Determinants of helix-loop-helix dimerization affinity. Random mutational analysis of SCL/tal.

Dimerization represents a key regulatory step in the function of basic helix-loop-helix transcriptional factors. In many instances tissue-specific basic helix-loop-helix proteins, such as the hematopoietic factor SCL/tal or the myogenic factor MyoD, interact with ubiquitously expressed basic helix-loop-helix proteins, such as E2A or E2-2. Such dimerization is necessary for high affinity, sequence-specific DNA binding. Previous biochemical and structural studies have shown the helix-loop-helix region to be necessary and sufficient for this interaction. In the present study, we analyzed the relative affinities of various helix-loop-helix interactions using the yeast two-hybrid system. The relative affinities of selected helix-loop-helix species for the partner protein E2-2 were as follows: Id2 > MyoD > SCL/tal. Mutants of SCL/tal with increased affinity for E2-2 were selected from a library of randomly mutated basic helix-loop-helix domains. The amino acid changes in these high affinity versions of SCL/tal introduced residues that resembled those in the corresponding positions of the Id proteins and MyoD. One of the mutants, SCL 12, also contained mutations in highly conserved residues previously thought to be necessary for dimerization. This mutant of SCL demonstrated diminished temperature sensitivity in in vitro interaction assays as compared with the wild type protein. Computational modeling of helix-loop-helix dimers provides an explanation for the increased dimerization affinity of SCL mutant 12.

Basic helix-loop-helix (bHLH) 1 transcriptional factors play a fundamental role in cell fate determination in eukaryotic organisms ranging from Caenorhabditis elegans to humans. This family of over 60 different proteins has been implicated in processes such as lineage commitment, differentiation programming, cell cycle regulation, and oncogenesis (1)(2)(3)(4)(5). Many of these activities arise from sequence-specific DNA binding by bHLH factors, followed by transcriptional activation of target genes. The structural motif shared by these proteins, the bHLH domain, mediates dimer formation as well as direct DNA contact (6 -8). The bHLH domain consists of a ϳ15-amino acid basic region, followed directly by two amphipathic helices separated by a loop. The basic domain inserts into the major groove of DNA to bind a specific half-site within the core recognition sequence CANNTG, also known as an E box. Stable DNA binding occurs only with dimeric complexes of bHLH proteins, each of which contributes a basic domain to a specific half-site. Dimerization of bHLH proteins occurs via the amphipathic helices of the helix-loop-helix (HLH) portion.
Dimerization plays a central role in regulating the function of bHLH proteins. The composition of the bHLH dimeric complexes determines which specific DNA sites will be recognized. Because each basic domain within a complex contributes a half-site specificity, combinatorial diversity of bHLH complexes expands the array of potential target sequences. In general, tissue-specific bHLH proteins, such as MyoD, SCL/tal, and MASH, poorly homodimerize and preferentially heterodimerize with the broadly expressed E bHLH proteins: E2A, E2-2, and HEB (9 -11). The E proteins, in particular E2A, may either homodimerize or heterodimerize with tissue-specific bHLH proteins. These patterns of bHLH dimer formation, with an array of tissue-specific bHLH proteins vying for common E protein partners, may account for the capacity of cells to make mutually exclusive developmental decisions (12). Another level of regulation is imposed by the dominant negative HLH proteins of the Id family (13). Id proteins contain an HLH domain without a DNA binding basic domain and preferentially heterodimerize with E proteins to form inactive complexes (14). By competing with tissue-specific bHLH proteins for limiting quantities of E protein partners, Id proteins may globally down-regulate bHLH mediated transcriptional activity (15)(16)(17).
The structural and biochemical features of HLH dimerization have not been thoroughly characterized. X-ray crystallographic studies indicate that the HLH regions dimerize in the form of a parallel four-bundle left-handed helix, with many of the dimerization contacts occurring within a hydrophobic core region (6,7). Biochemical studies indicate that for some dimers, e.g. E2A homodimers, intermolecular disulfide bond formation may be important in stabilizing interactions (18). Biochemical studies of MyoD heterodimerization with E12, using site-directed mutagenesis, indicate a role for non-conserved charged residues in helix 2 of MyoD forming ionic bonds with charged residues in helix 1 of E12 (19). Similar studies have identified a homodimerization inhibitory domain, amino-terminal to the basic domain in E12, which directs preferential heterodimerization with MyoD (20).
To further characterize the determinants of helix-loop-helix dimerization, we have employed the yeast two-hybrid system to analyze the heterodimerization of the hematopoietic bHLH protein SCL/tal with the E protein E2-2. E2-2 was chosen as the E protein partner because of its preferential formation of heterodimers with SCL as compared with E2-2 homodimer formation (11). In addition the E2-2 bHLH domain is highly homologous (95%) to that of E47, for which a crystal structure is available (6). The yeast two-hybrid system permits direct and accurate quantitation of protein-protein interactions (21) and has previously been applied toward analyzing helix-loop-helix dimerization (22). In our studies Id and MyoD proteins, as compared with SCL/tal, displayed significantly higher affinity for E2-2. The bHLH domain of SCL/tal was subjected to random mutagenesis, and mutants with normal or increased affinity for E2-2 were selected. Several of the changes in the high affinity SCL/tal mutants introduced amino acids identical or similar to those in the corresponding positions of the Id and MyoD proteins. One of the mutants, SCL/tal 12, contained non-conservative amino acid changes in the two most highly conserved positions in the bHLH family. Using an in vitro interaction assay, we found that the increased affinity of SCL 12 for E2-2 displayed a temperature dependence, manifesting at 30°C but not at 4°C. Computational modeling of SCL/tal binding to E proteins provides an explanation for the properties of SCL/tal mutant 12.

MATERIALS AND METHODS
Plasmid Constructions-The SCL/tal cDNA was provided by Ilan Kirsch (NCI, Bethesda, MD). The MyoD cDNA was provided by Harold Weintraub (Fred Hutchinson Cancer Research Center, Seattle, WA). The E2-2 cDNA was provided by Tom Kadesch (University of Pennsylvania School of Medicine, Philadelphia, PA). The parent vectors for the yeast two-hybrid analysis were kindly provided by the laboratory of Roger Brent (Massachusetts General Hospital, Boston, MA) (23). The bait plasmid pEG-E2-2 consists of a PCR-generated EcoRI-SalI fragment of E2-2, encoding amino acids 467-588, ligated into the EcoRI and SalI sites of pEG202 (23). pEG-E2-2 thus encodes a fusion protein between the DNA binding domain of LexA and the carboxyl terminus, including the bHLH domain, of E2-2. The SCL/tal prey plasmid pJG-SCL encodes SCL/tal amino acids 186 -242 (the bHLH domain) fused to the B42 acidic transcriptional activation domain in the parent vector pJG45 (23). The MyoD prey plasmid pJG-MyoD encodes MyoD amino acids 108 -163 (the bHLH domain) fused to the B42 activation domain in pJG45. For both pJG-SCL and pJG-MyoD, the PCR-generated inserts were cloned in-frame into the pJG45 parent vector as EcoRI-XhoI fragments. The Id2 prey plasmid pJG-Id2 was cloned from a HeLa cDNA library using the yeast two-hybrid screening technique; the HeLa cDNA library, in the pJG45 vector, was kindly provided by Roger Brent (23). Prey plasmids containing chimeras of mutant and wild type SCL/ tal were generated using three-way ligations, consisting of an EcoRI-SacI fragment encoding helix 1, a SacI-XhoI fragment encoding the loop and helix 2, and the pJG45 vector cut with EcoRI and XhoI. The sequences of the chimeric SCL/tal preys were verified by the dideoxy sequencing technique. For production of recombinant [ 32 P]E2-2 (amino acids 467-588), the bacterial expression construct pGEX-2TK-E2-2 was employed as we have previously described (11). For production of recombinant GST-SCLwt (amino acids 186 -242) and GST-SCLmut 12, EcoRI-XhoI inserts excised from the prey plasmids were ligated into the corresponding sites in pGEX-4T-1 (Pharmacia Biotech Inc.).
Library of Randomly Mutated SCL/tal-The segment of the SCL/tal cDNA encoding the bHLH domain (amino acids 186 -242) was amplified with the following primers: 5Ј-mer consisting of GCGAATTCGTTGT-GCGGCGTATCTTCACCAAC and 3Ј-mer consisting of CCGCTCGAGT-CACAGCTTGGCCAAGAAGTTGATATA. For a non-mutagenic control, standard PCR conditions with Taq polymerase were employed. To generate a mutational library, the PCR was performed following the guidelines of Leung et al. (24). After 30 cycles of mutagenic amplification, 1 l out of the 100-l reaction was subjected to a second 30 cycles of mutagenic amplification. Similarly, 1 l of the latter reaction was subjected to a third round of 30 cycles of mutagenic PCR. Equal quantities of PCR products resulting from 30, 60, and 90 cycles of mutagenic PCR were combined and cloned into the EcoRI-XhoI sites of the prey vector pJG45. Ligation products were transformed into XL1-blue ultracompetent cells (Stratagene) to yield 90,000 colonies. Analysis of 10 randomly picked bacterial colonies showed that 60% of the prey clones contained inserts of the correct size. Bacterial colonies were pooled, and the resultant plasmid preparation was transformed into the yeast strain YPH499 using the lithium acetate technique (25) to yield 80,000 colonies.
Yeast Two-hybrid Techniques-The yeast two-hybrid system developed in the laboratory of Roger Brent was employed (26). The yeast strains consist of EGY48 (MAT␣, ura3, trp1, his3, 6LexAop-LEU2) and YPH499 (MATa, ura3-52, trp1-⌬63, his3-⌬200, leu2-⌬1, lys2-801, ade2-101). The EGY48 strain contains six LexA binding sites replacing the upstream activating sequence of the LEU2 gene; therefore, growth on leucine-deficient media may serve as an indication of interaction between bait and prey. To create the bait strain, the LexA-E2-2 expression plasmid pEG-E2-2 was transformed into EGY48 along with a reporter plasmid, pSH18 -34. pSH18 -34 contains eight LexA binding sites upstream of the ␤-galactosidase gene (lacZ). The prey strains consist of prey plasmids transformed into YPH499. To assay for preybait interactions, the prey and bait strains were mated by coincubation in YPAD medium. The yeast were washed thoroughly and then incubated in Ϫ3 dropout medium (lacking uracil, histidine, and tryptophan) with 2% galactose and 1% raffinose. To assay the number of diploids, serial dilutions of yeast were plated on Ϫ3 dropout plates with 2% glucose. To isolate diploids with prey-bait interactions, the yeast were plated on Ϫ4 dropout plates (lacking uracil, histidine, tryptophan, and leucine) with 2% galactose and 1% raffinose. For relative quantitation of prey-bait interactions, liquid ␤-galactosidase assays were performed as described previously (23).
Western Blots-For quantitation of prey expression, equivalent quantities of yeast as judged by OD600 were subjected to Western blot analysis. Yeast grown to mid log phase in Ϫ3 dropout medium with 2% galactose and 1% raffinose were resuspended in SDS-PAGE loading buffer and boiled. Proteins resolved on 12% acrylamide gels by SDS-PAGE were electrotransferred to nitrocellulose membranes. To detect the hemagglutinin epitope-tagged prey proteins, the monoclonal antibody 12CA5 was employed.
In Vitro Interaction Assay-Production, 32 P labeling, and purification of E2-2 protein (residues 467-588) has been described previously (11). Immobilization of bacterially expressed GST fusion proteins on glutathione-Sepharose beads was performed as described previously (11). Mixtures contained immobilized GST fusion proteins and the indicated quantities of soluble [ 32 P]E2-2 in 500 l of protein interaction buffer, PIB (20 mM Tris-HCl, pH 8.0, 100 mM NaCl, 2.5 mM MgCl 2 , 0.5 mM dithiothreitol, 0.25% Nonidet P-40, 0.1 mg/ml bovine serum albumin). After 4 h of incubation at the indicated temperatures, the beads were washed repeatedly first with PIB and then with TEN (10 mM Tris-HCl, pH 8.0, 1 mM EDTA, 100 mM NaCl). Washes with TEN at room temperature were continued until minimal radioactivity in the supernatant was detectable by Geiger counter. The beads were then boiled in SDS-PAGE loading buffer and the resultant supernatant subjected to SDS-PAGE on 15% acrylamide gels. Autoradiography was carried out without intensifier screens for approximately 12 h.
Model Building of SCL/tal-Models of SCL/tal and E2-2 were initially built on an Evans & Sutherland PS390 computer graphics system with the program FRODO (27). This model was subsequently refined on a Silicon Graphics Indigo2 XZ workstation using the program O (28). This model is based upon the crystallographic coordinates of the E47 homodimer, kindly provided by Dr. Tom Ellenberger (Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, MA). Homology model building was straightforward due to the absence of any significant insertions or deletions in the amino acid sequence alignments of SCL/tal and E2-2 with E47. The most frequent rotamer conformation was selected for model building of mutant residues. Drawings of molecular models were made with the programs MOLSCRIPT (29) and RASTER3D (30).

Relative Affinities of HLH Proteins for E2-2-
To compare affinities of various HLH domains for the common partner protein E2-2, the yeast two-hybrid system was employed. In this system, the "bait" consists of the E2-2 bHLH domain produced as a fusion protein with the LexA DNA binding domain. The "preys" consist of various other HLH domains fused to the strong transcriptional activator peptide B42. The reporter genes consist of ␤-galactosidase (lacZ) and LEU2 with multiple upstream LexA binding sites. Activation of the lacZ reporter gene is measured with a quantitative, colorimetric enzyme assay; activation of the LEU2 reporter gene allows the yeast to grow on leucine-deficient medium. Alone, the LexA-E2-2 bait protein does not activate the reporter genes. However, binding of the B42-HLH prey proteins to the LexA-E2-2 bait activates both of the reporter genes. The strength of reporter gene activation, as measured by lacZ activity, directly reflects the affinity of the prey-bait interaction (21).
HLH preys, consisting of fusions of MyoD, Id2, and SCL/tal with the B42 transcriptional activation peptide, were coexpressed via yeast mating with the LexA-E2-2 bait protein.
Quantitative ␤-galactosidase assays were performed on equivalent numbers of yeast for each interaction. The relative affinities for E2-2 are shown in Fig. 1A. Using SCL/tal as a standard of comparison, MyoD has a slightly increased affinity for E2-2 (8.5-fold), and Id2 has considerably increased affinity for E2-2 (65.2-fold). These results correlate with previously published data using the Far Western blotting system, in which Id1 had greater affinity than SCL/tal for E2-2 (11). To further confirm that the results of the ␤-galactosidase assays reflected the affinities of the various preys for the E2-2 bait, as opposed to differential expression levels of the various preys, prey expression in the various strains was assayed by Western blot. Using a monoclonal antibody (12CA5) that recognizes an epitope tag common to all preys, Western blot shows roughly equivalent levels of expression of all the preys (Fig. 1B).
Screening a Library of Randomly Mutated SCL/tal for Variants with Enhanced Affinity for E2-2-To determine which residues might play a role in the enhanced affinity of Id2 and MyoD for E2-2, SCL/tal was subjected to random mutagenesis using error-prone PCR (24). SCL/tal mutants with enhanced affinity for E2-2 were selected by using the yeast two-hybrid mating/screening approach (23). Table I shows the relevant parameters in this library screening. When wild type SCL/tal was employed as the prey, 2% of diploid yeast (which contain prey, bait, and reporters) manifested growth on leucine-deficient medium. When randomly mutated SCL/tal was employed as prey, only 0.07% of diploid yeast manifested gowth on leucine-deficient medium. Therefore, an estimated 97% of the SCL/tal clones obtained by random mutagenesis have sustained mutations eliminating the capacity for dimerization with E2-2. Among those clones that retained the capacity for dimerization with E2-2, the majority displayed affinities similar to that of wild type SCL/tal. One such clone, SCL/tal mutant 4, was selected for further analysis. Three of the clones from the SCL/tal mutant pool (clones 9, 12, and 36) showed significantly enhanced affinity for E2-2 and were also selected for further study.
The relative affinities for E2-2 and the amino acid sequences of the SCL/tal mutants are displayed in Fig. 2A. SCL/tal mutant 4, with similar affinity for E2-2 as wild type SCL/tal, contained a single amino acid change, K234E in helix 2. SCL/ tal mutants 12 and 36, both with significantly increased affinity for E2-2, contain identical amino acid changes in the HLH domain and most likely derive from a common original clone. These mutants (designated SCL/tal mutant 12) contain a total of four amino acid changes, two in helix 1 (N204D and G205E) and two in helix 2 (K225E and K234E). Two of the amino acid changes in SCL/tal mutant 12, N204D and K225E, eliminate highly conserved residues in the HLH family, which normally interact to form an intramolecular hydrogen bond (6,13). SCL/ tal mutant 9, with moderately increased affinity for E2-2, contains a total of two amino acid changes, one in helix 1 (N202D) and one in helix 2 (M233I). An alignment of the high affinity SCL/tal mutants with MyoD and the Id family is shown in Fig.  2B. The residues shared in common by the high affinity SCL/tal mutants and the MyoD and Id proteins are highlighted. From this alignment it is evident that the majority of amino acid changes in the high affinity SCL/tal mutants introduce residues that are similar or identical to residues at corresponding positions in MyoD and the Id proteins.
Synergy of Mutations in Helix 1 and Helix 2 of High Affinity SCL/tal Mutants-In an attempt to determine which specific amino acid changes were responsible for the increased affinity for E2-2, chimeras between wild type and mutants of SCL/tal were analyzed for E2-2 binding. The sequences and E2-2 binding affinities of the various chimeras are shown in Fig. 3A. For SCL/tal mutant 9 neither the mutation in helix 1 (N202D) nor the mutation in helix 2 (M233I) alone can fully account for the increased affinity for E2-2 seen in the intact mutant. In fact, the relationship between the mutations in helix 1 and helix 2 of SCL/tal mutant 9 appears to be synergistic rather than additive. Similarly, for SCL/tal mutant 12 the mutations in helix 1 show a synergistic interaction with the mutations in helix 2 in augmenting the affinity for E2-2. Equivalent expression of SCL/tal wild type, mutants, and chimeras was documented by Western blot analysis (Fig. 3B). From these data we conclude that while single amino acid changes may modestly enhance  the affinity of SCL/tal for E2-2, marked enhancement of this affinity requires amino acid changes in both helix 1 and helix 2.
The Enhanced E2-2 Binding of Mutant 12 Is Only Manifested at Higher Temperatures-To verify the enhanced binding to E2-2 by SCL/tal mutant 12, in vitro protein interaction assays were performed with 32 P-labeled soluble E2-2 and immobilized GST fusion proteins. Fig. 4A (lanes 1 and 2) illustrates binding of soluble [ 32 P]E2-2 to the carboxyl-terminal half of SCL/tal (amino acids 200 -331) and absence of binding of [ 32 P]E2-2 to SCL/tal lacking an HLH domain (amino acids 255-331). At 4°C, the bHLH domain of wild type SCL/tal (lane 3) binds E2-2 with slightly increased affinity as compared with the bHLH domain of SCL/tal mutant 12 (lane 4). By contrast, when the interaction assay is carried out at 30°C (Fig. 4B), SCL/tal mutant 12 (lanes 2, 4, and 6) shows significantly enhanced binding to [ 32 P]E2-2 as compared with wild type SCL/tal (lanes 1, 3, and 5). As shown in Fig. 4B, the relative enhancement of [ 32 P]E2-2 binding by SCL/tal mutant 12 at 30°C occurs across a range of concentrations of E2-2. These results indicate a greater thermal stability for E2-2 dimerization with SCL/tal mutant 12 as opposed to E2-2 dimerization with SCL/tal wild type. DISCUSSION The Id HLH proteins are potent dominant negative inhibitors of bHLH proteins, exerting their effects through the se- FIG. 2. A, amino acid sequences of the HLH domain of wild type and mutant SCL/tal proteins. The SCL/tal mutants were selected, using the yeast two-hybrid approach, from a pool of randomly mutated HLH domains. Binding of these SCL/tal mutants to E2-2 is quantified by measuring the activation of the lacZ reporter plasmid as described in Fig. 1. The affinities are expressed relative to those of seven independent yeast colonies containing wild type SCL/tal prey. B, alignment of HLH domains from MyoD, Id proteins, wild type SCL/tal, and mutants of SCL/tal. Relative affinities for E2-2, as measured by the yeast two-hybrid system, are displayed as relative lacZ units. Relative lacZ units for each interaction were calculated from three independent ␤-galactosidase assays. The numbering of amino acid residues derives from the SCL/tal HLH domain. Amino acid residues shown in bold are those shared by the high affinity SCL/tal mutants and MyoD and Id proteins. FIG. 3. A, analysis of chimeras between wild type and mutant SCL/tal HLH domains. The indicated SCL/tal HLH domains were assayed for E2-2 binding as described in Figs. 1 and 2. B, Western blot analysis of prey protein expression. Equivalent quantities of yeast from Fig.  3A were subjected to SDS-PAGE followed by Western blot with the 12CA5 monoclonal antibody as described in Fig. 1B. questration of E proteins into inactive complexes (14). For Id proteins to function in an effective manner, they must efficiently outcompete the binding of tissue-specific bHLH proteins, such as SCL/tal, to E proteins. The significantly increased E2-2 binding affinity of Id proteins over SCL/tal, demonstrated previously and in this study, provides a mechanism for the reportedly effective antagonism of SCL/tal function by Id (11,17). Our current study also indicates a slightly increased E2-2 binding affinity of MyoD over SCL/tal. Through the isolation of SCL/tal mutants that bind E2-2 with increased affinities, comparable to those of the Id and MyoD proteins, we have identified amino acids that appear to behave as affinity determinants in HLH interactions. The amino acid alignment in Fig. 2B shows that these affinity determinants are present in several HLH proteins. Notably, these amino acids are present in both helix 1 and helix 2. Introduction of any one of the amino acids alone causes a modest increase in the affinity of SCL/tal for E2-2. Coordinate introduction of these amino acids into both helix 1 and helix 2 may lead to a marked increase in the affinity of SCL/tal for E2-2. Therefore, both helix 1 and helix 2 make significant contributions to the affinity of HLH interactions.
One of the amino acid changes in the high affinity SCL/tal mutants, G205E, introduces an acidic residue into helix 1 at a position where many other HLH proteins contain an acidic residue: MyoD and Id2 (Fig. 2B), as well as E12, E47, E2-2, HEB, Daughterless, Twist, myogenin, and MRF4. The crystallographic model of dimerization in Fig. 5A suggests that this amino acid is not close enough to E2-2 to form a bond (8.3 Å from Arg-199 of E2-2). However, previous biochemical studies clearly indicate that an acidic residue at this specific position strongly contributes to dimerization, possibly by forming an electrostatic bond with a basic residue in helix 2 of the partner protein (19). A similar effect might arise from the N202D mutation in helix 1 of SCL/tal mutant 9: Id1, Id3, E12, E47, E2-2, HEB, Daughterless, and Myc all possess an acidic residue at this position. The M233I mutation in helix 2 of SCL/mutant 9 increases the hydrophobicity at this position. Methionine has a Kyte-Doolittle index (K D ) value of 1.9, and isoleucine has a K D value of 4.5. Notably, most HLH proteins, including the Id, myogenic, E protein, and achaete-scute families, possess a highly hydrophobic residue at this position, either isoleucine (K D value of 4.5) or valine (K D value of 4.2). Structural data from x-ray crystallography indicate that the residue at this position contributes to a hydrophobic core at the dimerization interface (6). Correspondingly, we have shown that by simply changing methionine 233 to isoleucine one can reproducibly increase the affinity of SCL/tal for E2-2 by 3-fold.
In the SCL/tal mutant (mutant 12) with highest affinity for E2-2, asparagine 204 and lysine 225 are replaced by acidic residues, aspartic acid and glutamic acid, respectively. These changes represent non-conservative substitutions at the two most highly conserved residues in the entire HLH family (13). To analyze the structural consequences of these mutations, computational modeling was performed using the crystallographic coordinates of the E47 homodimer (see "Materials and Methods"). The model shows that N204 and K225 normally interact to form an intramolecular hydrogen bond, approximating helix 1 and helix 2 of SCL/tal (Fig. 5A). Our experimental data indicate that despite stringent evolutionary conservation, these residues are not required for HLH dimerization. Fig. 5 (B and C) shows a model predicting the effects of the combined N204D and K225E mutations in SCL/tal. By introducing acidic amino acids at positions 204 and 225 in SCL/tal, the intramolecular hydrogen bond between helix 1 and helix 2 is disrupted. However, the glutamic acid residue at position 225 of SCL/tal is optimally oriented to form an electrostatic bond with the guanido group of arginine 199 in helix 1 of the partner E protein.
The net result is that an intramolecular hydrogen bond is lost and an intermolecular ionic bond is gained. A functional corollary is that monomeric SCL/tal, destabilized by the loss of an intramolecular bond, becomes much less energetically favorable than the heterodimeric form which is stabilized by introduction of an additional intermolecular ionic bond. From this model, one might predict that heterodimers of E2-2 with SCL/ tal mutant 12 would show increased thermal stability, as compared with heterodimers of E2-2 with wild type SCL/tal (see data in Fig. 4). In order to confirm our hypothetical model, it will be necessary to perform direct crystallographic analyses on complexes of E2-2 with wild type and mutant versions of SCL/tal.
Our results show that the affinity of SCL/tal for E2-2 has not been maximized by natural evolution. Excessive affinity for an E protein partner may represent an undesirable property from an evolutionary standpoint. This excessive affinity might alter the functional characteristics, e.g. DNA binding, of SCL/tal-E protein complexes. Alternatively, the excessive affinity may diminish the reversibility of HLH complex formation, essen- tially locking cells into undesirable developmental pathways. In many systems of cellular differentiation, rapid dynamic alterations in the arrays of HLH complexes are required at various developmental stages (3,12). Usage of high affinity dimerization mutants of HLH proteins in these systems may provide insight into the function of normally transient, rapidly reversible HLH complexes. FIG. 5. A, ribbon model of the helixloop-helix dimerization domains of SCL/ tal and E2-2. The amino acid numbering scheme for SCL/tal is also applied to the bHLH domain of E2-2. Residues of interest are rendered as ball-and-stick figures. Within the wild type SCL/tal bHLH domain (red), the epsilon amino group of Lys 225 in helix 2 forms an intramolecular hydrogen bond with Asn-204 in helix 1. B, ribbon model for the dimerization of SCL/ tal mutant 12 with E2-2. As a result of the Lys-225 3 Glu mutation in SCL/tal mutant 12, a novel intermolecular salt bridge is formed between Glu-225 in SCL/tal mutant 12 and the guanido group of Arg-199 in E2-2. This interaction could contribute to the increased dimerization affinity of SCL/tal mutant 12. Glutamic acid residues 205 and 234 in SCL/tal mutant 12 are exposed to the solvent. C, closeup of the triad of residues: SCL/tal 204 and 225 and E2-2 199. Carbon atoms for wild type SCL/tal are depicted in black, and carbon atoms for SCL/tal mutant 12 are depicted in white. Bonds are indicated by broken lines, with bond distances in angstroms listed. Drawings were made with the programs MOLSCRIPT and RASTER3D, described in Refs. 29 and 30, respectively.