Regulation of Lymphocyte Development by Cell-Type-Specific Interpretation of Notch Signals

Abstract

Notch signaling pathways exert diverse biological effects depending on the cellular context where Notch receptors are activated. How Notch signaling is integrated with environmental cues is a central issue. Here, we show that Notch activation accelerates ubiquitin-mediated and mitogen-activated protein kinase (MAPK)-dependent degradation of E2A transcription factors and Janus kinases, molecules essential for both B- and T-lymphocyte development. However, these events occur in B lymphocytes, but not T lymphocytes, due to their different levels of MAPK, thus providing one mechanism whereby Notch inhibits B-cell development without impairing T-cell differentiation. Lymphoid progenitors expressing a Notch-resistant E2A mutant differentiated into B-lineage cells on stromal cells expressing Notch ligands and in the thymus of transplant recipients. Bone marrow transplant assays and examination of steady-state B lymphopoiesis also revealed that the expression of Notch-resistant E2A and constitutively active STAT5 in mice neutralized the effects of Notch-induced degradation, allowing B-cell development through a bone marrow-like program in the thymus. These findings illustrate that Notch function can be influenced by MAPKs, producing distinct outcomes in different cellular contexts.

Signaling from the Notch family of receptors is an ancient and important pathway, governing the differentiation of a variety of cell lineages in eukaryotes ranging from Drosophila to humans (2). Notch signaling is usually involved in determining binary decisions in cell fate choices, and the outcome often depends on the cellular context in which Notch receptors are stimulated. The activation of Notch receptors is achieved through their binding to two types of ligands, Jagged and Delta-like (DL), expressed on adjacent cells. This triggers the cleavage of Notch and translocation of its intracellular domain into the nucleus. The intracellular domains then function as coactivators to stimulate the transcription of genes, such as HES1, Deltex1, and c-myc (19, 41, 54). However, it remains unclear how Notch signaling integrates cues from the cellular environment to execute lineage decisions and the nature of molecular events that follow. We have now explored these fundamental questions by focusing on the mechanisms by which Notch regulates lymphoid development.

Notch signaling pathways are known to influence several critical steps in lymphoid differentiation. The receptors play a role in hematopoietic stem cell maintenance in the bone marrow, but Notch signaling is believed to be turned off during B-lineage differentiation due to diminished expression of ligands, such as Delta-like-1 (DL-1) (37). Constitutive activation of Notch in hematopoietic progenitors blocks B-cell differentiation and drives T-lineage-cell production in the bone marrow (35). Conversely, in the thymus, where relevant ligands exist, the disruption of the Notch1 gene results in the loss of T cells and accumulation of B cells (55). The molecular mechanism by which Notch exerts its diverse effects on developing lymphocytes is likely mediated through the coordinated regulation of multiple transcription factors and signaling molecules, which play critical roles in B- and T-cell development.

One such family of transcription factors contains the basic helix-loop-helix E proteins encoded by the E2A, HEB, and E2-2 genes. The complete loss of E-protein function blocks B- and T-cell commitment and subsequent differentiation (3, 23, 47, 60). We have recently shown that Notch signaling enhances the degradation of E2A transcription factors by a ubiquitin-mediated and mitogen-activated protein kinase (MAPK)-dependent process (15, 30). While Notch-induced E2A degradation might explain its inhibitory effect on B-cell differentiation, E proteins must be available in T cells to allow development to proceed (5, 23). Therefore, Notch signaling must elicit different effects on different cell types. Indeed, we have previously shown and further demonstrate here that Notch signaling accelerates E2A degradation in B cells, but not T cells, due to the fact that basal levels of MAPKs are high in B cells but low in T cells (30). Therefore, the selective degradation of E proteins in B cells could be one important mechanism whereby Notch signaling inhibits B-cell development. However, it is important to emphasize that E-protein degradation is unlikely to be the only mechanism governing B- versus T-lineage choice.

Cytokines play central roles in hematopoiesis and lymphopoiesis. In adult mice, the early stages of lymphocyte development are primarily regulated by interleukin-7 (IL-7). Mice deficient in IL-7 or its receptor are severely deficient in B and T cells (33, 53). Signaling from the IL-7 receptor involves the activation of the Janus kinases (Jaks), which in turn phosphorylate and activate the transcription factor STAT5. The absence of Jak3 and STAT5 also results in severe impairment of B-cell production and T-cell development (49, 58). We now report that Notch signaling stimulates the ubiquitin-mediated degradation of Jak proteins. As in the case of E2A degradation, Notch accelerates the degradation of Jak2 and -3 in B cells, but not in T cells. Once again, the differential Notch-induced degradation of Jak2 and -3 is due to differences in MAPK activities in B and T cells. Thus, the regulation of Jak levels represents another mechanism by which Notch mediates its selective inhibitory effect on B-cell development. Consistent with our findings, it has been shown that the expression of a constitutively active form of STAT5, which can bypass the need for activating cytokines and Jaks, increases the numbers of B cells in the thymus (11).

To further test our hypothesis about the role of Notch-induced degradation in B versus T cells, we generated knock-in mice with mutant E2A proteins lacking the obligatory MAPK target sites. As predicted by the results of our previous studies, these mutants exhibited resistance to Notch-induced degradation. Lymphoid progenitors from these mice are superior in B-cell differentiation on stromal cells expressing Notch ligands and in the thymus when transplanted. The mutant E2A also synergized with constitutively active STAT5 to promote B-cell differentiation in the thymus. Therefore, accelerating the degradation of both E2A transcription factors and Jak molecules in B cells, but not in T cells, may be one mechanism whereby Notch signaling effectively inhibits B-cell development without interfering with T-cell differentiation.

MATERIALS AND METHODS

Generation of the targeting construct and mice.

The genomic DNA clones pE2A.2x and pE2A51R1 contain DNA fragments with overlapping sequences of the 5′ and 3′ portions of the E2A gene (kindly provided by Yuan Zhuang of Duke University). Point mutations into exon 13 were introduced by using two-step PCR. The final PCR product was cloned into pGEM-T Easy vector (Promega, Madison, WI) and sequenced. The insert was then isolated by digestion with Asp718 and XhoI and cloned into Asp718- and EcoRI-digested pBSK vector together with an XhoI-EcoRI fragment containing a floxed neomycin resistance gene isolated from the ploxP vector (57). From this construct, an Asp718-EcoRI fragment was isolated and ligated with a 5.2-kb EcoRI-HindIII fragment containing the sequence downstream from the intron between exons 13 and 14 into the Asp718 and HindIII sites of a pBSK vector containing the DT-A gene cloned in its HindIII and SmaI sites. Subsequently, an Asp718-NotI fragment from this construct was cloned into the same sites of a ploxP vector already carrying a 2.0-kb EcoRI-Asp718 fragment including the E2A sequence upstream of exon 10. This final construct was linearized by NotI digestion and used for electroporation of 129X1/SvJ ES cells.

Five independent 129X1/SvJ ES cell lines harboring an appropriately targeted allele were identified by Southern blotting and PCR. The presence of point mutations was confirmed by sequencing PCR products amplified from the targeted allele. Two of the five independent 129X1/SvJ ES cell lines harboring an appropriately targeted allele were injected into blastocysts of C57BL/6J mice. Chimeras were screened for germ line transmission of the targeted allele. Positive progenies were crossed with EIIa-Cre transgenic mice on the C57BL/6 background to remove the neomycin resistance gene in germ cells (25). Mice carrying the knock-in allele and the EIIa-Cre transgene were identified by analyzing the tail DNA of black mice using PCR and Southern blotting and backcrossed with C57BL/6 mice. Mice which had the pGK-neo cassette deleted from the E2A^M/+ knock-in allele and lack the EIIa-Cre transgene were selected for subsequent backcrosses with C57BL/6 for at least 5 and up to 10 generations.

Preparation of lymphocytes and progenitors.

Total thymocytes were isolated from wild-type and E2A^M/M mice. To enrich for CD19⁺ cells, bone marrow cells were stained with purified antibodies against immunoglobulin M (IgM), Mac-1, Gr-1, and Ter119, followed by depletion with anti-rat IgG (Fc)-conjugated immunomagnetic beads (Dynalbeads, Invitrogen, Carlsbad, CA). All isolated cell populations were cultured in RPMI 1640 medium (phenol red-free) with l-glutamine supplemented with 10% fetal bovine serum and 50 μM β-mercaptoethanol for 4 h prior to treatment with or without 10 ng/ml SCF, 10 ng/ml Flt3-L, and 5 ng/ml IL-7 (R&D Systems, Inc., Minneapolis, MN). To isolate common lymphoid progenitors (CLPs), bone marrow cells were double-immunomagnetically depleted with purified antibodies against CD2, CD3, CD5, CD8, CD19, Mac-1, Gr-1, Ter-119, and B220 (BD-Pharmingen, San Diego, CA). Lineage-negative (Lin⁻) cells were then stained with anti-c-kit, Sca-1, and IL-7Rα, and c-kit^Low Scal-1⁺ IL-7R⁺ cells were sorted by using a FACS Aria (BD Biosciences, San Jose, CA). To isolate Flt3⁺ LSK cells, Lin⁻ marrow was sorted for Sca1⁺ c-kit^Hi Flt3⁺ cells.

Stromal cell coculture.

The OP42 stromal cell line (46) was transduced with retroviruses expressing the empty retrovector or human DL-1. Primary bone marrow cells were seeded at 8 × 10⁶ cells per well on stromal cells with or without DL-1. The cocultures were then centrifuged at 180 × g for 10 min before incubation at 37°C. To harvest cells in contact with stromal cells, the plates were gently shaken and unattached cells in the medium were removed and gently washed with precooled phosphate-buffered saline (PBS). One milliliter of 5 mM EDTA in PBS was then added to the wells, and attached lymphoid cells were immediately washed off by vigorously tapping the plates. The harvested cells were lysed in radioimmunoprecipitation buffer (1% NP-40, 0.5% sodium deoxycholate, and 0.1% sodium dodecyl sulfate in PBS) supplemented with protease and phosphatase inhibitors (200 μM phenylmethylsulfonyl fluoride, 2 μg/ml leupeptin, 2 μg/ml aprotinin, 2 μg/ml pepstatin A, 20 mM NaF, and 1 mM Na₃VO₄).

To examine B-cell differentiation from CLPs, the cells were cocultured on OP9-vector and OP9-DL-1 stromal cells as previously described (14). Briefly, 1,500 CLPs per well were seeded on stromal cells at 80% confluence in the alpha-minimal medium supplemented with 15% fetal calf serum and various concentrations of cytokines. The medium contained 5 ng/ml SCF, 5 ng/ml Flt3-L, and 1 ng/ml IL-7 during the first 2 days and 2 ng/ml Flt3-L and 0.5 ng/ml IL-7 for the following 5 days. Cells were then harvested by using 5 mM EDTA-PBS as described above for fluorescence-activated cell sorter (FACS) analyses. The number of B cells produced correlated inversely with the density of DL-1 on stromal cells.

Lymphoid progenitor transplant.

Flt3⁺ LSK cells were transferred intravenously into sublethally irradiated (6.5 Gy) C57BL/6.SJL hosts as indicated in Fig. 6A, with each host receiving 3,000 sorted progenitors. At the indicated times posttransplant, individual thymic lobes were collected and separately dissociated into single-cell suspensions in Hanks' balanced salt solution supplemented with 5% fetal calf serum, 10 mM HEPES buffer, and 10 mM NaN₃. Absolute cell numbers were determined for each sample by using a hemocytometer, and aliquots were immunofluorescently labeled for CD45.2 (donor marker), CD4, CD8, B220, and/or IgM. Flow cytometry analysis was conducted with either a FACSCalibur or an LSRII flow cytometer (BD).

FIG. 6.

Competitive bone marrow transplant assays for thymic B-cell reconstitution from E2A^M/M and/or STAT5^tg progenitors. (A) Experimental design. (B) Numbers of thymic B220⁺ cells derived from Flt3⁺ LSK cells of indicated genotypes. Parentheses contain average numbers from five to eight thymic lobes per donor genotype. Differences between wild-type (WT) donors and those of other genotypes are all significant based on Student's t test (P < 0.05). (C) Numbers of T cells recovered from the same transplant experiments. Error bars show standard errors of the means.

Immunoprecipitation and immunoblotting.

For immunoblotting, the cells were lysed in radioimmunoprecipation assay buffer containing proteasome and phosphatase inhibitors and 50 μg of the lysates was used for immunoblot analyses. Antibodies were purchased from the following vendors: phospho-STAT5, STAT5a, phospho-p38, p38, and ubiquitin from Zymed Laboratories (San Francisco, CA); phospho-Jak2, Jak2, Jak3, and phospho-Erk from Upstate Biotechnology (Lake Placid, NY); and E2A, E47, TFIIH, and Erk1 from Santa Cruz Biotechnology (Santa Cruz, CA).

RESULTS

Notch signaling enhances ubiquitin-mediated and proteasome-dependent degradation of Jak2 and -3 in B cells, but not T cells.

Since cytokines are essential for lymphocyte development, we explored the possible influence of Notch signaling on STAT activation following cytokine binding to its receptors. Using an IL-3-dependent cell line, BaF3, we initially found that STAT5 activation as measured by its phosphorylation was markedly diminished in cells transduced with a constitutively active form of Notch1, the intracellular domain of Notch1 (N1-IC) (Fig. 1A). However, the total amounts of STAT5 present in N1-IC-transduced cells were not significantly reduced, suggesting an inhibitory effect on STAT5 activation.

FIG. 1.

Notch-induced ubiquitination and degradation of Jak2 and Jak3. (A) BaF3 cells were infected with retroviruses expressing either N1-IC or vector. Infected cells were sorted based on enhanced GFP expression from the constructs. The sorted cells were starved of IL-3 for 4 h prior to restimulation for indicated lengths of time. Immunoblotting was carried out with indicated antibodies. (B) Immunoprecipitation (IP) was performed on whole-cell lysates from vector- or N1-IC-transduced BaF3 cells using anti-Jak2 or control antibodies (Ctr). The precipitates were analyzed by probing with antiubiquitin or anti-Jak2 antibodies. α, anti; IB, immunoblotting. (C) BaF3 cells were cocultured on vector- or DL-1-transduced stromal cells in the presence of MG132 (12.5 μM) or vehicle for 2 h. The level of Jak2 was determined by using immunoblotting. (D) OP42-vector or DL-1 stromal cells were cocultured with indicated populations of cells plus or minus cytokines (10 ng/ml SCF, 10 ng/ml Flt3-L, and 5 ng/ml IL-7) for 1.5 h. Lymphoid cells were then collected and probed for indicated proteins. WT, wild type. (E) Lin⁻ bone marrow cells were cocultured in the presence of cytokines with or without 25 μM γ-secretase inhibitor (GSI) XII (CalBiochem, San Diego, CA) for 2 h. Whole-cell lysates were used for immunoblotting of indicated proteins. Loading controls in all panels were conducted with anti-Erk1 or anti-TFII-H antibodies. +, present; −, absent; BM, bone marrow cells.

We then looked upstream at Jak2 and Jak3, which are responsible for STAT5 phosphorylation. BaF3 cells express undetectable levels of Jak3 but relatively high levels of Jak2, which becomes phosphorylated and activated upon IL-3 stimulation. Jak2 degradation is known to depend on phosphorylation at the Y1007 and Y1008 residues (50). We observed a dramatic decrease in phospho-Jak2 levels detected with antibodies from animals immunized with a peptide containing the two phospho-tyrosine residues (Fig. 1A). The expression of N1-IC also reduced the level of total Jak2 protein in IL-3-stimulated cells, further suggesting that Notch signaling likely inhibits STAT5 activation by regulating the turnover of Jak2 (Fig. 1A). Curiously, IL-3 appeared to enhance the production of Jak2 in vector-transduced cells but the expression of N1-IC itself upregulated Jak2 in the absence of IL-3 (Fig. 1A). Even so, a dramatic decrease in Jak2 occurred after IL-3 stimulation of N1-IC-transduced cells.

To determine if Notch signaling promotes Jak2 degradation through a ubiquitination-dependent and proteasome-mediated mechanism, we next measured the ubiquitination status of Jak2. Whole-cell lysates prepared from BaF3 cells transduced with vector or N1-IC retroviruses were immunoprecipitated with anti-Jak2 or control antibodies. The precipitates were analyzed by immunoblotting with antiubiquitin or anti-Jak2 antibodies. As shown in Fig. 1B, ubiquitination is stimulated by IL-3, which triggers the phosphorylation of Jak2. More importantly, N1-IC dramatically increased the amounts of ubiquitinated Jak2 in the presence of IL-3. Consistent with this finding, Notch-induced Jak2 degradation was blocked by treatment with a proteasome inhibitor, MG132 (Fig. 1C). These results suggest that Jak2, like E2A, is degraded through a ubiquitin-mediated process which is enhanced by Notch signaling.

To test if Notch signaling influences Jak turnover in developing lymphocytes, we cocultured various lymphocyte populations with vector- or DL-1-transduced stromal cells to stimulate endogenous Notch receptors on the surface of the lymphocytes (Fig. 1D). CD19⁺ B-lineage cells were collected by depletion with antibodies against Mac1, Gr1, Ter119, and IgM from wild-type bone marrow. The same depletion procedure used on RAG1^−/− bone marrow yielded mostly Lin⁻ hematopoietic progenitors, with some B220⁺ pro-B cells. In the presence of cytokines, which cause Jak phosphorylation, the Jak2 and Jak3 levels in wild-type CD19⁺ cells were dramatically reduced upon exposure to DL-1-transduced, but not vector-transduced, stromal cells (Fig. 1D). As a result, the levels of phospho-STAT5 were markedly reduced. We next examined Jak degradation in thymocytes isolated from RAG1^−/− mice, which are mostly CD4⁻ CD8⁻ cells capable of responding to cytokine stimulation, as indicated by STAT5 activation (Fig. 1D). Jak2 and Jak3 levels remained unchanged with or without DL-1 and cytokines. We also measured Jak2 and Jak3 levels in Lin⁻ bone marrow cells and found that they were significantly decreased when the cells were cocultured with stromal cells expressing DL-1 (Fig. 1E). However, incubation with a γ-secretase inhibitor prevented the DL-1-induced degradation of Jak2 and Jak3, suggesting that Notch activation is involved in this process. Together, these results indicate that Notch signaling induces Jak2 and Jak3 degradation in CD19⁺-enriched bone marrow and hematopoietic progenitor cells, but not in T-lineage cells, which could contribute to the inhibitory effect of Notch on B-cell development, but not T-cell development.

Notch-induced Jak degradation depends on MAPKs.

Further analyses showed that both p38 and Erk were active in CD19⁺-enriched marrow, but not in thymocyte preparations (Fig. 2A), which correlated with the differential effect of Notch signaling on Jak degradation in B and T cells. To test if the activity of MAPKs influences Jak2 degradation, we first treated vector- or N1-IC-transduced BaF3 cells with inhibitors of MEK1 and p38, PD98059 and SB203580, respectively, which led to reductions in the levels of phospho-Erk and phospho-p38 (probably by inhibiting its autophosphorylation) (Fig. 2B). N1-IC expression by itself decreased Jak2 levels by 70%. However, adding either MAPK inhibitor returned the Jak2 levels to normal, suggesting that both Erk and p38 are involved in Notch-induced Jak2 degradation. To further assess the involvement of MAPK, we cotransfected dominant-negative forms of MEK1 (MEK1-DN) or p38 (p38-DN) (8, 38) with Jak2 in the presence of N1-IC into HeLa cells, which have high levels of MAPK. A green fluorescent protein (GFP) expression construct was included to serve as a control for transfection efficiency. The expression of N1-IC resulted in the reduction of Jak2 to background levels (Fig. 2C, lane 2 versus lanes 1 and 5). However, the inclusion of either MEK1-DN or p38-DN rescued Jak2 proteins from N1-IC-induced degradation (lane 2 versus lane 3 or 4), confirming the requirement for both Erk and p38 in Notch-induced Jak2 degradation. Furthermore, the attenuation of MAPK activation by MEK1 or p38 inhibitors also prevented Jak2 degradation in primary CD19⁺ bone marrow cells cocultured with DL-1-transduced stromal cells (Fig. 2D), as well as in Lin⁻ bone marrow cells (data not shown).

FIG. 2.

Notch-induced degradation of Jak2 is MAPK dependent. (A) MAPK levels in primary lymphocytes treated with or without cytokines for 1 h were determined by immunoblotting with the indicated antibodies. +, present; −, absent. (B) Vector- or N1-IC-transduced BaF3 cells were treated with MAPK inhibitor PD98059 (PD) or SB203580 (SB) at a final concentration of 10 μM for 1.5 h. No-inhibitor control (−) was performed by treating the cells with the vehicle, dimethyl sulfoxide. Whole-cell lysates were used for immunoblotting with indicated antibodies. (C) HeLa cells were transiently transfected with indicated amounts of expression plasmids. Whole-cell lysates were analyzed by immunoblotting 36 h later. (D) Wild-type CD19⁺ IgM⁻-enriched bone marrow cells were cocultured on vector- or DL-1-transduced stromal cells with cytokines plus or minus inhibitors as described for panel B. Jak2 levels were determined by immunoblotting. Bar graphs show normalized levels of Jak2 relative to loading controls. (E) 16610D9 cells were transduced with vectors- or N1-IC-expressing retroviruses. Transduced cells were sorted based on GFP expression and cultured overnight before treatment with 10 nM phorbol myristate acetate (PMA) and 50 ng/ml ionomycin (Ion) for 1 h. Whole-cell lysates were used for immunoblotting against indicated proteins. (F) BaF3 cells were treated with PD98059 (PD) or SB203580 (SB) at a final concentration of 10 μM for 1 h and analyzed by immunoblotting. WT, wild type; BM, bone marrow; pErk1/2, phospho-Erk1/2; pp38, phospho-p38.

Conversely, the stimulation of a T-cell line, 16610D9, with phorbol myristate acetate and ionomycin dramatically activated both Erk and p38, as indicated by their phospho forms (Fig. 2E, lane 1 versus 3). Under this condition, transduction with N1-IC, but not vector, retroviruses led to a significant reduction of Jak3 in the cells cultured in IL-7 (Fig. 2E, lane 3 versus 4), which causes phosphorylation of Jak3 on the tyrosine residues required for its degradation. The overexpression of N1-IC in 16610D9 cells by itself modestly increased the levels of active MAPK and slightly reduced Jak3 levels (Fig. 2E, lane 1 versus 2). These results thus suggest that Jak proteins can be degraded upon Notch signaling if MAPK is available.

To explore the mechanisms underlying the MAPK dependence of Jak degradation, we examined the effects of MEK1 and p38 inhibitors on the expression of SOCS1, a crucial component of the E3 ubiquitin ligase complex for Jak proteins. The p38 inhibitor SB203580 diminished SOCS1 expression, whereas the inhibition of Erk activation by PD98059 did not have such an effect, which may imply that p38 and Erk serve nonoverlapping functions in Jak degradation (Fig. 2F).

Taken together, these data indicate that, like E2A degradation, Notch-induced Jak degradation is regulated by MAPK activities and suggest that differences in MAPK between B and T cells contribute to the differential effect of Notch signaling.

Mutations of MAPK sites in E2A render it resistant to Notch-induced degradation in vivo.

We next began to test the biological effects of Notch-induced degradation of E2A and Jak proteins in animals. To eliminate the effect of Notch on E2A degradation, we created knock-in mice expressing E2A proteins with mutations in the three MAPK recognition sites important for Notch-induced E2A degradation (30). To generate the targeting construct (Fig. 3A), a cluster of point mutations was introduced into exon 13 of the E2A gene. Mice derived from an appropriate ES cell line were backcrossed onto the C57BL/6J background for up to 10 generations. Homozygous knock-in mice, referred to as E2A^M/M mice, for MAPK sites mutant, were produced at an expected Mendelian frequency and exhibited no gross developmental abnormalities.

FIG. 3.

Notch-resistant E2A degradation and B-cell differentiation in E2A^M/M mice. (A) Schematic diagram of the strategy used to generate E2A^M/M mice. The portion of the E2A locus involved in the knock-in construct is represented by exons shown in gray boxes along with indicated exon numbers. The exon where mutations were introduced is depicted as a dark box, with wild-type sequence shown below. The three MAPK target sites are in bold, and residues changed to alanines are italicized. Deletion of the floxed pGK-neo cassette by EIIa-Cre leaves behind a loxP site (dark triangle). (B) CD19⁺ IgM⁻-enriched bone marrow cells from wild-type (WT) and E2A^M/M mice were cocultured on vector- (V) and DL-1-transduced stromal cells and analyzed by immunoblotting. (C) RAG1^−/− mice were intraperitoneally injected with 10 μg of the anti-CD3 (α-CD3) antibody per mouse and sacrificed at indicated time points to isolate thymocytes and assay for E2A levels. +, present; d, days. (D) Indicated thymocyte populations from WT and E2A^M/M mice were analyzed for E2A levels along with loading controls. Normalized E2A levels are shown in the bar graph. DP, double-positive thymocytes; SP, single-positive thymocytes. (E) Lin⁻ Sca1⁺ c-kit^Low IL-7Rα⁺ CLPs were collected from age-matched wild-type or E2A^M/M mice and cocultured on vector- or DL-1-transduced stromal cells with SCF, Flt3 ligand, and IL-7. FACS plots show cells produced from 1,500 CLPs after 7 days which are representative of five independent experiments. The percentage of cells in each quadrant is shown. The bar graph shows average numbers of CD19⁺ cells. Error bars indicate standard errors of the means. (F) CD4 and CD8 DN thymocytes from two-month-old WT and E2A^M/M mice were analyzed by staining with antibodies against CD19 and IgM. The bar graph shows numbers of B cells by population ± standard errors of the means. The number of mice analyzed is shown in parentheses. P values from Student's t test for the relevant pairs are shown above each set.

To test if E2A mutant proteins from these mice were resistant to Notch-induced degradation, we prepared CD19⁺-enriched populations from wild-type or E2A^M/M mouse bone marrow as described above and cocultured them with control or DL-1-expressing stromal cells. While the E2A level in wild-type cells was dramatically reduced upon exposure to the Notch ligand, the level remained the same in E2A^M/M cells, indicating that the E2A mutant is resistant to Notch-induced degradation in bone marrow-derived CD19⁺ B-lineage cells (Fig. 3B).

Notch-induced E2A degradation does not generally occur in T-lineage cells, due to low basal levels of MAPK (30). However, MAPKs are transiently activated following pre-T-cell receptor (pre-TCR) signaling or TCR signaling, which would allow E2A degradation in the thymus. To test this hypothesis, we injected RAG1-deficient mice with anti-CD3 antibodies to mimic pre-TCR signaling and measured E2A levels at different time points. Within the first 24 h, E2A levels gradually decreased, but they returned to normal 7 days later (Fig. 3C). This finding prompted us to examine E2A levels in CD4⁻ CD8⁻ CD44⁻ thymocytes (double-negative 3 [DN3] and DN4 subsets), which develop immediately after pre-TCR signaling. The level of wild-type E2A was obviously lower than that of mutant E2A (Fig. 3D), suggesting that E2A is degraded in these cells, probably due to the activation of both Notch and MAPK. In contrast, differences in protein levels between wild-type and E2A^M/M CD4 and CD8 double- or single-positive thymocytes were less significant (Fig. 3D). These results demonstrate the dependence of E2A degradation on MAPK. However, E2A^M/M mice do not exhibit significant phenotypic alterations in T-cell developmental profiles (data not shown), probably due to redundant mechanisms controlling overall E protein activity at each developmental checkpoint.

B-cell differentiation of E2A^M/M lymphoid progenitors in the presence of Notch signals.

To evaluate the ability of the mutant E2A to facilitate B-cell development in the presence of Notch signals, we examined B-cell differentiation on stromal cells expressing DL-1 from CLPs, which give rise to CD19⁺ B-lineage cells in vitro (24). CLPs isolated from wild-type and E2A^M/M mice were cocultured on control or DL-1-expressing OP9 stromal cells in medium containing SCF, Flt3-L, and IL-7 for 7 days. As expected, both wild-type and E2A^M/M CLPs differentiated into B-lineage cells efficiently on control stromal cells (Fig. 3E). In contrast, B-cell development from wild-type CLPs was completely inhibited when they were cultured on DL-1-transduced stromal cells, which is consistent with previous findings (14, 43). However, E2A^M/M CLPs continued to differentiate into B-lineage cells in the presence of Notch ligands, albeit at a reduced efficiency compared to their differentiation on vector-transduced stromal cells.

We next examined B-cell production in the thymuses of E2A^M/M mice. E2A^M/M mice exhibited a largely normal CD4 and CD8 profile and total cellularity (data not shown). The numbers of thymic B cells in young mice did not change significantly. However, by 2 months of age, the numbers of CD19⁺ IgM⁻ or CD19⁺ IgM⁺ B cells in E2A^M/M thymuses increased by twofold over the numbers in wild-type thymuses (Fig. 3F). Nevertheless, the steady-state numbers of B cells accumulated in E2A^M/M mice was small compared to the numbers in the thymuses of mice whose Notch1 gene is inducibly deleted (55), which suggested that additional mechanisms exist by which Notch signaling controls the lineage choice, proliferation, and/or survival of B cells.

Steady-state B-cell development in the thymus under the influence of constitutively active STAT5 and Notch-resistant E2A.

Since Notch signaling induces the degradation of not only E2A transcription factors but also Jaks, we tested if the neutralizing effects of Notch on both types of molecules could influence B-cell development in the thymus. Since Notch-resistant Jak mutants have not been identified, we made use of established STAT5 transgenic mice (STAT5^tg) expressing a constitutively active mutant of STAT5 in both B- and T-lineage cells (11). The dephosphorylation of this mutant has previously been shown to be attenuated, thus alleviating the demand for Jak-mediated phosphorylation (31). Indeed, we found that, in the presence of cytokines, the levels of phospho-STAT5 were much higher in CD19⁺ B-lineage cells derived from STAT5^tg bone marrow cells than in wild-type cells, even though the overall levels of STAT5 are not dramatically different (Fig. 4A). When these cells were cocultured with stromal cells transduced with DL-1-expressing retroviruses, the levels of wild-type phospho-STAT5 decreased by about 40% but those of the mutant phospho-STAT5 diminished by less than 10% (Fig. 4A). The residual levels of phospho-STAT5 were still higher than those found in wild-type cells without Notch activation, suggesting that sufficient levels of active STAT5 are available to support B lymphopoiesis. Therefore, STAT5^tg mice were crossed with E2A^M/M mice to obtain mice carrying either STAT5^tg or E2A^M/M alleles or both. First, we examined E2A degradation in B220⁺ IgM⁻ cells isolated from the thymuses of STAT5^tg and STAT5^tg E2A^M/M mice. The E2A level in STAT5^tg B cells was very low compared to that in STAT5^tg E2A^M/M B cells, supporting the hypothesis that E2A degradation is accelerated in B-lineage cells within the thymus (Fig. 4B). A small difference in E2A levels was also detected in STAT5^tg and STAT5^tg E2A^M/M mouse CD4⁻ CD8⁻ B220⁻ T cells, which is analogous to the finding that E2A is degraded following pre-TCR signaling (Fig. 3E). In contrast, the E2A levels were similar in B cells isolated from the bone marrow of mice regardless of their genotypes. Together, these results reveal a selective downregulation of E2A in thymic B cells, which might negatively influence their development.

FIG. 4.

Steady-state B lymphopoiesis in E2A^M/M and STAT5^tg mice. (A) Levels of phospho-STAT5 (pSTAT5) in STAT5^tg mice. CD19⁺ IgM⁻ cells from wild-type (WT) and STAT5^tg bone marrow were cultured with OP42 stromal cells transduced with vector (V)- or DL-1-expressing retroviruses for 2 h in the absence (−) or presence (+) of cytokines. Whole-cell lysates were used for immunoblotting with antibodies against phospho-STAT5 and STAT5a as indicated. (B) Selective E2A degradation in thymic B cells. E2A levels in the indicated cell populations from WT, E2A^M/M (M/M), STAT5^tg (ST5), and STAT5^tg E2A^M/M (ST5/M) mice were measured by immunoblotting. (C) Representative FACS plots showing B220 and IgM expression in CD4⁻ CD8⁻ thymocytes from mice of indicated genotypes. Numbers shown in parentheses and within each quadrant are total numbers and percentages of CD4⁻ CD8⁻ thymocytes, respectively. The bar graph shows total numbers of immature and mature B cells by genotype averaged from indicated numbers (n) of individual animals. Error bars indicate standard errors of the means. All mice were between 4 and 6 weeks old at the time of analysis. (D) Representative FACS plots showing BP-1 expression on CD4⁻ CD8⁻ B220⁺ thymocytes from STAT5^tg and STAT5^tg E2A^M/M mice. Percentages of the immature BP-1⁺ cell fraction are indicated above bars.

To examine steady-state B-cell development in the thymus, we stained DN thymocytes for B220 and IgM markers and determined the cellularity of different B-cell subsets. One-month-old wild-type and E2A^M/M mice had comparable numbers of thymic B cells, most of which are B220⁺ IgM⁺ mature B cells (Fig. 4C). As shown by Goetz et al., the numbers of B cells produced in STAT5^tg thymuses were dramatically increased (11). However, the majority of them were B220⁺ IgM⁻ B-lineage cells (Fig. 4C) and expressed CD19 (data not shown). When the E2A mutant was expressed in STAT5^tg mice, differentiation into B220⁺ IgM⁺ B cells was facilitated (Fig. 4C). The maturation of pro-B cells can also be measured by the downregulation of BP-1 as pro-B cells progress to immature stages (13). STAT5^tg B cells are largely BP-1 positive, while Notch-resistant E2A triggered the downregulation of BP-1 in 60% of STAT5^tg E2A^M/M B cells (Fig. 4D). These results suggest that the constitutive activation of STAT5 facilitates the survival or production of pro-/pre-B cells while Notch-resistant E2A promotes B-cell maturation in the thymus.

Constitutively active STAT5 and Notch-resistant E2A promote a bone marrow-like B-cell differentiation program in the thymus.

A small number of B cells are normally present in adult thymuses. However, they are distinct from bone marrow-derived B cells with respect to CD5 expression. Most B cells in the bone marrow or spleen are CD5^low or CD5⁻, but a significant portion of thymic B cells are CD5^hi and exhibit different proliferative responses to stimulators such as lipopolysaccharide and anti-IgM antibodies (6, 28). Thymic B cells are also distinct from B1 cells found in the peritoneal cavity (9). To characterize the thymic B cells of STAT5^tg or STAT5^tg E2A^M/M mice, we examined DN thymocytes for the expression of B220 and CD5. CD5^hi cells constituted only 5.5% and 3.65% of CD4⁻ CD8⁻ B220⁺ cells in these mice, similar to the percentage of CD5^hi cells in B220⁺ cells of wild-type bone marrow (Fig. 5A). In contrast, 46% and 39% of CD4⁻ CD8⁻ B220⁺ cells from wild-type and E2A^M/M thymuses, respectively, were CD5^hi. These results suggest that STAT5 activation enables a thymic B-cell differentiation program similar to that normally occurring in the bone marrow.

FIG. 5.

STAT5 promotes a bone marrow-like B-cell developmental program in the thymus. (A) FACS histograms showing CD5 expression on total CD4⁻ CD8⁻ B220⁺ thymocytes from mice of indicated genotypes. B220⁺ cells from wild-type (WT) bone marrow are shown for reference. Numbers above bars show percent of total B220⁺ cells within the CD5⁺ gate. Numbers in parentheses show total B220⁺ CD5⁺ cells in each representative thymus. (B) CD4⁻ CD8⁻ CD43⁺ thymocytes were plotted for B220 and BP-1 expression by FACS to assess accumulation of “fraction-C” B-cell progenitors, shown within the rectangular gates. Wild-type CD43⁺ bone marrow cells were analyzed for reference. Percentages of these cells are indicated, with the total numbers per thymus shown below in parentheses. (C) Thymocytes were collected from 4-week-old mice 18 h post-intraperitoneal injection with 100 μg of BrdU and analyzed by flow cytometry for BrdU incorporation. Representative plots show BrdU⁺ cells as the percentage of the B220^low population.

“Fraction C,” defined as B220⁺ CD43⁺ BP-1⁺, refers to a bone marrow B-cell developmental stage when pre-BCR receptors cooperate with cytokine signals to drive pre-B-cell expansion and differentiation (13). Therefore, we examined fraction-C cells in the thymuses of mice expressing STAT5 and/or Notch-resistant E2A by measuring the expression of B220 and BP-1 on CD4⁻ CD8⁻ CD43⁺ cells. While very few, if any, fraction-C cells were present in wild-type and E2A^M/M thymuses, STAT5^tg or STAT5^tg E2A^M/M mice have distinct populations of these cells (Fig. 5B). It is thus possible that the expansion of fraction-C cells is responsible for the dramatic increase in the numbers of thymic B cells.

To examine the proliferation status, we injected mice with BrdU and measured BrdU incorporation 18 h later by staining with anti-BrdU antibodies, as well as antibodies against B220, CD4, and CD8 (Fig. 5C). Since it is known that immature B cells proliferate as they differentiate, we focused on the percentages of B220^low cells that incorporated BrdU within the 18-h period. Wild-type and E2A^M/M mice had 24% and 18% BrdU⁺ cells in the B220^low population. However, the expression of STAT5 led to an increase in the percentage of BrdU-labeled cells to about 40% of B220^low cells in STAT5^tg or STAT5^tg E2A^M/M mice. This level of BrdU incorporation is similar to that seen in B220^low cells of wild-type bone marrow. These results suggested that larger fractions of B cells are capable of cycling in mice expressing the STAT5 transgene. This is consistent with the increase in the numbers of fraction-C-like cells, even though the rate of BrdU incorporation by these cells at any given time was not altered (11). Thus, the overall increase in the numbers of cycling B cells could contribute to the expansion of B cells in STAT5^tg and STAT5^tg E2A^M/M mouse thymuses.

Bone marrow transplant assays reveal cooperative effects of constitutively active STAT5 and Notch-resistant E2A in B-cell production.

To further evaluate the ability of constitutively active STAT5 and Notch-resistant E2A to promote B lymphopoiesis in the thymus, we followed the kinetics of B-cell production by conducting bone marrow transplant assays with Flt3⁺ Lin⁻ Sca1⁺ c-kit^Hi (Flt3⁺ LSK cell) progenitors that are thought to be biased toward B- and T-lymphoid lineages (1). These progenitors were isolated from wild-type, E2A^M/M, STAT5^tg, and E2A^M/M STAT5^tg mice and transplanted into sublethally irradiated congenic recipients, which allow the examination of lymphocyte repopulation in competitive situations (32) (Fig. 6A). Donor-derived thymocytes were analyzed for CD4 and CD8 expression, and DN thymocytes were gated for examination of the B220 marker (Fig. 6B). On day 19 after transplant, the numbers of B220⁺ cells generated from E2A^M/M and STAT5^tg donors were 3- and 4.5-fold higher, respectively, than those from wild-type donors. The difference between B-cell production by E2A^M/M and STAT5^tg donors was statistically insignificant. In contrast, E2A^M/M STAT5^tg progenitors produced 18-fold-more B cells than wild-type progenitors. This result suggests that constitutively active STAT5 and Notch-resistant E2A cooperate to promote B-cell development in the thymus. Interestingly, while the advantage of E2A^M/M over wild-type progenitors remained the same 34 days posttransplant, B cells derived from STAT5^tg donors expanded considerably, accumulating 18-fold more B cells. Despite this expansion, the combination of constitutively active STAT5 and Notch-resistant E2A led to a further increase in the numbers of thymic B cells (51-fold), reaffirming the synergistic effect of the two proteins.

In contrast to the effects on thymic B-cell production, T-cell repopulation benefited modestly from either constitutively active STAT5 or Notch-resistant E2A. By day 19, E2A^M/M donors had a threefold advantage over wild-type donors, but this diminished by day 34. This effect of mutant E2A is supported by the notion that E2A facilitates Notch signaling and T-cell commitment (16). STAT5^tg donors performed similarly to wild-type cells on day 19 but produced threefold-more T cells on day 34. Likewise, E2A^M/M STAT5^tg progenitors generated three- to fourfold more T cells when assayed at either time point. These results suggest that the Notch-induced degradation of E2A and Jak proteins specifically suppresses B-cell production.

DISCUSSION

The principal conclusion from this study is that the Notch and MAPK signaling pathways act in concert to inhibit B-cell differentiation while allowing T-cell development. Differences in MAPK activities in B and T lymphocytes account for their distinct responsiveness to Notch signaling, thus initiating and stabilizing their developmental fates. At least two groups of molecules, E2A transcription factors and Jaks, are selectively degraded in Notch-stimulated B-cell precursors. Other pathways are likely to be involved, but we can now begin to appreciate how a single stimulus elicits a binary fate decision (Fig. 7).

FIG. 7.

Cell-type-specific interpretation of Notch signals. Lymphoid progenitors originating from hematopoietic stem cells (HSC) acquire high and low levels of MAPK either stochastically or instructively. Those with high levels of MAPK normally differentiate into B cells, but the differentiation program is aborted upon Notch signaling due to MAPK-dependent degradation of E2A and Jak proteins. Those with low levels of MAPK proceed to T-cell differentiation, which is facilitated by E2A and Jak-STAT proteins available in these cells. Notch signaling also promotes T-cell development through additional mechanisms.

Mechanisms by which Notch signaling influences B-cell development.

Notch is known for its role in promoting T-cell development while inhibiting B-cell differentiation (26, 36, 52). Our findings that Notch signaling accelerates the ubiquitin-mediated degradation of E2A and Jak proteins suggest new mechanisms for limiting B-cell development and expansion. E2A transcription factors and Jak-STAT signaling pathways likely play different roles in B-cell development. E2A proteins are necessary for B-cell commitment, since E2A-deficient mice are devoid of early B-cell progenitors (4). Together with early B-cell factor, E2A stimulates Pax5 expression, and all three transcription factors initiate B-cell differentiation by regulating B-cell-specific gene expression and Ig gene rearrangement (12, 40, 44). This is consistent with our finding that E2A^M/M progenitors had a threefold advantage in generating thymic B cells when transplanted (Fig. 6B).

Jak-STAT pathways transmit signals from cytokine receptors. Signaling from c-kit, Flt3, and IL-7 receptors, which leads to the activation of Jaks and STAT5, is particularly vital for B-cell development. IL-7 is crucial for the differentiation of CLPs into pro-B cells and for pro-/pre-B-cell expansion (10, 29). Signaling from c-kit and Flt3 also contributes to lymphopoiesis prior to IL-7-dependent B- and T-cell development. Flt3-L-mediated signaling is necessary for the generation or maintenance of CLPs (45). Thus, the Notch-induced degradation of Jak2 and -3 likely prevents B-cell development in the thymus by inhibiting the proliferation or survival of progenitors or pro-/pre-B cells.

In STAT5^tg mice, the constitutively active protein could compensate for the diminution of Jak proteins in the thymus and stimulate the proliferation of progenitor cells and their differentiation into pre-B cells. Indeed, the B-cell-producing “DN1c” population defined by Petrie and colleagues is increased in STAT5^tg mice (11, 34). Furthermore, fraction-C-like B cells also accumulated in STAT5^tg mice (Fig. 5B). These observations could explain the significant increase in numbers of B-lineage cells seen in the thymuses of STAT5^tg mice. Goetz et al. also suggested that STAT5 could directly bind to the promoter of the Pax5 gene to activate its expression and initiate B-cell differentiation (11). However, Notch-resistant E2A also facilitated the differentiation of B cells, promoting the maturation of B220⁺ IgM⁻ to B220⁺ IgM⁺ B cells in STAT5^tg E2A^M/M mice (Fig. 4C). Under competitive conditions where bone marrow progenitors were transplanted, constitutively active STAT5 and stable E2A clearly cooperated to promote thymic B-cell production (Fig. 6B). It thus appears that both Jak-STAT signaling and E2A transcription factors are necessary to ensure robust thymic B-cell development and that the Notch-induced degradation of both E2A and Jaks contributes, at least in part, to the inhibition of B-cell development (Fig. 7).

The B-cell developmental program promoted by constitutively active STAT5 and Notch-resistant E2A in the thymus resembles that occurring in the bone marrow. Namely, thymic B cells in STAT5^tg and STAT5^tg E2A^M/M mice express low levels of CD5 and have large fraction-C populations. These B cells may originate from circulating B-cell progenitors, which normally cannot survive and develop in the thymus. Alternatively, robust B-cell development from normal thymus-seeding progenitors may be made possible by the availability of Jak-STAT signaling and E2A transcription factors, which are normally abundant in the bone marrow. These interesting issues await further investigation.

The Notch-controlled B- versus T-lineage decision probably involves additional mechanisms. For example, Notch signaling has been shown to be one of the earliest signals driving T-cell commitment from thymic seeding progenitors (42, 48). Therefore, Notch signaling likely pushes a majority of the progenitors to follow the T-lineage path in the thymus. In Notch-deficient thymuses, when the option to become T cells is eliminated, the stabilization of E2A and Jak proteins, along with other molecules, would allow B-cell differentiation and expansion. In contrast to Notch-deficient mice, genetic manipulations that counteract the Notch-accelerated degradation of E2A or Jak proteins would enable B-cell development to proceed, but it is important to note that the option of T-cell commitment and differentiation remains open. Therefore, the phenotypes of Notch-deficient mice are not expected to be completely recapitulated by alteration of the E2A and Jak-STAT pathways.

Integration of Notch signaling with cues from cellular environment.

How are basal levels of MAPKs regulated in a cell-type-specific manner? This may occur stochastically or instructively by specific control mechanisms. Since multiple signaling pathways converge to regulate MAPKs (7), it would be difficult to envision a rigid mechanism to specifically control them during hematopoiesis, thus making the stochastic model more attractive. Hematopoietic stem cells and multipotent progenitors differentiate under the influence of multiple signals originating from growth factors and cytokine receptors. For example, signaling from c-kit, Flt3, and IL-7R leads to the activation of tyrosine kinases, which can initiate a cascade of events activating MAPKs, including Erk, p38, and Jun N-terminal protein kinase (22). However, the steady-state levels of MAPKs in a given cell ultimately depend on the interplay between activating events and inactivating measures mediated by phosphatases. Perhaps stochastic events result in different levels of MAPKs in individual progenitor cells, setting the stage for other factors that dictate lineage choice (Fig. 7). For instance, when basal levels of MAPKs reach a certain threshold, B-cell differentiation is favored. However, when the activities are below this level, T-cell development proceeds. As shown in this report and others (39), thymocytes indeed have very low basal levels of MAPKs (Fig. 2A), whereas pro-/pre-B cells have much higher basal activities. Consistent with this, the inhibition of the Ras-MAPK pathway by the expression of a DN form of p21^N-ras dramatically impaired B-cell development but had a minimal effect on early T-cell development within the CD4⁻ CD8⁻ stage (17). However, it is difficult to directly measure the basal levels of MAPKs in discrete populations of progenitors due to the paucity of cells and short half-life of phospho-MAPKs. Assuming that the basal levels of MAPKs are high and low for B- and T-cell-biased progenitors, respectively, the role of Notch signaling in the B- versus T-lineage choice could be to selectively promote the differentiation of progenitors with lower MAPK activities into T cells and to eliminate the progenitors with higher MAPK activities by blocking their differentiation into B cells (Fig. 7).

The integration between Notch signaling and cellular environmental cues is not restricted to lymphocyte development. Notch and MAPK signaling pathways are highly conserved throughout evolution. In C. elegans, vulval development is controlled by the interplay of inductive signals provided through the epidermal growth factor receptor-MAPK pathway and lateral signals from the Notch receptor (Lin-12), which eliminates the primary cell fate and specifies a secondary cell fate (59). Similarly, Drosophila eye development also involves signaling mediated by both the epidermal growth factor receptor-MAPK and Notch pathways (51). Thus, the underlying molecular mechanisms in these disparate developmental systems might be conserved.

Biochemical mechanism by which Notch signaling induces protein ubiquitination.

Elucidating the biochemical mechanism by which Notch signaling induces the degradation of E2A and Jak proteins will have a broad impact on the understanding of downstream events. We have shown here and elsewhere that Notch signaling enhances the ubiquitination of E2A and Jak proteins (30). E2A ubiquitination is mediated by the E3 ubiquitin ligase SCF^Skp2, which includes Skp1, Skp2, Rbx1, and Cul1 (20, 30). Jak2 ubiquitination is known to involve an E3 ligase complex consisting of SOCS1, elongin B/C, Rbx1, and cullin 2 or 5 (21, 56). Given the fact that Jak2 and Jak3 share extensive sequence homology, a similar mechanism could be utilized for Jak3 ubiquitination. Our previous data suggested that the transcriptional activity of Notch1 is involved in stimulating E2A ubiquitination (30). It is conceivable that Notch signaling activates genes collectively regulating the ubiquitination function of a class of E3 ligases, such as the cullin-based E3 ligases. If this hypothesis is correct, one would predict that the degradation of additional substrates of cullin-based E3 ligases is also regulated by Notch signaling.

Furthermore, although the phosphorylation of E2A proteins themselves by Erk1 and -2 is necessary for their degradation (30), the phosphorylation of Jak2 and Jak3 by MAPKs per se may not be essential for their degradation. We have created mutations lacking the S523 phosphorylation site of Jak2 that is targeted by Erk (18, 27), as well as two other potential MAPK sites located in the C terminus. None of these mutations could render the protein resistant to Notch-induced degradation (data not shown). Instead, we have shown that p38, but not Erk, is important for the expression of SOCS1, essential for Jak ubiquitination. The fact that p38 and Erk act distinctly may explain how the inhibition of either p38 or Erk activation significantly prevents Jak degradation (Fig. 2). If p38 and Erk controlled the same step in Jak degradation, we would have not observed a dramatic effect by ablating either MAPK pathway. The regulation of protein degradation by MAPK activities may represent an important control mechanism in eukaryotic differentiation. Understanding how MAPKs influence Jak degradation will shed light on the turnover of other substrates regulated by Notch signaling.