Redundant Role for Early Growth Response Transcriptional Regulators in Thymocyte Differentiation and Survival¹

T cell development in the thymus progresses through well-defined stages characterized by the expression of the CD4 and CD8 cell surface markers. The earliest T cell precursors express neither CD4 nor CD8 (double-negative (DN)³ thymocytes), which progress through a series of maturational steps to form double-positive (DP) thymocytes that express both CD4 and CD8, and then finally to mature single-positive (SP) thymocytes that express either CD4 or CD8. DN thymocytes can be further subdivided into several substages of maturation on the basis of CD25 and CD44 expression, namely DN1 (CD44⁺ and CD25⁻), DN2 (CD44⁺ and CD25⁺), DN3 (CD44⁻ and CD25⁺), and DN4 (CD44⁻ and CD25⁻) thymocytes (1). As part of this complex maturational process, rearrangement of the TCRβ gene occurs at the DN3 stage during which cells express the successfully rearranged TCRβ gene along with the pre-TCRα (pTα) chain to form the functional pre-TCR receptor complex. Signaling through the pre-TCR receptor involves a variety of cellular responses, including robust proliferation, enhanced cell survival, down-regulation of the Rag1 and Rag2 recombinase genes, reduced CD25 expression, decreased pTα expression, induction of the TCRα chain, and allelic exclusion of the nonrearranged TCRβ locus (2). Differentiation through DN3 to DN4 is known as β-selection because only cells with a functionally rearranged TCRβ chain survive this critical transition.

Thymocyte development depends upon normal pre-TCR signaling to engage a variety of molecules, including members of the early growth response (Egr), E protein, TCF/LEF, NF-κB, and NFAT transcriptional regulators (3, 4, 5, 6, 7). Although their precise function and the target genes regulated by them are largely unknown (8), it appears likely that at least some of the proliferative and antiapoptotic regulatory pathways engaged during thymocyte maturation converge on the promoters of the cyclin D3 and the Bcl-2-related A1 genes (9, 10, 11).

Egr proteins have a particularly important role as effectors of pre-TCR signaling, yet the precise stage of thymocyte development and mechanisms through which they act are still poorly understood. There are four Egr genes (Egr1–4), three of which (Egr1–3) are expressed during thymocyte differentiation. Egr proteins contain highly conserved zinc-finger DNA-binding domains that can bind a number of common target gene promoters and potentially cooperate in regulating their expression (12). In thymocytes, Egr1, Egr2, and Egr3 are induced by pre-TCR signaling and overexpression of these proteins in pre-TCR signaling-deficient thymocytes can partially facilitate progression through the β-selection checkpoint (13, 14). In addition, Egr protein overexpression down-regulates Rag1, Rag2, and pTα, and up-regulates TCR-Cα in vitro, mimicking some aspects of endogenous pre-TCR signaling. Conversely, dominant-negative Egr proteins inhibit progression from DN3 to DN4 in vitro (13). In vivo, Egr1-deficient thymocytes have a small but significant defect in positive selection but have no apparent defect in β-selection (15). In contrast, Egr3-deficient thymocytes have a partial block at the DN3 stage that is largely explained by a reduced proliferation of Egr3-deficient DN4 thymocytes (16). Transgenic overexpression of Egr3 in Rag1-deficient thymocytes (which are normally blocked at DN3) leads to increased thymocyte proliferation and partial rescue of the β-selection defect in these cells. The transient kinetics of Egr3 expression appear to be critical for proper development, as sustained overexpression of Egr3 inhibits the expression of RORγT through induction of the transcriptional inhibitor Id3 (17, 18). In turn, repression of RORγT leads to decreased Bcl-x_L expression and increased apoptosis in DP thymocytes.

Although Egr1 and Egr3 single knockout mice have helped to define their role in thymocyte development, other Egr-dependent processes may still be uncharacterized due to potential overlapping function of these proteins, because Egr1, Egr2, and Egr3 are all up-regulated at the DN3/DN4 transition and during positive selection, they may regulate a common set of target genes. In this study, we examined the impact on thymocyte development of simultaneously abrogating Egr1 and Egr3 function in developing thymocytes. Egr1 and Egr3 double knockout mice (referred to as 1/3 DKO) had consistently severe thymic atrophy not seen in mice lacking only Egr1 or Egr3. Although the relative proportions of DN, DP, and SP thymocytes were normal in most 1/3 DKO mice, there was a partial block at the DN3/DN4 transition and a reduction of thymocyte proliferation. Strikingly, there was massive apoptosis of developing thymocytes in 1/3 DKO mice not seen in mice lacking only Egr1 or Egr3. Chimeric transplant studies showed that thymocyte death was a consequence of a cell autonomous survival defect of 1/3 DKO thymocytes. Although the gene regulatory networks modulated by Egr1 and Egr3 during progression from the DN to DP stage are likely to be very complex, we identified a number of genes deregulated in DN thymocytes lacking both Egr1 and Egr3. These results may provide new insights into the molecular mechanisms through which Egr transcriptional regulators facilitate normal T cell development in the thymus.

Egr1-deficient and Egr3-deficient mice were generated and genotyped as previously described (19, 20, 21, 22). Egr1-deficient mice were backcrossed 10 generations and Egr3-deficient mice were backcrossed 4 generations to the C57BL/6J strain. Heterozygous matings were established and littermate wild-type (WT) mice were used as controls for all experiments in which comparisons were made between WT and Egr gene-deficient mice. B6.SJL-Ptprca Pepcb/BoyJ mice were purchased from The Jackson Laboratory. All experimental procedures complied with protocols approved by the Northwestern University Institutional Animal Care and Use Committee.

For immunohistochemical analysis, anesthetized mice (ketamine, 150 mg/kg; xylazine 10 mg/kg; i.p.) received intracardiac PBS (0.1 M (pH 7.4)), followed by aldehyde perfusion (4% paraformaldehyde, 0.1 M phosphate buffer (pH 7.4)). The thymuses and spleens were removed, cryoprotected in 30% sucrose-PBS overnight, frozen in OCT embedding medium (VWR), and stored at −80°C. Tissue sections were cut on a freezing microtome at 16-μm thickness. The sections were blocked (3% normal serum-0.1% Triton X-100 in PBS) for 1 h at room temperature, incubated with affinity purified anti-activated caspase 3 Ab (Cell Signaling Technology) diluted in blocking buffer overnight at room temperature, and subsequently incubated with a biotinylated anti-rabbit secondary Ab (Jackson ImmunoResearch Laboratories). An avidin-biotin histochemistry reaction using diaminobenzidine as the chromagen was performed according to the manufacturer’s specifications (Vector Laboratories). Quantification of caspase 3⁺ cells and measurement of tissue section volume was performed using unbiased stereology and optical fractionator procedures (StereoInvestigator; MicroBrightField). TUNEL histochemistry was performed as described (23) using biotin-labeled dUTP and visualized using avidin-biotin histochemistry with diaminobenzidine as the chromagen.

Thymus tissue was fixed in 2.5% glutaraldehyde/2% paraformaldehyde/0.1 M cacodylate buffer (pH 7.2), postfixed in 1% osmic acid in PBS, dehydrated through graded ethanol, and embedded in Epon. Sections were cut at 1 μm and stained with toluidine blue.

All Abs were from eBioscience. Thymus cell suspensions were stained for 20 min in PBS/2% FBS with the following fluorophore conjugated mAbs: anti-CD4-allophycocyanin or anti-CD8a-PE. For identification of DN thymocyte subsets, cells were first stained with biotin-labeled anti-CD4, anti-CD8, anti-CD11b, anti-NK1.1, and anti-γδTCR, followed by streptavidin-allophycocyanin, anti-CD25-PE, and anti-CD44-FITC. For DNA content analysis, cells were first stained for surface markers and then resuspended in PBS/0.1% saponin (Sigma-Aldrich)/10 μg/ml 7-aminoactinomycin D (7-AAD) at room temperature for 30 min before FACS analysis. Intact cells were determined from forward and side scatter profiles.

OP9-DL1 monolayers were seeded with bone marrow hemopoietic stem cells (HSCs; cKit⁺Sca-1^highLin⁻) isolated by flow cytometry from Egr-deficient mice and cultured as described, except that IL-7 levels were decreased to 0.2 ng/ml (24, 25). Lineage exclusion markers included CD19, Ter119, NK1.1, Gr1, CD11b, CD4, and CD8. Briefly, 10,000 HSCs were seeded on day 0, fed with cytokines on day 4 (5 ng/ml Flt3 ligand and 0.2 ng/ml IL-7), and then 300,000 cells were transferred to fresh subconfluent monolayers on day 7. Thereafter, 300,000 cells were transferred to fresh subconfluent monolayers every 4 days and analyzed by flow cytometry using the indicated Ab.

Primary bone marrow cells were isolated from the tibias and femora of P21 WT and 1/3 DKO mice. B6.SJL-Ptprca Pepcb/BoyJ mice were used as recipients because they express a different CD45 isoform (CD45.1) than WT and 1/3 DKO (CD45.2) donor cells. Recipient mice were exposed to two sequential doses of gamma irradiation (550 rad/dose) 3–4 h apart. Immediately after the second irradiation dose, mice were injected with freshly isolated bone marrow cells (1 × 10⁶ cells suspended in 200 μl of PBS) from WT and 1/3 DKO littermates by tail vein injection. The mice were administered antibiotic (polymyxin B sulfate and neomycin sulfate) containing water in a pathogen-free vivarium. To examine thymocytes derived from transplanted bone marrow cells (CD45.2⁺), recipient CD45.1⁺ cells were excluded from the flow cytometric analysis.

DN thymocytes from three WT and three 1/3 DKO mice were sorted on a MoFlo cell sorter directly into TRIzol (Invitrogen Life Technologies). Total RNA (50 ng for each sample) was amplified using the MessageAmp II aRNA kit (Ambion). Microarray analysis was performed as described (26) using GeneChip mouse expression arrays (430A and 430B; Affymetrix). Briefly, 5 μg of antisense RNA was converted to cDNA using the Superscript Reverse Transcriptase (Invitrogen Life Technologies) and the T7-Oligo(dT) Promoter Primer kit (Affymetrix). The cDNA was purified using the GeneChip sample cleanup module (Affymetrix) and was then used for the in vitro synthesis of biotin-labeled cRNA using the GeneChip expression 3′-amplification reagents for in vitro transcription labeling (Affymetrix) at 37°C for 16 h. The cRNA was fragmented into 35- to 200-bp fragments using a magnesium acetate buffer (Affymetrix). A total of 10 μg of labeled cRNA were hybridized to the microarrays for 16 h at 45°C. The GeneChips were washed and stained according to the manufacturer’s recommendations (Affymetrix) using the GeneChips fluidics station (model 450; Affymetrix). This procedure included washing the chips with PE-streptavidin, performing signal amplification by a second staining with biotinylated anti-streptavidin, and performing a third staining with PE-streptavidin. Each chip was scanned using the GeneChip scanner (model 3000; Affymetrix). Signal intensity and detection calls were generated using the GeneChip operating software (Affymetrix). The absolute intensity values of each chip were scaled to the same target intensity value of 150 to normalize the data for interarray comparisons. Six chip comparisons were generated using the samples from three WT and three 1/3 DKO DN thymocytes. Up-regulated and down-regulated genes with a >2-fold change in expression were identified using statistical measures performed on the nine interarray comparisons. The annotated MAGE-ML and image fields are available for all arrays at http://arrayconsortium.tgen.org.

RNA isolation, reverse transcription, and qPCR analysis were performed as described (27). Quantification was performed by generating standard curves using control cDNA samples and all samples were normalized relative to expression of GAPDH. Primer sequences are available upon request.

Egr1 is induced at multiple stages of thymocyte development, yet Egr1-deficient mice exhibit only subtle defects in positive selection and normal β-selection during thymocyte development (15). To test the hypothesis that multiple Egr proteins may perform redundant roles during thymocyte differentiation, mice with simultaneous loss of both Egr1 and Egr3 (1/3 DKO) were generated. Compared with WT littermates, thymuses from 1/3 DKO mice were markedly smaller in gross appearance (Fig. 1, A and B). Thymocyte cell counts were obtained from WT, Egr1^−/−, Egr3^−/−, and Egr1/3 DKO mice during the first 4 wk of life to more precisely define the timing and extent of the thymic atrophy. Thymocyte numbers were equivalent between all genotypes at birth, but as early as postnatal day 6 (P6) a statistically significant reduction in cell number was seen in 1/3 DKO thymuses compared with WT controls (p = 0.0002, Fig. 1,C). This difference became more pronounced 2 and 3 wk after birth. By P21, thymocyte number in 1/3 DKO mice were reduced to 10% that of WT littermates. Although 1/3 DKO mice are ∼50% smaller than normal littermates by P21, the differences in thymocyte cell number were significant even after correction for body weight (data not shown), indicating that the thymic atrophy was not simply the result of smaller body size. At P21, Egr1^−/− and Egr3^−/− mice also showed reduced cellularity. However, thymic atrophy was significantly more pronounced in 1/3 DKO mice relative to Egr1^−/− and Egr3^−/− (p = 0.008 and 0.004, respectively). In addition, relative to WT thymuses (Fig. 1,D), histological analysis of 1/3 DKO thymuses at P21 revealed effacement of normal thymic architecture and numerous pyknotic nuclei (Fig. 1 E, arrows).

TUNEL histochemistry was performed on WT, Egr1^−/−, Egr3^−/−, and 1/3 DKO thymuses to evaluate the extent of apoptosis. In 1/3 DKO thymuses at P21, there was massive apoptosis not seen in WT, Egr1^−/−, or Egr3^−/− littermates (Fig. 2, A–D). Quantification of TUNEL-positive cells demonstrated that 1/3 DKO mice, but not Egr1^−/− or Egr3^−/− mice, had significantly increased apoptosis at P21 (Fig. 2,E, p < 0.001). Because TUNEL histochemistry labels cells in late stages of apoptosis when many thymocytes have been phagocytosed by macrophages (28), it is likely that the degree of apoptosis in the 1/3 DKO was underestimated. To determine whether increased apoptosis was present in very young mice, WT and 1/3 DKO littermates were analyzed at P6. To maximize detection of early apoptotic cells, immunohistochemistry for activated caspase 3, a sensitive and early marker of apoptosis, was performed. Consistent with the increased TUNEL reactivity observed in P21 mice, the number of activated caspase 3⁺ thymocytes was significantly increased in 1/3 DKO thymuses at P6, indicating that the decreased thymic cellularity was at least in part a consequence of increased apoptosis (Fig. 2 F).

To assess the effect of Egr1 and Egr3 deficiency on thymocyte differentiation, CD4 and CD8 expression were examined in single and double knockout thymocytes. Representative FACS analysis plots from P21 thymocytes are shown in Fig. 3,A. The percentages of DN, DP, and CD4 SP thymocytes were statistically indistinguishable between WT and 1/3 DKO thymocytes (Fig. 3,B). However, there was a small but significant decrease in the percentage of CD8 SP thymocytes in 1/3 DKO mice (2.2% in 1/3 DKO vs 4.0% in WT, p = 0.002). Thymocytes from 1/3 DKO thymuses at earlier ages, or in Egr1^−/− or Egr3^−/− mice at any age, were statistically similar to WT (data not shown). Because Egr genes have been shown to be essential for progression from the DN3 to the DN4 stage, substages of DN thymocyte development were examined. Consistent with previous data implicating Egr proteins in the progression through the β-selection checkpoint, there was a partial block at the DN3 stage in 1/3 DKO thymocytes (Fig. 4, A and B). This block was noted as early as P6, and it became more prominent by P21 (Fig. 4, C and D). By contrast, significant alteration in the DN4/DN3 ratio was not seen in Egr1^−/− or Egr3^−/− thymocytes (Fig. 4,D). To simultaneously assess proliferation and apoptosis in substages of DN maturation, thymocytes were stained with the DNA-binding dye 7-AAD (Fig. 5,A). Consistent with the increased apoptosis detected in situ in 1/3 DKO thymuses, 1/3 DKO DN4 cells exhibited increased numbers of cells with sub-G₀ DNA content compared with WT littermates (Fig. 5,B). Proliferating DN3 and DN4 cells were also decreased in the 1/3 DKO thymus, and a similar decrease was seen with BrdU immunohistochemistry on WT and 1/3 DKO thymuses (Fig. 5 C and data not shown). However, the magnitude of this decrease was not significantly different from previously reported for Egr3^−/− mice (16). No reproducible differences in DN1 or DN2 proliferation or apoptosis could be detected in any genotype (data not shown).

Most 1/3 DKO thymuses exhibited cell loss, increased apoptosis, and relatively normal CD4 and CD8 expression. However, three mice had severe cell loss (total thymic cellularity <10 million cells) and a significant decrease in the percentage of DP thymocytes (data not shown). These mice were suspected to have an intrinsic defect in thymocyte survival and differentiation, but also an extrinsic thymocyte killing, possibly mediated by glucocorticoids in response to stress or malnourishment. Such mice were excluded from the analysis shown above. However, to confirm that the phenotype observed was due to a thymocyte-intrinsic defect and not a secondary effect such as stress, bone marrow HSCs from WT, Egr1^−/−, Egr3^−/−, and 1/3 DKO mice were cultured on OP9-DL1 cells which are capable of inducing Notch signaling and T cell differentiation in vitro (25). After 15 days of coculture with OP9-DL1 cells, a relatively small percentage of marrow-derived cells expressed both CD4 and CD8 (DP) in all cultures (Fig. 6,A). However, after 23 days in culture, Egr3^−/− and 1/3 DKO cells had generated significantly fewer DP cells compared with WT (Fig. 6), a significant effect which was also seen after 27 days of culture (Fig. 6,B). Consistent with a lack of defects in β selection in vivo, Egr1^−/− cells generated DP cells with similar efficiencies compared with WT (p > 0.05 at all time points). These results support the hypothesis that impaired differentiation of the Egr3^−/− and 1/3 DKO thymocytes is due to a cell-autonomous defect. Importantly, after 27 days of culture, 1/3 DKO cells showed significantly impaired differentiation compared with cells lacking only Egr1 or Egr3, indicating that defects seen in the 1/3 DKO mice cannot be attributed solely to loss of Egr3 (p < 0.05; Fig. 6 B).

To demonstrate whether there was a cell-autonomous defect in vivo, bone marrow cells from 1/3 DKO and WT mice were transplanted into normal recipients. Donor cells carry the CD45.2 allele and recipient cells carry the CD45.1 alleles, which are distinguishable by flow cytometry. Therefore, recipient cells were excluded from analysis by gating on CD45.2⁺ cells derived from donor bone marrow cells. Total thymic cellularity was decreased ∼2-fold in mice transplanted with 1/3 DKO bone marrow, although due to high animal-to-animal variability this difference did not reach statistical significance (Fig. 7,A, p = 0.06). In addition, there was no significant difference between the number of DN, DP, and SP thymocytes between WT and 1/3 DKO-transplanted mice (Fig. 7,B). However, analysis of DN thymocyte subpopulations in transplanted mice receiving 1/3 DKO bone marrow-derived cells again revealed a partial block at the DN3 population relative to mice transplanted with cells from WT mice (Fig. 7, C and D, p = 0.004). An increased number of apoptotic cells was noted in thymuses reconstituted with 1/3 DKO relative to WT-derived bone marrow cells (Fig. 7,E). Moreover, caspase 3 reactivity was observed in both individual thymocytes (Fig. 7,E, black arrowheads) and in macrophages (Fig. 7,E, white arrowheads) which phagocytosed the apoptotic cells. In addition, there was a quantitative increase in the number of caspase 3⁺ cells and macrophages in thymuses from mice transplanted with 1/3 DKO-relative WT bone marrow-derived cells (Fig. 7 F, p < 0.005 for both comparisons).

The increased apoptosis observed in 1/3 DKO thymocytes suggests that Egr proteins may cooperate to regulate thymocyte survival factors. Because Bcl-2 has been proposed to be a downstream target of Egr1 (15), Bcl-2 was measured in DN, DP, and SP thymocytes using intracellular flow cytometry. However, levels of Bcl-2 protein in 1/3 DKO thymocytes were equal to or greater than the levels found in WT littermates (data not shown). Likewise, the mRNA expression of other antiapoptotic genes—including A1, Mcl-1, and Bcl-x_L—was equivalent in WT and 1/3 DKO thymocytes (data not shown).

To identify novel target genes induced by Egr proteins as thymocytes progress from DN3 to DN4, genome-wide expression analysis was performed on WT and 1/3 DKO DN thymocytes using Affymetrix microarray analysis. The signal values for all genes analyzed are listed in supplementary table I.⁴ Six hundred thirty-nine genes were significantly down-regulated ≥2-fold in 1/3 DKO thymocytes relative to WT. The fold-changes in levels of expression, as well as confirmatory qPCR analysis for a select population of potential target genes, are shown in Table I. In addition, analysis of promoter regions of several of these genes revealed evolutionarily conserved Egr-binding motifs, suggesting that they may be subject to direct Egr-mediated transcriptional regulation (Fig. 8). Notably, the transcriptional regulators Myc and Runx1, both of which are involved in β-selection, were down-regulated in 1/3 DKO thymocytes, as was the proliferation regulator CDK4. The apoptosis that occurs in 1/3 DKO thymocytes skewed the DN3 and DN4 populations by 1.3- and 1.5-fold, respectively, relative to WT. To confirm that this had no significant effect on the gene expression results, we isolated DN3 and DN4 cells from WT and 1/3 DKO thymuses and examined the expression of a few deregulated genes by qPCR. The results indicated that gene expression changes were far more dramatic than could be explained by alterations in DN3/DN4 cell populations due to apoptosis (Table II). Moreover, expression of CD25 which is expressed at high levels in DN3 cells and low levels in DN4 cells demonstrates the purity of the cells tested and no significant alteration in expression between WT and 1/3 DKO mice. Interestingly, another 858 genes were up-regulated ≥2-fold in 1/3 DKO thymocytes, including the proapoptotic genes Bcl2L11 (Bim), Bik, and FasL.

To identify functional classes of genes which were misregulated in thymocytes that lack both Egr1 and Egr3, significantly deregulated genes were clustered using Gene Ontology (GO) annotations with the EXPANDER software suite (29) (www.cs.tau.ac.il/∼rshamir/expander). Surprisingly, several classes of genes involved in basic metabolic processes were significantly overrepresented among the genes down-regulated in 1/3 DKO DN thymocytes (Fig. 9,A). Over 100 genes annotated as involved in cellular metabolism, comprising 13.1% of all genes in this class, were down-regulated in thymocytes lacking Egr1 and Egr3. These results suggest that Egr proteins, either directly or indirectly, regulate genes involved in maintenance of cellular metabolic activity. Among the genes up-regulated in 1/3 DKO thymocytes, only genes annotated as possessing hydrolase activity were enriched (Fig. 9 A).

The EXPANDER program was also used to identify transcription factor-binding sites (TFBS) in potential Egr target genes which were down-regulated in 1/3 DKO thymocytes. Binding sites for two proteins, E2F and Myc, were found to be significantly enriched in the set of down-regulated genes (p < 0.001 for both factors). E2F sites were identified in the promoters of 140 genes (or 1.6% of E2F promoter-containing genes represented on the array) and Myc sites were identified in nearly 80 genes (or 1.5% of Myc promoter-containing gene represented on the array; Fig. 9 B).

Although Egr proteins have been proposed to be critical mediators of thymocyte differentiation, defining their essential role in vivo has been difficult due to some functional overlap between Egr1, Egr2, and Egr3. For example, in genetic backgrounds where thymocytes are unable to generate a pre-TCR signal, overexpression of Egr1 can partially drive differentiation past the DN4 stage in vitro and in vivo (13, 30). However, Egr1^−/− mice show no defect in β-selection, indicating that the function of Egr1 is compensated by other Egr family members at this stage (15). Egr3-deficient mice exhibit thymic atrophy due to impaired proliferation after β-selection. In this study, we characterized thymocyte differentiation in mice lacking both Egr1 and Egr3, and found that these mice exhibit severe thymocyte loss due to impaired proliferation and increased apoptosis. The apoptotic thymocyte death was at least partially cell autonomous and occurred at the DN4 stage where Egr genes are first expressed. Although thymocyte cultures on OP9-DL1 cells confirmed a cell autonomous role for Egr1 and Egr3 in thymocyte survival, the overall thymic cellularity was not decreased to the same extent in the bone marrow chimeras as it was in 1/3 DKO mice. These results may suggest that Egr-dependent environmental or non-thymocyte-intrinsic factors also contribute to thymocyte survival in vivo. One possibility is that thymocyte conditioning by thymic stromal interactions may also depend upon Egr-mediated transcription. The extent to which Egr-dependent gene regulation in thymic stromal cells may influence thymocyte development is currently unknown.

Mice lacking both Egr1 and Egr3 also exhibited fewer CD8⁺ SP cells, although this could not be replicated in normal recipients transplanted with 1/3 DKO bone marrow. It may be that this represents a non-cell-autonomous defect. Conversely, because there was only a small reduction in CD8⁺ SP cells in the 1/3 DKO mice compared with WT mice, it is possible that the bone marrow reconstitution experiments lacked sufficient power to detect this difference.

Although these results confirm a role for Egr proteins in the progression through the β-selection checkpoint, we note some discrepancies with previously reported literature. Notably, we were unable to identify a significant block at the DN3 stage in Egr3^−/− thymocytes, although consistent with published reports we find that the Egr3^−/− thymus is smaller than WT littermates (16). This difference may be explained by differences in genetic background (C56Bl6/129SV vs B6.AKR). However, taken together, the combined results suggest that Egr3 is required for normal thymocyte differentiation and simultaneous loss of Egr1 and Egr3 leads to more profound defects.

Cell death in 1/3 DKO thymocytes was not accompanied by the loss of known survival molecules. To understand which genes depend upon the expression of Egr1, Egr3, or both proteins, Affymetrix microarray analysis was performed to identify novel target genes. The transcription factors Runx1 and Myc, both known to play important roles during β-selection, were down-regulated in 1/3 DKO DN thymocytes. Impairment of Runx1 function in DP thymocytes leads to an up-regulation of Bim and increased sensitivity to apoptosis (31). As Runx1 expression in the thymus is highest in DN thymocytes (32), it is possible that reduction of Runx1 in DN thymocytes similarly leads to increased apoptosis. Consistent with this possibility, proapoptotic Bim was up-regulated in 1/3 DKO DN thymocytes (Table I).

To identify functional classes of genes deregulated in 1/3 DKO DN thymocytes, GO annotations were assigned to genes with significant changes in expression compared with WT. Surprisingly, several classes of genes involved in basic metabolism were significantly enriched in the set of genes down-regulated in 1/3 DKO cells. These results suggest that Egr-mediated pre-TCR signaling may influence thymocyte survival by maintaining metabolic processes in a manner analogous to that proposed for signals downstream of Notch1 and TCR stimulation (33, 34). For example, spermidine, and the related polyamine spermine, are known to be critical regulators of cell survival (35). By qPCR analysis, both Srm and Sms, which encode spermine synthase, were significantly down-regulated in 1/3 DKO thymocytes. Several studies have indicated that intracellular spermine is protective against apoptosis in thymocytes, raising the possibility that reduction in polyamines may sensitize 1/3 DKO thymocytes to apoptosis (36, 37, 38). Alternatively, it is possible that down-regulation of metabolic genes represents a nonspecific consequence of apoptosis initiated through other mechanisms. However, because the expression of over 1200 genes annotated in the “cellular metabolism” GO category did not vary between WT and 1/3 DKO (expression ratios 0.9–1.1) the results do not appear to represent a global down-regulation of metabolism that would likely be associated with apoptosis.

To further characterize transcriptional networks affected by the loss of both Egr1 and Egr3, TFBS were analyzed in the promoters of some deregulated genes. Egr-binding sites were identified in the promoters of 68 genes, including evolutionarily conserved sites in the promoters of Runx1, Myc, E2F1, and cdk4. However, after applying the Bonferroni correction for multiple tests, the enrichment of Egr-binding sites was not considered significant compared with the total set of TFBS in the TRANSFAC database (data not shown). This may indicate that Egr proteins bind to promoter sites which differ from the canonical consensus site used in the TRANSFAC database. In support of this possibility, similar promoter analysis on the set of genes induced by overexpression of Egr3 in myotubes also did not yield significant enrichment of Egr-binding sites (Ref. 26 and data not shown), yet several directly regulated target genes have been identified and well-characterized from the array data set (26, 27). Alternatively, this may indicate that a large number of down-regulated genes are indirectly regulated by Egr1 and/or Egr3.

Because a number of transcriptional activators were down-regulated in 1/3 DKO thymocytes, the promoters of down-regulated genes were scanned for significant enrichment of binding sites for these proteins. For example, Myc- and E2F-binding sites were significantly overrepresented in promoter regions of the down-regulated genes. This was noteworthy as both Myc and E2F1 were down-regulated in cells lacking Egr1 and Egr3. Given that Myc is known to transactivate a large number of genes involved in metabolism and ribosome synthesis (39), it is possible that the down-regulation of metabolism-related genes may reflect an indirect effect through a decrease in Myc expression. Furthermore, E2F4/p130 complexes repress a large number of metabolic genes during growth arrest, making it possible that transcriptionally active E2F proteins contribute to the expression of these genes during periods of growth such as seen after β-selection (40). This raises the possibility that reductions in E2F and Myc contribute to further deregulation of gene expression, leading to functional impairments of 1/3 DKO thymocytes. It is also possible that Myc and E2F1 modulate each other’s expression or that of Egr1 based on previous overexpression studies and/or chromatin immunoprecipitation experiments (41, 42, 43, 44). A potential regulatory network involving Egr, E2F, Myc, Runx, and Bim is shown in Fig. 9 C.

Some genes previously proposed to be targets of Egr proteins appeared normal in 1/3 DKO thymocytes. For example, Bcl-2 was shown to be modestly reduced in 3A9Tg, Egr1^−/−CD4^low/CD8^low thymocytes after stimulation with anti-CD3 (15). Here, we report that Bcl-2 expression was equivalent or slightly increased in 1/3 DKO DN thymocytes (see supplementary table I) and this was confirmed at the protein level in DN, DP, CD4⁺, and CD8⁺ cells by intracellular flow cytometry (data not shown). Although it is possible that Egr proteins affect the kinetics of Bcl-2 induction during positive selection, our data suggest that neither Egr1 nor Egr3 are essential for normal expression of Bcl-2 in vivo. Similarly, Id3 was initially proposed to be a potential Egr target by Bain and colleagues (45) who showed that Egr1 and Id3 were both induced by MAPK signaling with kinetics showing Id3 expression followed Egr1. Further studies demonstrated that overexpression of Egr3 can induce Id3 and lack of Egr3 results in a partial block at DN3, providing a model where Egr3 serves to regulate proliferation after β-selection (16, 18). However, in this study, the level of Id3 was not reduced in 1/3 DKO thymocytes compared with WT littermates. This raises two possible explanations. First, although overexpression of Egr3 may be sufficient to induce Id3, Egr function may not be required for normal Id3 expression in vivo. Alternatively, remaining Egr2 protein may be able to compensate for a loss of Egr1 and Egr3. Further studies will be required to distinguish these two possibilities.

In summary, we demonstrate that Egr1 and Egr3 serve redundant roles in the maintenance of cell survival during thymocyte differentiation in vivo. The mechanism by which these transcription factors contribute to cell survival remains to be precisely defined, but it does not appear to depend upon the up-regulation of known antiapoptotic molecules. One possibility is that failure to up-regulate essential metabolic enzymes and subsequent depletion of macromolecule precursors may increase sensitivity to apoptotic stimuli. Given the number of proapoptotic molecules which are up-regulated in 1/3 DKO thymocytes, it is also possible that Egr proteins serve to actively repress genes which normally would kill cells which do not receive a pre-TCR signal. Additional studies will be needed to further define the role of Egr proteins, including Egr2, in mediating some of the diverse phenotypic changes which follow successful passage through T cell developmental checkpoints.

We thank members of the National Institute for Neurological Diseases microarray consortium (TGen) for performing and analyzing the Affymetrix microarrays. We thank members of the Tourtellotte laboratory for advice on the manuscript, and Dr. L. Li for technical assistance with bone marrow transplants.

The authors have no financial conflict of interest.