Defining a Transcriptional Fingerprint of Murine Splenic B-Cell Development

Home , BCLAF1, BHLHB2, BST1, CKAP2, CXCL7, ELF2

ORIGINAL ARTICLE Defining a transcriptional fingerprint of murine splenic B-cell development

I Debnath1, KM Roundy1, DM Dunn2, RB Weiss2, JJ Weis1 and JH Weis1 1Division of Cell Biology and Immunology, Department of Pathology, University of Utah School of Medicine, Salt Lake City, UT, USA and 2Department of Human Genetics, University of Utah School of Medicine, Salt Lake City, UT, USA

B-cell development occurs in a stepwise fashion that can be followed by the expression of B cell-specific surface markers. In this study, we wished to identify proteins that could contribute to the changes in expression of such markers. By using RNA from freshly isolated B220 þ cells, we hoped to reduce the effect of artifacts that occur during the isolation and amplification steps necessary to use flow cytometry analysis-sorted subsets in microarray experiments. Analyses comparing expression patterns from B220 þ 2-week bone marrow (pro-B, pre-B, immature B cells), 2-week spleen (predominantly transitional cells) and 8-week spleen (mainly mature B cells) yielded hundreds of genes. We also examined the B cell-activating factor (BAFF)- dependent effects on immature splenic B cells by comparing expression patterns in the spleen between 2-week A/J vs 2-week A/WySnJ mice, which lack functional BAFF receptor signaling. Genes that showed the expression differences between spleen and bone marrow samples were then analyzed through quantitative PCR on B-cell subsets isolated using two different sorting protocols. A comparison of the results from our study with the results from other analyses showed not only some overlap of preferentially expressed genes but also an expansion of other genes potentially involved in B-cell development. Genes and Immunity (2008) 9, 706–720; doi:10.1038/gene.2008.70; published online 11 September 2008

Keywords: B cells; gene expression; microarray; CD21; CD23; BAFF

Introduction The products of the mouse Cr2 (CD21) and Fcer2a (CD23) genes are widely used as cellular markers for The development of the B-cell lineage initiates in the differentiating splenic B-cell subsets.14–16 Our lab has a bone marrow when pro-B cells are generated from long-standing interest in understanding the immune common lymphoid progenitor cells.1 B cell-specific regulation and transcriptional control of the Cr2 and proteins such as the B-cell receptor (BCR), CD19, Pax5 Fcer2a genes. These two genes are expressed during the and the B220 form of CD45 allow for the further same window of B-cell development at the T1–T2 maturation of the B cell into a pre-B cell phenotype.2 transition phase, and their expression is extinguished Maturing B cells exit the marrow and colonize peripheral following the loss of Pax5, which occurs as B cells mature lymphatic tissues such as the spleen.3,4 Autoreactive B into terminally differentiated plasma cells.17,18 The Cr2 cells are eliminated, undergo receptor editing, or are and Fcer2a genes share a number of transcriptional anergized in the marrow5,6 and the spleen.7,8 The most control elements including those for Pax5, nuclear factor immature B cells of the spleen, termed transitional 1 cells (NF)-kB, nuclear factor of activated T cells family (T1),9 are susceptible to B-cell receptor-dependent cell members, E box protein E2A and the Notch signaling death/arrest. On the basis of the models proposed by family member retinol binding protein-Jk.19–22 Mature B Loder,10 Allman11 and others, T1 cells lead to another cells engineered to lack Pax5 (through a CD19-cre- transitional intermediate, the transition 2 (T2) cell. The dependent deletion) show decreased expression of the canonical B-cell differentiation pathway further suggests CR1/CR2 proteins and CD23.23 Earlier we have demon- that marginal zone (MZ) cells and follicular mature (FM) strated that Pax5 binds equally to the Cr2 and Fcer2a cells are the direct descendents of T2 B cells.10–12 This genes in expressing mature splenic B cells as well as highly ordered pathway of marrow and splenic B-cell marrow B cells, which do not express Cr2 and Fcer2a.19 development relies upon a complex network of various This finding suggests that these two genes either lack a environmental signals and transcription factor regula- key activating transcription factor to allow their expres- tory pathways.13 sion at the T1/T2 stages or are blocked from transcription in marrow B cells by a transcriptional silencing protein whose function is lost during B-cell differentia- Correspondence: Dr JH Weis, Department of Pathology, University tion. of Utah School of Medicine, Cell Biology and Immunology, 15 North In addition, it is known that the T1–T2 transition of Medical Drive East, Salt Lake City, UT 84112, USA. E-mail: [email protected] splenic B cells is dependent upon the cell survival and Received 11 July 2008; revised 5 August 2008; accepted 6 August maturation signals generated by the B cell-activating 2008; published online 11 September 2008 factor of the tumor necrosis factor family (BAFF).24–27 Gene expression analysis of splenic B-cell maturation I Debnath et al 707 BAFF binding to its receptor, BAFF-R, induces activation guidelines of the National Institutes of Health for the care of an alternative NF-kB signaling pathway (utilizing p52 and use of laboratory animals. Mice examined were age- homodimers)28–31 and downregulates the pro-apoptotic matched males and females, unless otherwise specified. protein Bim through a heightened activation of extracellular signal-regulated kinase.32 Animals deficient in Antibodies BAFF or BAFF-R show a profound loss in peripheral Fluorescein isothiocyanate-conjugated anti-mouse CD21 mature B-cell populations, but B-cell development in the (clone 7G6), phycoerythrin (PE)-conjugated anti-mouse marrow and the migration of T1 B cells into the spleen CD24 (clone M1/69), streptavidin (SA) PE-Cy5 (BD are not altered.15,24,27 The BAFF-R mutant A/WySnJ Phamingen, Inc., San Diego, CA, USA), biotin- and PE- mouse possesses a mutation within the cytoplasmic tail conjugated anti-mouse CD23 (clone B3B4) (eBioscience, of this receptor that eliminates its ability to bind to the San Diego, CA, USA), PE-conjugated anti-mouse AA4.1 intracellular signaling molecule tumor necrosis factor (CD93) (eBioscience) and Cy5-conjugated polyclonal anti- receptor-associated factor 3.33–37 These mice have defi- mouse IgM (Biomedia Corporation, Foster City, CA, USA) ciencies in T2, MZ and FM B-cell populations but their were used for flow cytometry analyses (FACS). To-Pro-3 bone marrow B-cell developmental stages and splenic T1 and 4’,6-diamidino-2-phenylindole were used for viability populations are normal and similar to those of BAFF- staining (Invitrogen Corporation, Carlsbad, CA, USA). deficient animals.24,27 The expression of Cr2 and Fcer2a has been suggested to be a direct effect of BAFF signaling Isolation of B-cell populations independent of the B-cell survival function of the Total splenocytes or marrow cell populations were cytokine16 (presumably through the enhanced transloca- isolated from various strains of mice at various ages as tion of NF-kB-p52 into the nucleus). However, we have indicated in figure legends. After red blood cell lysis, a B found BAFF-dependent expression levels of CD21 and cell-enriched population was obtained using B220 þ CD23 to be similar to those of CD19 and other B-cell magnetic beads depletion according to the manufac- marker proteins whose expression is not directly turer’s protocol (Miltenyi Biotech, Inc., Auburn, CA, dependent upon NF-kB-p52. Furthermore, NF-kB-p52 USA). B220 þ cells were stained with CD21-FITC, CD24- chromatin complexes are evident within Cr2 and Fcer2a PE, CD23-biotin/SA-PE-Cy5 or with IgM Cy5, AA4.1 PE genes from BAFF-R-defective and normal B cells.38 and CD23 PE-Cy7, for flow cytometry analyses and In this study, we sought to identify the key transcrip- resuspended at 5 106 cells per ml for cell sorting. tional regulatory proteins that control Cr2 and Fcer2a Magnetic sorting with B220 routinely provided samples expression during B-cell maturation. We utilized cells 95% or greater homogeneity. from wild-type A/J bone marrow, immature spleen and Various B-cell subsets (see Results) were sorted and mature spleen to create overlapping genomic screens to isolated using FACSVantage SE at the FACS Core Facility, identify gene products preferentially lost or enhanced University of Utah. Two different approaches to isolate during B-cell development. This approach allowed us to the T1, T2, MZ and FM splenic cells were utilized1,11,40 enrich for populations that vary in their expression of and described in figure legends. CD21 and CD23 instead of isolating marrow and splenic B-cell subsets and using amplification to obtain enough Microarray analyses RNA. We also compared A/J mice with the mutant A/ Total RNA was isolated from B220 þ B cells from 2-week WySnJ to investigate the downstream targets/genes of marrow (20 animals), 2-week (four animals)- or 8-week BAFF signaling that could contribute to CD21 and CD23 (four animals)-old spleen from A/J, or 2-week spleen expression. We have identified a number of immuno- from A/WySnJ mice. Pools of cells from five mice were modulatory proteins whose expression and function are used per sample. Total RNA from the pools was linked to B-cell development by obtaining gene grid quantified and used for Affymetrix array hybridization array data from a variety of similar models and according to the manufacturer’s instructions (Affymetrix, comparing those data with known positive and negative Santa Clara, CA, USA). cDNA was synthesized from controls for appropriate expression profiles. This list of 5–7 mg of total RNA and hybridized to GeneChip Mouse candidate proteins includes numerous transcription Genome 430 2.0 Array (Affymetrix). We utilized single- factors that could control target gene expression either chip hybridizations as duplicate chips in previous by direct induction of transcription or by alleviating analyses showed virtually no variability.41 To avoid transcriptional suppression. Thus, this study not only skewing the data due to the health of a single animal, details a broad gene expression analysis of B-cell all grid array analyses were carried out with cells pooled maturation that includes our specific targets of Cr2 and from at least four genetically identical, age-matched Fcer2a but also incorporates analyses of other genes animals. Data were processed using the Affymetrix and whose expression is similarly controlled. gcRMA (Robust Multiarray Average) packages.42–44 Tran- scripts showing a change of twofold or greater were considered differentially expressed, whereas other tran- Materials and methods scripts were considered not changed. The function of gene products was primarily determined with the National Mice Cancer Institute database, Gene Ontology annotations A/J, A/WySnJ, C57BL/6 and BALB/c mice were obtained according to Mouse Genome Informatics and public from Jackson Laboratory. C2ta (class II trans-activator)- databases such as NCBI/Pubmed and Ensembl. deficient mice39 were provided by Dr Peter E Jensen, University of Utah. All the animals were housed RNA preparation, cDNA synthesis and real-time PCR and bred in the Comparative Medicine Center at Uni- Total RNA was isolated from cells using Illustra RNA Spin versity of Utah Health Science Center according to the Mini kit (GE Healthcare, Buckinghamshire, UK) according

Genes and Immunity Gene expression analysis of splenic B-cell maturation I Debnath et al 708 to the manufacturer’s instructions. cDNA was synthesized serve to increase this bias, excluding the RNA contribu- using 2 mg of total RNA (or less in the case of isolated B-cell tion from smaller or less abundant transcripts. Other subsets). cDNA was purified using the Qiagen (Qiagen, groups have documented that the sensitivity of such a Inc., Valencia, CA, USA) PCR purification kit. Quantitative screen to detect low-abundance RNA, such as those real-time PCR (RT-PCR) was performed using Light Cycler encoding transcription factors, is compromised when (Roche Diagnostics, Indianapolis, IN USA). All shown using limited numbers of sorted, purified cells.51 transcript values are relative to b-actin expression and are A second caveat to this approach is that FACS sorting mean values±s.d. Primer sequences for quantitative PCR of cells is a time-consuming process and utilizes a variety analyses are listed in Table 2. Semiquantitative RT-PCR for of antibodies to identify key cell-surface proteins. These C2ta using P32-dCTP was performed as described before45 surface proteins, particularly IgM, may alter the activa- and PhosphorImager quantification was done (STORM820 tion state of the cell upon binding. It is possible that the PhosphorImager and Image Quant analysis; Molecular RNA population present in a subset of cells after sorting Diagnostics, Inc., Sunyvale, CA, USA) as detailed earlier.19 is no longer representative of the RNA population that was present immediately after isolation. A second approach would be to utilize differences in B-cell populations between various tissues and ages of Results mice to obtain larger numbers of cells, yielding more Establishing model systems for the analysis of B-cell gene significant amounts of RNA and eliminating the need for expression further amplification steps. In this approach, magnetic The goal of this study was to identify those genes that are beads would be used to swiftly isolate B220 þ B cells silenced or activated during splenic B-cell development, from freshly harvested bone marrow and spleen popula- with a specific focus upon those transcription factors that tions. Magnetic bead sorting for B220 þ cells is com- might regulate the expression of the CD21 and CD23 monly used to purify B cells and does not directly result genes in these cells. In designing such a screen, two in their activation.52–54 The major caveat to this second different protocols were considered. The first approach approach is that gene expression profiles obtained would be to use various B-cell surface markers to purify through this method would be representative of a mixed specific subsets of marrow and splenic B cells including population of cells, not a single subset; however, by T1, T2, MZ and FM cells of the spleen. RNA would then judiciously choosing the tissues to examine, we can be obtained from each subset and used in the grid highly enrich for specific cell populations. analysis. A number of groups have taken this approach As shown in Figure 1, distinct populations of B cells in examining gene expression profiles in subsets of can be discerned in wild-type 2-week marrow, and immune cells.46–50 There are, however, at least two major 2-week and 8-week spleens. Mouse bone marrow is caveats to this approach. populated, in part, by a series of maturing B cells First, the small number of cells that can be isolated including pro-, pre- and immature B-cell types. Cells using FACS sorting results in a low quantity of RNA that released from the marrow colonize peripheral sites such must be amplified to screen a grid array. The standard as the spleen. Genes differentially expressed in marrow B array process, based on having the amount of required cells include those destined to be expressed in mature B RNA without amplification, is already biased toward cells, such as Pax5 and CD19, as well as those genes preferential identification of transcripts due to abun- whose expression is extinguished in mature B cells, such dance, size or G/C content. Addition of another linear or as Il7r and terminal transferase (Dntt) (Supplementary exponential amplification step into the process would Figure S1). Marrow from mature animals also possesses a

2 week A/J 2 week A/WySnJ 2 week A/J 8 week A/J Bone Marrow Spleen Spleen Spleen 70 70 70 70 2.6% 13.5% 36.6% 74.9% Counts Counts Counts Counts

0 0 0 0 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 CD21

70 70 70 48.5% 70 85.6% 8.2% 16.4% Counts Counts Counts Counts

0 0 0 0 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 CD23 Figure 1 Phenotypic characterization of selected mouse spleen and marrow populations. Total cell populations were obtained from 2-week A/J bone marrow, 2-week A/WySnJ spleen, 2-week A/J spleen and 8-week A/J spleen. Cell were sorted for B220 þ populations and then analyzed for CD21 (top panels) or CD23 (bottom panels) expression (filled peaks). Isotype controls are shown as empty peaks. Values indicate the percent of positive cells. Data are representative of multiple independent experiments.

Genes and Immunity Gene expression analysis of splenic B-cell maturation I Debnath et al 709 sub-population of re-circulating mature B cells (CD21 þ ) demonstrated altered expression of 612 probe sets as well as plasma cells; however, marrow from 2-week (twofold or more difference over A/J samples) whereas animals lacks these cell types as seen by the low level of those of A/J showed 211 probe sets (twofold or more CD21 expression in Figure 1. Thus, the use of immature B difference over the A/WySnJ counterparts) (Supplemen- cells from 2-week bone marrow, when compared with tary Table S1). A selected set of genes showing differential maturing or mature splenic B cells, would provide more expression in this screen is shown in Table 1. Analysis of robust data than using marrow B-cell populations from the genes elevated in the A/WySnJ subset indicated an mature animals. abundance of genes encoding proteins involved in cell Upon migration into the spleen, immature B cells cycle arrest, apoptosis and cell division perhaps because develop into mature B cells, capable of fully expressing of growth arrest induced by the loss of BAFF-R signaling. CD21 and CD23. For instance, in the spleens of mature The genes demonstrating elevated expression in the A/J animals (8 weeks), B-cell subsets can be found in the samples encoded many different immune response and following percentages: T1 (5–10%), T2 (15–20%), MZ modulation proteins indicative of the healthy maturation (3–5%) and FM (65–80%).10 However, in younger of B cells into the mature phenotype including the animals, these percentages are skewed toward a more transcription factors C2ta, NFkB and Runx3. Those genes transitional, immature phenotype; at week 1, most encoding B-cell proteins expressed in both T1 and T2 splenic B cells are T1 but by week 2, T2 and FM B cells populations, such as CD19 or Pax5, did not show up in appear (Figure 1). The mouse spleen lacks mature this array analysis indicating that the lack of BAFF architecture and MZ B-cell phenotypes until 4–6 weeks signaling in the A/WySnJ sample did not significantly postbirth.38 Thus, comparing splenic cell populations extinguish expression of these genes. from an 8-week animal with those of a 2-week animal To confirm the differential expression of known and would be similar to comparing mature B cells with putative regulatory genes at the BAFF-dependent inter- immature B cells as those cells are predominant in the face, quantitative RT-PCR was carried out using freshly spleen at that time. In addition, comparing these two isolated 2-week B220 þ splenocytes from A/J or A/WySnJ splenic samples vs the 2-week marrow B-cell sample mice (Figure 2) (Supplementary Table S2). CD21 (Cr2)and would provide us with subsets of genes whose expression CD23 (Fcer2a) were used as controls and, as expected, their is extinguished or activated during B-cell maturation. expression level was distinctly lower in the A/WySnJ In addition to analyzing wild-type samples, cells sample. CD19 expression levels remained relatively similar obtained from mice possessing mutations that alter between the two samples indicating, in part, that the B-cell maturation can also be analyzed. For example, differential gene expressions are not due to the varying the A/WySnJ mouse lacks functional BAFF signaling number of B cells between the two populations. F4/80, a because of a mutation in the receptor. A/WySnJ mice are macrophage marker, was included as a control for deficient in T2, MZ and FM B cells but have normal macrophage contamination of isolated B cells. The two populations of T1 and B1 B cells in the spleen.15 FACS lower panels of Figure 2 demonstrate verification of analysis of B220 þ splenocytes from 2-week A/J and A/ differential gene expression between the A/J and A/ WySnJ animals (Figure 1) demonstrates a marked WySnJ samples. Runx3 and Gtf2i encode transcriptional reduction in CD21 þ cells and fewer CD23 þ cells in regulators Sema7a, Icosl (Icos ligand), Ccr6, and Bst1 encode the A/WySnJ samples compared with the A/J animal. cell-surface proteins whereas Rsad2 encodes a cellular The A/WySnJ 2-week spleen is highly enriched for reductase. These profiles indicated that differences in gene immature B cells that are unable to fully develop into expression identified in the gene grid array were mature splenic B cells, failing to significantly express confirmed by transcript analysis from independently CD21 and CD23. Therefore utilizing gene grid array to derived A/J and A/WySnJ RNA samples. screen transcripts from wild-type and A/WySnJ 2-week To further analyze the expression profiles of different splenic samples would allow us to define transcripts candidate genes obtained from the A/J vs A/WySnJ enhanced in the more immature splenic B cells (from the screen, we utilized RT-PCR analysis to screen mRNA A/WySnJ sample) vs those expressed in normal matur- samples obtained from B220 þ cells of the marrow of 2- ing B cells (the A/J wild-type sample). week-old A/J animals, B220 þ splenocytes from 2-week- We chose to use the second approach to examine gene old A/J mice and B220 þ splenocytes from 8-week-old expression in maturing B-cell populations because of its A/J mice. As shown in Figure 3, CD21 (Cr2), CD23 increased sensitivity (vs using limiting quantities of RNA (FceR2a), Icosl and C2ta show very similar profiles with isolated from aged, manipulated cells as described in the the highest expression in the most mature B-cell subsets. first approach) and the ability to obtain the highly A number of other genes that also demonstrated elevated enriched cell populations from the animal tissues expression in the A/J samples vs the A/WySnJ cells (for described above. This second approach would provide example, Ccr6 and Runx3) showed similar profiles to us with an extensive list of possible candidate genes that CD21. Genes that showed elevated expression in the A/ could then be validated by transcript analysis from WySnJ samples, such as Tcf12, Gfi1, Erg and Bst1, had the FACS-sorted B-cell subsets. greatest level of expression in the marrow samples and the least in the 8-week splenocyte sample. Other genes, Loss of BAFF-R signaling results in a block of expression of such as Cnr2 and Sema7a, showed very similar levels of mature B-cell immunomodulatory genes expression between the 2- and 8-week spleen samples, The purpose of the first screen was to identify genes suggesting that their expression levels do not change differentially expressed because of BAFF-dependent during splenic B-cell differentiation. Finally, a subset of signaling by utilizing B220 þ cells from 2-week-old A/J genes (typified by Rsad2) showed the greatest level of and A/WySnJ spleens as a source of RNA for probing expression in the immature 2-week spleen sample an Affymetrix mouse microarray. The A/WySnJ cells compared with the mature 8-week spleen or immature

Genes and Immunity Gene expression analysis of splenic B-cell maturation I Debnath et al 710 Table 1 Functional categorization of selected genes from A/Ja vs Table 1 Continued A/WySnJb screen FC FC Cell cycle, apoptosis Immune regulation Pten 2.6 CD21/Cr2 4.2 Akt1 2.6 Fcer2a/CD23 2.7 Dffb 2 Icosl 6 Ckap2 3.2 Gzme 2.2 Rad21 2.9 H2-Ab1 2.1 Pik3r1 2.9 H2-DMa 3.3 Bre 5.8 H2-Aa 4.6 Bub1 2.8 H2-Ea 3.4 Calm2 2.8 Sema7a 3.2 Cdc20 2.2 Cxcl1 2.3 Ube1C 3.3 Blr1 2.4 Ccnb1 2.6 Tnf 2.2 Slk 2.2 Lta 3.1 Aurkb 2.2 CD24a 2.4 Kif11 2.8 Pik3cd 3.9 Stag1 2.7 Bst1 5.1 Tlr4 3.1 aNormal type represents increased fold expression of genes from Ctse 6.9 A/J B220+ splenocytes compared with A/WySnJ B220+ spleno- Rsad2 6.6 cytes. Ctsb 2.5 bItalic bold type represents increased fold expression of genes from Transcription A/WySnJ B220+ splenocytes compared with A/J B220+ spleno- Nfkb2 2.9 cytes. C2ta 2.9 FC represents fold change. Gatad2b 2.1 Judm2/Jdp2 2.6 Runx3 2.2 Zfp318 2.4 Fbxl11 2.1 marrow samples. These data confirm that the genes Smad5 2.6 Fos 2.4 showing enriched expression in the A/WySnJ samples E2f5 2.8 are indicative of immature/maturing B cells whereas Sox4 3 those showing elevated expression in the A/J sample Atf2 2.4 represent products of mature B cells. Erg 2.3 Gfi1 2.7 Gtf2i 2.5 Age- and maturation-dependent analyses of differential gene Pknox1 5.7 expression of bone marrow and splenic B-cell populations Tcf12 3.2 The previous set of experiments identified genes differ- Fbxo30 4 entially expressed during BAFF-dependent splenic B-cell Zfpn1a1/Ikaros 2.2 maturation. A number of these genes were already Zfpn1a3/Aiolos 2.6 known to follow such expression profiles, such as Cr2 Cell signaling and C2ta, and thus served as positive controls. However, Cnr2 8.9 a number of genes not previously linked to B-cell Jak3 2.2 differentiation and maturation were implicated to have Dtx1 2.7 Dtx4 2.9 a function in that process. One liability of this analysis Itsn1 2.7 was the inclusion of the A/WySnJ cells that are, due to Ccr6 2.5 the lack of BAFF signaling, inappropriately arrested in Edg5 2.5 their development. Therefore, we sought to analyze Gpr45 2 similar samples from normal mice to validate candidate Csprs 2.7 genes obtained from the A/WySnJ screen, as well as to Sparc 3 Lpxn 2.2 identify additional genes whose expression is coincident Picalm 5.8 with the transitional stage of splenic B-cell development. Rgs12 2.5 To accomplish this, we utilized total RNA obtained from Syk 2.8 B220 þ cells from 2-week bone marrow and compared it Csnk1d 2.2 with RNA obtained from either B220 þ cells from 2-week Rhoh 2.1 or 8-week spleen (see Figure 1 and Supplementary Figure Ctnnb1 2.8 Dusp10 2.2 S1). Affymetrix grid analysis demonstrated that 1636 Adam17 2.6 probe sets demonstrated a twofold or greater expression Tulp4 4 in the marrow sample whereas 1480 probe sets demonstrated a greater than twofold expression in the 2-week Cell cycle, apoptosis Bcl2 7.5 spleen sample (Supplementary Table S3). Similarly, 1764 Lig3 2.1 probe sets showed elevated expression in the marrow Pglyrp1 2.6 sample compared with 1436 probe sets in the 8-week Clu 3.7 splenocyte population (Supplementary Table S4). The

Genes and Immunity Gene expression analysis of splenic B-cell maturation I Debnath et al 711 160

140 A/J 2wk B220+ Spl

120 A/WySnJ 2wk B220+ Spl

actin 100 β

Transcript/1000 40

0 CD21 CD23 CD19 F4/80

14 actin

β 12

6 Transcript/1000 4

0 Runx3 Gtf2i Rsad2 Sema7a Icosl

3.5

2.5 actin β 2

1.5

Transcript/1000 1

0.5

0 Ccr6 Bst1 Figure 2 RT-PCR verification of selected genes from the A/J vs A/WySnJ gene array using A/J and A/WySnJ samples. RNA was isolated from a pool of five different B220 þ splenocyte populations obtained from 2-week-old splenocytes from A/J (gray bars) or A/WySnJ (black bars) mice. Transcripts are all relative to b-actin. Data are represented as an average of three independent experiments (three different RNA preparations and RT-PCR analysis), and error bars represent the standard deviation. Gene-specific oligonucleotides are listed in Supplementary Table S2. RT-PCR, real-time PCR. gene expression profiles of the 2-week and 8-week Table 2 shows a subset of genes demonstrating spleens were also contrasted with each other (Supple- increased expression in the B220 þ marrow population mentary Table S5). The 2-week spleen sample demon- compared with the two splenocyte populations. Thus, as strated 951 probe sets with elevated expression compared B cells move from the marrow to the spleen and with the 8-week spleen sample, whereas 1311 probe sets differentiate into mature phenotypes, the expression of are elevated in the 8-week splenic sample. Cxcl4, Cxcl7 and Cxcl12 is lost, as is the expression of

Genes and Immunity Gene expression analysis of splenic B-cell maturation I Debnath et al 712 45 6000

40 A/J 2wk B220+ BM 5000 A/J 2wk B220+ SP 35 A/J 8wk B220+ SP

actin 30 4000 β 25 3000 20

15 2000 Transcript/1000 Transcript/1000 10 1000 5

0 0 CD21 CD23 Icosl C2ta 25 8

7 20 6 actin

β 5 15

10 3

Transcript/1000 Transcript/1000 2 5 1

0 0 Ccr6 Zfp318 Runx3 Cnr2 Tcf12 Pknox1 Gfi1 Erg Bst1 Sema7a Rsad2 Figure 3 RT-PCR verification of selected genes from the A/J vs A/WySnJ gene array using normal mouse tissues. RNA (from a pool of five animals) was isolated from B220 þ 2-week bone marrow (light bar), B220 þ 2-week splenocytes (dark gray bar) and B220 þ 8-week splenocytes (black bar) from A/J mice. Real-time PCR was performed on LightCycler (Roche), except for C2ta, which was done by semiquantitative RT-PCR and analyzed by PhosphorImager. Transcripts are relative to b-actin. Error bars represent the mean standard deviation of two independent experiments (including both RNA preparation from pooled cells and RT-PCR analysis). Gene-specific oligonucleotides are listed in Supplementary Table S2. RT-PCR, real-time PCR.

genes utilized in the immunoglobulin rearrangement levels in both 2-week and 8-week subsets (compared process such as pre-B lymphocyte 1 (Vpreb1), Dntt, Rag1 with the marrow sample), suggesting that their expres- and Rag2. Those genes showing virtually equivalent sion is maximally induced as soon as the maturing B cells changes in comparing the 2- and 8-week splenocyte enter the spleen and is maintained in the mature B-cell samples (Cxcl4, Cxcl12, Il2ra, Igfbp5, Fabp4, Ptp22, Cleg1b subsets. Class C showed the highest level of expression and Bzw2) suggest that the expression of these genes is in the 2-week splenocytes and subsequent loss in the extinguished before maturing B cells enter the spleen. 8-week sample. Genes of class C, such as Elf2, Gpbp1, However, those genes showing an enhanced expression Zfp62, Ets1 and Bptf, show the highest level of expression change between the marrow and the 2- and 8-week in B220 þ B cells immediately after migration into the samples (Ptpla, T2bp, Tmprss3, Bub1, Il7r, Sox4 and Tcf19) spleen with diminished expression as the resident B cells indicate that these genes are still being expressed in take on mature phenotypes. Obviously, these classes of immature B cells but are lost in the full B-cell maturation gene expression are not absolute, but suggest trends in state of the 8-week-old B220 þ splenocytes. gene expression profiles of this selected subset of genes Table 3 shows the expression profile of a subset of encoding known, or putative, transcriptional control genes known to encode transcriptional regulatory pro- proteins. teins. In this table, those genes showing elevated To confirm the specificity of expression of a number of expression in the 2- or 8-week B220 þ splenocyte candidate genes identified in the screen comparing populations compared with the 2-week bone marrow wild-type marrow and spleen samples, we utilized sample are shown. The expression patterns of this subset RT-PCR analysis using cDNA independently generated of genes can be divided into three general groups from a pool of B220 þ cells obtained from A/J 2-week (Figure 4). Class A, typified by Mef2c, Zfp318, C2ta, marrow, and 2-week and 8-week spleens. As shown Chd7, Tbx21, Stat4, Runx3 and Cbx7, demonstrated in Figure 5, a variety of gene expression profiles gradual increase from 2-week BM to 2-week spleen with were observed. Genes such as Prpf40a, Smc2 and E2f7 greatest expression in the 8-week spleen sample, showed a loss of expression during B-cell maturation. suggesting a required function in fully mature B cells. Others, such as Phf1, Tbx21 (Tbet), Stat4, Mef2c, The second subset, class B, typified by genes Spic, Rbm5, Spic, Klf12 and Hfe, demonstrated an increased Zfp385, Smc6 and Id2, demonstrated similar expression expression.

Genes and Immunity Gene expression analysis of splenic B-cell maturation I Debnath et al 713 Table 2 Genes demonstrating increased expression in 2-week Table 3 Differential expression of selected genes encoding bone marrow B220+ B cells vs B220+ splenic B cells from 2- or transcriptional control proteins between B220+ marrow cells vs 8-week-old mice B220+ 2- or 8-week splenocytes

Probe set Gene Fold change marrow Fold change marrow Gene Probe set 2 week spleen vs 8 week spleen vs Class symbol vs 2-week spleen vs 8-week spleen symbol 2 week marrow 2 week marrow

Chemokine family Mef2c 1421028_a_at 37.6 75 A 1418480_at Cxcl7 109.6 142.8 Spic 1449134_s_at 36.8 32.1 B 1448995_at Cxcl4 49 51.5 Bptf 1427311_at 29.5 9.2 C 1417574_at Cxcl12 34 43.5 Zfp318 1425347_a_at 21.4 30.7 A Ets1 1422027_a_at 13.5 7.9 C Immune function Elf2 1428045_a_at 13.3 3.1 C 1449869_at Vpreb1 4.3 1744.6 Foxp1 1456851_at 12.0 4.7 C 1450545_a_at Dntt 2.5 173.8 Fbxl11 1435329_at 9.5 4.7 C 1450680_at Rag1 2.7 187.5 Trim30 1456494_a_at 8.7 5.6 B 1418065_at Rag2 3.1 21.2 Gpbp1 1430981_s_at 8.7 2.8 C Zfp62 1451730_at 7.8 3 C Receptor signaling Bclaf1 1436023_at 7.8 3.2 C 1457434_s_at Ptpla 3.5 95.4 Maf 1447849_s_at 7.0 9.5 B 1448575_at Il7r 2.8 81.6 Zfp207 1438515_at 6.5 2.2 C 1420692_at Il2ra 12 12.5 Phf15 1455345_at 6 12.7 A 1421182_at Clec1b 11.2 7.8 Eomes 1435172_at 6.0 7.6 B 1426501_a_at T2bp 2.8 27.1 Jarid1c 1426498_at 5.3 o2C 1450194_a_at Myb 2.7 36.2 Creg1 1415948_at 5.3 5 B 1421714_at Tmprss3 7.9 196.3 Stat1 1450033_a_at 5.1 3.8 B 1448147_at Tnfrsf19 2.1 16.8 Zfp385 1418865_at 4.8 3.9 B Brwd1 1452322_a_at 4.8 7.9 A Metabolism Bcl11a 1446293_at 4.8 3.5 B 1452114_s_at Igfbp5 30.2 30.1 Arid5b 1420973_at 4.7 4.9 B 1423756_s_at Igfbp4 14.2 154.1 C2ta 1421210_at 4.7 23.2 A 1417023_a_at Fabp4 199 260.5 Chd4 1436343_at 4 o2C 1424046_at Bub1 7.3 152.3 Zfp458 1445824_at 3.9 3.3 B 1451128_s_at Kif22 4 93 Elf1 1439319_at 3.7 2 C 1437611_x_at Kif2c 3.3 243.3 Pparg 1420715_a_at 3.5 2.6 B 1415810_at Uhrf1 3 61.1 Bzw1 1450846_at 3.5 2.1 C 1452954_at Ube2c 3.1 81.1 Ebf1 1441627_at 3.4 o2C 1428304_at Esco2 7.3 98 Tcf25 1437204_a_at 3.3 3.4 B 1416961_at Bub1b 5 179.3 Klf7 1419356_at 3.1 5.1 A 1437580_s_at Nek2 5.9 52 Zbtb3 1440180_x_at 3.1 o2C Bhlhb2 1418025_at 3 4.4 B Transcription Tbx21 (tbet) 1449361_at o2 19.7 A 1424608_a_at Bzw2 13.4 13.27 Stat4 1448713_at o2 8.9 A 1417574_at Sox4 3.9 41.7 Phf1 1416542_at o2 6.6 A 1436186_at E2f8 2.6 40 Phf20 1452258_at 3 5.8 A 1423809_at Tcf19 2.8 38.4 Pou6f1 1452844_at o2 5.8 A 1437187_at E2f7 3.4 35 Nfat5 1439805_at 2.8 5.7 A 1428827_at Whsc1 2.9 11.1 Nr1d2 1416958_at o2 5.1 A 1439619_at Tcf12 3.4 3.4 Chd2 1445843_at 2.4 5 A 1445201_at Zfp53 8 o2 Runx3 1440275_at 2.1 4.4 A 1417679_at Gfi1 4.5 5.4 Batf 1419410_at o2 3.9 A Id2 1435176_a_at 2.8 3.7 B Zfp28 1439503_at o2 3.5 A Zfp69 1458274_at 2.3 3.4 A Analyses of genes implicated in B-cell maturation using Nfatc1 1428479_at 2.1 3.2 A BALB/c and C57BL/6 populations Nr4a1 1416505_at o2 3.1 A The earlier analyses were performed on cell populations Nfatc2 1439205_at o23A Gatad2b 1425075_at 2.7 o2C from the A/J mouse strain. To determine if such Klf3 1454666_at 2.7 8.9 A expression profiles were unique to the A/J strain or Klf12 1439847_s_at o2 22.6 A were strain independent, we tested the expression of Creb12 1435085_at o2 8.6 A candidate genes derived from the two previous gene array screens using sorted cell populations from BALB/c and C57BL/6 mice. B220 þ cells were purified from depressed expression in the 8-week spleen sample for pooled samples from 2-week marrow, 2-week and the C57BL/6 strain, and Mef2c expression in the 8-week 8-week spleens from the two strains and analyzed by BALB/c sample was muted. In general, the expression RT-PCR. As shown in Supplementary Figure S2, the profiles of the B-cell maturation candidate genes were control genes CD21 (Cr2) and CD23 (Fcer2a) showed strain independent. the same expression profiles as the A/J samples (see Figure 3). Similar expression profiles for most of the Defining B-cell subsets responsible for differential gene candidate genes were obtained from cell samples expression profiles derived from all three strains. A small subset of genes A number of the candidate genes obtained from the did show differences in expression; Spic showed previously described gene grid screens showed elevated

Genes and Immunity Gene expression analysis of splenic B-cell maturation I Debnath et al 714 expression in maturing B cells. One goal of this project fluctuation of CD21 and CD23 surface expression during was to define transcription factors whose expression B-cell maturation. We and others have shown that MZ pattern was either identical (transcriptional inducers) or cells are enriched for high levels of Cr2 but not Fcer2a opposite (transcriptional repressors) to that of Cr2 or transcripts (in which CD21 cell surface expression is Fcer2a in splenic B-cell subsets to account for the elevated but that of CD23 lost),19,48,55 whereas FM cells possess transcripts and proteins for both CD21 and CD23. To determine if the expression profiles of the candidate genes identified in the previous screens demonstrated differential expression in maturing/ma- Class B ture B-cell populations, we performed quantitative RT- PCR on RNA isolated from purified T1, T2, MZ and FM cells. The genes analyzed were those that demonstrated Class C Class A increased expression during B-cell maturation. To accomplish this task, we used two different published

Gene Expression protocols for the isolation of B-cell subsets from mature spleens. In the first screen, we FACS-sorted cells using relative expression of CD23, CD21 and CD24 on B220 þ splenic 1,10 Increasing B cell maturation cells to obtain T1, T2, MZ and FM populations. Specifically, 8-week C57BL/6 B220 þ splenocytes were Figure 4 Schematic model of splenic B-cell gene expression stained for expression of CD21, CD23 and CD24 patterns derived from marrow vs spleen array data. Class A (Figure 6a). CD23 cells (R2 population) and CD23 þ represents genes that show incremental upregulation during B-cell cells (R3 population) were analyzed for relative expres- maturation. Class B genes are elevated during bone marrow/T1–T2 high high transition phase with expression level remaining constant in the sion of CD21 and CD24. CD23 CD21 CD24 and mature B cells. Class C shows genes with elevated expression in the CD23 CD21-CD24high cells were sorted as MZ and T1 T1 subset and reduced expression in more mature B-cell types. cells, respectively. The CD23 þ CD21highCD24high and

35 A/J 2wk B220+ BM 30 A/J 2wk B220+ SP

25 A/J 8wk B220+ SP actin β 20

10 Transcript/1000

0 Prpf40a Smc2 E2f7 Zfp53

actin 5 β

Transcript/1000 2

0 K1f3 Phf1 Tbx21 Stat4 Mef2C Spic K1f12 Hfe Figure 5 RT-PCR verification of selected genes from the A/J 2-week marrow, 2-week spleen and 8-week spleen array analysis. RNA (from a pool of five different animals) was isolated from B220 þ 2-week bone marrow (light bar), B220 þ 2-week splenocytes (dark gray bar) and B220 þ 8-week splenocytes (black bar) from A/J mice. Transcripts are relative to b-actin. Error bars represent the mean standard deviation of two independent experiments (including RNA preparation from pooled cells and RT-PCR analysis). Gene-specific oligonucleotides are listed in Supplementary Table S2. RT-PCR, real-time PCR.

Genes and Immunity Gene expression analysis of splenic B-cell maturation I Debnath et al 715 CD23 þ CD21 þ CD24low cells were sorted as T2 and FM genes. As shown, the relative expression levels of the cells, respectively. two control genes, CD21 and CD23, followed the The expression of the CD21 (Cr2) and CD23 (Fcer2a) expected profile and were very similar to the levels genes in these B-cell subsets was as expected (Figure 6b). shown in the preceding Figure 6. Similar profiles were The expression of other genes varied widely with some also obtained for a number of other test genes. The data showing the highest level of expression in the MZ in Figure 6 and Supplementary Figure S3 indicate that population (Smc2, Spic and Klf12) whereas others genes encoding suspect transcription factors identified demonstrated elevated expression in the FM population from the gene grid array analysis demonstrated elevated (Zfp318 and Stat4). A third subset of genes showed expression in more mature B cells. equivalent expression in these mature cell types (Runx3, Mef2c and Phf1). Similar screening with a number of Gene expression profiling of C2ta-deficient animals suggests a other candidate genes (data not shown) also showed a model of multiple developmental pathways associated with the wide variety of expression profiles indicating a high level induction of genes during splenic T1–T2 transition of gene expression heterogeneity during B-cell differ- One difficulty in analyzing gene grid array experiments entiation into MZ and FM cells. is interpreting the massive amount of information that In a similar set of experiments, we utilized an results from performing whole genome analyses. The alternative protocol to isolate the T1, T2, FM and MZ preceding screens have provided us with potentially populations.11,40,56 This protocol relies upon the differ- hundreds of genes whose products could potentially ential staining of AA4.1, IgM and CD23 to resolve the alter B-cell maturation. Our goal is to sort these gene B-cell subsets (Supplementary Figure S3). AA4.1 þ cells products into pathways that influence the expression of (R3) were resolved into T1 and T2 populations by specific target genes such as Cr2 and Fcer2a. By utilizing differential expression of CD23 by IgM þ cells. AA4.1 known expression pathways, it should be possible to cull cells (R4) were separated into MZ and FM populations out previously described gene products and concentrate by the differential expression of both CD23 and IgM. upon those novel gene products obtained from the RNA was isolated from the subsets and probed, by Affymetrix gene arrays. One such known pathway is quantitative RT-PCR, for the expression of selected that of C2ta.

4 4 250 10 10 R2 R3 200 103 103

150 T2 FSC CD21 2 MZ 2 10 10 FM 100

R3 101 101 50 R2 T1 0 100 100 0 1 2 3 4 0 1 2 3 4 10 10 10 10 10 100 101 102 103 104 10 10 10 10 10 CD23 CD24

60 1.6

BL-6 T1 1.4 50 BL-6 T2 BL-6 MZ BL-6 FM 1.2 40 actin

β 1

30 0.8

0.6 20

Transcript / 1000 Transcript 0.4 10 0.2

0 0 CD21 CD23 Runx3 Zfp318 Smc2 Mef2c Phf1 Stat4 Ccr6 Spic K1f12 Figure 6 Analyses of selected gene expression using CD24/CD21 expression-defined splenic B-cell subsets. (a) FACS sorting of splenic B-cell subsets. B220 þ splenocytes pooled from three 8-week-old C57BL/6 mice (BL-6) were stained for CD21, CD23 and CD24. The CD23 cells (the R2 population) were sorted into MZ and T1 subsets, whereas the CD23 þ cells (the R3 population) were sorted into T2 and FM B cells. (b) Real-time PCR on selected genes using RNA obtained from the T1 (open bar), T2 (light gray bar), MZ (dark gray) and FM (black bar) B-cell populations. Transcript level of each gene is relative to b-actin. The data are an average of three independent cell sort experiments. Gene- specific oligonucleotides are listed in Supplementary Table S2. FACS, flow cytometry analysis; FM, follicular mature; MZ, marginal zone.

Genes and Immunity Gene expression analysis of splenic B-cell maturation I Debnath et al 716 Table 4 Differential expressions of selected mouse genes in Table 4 Continued comparison between B220+ splenic B cells from wild-type and C2ta-deficient mature animals Gene symbol FC

Gene symbol FC Unaffected genes in screen Tulp4 1.0 E2f5 1.2 Genes elevated in wild type Fbxl11 1.0 H2-Eb1 1990 Gatad2b 1.0 H2-Aa 844 Tcf12 1.0 C2ta 82 Jundm2 1.0 Hspa1a 9.4 Clu 1.3 H2-DMa 7.0 Sema7a 1.0 Ltbp3 4.4 Jak3 1.1 Wdfy1 3.3 Fbxo45 3.3 Zfp644 3.2 Rabgap1 3.0 Ctse 2.9 C2ta is the master regulator controlling the expression Zfp2 2.9 of major histocompatibility complex class II genes.57 As Zfp142 2.5 described above, the expression of C2ta was significantly Camk2b 2.4 induced in maturing B cells at the T1–T2 differentiation H2-0a 2.3 interface (coincident, for example, with Cr2 and Icosl Zfp324 2.3 expression) (Supplementary Figure S1). We also Genes elevated in C2ta KO observed increased expression of known C2ta target Slc25a37 41 genes such as H2-Ab1, H2-DMa, H2-Ea, H2-Oa and Ctse Rhced 20 in the 2-week and 8-week spleen (vs the 2-week marrow) Cldn13 13 Affymetrix screens. We therefore wondered if any of the Rhag 11 novel candidate genes identified in our Affymetrix Gypa 11 Trim10 9.0 screens above were also controlled by the C2ta protein Kell 8.8 and would thus show diminished expression in a C2ta- Spna1 8.4 deficient mouse. We chose B220 þ cells from 2-week-old Tmcc2 6.6 splenocytes for this analysis because, as shown above, Fcgr1 5.9 the 2-week wild-type spleen is enriched for transitional Tspan33 5.6 B cells and diminished (compared with older animals) Tal1 5.5 Timd4 5.1 for mature B-cell populations. B cells from C2ta-deficient C2 5.0 mice expressed wild-type levels of CD21 and CD23 Tlr4 4.2 proteins (data not shown). B220 þ cells were isolated Clec4a3 3.8 from a pool of C2ta-deficient spleens (and C57BL/6 Rsad2 3.8 strain-matched controls); RNA was obtained and used to Spic 2.5 screen a mouse Affymetrix gene array. Genes showing a Unaffected genes in screen twofold or greater difference in expression (the same Cr2 1.5 threshold for the previous analyses) were considered Itgb7 1.1 significant (Supplementary Table S6). In these analyses, Ccr6 1.0 357 genes were upregulated twofold or higher in C2ta- Mef2c 1.1 deficient cells in comparison to the WT counterpart and Ms4a4c 1.3 95 genes were found to be elevated twofold or more in Zfp318 1.2 CD86 1.3 the C57BL/6 sample when compared with the C2ta CD36 1.3 mutant animal. Elf2 1.1 Table 4 contains a partial list of genes identified in the Fcer2a 1.7 C2ta analysis highlighting a subset of genes showing Icosl 1.0 elevated expression in either the wild-type sample or the Ctsh 1.2 C2ta-deficient sample and ‘genes of interest’ (obtained Klf12 1.0 Phf1 1.0 from the previous screens) that demonstrated no Ubtf 1.1 difference in expression. As expected, the expression of Smc2 1.4 known C2ta target genes was reduced in the C2ta- Tbx21 (Tbet) 1.0 deficient sample including a variety of major histocom- Stat4 1.0 patibility complex class II genes. The C2ta protein has E2f7 1.0 been primarily described as a transcriptional activator of Zfp53 1.0 Klf3 1.1 class II genes; thus, it was interesting to observe that a Pknox1 1.0 large number of genes showed enhanced expression in Supt16h 1.2 the C2ta-deficient sample suggesting that the protein Atf2 1.0 may also directly (or indirectly) inhibit transcription of HdgfrP2 1.2 other genes. A number of these genes showing elevated Runx3 1.0 expression in the C2ta-deficient cells included those Gtf2i 1.0 Cnr2 1.3 encoding cell-surface glycoproteins and blood group Gfi1 1.5 antigens such as Rhced (the Rhesus blood group, D Bst1 1.3 antigen), Cldn13 (a member of the claudin family of

Genes and Immunity Gene expression analysis of splenic B-cell maturation I Debnath et al 717 extracellular matrix proteins), Rhag (Rhesus blood group- cells in the marrow of the immature animal. Older associated A glycoprotein) and Gypa (glycophorin). animals possess B220 þ CD21 þ mature B cells in the These data suggest that the elevated expression of these marrow either through specific retention or from blood proteins is reciprocal with that of the class II surface circulation, and the presence of such cells in the marrow glycoproteins. However, when we scanned the C2ta B-cell population would have reduced the screen’s grid array for altered expression of genes identified in sensitivity. The 2-week spleen sample was enriched for the previous screens as demonstrating differential immature splenic B cells whereas the 8-week spleen expression (A/WySnJ vs A/J, or 2-week marrow vs sample was highly enriched for mature B-cell types. The 2-week/8-week spleen analyses), we found that most of genes showing significantly greater expression in the 2- these genes showing stage-specific expression did not week and 8-week spleens compared with the 2-week demonstrate differences in expression between the marrow included many of those genes identified in the C2ta-deficient and wild-type B220 þ splenocytes. Other A/J vs A/WySnJ screen (enhanced in the A/J samples), genes, whose expression is dramatically up- or down- as well as a number of additional candidate genes. For regulated at the T1/T2 differentiation step but do not example, the A/WySnJ cells demonstrated an altered show alterations in expression in the absence of C2ta expression (compared with 2-week A/J B220 þ spleno- (genes such as Cr2, Ccr6, Mef2c, FceR2a, Pknox1, Gfi1, cytes) of 612 probe sets; of these, 221 genes also Ctsh, Icosl and so on), must be under the control of demonstrated elevated expression in the A/J bone other regulator protein(s). One candidate in the control marrow sample (compared with the 2-week and 8-week of Fcer2a is Mef2c in that it was recently reported spleen samples). In addition, the A/J 2-week spleen that B cells lacking this transcription factor lost the sample showed 198 probe sets with elevated expression expression of the CD23 protein.58 Therefore, the tran- compared with the A/WySnJ samples. Of these 198 scriptional induction of genes such as Icosl, Cr2 and markers, 89 also demonstrated increased expression in others (that are coincident with C2ta expression) must be the 2-week A/J spleen sample compared with the 2-week accomplished by positive induction or relaxation of A/J marrow sample. suppression through transcriptional pathways yet to be The primary focus of this work was to identify elucidated. potential transcriptional regulators that could activate transcription of Cr2 and Fcer2a genes during splenic B-cell maturation.19,59 One such protein, Pax5, has been implicated in controlling Cr2 and Fcer2a expression,23 Discussion and we have documented that Pax5 binds to the genes In this report, we have used gene grid array analysis in vivo using mature splenic B cells.19 Interestingly, Pax5 with a variety of mouse systems to define those genes is also bound to these two genes as well as CD19 in demonstrating differential expression during splenic immature marrow B cells that express the CD19 protein B-cell maturation. Although the impetus of this work but not the CR1/CR2 or CD23 proteins, suggesting that was to define those pathways that directly regulate the the Cr2 and Fcer2a genes may be transcriptionally expression of Cr2 and Fcer2a, the data obtained revealed repressed in such cells. a large number of genes with differential expression These Affymetrix arrays have provided us with a during B-cell differentiation. number of transcriptional control candidate proteins Our initial experiments addressed gene expression whose expression nearly parallels that of Cr2/Fcer2a (thus between cells arrested at the T1/T2 interface, due to the acting as potential activators) or whose expression is lack of BAFF signaling, with those of B cells isolated from opposite to that of Cr2/Fcer2a (thus potentially acting as an immature wild-type spleen. This block in differentia- transcriptional repressors). For example, Spic, Mef2c, tion in the A/WySnJ samples was confirmed by the Runx3, Zfp318 and Klf12 are genes that encode transcrip- relative loss of transcripts from the Cr2, Icosl and Bcl2 tional control proteins whose expression is lowest in the genes and the increase of transcripts encoding immature bone marrow cells and highest in the mature B cells, and B-cell markers such as Bst1. We were initially surprised could thus serve to regulate transcription of target genes to observe the large number of cell cycle/apoptosis genes in mature B-cell types. Our data show that Spic is most showing elevated expression in the A/WySnJ samples highly expressed in T2 and MZ cells, with less expression (Supplementary Table S1); however, the lack of BAFF in FM cells and only marginal expression in T1 cells. Spic signaling deprives B cells of a critical growth and is a member of the ets gene family and its gene product maintenance factor predisposing them to anergy and has been described as having opposing effects to PU.1 death.14 A number of genes showing elevated expression during B-cell maturation.60,61 Spic has also been impli- in the wild-type sample (compared with the A/WySnJ cated, with STAT-6, in regulating germline IgE trans- cells) were consistently shown to be associated with cription.62 Mef2c (myocyte enhancer factor 2C) has B-cell maturation and differentiation. Thus, the gene set previously been defined as a transcription factor showing elevated expression in the A/WySnJ sample found in B cells following inflammatory stimuli.63–65 included both immature B-cell products and genes Our data show Mef2c transcription to be evident in induced during BAFF deprivation whereas those show- maturing and mature B cells, with FM cells expressing ing elevated expression in the A/J sample were specific the highest levels. Mef2c was recently shown to be for mature B-cell products. critical in B-cell proliferation and survival responses To define additional genes showing differential ex- following antigen activation.58 Runx3 is a member pression during splenic B-cell maturation, we utilized of the runt domain-containing family of transcription Affymetrix analysis of A/J B220 þ cells from 2-week factors that can either activate or suppress trans- marrow, and 2-week and 8-week spleen samples. We cription.66 Runx3 has been shown to be upregulated in chose 2-week marrow because of the paucity of mature B maturing B cells, directly repressing the expression of

Genes and Immunity Gene expression analysis of splenic B-cell maturation I Debnath et al 718 Runx1.67 Our data show that Runx3 transcription Acknowledgements is also highest in the most mature B cells. Zfp318 encodes a zinc finger-containing protein (also known as We thank the University of Utah FACS and Oligonucleo- testicular zinc-finger protein) that is implicated in tide Core Facilities for their assistance. We also thank Dr spermatogenesis and potentially helps to augment or Peter Jensen for providing the C2ta-deficient animals. We repress expression of the androgen receptor through sincerely acknowledge the technical help and sugges- alternative splice variants.68–71 We observe Zfp318 tions from Dr Wayne Green for assistance in the isolation expression with elevated expression in FM splenic B of B-cell subsets. Finally, we also thank all the members cells (compared with marrow precursors) but low of our laboratories for their insightful critiques of this levels in MZ B cells. Finally, Klf12 is highly expressed work. in the most mature B cells analyzed (8-week spleen) and This work was supported by grants from the National shows the highest expression in the MZ population. Institute of Allergy and Infectious Diseases (AI-24158, Klf12 is a member of the basic Kruppel family JHW: AI-32223, JJW). The content is solely the respon- of transcriptional regulators and is implicated in sibility of the authors and does not necessarily represent transcriptional suppression during hematopoietic cell the official views of the Institute of Allergy and differentiation.72–75 Infectious Diseases or the National Institutes of Health. Another set of genes encoding transcriptional regulators are progressively downregulated during the transitional maturation of B cells and thus may allow the expression of genes in mature B cells (by alleviating References transcriptional repression) or result in the loss of proteins expressed in immature B cells (loss of positive regula- 1 Su TT, Rawlings DJ. Transitional B lymphocyte subsets tion). For example, the genes Tcf12, Gfi1, Erg and E2f7 operate as distinct checkpoints in murine splenic B cell development. J Immunol 2002; 168: 2101–2110. show the highest level of expression in the B220 þ 2 Busslinger M. Transcriptional control of early B cell develop- marrow cells with declining expression in the most ment. Annu Rev Immunol 2004; 22: 55–79. mature B cells. Tcf12 (also known as HeLa E-box-binding 3 Allman DM, Ferguson SE, Cancro MP. Peripheral B cell protein) is a member of the E-protein family of maturation. I. Immature peripheral B cells in adults are heat- transcription factors that can form heterodimers with stable antigenhi and exhibit unique signaling characteristics. basic helix–loop–helix proteins.76,77 Animals lacking the J Immunol 1992; 149: 2533–2540. Tcf12 protein die soon after birth but do show normal 4 Allman DM, Ferguson SE, Lentz VM, Cancro MP. Peripheral B B-cell development.78 Animals lacking both Tcf12 and cell maturation. II. Heat-stable antigen(hi) splenic B cells are E2A, however, show a more profound B-cell defect than an immature developmental intermediate in the production of long-lived marrow-derived B cells. J Immunol 1993; 151: either mutation alone, suggesting that B-cell develop- 78 4431–4444. ment requires a combined dosage of E2A and Tcf12. 5 Goodnow CC. Balancing immunity and tolerance: deleting The specific gene targets of Tcf12 in maturing B cells is and tuning lymphocyte repertoires. Proc Natl Acad Sci USA not known although CD4 is a target of Tcf12 control in T 1996; 93: 2264–2271. cells.79 The Gfi1 gene encodes a SNAG domain-contain- 6 Melchers F. Anergic B cells caught in the act. Immunity 2006; ing zinc finger transcriptional repressor that has been 25: 864–867. implicated in T-cell development and growth, the 7 Russell DM, Dembic Z, Morahan G, Miller JF, Burki K, differentiation of neutrophils and immature hematopoie- Nemazee D. Peripheral deletion of self-reactive B cells. Nature tic stem cell differentiation.80–83 Animals lacking Gfi1 1991; 354: 308–311. show reduced numbers of B220 þ marrow cells, which 8 Kench JA, Russell DM, Nemazee D. Efficient peripheral clonal elimination of B lymphocytes in MRL/lpr mice bearing has been shown to be because of reduced expression of, autoantibody transgenes. J Exp Med 1998; 188: 909–917. 84 and signal transduction from, the IL-7R. Erg is in the 9 Carsetti R, Rosado MM, Wardmann H. Peripheral develop- Erg/Fli-1 subfamily of the ets gene family of transcription ment of B cells in mouse and man. Immunol Rev 2004; 197: factors.85 The ERG protein has been most highly studied 179–191. in the development of cartilage formation in chick 10 Loder F, Mutschler B, Ray RJ, Paige CJ, Sideras P, Torres R et al. embryos: its potential role in B-cell development is B cell development in the spleen takes place in discrete steps unknown.86,87 E2f7 is a member of the E2F transcription and is determined by the quality of B cell receptor-derived factor family that includes a number of transcriptional signals. J Exp Med 1999; 190: 75–89. activating and repressing proteins (E2F7) that regulate 11 Allman D, Lindsley RC, DeMuth W, Rudd K, Shinton SA, Hardy RR. Resolution of three nonproliferative immature cell cycle progression and activation (suppression) of 88–91 splenic B cell subsets reveals multiple selection points during target genes. Its role in immature B-cell development peripheral B cell maturation. J Immunol 2001; 167: 6834–6840. is not known. 12 Meyer-Bahlburg A, Andrews SF, Yu KO, Porcelli SA, Rawlings The description above of only a small subset of DJ. Characterization of a late transitional B cell population transcriptional regulators identified in these gene grid highly sensitive to BAFF-mediated homeostatic proliferation. array screens indicates that our list of candidate genes J Exp Med 2008; 205: 155–168. includes those genes with a known role in B-cell 13 Matthias P, Rolink AG. Transcriptional networks in develop- development and function, and a large subset of genes ing and mature B cells. Nat Rev Immunol 2005; 5: 497–508. for which no such role(s) has been identified. Using 14 Rolink AG, Tschopp J, Schneider P, Melchers F. BAFF is a conditional animal knockout models, anti-sense/over- survival and maturation factor for mouse B cells. Eur J Immunol 2002; 32: 2004–2010. expression assays and chromatin immunoprecipitation 15 Sasaki Y, Casola S, Kutok JL, Rajewsky K, Schmidt-Supprian assays, the functions of these proteins involved in B-cell M. TNF family member B cell-activating factor (BAFF) development can be elucidated and the genes they receptor-dependent and -independent roles for BAFF in B regulate be identified. cell physiology. J Immunol 2004; 173: 2245–2252.

Genes and Immunity Gene expression analysis of splenic B-cell maturation I Debnath et al 719 16 Gorelik L, Cutler AH, Thill G, Miklasz SD, Shea DE, Ambrose 35 Lentz VM, Cancro MP, Nashold FE, Hayes CE. Bcmd C et al. Cutting edge: BAFF regulates CD21/35 and CD23 governs recruitment of new B cells into the stable peripheral expression independent of its B cell survival function. B cell pool in the A/WySnJ mouse. J Immunol 1996; 157: J Immunol 2004; 172: 762–766. 598–606. 17 Lin KI, Angelin-Duclos C, Kuo TC, Calame K. Blimp-1- 36 Lentz VM, Hayes CE, Cancro MP. Bcmd decreases the life dependent repression of Pax-5 is required for differentiation of span of B-2 but not B-1 cells in A/WySnJ mice. J Immunol 1998; B cells to immunoglobulin M-secreting plasma cells. Mol Cell 160: 3743–3747. Biol 2002; 22: 4771–4780. 37 Miller DJ, Hayes CE. Phenotypic and genetic characterization 18 Shaffer AL, Shapiro-Shelef M, Iwakoshi NN, Lee AH, Qian SB, of a unique B lymphocyte deficiency in strain A/WySnJ mice. Zhao H et al. XBP1, downstream of Blimp-1, expands the Eur J Immunol 1991; 21: 1123–1130. secretory apparatus and other organelles, and increases 38 Debnath I, Roundy KM, Weis JJ, Weis JH. Analysis of the protein synthesis in plasma cell differentiation. Immunity regulatory role of BAFF in controlling the expression of CD21 2004; 21: 81–93. and CD23. Mol Immunol 2007; 44: 2388–2399. 19 Debnath I, Roundy KM, Weis JJ, Weis JH. Defining in vivo 39 Chang CH, Guerder S, Hong SC, van Ewijk W, Flavell RA. transcription factor complexes of the murine CD21 and CD23 Mice lacking the MHC class II transactivator (CIITA) show genes. J Immunol 2007; 178: 7139–7150. tissue-specific impairment of MHC class II expression. 20 Tinnell SB, Jacobs-Helber SM, Sterneck E, Sawyer ST, Conrad Immunity 1996; 4: 167–178. DH. STAT6, NF-kappaB and C/EBP in CD23 expression and 40 Srivastava B, Quinn III WJ, Hazard K, Erikson J, Allman D. IgE production. Int Immunol 1998; 10: 1529–1538. Characterization of marginal zone B cell precursors. J Exp Med 21 Visan I, Goller M, Berberich I, Kneitz C, Tony HP. Pax-5 is a 2005; 202: 1225–1234. key regulator of the B cell-restricted expression of the CD23a 41 Crandall H, Dunn DM, Ma Y, Wooten RM, Zachary JF, Weis JH isoform. Eur J Immunol 2003; 33: 1163–1173. et al. Gene expression profiling reveals unique pathways 22 Kneitz C, Goller M, Tony H, Simon A, Stibbe C, Konig T et al. associated with differential severity of lyme arthritis. The CD23b promoter is a target for NF-AT transcription J Immunol 2006; 177: 7930–7942. factors in B-CLL cells. Biochim Biophys Acta 2002; 1588: 42 Bolstad BM, Irizarry RA, Astrand M, Speed TP. A comparison 41–47. of normalization methods for high density oligonucleotide 23 Horcher M, Souabni A, Busslinger M. Pax5/BSAP maintains array data based on variance and bias. Bioinformatics 2003; 19: the identity of B cells in late B lymphopoiesis. Immunity 2001; 185–193. 14: 779–790. 43 Irizarry RA, Bolstad BM, Collin F, Cope LM, Hobbs B, Speed 24 Schiemann B, Gommerman JL, Vora K, Cachero TG, Shulga- TP. Summaries of Affymetrix GeneChip probe level data. Morskaya S, Dobles M et al. An essential role for BAFF in the Nucleic Acids Res 2003; 31: e15. normal development of B cells through a BCMA-independent 44 Gautier L, Cope L, Bolstad BM, Irizarry RA. affy–analysis of pathway. Science 2001; 293: 2111–2114. Affymetrix GeneChip data at the probe level. Bioinformatics 25 Smith SH, Cancro MP. Cutting edge: B cell receptor signals 2004; 20: 307–315. regulate BLyS receptor levels in mature B cells and their 45 Tan SS, Weis JH. Development of a sensitive reverse immediate progenitors. J Immunol 2003; 170: 5820–5823. transcriptase PCR assay, RT-RPCR, utilizing rapid cycle times. 26 Patke A, Mecklenbrauker I, Erdjument-Bromage H, Tempst P, Genome Res 1992; 2: 137–143. Tarakhovsky A. BAFF controls B cell metabolic fitness through 46 Terahara K, Yoshida M, Igarashi O, Nochi T, Pontes GS, Hase a PKC beta- and Akt-dependent mechanism. J Exp Med 2006; K et al. Comprehensive gene expression profiling of Peyer’s 203: 2551–2562. Patch M cells, villous M-like cells, and intestinal epithelial 27 Gross JA, Dillon SR, Mudri S, Johnston J, Littau A, Roque R cells. J Immunol 2008; 180: 7840–7846. et al. TACI-Ig neutralizes molecules critical for B cell 47 Bhattacharya D, Cheah MT, Franco CB, Hosen N, Pin CL, Sha development and autoimmune disease. impaired B cell WC et al. Transcriptional profiling of antigen-dependent maturation in mice lacking BLyS. Immunity 2001; 15: 289–302. murine B cell differentiation and memory formation. 28 Kayagaki N, Yan M, Seshasayee D, Wang H, Lee W, French J Immunol 2007; 179: 6808–6819. DM et al. BAFF/BLyS receptor 3 binds the B cell survival factor 48 Kin NW, Crawford DM, Liu J, Behrens TW, Kearney JF. DNA BAFF ligand through a discrete surface loop and promotes microarray gene expression profile of marginal zone versus processing of NF-kappaB2. Immunity 2002; 17: 515–524. follicular B cells and idiotype positive marginal zone B cells 29 Claudio E, Brown K, Park S, Wang H, Siebenlist U. BAFF- before and after immunization with Streptococcus pneumoniae. induced NEMO-independent processing of NF-kappa B2 in J Immunol 2008; 180: 6663–6674. maturing B cells. Nat Immunol 2002; 3: 958–965. 49 Lindvall JM, Blomberg KE, Berglof A, Smith CI. Distinct gene 30 Hatada EN, Do RK, Orlofsky A, Liou HC, Prystowsky M, expression signature in Btk-defective T1 B-cells. Biochem MacLennan IC et al. NF-kappa B1 p50 is required for BLyS Biophys Res Commun 2006; 346: 461–469. attenuation of apoptosis but dispensable for processing of NF- 50 Hoffmann R, Seidl T, Neeb M, Rolink A, Melchers F. Changes kappa B2 p100 to p52 in quiescent mature B cells. J Immunol in gene expression profiles in developing B cells of murine 2003; 171: 761–768. bone marrow. Genome Res 2002; 12: 98–111. 31 Morrison MD, Reiley W, Zhang M, Sun SC. An atypical tumor 51 van Haaften RI, Schroen B, Janssen BJ, van Erk A, Debets JJ, necrosis factor (TNF) receptor-associated factor-binding motif Smeets HJ et al. Biologically relevant effects of mRNA of B cell-activating factor belonging to the TNF family (BAFF) amplification on gene expression profiles. BMC Bioinformatics receptor mediates induction of the noncanonical NF-kappaB 2006; 7: 200. signaling pathway. J Biol Chem 2005; 280: 10018–10024. 52 Grimaldi CM, Cleary J, Dagtas AS, Moussai D, Diamond B. 32 Craxton A, Draves KE, Gruppi A, Clark EA. BAFF regulates B Estrogen alters thresholds for B cell apoptosis and activation. cell survival by downregulating the BH3-only family member J Clin Invest 2002; 109: 1625–1633. Bim via the ERK pathway. J Exp Med 2005; 202: 1363–1374. 53 Fink K, Manjarrez-Orduno N, Schildknecht A, Weber J, Senn 33 Yan M, Brady JR, Chan B, Lee WP, Hsu B, Harless S et al. BM, Zinkernagel RM et al. B cell activation state-governed Identification of a novel receptor for B lymphocyte stimulator formation of germinal centers following viral infection. that is mutated in a mouse strain with severe B cell deficiency. J Immunol 2007; 179: 5877–5885. Curr Biol 2001; 11: 1547–1552. 54 Suzuki A, Kaisho T, Ohishi M, Tsukio-Yamaguchi M, Tsubata 34 Amanna IJ, Dingwall JP, Hayes CE. Enforced bcl-xL gene T, Koni PA et al. Critical roles of Pten in B cell homeostasis and expression restored splenic B lymphocyte development in immunoglobulin class switch recombination. J Exp Med 2003; BAFF-R mutant mice. J Immunol 2003; 170: 4593–4600. 197: 657–667.

Genes and Immunity Gene expression analysis of splenic B-cell maturation I Debnath et al 720 55 Kuroda K, Han H, Tani S, Tanigaki K, Tun T, Furukawa T et al. basal and IFN-gamma regulated expression of the human Regulation of marginal zone B cell development by MINT, a complement C4 promoter. J Immunol 2000; 164: 300–307. suppressor of Notch/RBP-J signaling pathway. Immunity 2003; 74 Turner J, Crossley M. Basic Kruppel-like factor functions 18: 301–312. within a network of interacting haematopoietic transcription 56 Shahaf G, Allman D, Cancro MP, Mehr R. Screening of factors. Int J Biochem Cell Biol 1999; 31: 1169–1174. alternative models for transitional B cell maturation. Int 75 Zhang P, Basu P, Redmond LC, Morris PE, Rupon JW, Ginder Immunol 2004; 16: 1081–1090. GD et al. A functional screen for Kruppel-like factors that 57 Reith W, LeibundGut-Landmann S, Waldburger JM. Regula- regulate the human gamma-globin gene through the CACCC tion of MHC class II gene expression by the class II promoter element. Blood Cells Mol Dis 2005; 35: 227–235. transactivator. Nat Rev Immunol 2005; 5: 793–806. 76 Parker MH, Perry RL, Fauteux MC, Berkes CA, Rudnicki MA. 58 Wilker PR, Kohyama M, Sandau MM, Albring JC, Nakagawa MyoD synergizes with the E-protein HEB beta to induce O, Schwarz JJ et al. Transcription factor Mef2c is required for B myogenic differentiation. Mol Cell Biol 2006; 26: 5771–5783. cell proliferation and survival after antigen receptor stimula- 77 Quong MW, Romanow WJ, Murre C. E protein function in tion. Nat Immunol 2008; 9: 603–612. lymphocyte development. Annu Rev Immunol 2002; 20: 301–322. 59 Zabel MD, Weis JH. Cell-specific regulation of the CD21 gene. 78 Zhuang Y, Cheng P, Weintraub H. B-lymphocyte development Int Immunopharmacol 2001; 1: 483–493. is regulated by the combined dosage of three basic helix- 60 Schweitzer BL, Huang KJ, Kamath MB, Emelyanov AV, loop-helix genes, E2A, E2-2, and HEB. Mol Cell Biol 1996; 16: Birshtein BK, DeKoter RP. Spi-C has opposing effects to 2898–2905. PU.1 on gene expression in progenitor B cells. J Immunol 2006; 79 Sawada S, Littman DR. A heterodimer of HEB and an E12- 177: 2195–2207. related protein interacts with the CD4 enhancer and regulates 61 Bemark M, Martensson A, Liberg D, Leanderson T. Spi-C, a its activity in T-cell lines. Mol Cell Biol 1993; 13: 5620–5628. novel Ets protein that is temporally regulated during B 80 Gilks CB, Bear SE, Grimes HL, Tsichlis PN. Progression of lymphocyte development. J Biol Chem 1999; 274: 10259–10267. interleukin-2 (IL-2)-dependent rat T cell lymphoma lines to IL- 62 Carlsson R, Thorell K, Liberg D, Leanderson T. SPI-C and 2-independent growth following activation of a gene (Gfi-1) STAT6 can cooperate to stimulate IgE germline transcription. encoding a novel zinc finger protein. Mol Cell Biol 1993; 13: Biochem Biophys Res Commun 2006; 344: 1155–1160. 1759–1768. 63 Martin JF, Schwarz JJ, Olson EN. Myocyte enhancer factor 81 Hock H, Hamblen MJ, Rooke HM, Schindler JW, Saleque S, (MEF) 2C: a tissue-restricted member of the MEF-2 family Fujiwara Y et al. Gfi-1 restricts proliferation and preserves of transcription factors. Proc Natl Acad Sci USA 1993; 90: functional integrity of haematopoietic stem cells. Nature 2004; 5282–5286. 431: 1002–1007. 64 Swanson BJ, Jack HM, Lyons GE. Characterization of myocyte 82 Zeng H, Yucel R, Kosan C, Klein-Hitpass L, Moroy T. enhancer factor 2 (MEF2) expression in B and T cells: MEF2C Transcription factor Gfi1 regulates self-renewal and engraft- is a B cell-restricted transcription factor in lymphocytes. Mol ment of hematopoietic stem cells. EMBO J 2004; 23: 4116–4125. Immunol 1998; 35: 445–458. 83 Hock H, Hamblen MJ, Rooke HM, Traver D, Bronson RT, 65 Han J, Jiang Y, Li Z, Kravchenko VV, Ulevitch RJ. Activation of Cameron S et al. Intrinsic requirement for zinc finger the transcription factor MEF2C by the MAP kinase p38 in transcription factor Gfi-1 in neutrophil differentiation. inflammation. Nature 1997; 386: 296–299. Immunity 2003; 18: 109–120. 66 Whiteman HJ, Farrell PJ. RUNX expression and function in 84 Rathinam C, Klein C. Transcriptional repressor Gfi1 integrates human B cells. Crit Rev Eukaryot Gene Expr 2006; 16: 31–44. cytokine-receptor signals controlling B-cell differentiation. 67 Spender LC, Whiteman HJ, Karstegl CE, Farrell PJ. Transcrip- PLoS ONE 2007; 2: e306. tional cross-regulation of RUNX1 by RUNX3 in human B cells. 85 Sharrocks AD. The ETS-domain transcription factor family. Oncogene 2005; 24: 1873–1881. Nat Rev Mol Cell Biol 2001; 2: 827–837. 68 Tao RH, Kawate H, Ohnaka K, Ishizuka M, Hagiwara H, 86 Iwamoto M, Tamamura Y, Koyama E, Komori T, Takeshita N, Takayanagi R. Opposite effects of alternative TZF spliced Williams JA et al. Transcription factor ERG and joint and variants on androgen receptor. Biochem Biophys Res Commun articular cartilage formation during mouse limb and spine 2006; 341: 515–521. skeletogenesis. Dev Biol 2007; 305: 40–51. 69 Ishizuka M, Kawate H, Takayanagi R, Ohshima H, Tao RH, 87 Yang L, Xia L, Wu DY, Wang H, Chansky HA, Schubach WH Hagiwara H. A zinc finger protein TZF is a novel corepressor et al. Molecular cloning of ESET, a novel histone H3-specific of androgen receptor. Biochem Biophys Res Commun 2005; 331: methyltransferase that interacts with ERG transcription factor. 1025–1031. Oncogene 2002; 21: 148–152. 70 Inoue A, Ishiji A, Kasagi S, Ishizuka M, Hirose S, Baba T et al. 88 Logan N, Delavaine L, Graham A, Reilly C, Wilson J, The transcript for a novel protein with a zinc finger motif is Brummelkamp TR et al. E2F-7: a distinctive E2F family expressed at specific stages of mouse spermatogenesis. member with an unusual organization of DNA-binding Biochem Biophys Res Commun 2000; 273: 398–403. domains. Oncogene 2004; 23: 5138–5150. 71 Ishizuka M, Ohshima H, Tamura N, Nakada T, Inoue A, 89 Di Stefano L, Jensen MR, Helin K. E2F7, a novel E2F featuring Hirose S et al. Molecular cloning and characteristics of a novel DP-independent repression of a subset of E2F-regulated zinc finger protein and its splice variant whose transcripts genes. EMBO J 2003; 22: 6289–6298. are expressed during spermatogenesis. Biochem Biophys Res 90 de Bruin A, Maiti B, Jakoi L, Timmers C, Buerki R, Leone G. Commun 2003; 301: 1079–1085. Identification and characterization of E2F7, a novel mamma- 72 Roth C, Schuierer M, Gunther K, Buettner R. Genomic lian E2F family member capable of blocking cellular prolifera- structure and DNA binding properties of the human zinc tion. J Biol Chem 2003; 278: 42041–42049. finger transcriptional repressor AP-2rep (KLF12). Genomics 91 Maiti B, Li J, de Bruin A, Gordon F, Timmers C, Opavsky R 2000; 63: 384–390. et al. Cloning and characterization of mouse E2F8, a novel 73 Ulgiati D, Subrata LS, Abraham LJ. The role of Sp family mammalian E2F family member capable of blocking cellular members, basic Kruppel-like factor, and E box factors in the proliferation. J Biol Chem 2005; 280: 18211–18220.

Supplementary Information accompanies the paper on Genes and Immunity website (http://www.nature.com/gene)

Genes and Immunity