Modulation of basic helix-loop-helix transcription complex formation by Id proteins during neuronal differentiation

In the present study, we analyzed expression of the Id genes in relation to some tissue-specific bHLH factors that are expressed in differentiating neuroblastoma cells. We found varying levels of Id1, Id2, and Id3, but not Id4, in neuroblastoma cell lines, and the expression of these proteins was downregulated when differentiation was induced. To further elucidate the dynamics of the bHLH network, we also performed experiments to ascertain the capacity of Id proteins to interact with bHLH proteins involved in neurogenesis. The results show that the Id proteins form complexes with HES-1, both in vitro and in vivo , which demonstrates novel functions of HES-1 and the Id proteins and provides evidence of an additional level of regulation within the bHLH network. This could have implications for understanding of the formation of the neural crest and the lineage determination of neuronal cells, as well as the function of these proteins in general. Coimmunoprecipitation . (A) CHO cells were transfected with the indicated plasmids, using pEGFP-HASH-1, pBIND-HASH-1, and pBIND as negative controls. Cell extracts were immunoprecipitated with a monoclonal anti-GAL4 DNA-binding domain antibody, and the immunoprecipitated protein complexes were subjected to 10% SDS-PAGE and blotted to a PVDF filter. Crude cell extract from CHO cells transfected with pEGFP-HES-1 and pBIND-Id2 was run in a parallel lane. Coimmunoprecipitated EGFP-HES-1 fusion protein was detected by Western blotting with a polyclonal anti-GFP antiserum. (B) Coimmunoprecipitation of endogenous HES-1 and Id1 proteins from extracts of PC12 pheochromocytoma cells. The proteins were immunoprecipitated with a polyclonal anti-Id1 antiserum or with normal rabbit immunoglobulin (Ig) as a negative control. The immunoprecipitate was subjected to 12.5% SDS-PAGE, and proteins were transferred to a PVDF filter. Crude cell extract from PC12 cells was run in a parallel lane. Immunodetection was done with polyclonal anti-HES-1 or anti-Id1 antiserum, as indicated.


Introduction
network. The Id proteins are dominant-negative inhibitors since they lack the basic DNA-binding domain (8). The four mammalian Id proteins (Id1-Id4) form transcriptionally inactive heterodimers, primarily with the E proteins, and thereby prevent the E proteins from forming functional heterodimers with tissue-specific bHLH proteins (9).
Mammalian achaete-scute homologue-1 (MASH-1 in the mouse and HASH-1 in humans) is a vital bHLH protein in the developing sympathetic nervous system (SNS). MASH-1 is expressed in restricted regions of the embryonic brain and in sympathetic and enteric precursor cells (10). Gene-targeting experiments have shown that MASH-1 is needed for proper development of autonomic and olfactory neuroblasts, neuroendocrine cells of the lung, and certain regions of the telencephalon (11)(12)(13). Another tissue-specific bHLH protein is dHAND, which is expressed in the embryonic sympathetic and enteric nervous systems (14)(15)(16).
Neuroblastoma cells show characteristics of developing sympathetic neuroblasts, and we have previously studied the expression of dHAND and HASH-1, in both primary tumors and neuroblastoma cell lines. At the mRNA level, we detected expression of these two genes in all neuroblastoma cell lines and a majority of the primary tumors we analyzed (16)(17)(18). Furthermore, during induced differentiation of neuroblastoma cells, we observed rapid downregulation of HASH-1 that was accompanied by a transient upregulation of HES-1 (17). HES-1 is an important protein in the Notch-1 signaling cascade that has been shown to bind the HASH-1 promoter and thereby inhibit expression of HASH-1 (19). Accordingly, downregulation of HASH-1 induced by increased expression of HES-1 may be a prerequisite for neuronal differentiation of neuroblastoma cells (17). 5 In the present study, we analyzed expression of the Id genes in relation to some tissuespecific bHLH factors that are expressed in differentiating neuroblastoma cells. We found varying levels of Id1, Id2, and Id3, but not Id4, in neuroblastoma cell lines, and the expression of these proteins was downregulated when differentiation was induced.
To further elucidate the dynamics of the bHLH network, we also performed experiments to ascertain the capacity of Id proteins to interact with bHLH proteins

Mammalian two-hybrid analysis
Mammalian two-hybrid analyses were performed using the Checkmate system (Promega). Constructs encoding VP16 transactivating (pACT) and GAL4-DNAbinding (pBIND) proteins were created by cloning PCR-generated fragments into BamHI/SalI digested vectors. The fragments comprised the following amino acids: dHAND, 1-217; E2-2, 1-668; E47, 508-654 (spanning the bHLH region); HASH-1, 1-180; HES-1, 90-281; Id1, 1-154; Id2, 1-134; Id3, 1-119; Id4, 1-162. The pBIND-MyoD vector was provided in the Checkmate kit (Promega). The pG5luc reporter plasmid contains five GAL4 binding sites upstream of the coding sequence for firefly luciferase. The pBIND vector also encodes renilla luciferase downstream of a constitutively active CMV promoter, hence it was possible to use a Dual Luciferase kit (Promega) for both luminometric measurement of the interaction and determination of transfection efficiency. The analyses were done on CHO cells (1.5 × 10 5 per 35-mm dish), which were transfected with the appropriate vectors and the pG5luc reporter (0.6 µg of each plasmid per dish) using Lipofectamine (Life Technologies). The experiments were performed in triplicate, and the results were recorded as the relative luciferase activity (i.e., the ratio of firefly luciferase activity to renilla luciferase activity). The expression plasmids were sequenced, and expression of all proteins was verified by Western blot analysis. The mammalian two-hybrid system was also employed to assess the ability of HES-1 and Id2 to interfere with the formation of bHLH dimers in a dominant-negative manner. CHO cells were cotransfected with pACT-E2-2 and pBIND-HASH-1 or pBIND-Id2 (0.2 µg of each plasmid per dish), along with increasing amounts of pMYC-HES-1 or pEGFP-Id2 (0.2-0,8 µg of plasmid per dish). The total amount of DNA was equalized with either pMYC or pEGFP. The pMYC-HES-1 and pEGPP-Id2 were made as described above, but, for HES-1, the full-length sequence (amino acids 1-281) was used. The results are given as percentage of the initial activity.

Expression of Id proteins in neuroblastoma cell lines
The expression patterns of Id1, Id2, Id3, Id4, HASH-1, and HES-1 were

Expression of Id proteins in differentiating neuroblastoma cells
In many cell systems, Id expression decreases during differentiation (9). To study the expression pattern of HASH-1, HES-1, and the Id proteins in differentiating neuroblastoma cells, we used two different cell systems: SH-SY5Y, which lacks N-myc amplification; and SK-N-BE(2), in which each cell contains approximately 85 copies of the N-myc gene (21). The SH-SY5Y cells were induced to undergo robust and well-characterized sympathetic neuronal differentiation by exposure to 16 nM TPA (22,23), and the SK-N-BE(2) cells were induced to differentiate with 10 µM RA, which led to a less distinct neuronal differentiation (17,18). Within 2 h of TPA treatment, the level of HES-1 in the SH-SY5Y cells reached a maximum, and there was a concurrent downregulation of HASH-1 expression (Fig. 2); neither of these proteins was detected after 96 h of differentiation. In accordance with these findings, we have previously studied the corresponding genes and observed a similar rapid downregulation of HASH-1 at the mRNA level, with a concomitant transient upregulation of HES-1 (17). In differentiating SH-SY5Y cells, the Id1 level remained constant up to 24 h, but had decreased markedly after 96 h (Fig. 2). By comparison, Id2 and Id3 were downregulated more quickly: both showed a distinct decrease after only 8 h, and, after 96 h, Id3 was undetectable, and the level of Id2 was low (Fig. 2).
After 2 h of differentiation induced by 10 µM RA, the SK-N-BE(2) cells exhibited a decrease in the level of HASH-1, and this was accompanied by an increase in HES-1 (Fig. 3). After 96 h, the former protein was barely detectable, and the level of HES-1 had decreased to baseline, which reflects our previously reported mRNA data (17). The changes in Id levels that occurred in this cell line with progressing differentiation were not as marked as those observed in SH-SY5Y cells undergoing TPA-induced differentiation. Moreover, in the SK-N-BE(2) cells, the level of Id1 was virtually constant, while the level of Id2 decreased to some extent, and there was a slight transient upregulation of Id3 expression after 8 h (Fig. 3). Thus, the SK-N-BE(2) and SH-SY5Y cells were similar in regard to regulation of expression of HASH-1 and HES-1, but they exhibited slightly different Id expression patterns.

Mammalian two-hybrid analysis of interactions between Id proteins and neuronal bHLH transcription factors
We used the mammalian two-hybrid system to investigate the ability of the Id proteins to form complexes with bHLH transcription factors (24). Expression The ability of HES-1 to dimerize with both the Id and the E proteins distinguished it from the other tissue-specific bHLH transcription factors we investigated. The difference between dHAND and HES-1 in regard to interaction capacity was further illustrated by coexpression of VP16-dHAND or VP16-HES-1 with GAL4 fusion proteins of a panel of HLH/bHLH factors. We thereby generated a "fingerprint" of the abilities of dHAND and HES-1 to interact with other proteins of the HLH network ( Fig. 4C and D). The background activity of each GAL4 fusion protein was assessed in combination with the empty pACT vector. These analyses highlighted the following characteristic difference in the interaction patterns of these two proteins: HES-1 formed complexes with the E and the Id proteins, whereas dHAND dimerized only with the E proteins.

Mammalian two-hybrid analysis of dominant-negative effects of HES-1 and Id2
We extended the mammalian two-hybrid analyses to ascertain whether HES-1 and Id2 interfere with the formation of HLH dimers in a dominant-negative manner.
Increasing amounts of HES-1 had no effect on dimerization between HASH-1 and E2-2 (Fig. 5A). In contrast, Id2 decreased the reporter gene activity even at the initial concentration we tested (Fig. 5B). At this level, the EGFP-Id2 fusion protein was hardly detectable by Western blot analysis. This effect was most likely due to the formation of E2-2/Id2 dimers, since the Id proteins do not interact with HASH-1 ( Fig.   4B and data not shown). Accordingly, we investigated the effect of HES-1 on formation of the E2-2/Id2 dimer and discovered that reporter gene activation decreased with increasing amounts of HES-1 (Fig. 5C). Theoretically, this may have been due to binding of either Id2 or E2-2 to HES-1. However, HES-1 did not affect the interaction between E2-2 and HASH-1 (Fig. 5A), hence we suggest that this inhibitory effect of HES-1 was mainly the result of complex formation with Id2.
Thus, in these experiments in vivo, HES-1 sequestered Id2 and in that way prevented it from dimerizing with the E protein E2-2.

Coimmunprecipitation of Id2 and HES-1
We performed coimmunoprecipitation assays to prove that HES-1 and Id proteins can interact in mammalian cells. CHO cells were transfected with expression plasmids for EGFP-tagged HES-1 or HASH-1 and GAL4-tagged Id2 (Fig. 6A), and Western blotting revealed that EGFP-HES-1 and EGFP-HASH-1 were expressed at similar levels (data not shown). Sixteen hours after transfection, the cells were lysed, and Id2 with associated proteins were immunoprecipitated with an anti-GAL4 antibody conjugated to protein G Sepharose beads. The precipitates were subjected to Western blot analysis using a polyclonal anti-EGFP antiserum, which showed that HES-1, but not HASH-1, was coimmunoprecipitated with Id2 (Fig. 6A). Use of an empty GAL4-expressing vector showed that the Id2 part of the fusion protein was necessary for coimmunoprecipitation to occur. We included the E proteins E2-2 and E47 as positive controls in these experiments. Our findings confirmed the results obtained using the mammalian two-hybrid system, that is, both E2-2 and E47 formed complexes with HES-1, whereas there was no detectable association between HASH-1 and HES-1 or between HASH-1 and Id2 (Fig. 6A). Apparently, the amounts of HES-1 immunoprecipitated by Id2, E47, and E2-2, respectively, were to some extent correlated with the level of reporter gene activation observed in the mammalian twohybrid system, because we found that more HES-1 was coimmuniprecipitated with Id2 than with the E-proteins. Furthermore, E47 was less efficient in precipitating HES-1 than E2-2 ( Fig. 4B and 6A). It will be necessary to use other experimental techniques to determine whether these observations reflect different affinities between the various HLH proteins. We continued our work by examining the ability of HES-1 to form complexes with the Id proteins in cells derived from the sympathetic nervous system. We chose to analyze Id1 instead of Id2 due to the superior immunoprecipitating capacity of the anti-Id1 compared to the anti-Id2 antiserum. The rat pheochromocytoma cell line PC12 was used, because these cells express high levels of HES-1 and Id1. We found that the anti-Id1 antiserum did immunoprecipitate HES-1 from PC12 extracts (Fig.   6B), which confirmed the results of the coimmunoprecipitations using extracts from transfected cells. Therefore, we conclude that Id1 can complex with HES-1 in sympathetic neuronal cells.

Id2 inhibits binding of HES-1 to an N-box oligonucleotide
HES-1 and other hairy-related bHLH proteins bind to a distinct hexameric sequence called the N-or C-box (CACNAG) (6). For electrophoretic mobility shift assays, we transfected CHO cells with a VP16-tagged HES-1 expression construct and then prepared cell lysates. A prominent complex appeared when 32 P-labeled N-boxcontaining oligonucleotide was mixed with cell extract from HES-1-transfected CHO cells, whereas no such complex was detected when this oligonucleotide was mixed with extract from untransfected cells (Fig. 7A, lanes 2 and 4). The complex was supershifted when we mixed extract from HES-1-transfected cells with an anti-HES-1 serum, which confirms that this complex arose due to HES-1 DNA-binding activity (Fig. 7A, lane 3). To verify the specificity of this binding, we performed competition experiments, in which an excess of unlabeled wt or N-box-mutated oligonucleotide was added to the reaction mixtures, and found that only the wt oligonucleotide could diminish the DNA-binding of HES-1 (Fig. 7A, lanes 5 and 6). To study the effect of Id2 on DNA binding by HES-1, we pre-incubated extract from HES-1-transfected cells with Id2 translated in vitro or, as a control, with an equal amount of a lysate of mock-translated rabbit reticulocyte lysate. The results show that binding of HES-1 to the N-box oligonucleotide was decreased by the in vitro translated Id2, but not by the control mock-translated lysate (Fig. 7B, lanes 2 and 3). Extending these experiments, we mixed increasing amounts of cell extract from Id2-transfected CHO cells with a constant amount of cell extract from HES-1 transfected CHO cells, and found that binding of HES-1 to the N-box-containing oligonucleotide decreased with increasing Id2 concentration (Fig. 7C). Together, these results show that HES-1 can bind an Nbox-containing oligonucleotide, and that addition of Id2 translated in vitro or extract from Id2-transfected cells decreases this DNA-binding activity of HES-1.

Discussion
To gain a better understanding of the biological activity and function of the different Id proteins, we analyzed these proteins in regard to their expression patterns and their ability to form complexes with bHLH factors expressed in differentiating neuroblastoma cells. All the neuroblastoma cell lines we investigated expressed Id1 and Id2, and most of them also expressed Id3.
It has been reported that, in a number of model systems, Id levels are downregulated during induced differentiation, and there is concomitant cell cycle withdrawal (27). Consistent with those findings, we have previously noted that all Id genes expressed in our experiments (Id1, Id2, and Id3) were downregulated during TPA-induced differentiation and growth inhibition in SH-SY5Y cells (28), but it is not clear whether the downregulation of Id proteins is a prerequisite for the decreased proliferation. By comparison, in the present study, we detected only modest changes in Id levels in SK-N-BE(2) cells that had been induced to differentiate. The differences in Id levels may reflect the varying extent of neuronal differentiation in these two cell lines, with a more robust neuronal phenotype in the SH-SY5Y cells than in the SK-N-BE(2) cells. Considering the decreased proliferation during induced differentiation, it should be mentioned that Id proteins also have several effects on cell cycle regulatory proteins. One such effect is the direct interaction between Id2 and proteins of the retinoblastoma family (pRB, p107, and p130) (29). It is assumed that Id2 interferes with the association between pRB and E2F-DP1 by binding hypophosphorylated pRB, which results in E2F-DP1-driven transcription of genes required for S-phase progression. Thus, a possible explanation for the reduced proliferation seen in neuroblastoma cells undergoing induced differentiation is that a lower level of Id2 increases the pool of free hypophosphorylated pRB. In support of this concept, ectopic expression of the Id proteins leads to a blocked differentiation and increased proliferation in several model systems (9). However, we could not determine whether this is also the case in neuroblastoma cells, because they died upon overexpression of Id proteins (data not shown). Of specific interest for the genesis of neuroblastoma is a study indicating that Id2 is a transcriptional target of N-myc (30).
Since N-myc amplification, and thereby dysregulated expression, is a cardinal feature of high stage neuroblastoma tumors these findings could shed light on some puzzling features of neuroblastomas, such as the lack of mutations in the pRB pathway (31).
Within the bHLH network, the Id proteins exert their effect primarily by binding to E proteins. This was readily revealed by the mammalian two-hybrid analyses, which showed that all four Id proteins interacted with E47 and E2-2 ( Fig.   4B and data not shown). We observed that the Id proteins interacted with MyoD, as also reported by other investigators (8,26), but not with the proneuronal bHLH proteins HASH-1 and dHAND. These data corroborate the concept that the Id proteins act by sequestering the E proteins, thereby preventing complex formation between proneuronal bHLH proteins and their obligatory protein dimerization partners (4). Furthermore, we found that HES-1 interacted with all four Id proteins, and the difference in interaction pattern between HES-1 and the proneuronal bHLH protein dHAND was clearly demonstrated by generating mammalian two-hybrid interaction "fingerprints" (Fig. 4C and D). Whereas dHAND interacted solely and specifically with the E proteins, HES-1 formed complexes with both the E and Id proteins. Although the level of reporter gene activation in the mammalian two-hybrid system does not necessarily reflect the exact strength of the interaction, our coimmunoprecipitation data support the suggestion that the association between HES-experiments of endogenous proteins showed that a considerable fraction of HES-1 in PC12 cells is complexed with Id1 (Fig. 6B), which is the predominant Id protein in this SNS-derived cell line. This clearly suggests that the interaction influences the activity of both proteins in vivo.
We extended the mammalian two-hybrid analyses to include a third factor that was expressed together with the GAL4 and VP16 fusion proteins. We reasoned that if this additional protein bound either of the two fusion proteins, we would detect decreased reporter gene activation. Indeed, introduction of increasing amounts of Id2 to cells with a transactivating HASH-1/E2-2 complex showed that Id2 had a dominant-negative effect on the HASH-1/E2-2 reporter gene activation (Fig. 5B). We also found that HES-1 can interfere with the dimerization between Id2 and E2-2, but not that between HASH-1 and E2-2, which suggests that HES-1 in vivo can modulate the dominant-negative effect that Id2 has on an E protein without directly interfering with dimerization between HASH-1 and E2-2 (Fig. 5). This reveals a novel level of regulation within the HLH network: in an in vivo situation in which the four proteins compared in this experiment are present, Id2 may act as a dominant-negative regulator of the HASH-1/E2-2 complex, and HES-1 will, in turn, act as a negative regulator of Id2, without affecting the association between HASH-1 and E2-2. Our results may also have implications beyond the bHLH network, since HES-1 might interfere with other functions of the Id proteins, such as the interaction between Id and pRB (29), and in that way directly affect cell proliferation. Furthermore, Id2 decreased the binding of HES-1 to an N-box-containing oligonucleotide (Fig. 7C).
Thus, Id2 can sequester HES-1 and prevent it from binding DNA in a manner similar to the dominant-negative effect that the Id proteins exert on the E proteins. Mutual regulation of HES-1 and Id proteins may also be important in other tissues and cell types in which involvement of HES-1 has been implicated, for instance the developing CNS (39), β cells of the pancreas (40), and tumors such as small cell lung cancer (41). During differentiation in SH-SY5Y neuroblastoma cells, HES-1 is transiently upregulated before the Id proteins are downregulated (Fig. 2).
Consequently, it is possible that differentiation is initiated by the short-lived upregulation of HES-1 and that the subsequent downregulation of Id proteins is necessary during later stages of differentiation and for cessation of proliferation. In     factors. The degree of interaction is given as the "relative luciferase activity", which was calculated as the ratio of firefly luciferase activity to renilla luciferase activity.
The interaction patterns obtained in experiments with Id2, Id3, and Id4 (data not shown) were identical to the pattern for Id1. The Western blot panel shows the expression levels of the VP16 fusion proteins. The mammalian two-hybrid system was used to analyze VP16 fusion proteins of either HES-1 (C) or dHAND (D) regarding their ability to form complexes with a panel of GAL4-fused HLH proteins.
The shaded columns represent the relative luciferase activity recorded when pACT-HES-1 or pACT-dHAND was cotransfected with different pBIND-HLH plasmids.
The unshaded columns represent the background relative luciferase activity detected when the various pBIND-HLH vectors were cotransfected with empty pACT vector.
Expression of GAL4 fusion protein was analyzed by Western blotting (C).