Phosphorylation of Ser 2078 Modulates the Notch2 Function in 32D Cell Differentiation*

Notch signaling is involved in the regulation of many cell fate determination events in both embryonic development and adult tissue homeostasis. We previously demonstrated that Notch1 and Notch2 molecules inhibit myeloid differentiation in a cytokine-specific manner and that the Notch cytokine response domain is necessary for this functional specificity. We have now investigated the putative role of phosphorylation in the activity of Notch in response to cytokine signals. Our results show that the granulocyte colony-stimulating factor (G-CSF) stimulation of 32D cells expressing the intracellular Notch2 protein induces phosphorylation at specific sites of this molecule, rendering the molecule inactive and permitting differentiation of these cells. In contrast, when cells are stimulated with granulocyte macrophage colony-stimulating factor (GM-CSF), intracellular notch2 is not phosphorylated at these residues and differentiation is inhibited. We also show that deletion of the Ser/Thr-rich region between amino acids 2067 and 2099 abrogates G-CSF-induced phosphorylation and results in a molecule that inhibits differentiation in response to either G-CSF or GM-CSF. Our results further indicate that Ser in 32D Myeloid Progenitor Cells— We have previously shown that Notch1 and Notch2 are capable of selectively inhibiting 32D myeloid differ-Notch2

Notch signaling is involved in the regulation of many cell fate determination events in both embryonic development and adult tissue homeostasis. We previously demonstrated that Notch1 and Notch2 molecules inhibit myeloid differentiation in a cytokine-specific manner and that the Notch cytokine response domain is necessary for this functional specificity. We have now investigated the putative role of phosphorylation in the activity of Notch in response to cytokine signals. Our results show that the granulocyte colony-stimulating factor (G-CSF) stimulation of 32D cells expressing the intracellular Notch2 protein induces phosphorylation at specific sites of this molecule, rendering the molecule inactive and permitting differentiation of these cells. In contrast, when cells are stimulated with granulocyte macrophage colony-stimulating factor (GM-CSF), intracellular notch2 is not phosphorylated at these residues and differentiation is inhibited. We also show that deletion of the Ser/Thr-rich region between amino acids 2067 and 2099 abrogates G-CSF-induced phosphorylation and results in a molecule that inhibits differentiation in response to either G-CSF or GM-CSF. Our results further indicate that Ser 2078 is a critical residue for phosphorylation and modulation of Notch2 activity in the context of G-CSF-induced differentiation of 32D cells.
Notch molecules are highly conserved transmembrane proteins that, through their participation in cell-cell interactions, play critical roles in cell fate decisions during many developmental processes (reviewed in Ref. 1). Four distinct Notch homologs have been identified in mammals (2)(3)(4)(5) and the extent to which they have specific functions is a theme of current investigation. The Notch1 and 2 molecules are synthesized as single precursor proteins that are cleaved in the Golgi to become functional heterodimeric receptors present on the cell surface. Notch ligands are expressed on the surface of neighboring cells and also on the same cell as the receptor, with the relative dosage of receptor and ligand being an important determinant of Notch activity (6). The productive interaction of Notch receptor with ligand results in further proteolytic cleavage, with release and nuclear translocation of the intracellular domain of Notch (Notch-IC), 1 which functions as a transcriptional regulator (reviewed in Ref. 7). Truncated Notch molecules corresponding to Notch-IC (i.e. lacking the transmembrane and extracellular domains) behave as constitutively active molecules, and thus have been used extensively to study Notch function in many vertebrate and invertebrate systems (8 -11). Notch activity generally, but not exclusively, leads to the inhibition of a differentiating signal, thereby preserving progenitors that are capable of responding to subsequent signals and generating cell diversity.
Despite the extraordinary conservation of Notch structural domains, several observations support the lack of complete functional redundancy. Inactivation of different Notch genes in mice has been very instructive in this regard; for instance, Notch1 and Notch2 null mice are embryonic lethals before day 11.5 (12,13), and they show distinct phenotypic defects indicating that they cannot compensate for the inactivated molecule (14 -16). Although the biochemical mechanisms remain elusive (17,18), these studies provide evidence for the specificity of Notch homolog function during mammalian development.
Notch1, Notch2, and Notch3, as well as multiple DSL ligands, are expressed in hematopoietic cells (19 -21). Notch signaling has been shown to influence the development of lymphoid (22)(23)(24), myeloid (11,25,26,27), and erythroid cells (28,29) and to participate in the maintenance of the hematopoietic progenitor pool (30 -33). Despite the emerging evidence implicating Notch as a key mediator in hematopoiesis, the precise role of the different Notch homologs in regulating specific hematopoietic cell fate decisions remains unclear. Within the lymphoid system, Notch1 and Notch2 are preferentially expressed in thymus and spleen, respectively (3). However, T and B lymphocytes express both Notch1 and 2, and in other hematopoietic cell types Notch1, 2, and 3 are frequently coexpressed, 2 suggesting that more than one Notch molecule may function simultaneously in a given cell.
Recently it has been shown that Notch molecules are posttranslationally modified by glycosylation (34) and phosphorylation (35), adding further complexity to the regulation of Notch signaling. Glycosylation of Notch EGF repeats by fringe molecules is important in modulating Notch-ligand interactions (36,37) and it is likely to contribute to ligand binding specificity of the different Notch homologs (38).
Phosphorylation is a widely used mechanism for regulating activity through signal transduction pathways. Recent reports have shown that a nuclear form of N1IC 1 (Notch1-IC) is phosphorylated (39) and that ligand binding to Notch2 induces hyperphosphorylation and nuclear translocation of this molecule (40). Furthermore, both phosphorylation and nuclear translocation appear to be necessary for oncogenic transformation mediated by Notch (41). Together, these findings suggest that phosphorylation may be an important factor in modulating intracellular signal transduction through the Notch pathway. Previously, we have reported that either expression of truncated Notch1 (N1IC ⌬OP ) or ligand activation of the full-length Notch1 receptor inhibits G-CSF-induced differentiation of 32D myeloid progenitors (11,42). However, N1IC ⌬OP does not inhibit differentiation induced by GM-CSF; and conversely, truncated Notch2 (N2IC ⌬OP ) does not affect G-CSF-induced differentiation, but inhibits differentiation in response to GM-CSF (42). These latter studies also revealed that a specific region of the Notch molecule, designated the NCR (Notch cytokine response) region, confers this cytokine specificity. In the present study, we have investigated whether phosphorylation plays a role in regulating activity of the N2IC ⌬OP molecule. We find that N2IC ⌬OP is differentially phosphorylated in 32D cells exposed to G-CSF or GM-CSF. In the context of G-CSF stimulation, phosphorylation of N2IC ⌬OP correlates with its inability to inhibit differentiation. At least one of the N2IC ⌬OP residues phosphorylated in G-CSF-stimulated cells mapped to a serine/ threonine-rich (STR) region within the NCR domain. Further characterization of this STR region included analyses of deletion mutants and N2IC ⌬OP molecules containing point mutations; these studies indicate that phosphorylation of Ser 2078 is a significant factor in the regulation of Notch2 function in this system.
Cell Cultures-32D cells (43) were maintained in Iscove's modified Dulbecco's medium with 10% fetal calf serum and 10% WEHI Conditioned Medium (WCM) as a source of interleukin-3 (IL3). Cells were maintained undifferentiated, mycoplasma free and were checked regularly for the capacity to differentiate in G-CSF. To obtain-Notch expressing clones, 4 ϫ 10 6 32D cells were electroporated at 295 V and 1050 F in a Gene Pulse II apparatus (Bio-Rad) with 10 -20 g of Notch construct and 1-5 g of pCDNA.3 vector to allow G418 selection. Individual clones were isolated by limiting dilution and evaluated for Notch construct expression by Western blot using the 9E10 (␣-Myc tag) anti-body. Point mutant transcripts were corroborated by sequencing RT-PCR products from some clones.
Differentiation Cultures-Differentiation cultures were performed as previously described (25) with minor modifications. Cells were plated at 1-1.5 ϫ 10 5 cells/ml in 6-well plates (3-4 ml) in 1-0.5% WCM plus 10 ng/ml rhG-CSF (Amgen) or 0.4 ng/ml rmIL3 (Strathmann Biotech GmbH) plus 5 ng/ml rmGM-CSF (Strathmann Biotech GmbH). Slides were prepared in a cytospin (Shandon) and Wright-stained (Quick Panoptic). Cells were evaluated for granulocytic differentiation using the conventional criteria of nuclear segmentation, increased cytoplasmic/nuclear ratio, and increased eosinophilia. All cell counts were performed in a blind fashion by the same investigator. For protein stability experiments, cells were cultured in the G-CSF differentiation medium with 10 g/ml of cycloheximide (Sigma).
In Vivo Phosphorylation-5 ϫ 10 6 32D cells were incubated in 0.5 ml of phosphate-free Dulbecco's modified Eagle's medium (Life Technologies, Inc.) and 10% dialyzed fetal calf serum for 1 h at 37°C and 5% CO 2 . Then, 0.2 mCi [ 32 P]orthophosphate (Amersham Pharmacia Biotech) was added for the next 3 h. Cells were exposed during the last 5, 15, 30, 60 min to different stimuli (400 ng/ml of G-CSF or GM-CSF or no cytokine). Cells were extensively washed and lysed in 1% Nonidet P-40, 10 mM Tris-HCl pH 7,5, 140 mM NaCl, 5 mM EDTA, 50 mM sodium fluoride, 0.4 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride and 10 g/ml leupeptin and aprotinin. N2IC ⌬OP molecules were immunoprecipitated with 9E10 antibody (anti-Myc tag) and electrophoresed through an 8% SDS-PAGE gel, transferred to a polyvinylidene difluoride membrane and autoradiographed and immunostained with the 9E10 antibody. Densitometric values were obtained with the Phoretix software by measuring the intensity of the labeled protein compared with the total immunoprecipitated protein.
In Vivo Phosphopeptide Mapping-Phosphorylation of N2IC ⌬OP for tryptic maps was performed as described above with minor modifications. Briefly, 1.5 ϫ 10 7 32D cells were incubated in phosphate-free medium for 1 h, followed by the addition of 1.8 mCi of [ 32 P]orthophosphate and incubation for another 3 h. Cells were exposed to cytokine stimuli for the last 15 min. Phosphopeptide mapping was performed as previously described (44). Briefly, radiolabeled N2IC ⌬OP molecules were excised from dried 8% PAGE and swollen in 400 l of H 2 O/methanol/ acetic acid (8:3:1) for 30 min, washed in 50% methanol and H 2 0 and dried. Pellets were incubated with 400 g of trypsin (Worthington) in 0.05 mM NH 4 HCO 3 overnight. Peptides were oxidized in 50 l of ice-cold performic acid for 60 min, diluted in 400 l of water and dried. Peptides were dissolved in electrophoresis buffer (pH 4.72) and run in a Hunter thin-layer system on a TLC plate (Merck). Chromatography was performed overnight in n-butyl alcohol/pyridine/acetic/H 2 O (75:50:15:60) buffer.
Transactivation Assay-NIH-3T3 cells were transiently transfected by calcium phosphate precipitation. Notch constructs in pCS2ϩMT or the vector alone were cotransfected with the Hes-1-luc plasmid (kindly provided by A. Israel). ␤-galactosidase expression was used for normalization of activity. After 48 h, lysates were prepared using the cell lysis buffer (Promega) following manufacturer's specifications. Results are reported as the ratio between individually normalized luciferase activities and the control empty vector. A Student's t test was used to validate the differences.

RESULTS
Notch-IC Molecules Are Phosphorylated in 32D Myeloid Progenitor Cells-We have previously shown that Notch1 and Notch2 are capable of selectively inhibiting 32D myeloid differ-Notch2 Phosphorylation in 32D Differentiation entiation in response to G-CSF or GM-CSF. 32D cells expressing the truncated N2IC ⌬OP molecule do not differentiate in the presence of GM-CSF but differentiate normally when stimulated with G-CSF. In those studies, we demonstrated that exchange of the NCR region (amino acids 2067-2155 of Notch2) between the N1IC ⌬OP and N2IC ⌬OP molecules resulted in a corresponding exchange in their effects. Moreover, exchange of the nuclear localization signal portion of the NCR domain (amino acids 2100 -2155) left their original activities intact, suggesting that the intermediate region (amino acids 2067-2099) was responsible for the cytokine-specific effects of N1IC ⌬OP and N2IC ⌬OP (25).
Comparison of the predicted amino acid sequences of the different Notch homologs revealed that the intermediate regions of Notch1 and Notch2 were highly conserved and constitute a STR region (Fig. 1A). The Notch2 STR comprises 33 amino acids including 5 serine and 4 threonine residues, and only four of these residues are conserved between the Notch1 and Notch2 homologs. We have now investigated whether N2IC ⌬OP function in the context of specific cytokines is modulated by phosphorylation of the STR domain.
To evaluate phosphorylation of the N2IC ⌬OP molecule, we labeled 32D-N2IC ⌬OP cells with [ 32 P]orthophosphate in the presence of G-CSF, GM-CSF or medium alone, and analyzed the immunoprecipitated N2IC ⌬OP proteins in 8% PAGE (Fig. 1, B and C). We detected a radiolabeled N2IC ⌬OP molecule under all three conditions, indicating a basal level of phosphorylation. The degree of phosphorylation of N2IC ⌬OP did not change significantly in cells stimulated with GM-CSF (Fig. 1B) or medium alone (data not shown). In cells incubated with G-CSF, a relative increase in phosphorylation of N2IC ⌬OP was observed within 5 min and persisted for at least 60 min following stimulation (Fig. 1B). By contrast, N1IC ⌬OP molecule showed a basal phosphorylation that persisted in the presence of GM-CSF and dramatically decreased after 30 min of incubation with G-CSF (Fig. 1C).
Phosphorylation Pattern of Notch2-IC Is Different in 32D Cells Stimulated with G-CSF or GM-CSF-To investigate whether differences in N2IC ⌬OP phosphorylation might be associated with its cytokine-specific activity, we analyzed the phosphorylation patterns of N2IC ⌬OP proteins obtained from 32D cells incubated with medium, G-CSF, or GM-CSF. We purified metabolically labeled [ 32 P]N2IC ⌬OP from 32D cells under each condition and performed tryptic phosphopeptide maps. As seen in Fig. 2, A-C, most of the phosphopeptides generated under the three conditions were equivalent, suggesting that many residues are constitutively phosphorylated in Notch-IC molecules as previously reported (39). However, three phosphorylated peptides (designated A, B, and C) were predominantly seen in cells exposed to G-CSF (Fig. 2C). Together, the tryptic maps and the observation that N2IC ⌬OP inhibits GM-CSF-induced differentiation but fails to inhibit G-CSFinduced differentiation (Ref. 6 and Fig. 3B) suggested that phosphorylation of peptides A, B, and C might contribute to regulation of Notch2 activity in the context of G-CSF stimulation. Although NothIC ⌬OP has been extensively used to mimic the activated Notch phenotype in different systems (10,25,45), we confirmed the presence of phosphopeptides A, B, and C in the full intracellular domain of Notch2 (N2IC) in stably transfected 32D cells exposed to G-CSF (Fig. 2D). Similarly to the observed effect of N2IC ⌬OP , clones expressing N2IC failed to block G-CSF-induced differentiation ( Table I).
Deletion of Amino Acids 2067-2099 Permits Cytokine-independent Notch2-IC Activity and Abrogates G-CSF-specific Phosphorylation-Because our previous studies suggested that the STR region (Fig. 1A) of the NCR domain conferred cytokine specificity to the Notch1 molecule (6), we postulated that the STR region might also be responsible for cytokine-specific effects of Notch2. To address this possibility, we deleted the STR region (amino acids 2067-2099) from the N2IC ⌬OP molecule and expressed the mutant N2IC ⌬2067-2099 construct in 32D cells. Multiple individual clones were isolated and evaluated for the capacity to differentiate in response to G-CSF and GM-CSF. Fig. 3 represents results from two experiments, each evaluating G-CSF and GM-CSF induced differentiation of five N2IC ⌬2067-2099 clones (Fig. 3C), two N2IC ⌬OP clones (Fig. 3B), and wild type 32D cells (Fig. 3A) for comparison. In the presence of G-CSF, an average of 74 Ϯ 4% of the cells expressing N2IC ⌬2067-2099 remained undifferentiated after 6 days, compared with 4.5 Ϯ 1% of the cells expressing N2IC ⌬OP . In simultaneous cultures containing GM-CSF, over 70% of the cells expressing N2IC ⌬2067-2099 remained undifferentiated after 6 days, comparable with those expressing N2IC ⌬OP and consistent with our previous results. Including all experiments, seven different N2IC ⌬2067-2099 clones remained primarily undifferentiated after culture in either G-CSF or GM-CSF, suggesting that this molecule is constitutively active in a cytokine-independent manner.
We next analyzed the phosphopeptide maps generated from N2IC ⌬2067-2099 -expressing cells incubated in the presence of G-CSF or GM-CSF (Fig. 3C). In contrast to the parental N2IC molecule (Figs. 2D and 3B), peptides A, B, and C were undetectable in N2IC ⌬2067-2099 -expressing cells stimulated with G-CSF. The N2IC ⌬2067-2099 molecule exhibits almost identical tryptic maps following exposure to G-CSF or GM-CSF (Fig. 3C), correlating its phosphorylation status with differentiation phenotypes. In addition, these phosphopeptide maps are indistinguishable from that of the N2IC ⌬OP molecule from cells exposed to GM-CSF (Fig. 3B), further correlating phosphorylation status with activity. These experiments suggest that Notch2 function is modulated by phosphorylation of peptides A, B, and C. These phosphopeptides only appear when the region comprised between amino acids 2067-2099 is present, and its deletion correlates with a cytokine-independent Notch2 inhibitory phenotype.
Phosphorylation of Ser 2078 Is Crucial for the Regulation of Notch2 Function in Myeloid Differentiation-To identify the specific residues of Notch2, which are phosphorylated in the presence of G-CSF and modulate Notch2 activity, we performed small deletions and point mutations in the STR domain of N2IC ⌬OP (Table I). We then evaluated the phosphorylation pattern and functional activity of these mutant molecules expressed in 32D cells. Among the small deletion mutants, we found that deletion of amino acids 2077-2081 was most informative. 32D cells expressing N2IC ⌬2077-2081 showed very weak phosphorylation of peptide B and the absence of peptide A (Fig.  4B) and did not differentiate in the presence of G-CSF ( Fig. 4F and Table I). Analysis of N2IC molecules containing different point mutations in the STR domain revealed that a Ser to Ala mutation of residue 2078 was sufficient to prevent phosphorylation of peptide A (Fig. 4D) and to permit Notch2 inhibitory activity during G-CSF induced differentiation of 32D cells (Fig.  4H). To further corroborate this observation, the same phenotype was obtained with the double mutation S2078A/S2090A whereas the phosphorylation and differentiation phenotype obtained with the single mutant S2090A resembled the N2IC ⌬OP wild type molecule ( Table I).
Deletion of the STR Domain Does Not Alter Protein Stability or Subcellular Localization of the N2IC ⌬OP Molecule-We next addressed whether deletion of the STR domain might increase the stability of the Notch protein thus enhancing its ability to inhibit of cell differentiation. N2IC ⌬OP and N2IC ⌬2067-2099 32D expressing clones were incubated in cycloheximide and G-CSF-differentiating media, and cell extracts were obtained at different time points. As shown in Fig. 5A, the stability of the N2IC ⌬OP molecule is even higher than the stability of the mutant lacking the STR domain. Moreover, after 3 days of culture in G-CSF comparable levels of N2IC ⌬OP or N2IC ⌬2067-2099 were detected (Fig. 5B). Together these results indicate that the ability of different mutants to block G-CSF-induced differentiation does not correlate with protein levels.
Then we examined the possibility that the immediate consequence of STR phosphorylation was a change in the nuclear localization of N2IC ⌬OP molecule. Analysis of the N2IC ⌬OP , N2IC ⌬2067-2099 (Fig. 5C), as well as N2IC S2078A (data not shown) molecules by immunofluorescence showed that, after a 15-min incubation in G-CSF, all localized primarily to the nucleus of 32D cells. Nevertheless, after a longer G-CSF incubation (48 h) N2IC ⌬OP molecule was diffusely localized both in the nucleus and cytoplasm (25).
Finally, we have determined the ability of the different mutant Notch molecules to transactivate the Hes-1 promoter in NIH-3T3 cells. Results in Table I reflect that the cytokine-independent inhibitory phenotype of the mutant N2IC ⌬OP molecules does not correlate with a higher Hes-1 transactivation capacity. DISCUSSION The Notch signaling pathway participates in many cell fate decisions during embryonic and adult tissue development (for reviews see Refs. 1, 46, 47). The regulation of such a widely

Notch2 Phosphorylation in 32D Differentiation
used pathway is complex and involves different mechanisms. Phosphorylation of the Notch molecule was first described in Drosophila (35) and has more recently been reported in mammalian cells (39,40). Here we have presented evidence that phosphorylation is important in the regulation of Notch2 activity in the context of specific cytokine signals.
Intracellular Notch1 and Notch2 have been shown to function as activated molecules in several systems. Although both homologs display similar biochemical activity when assayed in in vitro systems (17,18), when overexpressed in hematopoietic cells they have been shown to exhibit functional specificity (25). We have now demonstrated that deletion of a STR region between amino acids 2067-2099 abrogates the cytokine-restricted activity of Notch2 on 32D myeloid differentiation.
Within this region, we have determined that phosphorylation of the Ser 2078 residue in the context of G-CSF stimulation is important in regulating N2IC ⌬OP activity.
Notch2 Function Is Regulated by Phosphorylation-Regulation of Notch activity occurs at a variety of different levels during the protein lifespan. Prior to placement on the cell surface, the Notch protein undergoes proteolytic processing to generate a heterodimeric receptor (48) and glycosylation by fringe to modulate ligand binding specificity (38). Following ligand binding, Notch is cleaved by at least 2 different proteases, ␥-secretase/presenilins (49, 50) and disintegrinmetalloproteases (ADAM) (51,52), to release the intracellular domain. Finally, intracellular Notch is modified by phosphoryl- a Relative intensity of phosphopeptides A and B in tryptic maps from different molecules expressed in 32D and exposed to G-CSF. Notch2 Phosphorylation in 32D Differentiation ation in response to ligand binding (40) and to differentiation signals (this report).
It has been reported that phosphorylation of Drosophila Notch is necessary for nuclear translocation and interaction with the transcription factor Suppressor of Hairless (Su(H)) (35). Several recent studies also support a role for phosphorylation in the regulation of subcellular localization and activity of intracellular Notch. For example, Notch-IC proteins have been found in the nucleus of dendritic cells in distinct phosphorylated forms (39). In other reports, hyperphosphorylation of Notch has been associated with Notch activation, nuclear translocation, and transformation ability (41,40). We previously observed that N2IC ⌬OP molecule is able to inhibit differentiation when induced by GM-CSF but fails to block differentiation in the presence of G-CSF (25). In the present study, we have found that N2IC ⌬OP is phosphorylated in 32D cells in IL3 or GM-CSF medium and increased phosphorylation of N2IC ⌬OP occurs when cells are incubated in G-CSF. Together these findings suggest the possibility that G-CSF induces specific phosphorylation of N2IC ⌬OP , thereby rendering the molecule inactive. By contrast, when cells are incubated with G-CSF, N1IC ⌬OP molecule decreased its phosphorylation suggesting that both molecules may be differentially regulated in response to the same cytokine.
Our results show increased phosphorylation of at least three spots (A, B, and C) in the N2IC and N2IC ⌬OP after G-CSF stimulation. In all three mutants that are able to block G-CSFinduced differentiation peptide A is absent, thus strongly suggesting that Ser 2078 is critical in regulating the Notch2 cytokine specific activity in 32D cells. The fact that peptide A is completely absent in the deletion mutant ⌬2077-2081, the double point mutant S2078A/S2090A, and the S2078A single mutant suggests that this peptide includes Ser 2078 . When a theoretical trypsin digest map for N2IC ⌬OP is generated, the peptide containing Ser 2078 is 29 amino acids long (Leu-Leu-Asp-Glu-Tyr-Asn-Val-Thr-Pro-Ser-Pro-Pro-Gly-Thr-Val-Leu-Thr-Ser 2078 -Ala-Leu-Ser-Pro-Val-Leu-Cys-Gly-Pro-Asn-Arg) and its predicted mobility coincides with the actual migration of peptide A. We cannot exclude that spots B and C are multiple phosphorylation states of the same peptide or partial digestions containing Ser 2078 thus explaining their decreased intensity in the tryptic maps lacking peptide A. Moreover, we have shown that different mutant molecules that lack Ser 2078 are able to inhibit G-CSF induced differentiation. These results strongly suggest that phosphorylation of this residue is crucial for regulating Notch2 cytokine specific activity in 32D cells. Of note, Ser 2078 is unique to Notch2.
We have focused on phosphopeptides A, B, and C in this study as these were the only peptides to reproducibly correlate with Notch2 function and cytokine stimulus. However, in individual experiments, different peptides showed variable differences in intensity following exposure to G-CSF or GM-CSF. Thus, we cannot exclude the possibility that phosphorylation of other residues in the N2IC ⌬OP molecule may also influence its activity. It also remains possible that other residues in addition to Ser 2078 contribute to the appearance or disappearance of phosphopeptide A. It is worth noting that initial phosphorylation of Ser 2078 may be required to establish target motifs for subsequent phosphorylation by CKI (-S(P)-A-L-S-) or GSK3 (-T-V-L-T-S(P)-A-L-S). In this sense and using one of the available proteomics data bases (PhophoBase v2.0; Center for Biological Sequence Analysis; www.cbs.dtu.dk), we searched for potential phosphorylation sites in the N2IC ⌬OP sequence. This search revealed consensus sequences for calmodulin kinase II, (CaMKII), casein kinase I (CKI), casein kinase II (CKII), glycogen synthase kinase 3 (GSK3), protein kinase A (PKA), and protein kinase C (PKC). The specific kinase responsible for phosphorylation of Ser 2078 remains to be determined and is a subject of current investigation. Molecular pathways that might directly link G-CSF signal transduction to phosphorylation of Notch2 are particularly intriguing. Many cytokines, including G-CSF and GM-CSF, activate kinase signaling cascades (53), some of which also interact with the Notch pathway. For example, interactions between the Notch and Ras/MAP kinase signaling pathways have been shown to result in reciprocal modulation of the two pathways: in some cases, Ras signaling is affected by Notch inhibition of MAP kinase activity (18, 54 -56); in others, Notch signaling is modulated by MAPKinduced phosphorylation of molecular components of the Notch pathway (reviewed in Ref. 1). MAPK consensus sequences are particularly ambiguous, but include a Thr/Ser residue followed by a Pro, which are also present in the STR of Notch1 and Notch2 (Fig. 1).
A Role for Notch Homologs in Myeloid Differentiation-Over the past few years it has become clear that Notch signaling is involved in the regulation of hematopoiesis (reviewed in Ref. 47). Although many details are still lacking, the influence of Notch1 at multiple differentiation branch points during lymphoid development has been well documented. For example, Notch1 first influences the development of cells along the T or B cell lineage (23) and then participates in the determination of sequential cell fates during T cell differentiation (22,57,24).
Definitive studies of Notch function in myeloid development (including erythroid, megakaryocytic, granulocytic, and monocytic lineages) have been more elusive, possibly because of the relative complexity of Notch signaling in this system. However, a number of studies indicate that Notch signaling is involved in myelopoiesis, and provide clues as to specific functions. Expression of active (Notch-IC) molecules or ligand-induced activation of Notch in myeloid cell lines such as 32D (11,42) and HL60 (26), or in primary hematopoietic cells (30 -32) generally result in undifferentiated phenotypes. However, in both invertebrate and mammalian systems, the effects of Notch have been shown to be both multifaceted and highly context-dependent. Thus, depending on the specific cell type and environmental signals, Notch activity may promote or inhibit differentiation, prevent or induce apoptosis (58,59), and promote cell cycle progression and mitosis or induce cell cycle arrest (60,61). In the hematopoietic system, context-dependent effects of Notch have been observed in progenitor populations (62), and during lineage differentiation of T cells (reviewed in Ref. 63), monocytes (59), myeloid (25), and erythroid cells (28,29). For example, in the K562 erythroleukemia cell line, expression of Notch1-IC blocks erythroid but not megakaryocytic differentiation (28), whereas in the MEL erythroleukemia cell line Notch1 inhibits apoptosis and is necessary for erythroid differentiation (29). Although most studies in 32D cells indicate that Notch activity inhibits terminal differentiation and enhances survival of progenitors (11,42,64), one group has reported that transient expression of Notch1 promotes granulocytic differentiation through a CSLdependent pathway (27). In the present studies, it is formally possible that G-CSF phosporylation is converting Notch2 from an inhibitory to a pro-differentiative molecule. However, we observe comparable rates and extent of differentiation among 32D clones expressing N2IC, N2IC ⌬OP or control constructs. Nonetheless, mutations of Ser 2078 to Asp or Glu could be very informative in this regard.
Notch interacts with many different intracellular molecules (for review see Ref. 65). Although CSL proteins appear to function as the primary effectors of Notch signaling in many developmental systems, it is now clear that alternative, CSLindependent pathways are also important for Notch signal transduction (66,67,39). When assayed in NIH-3T3 cells, deletion of the STR and point mutation of different residues in this domain do not dramatically modify the transactivation of the Hes-1 promoter compared with the N2IC ⌬OP molecule. Despite the fact that NIH-3T3 cells are not a cytokine-dependent model, these results might suggest that inhibition of 32D cells differentiation by Notch may involve a CSL-independent pathway.
Our previous studies demonstrating the cytokine specific activities of Notch1 and 2 on 32D differentiation further illustrate the subtleties of Notch signaling. Even in a defined system in which inductive signals can be controlled and potential cell fates are limited, the effects of Notch signaling are complex. Whereas the present studies do not resolve all of the issues pertaining to Notch function in myeloid differentiation, they do provide some insights into the molecular regulation of Notch activity in 32D cells. The precise mechanism by which G-CSFinduced phosphorylation influences Notch2 activity remains to be determined. However, we have demonstrated that protein levels and nuclear localization are comparable for N2IC molecules that differ in their phosphorylation states (N2IC ⌬OP and N2IC ⌬2067-2099 ). Thus, it is unlikely that phosphorylation results in enhanced protein degradation, impaired nuclear translocation, or cytoplasmic sequestration of N2IC. We speculate that phosphorylation of N2IC ⌬OP may be determining a conformational change and/or facilitating protein interactions in the nucleus of 32D cells to ensure the inactivity of the molecule after stimulation with G-CSF. The fact that full-length transmembrane Notch does not show phosphorylation in the presence of G-CSF (data not shown) strongly suggests that only when Notch-IC is cleaved, phosphorylation events may regulate its function.
In addition to the cdc10/ankyrin repeats, which are essential for Notch function, other regions of the Notch-IC molecule are important for defining specific activities. For example, the RAM23 domain has high binding affinity for CSL proteins and enhances signaling through CSL-dependent pathways (68). The region C-terminal to the ankyrin repeats has been associated with specific protein-protein interactions (69), subcellular trafficking (10), and transforming activity (70). We previously designated the region between the ankyrin repeats and the C-terminal nuclear localization signals as the NCR region (which includes the STR) because of its association with the cytokine-specific effects of Notch on myeloid differentiation. Observations from those studies are complementary to results of the present study, and together indicate that differential phosphorylation within the STR is important to determine the function of Notch2 during G-CSF and GM-CSF-induced differentiation of 32D cells. More precisely, phosphorylation of Ser 2078 is crucial for inactivating the Notch2 molecule in the context of G-CSF induction.