechnology. The mAb against hemagglutinin (HA) epitope Fig. 1A, the pattern of CD23b GAS-binding complexes in U266 (clone 12CA5) was from Boehringer Mannheim (Indianapolis, IN). The extracts was markedly different from that detected in extracts from hybridomas secreting the anti-CD40 mAb G28-5 (IgG1) or an isotype- matched control mAb were obtained from ATCC. Ramos cells, which phenotypically resemble germinal center B cells and contain BCL-6, but not Blimp1/PRDI-BF1 (data not DNA constructs shown) (31). As we previously demonstrated (4), in Ramos cells, The human IRF-4 expression plasmid (pCEP4-IRF-4) and the GST-IRF-4 IRF-4 participates in the formation of a slow mobility complex, expression plasmid were previously described (4). Various deletion mu- whose visualization is normally obscured by the presence of a 6106 MODULATION OF IRF-4 FUNCTION BY BLIMP1

Table I. Sequence comparison of IRF-4 and Blimp1 binding sites Oligonucleotide competition experiments with a panel of CD23b GAS mutant elements (Fig. 2B) revealed that the CD23b GAS M2 Binding site Sequence (5Ј to 3Ј) mutant, which contains mutations in the IRF-4 binding site (4), is unable to efficiently compete the Blimp1 complex. The competi- IRF-4 ␬3Ј Enhancer CTTTGA GGAACTGAAA ACAGA ␭B Enhancer ATAAAA GGAAGTGAAA CCAAG tion pattern for the Blimp1 complex is furthermore distinct from Blimp 1 MYC-PRF GGTACA GAAAGGGAAA GGACT that we have previously shown for BCL-6, which is competed by CIITA CAGTAA GGAAGTGAAA TTAAT the M2, but not by the M1 or M3 mutants (4). These results thus suggest that at different stages of B cell differentiation, distinct Kru¨ppel-type zinc finger proteins target the CD23b GAS. strong BCL-6-containing complex. In contrast, extracts from U266 To assess whether the ability of BCL-6, Blimp1, and IRF-4 to cells contained a clearly visible IRF-4 complex and lacked the target a common DNA element was confined to the CD23b GAS, BCL-6 complex. Interestingly, U266 contained a distinct CD23b we then subjected extracts from 293T transfectants expressing GAS-binding complex (complex X), which was absent in Ramos either BCL-6 or Blimp1 to an oligonucleotide competition exper- cells. The mobility of this complex was faster than that of the iment with a panel of DNA elements from different promoters/ BCL-6-containing complex. Furthermore, BCL-6 expression is enhancers. The elements used in this experiment included: 1) sites normally down-regulated upon CD40 stimulation (4, 18, 20). In known to bind Blimp1 (MYC-PRF, CIITA) (23, 24), 2) sites known ␬ Ј contrast, stimulation of U266 cells with an anti-CD40 Ab did not to bind IRF-4 ( 3 enhancer, CD20) (8, 9), and 3) a site known to ⑀ affect the appearance of complex X, despite appropriately decreas- bind BCL-6 (I GAS) (32). Self-competition with the CD23b GAS ing the intensity of the BCL-6 complex in Ramos cells (Fig. 1A). element was included as control. As shown in Fig. 2C, this experi- Downloaded from ␬ Ј Supershifting experiments confirmed the presence of IRF-4 in ment indicated that all three known IRF-4 binding sites ( 3 en- U266 cells and demonstrated that complex X did not contain either hancer, CD20, and CD23b GAS) could efficiently block the bind- IRF-4 or BCL-6 (Fig. 1B and data not shown). EMSA analysis of ing of both the BCL-6 as well as the Blimp1 complexes to the three additional myeloma cell lines revealed a similar pattern of CD23b GAS. In contrast, the known Blimp1 binding sites (MYC- CD23b GAS-binding complexes (data not shown). PRF, CIITA) only competed the Blimp1, but not the BCL-6 com-

To determine whether the faster CD23b GAS-binding complex plex, whereas the addition of an oligonucleotide containing the http://www.jimmunol.org/ ⑀ present in U266, but not in Ramos cells contained Blimp1/PRDI- BCL-6 binding site (I GAS) only blocked binding of the BCL-6 BF1, we subjected U266 extracts to oligonucleotide competition complex, but not that of the Blimp1 complex. These data thus experiments using the known Blimp1 binding site from the myc suggest that both BCL-6 and Blimp1 can also target other IRF-4 promoter (MYC-PRF site) (23) as a cold competitor of the radio- binding sites in addition to the CD23b GAS. However, the targets labeled CD23b GAS wt probe. These oligonucleotide competition of the two Kru¨ppel-type zinc finger proteins do not completely experiments revealed that the MYC-PRF element could efficiently overlap. Of particular interest is the finding that the Blimp1 com- ⑀ compete both the IRF-4-containing complex as well as the faster plex was not competed by the I GAS, which is known to bind mobility complex, suggesting that the latter complex might contain Stat6 in addition to BCL-6 (32). In contrast to the CD23b GAS, the ⑀ by guest on September 29, 2021 Blimp1/PRDI-BF1 (Fig. 2A, left panel). Competition of IRF-4 was I GAS does not appear to be targeted by IRF-4, because it fails to not surprising because MYC-PRF has been described as an IFN- compete IRF-4 complexes in oligonucleotide competition assays stimulated response-like element (23). To confirm this observa- (data not shown). This result thus suggests that the known ability tion, we then proceeded to directly test whether rBlimp1 could of BCL-6 to modulate Stat6 activity may not be shared by Blimp1. target the CD23b GAS. 293T cells (which lack BCL-6, Blimp1, and IRF-4) were transiently transfected with an expression plasmid IRF-4 physically interacts with Blimp1 encoding either Blimp1 or BCL-6. Cells transfected with an empty In our previous studies, we had detected a very strong interaction vector were included as control. Extracts from the various 293T between BCL-6 and IRF-4 (4). To determine whether Blimp1 transfectants were then subjected to EMSA experiments using the could also physically associate with IRF-4, we first performed CD23b GAS as a probe (Fig. 2A, right panel). These experiments pull-down assays with a GST-IRF-4 fusion protein. As shown in clearly revealed that 293T transfectants expressing Blimp1 or Fig. 3A, incubation of a GST-IRF-4 fusion protein with extracts BCL-6 contained CD23b GAS-binding complexes, which were ab- from 293T cells transfected with an HA epitope-tagged Blimp1 sent in 293T cells tranfected with an empty vector. The mobility expression vector revealed that IRF-4 can indeed associate with exhibited by rBlimp1 was similar to that of the CD23b GAS-bind- Blimp1. No interaction was observed with the GST moiety alone ing complex, which we had detected in U266 cells, but not in or upon incubation of the GST-IRF-4 with a control extract Ramos, and was clearly distinct from that displayed by rBCL-6. (Ramos) or with 293T cells transfected with an empty vector. We

FIGURE 1. Different patterns of CD23b GAS-bind- ing complexes are displayed by human cell lines cor- responding to distinct stages of B cell differentiation. A, Nuclear extracts from U266 (a myeloma cell line) were compared with those from Ramos (a Burkitt’s lym- phoma) by EMSA using a 32P-labeled CD23b GAS wt probe. Ramos and U266 cells were either unstimulated or stimulated with an anti-CD40 Ab (1 ␮g/ml) or a control Ab (1 ␮g/ml) for 24 h at 37°C. B, Ab interfer- ence mobility shift assays were conducted by addition of antisera against IRF-4 or control. All antisera were added at a final dilution of 1/20 for 30 min at 4°C be- fore incubation with the probe for 20 min at 25°C. U266 cells were unstimulated. The Journal of Immunology 6107

FIGURE 2. Blimp1 can target the CD23b GAS. A, Nuclear extracts from unstimulated U266 cells were obtained and assayed as de- scribed in Fig. 1. Oligonucleotide competi- tion assays were performed either in the ab- sence or in the presence of a 100-fold molar excess of a cold MYC-PRF or CD23b GAS oligonucleotide added to the shift reaction as indicated (left panel). 293T cells were tran- siently transfected with an empty vector or with an expression vector encoding either BCL-6 or Blimp1. Nuclear extracts were then prepared and analyzed by EMSA using a 32P-labeled CD23b GAS wt probe (right panel). B, 293T cells were transiently trans- fected with an empty vector or with an ex- pression vector encoding Blimp1 and then assayed by EMSA, as described above. Oli- Downloaded from gonucleotide competition assays were per- formed either in the absence or in the pres- ence of a 100-fold molar excess of cold wt or mutant CD23b GAS oligonucleotides added to the shift reaction as indicated. C, 293T cells were transiently transfected with an empty vector or with an expression vector http://www.jimmunol.org/ encoding either BCL-6 or Blimp1 and then assayed as described above. Oligonucleotide competition assays were performed either in the absence or in the presence of a 100-fold molar excess of cold oligonucleotides corre- sponding to previously described Blimp1, IRF-4, or BCL-6 binding sites. Cold compet- itors were added to the shift reaction as

indicated. by guest on September 29, 2021

also cotransfected 293T cells with both IRF-4 as well as HA letion of this region (⌬151–199) resulted in a diminished inter- epitope-tagged Blimp1 expression vectors and then subjected the action of the mutant protein with Blimp1. When the GST-IRF-4 extracts to immunoprecipitation assays with either an anti-IRF-4 or deletion mutants were incubated with extracts from Ramos cells anti-HA Ab (Fig. 3B). Consistent with our GST pull-down assays, and the IRF-4/BCL-6 interaction visualized by Western blotting presence of the IRF-4 protein could be detected in anti-HA im- with an anti-BCL-6 antiserum, the aa 1–150 region of IRF-4 munoprecipitates of 293T cells cotransfected with both IRF-4 and was again found to be essential for its interaction with BCL-6 HA epitope-tagged Blimp1 (Fig. 3B, upper panel), but not of con- (Fig. 4A, middle panel). Similarly to the Blimp1/IRF-4 inter- trol transfectants. Stripping and reprobing of the filter with an anti- action, the IRF-4 mutant lacking the proline-rich region (⌬151– HA Ab demonstrated the presence of HA-Blimp1 in the IRF-4 199) also displayed a decreased ability to associate with BCL-6. immunoprecipitates of 293T transfectants expressing both IRF-4 However, in striking contrast to the results obtained with and HA epitope-tagged Blimp1, but not of 293T control transfec- Blimp1, association of IRF-4 with BCL-6 also required the aa tants (Fig. 3B, lower panel). Taken together, these data indicate 200–360 region of IRF-4. We have previously shown that, in that IRF-4 can physically interact not only with BCL-6, but also Ramos cells, IRF-4 can interact not only with BCL-6, but also with Blimp1. with Stat6 (4). Interestingly, stripping and reprobing of this To further dissect the interaction of IRF-4 with the two Kru¨p- Western blot with a Stat6 Ab (Fig. 4A, lower panel) demon- pel-type zinc finger proteins, we proceeded to map the IRF-4 ⌬ domains involved in the association of IRF-4 with BCL-6 and strated that the IRF-4 1–150 interacted normally with Stat6. with Blimp1. We thus generated a panel of different GST-IRF-4 Association of Stat6 with IRF-4 was instead found to require mutant proteins, and compared them with wt IRF-4 in GST the aa 200–360 region. As summarized in Fig. 4B, this exper- pull-down assays (Fig. 4A). Incubation of the different GST- iment thus revealed that the aa 1–150 region of IRF-4, which IRF-4 deletion mutants with extracts from 293T transfectants contains the DNA of IRF-4, is required for its expressing HA epitope-tagged Blimp1 revealed that interaction interaction with both BCL-6 and Blimp1, but not with Stat6. In of IRF-4 with Blimp1 requires the aa 1–150 region of IRF-4, contrast, the aa 200–360 region of IRF-4, which contains res- which is known to contain its DNA binding domain (Fig. 4A, idues previously shown to be critical for its interaction with upper panel) (1). Contribution of the proline-rich region of PU.1 (33), is necessary for the association of IRF-4 with BCL-6 IRF-4 to its interaction with Blimp1 is also likely because de- and Stat6, but not with Blimp1. 6108 MODULATION OF IRF-4 FUNCTION BY BLIMP1

These studies thus indicate that Blimp1/PRDI-BF1 can block IRF-4 function. To further examine the functional role of Blimp1 on the regu- lation of CD23b, we then proceeded to determine the effects of Blimp1 cotransfection on the IL-4 inducibility of CD23b (Fig. 5C). As we previously demonstrated, similar inducibility of the CD23b promoter construct can be achieved upon IL-4 stimulation of U937 or upon cotransfection of IRF-4. However, when cotransfection of IRF-4 is conducted in the presence of IL-4 stimulation, higher levels of luciferase activity are detected, suggesting that optimal CD23b inducibility requires cooperation between Stat6 and IRF-4 (4). Consistent with previous results, addition of BCL-6 blocked the IRF-4-mediated transactivation of this reporter construct as well as the IL-4 inducibility of CD23b (4, 34). Cotransfection of Blimp1 blocked the ability of IRF-4 to drive CD23b promoter activity and to augment the IL-4 inducibility of this reporter con- struct. However, in contrast to BCL-6, presence of Blimp1 did not affect the IL-4 inducibility of CD23b, regardless of the presence of

IRF-4. These findings thus suggest that, in contrast to BCL-6, Downloaded from Blimp1 selectively targets IRF-4 function and does not interfere with Stat6 activity.

Discussion IRF-4, an IRF family member whose expression is mostly re-

stricted to lymphocytes (1–3), has been shown by genetic studies http://www.jimmunol.org/ to play a crucial role in mature B cell function (7). Consistent with the phenotype displayed by the IRF-4-deficient mice, IRF-4 ex- pression is up-regulated in response to B cell activation stimuli (2, 4–6). Our previous studies have indicated that IRF-4 function can be regulated by BCL-6, a Kru¨ppel-type zinc finger transcriptional repressor (27, 35, 36). In this study, we report that IRF-4 function FIGURE 3. Blimp1 interacts with IRF-4. A, GST-IRF-4 fusion protein can also be modulated by a distinct Kru¨ppel-type zinc finger pro- binds Blimp1. Nuclear extracts from 293T cells transiently transfected ei- tein, Blimp1/PRDI-BF1, previously shown to be a critical regula- ther with an empty vector or with an HA epitope-tagged Blimp1 expression by guest on September 29, 2021 vector were prepared and incubated with immobilized GST-IRF-4 fusion tor of terminal B cell differentiation (22). Whereas expression of protein. Bound proteins were eluted, resolved by 7% SDS-PAGE, blotted BCL-6 is mostly restricted to germinal center B cells (19), high onto a nitrocellulose membrane, and then probed with an anti-HA Ab. levels of Blimp-1 are primarily found in terminally differentiated Pull-downs with GST-alone or from Ramos extracts are shown as controls. cells (22). Thus, the ability of IRF-4 to pair with developmentally B, Coimmunoprecipitation of Blimp1 with IRF-4. The 293T cells were restricted sets of Kru¨ppel-type zinc finger proteins may regulate cotransfected with an HA epitope-tagged Blimp1 expression vector as well IRF-4 activity in a stage-appropriate manner. Furthermore, be- as an IRF-4 expression vector. 293T cells cotransfected with empty vectors cause Blimp1 expression can be detected in nonhemopoietic cells were simultaneously set up as control. Nuclear extracts were then prepared (25), interaction of Blimp1 with IRF-4 may confer lineage speci- and immunoprecipitated with either an IRF-4 antiserum or an anti-HA Ab. ficity to the actions of Blimp1. The immunoprecipitated proteins were resolved by 7% SDS-PAGE and The interaction of IRF-4 with Blimp1 may have broad biolog- then analyzed by Western blotting using an anti-IRF-4 Ab (upper panel). The blot was later stripped and reprobed with an anti-HA Ab (lower panel). ical implications. Indeed, our competition experiments suggest that this interaction is not restricted solely to the regulation of CD23 gene expression, but may extend to other IRF-4 target genes. In- terestingly, one of these genes is CD20, a B cell surface marker, Blimp1 selectively blocks the ability of IRF-4 to transactivate which, in humans, is differentially expressed in memory cells CD23b (CD20ϩ) vs plasma cells (CD20Ϫ) (37). Because IRF-4 has been To determine whether Blimp1/PRDI-BF1 could repress the ability shown to participate in the control of CD20 (9), this finding sug- of IRF-4 to transactivate the CD23b promoter, we performed tran- gests that modulation of IRF-4 function by Blimp1 may constitute sient transfection assays in U937 cells, a monocytic cell line that one of the molecular events driving activated B cells toward a is capable of activating Stat6 in response to IL-4, but lacks IRF-4, specific cell fate. BCL-6, and Blimp1 (Fig. 5A). Consistent with our previous results Our studies reveal not only similarities between the two Kru¨p- (4), cotransfection of an IRF-4 expression vector with a luciferase pel-type Zinc finger proteins, but also profound differences be- reporter construct driven by the CD23b promoter resulted in a tween them. In both cases, association with IRF-4 maps to a region 3-fold induction in luciferase activity. As in the case of BCL-6, of IRF-4 that contains its DNA binding domain (1). However, coexpression of Blimp1 with IRF-4 repressed the ability of IRF-4 physical interaction of IRF-4 with BCL-6, but not with Blimp1, to transactivate the CD23b promoter reporter construct. A similar also requires a region of IRF-4 that mediates its association with inhibitory effect was also detected when the human homologue of cofactors such as PU.1 (33) or Stat6 (Fig. 4A). This finding sug- Blimp1, PRDI-BF1, was cotransfected with IRF-4. The effect of gests that BCL-6 may modulate not only the DNA-binding ability Blimp1 on IRF-4-mediated transactivation was dose dependent be- of IRF-4, but also its ability to interact with cofactors and possibly cause addition of increasing amounts of the Blimp1 expression to assemble into multiprotein complexes. In contrast, Blimp1 may vector led to a progressive decrease in reporter activity (Fig. 5B). primarily target the IRF-4 DNA binding domain. Because plasma The Journal of Immunology 6109

FIGURE 4. Identification of IRF-4 do- main sequences important for interaction with Blimp1 and BCL-6. A, Binding of various GST-IRF-4 mutant proteins with Blimp1 or BCL-6. To examine the IRF-4 domains mediating the IRF-4/Blimp1 in- teraction, nuclear extracts from 293T cells transfected with an HA epitope-tagged Blimp1 expression vector were incubated with various GST-IRF-4 mutant proteins immobilized onto reduced glutathione-aga- rose beads, as indicated. Bound proteins were eluted, fractionated by 7% SDS- PAGE, transferred to a nitrocellulose membrane, and then probed with an anti- HA Ab (upper panel). Binding to immobi- lized GST alone is shown as a control. To assess the IRF-4 regions involved in the Downloaded from IRF-4/BCL-6 interaction, nuclear extracts from unstimulated Ramos cells were incu- bated with the same panel of GST-IRF-4 mutant proteins as above. Bound proteins were eluted, fractionated by 7% SDS- PAGE, transferred to a nitrocellulose

membrane, and then probed with an anti- http://www.jimmunol.org/ BCL-6 Ab (middle panel). The blot was then stripped and reprobed with a Stat6 Ab (lower panel). B, Summary of the deletion mutation analysis. The IRF-4 mutant pro- teins used in the GST pull-down assays are diagrammed with the deleted sequences in- dicated on the left of each mutant. The po- sitions of the DNA binding domain (DBD), the proline-rich region (PRO-RICH), and the glutamine-rich region (GLN-RICH) are by guest on September 29, 2021 indicated. The relative ability of each mu- tant to interact with Blimp1, BCL-6, or Stat6 is shown on the right of each mutant. ϩϩ indicates binding comparable with that demonstrated by wt IRF-4; ϩ indi- cates a diminished interaction; Ϫ indicates no interaction.

cells express high levels of both IRF-4 and Blimp1, we suspect that CD23a and CD23b promoters can efficiently compete Blimp1, we the interaction of IRF-4 with Blimp1 does not necessarily lead to have so far been unable to detect any targeting of the CD23a iso- repression of IRF-4 function, but it may rather direct IRF-4 toward form by BCL-6 (S.G., unpublished observations). A selective ef- stage-appropriate targets by modulating its DNA-binding proper- fect of BCL-6 only on the CD23b (ITIM-less) isoform may thus ties. Thus, one may predict that, in a different promoter context, explain the failure of microarray analysis to detect CD23 as a cooperative interactions between IRF-4 and Blimp1 may also be BCL-6 target gene (38). It is intriguing to speculate that employ- observed. Another important distinction uncovered by our studies ment of different mechanisms for regulating the two CD23 iso- is that, in contrast to BCL-6 (34), Blimp1 selectively represses forms may allow a B cell to modulate the ratio, and thus the sig- IRF-4 function and neither targets Stat6 binding sites nor blocks naling outcome, of this pair of potentially inhibitory/activating Stat6-mediated transactivation. Thus, it will be interesting to de- receptors. termine whether terminally differentiated B cells possess distinct Consistent with the central role exerted by IRF-4 in B cell ac- mechanisms to modulate Stat6 activity. tivation, deregulation of IRF-4 expression has been postulated to Although our studies suggest that all of the IRF-4 binding sites play a role in a variety of lymphoid malignancies (5, 39). Trans- tested could serve as targets for both BCL-6 and Blimp1 (Fig. 2C), locations of the BCL-6 gene leading to inappropriate expression of the regulation of IRF-4-mediated gene expression is likely to dis- BCL-6 are a common event in non-Hodgkin’s lymphomas (35, play additional complexities. In particular, different subsets of 40–42). Interestingly, deletion of chromosome 6q21-q22.1, which IRF-4 target genes may exist, which can be selectively targeted by contains the PRDI-BF1/Blimp1 gene, has been consistently de- different combinations of Kru¨ppel-type zinc finger proteins. Con- tected in high-grade non-Hodgkin’s lymphoma and PRDI-BF1/ sistent with this notion, although the IRF-4 binding sites in both Blimp1 has been suggested to be a candidate B-NHL suppressor 6110 MODULATION OF IRF-4 FUNCTION BY BLIMP1

FIGURE 5. Effects of Blimp1 on the transactivation of CD23b. A, U937 cells were cotransfected with a CD23b promoter luciferase reporter construct and an IRF-4 expression plasmid in the presence of a BCL-6, a Blimp1, or a PRDI-BF1 expres- sion vector, or equivalent amounts of empty vector, as indicated. The transfected cells were left unstimulated for 16 h. The data are presented relative to the activity of the reporter construct in control U937 cells, which was set to 1, as indicated, in each experiment. Results show the mean Ϯ SE of four independent experiments. B, U937 cells were cotransfected with a Downloaded from CD23b promoter luciferase reporter con- struct and an IRF-4 expression plasmid in the presence of increasing amounts of a Blimp1 expression vector or equivalent amounts of empty vector, as indicated. The data were analyzed as described above. Re- Ϯ

sults show the mean SE of three inde- http://www.jimmunol.org/ pendent experiments. C, U937 cells were cotransfected with a CD23b promoter lu- ciferase reporter construct and either an IRF-4 expression vector or an empty vec- tor. The cells were cotransfected in the presence of an expression plasmid encod- ing either BCL-6 (middle panel) or Blimp1 (right panel), or of equivalent amounts of control vector (left panel). The transfected cells were equally split into two 2-ml ali- by guest on September 29, 2021 quots and then incubated for 16 h in the presence or absence of IL-4 (10 ng/ml). The data are presented relative to the ac- tivity of the reporter construct in control U937 cells, which was set to 1, as indi- cated, in each experiment. Results show the mean Ϯ SE of four independent experiments.

gene (43). Our findings suggest that either persistence of BCL-6 References expression or loss of Blimp1 may lead to the inappropriate regu- 1. Eisenbeis, C., H. Singh, and U. Storb. 1995. Pip, a novel IRF family member, is lation of IRF-4 function. This may result in an activated or par- a lymphoid-specific, PU.1-dependent transcriptional activator. Genes Dev. tially activated B cell, which cannot successfully complete its dif- 9:1377. 2. Matsuyama, T., A. Grossman, H. Mittrucker, D. Siderovski, F. Kiefer, ferentiation program. In addition to the recently described ability T. Kawakami, C. Richardson, T. Taniguchi, S. Yoshinaga, and T. Mak. 1995. of BCL-6 to repress Blimp1 expression and thus block terminal B Molecular cloning of LSIRF, a lymphoid-specific member of the interferon reg- cells differentiation (38), disturbances in the IRF-4/Kru¨ppel inter- ulatory factor family that binds the interferon-stimulated response element (ISRE). Nucleic Acids Res. 23:2127. action may thus also contribute to lymphomagenesis. 3. Yamagata, T., J. Nishida, T. Tanaka, R. Sakai, K. Mitani, T. Taniguchi, Y. Yazaki, and H. Hirai. 1996. A novel interferon regulatory factor family tran- Acknowledgments scription factor, ICSAT/Pip/LSIRF, that negatively regulates the activity of in- terferon-regulated genes. Mol. Cell. Biol. 16:1283. We thank Dr. Hirai for the anti-IRF-4 antiserum. We are grateful to Man 4. Gupta, S., M. Jiang, A. Anthony, and A. Pernis. 1999. Lineage-specific modu- Jiang for her technical assistance and Drs. Dalla-Favera and Calame and lation of IL-4 signaling by interferon regulatory factor 4. J. Exp. Med. 190:1837. members of their laboratories for providing us with cell lines, the BCL-6, 5. Alizadeh, A. A., M. B. Eisen, R. E. Davis, C. Ma, I. S. Lossos, A. Rosenwald, and the Blimp1 constructs and critical discussions. We also thank Dr. G. J. C. Boldrick, H. Sabet, T. Tran, X. Yu, et al. 2000. Distinct types of diffuse large Siu for his critical reading of this manuscript. B-cell lymphoma identified by gene expression profiling. Nature 403:503. The Journal of Immunology 6111

6. Glynne, R., S. Akkaraju, J. I. Healy, J. Rayner, C. C. Goodnow, and D. H. Mack. 27. Chang, C., B. Ye, R. Chaganti, and R. Dalla-Favera. 1996. BCL-6, aPOZ/zinc- 2000. How self-tolerance and the immunosuppressive drug FK506 prevent B-cell finger protein, is a sequence-specific transcriptional repressor. Proc. Natl. Acad. mitogenesis. Nature 403:672. Sci. USA 93:6947. 7. Mittrucker, H., T. Matsuyama, A. Grossman, T. Kundig, J. Potter, A. Shahinian, 28. Perkel, J. M., and M. L. Atchison. 1998. A two-step mechanism for recruitment A. Wakeham, B. Patterson, P. Ohashi, and T. Mak. 1997. Requirement for the of Pip by PU.1. J. Immunol. 160:241. factor LSIRF/IRF4 for mature B and T lymphocyte function. Sci- 29. Kotanides, H., and N. C. Reich. 1993. Requirement of tyrosine phosphorylation ence 275:540. for rapid activation of a DNA binding factor by IL-4. Science 262:1265. 8. Pongubala, J. M. R., S. Nagulapalli, M. J. Klemsz, S. R. McKercher, R. A. Maki, 30. Goldstein, M., J. Hoxie, D. Zembryki, D. Matthews, and A. I. Levinson. 1985. and M. L. Atchinson. 1992. PU.1 recruits a second nuclear factor to a site im- Phenotypic and functional analysis of B cell lines from patients with multiple portant for immunoglobulin ␬ 3Ј enhancer activity. Mol. Cell. Biol. 12:368. 9. Himmelmann, A., A. Riva, G. L. Wilson, B. P. L. C. Thevenin, and J. H. Kehrl. myeloma. Blood 66:444. 1997. PU.1/Pip and basic helix loop helix zipper transcription factors interact 31. Tuscano, J. M., K. M. Druey, A. Riva, J. Pena, C. B. Thompson, and J. H. Kehrl. with binding sites in the CD20 promoter to help confer lineage- and stage-specific 1996. Bcl-x rather than Bcl-2 mediates CD40-dependent centrocyte survival in expression of CD20 in B lymphocytes. Blood 90:3984. the germinal center. Blood 88:1359. 10. Bonnefoy, J.-Y., S. Lecoanet-Henchoz, J.-P. Aubry, J.-F. Gauchat, and P. Graber. 32. Harris, M. B., C.-C. Chang, M. T. Berton, N. N. Danial, J. Zhang, D. Kuehner, 1995. CD23 and B-cell activation. Curr. Opin. Immunol. 7:355. B. H. Ye, M. Kvatyuk, P. P. Pandolfi, G. Catoretti, et al. 1999. Transcriptional 11. Conrad, D. 1990. Fc⑀RII/CD23: the low affinity for IgE. Annu. Rev. repression of Stat6-dependent interleukin-4-induced genes by BCL-6: specific Immunol. 8:623. regualtion of I⑀ transcription and immunoglobulin E switching. Mol. Cell. Biol. 12. Richards, M., and D. Katz. 1991. Biology and chemistry of the low affinity IgE 19:7264. receptor (Fc⑀RII/CD23). Crit. Rev. Immunol. 11:65. 33. Ortiz, M. A., J. Light, R. A. Maki, and N. Assa-Munt. 1999. Mutation analysis 13. Raulet, D., and W. Held. 1995. Natural killer receptors: the offs and ons of NK of the Pip interaction domain reveals critical residues for protein-protein inter- cell recognition. Cell 82:697. actions. Proc. Natl. Acad. Sci. USA 96:2740. 14. Yokoyama, W. 1997. What goes up must come down: the emerging spectrum of 34. Dent, A. L., A. L. Shaffer, X. Yu, D. Allman, and L. M. Staudt. 1997. Control of inhibitory receptors. J. Exp. Med. 186:1803. inflammation, cytokine expression, and germinal center formation by BCL-6. 15. Kikutani, H., and T. Kishimoto. 1990. Molecular and biology of two Science 276:589. different species of Fc⑀RII. Res. Immunol. 141:249. Downloaded from 35. Ye, B., F. Lista, F. L. Coco, D. Knowles, K. Offit, R. Chaganti, and 16. Vivier, E., and M. Daeron. 1997. Immunoreceptor tyrosine-based inhibition mo- R. Dalla-Favera. 1993. Alterations of a zinc finger-encoding gene, BCL-6, in tifs. Immunol. Today 18:286. diffuse large-cell lymphoma. Science 262:747. 17. Unkeless, J., and J. Jin. 1997. Inhibitory receptors, ITIM sequences and phos- phatases. Curr. Opin. Immunol. 9:338. 36. Seyfert, V., D. Allman, Y. He, and L. Staudt. 1996. Transcriptional repression by 18. Allman, D., A. Jain, A. Dent, R. Maile, T. Selvaggi, M. Kehry, and L. Staudt. the proto-oncogene BCL-6. Oncogene 12:2331. 1996. BCL-6 expression during B-cell activation. Blood 87:5257. 37. Liu, Y.-J. 1997. Tracing antigen-driven B cell development in humans. Chem. 19. Catoretti, G., C.-C. Chang, K. Cechova, J. Zhang, B. H. Ye, B. Falini, Immunol. 67:14.

D. C. Louie, K. Offit, R. S. K. Chaganti, and R. Dalla-Favera. 1995. BCL-6 38. Shaffer, A. L., X. Yu, Y. He, J. Boldrick, E. P. Chan, and L. M. Staudt. 2000. http://www.jimmunol.org/ protein is expressed in germinal center B cells. Blood 1:45. BCL-6 represses genes that function in lymphocyte differentiation, inflammation, 20. Catoretti, G., C. Chang, K. Cechova, J. Zhang, B. Ye, B. Falini, D. Louie, and cell cycle control. Immunity 13:199. K. Offit, R. Chaganti, and R. Dalla-Favera. 1997. Down-regulation of BCL-6 39. Iida, S., P. Rao, M. Butler, P. Corradini, M. Boccadoro, B. Klein, R. Chaganti, gene expression by CD40 and EBV latent membrane protein-1 (LMP1) and its and R. Dalla-Favera. 1997. Deregulation of MUM1/IRF4 by chromosomal trans- block in lymphoma carrying BCL-6 rearrangements. Blood 90(Suppl. 1):175a. location in multiple myeloma. Nat. Genet. 17:226. 21. Banchereau, J., and F. Rousset. 1992. Human B lymphocytes: phenotype, pro- 40. Baron, B. W., G. Nucifora, N. McCabe, R. Espinosa, M. M. L. Beau, and liferation, and differentiation. Adv. Immunol. 52:125. T. W. McKeithan. 1993. Identification of the gene associated with the recurring 22. Turner, C. A., D. Mack, and M. M. Davis. 1994. Blimp-1, a novel zinc finger- chromosomal translocations t(3;14)(q27;q32) and t(3;22)(q27;q11) in B-cell lym- containing protein that can drive the maturation of B lymphocytes into immu- phomas. Proc. Natl. Acad. Sci. USA 90:5262. noglobulin-secreting cells. Cell 77:297. 23. Lin, Y., K. Wong, and K. Calame. 1997. Repression of c-myc transcription by 41. Kerckaert, J. P., C. Deweindt, H. Tilly, S. Quief, G. Lecocq, and C. Bastard. 1993. LAZ3, a novel zinc-finger encoding gene, is disrupted by recurring chro- Blimp1, an inducer of terminal B cell differentiation. Science 276:596. by guest on September 29, 2021 24. Piskurich, J. F., K. Lin, Y. Lin, Y. Wang, J. P.-Y. Ting, and K. Calame. 2000. mosome 3q27 translocations in human lymphomas. Nat. Genet. 5:66. Blimp1/PRDI-BF1 mediates extinction of major histocompatibility class II trans- 42. Miki, T., N. Kawamata, S. Hirosawa, and N. Aoki. 1994. Gene involved in the activator (CIITA) expression in plasma cells. Nat. Immun. 1:526. 3q27 translocation associated with B-cell lymphoma, BCL5, encodes a Kru¨ppel- 25. Keller, A. D., and T. Maniatis. 1991. Identification and characterization of a like zinc-finger protein. Blood 83:26. novel repressor of ␤-interferon gene expression. Genes Dev. 5:868. 43. Mock, B. A., L. Liu, D. L. Paslier, and S. Huang. 1996. The B-lymphocyte 26. Angelin-Duclos, C., G. Catoretti, K.-I. Lin, and K. Calame. 2000. Commitment maturation promoting BLIMP1/PRDI-BF1 maps to D6S447 of B lymphocytes to a plasma cell fate is associated with Blimp-1 expression in on human chromosome 6q21–q22.1 and the syntenic region of mouse chromo- vivo. J. Immunol. 165:5462. some 10. Genomics 37:24.