Identification of Pax5 Target Genes in Early B Cell Differentiation¹

The B lymphocytes are produced in a stepwise process from self-renewing hemopoietic stem cells (HSCs)⁵ in the fetal liver and postnatal bone marrow (BM). In recent years it has become apparent that this process is controlled by a complex transcription factor network that both activates lineage-specific gene expression (lineage specification) and restricts the differentiation options of HSCs and their progeny (lineage commitment) (1, 2, 3). Although there has been extensive analysis of the transcription factors that regulate the initial steps in B lymphopoiesis, relatively little is known about the molecular targets of these factors that ultimately mediate the commitment process.

The transcription factor Pax5 is essential for B lymphopoiesis, as development is arrested at an early pro-B cell stage in the BM of Pax5-deficient mice (4, 5). These Pax5^−/− pro-B cells can be propagated in vitro in the presence of IL-7 and stromal cells and maintain an early B cell phenotype characterized by the expression of B cell-specific transcripts such as Cd79b (B29), Igll5 (λ5), and VpreB1 and D-J recombination events at the Igh locus (4, 6). Pax5-deficient pro-B cells, however, display a remarkable phenotype in that they are not committed to the B cell lineage and are able to differentiate into virtually all hemopoietic cell lineages in vitro and in vivo (7, 8, 9, 10).

Pax5 promotes B lymphopoiesis by activating B cell-specific genes such as those involved in pre-BCR signaling, including Cd19 (11), Cd79a (mb-1) (4, 12), and Blnk (13), as well as Igll5 and VpreB1 (14). Although the inability to express the pre-BCR was potentially the cause of the developmental block in the absence of Pax5, the introduction of functionally rearranged Igh and chimeric Igh-Igβ transgenes into the Pax5 mutant background was unable to progress B cell development beyond the early pro-B cell stage (15). Pax5 also functions to repress genes whose expression is not usually associated with the B cell program. The Pax5-dependent repression of the csf1r and Notch1 genes illustrates at the molecular level how developmental options are suppressed in committed B lymphocytes, as these cells are no longer responsive to the myeloid cytokine M-CSF or to the T cell-inducing Notch1 ligand Delta-like 1 (7, 16, 17). Pax5 also functions to repress genes associated with multipotency such as Flt3, which, although required for early hemopoiesis, must be silenced by Pax5 to allow B lymphopoiesis to proceed (18). A more global approach to identifying target genes using cDNA microarray technology has confirmed that many Pax5-repressed genes are normally expressed in non-B cell lineages and interestingly found that a number of those are reactivated during the physiological down-regulation of Pax5 during plasma cell differentiation, whereas many B cell-specific genes are positively regulated by Pax5 (19, 20).

As an alternative approach to identify potential Pax5 target genes, we performed a screen of a mature B cell cDNA microarray to compare gene expression between wild-type and Pax5^−/− pro-B cells. This screen has resulted in the identification of >100 genes representing both Pax5-activated and -repressed targets, many of which were not detected in the previous studies. These genes are known or predicted to perform a diverse range of functions within the cell and highlight the dual function of Pax5 to repress inappropriate gene expression while further activating essential components of the B cell program.

Pax5^−/− (5) mice were maintained on a C57BL/6 background and genotyped as described (4).

Pro-B cell lines were derived and propagated as described (4). The Pax5-estrogen receptor (ER) fusion protein (Pax5ER) and control ER retroviral vectors were introduced into Pax5^−/− pro-B cells and activated by 1 μM β-estradiol as previously reported (6). Some experiments were performed in the presence of 50 μg/ml cycloheximide (Sigma-Aldrich).

A B lymphocyte cDNA library was a generous gift from C. Estes and M. Bento Soares of the University of Iowa (Iowa City, IA) (21). The microarray was prepared at the Australian Genome Research Facility (Melbourne, Australia), where the inserts from 10,000 randomly generated clones were PCR amplified, purified, and spotted onto glass slides in duplicate. A second screen with an identical strategy used microarrays printed with the 15,000 clone mouse cDNA library of the National Institute on Aging (NIA 15k) (22). Total RNA was isolated from four pro-B cell lines (10⁸ cells) using TRIzol (Invitrogen Life Technologies) and mRNA isolated using the Oligotex mRNA isolation kit (Qiagen). mRNA (4 μg) was reverse transcribed using SuperScript II (Invitrogen Life Technologies) according to instructions from the manufacturer. The purified cDNA samples were labeled with the fluorescent dye esters N-hydroxysuccinimide (NHS)-Cy3 and NHS-Cy5 (Amersham Biosciences), and the microarrays were hybridized for 16 h at 42°C with each of the fluorescently labeled probes in 25% formamide, 5× SSC, 0.1% SDS, 5 μg/ml Cot-1 DNA (Invitrogen Life Technologies), 10 mg/ml poly(A) (Sigma-Aldrich), and 10 mg/ml salmon sperm DNA (Sigma-Aldrich). Slides were washed at room temperature with 1× SSC plus 0.2% SDS and then 0.1× SSC plus 0.2% SDS followed by two stringent washes (0.1× SSC) before scanning on a GenePix 4000B (Molecular Devices). For the B lymphocyte screen, nine arrays were used to compare one Pax5^+/+ and three Pax5^−/− pro-B cell lines. For the NIA15k screen, 12 arrays were used to compare Pax5^+/+, Pax5^−/−, and Rag1^−/− pro-B cell lines in a saturated loop design. The scanned microarray images were quantified using SPOT image analysis software (www.hca-vision.com). Statistical analysis of the microarray data used the Bioconductor software package LIMMA (www.bioconductor.org). Quality assessment using MA plots resulted in three of the NIA 15k arrays being discarded due to apparent hybridization problems. Expression values were print-tip loess normalized, with spot quality weights used to down-weight spots with unusually small or large scanned sizes (23). The duplicate spots for each probe were combined using a common correlation algorithm that uses both within and between array variability to assess statistical significance (24). Clones were ranked for differential expression between Pax5^−/− and Pax5^+/+ using empirical Bayes modified t statistics (25). p values were adjusted for multiple testing using the Benjamini and Hochberg algorithm to control the false discovery rate. Raw data from the NIA 15k arrays is available from the Gene Expression Omnibus (GEO) database (www.ncbi.nlm.nih.gov/projects/geo) as series GSE9345.

Plasmid DNA from differentially expressed clones was prepared and the insert sequenced. Clones identities were determined using the basic local alignment search tool (BLAST).

Total RNA was isolated from sorted B cell populations using TRIzol (Invitrogen Life Technologies) and reverse-transcribed with Moloney murine leukemia virus reverse transcriptase (Promega) according to the manufacturer’s instructions. cDNA was amplified using the gene-specific primers listed in Table I. cDNA input was normalized to Hprt.

The labeled mAbs CD157, CXCR5 (Blr1), and Syndecan-4 (Sdc4) were purchased from BD Pharmingen. The mAbs against B220, CD19, Flt3 c-Kit, CD24a, and Embigin (Emb) were purified and conjugated in our laboratory. HSC, common lymphoid progenitor (CLP), and B cell populations were sorted as previously described (26, 27). For flow cytometry, single-cell suspensions were prepared from the BM of 12-day old mice and stained with the appropriate mAb in PBS containing 2% FCS. Biotinylated mAb were revealed by PE-streptavidin (SouthernBiotech). Cells were analyzed on a FACScan flow cytometer (BD Biosciences) and cell sorting was conducted on high-speed flow cytometers (Vantage SE DiVa; BD Biosciences). Dead cells were excluded by propidium iodide staining.

G7.43.1, the anti-Emb mAb, was generated by immunizing a rat with human 293T cells infected with a retrovirus-expressing mouse Emb. Supernatants from hybridoma clones were screened against Pax5^−/− pro-B cells. The specificity of G7.43.1 was determined by Western blot analysis.

Tissues were embedded, stored, sectioned, and stained as described (28). Abs used were unlabeled anti-Emb and biotinylated anti-CD19. Anti-GL117 was an isotype control. Emb and GL117 were detected with mouse anti-rat Abs conjugated to HRP (BD Pharmingen) and anti-CD19 with streptavidin-alkaline phosphatase (SouthernBiotech). Alkaline phosphatase was visualized with the Fast Blue kit (Vector Laboratories), with endogenous phosphatases blocked by the addition of 2 mM levamisole (Sigma-Aldrich). HRP was visualized with the 3-amino-9-ethylcarbazole substrate kit (Vector Laboratories).

To further our understanding of gene regulation by Pax5 in early B cell development, we performed cDNA microarray analysis on wild-type and Pax5^−/− pro-B cells. As starting material we took advantage of the ability to grow pure populations of pro-B cells in long-term cultures in the presence of IL-7 and a ST-2 stromal layer. Previous analysis has confirmed that these pro-B cells retain many of the features of the equivalent populations in vivo (4). The experimental setup involved a pairwise comparison of three Pax5^−/− pro-B cells lines and a single wild-type control, SB^+/+ (7). Hybridizations were performed on microarrays printed with two mouse cDNA libraries, a custom cDNA microarray consisting of 10,000 randomly selected cDNAs from an IgM⁺ mature B cell library and the NIA 15k library (22) derived from diverse sources including embryonic, neural, and malignant samples. The combination of these two libraries allowed us to maximize the identification of both B cell and non-B cell-specific Pax5 target genes. All hybridizations were performed in duplicate with the Cy3 and Cy5 fluorochromes swapped between each experiment and with three independent Pax5^−/− pro-B cell lines.

The custom B cell microarray screens yielded a large number of clones with statistically significant differential expression in Pax5^−/− (4919 probes at a false discovery rate of <5%). Approximately 200 of the most highly differentially expressed clones were sequenced and identified by database searches. As expected from a library comprised of random clones, the differentially expressed genes were represented by multiple cDNA clones (16 genes were present in two or three copies in the initial 200 clones). The NIA 15k microarray screens yielded another 1491 differentially expressed clones at a false discovery rate of <5%. Fifty-two of the most significant clones were sequence confirmed. Overall, ∼60% of the clones had reduced expression in Pax5^−/− pro-B cells, representing potential Pax5-activated genes, and 40% were more strongly expressed in the absence of Pax5, representing Pax5-repressed genes. The sequence-validated candidates are shown in Tables II and III. Approximately 50% (54 of 109) has been confirmed by independent methodologies as outlined in the Tables II and III and Figs. 1 and 2. This list includes a number of previously identified Pax5 regulated genes, most notably the canonical Pax5 target gene Cd19 (4, 11) which was the most differentially expressed transcript on the B cell array (Table II). As expected Pax5 itself was also differentially expressed. Several other known Pax5 targets, including Blnk (13), Lef1, Cd79a (6), Notch1 (16), Ikzf3 (29), and Ebf1 (29, 30), were also identified, confirming the validity of the screen. Importantly, analysis of the expression of a selection of the potential Pax5 target genes confirmed the differential expression in three independent Pax5^−/− pro-B cell lines (Fig. 1 B and data not shown).

Of the potential Pax5-activated genes, a striking 28% were nuclear proteins involved in various aspects of transcription, 15% were cell surface receptors, 15% were involved in intracellular signaling pathways, and a further 15% were of unknown function. The remaining genes have been implicated in a variety of cellular functions such as metabolism, transport, and cytoskeletal integrity. The Pax5-repressed genes contained a relatively large proportion of cell surface (26%) and intracellular signal-transducing (17%) proteins, with nuclear proteins (21%) and metabolic enzymes (12%) making up a significant fraction of the clones.

Although the pro-B cell culture system provided an abundant and pure source of mRNA with which to conduct the array screening, analysis of differentially expressed genes in pro-B cells in vivo provided an important additional validation to exclude any differences between the genotypes that resulted from differences in the ability of the cells of the two genotypes to be propagated in culture. Indeed Lmo2, which was identified as a Pax5-repressed gene from the B cell array and confirmed as being differentially expressed in pro-B cell cultures, was not differentially regulated on ex vivo isolated pro-B cells (data not shown). Interestingly, the expression of Tcfe2a, which encodes the essential regulator of early B cell development, E2A, was also reduced in Pax5-deficient pro-B cell cultures, yet the expression of Tcfe2a was actually increased in mutant pro-B cells ex vivo (Table II and Fig. 2 C).

Semiquantitative RT-PCR on cDNA isolated from FACS-sorted B220⁺c-Kit⁺ pro-B cells from either genotype confirmed the differential expression of the many of the candidates (Fig. 2). In a number of cases, the putative target genes were cell surface molecules against which mAbs were available and expression could be assessed by flow cytometry. In this manner, CD157, CD24a, and Sdc4 were shown to rely to varying degrees on Pax5 (Fig. 3), whereas CXCR5 (Blr1) was confirmed as being expressed in Pax5-deficient pro-B cell cultures (Fig. 1 C).

We have previously shown that Pax5 directly regulates the expression of a variety of genes in pro-B cells including Cd19 and Flt3 (6, 18). In those cases, direct regulation was initially inferred by the complementation of Pax5^−/− pro-B cells with a retrovirus expressing an inducible Pax5-ER fusion protein and subsequently proven using direct promoter analysis. To assess to what extent the differentially expressed genes identified here were direct Pax5 targets, we used the Pax5-ER system to induce Pax5 activity in the presence and absence of the protein translation inhibitor cycloheximide and assessed the expression of the relevant genes by RT-PCR. In this manner the transcription factors Ikzf3 (encoding for Aiolos) and Spib, which play important roles at the pre-B and mature B cell stages of differentiation, as well as the E3 ubiquitin ligase Syvn1 appear to be direct Pax5 targets (Fig. 4). Interestingly, Irf8 (Icsbp), which was identified using a candidate approach by us and others (20), also appeared to be directly regulated by Pax5 (Fig. 4). Despite its reduced expression in the absence of Pax5, we were unable to induce Ebf1 using this system.

One gene that was strongly and specifically expressed in Pax5-deficient pro-B cells in vitro and in vivo was Emb (Fig. 1). Emb is a transmembrane glycoprotein belonging to the Ig superfamily that is expressed in the early stages of mouse embryogenesis and is postulated to mediate integrin-cell-substratum adhesion (31). Analysis of Emb during normal hemopoiesis revealed expression in T cell and myeloid cells but not B cells (Fig. 5,A and data not shown). Specific analysis of progenitors showed that Emb was strongly expressed in HSC- and CLP-enriched populations and later repressed in pro- and pre-B cells (Fig. 5 A). This was a virtually identical expression pattern to that of Flt3, which we have shown is directly repressed by Pax5 upon B cell lineage commitment (18).

To further analyze the Emb expression domain, we generated a specific mAb by immunizing rats with the 293T cell line expressing full-length mouse Emb. Supernatants from hybridoma clones were initially screened by flow cytometry on wild-type and Pax5^−/− pro-B cells lines. A single clone, G7.43.1, with specific reactivity against Pax5-deficient pro-B cells was identified and further characterized. The specificity of G7.43.1 was confirmed by Western blotting of parental and Emb-transfected 293T cells (data not shown).

Analysis of Pax5^−/− BM confirmed the differential expression of Emb protein on pro-B cells in vivo (Fig. 3). A detailed flow cytometric analysis of the hemopoietic compartment using biotinylated anti-Emb Ab revealed very strong expression in thymocytes, a finding that was confirmed by immunohistochemistry (Fig. 5,B). Emb was also expressed at a moderate level in mature peripheral CD4⁺ and CD8⁺ T cells and NK cells, and at a very high level in myeloid cells (Fig. 5,B and data not shown). In keeping with the RT-PCR data, the majority of BM, spleen, and lymph node B cells did not express Emb (Fig. 5, B and C). Initial attempts to use the anti-Emb Ab to assess the function of this poorly characterized molecule in non-B cells has not been successful, as the addition of the Ab did not affect thymopoiesis using fetal thymic organ culture, mature T cell proliferation, or NK cell cytolytic function (S. L. Nutt and M. Polli, unpublished observations). Anti-Emb did however partially deplete CD8⁺ T cells when injected in vivo (S. L. Nutt and M. Polli, unpublished observations). Interestingly, we and others have recently shown that Emb is re-expressed in activated B cells and plasma cells and that this expression correlated with the posttranslational inhibition of Pax5 function (32). Thus Emb expression in B cells represents a convenient marker for impaired Pax5 function.

Pax5 is essential for many aspects of B cell biology, including the initiation of B cell lineage commitment in the BM, V-DJ recombination of the Igh locus, and the maintenance of the B cell fate in more mature cells (33). In an attempt to understand the molecular events mediated by Pax5 in lineage commitment and early B cell differentiation, we used a cDNA microarray approach to identify genes regulated by Pax5. The putative Pax5 target genes identified here, as well as those isolated in the recent studies (19, 20), provide a wealth of material to further define existing roles and identify new functions of Pax5 in B cells.

Analysis of the microarray data revealed that Pax5 potentially activated the majority of targets identified thus far, as expression was relatively higher in wild-type pro-B cells. Numerous genes involved in signaling have been previously identified as Pax5 targets, including Cd19, Cd79a, Lef1 (6), and Blnk (13), were confirmed in the current screen. Other signaling molecules include Cd97, a transmembrane receptor involved in intracellular signaling in association with G proteins in immune cells (34), and Cd24a (heat-stable Ag or HSA), a surface glycoprotein expressed on immature hemopoietic cells involved in cell adhesion and activation that has a minor role in B lymphopoiesis in the BM (35). Pax5 also activates Bst1 (encoding CD157) and Sdc4, two receptors that play roles in mature B cells. Bst1^−/− mice have delayed B1 B cell development and impaired T-independent Ab responses (36), whereas Sdc4 is expressed broadly in the B cell lineage and is involved in adhesion to the extracellular matrix (37). Pax5 also activates a variety of intracellular signaling molecules including Rasgrp3, which is required for BCR-induced signaling and proliferation (38), along with a number of other molecules with less characterized roles in lymphocyte signaling including protein kinase D2 (Prkd2), Lrmp, Stk3, Stk4, and Wnk1.

The potential outcome of the altered expression of a group of adhesion receptors and intracellular signaling molecules may be manifested in a defect in homing or adhesion. Indeed, lymphoid progenitors undergo a dramatic change in cell migration and adhesion at B cell commitment, where committed early B cells adhere to a specialized BM niche consisting of IL-7-expressing stromal cells. In contrast, uncommitted progenitors adhere to IL-7-negative BM stroma (39). Interestingly, Nedd9 (also called HEF1), which is an important component of the cytoskeleton-linked signaling pathway initiated by ligation of the BCR (40), was dependent on Pax5 for its expression. In agreement with these results, Pax5^−/− pro-B cells have recently been shown in a Transwell assay to have increased CXCL12-induced migration and, most significantly, have greatly impaired integrin-mediated substrate adhesion compared with wild-type pro-B cells (20). Pax5 is also regulates the cellular proliferation that is required for the expansion phase of BM lymphopoiesis, first by IL-7 and then as a result of signals received from the pre-BCR, as Pax5 regulates Ccnd3 (cyclin D3), which is crucial for the proliferation of early B cells (41).

It is striking that many of the Pax5-activated genes identified here were nuclear proteins. A number of the genes identified were well-known regulators of B cell differentiation and function, including Ikzf3 (Aiolos), Spib, and Irf8 (Icsbp), as well as the previously identified regulators Lef1 (6) and Ebf1 (30). In most cases this regulation appeared to be direct, suggesting that Pax5 functions in part to reinforce the B cell program by further activating the downstream transcriptional cascade. Interestingly, Ikzf3 and Ebf1 expression was also dependent on Pax5 in the chicken DT40 B cell line, although that study did not determine whether the regulation was direct (29). Although these factors have many functions within the B cell lineage, it is interesting to note that they all have roles in early B cell development, particularly at the pre-B cell stage (2, 3, 42). The genes encoding two E proteins, Tcfe2a (encoding E2A) and Tcf4 (encoding E2-2), were also differentially expressed, as was the inhibitor of E proteins, Id3. E2-2 is required for optimal pro-B cell expansion (43), whereas Id3 functions to inhibit growth and induce apoptosis in pro-B cells as well as late stage B cells (44, 45). The differential expression of Tcfe2a and Id3 was surprising, as previous studies have not found these genes to be Pax5 dependent (4, 46). Although the available genetic evidence suggests that Tcfe2a acts upstream of Pax5 during B cell commitment (47), the recent finding that Ebf1 also acts both upstream and downstream of Pax5 raised the possibility that a similar situation occurred with Tcfe2a (30). Comparison of cultured and ex vivo isolated pro-B cells revealed that the differential expression of Tcfe2a was an artifact of adaptation of the cells to in vitro propagation, as the expression of Tcfe2a in pro-B cell analyzed ex vivo did not depend on Pax5 and was actually slightly increased in Pax5^−/− pro-B cells. The factors that regulate Tcfe2a expression are unknown, but IL-7 has been predicted to promote expression in early B cells (1). Because both the pro-B cell cultures and a number of Pax5 target genes, including Nmyc, are exquisitely sensitive to IL-7 (6), small changes in the culture conditions may result in this apparent differential expression. Similarly, while our previous studies did not find Id3 to be directly regulated by Pax5 (6), it was subsequently reported that Id3 is induced by pre-BCR signaling, a process that is also Pax5-dependent (13). These findings highlight the fact that the regulation of specific targets may be context specific and confirm the importance of assessing differential gene expression using equivalent primary cell populations directly isolated from the BM.

A number of other transcription factors whose function in B cells is less well known were also identified. Of these, the calcium-inducible factor Nfatc1 is important for T cell function. Nfatc1^−/− lack B1a B cells (48), and their follicular B cells are hyporesponsive to BCR and CD40L-mediated signals (49). It is possible that NFATc1 fulfills a function in pre-BCR signaling, although this has not been investigated.

Several cofactors and chromatin regulators were also differentially expressed, including Cbfb (core-binding factor-β). Cbfβ is a transcriptional coactivator that interacts with Runx family members, such as Runx1, a factor that is essential for many aspects of hemopoiesis including B cell development (50). The gene coding for the catalytic subunit of the SWI/SNF-related complex, Smarca4 (also known as Brg1), which is essential for T cell development (51), was moderately differentially expressed in both arrays. Taken together, the number of nuclear proteins that are activated by Pax5 suggests that a key function of this factor is to activate the expression of a panel of genes that further reinforce the B cell transcriptional program.

The expression of lineage inappropriate genes such as Csf1r, Flt3, and Notch1 in Pax5^−/− pro-B cells (reviewed in Ref. 52) has provided support for the concept of lineage priming and led to a model whereby Pax5 functions to repress this promiscuous gene expression and thus restrict progenitors to the B cell fate. The Pax5-repressed genes identified here include a number of genes associated with broad hemopoietic expression such as Tmsb10 (thymosin β10), Cd47, and Lat2. Two genes involved in neutrophil migration were also identified as potential Pax5-repressed genes: l-plastin (murine Lcp1) (53) and Lsp1 (54). In addition, glutathione peroxidase-1 (Gpx1) expression was up-regulated in Pax5 deficient pro-B cells. This enzyme is primarily responsible for the intracellular degradation of hydrogen peroxide and is the predominant antioxidant enzyme expressed by osteoclasts and BM macrophages (55).

The gene encoding the Ig superfamily receptor Emb appears to be a particularly interesting Pax5-repressed gene, as it was expressed both in progenitor populations as well as in non-B cells. To further our study of Emb, we produced a mAb that detects Emb in flow cytometry, immunohistochemistry and Western blotting. These studies demonstrate that while virtually all B cells are Emb-low or negative, myeloid cells express an extremely high level of Emb whereas T cells are Emb-intermediate. Finally, Emb is induced during the terminal differentiation of conventional B cells and its expression is maintained in plasma cells, whereas peritoneal B1 B cells and splenic marginal zone B cells constitutively express Emb (32). Emb is thought to be involved in mediating adhesion to the extracellular matrix and is a member of a small gene family that also contains Basigin (CD147) a protein expressed on a variety of cell lineages including erythrocytes and T cells (56). Future loss-of-function studies are required to test the importance of the Pax5-mediated down-regulation of Emb for B cell differentiation.

Several Pax5-repressed genes have important functions in B cells. This includes Cd180 (also known as RP105), which encodes a receptor for LPS that triggers Ig class switching and secretion (57). The Pten-encoded phosphatase functions as a negative regulator of the PI3K signaling pathway upon engagement of the TCR or BCR (reviewed in Ref. 58). Although BM B lymphopoiesis occurred normally after the conditional deletion of Pten in B cells, the resultant lower signaling threshold through the BCR promoted marginal zone and B1 B cell development and rescued germinal center formation in signaling deficient, Cd19^−/− mice (59). As well as being essential for many aspects of peripheral B cell biology, the inhibition of Pax5 function and the down-regulation of its expression is a crucial event in initiating plasma cell differentiation (29, 32, 60). In line with this, it has been shown that many Pax5-repressed genes, including, Flt3, Mef2c, Igj, and Emb are re-expressed during late B cell differentiation (19, 32).

The microarray data set presented in this study complements two recent screens for Pax5 target genes performed by the Busslinger laboratory (19, 20). These studies also used cDNA microarrays to compare the transcriptomes of in vitro cultivated Pax5^−/− and wild-type pro-B cells. The initial study used subtractive hybridization to enrich for differentially expressed genes and found 110 putative Pax5-repressed genes (19), whereas that of Schebesta et al. used the mouse “lymphochip” (61) to identify 170 Pax5-activated transcripts (20). A variety of models of Pax5 deficiency were then used to extensively validate the target genes. In comparison, our study has used two different cDNA microarrays for the screening, a cDNA clone set from the National Institute of Aging that does not contain specifically selected hemopoietic cDNAs and a custom array produced from a mature IgM⁺ normalized cDNA library. In addition, other small variables would be expected to slightly alter the gene lists; for example the prior studies have used a 3-fold cutoff for differential expression whereas our studies have used statistical significance only.

Comparison of the data presented here on Pax5-activated genes revealed that 18 of the 66 genes listed in Table II (or 27%) were found in the previous report (20). Similarly, 12 of the 43 repressed genes (28%) were found in the prior study (19). The overlap between the lists can, however, be further enhanced by only comparing those genes in our study that were validated by an independent methodology (RT-PCR and/or flow cytometry), as only 50% of the genes in our study were further examined. Using this approach, the overlap between these studies was now 39%. Thus, it appears that the use of distinct screen techniques have, while identifying many common genes such as Notch1, Igf2, Emb, Spib, Ikzf3, Sdc4, and Ccnd3, proven complementary by detecting additional unique target genes, for example Syvn1, Frm4, and Tmsb10.

In summary, we have identified numerous novel Pax5 target genes whose expression is potentially activated or repressed by Pax5. These genes mediate diverse biological functions in B cells such as migration, signaling, and the regulation of gene expression and demonstrate the plethora of processes that are regulated by Pax5 to promote B cell lineage commitment and subsequent differentiation.

We thank Jason Brady, Kate Elder, Jaclyn Gilbert, and Matthew Ritchie for technical assistance.

The authors have no financial conflict of interest.