The transit of T cell-activated B cells through the germinal center (GC) is controlled by sequential activation and repression of key transcription factors, executing the pre- and post-GC B cell program. B cell lymphoma (BCL) 6 and IFN regulatory factor (IRF) 8 are necessary for GC formation and for its molecular activity in Pax5+PU.1+ B cells. IRF4, which is highly expressed in BCL6 GC B cells, is necessary for class switch recombination and the plasma cell differentiation at exit from the GC. In this study, we show at the single-cell level broad coexpression of IRF4 with BCL6, Pax5, IRF8, and PU.1 in pre- and post-GC B cells in human and mouse. IRF4 is down-regulated in BCL6+ human GC founder cells (IgD+CD38+), is absent in GC centroblasts, and is re-expressed in positive regulatory domain 1-positive centrocytes, which are negative for all the B cell transcription factors. Activated (CD30+) and activation-induced cytidine deaminase-positive extrafollicular blasts coexpress Pax5 and IRF4. PU.1-negative plasma cells and CD30+ blasts uniquely display the conformational epitope of IRF4 recognized by the MUM1 Ab, an epitope that is absent from any other IRF4+PU.1+ lymphoid and hemopoietic subsets. Low grade B cell lymphomas, representing the malignant counterpart of pre- and post-GC B cells, accordingly express IRF4. However, a fraction of BCL6+ diffuse large B cell lymphomas express IRF4 bearing the MUM1 epitope, indicative of a posttranscriptional modification of IRF4 not seen in the normal counterpart.

The development and maturation of B cells from multipotential stem cells is dictated by multiple interacting transcription factors (TF),2 active at different defining steps throughout the process (1, 2, 3). Pax5 is the single TF necessary and sufficient for B cell development past the pro-B cell stage (4), a negative regulator of the B cell to plasma cell transition (5), and is expressed throughout all B cell stages, except in plasma cells (6). PU.1, a member of the large ets family of TF (7), enforces the B cell lineage commitment by limiting other differentiation choices (7, 8), and is then expressed along the B cell, but not the plasma cell lineage (9), together with other B cell-specific TF (9, 10). PU.1 heterodimerizes with members of the IFN regulatory factor family (IRFs) (10, 11, 12) in cell- and differentiation stage-specific combinations. IRF8 promotes genes crucial for germinal center (GC) development and function (B cell lymphoma (BCL) 6 and activation-induced cytidine deaminase (AID)) (13) and, with IRF4, is needed for pre-B to B cell transition (14). IRF4 has been shown to be expressed at high levels in centrocytes, believed to be at the preplasma cell stage (15). Lack of IRF4 prevents full mature B cell transit and plasma cell differentiation but not memory B cell development (16, 17). BCL6, which is highly expressed and necessary for GC B cell development (18, 19), has an inverse distribution with regards to IRF4 inside the GC (1, 15). The relative expression of these TF correlates roughly with the different molecular subtypes of diffuse large B cell lymphomas (DLBCL), including a group of GC-derived neoplasms characterized by V gene mutation and lack of terminal differentiation (20). Detailed molecular analysis has shown that IRF4 expression defines a group of DLBCL characterized by the expression of activation-associated genes (21, 22), low BCL6 expression (21), with frequent polysomy or rearrangement of chromosome 3q27 (23, 24, 25) and deletion of positive regulatory domain 1 (PRDM1) (26). Coexpression of BCL6 and IRF4 can be seen in these DLBCLs (15), as opposed to the normal GC, where the two are rigorously mutually exclusive (15). Pre- (mantle cell) and post-GC (marginal zone) type lymphomas are usually negative for BCL6 (27, 28) and MUM1 (29, 30), similar to the normal counterpart B cells (mantle and marginal zone B cells, respectively) (15), which are negative for both by routine immunohistochemistry (IHC).

Comparison of mRNA transcription with posttranslational expression of the final protein product of these TF shows discrepant results. In normal B cells, BCL6 shows posttranscriptional down-regulation outside the GC (31), thus accounting for the general lack of BCL6 detection in a range of mature B cells, including pre- and post-GC B cells, which are known to possess mRNA (18, 32, 33, 34). Similarly, IRF4, which is necessary for mature naive B cell and dendritic cell development (14, 35), is usually negative in these cells by IHC staining using the MUM1 mAb (15).

In this study, we show the expression of several TF in mouse and human lymphoid cells, with a special emphasis on BCL6 and IRF4. We show broad coexpression of all TF before, but not within, the GC reaction. In addition, we demonstrate that IRF4 is expressed at the protein level in a much broader range of mature B cell types, as detected by other Abs not limited to a conformational epitope recognized by the MUM1 Ab.

IRF4 knockout (17), BCL6 knockout mice (19), and their wild-type littermates (C57BL6 and F1 from C57BL6 × 129Sv) were gifts from U. Klein and R. Dalla-Favera (Columbia University, New York, NY). Anonymous tonsil cell suspensions and tumor samples were provided and processed as published (36). A tissue microarray containing B cell lymphomas and reference tissues was published previously (37).

Abs used in this study are listed in Table I. Multiple different anti-BCL6 Abs, which all reacted identically in various assays and combinations, will be referred to collectively as “BCL6.” The term “MUM1” will be restricted to anti-IRF4 Abs directed against aa 128–267 or aa 144–451 of the human protein.

Table I.

Abs used

AbClone or SerumImmunogenaReactive onbSpeciesSourceDilution
PU.1 (Spi1) sc-352 C-term mouse h, m Rabbit SCBT 1 μg/ml 
PU.1 (Spi1) sc-5949 N-term mouse Goat SCBT 0.1 μg/ml 
PU.1 (Spi1) G148-74 Full length h, m Mouse BD 1 μg/ml 
IRF4 sc-6059 C-term mouse h, m Goat SCBT 0.2 μg/ml 
IRF4 No. 4964 Asp175 Rabbit CST 1/100 
IRF4 sc-28696 AA 128–267h h, m Rabbit SCBT 1 μg/ml 
IRF4 MUM1 AA 144–451h Mouse B. Falini 1/50 
IRF8 (ICSBP) sc-6058 C-term mouse h, m Goat SCBT 1 μg/ml 
BCL6 PIF6 N-term human h, m Mouse Novocastra 1/100 
BCL6 PGB6 276 N-term human h, m Mouse B. Falini 1/10 
BCL6 PGB6p 594 N-term human Mouse B. Falini 1/10 
BCL6 sc-858 N-term human h, m Rabbit SCBT 0.1 μg 
BCL6 No. 4242 C-term human h, m Rabbit CST 1/100 
Pax-5 24 AA 151–306h h, m Mouse BD 1 μg/ml 
Pax-5 sc-1974 C-term human h, m Goat SCBT 1 μg/ml 
Pax-5 RB-9406 C-term human h, m Rabbit LabVision 1 μg/ml 
Oct-2 sc-233 C-term human h, m Rabbit SCBT 0.2 μg/ml 
CD20 RB-9013 C-term human Rabbit LabVision 1/200 
CD20 L26 Cytoplasm. CD20 Mouse DakoCytomation 1/200 
MCM7 47DC141 hCDC47 h, m Mouse LabVision 0.5 μg/ml 
Ki-67 SP6 C-term human h, m Rabbit LabVision 1/100 
Ki-67 MIB 1 rKi-67 Mouse J. Gerdes 1/50 
PRDM1 Serum AA 1–350 h, m Rabbit H.-M. Jäck 1/5000 
PRDM1 3H2E8 AA 199–409 h, m Mouse K. L. Calame 1/30 
PRDM1 6D3 rBlimp-1 h, m Rat L. Corcoran 1 μg/ml 
AID EK2-5G9 AA 185–198 Rat E. Kremmer 1/200 
IRTA1 mIRTA1 AA 102–373 Mouse B. Falini 1/10 
Neg. control    Rabbit Sigma-Aldrich 1 μg/ml 
Neg. control    Mouse Sigma-Aldrich 1 μg/ml 
AbClone or SerumImmunogenaReactive onbSpeciesSourceDilution
PU.1 (Spi1) sc-352 C-term mouse h, m Rabbit SCBT 1 μg/ml 
PU.1 (Spi1) sc-5949 N-term mouse Goat SCBT 0.1 μg/ml 
PU.1 (Spi1) G148-74 Full length h, m Mouse BD 1 μg/ml 
IRF4 sc-6059 C-term mouse h, m Goat SCBT 0.2 μg/ml 
IRF4 No. 4964 Asp175 Rabbit CST 1/100 
IRF4 sc-28696 AA 128–267h h, m Rabbit SCBT 1 μg/ml 
IRF4 MUM1 AA 144–451h Mouse B. Falini 1/50 
IRF8 (ICSBP) sc-6058 C-term mouse h, m Goat SCBT 1 μg/ml 
BCL6 PIF6 N-term human h, m Mouse Novocastra 1/100 
BCL6 PGB6 276 N-term human h, m Mouse B. Falini 1/10 
BCL6 PGB6p 594 N-term human Mouse B. Falini 1/10 
BCL6 sc-858 N-term human h, m Rabbit SCBT 0.1 μg 
BCL6 No. 4242 C-term human h, m Rabbit CST 1/100 
Pax-5 24 AA 151–306h h, m Mouse BD 1 μg/ml 
Pax-5 sc-1974 C-term human h, m Goat SCBT 1 μg/ml 
Pax-5 RB-9406 C-term human h, m Rabbit LabVision 1 μg/ml 
Oct-2 sc-233 C-term human h, m Rabbit SCBT 0.2 μg/ml 
CD20 RB-9013 C-term human Rabbit LabVision 1/200 
CD20 L26 Cytoplasm. CD20 Mouse DakoCytomation 1/200 
MCM7 47DC141 hCDC47 h, m Mouse LabVision 0.5 μg/ml 
Ki-67 SP6 C-term human h, m Rabbit LabVision 1/100 
Ki-67 MIB 1 rKi-67 Mouse J. Gerdes 1/50 
PRDM1 Serum AA 1–350 h, m Rabbit H.-M. Jäck 1/5000 
PRDM1 3H2E8 AA 199–409 h, m Mouse K. L. Calame 1/30 
PRDM1 6D3 rBlimp-1 h, m Rat L. Corcoran 1 μg/ml 
AID EK2-5G9 AA 185–198 Rat E. Kremmer 1/200 
IRTA1 mIRTA1 AA 102–373 Mouse B. Falini 1/10 
Neg. control    Rabbit Sigma-Aldrich 1 μg/ml 
Neg. control    Mouse Sigma-Aldrich 1 μg/ml 
a

C-term, C-terminal; N-term, N-terminal; Cytoplasm., Cytoplasmic.

b

h, Human; m, murine.

Sections were essentially stained as previously published (36, 38), including double staining on 1 mM EDTA (pH 8) Ag-retrieved slides. Species-specific, alkaline phosphatase-conjugated secondary Abs, prescreened for specificity and absence of cross-reactivity, were developed with nitroblue tetrazolium chloride/5-bromo-4-chloro-3-indolyl phosphate (Roche). The percentage and intensity of staining of CD20+ neoplastic B cells were independently scored in the DLBCL tissue microarray (TMA), each using a 10-tiered scale (0–9). The product of both was used as a case score and a value of 10 or greater was considered positive for χ2 calculations.

Double immunofluorescence was performed with primary Abs raised in different species or of different isotypes, counterstained with FITC- or Cy3-conjugated, species- or isotype-specific secondary Abs (Jackson ImmunoResearch Laboratories) or biotin-conjugated secondary Abs, followed by conjugated avidin (Jackson ImmunoResearch Laboratories). Tyramide signal amplification for single or double labeling (39) was performed according to the manufacturer’s instructions (PerkinElmer): briefly, avidin-HRP-counterstained immunostains were followed by tyramide-FITC or -biotin and Avidin Cy-3 (Jackson ImmunoResearch Laboratories) or Avidin-Alexa350 (Molecular Probes). The slides were counterstained with 4′,6′-diamidino-2-phenylindole (DAPI) (Molecular Probes) when indicated and mounted.

Gray scale or color images were taken on an E600-Nikon Microscope, fitted with Planachromat ×4/0.10/30.0, ×10/0.25/10.5, ×20/0.40/1.3, PlanApo ×40/0.095/0.12–0.16 (light microscopy) or PlanFluor ×10/0.30/16.0, ×40/0.75/0.72, ×60/0.80/0.3 (immunofluorescence) objectives, with a SPOT-2 charge-coupled device camera and software (Diagnostic Instruments). A Zeiss LSM 510 NLO Multiphoton Confocal Microscope (Carl Zeiss) equipped with ×10/0.3 and ×40/1.3 Fluor objectives, one 25-mW argon laser exciting at 458, 488, and 514 nm and one 1 mW helium-neon laser exciting at 543 nm, and proprietary image acquisition software was used for confocal analysis.

All images were edited for optimal color contrast with Adobe Photoshop 7 and Adobe Illustrator 10 (Adobe Systems), on a G4 Apple computer.

Live human or murine cell suspensions were stained with appropriate combinations of fluorochrome- or biotin-conjugated primary Abs for 15 min on ice in PBS-BSA-NaN3, washed twice with the same medium and further processed for nuclear staining. For BCL6 staining, cells were fixed with FACS lysing solution (BD Biosciences) for 10 min on ice, washed twice in PBS-BSA-NaN3, incubated overnight with negative mouse IgG1 (1 μg/ml) or PGB6 clone 276 (1/1000) + P1F6 (1/100) mouse IgG1 anti BCL6 mixture. After two washes in PBS-BSA-NaN3, the cells were counterstained with rat anti-mouse IgG1 PE-(1/200) or biotin-conjugated (1/300; BD Pharmingen), this latter followed by fluorochrome-conjugated avidin (1/500). IRF4, Pax5, and PU.1 Ab stainings requires a modified fixation, which destroys prebound fluorochromes and some surface Ags, but not biotin. Therefore, cells were first stained with biotin-conjugated Abs, fixed as above, washed in PBS, fixed in absolute cold methanol for 10 min, washed in PBS, transferred to PBS-BSA-NaN3, and processed for intracellular staining. Counterstaining of the nuclear Ag was followed by fluorochrome-avidin staining of the biotin-labeled Ab and staining for Ags which survive the double fixation (e.g., IgD, CD38). IgD was stained with a PE-conjugated rat anti-mouse IgD (Southern Biotechnology Associates) or with a PE-conjugated mouse anti-human IgD (BD Biosciences), preceded by excessive cold mouse Ig blocking of residual anti-mouse IgG moieties. BCL6 staining is not affected by the methanol postfixation.

Goat and rabbit primary Abs (IRF4, PU.1, negative control) were used at 1 μg/ml, incubated overnight, and counterstained with either donkey anti-rabbit Cy5 F(ab′)2 (1/500; Jackson ImmunoResearch Laboratories) or donkey anti-goat-FITC (1/500; Jackson ImmunoResearch Laboratories). For double IRF4, PU.1 and BCL6 stainings, human cells were surface-stained first with mouse anti CD3-biotin (BD Pharmingen), followed by nuclear staining, exclusion by gating of the CD3+ cells, and analysis of the non-T cells (>95% B cells).

More than 30,000 events were acquired on a FACSCalibur (BD Biosciences) and CellQuest acquisition software and analyzed with FlowJo 6.4 (Tree Star).

Data were obtained from published databases (1), obtained by Affymetrix U95A chip analysis of normal tonsil subsets.

LSI BCL6 dual-color, break-apart rearrangement, and Spectrum Green-labeled CEP 8 probes were obtained from Abbott Molecular. Four 5-μM thick tissue sections of two TMAs were cut onto adhesive-coated slides. Paraffin-sections were baked overnight at 60°C before hybridization. TMA slides were subjected to protease treatment using paraffin pretreatment kit (Abbott Molecular). Fluorescence in situ hybridization was performed by standard methods and hybridization signals were scored on at least 200 interphase nuclei on DAPI-stained slides (26). The sensitivity of hybridization on paraffin-embedded tissues was determined by performing the same analysis on analogous sections from normal tonsils and similarly processed cell lines of known genomic asset. Cases were diagnosed as rearranged if the fraction of cells showing the rearrangement in >5% cells.

The distribution of the transcription factors Pax-5, Oct-2, PU.1, and IRF8 (Fig. 1) in mouse and human lymphoid tissues is broad and overlaps among different B cell subsets (morphologically and phenotypically defined follicular mantle, marginal, and GC cells) as shown by in situ immunostaining with Abs validated for specificity (Fig. 2). BCL6 and IRF4 instead show a mutually exclusive distribution; BCL6 is highly expressed in GC, IRF4 instead is positive in mantle and marginal zone B cells, and negative in most GC B cells. The expression of IRF4 in IRF8+, PU.1+, Oct-2+, and Pax5+ follicular mantle B cells could be seen only with Abs directed against the C-terminal and Asp175, and not with two other Abs directed against aa 128–267 of the IRF4 protein (hence named MUM1 epitope). Abs against this region of IRF4 detect only plasma cells and a fraction of GC centrocytes (15).

FIGURE 1.

B cell TF staining in mouse spleen and human tonsil. A, Serial sections of immunized mouse spleen are stained for B cell TF and proliferation-associated Ag (MCM7). Low-power (×10) images are on top, high power (×40) on bottom. oPALS, outer periarteriolar lymphoid sheet; MC, mantle cells; MZ, marginal zone. Note the reciprocal staining pattern of IRF4 vs the other TF in the GC, and the coexpression in the MZ and MC. B, Serial sections of human tonsil are stained for B cell TF and proliferation-associated Ag (MCM7). Low-power (×4) images are on top, high power (×40) in the middle. Lower panel, The U95 Affymetrix cDNA chip raw values, depicted as bars, indicating cDNA expression for the corresponding transcription factors in memory (Mem), naive, and GC-purified cells. IE, intraepithelial; SE, subepithelial; MC, mantle cell; GC, germinal center. Note the reciprocal staining pattern of IRF4 vs the other TF in the GC, and the coexpression in the MC and IE.

FIGURE 1.

B cell TF staining in mouse spleen and human tonsil. A, Serial sections of immunized mouse spleen are stained for B cell TF and proliferation-associated Ag (MCM7). Low-power (×10) images are on top, high power (×40) on bottom. oPALS, outer periarteriolar lymphoid sheet; MC, mantle cells; MZ, marginal zone. Note the reciprocal staining pattern of IRF4 vs the other TF in the GC, and the coexpression in the MZ and MC. B, Serial sections of human tonsil are stained for B cell TF and proliferation-associated Ag (MCM7). Low-power (×4) images are on top, high power (×40) in the middle. Lower panel, The U95 Affymetrix cDNA chip raw values, depicted as bars, indicating cDNA expression for the corresponding transcription factors in memory (Mem), naive, and GC-purified cells. IE, intraepithelial; SE, subepithelial; MC, mantle cell; GC, germinal center. Note the reciprocal staining pattern of IRF4 vs the other TF in the GC, and the coexpression in the MC and IE.

Close modal
FIGURE 2.

Validation of the BCL6 and IRF4 staining. A, A murine wild-type (wt) lymph node (center left) is stained by BCL6 Ab in a GC (strongly) and in the follicles (weakly). The negative control Ab and the BCL6 Ab on a BCL6 knockout (KO) lymph node are negative. A wild-type and knockout mouse spleen are stained by IRF4; the knockout spleen is negative. B, Four mice, each bearing none, one, two BCL6 alleles, and one with one C-terminally truncated allele, are analyzed in the thymus and spleen for BCL6 expression by flow cytometry. The gray histogram is an isotype-matched negative control. Note the increased intensity (rightward shift of the peak fluorescence) of BCL6 staining with increased genetic dosage. The gating for the two B cell subsets are indicated at right. The truncated allele produces increased amounts of protein, because of the lack of autoregulation (Wang et al. (52 )). C, Spleen cells from an IRF4 knockout mouse (KO) and a wild-type littermate (wt) were stained for B220, CD11b, negative control (gray histogram), or IRF4 (black line). Note absence of specific staining in the IRF4 KO mouse.

FIGURE 2.

Validation of the BCL6 and IRF4 staining. A, A murine wild-type (wt) lymph node (center left) is stained by BCL6 Ab in a GC (strongly) and in the follicles (weakly). The negative control Ab and the BCL6 Ab on a BCL6 knockout (KO) lymph node are negative. A wild-type and knockout mouse spleen are stained by IRF4; the knockout spleen is negative. B, Four mice, each bearing none, one, two BCL6 alleles, and one with one C-terminally truncated allele, are analyzed in the thymus and spleen for BCL6 expression by flow cytometry. The gray histogram is an isotype-matched negative control. Note the increased intensity (rightward shift of the peak fluorescence) of BCL6 staining with increased genetic dosage. The gating for the two B cell subsets are indicated at right. The truncated allele produces increased amounts of protein, because of the lack of autoregulation (Wang et al. (52 )). C, Spleen cells from an IRF4 knockout mouse (KO) and a wild-type littermate (wt) were stained for B220, CD11b, negative control (gray histogram), or IRF4 (black line). Note absence of specific staining in the IRF4 KO mouse.

Close modal

We then compared the levels and distribution of B cell TF proteins in human tonsil tissue with the RNA levels, analyzed by gene expression profiling in purified B cell subsets (1). As shown in Fig. 1, the RNA levels of each TF in GC, mantle, and marginal zone B cells correspond to the distribution of the respective protein Ags, including IRF4 by C-terminal and Asp175 Ab staining.

Single-color staining and tissue RNA or protein extraction do not allow a detailed analysis of the coexpression of two given Ags on a cell-by-cell basis. Therefore, we analyzed by double immunofluorescence human (Fig. 3, D–N) and murine (data not shown) samples. As predicted from the observed distribution by single-color staining, IRF4 was largely mutually exclusive with BCL6 (15), PU.1, IRF8, and Pax5 in GC, while coexpressed on PU.1+, IRF8+, and Pax5+ mantle cells.

FIGURE 3.

Coexpression pattern of IRF4 with other B cell TF in human tonsil. A, AID (brown) decorates IRF4-GC cells (✬) and extrafollicular blast with variable amounts of IRF4 (purple). Light zone IRF4+ centrocytes are AID. Original magnification, ×10; scale bar, 15 μm. B, Enlargement of portion of extrafollicular area detailing AID+ (brown) cells containing variable amounts of IRF4 (purple). The edge of a GC is shown (arrow). Original magnification, ×40; scale bar, 15 μm. C, IRTA1+ (brown) lymphocytes express IRF4 (purple) in the mantle/marginal zone (yellow arrows, enlarged in the inset), but not in the intraepithelial location (far right). ✬, GC. Original magnification, ×40; scale bar, 15 μm. D–G, Low-power (left) and high-power LZ detail (right) of coexpression or IRF4 (green) with PU.1, IRF8, and BCL6 (all red) and of PU.1 (green) and BCL6 (red). Note in D coexpression of IRF4 and PU.1 in mantle zone B cells (✬) but not in centrocytes. where IRF4, BCL6, and IRF8 are mutually exclusive. PU.1 and BCL6 show coexpression in part of the GC cells. Left, Original magnification, ×4; scale bar, 100 μm. Right, Original magnification, ×40; scale bar, 5 μm. H, CD30+ (blue) interfollicular blasts coexpress IRF4 (green) and Pax5 (red). The boxed area is magnified and the color split on the right. Note two IRF4+Pax5+CD30 blasts (arrows) at the edge of the GC (✬). Original magnification ×40, scale bar 5 μm. I, Detail of GC light zone, containing centrocytes variably expressing IRF4 (green) and PRDM1 (red). Two cells also express Pax5 (blue, arrows). One Pax5+PRDM1+ blast is shown (arrowhead). K, Plasma cells outside the GC coexpress IRF4 (green) and PRDM1 (red) but not Pax5 (blue). Note the heterogeneity in Ag expression. A plasmacytoid cell is PRDM1+IRF4. L, Light zone centrocytes are IRF4+ (green), PRDM1+ (red), and a minority express MUM1 (blue). I, K, and L, magnification, ×40; scale bar, 15 μm. M, Confocal analysis of IRF4 (green), MUM1 (red), and DAPI (blue) staining in GC light zone, showing two IRF4-only centrocytes (arrows) and, on the right, nonidentical distribution of IRF4 and MUM1 in the nuclei. Original magnification, ×40; scale bar, 5 μm. N, Light zone centrocytes show relocation of cREL (red, arrows) in IRF4 nuclei (green). Nuclei are stained with DAPI (blue). Original magnification, ×40; scale bar, 15 μm.

FIGURE 3.

Coexpression pattern of IRF4 with other B cell TF in human tonsil. A, AID (brown) decorates IRF4-GC cells (✬) and extrafollicular blast with variable amounts of IRF4 (purple). Light zone IRF4+ centrocytes are AID. Original magnification, ×10; scale bar, 15 μm. B, Enlargement of portion of extrafollicular area detailing AID+ (brown) cells containing variable amounts of IRF4 (purple). The edge of a GC is shown (arrow). Original magnification, ×40; scale bar, 15 μm. C, IRTA1+ (brown) lymphocytes express IRF4 (purple) in the mantle/marginal zone (yellow arrows, enlarged in the inset), but not in the intraepithelial location (far right). ✬, GC. Original magnification, ×40; scale bar, 15 μm. D–G, Low-power (left) and high-power LZ detail (right) of coexpression or IRF4 (green) with PU.1, IRF8, and BCL6 (all red) and of PU.1 (green) and BCL6 (red). Note in D coexpression of IRF4 and PU.1 in mantle zone B cells (✬) but not in centrocytes. where IRF4, BCL6, and IRF8 are mutually exclusive. PU.1 and BCL6 show coexpression in part of the GC cells. Left, Original magnification, ×4; scale bar, 100 μm. Right, Original magnification, ×40; scale bar, 5 μm. H, CD30+ (blue) interfollicular blasts coexpress IRF4 (green) and Pax5 (red). The boxed area is magnified and the color split on the right. Note two IRF4+Pax5+CD30 blasts (arrows) at the edge of the GC (✬). Original magnification ×40, scale bar 5 μm. I, Detail of GC light zone, containing centrocytes variably expressing IRF4 (green) and PRDM1 (red). Two cells also express Pax5 (blue, arrows). One Pax5+PRDM1+ blast is shown (arrowhead). K, Plasma cells outside the GC coexpress IRF4 (green) and PRDM1 (red) but not Pax5 (blue). Note the heterogeneity in Ag expression. A plasmacytoid cell is PRDM1+IRF4. L, Light zone centrocytes are IRF4+ (green), PRDM1+ (red), and a minority express MUM1 (blue). I, K, and L, magnification, ×40; scale bar, 15 μm. M, Confocal analysis of IRF4 (green), MUM1 (red), and DAPI (blue) staining in GC light zone, showing two IRF4-only centrocytes (arrows) and, on the right, nonidentical distribution of IRF4 and MUM1 in the nuclei. Original magnification, ×40; scale bar, 5 μm. N, Light zone centrocytes show relocation of cREL (red, arrows) in IRF4 nuclei (green). Nuclei are stained with DAPI (blue). Original magnification, ×40; scale bar, 15 μm.

Close modal

Staining with all IRF4 Abs was both nuclear and cytoplasmic, the latter was least evident with Abs directed against aa 128–267.

The GC contains morphologically and phenotypically different and well-defined zones, comprised predominantly of B cells, which correspond to different functional and/or maturational subsets. Extrafollicular B cells, however, are more dispersed and therefore better defined by double staining (36, 40) rather than by topographic location. We analyzed the coexpression of IRF4 and other B cell TF in four B cell subsets: the CD30+ activated cells, the extrafollicular AID+ blasts (36), the centrocytes in the GC light zone, and the extrafollicular memory/marginal zone IRTA1+ cells (41).

CD30+ extrafollicular blasts are largely B cells showing evidence of BCR plus cofactor-mediated acute activation (36), and IRF4 has been shown to be part of this signature (42). Accordingly, CD30+ cells were largely IRF4+ and Pax5+. The intensity of Pax5 expression was inversely related to intensity of IRF4 expression (Fig. 3 H).

B cells express AID upon in vitro activation (43) and IRF4 is necessary for AID expression (17). Approximately half of the AID+ extrafollicular blasts were IRF4+, MUM1 (Fig. 3, A and B), with the intensity of IRF4 often noticeably dim, compared with that of CD30+ extrafollicular blasts. This was in sharp contrast to the GC where AID and IRF4 were rigorously mutually exclusive both in the dark and outer zone (Fig. 3 A). Thus, appreciable levels of IRF4 are detectable in extrafollicular AID+ cells, possibly before they enter the GC reaction, when IRF4 is shut off.

Commitment to the plasma cell lineage within the GC is marked by PRDM1 expression, at first in Pax5+ centrocytes, then in CD20, Pax5, and CD138 preplasma cells (38). The bulk of bright IRF4+ centrocytes coexpress PRDM1 and are Pax5 negative (Fig. 3, I, L, and M). In addition, a minority of Pax5-positive or Pax5 weakly positive centrocytes express IRF4 and/or PRDM1 (Fig. 3,I). The MUM1 epitope is not detected on these latter cells (Fig. 3 L).

The fourth B cell subset we focused on is defined by the FCRL4/IRTA1 Ag (41, 44). IRTA1+ cells are monocytoid B cells with evidence of Ag selection and variable Ig gene mutation (44), thus memory B cells, located within the tonsil festooned epithelium and scattered in the interfollicular areas. Although intraepithelial IRTA1+ cells were largely IRF4, interfollicular and intrafollicular IRTA1+ cells contained IRF4+ (Fig. 3,C). This is consistent with the reported expression of IRF4 by gene expression profiling in purified memory B cells (1) (Fig. 1). IRTA1+ B cells were previously reported negative for IRF4 with the MUM1 Ab.

In summary, IRF4 is coexpressed with the other B cell TF in pre- and post-GC B cells, but not inside the GC, where it is specifically down-regulated during transit.

BCL6 staining in the follicle mantle is usually negative, however, occasionally we are able to detect it by IHC only in selected mouse samples (Fig. 2). BCL6 is among the most variably expressed genes in the mouse (45) and our inconsistent results may be due to either sample-to-sample variation and/or insufficient sensitivity. To assess in a quantitative fashion BCL6 and IRF4, we developed and validated (Fig. 2) a flow cytometry (FCM) assay, which has the additional advantage of allowing multiparameter phenotypic characterization of B cell subsets.

Mouse cells of defined BCL6 genomic dosage showed specific, gene-dependent, near-ubiquitous biallelic BCL6 staining in most B cell subsets (Fig. 2), thymus and myeloid cells (data not shown). In particular, follicular B cells (B220+IgD+) showed detectable BCL6, consistent with the occasional staining seen by IHC. IRF4 knockout splenic B cells were negative for IRF4 staining, compared with positive wild-type littermate B cells (Fig. 2).

As predicted by gene expression profiling of human tonsils, BCL6 and IRF4 were expressed in mantle zone B cells (CD20+, IgD+CD38/dim) (Fig. 4). Plasmacytoid cells (CD20+, IgD, CD38++) were BCL6IRF4++. On the contrary, GC cells (CD20+, IgD, CD38+) were BCL6++,IRF4. Memory B cells (CD20+, IgD, CD38) showed IRF4 positivity and traces of BCL6. Interestingly, GC founder cells (CD20+, IgD+, CD38++) (46) contained lower levels of IRF4 than mantle zone or memory B cells, and two populations of BCL6low and BCL6high B cells (Fig. 4).

FIGURE 4.

BCL6 expression in human tonsil B cells subsets. Five human B cells subsets, defined by CD38 and IgD staining on CD3-negative, CD20+ B cells, are analyzed for BCL6, IRF4, and a negative control (gray histograms). Note expression of both BCL6 and IRF4 in mantle zone B cells (IgD+CD38), reduced IRF4 and emergence of a strong BCL6+ peak in GC founder cells (IgD+CD38+), disappearance of IRF4 in GC cells (IgDCD38+), expression of IRF4 but minimal BCL6 in memory B cells (IgDCD38) and high IRF4 expression in BCL6- plasmacytoid cells (IgDCD38++).

FIGURE 4.

BCL6 expression in human tonsil B cells subsets. Five human B cells subsets, defined by CD38 and IgD staining on CD3-negative, CD20+ B cells, are analyzed for BCL6, IRF4, and a negative control (gray histograms). Note expression of both BCL6 and IRF4 in mantle zone B cells (IgD+CD38), reduced IRF4 and emergence of a strong BCL6+ peak in GC founder cells (IgD+CD38+), disappearance of IRF4 in GC cells (IgDCD38+), expression of IRF4 but minimal BCL6 in memory B cells (IgDCD38) and high IRF4 expression in BCL6- plasmacytoid cells (IgDCD38++).

Close modal

Furthermore, tonsil cell suspensions were stained for BCL6, IRF4, Pax5, CD3, and IgD, to define total (CD3), GC (IgD), and mantle zone B cells (IgD+) (Fig. 5). Pax5+,IgD GC B cells contained no IRF4 and high levels of BCL6. The Pax5+,IgD+ fraction (containing mantle zone and IgD+ GC founder cells) contained a BCL6+,IRF4+ population, corresponding to mantle zone B cells and a BCL6++, IRF4 negative or weak population, corresponding to GC founder cells. MUM1 Ab, shown to be reactive with IRF4 by FCM on lymphoma cell lines (data not shown), was negative on all these B cell fractions.

FIGURE 5.

Coexpression of BCL6, IRF4, and Pax5 in IgD+ and IgD tonsil B cells. Two fractions of tonsil CD3 B cells (IgDneg and IgDpos) are costained for IRF4, negative control, BCL6, and Pax5. The shaded gray area indicate IRF4 vs negative control. The contour image superimposed shows IRF4 vs a second marker distribution. At each side, the single distribution of the negative isotype control (gray histogram) and the respective positive Ab (black line) are represented. Note two populations, one BCL6+ IRF4 weak/negative and one BCL6 weak/IRF4 weak in the IgD+ fraction.

FIGURE 5.

Coexpression of BCL6, IRF4, and Pax5 in IgD+ and IgD tonsil B cells. Two fractions of tonsil CD3 B cells (IgDneg and IgDpos) are costained for IRF4, negative control, BCL6, and Pax5. The shaded gray area indicate IRF4 vs negative control. The contour image superimposed shows IRF4 vs a second marker distribution. At each side, the single distribution of the negative isotype control (gray histogram) and the respective positive Ab (black line) are represented. Note two populations, one BCL6+ IRF4 weak/negative and one BCL6 weak/IRF4 weak in the IgD+ fraction.

Close modal

We conclude that each B cell subset coexpressing BCL6, IRF4, and the other B cell TF is characterized by discrete protein levels of such TF, which are typical of each cell type.

We then evaluated by IHC the mutual distribution of IRF4 and BCL6 in a series of 40 DLBCL. We found 45% (18 of 40) coexpressed BCL6 and IRF4, 45% were BCL6+ only, 7.5% (3 of 40) were IRF4+ only, and 2 were negative for both (Fig. 6). A semiquantitative evaluation of the staining showed two distinctly distributed groups of predominantly BCL6- or IRF4-positive cases and a heterogeneous cluster of coexpressing cases (Fig. 6 A). A similar inversely related distribution has been obtained for the RNA expression levels by gene expression profiling (Ref. 47 and data not shown). Cases with known BCL6 translocation were low in BCL6 and (3 of 4) coexpressed IRF4.

FIGURE 6.

Distribution and coexpression of BCL6 and IRF4 (MUM1 and polyclonal Ab) in DLBCL. A, Bivariate dot plot showing the distribution of BCL6 and IRF4 in 40 DLBCL cases, according to the score of positive cells. ▦, The 10% lower limit for positivity for score (see Materials and Methods). The number of cases in each quadrant is shown in italics. Gray symbols represent cases polysomic for 3q27; asterisks represent cases with 3q27 breaks (BCL6 rearrangement). Other cases are either normal or not tested (n = 11). B, The correlation of IRF4 and MUM1 staining in 40 DLBCL cases is shown. ▦, The 10% lower limit for positivity for score.

FIGURE 6.

Distribution and coexpression of BCL6 and IRF4 (MUM1 and polyclonal Ab) in DLBCL. A, Bivariate dot plot showing the distribution of BCL6 and IRF4 in 40 DLBCL cases, according to the score of positive cells. ▦, The 10% lower limit for positivity for score (see Materials and Methods). The number of cases in each quadrant is shown in italics. Gray symbols represent cases polysomic for 3q27; asterisks represent cases with 3q27 breaks (BCL6 rearrangement). Other cases are either normal or not tested (n = 11). B, The correlation of IRF4 and MUM1 staining in 40 DLBCL cases is shown. ▦, The 10% lower limit for positivity for score.

Close modal

IRF4 and MUM1 staining were highly correlated (Fig. 6,B), with only two IRF4+ cases not expressing the MUM1 epitope (Fig. 6 B).

To understand the mutual relationship of BCL6 and IRF4 at the single-cell level, we costained for both selected double-positive DLBCL cases. A range of coexpression patterns was obtained, from mutual exclusion to substantial costaining (Fig. 7).

FIGURE 7.

Coexpression of BCL6 and IRF4 in DLBCL. Four representative cases of DLBCL (A–D), double stained for BCL6 and IRF4 (MUM1), are shown, the two stains are split and shown in black and white. Arrowheads indicate nuclei with mutually exclusive staining, arrows coexpression. Note the variety of ratios of the two Ags.

FIGURE 7.

Coexpression of BCL6 and IRF4 in DLBCL. Four representative cases of DLBCL (A–D), double stained for BCL6 and IRF4 (MUM1), are shown, the two stains are split and shown in black and white. Arrowheads indicate nuclei with mutually exclusive staining, arrows coexpression. Note the variety of ratios of the two Ags.

Close modal

A small group of low-grade lymphomas representative of the neoplastic counterpart of pre- and post-GC cells was also evaluated. Six cases of chronic lymphocytic leukemia (CLL), two mantle cell lymphomas, and two marginal zone lymphomas, showed more extensive but not consistent IRF4 staining, compared with MUM1 (Fig. 8). One CLL case was BCL6+ by IHC (data not shown). BCL6 and IRF4 expression was confirmed by FCM in two representative CLL cases, one of which showed coexpression (data not shown).

FIGURE 8.

Comparison of MUM1 and IRF4 polyclonal Ab staining on a sample of low grade non-Hodgkin lymphoma. Selected cases of low grade B cell lymphomas (CLL; MC, mantle cell lymphoma, MZ, marginal zone lymphoma) and tonsil are stained respectively with the MUM1 and the polyclonal IRF4 Ab. Identical fields on serial sections are shown. A low-power field (×4) is shown at the sides. The polygon marks the tonsil GC, the star the mantle. Note the more frequent positivity of the polyclonal IRF4 Ab.

FIGURE 8.

Comparison of MUM1 and IRF4 polyclonal Ab staining on a sample of low grade non-Hodgkin lymphoma. Selected cases of low grade B cell lymphomas (CLL; MC, mantle cell lymphoma, MZ, marginal zone lymphoma) and tonsil are stained respectively with the MUM1 and the polyclonal IRF4 Ab. Identical fields on serial sections are shown. A low-power field (×4) is shown at the sides. The polygon marks the tonsil GC, the star the mantle. Note the more frequent positivity of the polyclonal IRF4 Ab.

Close modal

Thus, B cell lymphomas, as their normal B cell counterparts, show coexpression of BCL6 and IRF4. However, in tumors of putative GC origin, such as a group of DLBCL, IRF4 bears the MUM1 conformational epitope in BCL6+ cells.

BCL6 is a potent transcriptional repressor expressed at high levels in GC cells, and presumably prevents terminal B cell differentiation by repressing key genes, such as PRDM1/Blimp-1 (48). BCL6 may also have other important functions, such as antiapoptotic activity (49). Recently, Pax5 has been shown to have an essential role in maintaining the B cell phenotype in GC cells (5). IRF8 and IRF4 are also needed to complete essential activities within the GC, such as somatic mutation and class switch recombination of Ig genes through AID induction (13, 17). The emerging picture requires an understanding of the topographic distribution of all these actors before, after, and through the GC reaction, such knowledge has been lacking or largely incomplete. It is also important to know not only the distribution but also the abundance of these factors. PU.1 is an example of that, because levels PU.1 are critical for normal physiology of multiple cell types (50). These data have deep implications in interpretation of molecular interaction data in the right normal physiologic context.

In the present study, we show that BCL6 and IRF4 are broadly coexpressed, together with Pax5, PU.1, and Oct-2, in mature B cells, before and after entering the GC reaction, a finding largely expected according to previously published molecular and immunohistochemical studies (1, 31, 51). The mutual relationship of these TF outside the GC is fine tuned by the amount of each TF in a cell type-specific fashion, as demonstrated by quantitative flow cytometry.

The role of BCL6 in pre- and post-GC B cells is unknown and probably redundant for B cell maturation and survival, because mantle zone B cells in BCL6 knockout mice are unaffected (19). In addition, the protein levels of BCL6 in pre-GC cells are maintained low by posttranscriptional down-regulation (31) and transcriptional-negative autoregulation (52). However, unimmunized BCL6-transgenic mice with only moderately raised baseline BCL6 levels in pre-GC cells experience GC formation to a level comparable to that of a potent polyclonal immunization (53), suggesting a “tonic” role for such low protein levels. What is probably crucial is the ability to rapidly increase BCL6 levels upon activation. We suggest this increase to happen in GC founder cells.

Low levels of BCL6 are also physiologically relevant in myeloid-derived cells, in which BCL6, undetectable by routine IHC, but detected by FCM, is necessary to repress an otherwise lethal immune dysregulation (33) and to control IL-6 signaling (32). Thus, BCL6 is continuously expressed at various levels before and during the GC reaction, and is eventually down-regulated in terminally differentiated plasma cells.

The presence of IRF4 in pre-GC B cells is expected because it is required for IgD+IgM+ B cell phenotypic maturation (16). The levels of IRF4 in human and murine mantle and marginal zone B cells are quantitatively lower than the levels found in plasma cells and, notably, qualitatively different. The large majority of IRF4+ cells in mouse and human tissue cannot be detected by MUM1-like Abs directed against aa 128–267 of IRF4. The fact that these Abs recognize denatured IRF4 in a Western blot indicates that the molecule is in a peculiar conformational configuration which masks the MUM1-like epitope in most B cells and myelomonocytic cells.

IRF4 is structurally modified by an immunophilin (FKBP52 or FKBP4) with peptidyl-prolyl isomerase and chaperone-like activity (54), broadly expressed in all B cell subsets at the mRNA level (data obtained from Ref. 1) which binds to IRF4 in a segment centered on aa 150–237. FKBP4 binding induces a structural conformation which reduces the interaction of IRF4 with PU.1 and the transcriptional activity on PU.1-binding target genes (54). FKBP4 binding is inhibited when the C-terminal autoinhibitory domain folds back on the N-terminal DNA-binding domain (54).

Interestingly enough, the aa 128–267/MUM1 epitope is found predominantly on PU.1-negative cells (T cells, plasma cells, and melanocytes) (15, 55) and never present in myelomonocytic cells (G. Cattoretti, manuscript in preparation) or mantle zone B cells, where IRF4 has been shown to be functionally active and required, in cooperation with PU.1 (16, 56). This suggests that the acquisition of the MUM1+ conformation by IRF4 is associated with a change in interacting partners or with reduced interaction with PU.1. Future studies with biochemical assays are needed to identify the interacting partners and the target DNA sequences.

At the entry of the GC, IRF4 is quickly down-regulated, while BCL6 is up-regulated, as demonstrated by the absence of double IRF4-BCL6-positive cells by FCM and IHC inside the GC. The situation is reversed in the light zone centrocytes committed to the plasma cell lineage. There, down-regulation of BCL6 is followed by PRDM1 expression in cells which still express Pax5. Then, Pax5, IRF8, and PU.1 are gradually down-regulated and IRF4 is substantially up-regulated in cells which have already lost the B cell program (38, 48) (Fig. 9). BCL6 is eventually totally lost on these cells. The factors initiating the exit process are unknown at the present time. We detected nuclear relocation of cREL, an NF-κB member, in BCL6+ and negative centrocytes in the light zone (Ref. 57 and data not shown), largely in IRF4-negative but also in some IRF4+ and PRDM1+ cells (38). Unfortunately, we could not investigate nuclear relocation of other NF-κB members, therefore, the role of this group of TF in down-regulating BCL6 remains elusive. Recent studies have suggested that Pax5, not BCL6, may be the key regulator to be switched off before plasma cell differentiation (5), and PRDM1 would be the TF that seals the plasma cell fate (58).

FIGURE 9.

Scheme of B cell TF distribution across the human GC transit. A, B cell Ags and TF distribution is shown from naive mantle cell (MC) B cells, through the outer and dark zone (OZ, DZ) of the GC, the GC light zone (LZ), and post GC. Only the plasma cell exit is shown. T, T cell; FDC, follicular dendritic cell; NcREL, nuclear cRel. B, B cell Ags and TF distribution in the GC light zone (LZ) and in the memory cell (left) and plasma cell (right) arms of post-GC differentiation. Extrafollicular CD30+ cells are shown on the left as precursors of both the GC and the extrafollicular memory B cell pathway. On the right, a CD30+ intermediate is hypothesized (dashed) although no evidence of plasma cell commitment has been identified so far. Vertical gray bars are drawn to identify boundaries of Ag expression per cell type or stage of differentiation. Extent of coexpression is approximate and not quantitative.

FIGURE 9.

Scheme of B cell TF distribution across the human GC transit. A, B cell Ags and TF distribution is shown from naive mantle cell (MC) B cells, through the outer and dark zone (OZ, DZ) of the GC, the GC light zone (LZ), and post GC. Only the plasma cell exit is shown. T, T cell; FDC, follicular dendritic cell; NcREL, nuclear cRel. B, B cell Ags and TF distribution in the GC light zone (LZ) and in the memory cell (left) and plasma cell (right) arms of post-GC differentiation. Extrafollicular CD30+ cells are shown on the left as precursors of both the GC and the extrafollicular memory B cell pathway. On the right, a CD30+ intermediate is hypothesized (dashed) although no evidence of plasma cell commitment has been identified so far. Vertical gray bars are drawn to identify boundaries of Ag expression per cell type or stage of differentiation. Extent of coexpression is approximate and not quantitative.

Close modal

The light zone contains Pax5+ centrocytes which lack both high levels of BCL6, as well as PRDM1 and IRF4. Some of these may be memory B cell precursors, which did not acquire yet memory B cell Ags, such as IRTA1 (41), usually not detectable in the GC. However, the lack of IRF4 would be in contrast with the gene expression and flow cytometry data that we have generated, indicating that memory B cells are IRF4+. One possibility is that Pax5+ memory B cell precursors are rapidly exported from the GC and they acquire memory B cell Ags and IRF4 outside the GC boundaries. The detection of IRF4 in IRTA1+ cells at the edge of the GC may be evidence of such maturation. Another nonexclusive possibility is that centrocytes may be quite promiscuous in terms of lineage commitment, once they reach the BCL6-negative stage. IRF4+ and/or PRDM1+ cells may still be able to enter the memory B cell lineage and would then quickly exit the GC. Promiscuity in lineage commitment is not a new concept in B and myeloid TF literature (59). The hypothesis that PRDM1 is expressed in a common memory and plasma cell precursor has been published in the past (60). Both hypotheses would require a “common centrocyte” (Fig. 9) which would then give rise to either memory or plasma cells or both.

This putative “common centrocyte” is an AID-negative cell. We have previously shown that AID+ cells in the dark and outer zone of the GC are MUM1 negative and now we confirm these data with a broader anti-IRF4 reagent. Because IRF4 is needed for class switch recombination (17), its absence in AID+ cells through the GC is in contrast with the common knowledge that the light zone is the site where such activity occurs. It is possible that IRF4 function in the GC is replaced by another TF of the same class; IRF8 and IRF4 have been previously shown to be partially redundant (56), thus IRF8 (which promotes AID expression (13)) may be a candidate. Yet, the absence of AID in centrocytes suggests that the molecular lesions initiating class switch recombination occur in a cell upstream, which would be AID+ and may express IRF4 at a certain point. We have identified putative cells fulfilling such criteria. IRF4 is expressed in rare AID+ blasts at the edge of the GC and, more conspicuously, in extrafollicular AID+ cells. We have shown that these cells are losing the phenotype of the acute BCR stimulation typical of CD30+ blasts (Myc, JunB, CCL22) and acquire characteristics that are similar to GC cells (36). Once these cells enter the GC reaction, if they do, then they may start the molecular processes, completed later in the light zone. The remarkable finding is that IRF4 is broadly coexpressed with other B cell TF outside the GC but it is repressed through most of the GC transit except for the plasma cell exit, where it acquires posttranscriptional modifications which may change its function. A fraction of diffuse large B cell lymphomas, some of which are identified as “activated B cell type” (22, 61), coexpress BCL6 and IRF4.

This phenotype, as we demonstrate in this manuscript, is not tumor specific per se, except for two distinguishing features: the levels of BCL6 are intermediate between GC (very high) and mantle zone B cells (undetectable by conventional IHC) and IRF4 is almost exclusively bearing the MUM1 epitope. In addition they are often AID+ and IgM+ (Ref. 62 and G. Cattoretti, unpublished data), suggestive of a centroblastic or outer zone origin rather than centrocytic (36).

The conformation of IRF4 resulting in the exposure of the MUM1 epitope may reduce the transcriptional activity of PU.1, as suggested by in vitro studies (54). In addition, binding of Krüppel-type zinc finger proteins such as BCL6 and PRDM1/Blimp-1 to IRF4 is mediated by partially overlapping portions of the molecule, whose posttranscriptional modification may change both the interacting partners and the effect of IRF4-transactivating ability (63).

The biological consequences of a modified IRF4 conformation are unknown. Pre-GC cells need IRF4 for terminal differentiation and never express the MUM1-like form of it (except for CD30+ blasts and some AID+ cells); IRF4 is also needed for plasma cell differentiation (17), but is unclear whether the required form in this case is the modified type found in PU.1 cells. To add to the complexity, DLBCL containing a modified IRF4 may at the same time contain deregulated BCL6 or lack PRDM1 expression, the other necessary factor to drive plasma cell differentiation (26, 63, 64). Finally, the mutual interactions between all these TF is only partially elucidated and posttranscriptional regulation will add yet another layer of complexity to this multiplayer system.

Low-grade, pre-, or post-GC human B cell lymphomas (in contrast to DLBCL) resemble their normal counterpart by expressing low levels of IRF4 and BCL6. However, there is significant variability among different cases, possibly reflecting the state of activation or the fluctuation of baseline BCL6 expression (45). From a practical point of view, use of MUM1 Ab for prognostication in DLBCL should remain valid (61), but should be reconsidered for any non-GC, PU.1+ cell type.

We thank Ulf Klein, Masumichi Saito, Riccardo Dalla-Favera, and Michael L. Shelanski for discussion, scientific support, continuous encouragement, and advice. Attilio Orazi (Indiana University, Indianapolis, IN) contributed cases to the TMA. Maryellen Benito provided outstanding technical help. Lin Yang, the Molecular Pathology Facility and the Optical Microscopy Facility, Herbert Irving Comprehensive Cancer Center, Columbia University, provided superb histology service and excellent confocal analysis. Brunangelo Falini (Perugia University, Perugia, Italy), Lynn Corcoran (Walter and Eliza Hall Institute, Melbourne, Australia), Elisabeth Kremmer (GSF-National Research Center for Environment and Health, Münich, Germany), and Johannes Gerdes (Molecular Immunology, Borstel, Germany) generously provided Abs.

The authors have no financial conflict of interest.

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 with 18 U.S.C. Section 1734 solely to indicate this fact.

2

Abbreviations used in this paper: TF, transcription factor; IRF, IFN regulatory factor; GC, germinal center; AID, activation-induced cytidine deaminase; BCL, B cell lymphoma; DLBCL, diffuse large BCL; IHC, immunohistochemistry; DAPI, 4′,6′-diamidino-2-phenylindole; TMA, tissue microarray; FCM, flow cytometry; CLL, chronic lymphocytic leukemia; PRDM1, positive regulatory domain 1.

1
Klein, U., Y. Tu, G. A. Stolovitzky, J. L. Keller, J. Haddad, Jr, V. Miljkovic, G. Cattoretti, A. Califano, R. Dalla-Favera.
2003
. Transcriptional analysis of the B cell germinal center reaction.
Proc. Natl. Acad. Sci. USA
100
:
2639
-2644.
2
Matthias, P., A. G. Rolink.
2005
. Transcriptional networks in developing and mature B cells.
Nat. Rev. Immunol.
5
:
497
-508.
3
Shaffer, A. L., A. Rosenwald, E. M. Hurt, J. M. Giltnane, L. T. Lam, O. K. Pickeral, L. M. Staudt.
2001
. Signatures of the immune response.
Immunity
15
:
375
-385.
4
Nutt, S. L., A. M. Morrison, P. Dorfler, A. Rolink, M. Busslinger.
1998
. Identification of BSAP (Pax-5) target genes in early B-cell development by loss- and gain-of-function experiments.
EMBO J.
17
:
2319
-2333.
5
Nera, K. P., P. Kohonen, E. Narvi, A. Peippo, L. Mustonen, P. Terho, K. Koskela, J. M. Buerstedde, O. Lassila.
2006
. Loss of Pax5 promotes plasma cell differentiation.
Immunity
24
:
283
-293.
6
Horcher, M., A. Souabni, M. Busslinger.
2001
. Pax5/BSAP maintains the identity of B cells in late B lymphopoiesis.
Immunity
14
:
779
-790.
7
Oikawa, T., T. Yamada.
2003
. Molecular biology of the Ets family of transcription factors.
Gene
303
:
11
-34.
8
Nutt, S. L., D. Metcalf, A. D’Amico, M. Polli, L. Wu.
2005
. Dynamic regulation of PU.1 expression in multipotent hematopoietic progenitors.
J. Exp. Med.
201
:
221
-231.
9
Nagy, M., B. Chapuis, T. Matthes.
2002
. Expression of transcription factors Pu. 1, Spi-B, Blimp-1, BSAP and oct-2 in normal human plasma cells and in multiple myeloma cells.
Br. J. Haematol.
116
:
429
-435.
10
Su, G. H., H. S. Ip, B. S. Cobb, M. M. Lu, H. M. Chen, M. C. Simon.
1996
. The Ets protein Spi-B is expressed exclusively in B cells and T cells during development.
J. Exp. Med.
184
:
203
-214.
11
Lohoff, M., T. W. Mak.
2005
. Roles of interferon-regulatory factors in T-helper-cell differentiation.
Nat. Rev. Immunol.
5
:
125
-135.
12
Sato, M., T. Taniguchi, N. Tanaka.
2001
. The interferon system and interferon regulatory factor transcription factors—studies from gene knockout mice.
Cytokine Growth Factor Rev.
12
:
133
-142.
13
Lee, C. H., M. Melchers, H. Wang, T. A. Torrey, R. Slota, C. F. Qi, J. Y. Kim, P. Lugar, H. J. Kong, L. Farrington, et al
2005
. Regulation of the germinal center gene program by interferon (IFN) regulatory factor 8/IFN consensus sequence-binding protein.
J. Exp. Med.
203
:
63
-72.
14
Lu, R., K. L. Medina, D. W. Lancki, H. Singh.
2003
. IRF-4,8 orchestrate the pre-B-to-B transition in lymphocyte development.
Genes Dev.
17
:
1703
-1708.
15
Falini, B., M. Fizzotti, A. Pucciarini, B. Bigerna, T. Marafioti, M. Gambacorta, R. Pacini, C. Alunni, L. Natali-Tanci, B. Ugolini, et al
2000
. A monoclonal antibody (MUM1p) detects expression of the MUM1/IRF4 protein in a subset of germinal center B cells, plasma cells, and activated T cells.
Blood
95
:
2084
-2092.
16
Mittrucker, H. W., T. Matsuyama, A. Grossman, T. M. Kundig, J. Potter, A. Shahinian, A. Wakeham, B. Patterson, P. S. Ohashi, T. W. Mak.
1997
. Requirement for the transcription factor LSIRF/IRF4 for mature B and T lymphocyte function.
Science
275
:
540
-543.
17
Klein, U., S. Casola, G. Cattoretti, Q. Shen, M. Lia, T. Mo, T. Ludwig, K. Rajewsky, R. Dalla-Favera.
2006
. Transcription factor IRF4 controls plasma cell differentiation and class-switch recombination.
Nat. Immunol.
7
:
773
-82.
18
Cattoretti, G., C. C. Chang, K. Cechova, J. Zhang, B. H. Ye, B. Falini, D. C. Louie, K. Offit, R. S. Chaganti, R. Dalla-Favera.
1995
. BCL-6 protein is expressed in germinal-center B cells.
Blood
86
:
45
-53.
19
Ye, B. H., G. Cattoretti, Q. Shen, J. Zhang, N. Hawe, R. de Waard, C. Leung, M. Nouri-Shirazi, A. Orazi, R. S. Chaganti, et al
1997
. The BCL-6 proto-oncogene controls germinal-centre formation and Th2-type inflammation.
Nat. Genet.
16
:
161
-170.
20
Jaffe, E. S., N. L. Harris, H. Stein, W. V. Vardiman.
2001
.
Pathology and Genetics of Tumors of Haematopoietic and Lymphoid Tissues
IARC Press, Lyon, France.
21
Alizadeh, A. A., M. B. Eisen, R. E. Davis, C. Ma, I. S. Lossos, A. Rosenwald, J. C. Boldrick, H. Sabet, T. Tran, X. Yu, et al
2000
. Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling.
Nature
403
:
503
-511.
22
Shaffer, A. L., A. Rosenwald, L. M. Staudt.
2002
. Lymphoid malignancies: the dark side of B-cell differentiation.
Nat. Rev. Immunol.
2
:
920
-932.
23
Jardin, F., G. Buchonnet, F. Parmentier, N. Contentin, S. Lepretre, P. Lenain, J. M. Picquenot, S. Laberge, P. Bertrand, A. Stamatoullas, et al
2002
. Follicle center lymphoma is associated with significantly elevated levels of BCL-6 expression among lymphoma subtypes, independent of chromosome 3q27 rearrangements.
Leukemia
16
:
2318
-2325.
24
Bea, S., A. Zettl, G. Wright, I. Salaverria, P. Jehn, V. Moreno, C. Burek, G. Ott, X. Puig, L. Yang, et al
2005
. Diffuse large B-cell lymphoma subgroups have distinct genetic profiles that influence tumor biology and improve gene-expression-based survival prediction.
Blood
106
:
3183
-3190.
25
Skinnider, B. F., D. E. Horsman, B. Dupuis, R. D. Gascoyne.
1999
. Bcl-6 and Bcl-2 protein expression in diffuse large B-cell lymphoma and follicular lymphoma: correlation with 3q27 and 18q21 chromosomal abnormalities.
Hum. Pathol.
30
:
803
-808.
26
Pasqualucci, L., M. Compagno, J. Houldsworth, S. Monti, A. Grunn, S. V. Nandula, J. C. Aster, V. V. Murty, M. A. Shipp, R. Dalla-Favera.
2006
. Inactivation of the PRDM1/BLIMP1 gene in diffuse large B cell lymphoma.
J. Exp. Med.
203
:
311
-317.
27
Capello, D., F. Fais, D. Vivenza, G. Migliaretti, N. Chiorazzi, G. Gaidano, M. Ferrarini.
2000
. Identification of three subgroups of B cell chronic lymphocytic leukemia based upon mutations of BCL-6 and IgV genes.
Leukemia
14
:
811
-815.
28
Camacho, F. I., J. F. Garcia, J. C. Cigudosa, M. Mollejo, P. Algara, E. Ruiz-Ballesteros, P. Gonzalvo, P. Martin, C. Perez-Seoane, J. Sanchez-Garcia, M. A. Piris.
2004
. Aberrant Bcl6 protein expression in mantle cell lymphoma.
Am. J. Surg. Pathol.
28
:
1051
-1056.
29
Natkunam, Y., R. A. Warnke, K. Montgomery, B. Falini, M. van De Rijnqq.
2001
. Analysis of MUM1/IRF4 protein expression using tissue microarrays and immunohistochemistry.
Mod. Pathol.
14
:
686
-694.
30
Petit, B., M. P. Chaury, C. Le Clorennec, A. Jaccard, N. Gachard, S. Moalic-Judge, F. Labrousse, M. Cogne, D. Bordessoule, J. Feuillard.
2005
. Indolent lymphoplasmacytic and marginal zone B-cell lymphomas: absence of both IRF4 and Ki67 expression identifies a better prognosis subgroup.
Haematologica
90
:
200
-206.
31
Allman, D., A. Jain, A. Dent, R. R. Maile, T. Selvaggi, M. R. Kehry, L. M. Staudt.
1996
. BCL-6 expression during B-cell activation.
Blood
87
:
5257
-5268.
32
Yu, R. Y., X. Wang, F. J. Pixley, J. J. Yu, A. L. Dent, H. E. Broxmeyer, E. R. Stanley, B. H. Ye.
2005
. BCL-6 negatively regulates macrophage proliferation by suppressing autocrine IL-6 production.
Blood
105
:
1777
-1784.
33
Toney, L. M., G. Cattoretti, J. A. Graf, T. Merghoub, P. P. Pandolfi, R. Dalla-Favera, B. H. Ye, A. L. Dent.
2000
. BCL-6 regulates chemokine gene transcription in macrophages.
Nat. Immunol.
1
:
214
-220.
34
Flenghi, L., B. H. Ye, M. Fizzotti, B. Bigerna, G. Cattoretti, S. Venturi, R. Pacini, S. Pileri, F. Lo Coco, E. Pescarmona, et al
1995
. A specific monoclonal antibody (PG-B6) detects expression of the BCL-6 protein in germinal center B cells.
Am. J. Pathol.
147
:
405
-411.
35
Suzuki, S., K. Honma, T. Matsuyama, K. Suzuki, K. Toriyama, I. Akitoyo, K. Yamamoto, T. Suematsu, M. Nakamura, K. Yui, A. Kumatori.
2004
. Critical roles of interferon regulatory factor 4 in CD11bhighCD8α-dendritic cell development.
Proc. Natl. Acad. Sci. USA
101
:
8981
-8986.
36
Cattoretti, G., M. Buttner, R. Shaknovich, E. Kremmer, B. Alobeid, G. Niedobitek.
2006
. Nuclear and cytoplasmic AID in extrafollicular and germinal center B cells.
Blood
107
:
3967
-3975.
37
Shaknovich, R., A. Celestine, L. Yang, G. Cattoretti.
2003
. Novel relational database for tissue microarray analysis.
Arch. Pathol. Lab. Med.
127
:
492
-494.
38
Cattoretti, G., C. Angelin-Duclos, R. Shaknovich, H. Zhou, D. Wang, B. Alobeid.
2005
. PRDM1/Blimp-1 is expressed in human B-lymphocytes committed to the plasma cell lineage.
J. Pathol.
206
:
76
-86.
39
Hunyady, B., K. Krempels, G. Harta, E. Mezey.
1996
. Immunohistochemical signal amplification by catalyzed reporter deposition and its application in double immunostaining.
J. Histochem. Cytochem.
44
:
1353
-1362.
40
Marafioti, T., M. Jones, F. Facchetti, T. C. Diss, M. Q. Du, P. G. Isaacson, M. Pozzobon, S. A. Pileri, A. J. Strickson, S. Y. Tan, et al
2003
. Phenotype and genotype of interfollicular large B cells, a subpopulation of lymphocytes often with dendritic morphology.
Blood
102
:
2868
-2876.
41
Falini, B., E. Tiacci, A. Pucciarini, B. Bigerna, J. Kurth, G. Hatzivassiliou, S. Droetto, B. V. Galletti, M. Gambacorta, A. Orazi, et al
2003
. Expression of the IRTA1 receptor identifies intraepithelial and subepithelial marginal zone B cells of the mucosa-associated lymphoid tissue (MALT).
Blood
102
:
3684
-3692.
42
Glynne, R., G. Ghandour, J. Rayner, D. H. Mack, C. C. Goodnow.
2000
. B-lymphocyte quiescence, tolerance and activation as viewed by global gene expression profiling on microarrays.
Immunol. Rev.
176
:
216
-246.
43
Dedeoglu, F., B. Horwitz, J. Chaudhuri, F. W. Alt, R. S. Geha.
2004
. Induction of activation-induced cytidine deaminase gene expression by IL-4 and CD40 ligation is dependent on STAT6 and NFκB.
Int. Immunol.
16
:
395
-404.
44
Lazzi, S., C. Bellan, E. Tiacci, N. Palummo, R. Vatti, M. Oggioni, T. Amato, K. Schuerfeld, T. Tonini, P. Tosi, et al
2006
. IRTA1+ monocytoid B cells in reactive lymphadenitis show a unique topographic distribution and immunophenotype and a peculiar usage and mutational pattern of IgVH genes.
J. Pathol.
209
:
56
-66.
45
Pritchard, C. C., L. Hsu, J. Delrow, P. S. Nelson.
2001
. Project normal: defining normal variance in mouse gene expression.
Proc. Natl. Acad. Sci. USA
98
:
13266
-13271.
46
Lebecque, S., O. de Bouteiller, C. Arpin, J. Banchereau, Y. J. Liu.
1997
. Germinal center founder cells display propensity for apoptosis before onset of somatic mutation.
J. Exp. Med.
185
:
563
-571.
47
Klein, U., A. Gloghini, G. Gaidano, A. Chadburn, E. Cesarman, R. Dalla-Favera, A. Carbone.
2003
. Gene expression profile analysis of AIDS-related primary effusion lymphoma (PEL) suggests a plasmablastic derivation and identifies PEL-specific transcripts.
Blood
101
:
4115
-4121.
48
Shaffer, A. L., K. I. Lin, T. C. Kuo, X. Yu, E. M. Hurt, A. Rosenwald, J. M. Giltnane, L. Yang, H. Zhao, K. Calame, L. M. Staudt.
2002
. Blimp-1 orchestrates plasma cell differentiation by extinguishing the mature B cell gene expression program.
Immunity
17
:
51
-62.
49
Phan, R. T., R. Dalla-Favera.
2004
. The BCL6 proto-oncogene suppresses p53 expression in germinal-centre B cells.
Nature
432
:
635
-639.
50
DeKoter, R. P., H. Singh.
2000
. Regulation of B lymphocyte and macrophage development by graded expression of PU.1.
Science
288
:
1439
-1441.
51
Tumang, J. R., R. Frances, S. G. Yeo, T. L. Rothstein.
2005
. Cutting edge: spontaneously Ig-secreting B-1 cells violate the accepted paradigm for expression of differentiation-associated transcription factors.
J. Immunol.
174
:
3173
-3177.
52
Wang, X., Z. Li, A. Naganuma, B. H. Ye.
2002
. Negative autoregulation of BCL-6 is bypassed by genetic alterations in diffuse large B cell lymphomas.
Proc. Natl. Acad. Sci. USA
99
:
15018
-15023.
53
Cattoretti, G., L. Pasqualucci, G. Ballon, W. Tam, S. V. Nandula, Q. Shen, T. Mo, V. V. Murty, R. Dalla-Favera.
2005
. Deregulated BCL6 expression recapitulates the pathogenesis of human diffuse large B cell lymphomas in mice.
Cancer Cell
7
:
445
-455.
54
Mamane, Y., S. Sharma, L. Petropoulos, R. Lin, J. Hiscott.
2000
. Posttranslational regulation of IRF-4 activity by the immunophilin FKBP52.
Immunity
12
:
129
-140.
55
Grossman, A., H. W. Mittrucker, J. Nicholl, A. Suzuki, S. Chung, L. Antonio, S. Suggs, G. R. Sutherland, D. P. Siderovski, T. W. Mak.
1996
. Cloning of human lymphocyte-specific interferon regulatory factor (hLSIRF/hIRF4) and mapping of the gene to 6p23–p25.
Genomics
37
:
229
-233.
56
Tamura, T., P. Tailor, K. Yamaoka, H. J. Kong, H. Tsujimura, J. J. O’Shea, H. Singh, K. Ozato.
2005
. IFN regulatory factor-4 and -8 govern dendritic cell subset development and their functional diversity.
J. Immunol.
174
:
2573
-2581.
57
Basso, K., U. Klein, H. Niu, G. A. Stolovitzky, Y. Tu, A. Califano, G. Cattoretti, R. Dalla-Favera.
2004
. Tracking CD40 signaling during germinal center development.
Blood
104
:
4088
-4096.
58
Shapiro-Shelef, M., K. Calame.
2005
. Regulation of plasma-cell development.
Nat. Rev. Immunol.
5
:
230
-242.
59
Shigematsu, H., B. Reizis, H. Iwasaki, S. Mizuno, D. Hu, D. Traver, P. Leder, N. Sakaguchi, K. Akashi.
2004
. Plasmacytoid dendritic cells activate lymphoid-specific genetic programs irrespective of their cellular origin.
Immunity
21
:
43
-53.
60
Shapiro-Shelef, M., K. I. Lin, L. J. McHeyzer-Williams, J. Liao, M. G. McHeyzer-Williams, K. Calame.
2003
. Blimp-1 is required for the formation of immunoglobulin secreting plasma cells and pre-plasma memory B cells.
Immunity
19
:
607
-620.
61
Hans, C. P., D. D. Weisenburger, T. C. Greiner, R. D. Gascoyne, J. Delabie, G. Ott, H. K. Muller-Hermelink, E. Campo, R. M. Braziel, E. S. Jaffe, et al
2004
. Confirmation of the molecular classification of diffuse large B-cell lymphoma by immunohistochemistry using a tissue microarray.
Blood
103
:
275
-282.
62
Lossos, I. S., R. Levy, A. A. Alizadeh.
2004
. AID is expressed in germinal center B-cell-like and activated B-cell-like diffuse large-cell lymphomas and is not correlated with intraclonal heterogeneity.
Leukemia
18
:
1775
-1779.
63
Gupta, S., A. Anthony, A. B. Pernis.
2001
. Stage-specific modulation of IFN-regulatory factor 4 function by Kruppel-type zinc finger proteins.
J. Immunol.
166
:
6104
-6111.
64
Tam, W., M. Gomez, A. Chadburn, J. W. Lee, W. C. Chan, D. M. Knowles.
2006
. Mutational analysis of PRDM1 indicates a tumor-suppressor role in diffuse large B-cell lymphomas.
Blood
107
:
4090
-4100.