Elucidation of the Interaction of Calmodulin with the IQ Motifs of IQGAP1*

Calmodulin regulates the function of numerous proteins by binding to short regions on the target molecule. IQ motifs, which are found in over 100 human proteins, appear in tandem repeats and bind calmodulin in the absence of Ca2+. One of these IQ-containing proteins, IQGAP1, interacts with several targets, including Cdc42, β-catenin, E-cadherin, and actin, in a calmodulin-regulated manner. To elucidate the molecular mechanism by which apocalmodulin and Ca2+/calmodulin differentially regulate IQGAP1, a series of constructs of IQGAP1 with selected point mutations of the four tandem IQ motifs were generated. Mutating the basic charged arginine residues in all four IQ motifs abrogated binding of IQGAP1 to apocalmodulin, but had no effect on its interaction with Ca2+/calmodulin. Analysis of IQGAP1 constructs with point mutations in single, double, or triple IQ motifs revealed that apocalmodulin bound only to IQ3 and IQ4. By contrast to the arginine mutant constructs, mutation of selected hydrophobic residues in the IQ motifs produced an IQGAP1 protein incapable of binding either apocalmodulin or Ca2+/calmodulin. These results, which differ from the conventional model of Ca2+-independent binding of calmodulin to IQ motifs, provide insight into the complexity of the molecular interactions between calmodulin and IQ motifs.

Calmodulin is a multifunctional signaling protein that elicits myriad effects in cells by modulating the function of target proteins (1)(2)(3). A diverse array of proteins are regulated by Ca 2ϩ /calmodulin, ranging from the classic kinases (such as myosin light chain kinase and the Ca 2ϩ /calmodulin kinase family) (1) to ion channels and anthrax (4,5). The calmodulin targets have short (ϳ14 -26 amino acid residues) regions to which calmodulin binds. Although these domains exhibit little sequence conservation, many adopt an amphiphilic ␣-helical conformation (6). In addition to these Ca 2ϩ -dependent targets, proteins that bind to calmodulin in the absence of Ca 2ϩ were subsequently identified (7,8). These targets contain a sequence called the IQ motif (9). Initially described in neuromodulin and unconventional myosins (9), examination of the Pfam data base reveals IQ motifs in over 100 human proteins. The IQ motif comprises 20 -25 amino acids, with the core fitting the consen-sus IQXXXRGXXXR (where X is any amino acid) (9 -11). IQ motifs frequently appear in tandem repeats that bind multiple calmodulin molecules with highest affinity in the absence of Ca 2ϩ (12).
IQGAP1, a ubiquitous 190-kDa protein, contains several protein recognition motifs through which it interacts with targets (13,14). Proteins that bind to and are regulated by IQGAP1 include E-cadherin (15,16), ␤-catenin (15,17), Cdc42 (18 -21), and actin (19,22,23). In addition, calmodulin binds to the IQ region of IQGAP1 both in the presence and absence of Ca 2ϩ (19,24). In contrast to most IQ-containing proteins, IQGAP1 exhibits an affinity for Ca 2ϩ /calmodulin ϳ2-fold higher than for apocalmodulin (19,24). Investigation of the functional sequelae of the interaction reveals that calmodulin modulates the binding of IQGAP1 to its other targets (16,17,19,23,24). Interestingly, calmodulin attenuates some of the IQGAP1-target interactions only in the presence of Ca 2ϩ (19,23). To elucidate the molecular mechanism by which Ca 2ϩ /calmodulin and apocalmodulin differentially regulate IQGAP1 function, we have generated a series of constructs of IQGAP1 with selected point mutations in each of the four tandem IQ motifs. Analysis of the binding of these constructs to calmodulin provides insight into our understanding of the mode of interaction between calmodulin and IQ motifs.

EXPERIMENTAL PROCEDURES
IQGAP1 Plasmid Construction-A Myc-tagged human IQGAP1 in pcDNA3 vector (24) was used. Construction of IQGAP1⌬CHD (residues 35-265 deleted), IQGAP1⌬WW (residues 643-744 deleted), IQGAP1⌬IQ (residues 699 -905 deleted), and IQGAP1⌬GRD (residues 1122-1324 deleted) mutants was described previously (16,20). All deletion mutants migrated to the expected position on SDS-PAGE (see Fig. 1A). To perform site-directed mutagenesis, a PacI linker was inserted into pBluescript KS at an SspI site to produce pBluescript-PacI. An ϳ2-kilobase PacI-ClaI fragment containing the IQ region of IQGAP1 was isolated from pcDNA3-IQGAP1 and inserted into pBluescript-PacI digested with PacI and ClaI to produce pBluescript-IQ. Site-directed mutagenesis was performed with the QuikChange site-directed mutagenesis kit (Stratagene). After mutagenesis, the PacI-ClaI fragment of pBluescript-IQ was re-inserted into pcDNA3-IQGAP1 from which the wild type IQ region had been removed. The sequence of all constructs was confirmed by DNA sequencing. Plasmids were purified with a QIAprep Spin Miniprep kit (Qiagen) according to the manufacturer's instructions.
Cell Culture and Transient Transfection-COS and MCF-7 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal bovine serum in a 37°C humidified incubator with 5% CO 2 . Transient transfections of wild type or mutant IQGAP1 constructs were performed with FuGENE 6 (Roche Molecular Biochemicals) as instructed by the manufacturer. Briefly, cells were grown to 70 -80% confluence in 100-mm dishes. Five micrograms of plasmid DNA was mixed with 15 l of FuGENE 6 and added to the cells. After 24 to 48 h, cells were harvested, lysed, and processed as described below.
Binding Analysis-Mutant and wild type IQGAP1 cDNAs were subcloned into the pGEX-2T vector. Glutathione S-transferase (GST) 1 fusion constructs of wild type and the indicated mutant constructs of IQGAP1 were expressed in Escherichia coli and isolated by glutathione-Sepharose essentially as described previously (19). All GST-IQGAP1 constructs were Ͼ90% pure (data not shown). Cells were lysed in 500 l of buffer A (150 mM NaC1, 1% Triton X-100, and 50 mM Tris, pH 7.4) containing 1 mM CaC1 2 or 1 mM EGTA. Equal amounts of protein lysate were precleared with glutathione-Sepharose beads for 1 h at 4°C. Lysates were then incubated with 500 ng of GST-IQGAP1 on glutathione-Sepharose beads for 3 h at 4°C. In all cases, GST alone was used as control. After sedimentation by centrifugation, samples were washed, resolved by SDS-PAGE, and transferred to polyvinylidene difluoride membrane. The resultant Western blots were probed with anti-Myc (16) and anti-calmodulin (25) primary antibodies, followed by the appropriate horseradish peroxidase-conjugated secondary antibody and developed by enhanced chemiluminescence (ECL). Where indicated, GST fusion proteins were incubated with 2 g of calmodulin. Following SDS-PAGE, the gel was cut in half; the top portion (containing IQGAP1) was stained with Coomassie Blue and the bottom half was processed by blotting for calmodulin.
Binding to calmodulin was evaluated by calmodulin-Sepharose chromatography as described (24). Briefly, after preclearing with Sepharose beads for 1 h at 4°C, equal amounts of protein lysate were incubated with 40 l of calmodulin-Sepharose (ϳ20 g of calmodulin) (or Sepharose without calmodulin as control) and incubated on a rotator at 4°C for 3 h. The calmodulin-Sepharose was washed five times in buffer A containing either 1 mM CaC1 2 or 1 mM EGTA as appropriate and resuspended in SDS-PAGE sample buffer (20 mM Tris-HCl, pH 7.5, 2% (w/v) sodium dodecyl sulfate, 2% (v/v) ␤-mercaptoethanol, 0.01% (w/v) bromphenol blue, 0.25 M sucrose, and 2 mM EDTA). Samples were heated at 100°C for 5 min and processed by immunoblotting as described above.
Miscellaneous-Densitometry of ECL signals was analyzed with UN-SCAN-IT software (Silk Scientific Corp.). Statistical analysis was performed by Student's t test, using InStat software (GraphPad Software, Inc.). Protein concentrations were measured with the DC protein assay (Bio-Rad).

Binding of Mutant IQGAP1
Constructs to Calmodulin-Initial analysis was performed with four deletion mutant IQGAP1 constructs. As previously demonstrated with endogenous IQGAP1 (24), transfected wild type IQGAP1 bound readily to calmodulin in the absence of Ca 2ϩ (Fig. 1A). Deletion of the CHD or the WW domains from IQGAP1 did not significantly change its affinity for calmodulin. By contrast, deletion of all four IQ motifs (IQGAP1⌬IQ) essentially eliminated binding ( Fig. 1). Surprisingly, IQGAP1⌬GRD had significantly reduced binding to calmodulin, although the region deleted is more than 200 amino acids distal to the IQ domain. The reason is not known, but is presumably because of effects on the tertiary conformation of IQGAP1. The reduced binding is not caused by total disruption of IQGAP1 structure because both IQGAP1⌬IQ (17) and IQGAP1⌬GRD (data not shown) bind to ␤-catenin with an affinity similar to that of wild type IQGAP1. Interestingly, the conformational "cross-talk" between the IQ and GRD regions seems to be reciprocal. We previously observed that deletion of the IQ motifs prevents Cdc42, which binds to the GRD and adjacent residues in the C-terminal half of the molecule (13), from co-immunoprecipitating with IQGAP1 (20). Binding was specific for calmodulin as wild type IQGAP1 did not bind to Sepharose alone (Fig. 1A). Essentially identical findings were observed in the presence of Ca 2ϩ (data not shown).
Selected Point Mutations of Basic Residues in the IQ Motifs of IQGAP1 Decrease Binding to Apocalmodulin, but Not to Ca 2ϩ / Calmodulin-Because deletion of all four IQ motifs abrogated the binding of IQGAP1 to calmodulin, point mutations were generated of individual residues in each IQ motif. A prior publication (26) indicated that substitution of Gln for the conserved Arg residues in the IQ motif of Ras-GRF prevented calmodulin binding. Moreover, binding of light chains to the IQ region of scallop myosin demonstrates that the conserved Arg residues in the IQ form important hydrogen bonds (27). For these reasons, we chose to mutate to Gln the Arg residues in the IQ motifs of IQGAP1. IQ1, IQ3, and IQ4 are complete IQ motifs, whereas IQ2 is incomplete as it lacks Arg at position 11 (11) (Fig. 2). Therefore, for IQ1, IQ3, and IQ4 we mutated the two Arg (at positions 6 and 11) to Gln, whereas for IQ2 the single Arg and proximal Gln (at position 2) were replaced by Gln and Ala, respectively (Fig. 2). These constructs were termed the IQR series, because the Arg residues were mutated.
Replacement of the charged Arg residues in all four IQ motifs (IQ1,2,3,4R) essentially eliminated binding to apocalmodulin (Fig. 3). To ascertain the relative contribution of individual IQ motifs to overall calmodulin binding, analysis was also performed using constructs with point mutations in single or multiple IQ motifs. Interestingly, replacement of the Arg residues in only IQ3 and IQ4 (IQ3,4R) yielded a lack of binding essentially identical to that seen with mutation of all four IQ motifs (Fig. 3). By contrast, mutation of both IQ1 and IQ2 (IQ1,2R) resulted in binding to calmodulin that was not significantly different from that of wild type IQGAP1. The constructs with mutation of IQ3 alone (IQ3R) or IQ4 alone (IQ4R) bound calmodulin with an affinity ϳ50% of that seen with wild type After 48 h, cells were lysed in buffer containing 1 mM EGTA and equal amounts of protein were incubated with calmodulin-Sepharose or Sepharose alone. Both unfractionated lysates (Lysate) and isolated complexes (CaM-Sepharose or Sepharose) were resolved by SDS-PAGE and transferred to polyvinylidene difluoride. Western blots were probed with anti-Myc antibody (to detect the Myc-tagged IQGAP1), followed by horseradish peroxidase-conjugated secondary antibodies and developed by ECL (panel A). B, the relative amount of Myc-IQGAP1 that bound calmodulin was quantified by laser scanning densitometry and corrected for the amount of IQGAP1 in the lysate. Results are expressed relative to wild type IQGAP1 (clear bar) and represent the mean Ϯ S.E. from three independent experimental determinations. *, significantly different from wild type (p Ͻ 0.01); **, significantly different from wild type (p Ͻ 0.0001). IQGAP1, whereas mutation of IQ1 alone (IQ1R) had no effect on binding (Fig. 3). These data imply that apocalmodulin binds only to IQ3 and IQ4, with approximately equal affinity for each of the two IQ motifs. This interpretation is supported by the observation that mutation of IQ1, IQ2, and IQ4 (IQ1,2,4R) yields a construct that binds calmodulin indistinguishably from IQ4R (Fig. 3), indicating that IQ1 and IQ2 do not bind apocalmodulin.
A markedly different effect was observed with Ca 2ϩ /calmodulin. Mutation of Arg residues in one, two, three, or all four IQ motifs had no significant effect on the interaction of IQGAP1 with Ca 2ϩ /calmodulin (Fig. 4). Ca 2ϩ /calmodulin bound to IQ1,2,3,4R IQGAP1 with the same affinity as to wild type IQGAP1. The binding properties of IQ1,2,3,4R led us to rename this construct IQGAP1⅐apoCaM (Ϫ) .
Mutation of Hydrophobic Residues in the IQ Motifs of IQGAP1 Attenuates Calmodulin Binding-Because replacing the basic charged Arg residues in the IQ motifs with Gln failed to reduce binding to Ca 2ϩ /calmodulin, an alternative strategy was adopted. Based on the mechanism by which calmodulin recognizes target peptides in the presence of Ca 2ϩ , hydrophobic amino acids in the IQ motifs were targeted (Fig. 5A). The prediction was that one of the lobes of calmodulin would interact with the H-1 residues, whereas the other calmodulin lobe would bind to the H-2 residues. Therefore, selected hydrophobic residues distal to the Gln (Q) of the IQ motif were mutated to Asp (Fig. 5A). All the H-1 and the distal H-2 (at position 14) residues were mutated. These point mutant constructs are termed the QH series (H for hydrophobic). Replacement of these hydrophobic residues (at positions 5, 8, and 14) in all four IQ motifs prevented binding to apocalmodulin (Fig. 5B).
Unexpectedly, the IQGAP1 construct with all four IQ motifs mutated as described (QH) still bound to Ca 2ϩ /calmodulin, although the affinity was attenuated (Fig. 5C). Therefore, an additional point mutation was performed in the residue immediately proximal to the Gln (the "I" of the IQ). Leu-752 in IQ1 was changed to Glu, whereas the Ile in IQ2, IQ3, and IQ4 (Ile-782, Ile-812, and Ile-842, respectively) were converted to Asp (Fig. 6A). These constructs are termed the IQH series. Analysis of the IQ1,2,3,4H mutant with all four IQ motifs mutated revealed a complete absence of binding to apocalmodulin (Fig. 6B). Similarly, mutation of all four IQ motifs abrogated binding to Ca 2ϩ /calmodulin. The IQ1,2,3,4H was re- named IQGAP1⅐CaM (Ϫ) because it does not bind calmodulin regardless of whether Ca 2ϩ is absent or present. As previously observed (24), more wild type IQGAP1 bound to Ca 2ϩ /calmodulin than to apocalmodulin (Fig. 6, compare lane 1 (WT) in calmodulin-Sepharose pull-downs in panels B and C).
Binding of Calmodulin to Purified IQGAP1-To verify these findings, the ability of calmodulin to bind to purified IQGAP1 was evaluated. GST fusion proteins of wild type, IQGAP1⅐apoCaM (Ϫ) , and IQGAP1⅐CaM (Ϫ) were expressed in E. coli and their ability to bind endogenous calmodulin was evaluated. As seen with the calmodulin-Sepharose analysis, purified IQGAP1⅐CaM (Ϫ) was unable to bind calmodulin regardless of whether Ca 2ϩ was present or absent (Fig. 7A).
Moreover, consistent with the data obtained with calmodulin-Sepharose analysis, purified IQGAP1⅐apoCaM (Ϫ) was unable to interact with calmodulin in the absence of Ca 2ϩ , but readily bound to Ca 2ϩ /calmodulin. The presence of the GST-IQGAP1 construct in each sample was validated by probing blots for IQGAP1 (data not shown).
Analysis was also performed in vitro with purified proteins. Pure calmodulin was incubated with GST fusion proteins of wild type and mutant IQGAP1 in the presence or absence of Ca 2ϩ . Congruent with the previous results, IQGAP1⅐CaM (Ϫ) bound no calmodulin, whereas IQGAP1⅐apoCaM (Ϫ) bound Ca 2ϩ /calmodulin but not apocalmodulin (Fig. 7B). These data validate our findings and indicate that other proteins in the cell are not responsible for the altered calmodulin binding to the IQGAP1 mutants. DISCUSSION IQ motifs, first recognized as calmodulin-binding domains in neuromodulin (7,8), have been identified in many proteins with a diverse array of functions (10). One of these is the scaffolding protein IQGAP1. In this paper, we generated two different series of point mutant constructs of the four tandem IQ motifs of IQGAP1, namely the IQR (Arg residues mutated) and IQH (hydrophobic residues mutated) series. Analysis of these constructs reveals that apocalmodulin binds to only IQ3 and IQ4, whereas Ca 2ϩ /calmodulin binds to all four IQ motifs of IQGAP1 (note that the stoichiometry of CaM:IQGAP1 is 4:1 (23)). We previously documented that Ca 2ϩ /calmodulin and apocalmodulin bind to fusion proteins of the IQ region of IQGAP1 (19). A low affinity binding site for Ca 2ϩ /calmodulin, but not apocalmodulin, was also identified in the calponin homology domain (CHD). The CHD appears to make a small contribution to total calmodulin binding as deletion of this region did not significantly reduce the binding of IQGAP1 to calmodulin. Moreover, essentially no binding of calmodulin was detected to IQGAP1 constructs with point mutations in the IQ motifs, despite the presence of an intact CHD. X-ray and NMR structures of calmodulin bound to "classic" target peptides reveal that calmodulin is compact, with the globular domains close together, joined by a loop of the extended flexible central helix (reviewed in Ref. 6). Whereas the classic calmodulin target peptides adopt an amphiphilic ␣-helical conformation (6,28), recent evidence reveals that calmodulin (and the target peptides) can adopt a variety of conformations when interacting with different targets. For example, Ca 2ϩ /calmodulin induced dimerization of the gating domain of a Ca 2ϩ -activated K ϩ channel (4). In this structure, the calmodulin-binding domain forms an elongated dimer with a calmodulin molecule bound at each end. In addition, when bound to anthrax adenylyl cyclase, calmodulin adopts an extended conformation (5), rather than the compact conformation seen with other targets. These recent surprising findings accentuate the diversity and variability in the interaction of calmodulin with its target molecules.
Although IQ motifs were described 10 years ago (9), a detailed structure of the calmodulin-IQ complex has not yet been published. A model of calmodulin bound to an IQ motif has been constructed, using the crystal structure of the regulatory domain of scallop myosin (11,29). The highly conserved portion of the IQ motif (IQXXXR; the first residue may be Ile, Leu, or Val (10)) is the most critical region and determines both the conformation and positioning of the C-terminal lobe of calmodulin (29). The second part of the IQ motif core (GXXXR) is not well conserved and has a minor role in fixing the position of the calmodulin N-terminal lobe. IQ motifs having both parts are termed "complete," whereas those lacking the second part are termed "incomplete." This distinction is important because binding of calmodulin to complete IQ motifs does not require Ca 2ϩ (29). The C-terminal lobe of calmodulin is expected to be semi-open and the N-terminal lobe closed when bound to a complete IQ motif. By contrast, when bound to an incomplete IQ motif, the N-terminal lobe would adopt an open conformation provided that Ca 2ϩ is present (29). It is thought that apocalmodulin and Ca 2ϩ /calmodulin are likely to bind different sites in the IQ motif (11).
The results presented here complement the model. Some of our data support the scheme, whereas other observations differ from those anticipated from the model. For example, consistent with the predictions of Houdusse and Cohen (11), both hydrophobic and electrostatic interactions appear to contribute to the binding of apocalmodulin and both seem necessary; disruption of either mode of association by mutation of critical residues reduces binding. The expectation from the model was that apocalmodulin would bind all the complete IQ motifs. However, our data revealed that apocalmodulin did not bind to IQ1, although it fulfills the criteria of a complete IQ motif. As indicated by our results, caution should be exercised in extrapolating data obtained with model IQ peptides because these do not always mimic the behavior of the intact protein. For example, neuromodulin and neurogranin bind calmodulin only in the absence of Ca 2ϩ , but isolated peptides of the calmodulinbinding domains bind Ca 2ϩ /calmodulin (30 -32). Similarly, the interaction of calmodulin with the complete catalytic domain of edema factor of adenylyl cyclase was different from its interaction with the peptide of the calmodulin-binding domain; the peptide induced a conformation of calmodulin opposite to that caused by the whole catalytic domain (5). These observations emphasize the importance of studies with intact proteins as conducted here.
It is generally believed that calmodulin targets that contain IQ motifs have a higher affinity for the Ca 2ϩ -free form of calmodulin (8,12,31,33,34). Moreover, for some proteins such as brush border myosin I (35), Ca 2ϩ induces the dissociation of Complexes were pelleted by centrifugation, washed, and resolved by SDS-PAGE. A, after transfer to polyvinylidene difluoride, Western blots were probed for calmodulin and developed by ECL. B, after SDS-PAGE, the gel was cut into two pieces; the top portion (containing IQGAP1) was stained with Coomassie Blue, whereas the bottom half was transferred to polyvinylidene difluoride and probed for calmodulin. Representative experiments of two to four determinations are shown. Relative intensities of the images for Ca 2ϩ and EGTA samples cannot be compared with one another because they are from separate exposures adjusted to demonstrate the relative intensities of wild type and mutant constructs. bound calmodulin. Our mutagenesis data reveal that the interaction between calmodulin and IQ motifs is substantially more complex than current thinking and selected IQ motifs may have higher affinity for Ca 2ϩ /calmodulin. This observation is supported by the reports from both our laboratory and others that the IQ-containing proteins IRS-1 (36) and NinaC myosin (37) bind Ca 2ϩ /calmodulin with higher affinity than apocalmodulin. The results presented here extend previous findings, indicating that the binding of Ca 2ϩ /calmodulin to IQ motifs exhibits several features different from those of apocalmodulin. First, Ca 2ϩ /calmodulin binds to all four IQ motifs of IQGAP1, whereas apocalmodulin does not interact appreciably with the first two (IQ1 or IQ2) IQ motifs. These findings provide the molecular mechanism for our initial observation that 2-fold more IQGAP1 bound calmodulin in the presence of Ca 2ϩ (24). Second, electrostatic interactions appear less important for Ca 2ϩ /calmodulin binding than hydrophobic interactions. Substitution of the polar but uncharged Gln for critical basic Arg residues in all four IQ motifs did not attenuate the interaction of Ca 2ϩ /calmodulin, but was sufficient to eliminate apocalmodulin binding. Third, individual hydrophobic residues have critical roles that differ between apocalmodulin and Ca 2ϩ /calmodulin. The I of the IQ motif appears to be essential for binding Ca 2ϩ /calmodulin, but has a less significant role in the interaction with apocalmodulin. Binding to apocalmodulin, but not Ca 2ϩ /calmodulin, could be eliminated without altering this Ile.
This study employed direct experimentation to test models of the interaction of calmodulin with IQ motifs. Nevertheless, some caveats of our work should be borne in mind. Fourteen amino acids had to be substituted in the four IQ motifs of IQGAP1 to abrogate binding of Ca 2ϩ /calmodulin. It is conceivable that these substitutions could modify secondary or tertiary structure, thereby altering the orientation of the IQ motifs to one another or to other regions of IQGAP1. The observation that IQGAP1⅐apoCaM (Ϫ) and IQ1,2,3,4R IQGAP1 (which have 7 and 10 substitutions, respectively) bind to Ca 2ϩ /calmodulin with the same affinity as to wild type IQGAP1 renders this possibility less likely. It remains possible, however, that the mutant IQGAP1 could bind to calmodulin by an interaction different from that in the native protein. This premise can be eliminated only by solving the structures of wild type and mutant IQGAP1.
The functional sequelae of calmodulin binding to IQ motifs remain incompletely understood. For the unconventional myosins, IQ motifs are thought to influence the chemomechanical properties of the myosins (10). Binding of calmodulin has a substantial effect on IQGAP1, modulating its interaction with other targets (16,17,24). Previous analysis documented that Ca 2ϩ is required for calmodulin to block the binding of Cdc42 to IQGAP1 (24), implying that Ca 2ϩ /calmodulin induces a conformation in IQGAP1 different from that produced by apocalmodulin. The findings presented here contribute to our understanding of the molecular mechanism underlying these functional effects. IQ3 and IQ4 of IQGAP1 bind both apocalmodulin and Ca 2ϩ /calmodulin, whereas IQ1 and IQ2 bind only Ca 2ϩ / calmodulin. Therefore, the conformation adopted by IQGAP1 bound to Ca 2ϩ /calmodulin is likely to result from the interaction of Ca 2ϩ /calmodulin with IQ1 and/or IQ2. A second possibility is that simultaneous occupation of all four IQ motifs is necessary to eliminate Cdc42 binding. Third, Ca 2ϩ /calmodulin may interact with amino acid residues in the IQ motifs different from those recognized by apocalmodulin, inducing a different shape in IQGAP1. These mechanisms are not mutually exclusive and more than one may be operative. Together, the findings obtained by mutating individual residues in sequential IQ motifs in an intact protein enhance our comprehension of the molecular interactions between calmodulin and IQ motifs, as well as elucidating the modulatory role of Ca 2ϩ . We look forward to the crystal structures of IQ motifs bound to Ca 2ϩ / calmodulin and apocalmodulin to yield further insight into these important protein regulatory motifs.