Calcium/Calmodulin Inhibition of Transcriptional Activity of E-proteins by Prevention of Their Binding to DNA*

The Ca 2 (cid:1) sensor protein calmodulin can interact with the DNA binding basic helix-loop-helix (bHLH) domain of E12, E47, and SEF2-1 (E2-2), which belong to the E-protein subclass of bHLH transcription factors. This interaction inhibits the DNA binding of these bHLH proteins in vitro , and an ionophore that increases intracellular Ca 2 (cid:1) can inhibit transcriptional activation by the E-proteins. Here we have attempted to determine if these phenomena reflect a direct calmodulin-depend-ent inhibition of DNA binding by E-proteins in vivo . We show that calmodulin overexpression inhibits the transcriptional activity of E12, E47, and SEF2-1. We have compared calmodulin effects on DNA binding in vitro and on activation of transcription in vivo using a series of E12 mutants harboring defined alterations within the basic sequence of the bHLH domain that reduce their ability to bind calmodulin to varying degrees. We find a striking direct correlation between the ability of calmodulin to inhibit their DNA binding in vitro and the ability of overexpressed calmodulin or cellular Ca 2 (cid:1) mobilization to inhibit their transcriptional activity in vivo . Furthermore, E12 and overexpressed calmodulin were co-localized in the nucleus, and calmodulin pull-down experiments with cell extracts showed a Ca 2 (cid:1) -de-pendent interaction between calmodulin and E12 but not with a calmodulin inhibition-deficient E12 mutant. Chromatin immunoprecipitation showed that calmodulin The pre-cleared was incubated 2 h further at (cid:1) 4 °C in presence of 2 (cid:2) g/ml anti-E47 (N-649) antibody (Santa Cruz) for E47 immunoprecipitation and 2 (cid:2) g/ml anti-FLAG antibody M2 for FLAG immunoprecipi- tation. Immunocomplexes were bound for 1 h to 15 (cid:2) l of protein G-Sepharose beads blocked in 0.3 mg/ml sonicated herring sperm DNA. Beads were washed times with radioimmune precipitation assay lysis and three with Tris-EDTA, 8.0. After every third wash, the beads were moved to clean tubes to prevent nonspecific binding of to the tube walls. Cross-links were reversed with SDS in Tris-EDTA, at overnight. DNA was phenol-chloroform-extracted, ethanol-precipitated, and PCR-amplified for 28 cycles with the primer pair 5 (cid:3) -TGTATCTTATGGTACTGTAACTG-3 (cid:3) and 5 (cid:3) -CTT-TATGTTTTTGGCGTCTTCCA-3 (cid:3) that flanks the promoter sequence of the E-box luciferase reporter plasmid.

Calmodulin is a ubiquitously expressed regulator of numerous cellular processes including cell cycle progression, cell mobility and contraction, ion homeostasis, and transcription. Most target proteins bind to the Ca 2ϩ -loaded form of calmodulin. Therefore, upon signals leading to increased intracellular Ca 2ϩ concentration, calmodulin is able to bind to a new set of target proteins and modulate their activity (1). Several transcriptional regulators are direct targets for calmodulin or are regulated by calmodulin-dependent kinases or the calmodulin-dependent phosphatase calcineurin (2,3).
Basic helix-loop-helix (bHLH) 1 proteins are a large family of transcription factors that are important regulators of many cellular functions including differentiation processes such as myogenesis, hematopoiesis, and neurogenesis (4). bHLH transcription factors can be divided into several classes based on their structure, dimerization properties, function, and expression pattern. E-proteins (class I bHLH transcription factors) are encoded by three mammalian genes, E2A, SEF2-1 (also denoted E2-2), and HEB. The E2A gene encodes the alternatively spliced protein products E12 and E47 (4). E-proteins usually function as heterodimers with members of a large family of tissue specific (class II) bHLH transcription factors. However, in B-cell development, where E-proteins are essential, the E-proteins are believed to function as dimers within the E-protein subfamily since no B-cell-specific class II bHLH transcription factor has been found (4).
E-box DNA sequences, the recognition sequences for heterodimers of tissue-specific bHLH transcription factors and Eproteins, are also binding sites for E-protein homodimers (5). Potentially, binding of these homodimers to the E-boxes of tissuespecific promoters could thereby block binding of the appropriate tissue-specific bHLH factor/E-protein heterodimers. Therefore, mechanisms that regulate the binding of distinct bHLH dimers to these E-boxes are of significant interest.
SEF2-1 and the E2A proteins E12 and E47 all bind calmodulin in vitro at physiologically relevant calmodulin concentrations (6). The DNA binding basic sequence of these bHLH proteins also binds to calmodulin, and this interaction is modulated by sequences directly N-terminal to the basic domain (7). Because the binding domains overlap, calmodulin bound E-proteins are inhibited from binding to DNA in vitro (6). Significantly, increased intracellular Ca 2ϩ concentration induced by the ionophore ionomycin is a potent inhibitor of the transcriptional activity of E47 and SEF2-1 on an E-box containing reporter plasmid (6). However, whether this correlation reflects an inhibition of the DNA binding of the E-proteins in vivo by calmodulin was not analyzed.
Here we show that calmodulin overexpression inhibits the transcriptional activity of E12, E47, and SEF2-1. Using E12 mutants with amino acid substitutions in the calmodulin binding basic sequence, we also report a striking direct correlation between the levels of calmodulin inhibition of DNA binding in vitro and the effects of calmodulin overexpression and cellular Ca 2ϩ mobilization on transcription. We also show nuclear colocalization of E12 and overexpressed calmodulin and a Ca 2ϩdependent interaction in cell extracts between calmodulin and E12 but not a calmodulin inhibition-deficient E12 mutant. Furthermore, DNA binding in vivo of E12, but not a calmodulinresistant E12 mutant, is inhibited by overexpression of calmod-ulin. The data suggest that Ca 2ϩ signaling inhibits the transcriptional activity of an E-protein by establishing an Eprotein-calmodulin complex that prevents the E-protein from interaction with target DNA.

EXPERIMENTAL PROCEDURES
Plasmids and Mutagenesis-The cDNAs coding for murine E12 and E47 (6) were inserted into pcDNAI/amp (Invitrogen) by excision with XhoI and BamHI and subsequent blunt end ligation to the EcoRV site of pcDNAI/amp. The MyoD cDNA was inserted as an EcoRI fragment from pMSV-MyoD (6) into the EcoRI site of pcDNAI/amp. The cDNA coding for human SEF2-1 (E2-2) was isolated from a human ZAP library containing cDNA inserts from the Burkitt's lymphoma cell line BL-29 (8). For expression vector construction, a SEF2-1 cDNA of the SEF2-1B type (9) was isolated from this cDNA library using a SEF2-1B probe (9), and it was subcloned as a blunt-ended EcoRI fragment into blunt-ended XhoI-BalI-digested pKGE346 (10), thereby replacing the chloramphenicol acetyltransferase-encoding fragment. The luciferase reporters for E12, 6ϫ(E5ϩE2)-luciferase, and MyoD, 4ϫ(MEF1)luciferase and the hCMV-␤gal plasmid for normalization of transfections have been described previously (6). The expression vector for the mammalian calmodulin (CaM), pcDNAI/amp mCaM, has also been described previously (11). E12 basic sequence mutants were generated by standard PCR mutagenesis and substituted into full-length E12 in pcDNAI/amp E12 as a DrdI-StuI fragment. For construction of Cterminal FLAG derivatives of E12 and the E12 mutant m8-4-7, BamHI, BglII, and XbaI sites were introduced at the stop codon of E12 and m8-4-7. Double-stranded oligonucleotides coding for two tandem copies of the FLAG epitope sequence were inserted between the BamHI and BglII sites.
Cell Culture and Transfections-DG75 B-cells were maintained in RPMI medium supplemented with 5% fetal calf serum and antibiotics. DG75 cells were transiently transfected by electroporation as described (12). The IMR-32 neuroblastoma cell line was maintained in Hepesbuffered minimum essential medium supplemented with 10% fetal calf serum, nonessential amino acids, and antibiotics. IMR-32 cells were transfected using polyethyleneimine (13). 10 6 cells were seeded per 5-cm plate 1 day before transfection. DNA-polyethyleneimine complexes were formed in 1 ml of serum-free minimum essential medium for 10 min and added on cells with the addition of 1 ml of serum-free minimum essential medium. Cells were exposed to DNA-polyethyleneimine complexes for 2 h, after which the medium was replaced with serum-containing minimum essential medium with or without 5 mM caffeine. Cells were harvested 20 h after the start of transfection and analyzed for luciferase and ␤-galactosidase activity as described (12).
Western Blot with Nuclear and Cytoplasmic Extracts-Cells were pelleted and washed once with cold phosphate-buffered saline. The cell pellet was resuspended in 10 pellet volumes of hypotonic buffer (10 mM Hepes, pH 8, 10 mM KCl, 0.1% Nonidet P-40, 1 mM EDTA) supplemented with protease inhibitor mixture (Roche Applied Science) and kept on ice for 10 min. The cells were then vortexed for 10 s, the lysate was pelleted by centrifugation at 10,000 ϫ g for 1 min at ϩ4°C, and the supernatant containing the cytoplasmic fraction was collected. The pellet was washed once with 1 ml of the hypotonic buffer, and nuclear proteins were extracted with one pellet volume of high salt buffer (20 mM Hepes, pH 8, 400 mM NaCl, 10% glycerol, 1 mM EDTA) supplemented with protease inhibitor mixture on ice for 30 min with occasional flicking of the tubes. The insoluble nuclei were pelleted by centrifugation at 20,000 ϫ g for 3 min at ϩ4°C, and the supernatant with the nuclear extract was collected. Western blot analysis was performed using SuperSignal (Pierce) according to the manufacturer's instructions using anti-E12 antibody (H-208, Santa Cruz Biotechnology) and horseradish peroxidase-conjugated secondary antibody (Amersham Biosciences).
Analysis of DNA Binding-E12 wild type and basic sequence-mutated E12 proteins containing the amino acid sequence from 483 to 652 were produced for DNA binding analyses as previously described (7). E12 wild type and most mutants were produced in Escherichia coli, whereas mutants m1-m7 were produced by in vitro translation. Electrophoretic mobility shift assays (EMSAs) of the E12 proteins were performed as previously described using an end-labeled DNA sequence containing the E5 E-box as probe (7). DNA binding was quantified with a PhosphorImager and the ImageQuant software (Amersham Biosciences).
Immunocytochemistry-Immunostaining and confocal microscopy of transfected DG75 cells was performed essentially as described (14) with the exception that cells were fixed with 2% paraformaldehyde in phosphate-buffered saline for 15 min. Primary antibodies used were rabbit anti-E12 (H-208) and goat anti-calmodulin (N-19) (both from Santa Cruz Biotechnology). The secondary antibodies were fluorescein isothiocyanate-conjugated donkey anti-rabbit IgG and rhodamine red X-conjugated donkey anti-goat IgG (both from Jackson ImmunoResearch Laboratories). All antibodies were used as 1:50 dilutions. Higher laser power and detector sensitivity was used to detect endogenous proteins compared with overexpressed proteins in the confocal microscopy.
Calmodulin-Sepharose Pull-down-DG75 cells were transiently transfected with the expression plasmid for E12, the m8-4-7 mutant, or the empty vector. Whole cell lysates were prepared with Nonidet P-40 lysis buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1% Nonidet P-40) and diluted 5-fold in EMSA binding buffer with either 0.5 mM CaCl 2 or 1 mM EDTA. Diluted extracts were incubated with calmodulin-Sepharose (Stratagene) for 30 min, and the beads were washed three times with the Ca 2ϩ -or EDTA-containing buffer. Calmodulin-Sepharosebound proteins were separated by SDS-PAGE, blotted, and detected with the anti-E12 (H-208) antibody or, in the case of the FLAG tagged proteins, with the anti-FLAG antibody M2 (Sigma).
Chromatin Immunoprecipitations-10 7 DG75 cells were transiently transfected with the E-box luciferase reporter and with expression vector for either E47, E12, or the E12 mutant m8-4-7 and for calmodulin where indicated. 20 h after transfection, cells were cross-linked with 1% formaldehyde for 10 min at room temperature. Cross-linking was quenched with a 1 ⁄20 volume of 2.5 M glycine, and cells were washed with Tris-buffered saline. Cells were lysed on ice with 0.5 ml of radioimmune precipitation assay lysis buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1% Nonidet P-40, 0.1% SDS, 0.5% sodium deoxycholate, and 5 mM EDTA) supplemented with protease inhibitor mixture (Roche Applied Science). The lysate was sonicated, clarified by centrifugation at ϩ4°C, and pre-cleared overnight with 15 l of protein G-Sepharose. The precleared extract was incubated 2 h further at ϩ4°C in presence of 2 g/ml anti-E47 (N-649) antibody (Santa Cruz) for E47 immunoprecipitation and 2 g/ml anti-FLAG antibody M2 for FLAG immunoprecipitation. Immunocomplexes were bound for 1 h to 15 l of protein G-Sepharose beads blocked in 0.3 mg/ml sonicated herring sperm DNA. Beads were washed 6 times with radioimmune precipitation assay lysis buffer and three times with Tris-EDTA, pH 8.0. After every third wash, the beads were moved to clean tubes to prevent nonspecific binding of DNA to the tube walls. Cross-links were reversed with 1% SDS in Tris-EDTA, pH 8.0, at 65°C overnight. DNA was phenol-chloroformextracted, ethanol-precipitated, and PCR-amplified for 28 cycles with the primer pair 5Ј-TGTATCTTATGGTACTGTAACTG-3Ј and 5Ј-CTT-TATGTTTTTGGCGTCTTCCA-3Ј that flanks the promoter sequence of the E-box luciferase reporter plasmid.

RESULTS
Transcriptional Activity of E12, E47, and SEF2-1 Is Inhibited by Calmodulin Co-expression-Binding of calmodulin has been shown to inhibit DNA binding of E-proteins in vitro (6,7,15), and a Ca 2ϩ ionophore that increases the intracellular Ca 2ϩ concentration has been shown to inhibit activation of transcription by E47 (6). On the way to determining if these phenomena are linked, we analyzed whether overexpression of calmodulin can also inhibit the transcriptional activity of E-proteins. The B-cell line DG75 was co-transfected with an E-box luciferase reporter and an expression vector for either E12, E47, SEF2-1, or the empty vector together with either the calmodulin expression vector or the corresponding empty expression vector. Calmodulin overexpression was found to inhibit E12, E47, and SEF2-1-induced E-box reporter activity in a dose-dependent manner (Fig. 1A). Co-transfection of 10 g of calmodulin expression plasmid decreased transcriptional activation by E12 more than 15-fold, and E47-and SEF2-1-induced E-box reporter activity was also strongly inhibited by calmodulin overexpression. In contrast, transcriptional activation by MyoD, a bHLH protein whose DNA binding shows little calmodulin sensitivity in vitro (6,7,15), was not significantly affected by calmodulin overexpression (Fig. 1A). Western blot analysis of cytoplasmic and nuclear cell extracts revealed that calmodulin overexpression did not affect the level of expressed E12 protein or its nuclear/ cytoplasmic distribution (Fig. 1B). Thus, calmodulin overexpression can selectively inhibit the transcriptional activity of an E-protein without affecting the nuclear level of the protein.

E12 Mutants with Decreased Sensitivity to Calmodulin-To
analyze if calmodulin binding to the basic sequence of the bHLH domain of E12 causes the inhibition of the transcriptional activation by cellular Ca 2ϩ mobilization and calmodulin overexpression, we generated E12 mutants with amino acid substitutions in the basic sequence that bind both DNA and calmodulin. With the aim of generating E12 mutants with disrupted binding to calmodulin but not to DNA, we used the known structure of the E47-DNA complex (16) to predict those amino acids likely not to be critical for the DNA binding by E12. These residues were preferentially mutated to alanine or, when necessary, to glutamine to produce the E12 mutants m1 to m8 ( Fig. 2A).
Three of the eight point mutants, m1, m2, and m3 (I562A, R560A and V559A, respectively), showed no detectable DNA binding in EMSA (Fig. 2B) and were subsequently excluded from further study. However, the remaining five point mutants all retained the DNA binding (Fig. 2B). Therefore, the effect of calmodulin on their DNA binding was analyzed in detail (Fig.  3). None of the five single amino acid substitutions led to a calmodulin-resistant E12 mutant protein. Compared with wild type E12, four of these mutants, m4, m5, m6, and m7, showed a slight decrease in sensitivity to calmodulin that was more evident at the higher calmodulin concentration (1 M). Moreover, even though the fifth mutant, m8 (D561A), was less sensitive to calmodulin than the other four mutants, much of the sensitivity still remained (Fig. 3). In view of our recent NMR structural study, highlighting that the binding of the E12 basic sequence to calmodulin involves multiple alternative amino acid contacts, 2 we proposed that combinations of amino acid substitutions would lead to less calmodulin-sensitive mutants of E12. Thus, the amino acid substitution of the m8 mutant (D561A) was combined with m5 or m7 to produce the double mutants m8-5 and m8-7 or with those of both m4 and m7 to produce the triple mutant m8-4-7 ( Fig. 2A). As expected, the DNA binding of the double mutants m8-5 and m8-7 was FIG. 1. Calmodulin inhibition of the transcriptional activity of E12, E47, and SEF2-1. A, expression vectors for bHLH proteins (2.5 g) were transiently expressed in the B-cell line DG75 together with 2.5 g of a luciferase reporter plasmid for E-proteins, 6ϫ(E5ϩE2)-luciferase, or for MyoD, 4ϫ(MEF1)-luciferase. In addition, the hCMV-␤gal plasmid (2.5 g) was included to normalize each assay. Where indicated, expression vectors for bHLH proteins were co-transfected with either 2.5 or 10 g of the CaM expression plasmid pcDNAI/amp mCaM. Where 2.5 g or no pcDNAI/amp mCaM was used, total transfected DNA was equalized with 7.5 or 10 g of pcDNAI/amp vector DNA, respectively. As a control, the empty expression vector was used in place of the bHLH protein expression vectors (co). The luciferase reporter activities of harvested cells were measured and then normalized to the ␤-galactosidase activity from the co-transfected hCMV-␤gal plasmid. Transcriptional activities of bHLH proteins in the absence of calmodulin overexpression were set as 100%. The bars show the mean values Ϯ S.E. of three transfections. B, Western blot analysis of E12 protein levels in cytoplasmic and nuclear extracts of DG75 cells transfected as indicated. Equal amounts of protein from the cytoplasmic and nuclear extracts were separated by SDS-PAGE, and E12 was detected by Western blot with anti-E12 antibody (H-208, Santa Cruz).

FIG. 2. E12 basic sequence site-directed mutants used in the study.
A, amino acids in the basic sequence of mouse E12 were mutated to decrease inhibition of the DNA binding by calmodulin. The amino acid substitutions are in bold. For DNA binding analyses, E12 wild type (wt) and basic sequence-mutated E12 proteins containing the amino acid sequence from 483 to 652 were used. For transient transfections, the mutations were incorporated into the full-length E12 expression vector ("Experimental Procedures"). The asterisks indicate the three mutants that showed no detectable DNA binding in EMSA. B, DNA binding of wild type E12 and E12 basic sequence mutants to E5 E-box probe analyzed by EMSA. slightly less sensitive to calmodulin than that of the single mutant m8 (Fig. 3). The triple mutant m8-4-7 was the least calmodulin-sensitive mutant (Fig. 3). We also created mutants m10, m11, and m12 by combining glutamine substitutions in the amino acids substituted in the four single point mutants m4-m7 ( Fig. 2A). Although these new mutants were all more resistant to calmodulin than wild type E12, they were not more resistant than the single mutant m8. Therefore, we generated mutant m13, which combined the D561A mutation of mutant m8 with glutamine substitutions at all four amino acid positions mutated in m4-m7. The m13 mutant and mutant m8-4-7 were the most resistant mutants in the series, being ϳ20-fold less sensitive than wild type E12 to 1 M calmodulin (Fig. 3). The set of mutants provides ideal molecular tools to probe the connection between Ca 2ϩ /calmodulin inhibition of E-protein DNA binding in vitro and activation of transcription in vivo.
Decreased Inhibition of Transcriptional Activities of E12 Mutants by Ca 2ϩ /Calmodulin-We compared the effects of calmodulin overexpression on activation of transcription by the series of E12 mutants that represent a gradient of sensitivity to calmodulin inhibition of the DNA binding in vitro. DG75 cells were co-transfected with an expression vector for each E12 mutant in the presence of the calmodulin or empty expression vector. The relative effects of calmodulin overexpression on transcriptional activation by the various E12 mutants correlated to their ability to bind DNA in vitro in the presence of calmodulin (cf. Figs. 3 and 4A). Transcriptional activation by the single amino acid substitution mutants m4, m6, and m7, but not mutant m5, were slightly less inhibited by calmodulin overexpression than that of wild type E12. Moreover, among the single amino acid mutants tested, transcriptional activity by mutant m8 was the least inhibited by calmodulin overexpression (Fig. 4A). The transcriptional activity of the multiple amino acid substitution mutant m13 was inhibited by calmodulin overexpression to a similar extent as m8, whereas the activities of mutants m8-7 and m8-4-7 were nearly unaffected by calmodulin overexpression (Fig. 4A).
We next analyzed the effect of increased intracellular Ca 2ϩ on transcriptional activation in transfected DG75 B-cells by E12 and the most calmodulin-resistant mutant of E12, m8-4-7, by treatment with the Ca 2ϩ pump blocker thapsigargin that leads to Ca 2ϩ signaling in lymphocytes. The activity of the E-box reporter without overexpression of E12 was inhibited by thapsigargin treatment (Fig. 4B). This thapsigargin-sensitive activity is mainly due to E12/E47 since this is the predominant E-protein in DG75 (data not shown). Transcriptional activation by overexpressed E12 was as thapsigargin-sensitive as the endogenous E-protein activity (Fig. 4B). In contrast, the increase in transcriptional activation by overexpression of the mutant m8-4-7 was not affected by the Ca 2ϩ inducer. We conclude that the transcriptional activity of E12 can be inhibited by Ca 2ϩ stimulation in the presence of endogenous levels of calmodulin in DG75 cells, whereas transcriptional activation by a calmodulin-resistant mutant of E12 is not inhibited by Ca 2ϩ stimulation in the B-cell line.
Treatment of cells with caffeine has been shown to increase the intranuclear Ca 2ϩ concentration in some cell types (17,18). Therefore, to further study the effects of increased intracellular Ca 2ϩ on transcriptional activity of E12 and E12 mutants, transfected IMR-32 neuroblastoma cells were stimulated with caffeine. The Ca 2ϩ inducer caffeine decreased the transcriptional activity of wild type E12 to 40% of that of the activity in non-stimulated cells (Fig. 4C). Moreover, the effects of caffeine on the transcriptional activities were not significantly different from the wild type E12 for most mutants. However, the transcriptional activities of the multiple amino acid substitution mutants m8-7, m8-4-7, and m13 were all clearly less affected by caffeine than wild type E12. Thus, the three multiple amino acid substitution mutants whose DNA binding showed low FIG. 4. A, effects of calmodulin overexpression on the transcriptional activity of E12 and E12 mutants in DG75 cells. 2.5 g of expression vector for either E12 or individual E12 mutants were co-transfected with 2.5 g each of expression vector for calmodulin, CMV-␤gal expression vector for normalization of transfections, and E-box luciferase reporter. The transcriptional activities of E12 and E12 mutants cotransfected with the calmodulin expression vector are expressed as percent of their activities when co-transfected with the empty pcDNAI/amp vector. B, effects of thapsigargin on the transcriptional activity of E12 and mutant m8-4-7 in transfected DG75 cells. 0.5 g of expression vector for E12, m8-4-7 or the corresponding empty vector was co-transfected with 4.5 g of the empty pcDNAI/amp vector with 2.5 g of both hCMV-␤gal expression vector and the E-box luciferase reporter. 5 h after electroporation, the cells were treated with thapsigargin (Calbiochem) or left untreated. Cells were harvested 3 h later for analysis. The bars show the increases in luciferase activity relative to untreated E12transfected cells during the 3 h period of stimulation. C, effects of the Ca 2ϩ inducer caffeine on the transcriptional activity of E12 and E12 mutants in IMR-32 neuroblastoma cells. 2 g of expression vector for E12 or E12 mutant was co-transfected with 1.5 g of the E-box luciferase reporter and 0.5 g of hCMV-␤gal. Duplicate transfections were treated with 5 mM caffeine or left untreated. The effects of caffeine on transcriptional activity were calculated as percent transcriptional activity in caffeine-treated cells compared with non-treated cells. The bars show the mean values Ϯ S.E. of at least three independent transfections. calmodulin sensitivity in vitro all showed low caffeine and calmodulin overexpression sensitivity in vivo (cf. Figs. 3 and 4).

Direct Correlation between Calmodulin Sensitivity of E12 Mutants in Vitro and in Vivo-
The transcriptional activities of some of the E12 mutants were altered compared with the wild type E12 (data not shown). However, differences in the DNA binding abilities or transcriptional activities were not the cause of the differences between the E12 variants in the effects of calmodulin or caffeine, since there was no correlation between the transcriptional activities of wild type and mutants of E12 and the effects of calmodulin overexpression (r 2 ϭ 0.10, p ϭ 0.28) or caffeine stimulation (r 2 ϭ 0.04, p ϭ 0.52) on their transcriptional activity. Furthermore, the least calmodulinsensitive mutant, m8-4-7, possessed a similar DNA binding ability and transcriptional activity in the absence of calmodulin overexpression or caffeine stimulation as the wild type E12 (data not shown).
To analyze the correlation between the Ca 2ϩ /calmodulin sensitivities of E12 mutants in vitro and in vivo, the transcription levels of each E12 variant in the presence of overexpressed calmodulin and the Ca 2ϩ inducer caffeine were expressed as a function of the DNA binding in vitro in the presence of 200 nM CaM ( Fig. 5; regression lines are in gray and black, respectively). Linear regression analysis showed a striking direct correlation between calmodulin inhibition of DNA binding by E12 variants in vitro and the effects of calmodulin overexpression (r 2 ϭ 0.97, p Ͻ 0.0001) and caffeine stimulation (r 2 ϭ 0.96, p Ͻ 0.0001) on their transcriptional activities in vivo. Importantly, no individual mutant displayed any significant lack of correlation between Ca 2ϩ /calmodulin sensitivity in vitro and in vivo. Furthermore, the activities of the mutants upon calmodulin overexpression correlated well with their activities upon caffeine stimulation (r 2 ϭ 0.97, p Ͻ 0.0001). Thus, Ca 2ϩ /calmodulin sensitivity of E12 in vitro and in vivo shows indistinguishable amino acid sequence specificity.
Nuclear Co-localization of Overexpressed E12 and Calmodulin-Because overexpression of calmodulin could inhibit the transcriptional activity of E12, we wondered whether these proteins were co-localized in cells. We immunostained calmodulin and/or E12 expression plasmid-transfected DG75 cells with antibodies against calmodulin and E12 and with fluorescent secondary antibodies. The localization of the proteins was analyzed by confocal microscopy. As expected, endogenous and overexpressed wild type E12 transcription factor and the E12 mutant m8-4-7 were all exclusively nuclear (Fig. 6). The cal-modulin localization in this cell line was primarily cytoplasmic, although some nuclear calmodulin could also be detected. However, overexpressed calmodulin was mainly localized in the nucleus. Co-expression of E12 and calmodulin resulted in a striking overlap in the subnuclear localization of these two proteins (observed as yellow in Fig. 6, A and B). In contrast, the calmodulin binding-deficient E12 mutant m8-4-7 co-localized poorly with calmodulin in the nucleus on the basis of limited overlap between areas of red fluorescence and green fluorescence (Fig. 6, A and B).
Binding of E12 to Calmodulin-Sepharose-To study if wild type E12, but not a calmodulin binding deficient mutant, interacts with calmodulin in a cell extract, DG75 cells were transfected with an expression vector for either E12 or the calmodulin-insensitive mutant m8-4-7. Pull-down analyses of the resulting cell extracts were subsequently performed with calmodulin-Sepharose. Both endogenous E12 and overexpressed E12 bound to calmodulin-Sepharose in the presence of FIG. 6. Co-localization of overexpressed calmodulin and wild type E12 in the nucleus. A, DG75 cells were transiently transfected with 5 g of empty pcDNAI/amp vector DNA or an expression plasmid for wild type E12 or E12 mutant m8-4-7 and/or CaM. The total amount of expression plasmid DNA was standardized to 10 g by the addition where necessary of the empty vector DNA. Cells were fixed 20 h after transfection and double-stained for E12 (green fluorescence) and calmodulin (red fluorescence). The immunostainings were analyzed by confocal microscopy with sequential scanning. Higher laser power and detector sensitivities were used for detection of endogenous proteins compared with the overexpressed proteins. B, confocal microscopy sections at higher magnification of two additional representative cell nuclei of DG75 cells co-transfected with the expression plasmids for calmodulin and E12 or the m8-4-7 mutant. Overlap in the localization of E12 and calmodulin is observed as a yellow color.
Ca 2ϩ but not in the presence of EDTA (Fig. 7A). However, at least 10-fold more E12 antibody-reactive protein was eluted from the calmodulin-Sepharose when using cell extracts of E12-overexpressing cells compared with expression vector control-transfected cells (Fig. 7A). In contrast, equal levels of E12 antibody-reactive proteins from m8-4-7-transfected cells as from vector control-transfected cells was eluted from the calmodulin-Sepharose. Therefore, this binding likely represents endogenous E12 protein or a co-migrating E12 antibody-reactive protein. To further analyze the difference between E12 and m8-4-7 in binding to calmodulin-Sepharose, we introduced FLAG epitope-tagged E12 constructs into the DG75 cell line. Transfected FLAG-tagged E12 bound efficiently to the calmodulin-Sepharose, whereas no binding of FLAG-tagged m8-4-7 to the calmodulin-Sepharose was detected (Fig. 7B).
Calmodulin Inhibition of in Vivo DNA Binding of E12 and E47-To study the effect of calmodulin overexpression on in vivo DNA binding of E47, E12 and the calmodulin-insensitive E12 mutant m8-4-7, chromatin immunoprecipitations were performed. Calmodulin overexpression significantly inhibited in vivo DNA binding of E47 (Fig. 8A) and wild type E12 (Fig.  8B) to the E-boxes located on the luciferase reporter plasmid. In contrast, DNA binding of the calmodulin-resistant mutant m8-4-7 was not decreased by calmodulin overexpression (Fig. 8B). DISCUSSION Calmodulin binding to the basic sequence of the bHLH transcription factors E12, E47, and SEF2-1 and how this inhibits their DNA binding in vitro has been extensively characterized (6,7,15,19). Furthermore, increasing intracellular Ca 2ϩ by ionomycin treatment inhibits the transcriptional activity of E47 and SEF2-1 (6). Before this study, however, a correlation between these results through the inhibition of the DNA binding of the E-protein in vivo by calmodulin had not yet been established. In this report we provide evidence for the hypothesis that Ca 2ϩ signaling through calmodulin can inhibit the transcriptional activity of E-proteins in vivo through direct binding of calmodulin to the basic DNA binding sequence.
Notably, an increase in intracellular Ca 2ϩ was not needed for the inhibitory effect of overexpressed calmodulin on the transcriptional activity of the E-proteins. A likely explanation is that the non-induced Ca 2ϩ level is sufficient for a proportion of cellular calmodulin to be Ca 2ϩ -bound. This is supported by the fact that calmodulin overexpression in absence of Ca 2ϩ stimulation can inhibit the transcriptional activity of a Smad-responsive promoter, an effect that was proposed to occur through Ca 2ϩ -dependent interaction of calmodulin with Smad proteins (20, 21) Furthermore, we have shown that the intracellular Ca 2ϩ level of DG75 cells in the absence of ionomycin stimulation is sufficient for potentiation of phorbol ester-activated transcription from a granulocyte-macrophage colony-stimulating factor promoter by the Ca 2ϩ /calmodulin-dependent phosphatase calcineurin (22). Thus, higher calmodulin levels would likely lead to an increase in the level of Ca 2ϩ -loaded calmodulin in the cell.
Calmodulin can accumulate rapidly in the nucleus upon an increase in intracellular Ca 2ϩ (23,24), presumably through the action of nuclear calmodulin-binding proteins that preferentially sequester the Ca 2ϩ -loaded form (24,25). In some neurons calmodulin is primarily nuclear. In these cells, CREB (cAMPresponse element-binding protein) phosphorylation by Ca 2ϩ / calmodulin-dependent kinase IV is not dependent on the nuclear import of any protein component but only on increased cellular Ca 2ϩ (26). Therefore, we assume that an increased Ca 2ϩ /calmodulin effect on transcriptional activity of E-proteins can occur either through nuclear accumulation of calmodulin that is dependent on an elevated Ca 2ϩ concentration or through activation by Ca 2ϩ binding of pre-existing nuclear calmodulin. Although Ca 2ϩ stimulation by ionomycin (6) and thapsigargin (Fig. 4B) inhibited the luciferase activity from the E-box reporter in the B-cell line DG75, we did not detect any increase in nuclear calmodulin in the cell line upon either treatment (data not shown). This argues that Ca 2ϩ loading of the pre-existing amount of nuclear calmodulin upon increase in intracellular Ca 2ϩ concentration is sufficient to inhibit E-protein transcriptional activity.
Endogenous calmodulin was mainly localized in the cytoplasm in DG75 cells, whereas overexpressed calmodulin showed preferential nuclear localization (Fig. 6). The change in calmodulin distribution upon overexpression could be dependent on the high amount of calmodulin in the transfected cells. Because calmodulin is considered to be a limiting factor in cells, overexpression could lead to the binding of new nuclear targets with low affinity but high abundance.
E12 and overexpressed CaM were co-localized in the nucleus in DG75 cells (Fig. 6). In contrast to E12, the mutant m8-4-7 showed only partial sub-nuclear co-localization with overexpressed calmodulin. These data together with the differential effects of overexpressed calmodulin on transcriptional activity of wild type E12 and some of the mutants most likely suggests that nuclear calmodulin binds wild type E12 through basic domain interactions and thereby forms complexes that are unable to bind DNA and, therefore, are transcriptionally inactive. Further evidence supporting this interpretation is demonstrated in the chromatin immunoprecipitation experiment, where calmodulin overexpression resulted in decreased binding of E12 and E47, but not the calmodulin binding-deficient mutant m8-4-7, to the E-box recognition sequences in vivo (Fig. 8).
An alternative explanation for calmodulin inhibition of E12 activity could be that calmodulin expression leads to a modification or protection from modification of E12 or another protein(s) involved in the E12 transcriptional complex. Many transcriptional activators are targets for calmodulin-dependent kinases or the calmodulin-dependent phosphatase calcineurin (2,3). The transcriptional co-activator CBP is also regulated by calmodulin-dependent kinase IV (27). However, overexpression of constitutively active variants of calmodulin-dependent kinase II or IV or calcineurin did not have any significant effect on the transcriptional activity of E12 or E47 (data not shown). Furthermore, specific inhibitors of calmodulin-dependent kinases II and IV and calcineurin did not affect ionomycin-induced inhibition of the transcriptional activity of E47 (28). Collectively, these data suggest that no calmodulin-dependent change in the phosphorylation status of E12/E47 or any other component in the E12/E47 transcription complex contributes to the Ca 2ϩ regulation of the transcriptional activity of E12/E47. Furthermore, analysis of partially purified E47 from the nuclear extracts of transfected cells revealed that overexpression of calmodulin did not affect DNA binding of E47 to an E-box probe in the absence of Ca 2ϩ /calmodulin (data not shown). Therefore, we believe that calmodulin-dependent inhibition of the DNA binding activity of bHLH proteins is not mediated by protein modification. Another alternative mechanism for calmodulin inhibition of E-protein activity would be a potential activation of Id expression by calmodulin. This we find unlikely because calmodulin overexpression in DG75 cells did not lead to any increase in Id1, Id2, or Id3 transcripts in reverse transcription-PCR analysis (data not shown). Moreover, because calmodulin overexpression did not lead to a decrease in nuclear level of E12 (Fig. 1B), inhibition of E12 transcriptional activity by calmodulin occurs in the nucleus through the formation of an E12-calmodulin complex that cannot bind DNA. In addition, bound calmodulin may also prevent further protein interactions with the basic sequence region potentially important for the formation of a functional bHLH transcriptional protein complex. In this context it is notable that the positive effect of the transcriptional co-activator P300 on the transcriptional activity of E47 is mediated through the bHLH domain (29). However, the Ca 2ϩ /calmodulin sensitivity of DNA binding of E12 in vitro and transcriptional activation by E12 in vivo were affected in an indistinguishable way by amino acid substitutions (Figs. [3][4][5]. Thus, the results strongly favor inhibition of the DNA binding as the dominant if not the only mechanism of calmodulin inhibition of E-proteins. T-cell receptor stimulation could potentially inhibit E-proteins through calmodulin, since T-cell receptor activation is well known to increase intracellular Ca 2ϩ . Bain et al. (30) have shown that T-cell receptor activation leads to a decreased DNA binding activity of E47, mediated by an increased transcription of the inhibitory Id3 HLH-protein that forms a non-DNA binding complex with E47. It is nevertheless still possible that a direct Ca 2ϩ /calmodulin interaction may partially contribute to the E47 inhibition after T-cell receptor activation.
Various bHLH proteins show differential sensitivity to calmodulin (6,7). Therefore, Ca 2ϩ signaling through calmodulin could selectively lead to exclusion of certain bHLH protein dimers from E-boxes of target genes. This regulatory mechanism may have positive effects in differentiation through inhibiting E-protein dimers and, thereby, favor dimer formation of differentiation specific Class II/E-protein heterodimers. In line with this hypothesis, it was recently shown that decreased intracellular Ca 2ϩ inhibits maturation of a myoblast cell line into myotubes without affecting the level of the myogenic myf5 bHLH transcription factor (31). This indicates that elevated Ca 2ϩ could be a determinant for myogenesis through calmodulin inhibition of E-protein dimers and thereby favoring of DNA binding of myogenic bHLH protein/E-protein heterodimers. Ca 2ϩ signaling has already been linked to the positive regulation of myogenesis, since the myogenic transcription factor MEF2 is activated by Ca 2ϩ /calmodulin-dependent kinases (32). In this context it is also interesting that a domain in E-proteins, designated the REP domain, was recently identified to play a role in keeping E-protein homodimers inactive on myogenic enhancers (33).
The calmodulin binding basic sequence shows high homology between the three E-proteins. In fact, this sequence is identical in SEF2-1 and in the third E-protein HEB. Therefore, we suggest that the transcriptional activities of all E-proteins and heterodimers within the E-protein subfamily can potentially be inhibited directly by Ca 2ϩ signaling through calmodulin binding. There are a high number of potential bHLH protein dimers with different sensitivities to calmodulin that could potentially regulate differentiation and/or other cellular functions in various tissues. In this report, identification of combinations of amino acid substitutions that block calmodulin regulation of E-proteins enables future mutant analyses to determine the role of calmodulin/bHLH-protein interactions in regulation of various target genes and cellular functions.