Structural and Biophysical Characterization of the Interactions between the Death Domain of Fas Receptor and Calmodulin*

Background: Calmodulin (CaM) is recruited into the death-inducing signaling complex in cholangiocarcinoma cells. Results: CaM binds to FasDD in a 2:1 CaM:FasDD model. CaM antagonists abolish FasDD-CaM interactions. Conclusion: Data offer a structural basis for Fas-CaM interactions and mechanisms of inhibition. Significance: Elucidating the structural determinants of Fas-CaM interaction is critical to understanding the functional role of CaM in Fas-mediated apoptosis. The extrinsic apoptotic pathway is initiated by cell surface death receptors such as Fas. Engagement of Fas by Fas ligand triggers a conformational change that allows Fas to interact with adaptor protein Fas-associated death domain (FADD) via the death domain, which recruits downstream signaling proteins to form the death-inducing signaling complex (DISC). Previous studies have shown that calmodulin (CaM) is recruited into the DISC in cholangiocarcinoma cells, suggesting a novel role of CaM in Fas-mediated signaling. CaM antagonists induce apoptosis through a Fas-related mechanism in cholangiocarcinoma and other cancer cell lines possibly by inhibiting Fas-CaM interactions. The structural determinants of Fas-CaM interaction and the underlying molecular mechanisms of inhibition, however, are unknown. Here we employed NMR and biophysical techniques to elucidate these mechanisms. Our data show that CaM binds to the death domain of Fas (FasDD) with an apparent dissociation constant (Kd) of ∼2 μm and 2:1 CaM:FasDD stoichiometry. The interactions between FasDD and CaM are endothermic and entropically driven, suggesting that hydrophobic contacts are critical for binding. We also show that both the N- and C-terminal lobes of CaM are important for binding. NMR and surface plasmon resonance data show that three CaM antagonists (N-(6-aminohexyl)-5-chloro-1-naphthalene sulfonamide, tamoxifen, and trifluoperazine) greatly inhibit Fas-CaM interactions by blocking the Fas-binding site on CaM. Our findings provide the first structural evidence for Fas-CaM interactions and mechanism of inhibition and provide new insight into the molecular basis for a novel role of CaM in regulating Fas-mediated apoptosis.

Apoptosis is a strictly regulated process by which abnormal cells are removed from the body without altering the immune system or generating an inflammatory response. Inappropriate apoptosis (enhanced or diminished) is linked to many human diseases including neurodegenerative and autoimmune disorders, AIDS, and many types of cancers (1). Cells from a wide variety of human malignancies show a decreased ability to undergo apoptosis in response to various stimuli, which may contribute to the clonal expansion of cancer cells (2). The apoptotic pathway is normally initiated by cell surface death receptors, which belong to a tumor necrosis factor (TNF) super family of receptors (3)(4)(5). The cytoplasmic regions of two major receptors, Fas (also called CD95/Apo1) and TNF-receptor-1, share a homologous cytoplasmic region of ϳ80 amino acids called the death domain (DD) 4 (5,6). These receptors undergo a conformational change in response to their cognate ligands, allowing them to interact with adaptor proteins such as Fasassociated death domain (FADD), which also contains a homologous DD region (7,8). Engagement of Fas by Fas ligand (FasL) initiates a cascade of interactions that lead to activation of specific proteases called caspases. Caspase activation can be achieved through two distinct but ultimately converging apoptotic pathways, extrinsic and intrinsic (9). In the extrinsic pathway, Fas-FasL interaction leads to activation of caspase 8 (by FADD) and formation of death-inducing signaling complex (DISC). Active caspase 8 then cleaves and activates caspases 3, 6, and 7, which target cellular substrates and ultimately lead to cell death (8 -10). DISC formation and subsequent protein recruitment is a critical initial step in regulating Fas-mediated apoptosis.
There is compelling evidence that the DD of Fas (FasDD) interacts with various molecules, suggesting that Fas signaling is complex and regulated by multiple proteins (5). Previous studies have shown that calmodulin (CaM) is recruited into DISC in cholangiocarcinoma (11)(12)(13)(14)(15)(16)(17) and pancreatic cancer cells (18), suggesting a novel role of CaM in Fas signaling. It has been hypothesized that Fas-CaM interaction may affect Fas-FADD interaction and thus regulates DISC assembly and inhibits apoptosis in cholangiocarcinoma cells (14). Fas has been shown to interact with CaM in cholangiocarcinoma cells in a calcium-dependent manner; the amount of CaM recruited into DISC has increased upon Fas stimulation (11). Elevated levels of CaM have been linked to diseases characterized by pathological, unregulated cell growth, such as cancer (19). CaM is a ubiquitous and highly conserved calcium-binding protein expressed in all eukaryotic cells and is implicated in a variety of cellular functions (20 -25). Binding of Ca 2ϩ to CaM triggers major structural rearrangements in the N-and C-terminal lobes resulting in the opening of large binding pockets on the surface of each domain consisting of hydrophobic residues that are essentially buried in the apoprotein (21,24,25). The CaM protein has a "dumbbell-like" architecture with the N-and C-terminal lobes (CaM-N and CaM-C, respectively) connected by a flexible helix called the central linker. CaM-N and CaM-C each possess two helix-loop-helix motifs called "EF-hands" (26 -28).
Based on mutagenesis studies, it was suggested that the CaM-binding site in FasDD is located between residues 215 and 238 (numbered 231-254 in that study) (13). The NMR structure of Fas shows that these residues form helices I and II (␣1 and ␣2) (29). Mutations that may alter the structure and/or function of Fas have been detected in many types of cancers (30,31). In particular, mutation of Val-238 to asparagine (V238N, named V254N in other studies (13,32)) is analogous to the identified mutant allele of Fas in lpr-autoimmune mice that have a deficiency in Fas-mediated apoptosis (33). Although it has been shown that the Fas V254N mutant reduced CaM binding (13), it is not yet known whether residue Val-238 is directly involved in CaM binding or if it exerts an allosteric effect that affects Fas binding to CaM.
Several molecules with distinct chemical structures have been found to inhibit CaM-mediated processes (12,14). N-(6-Aminohexyl)-5-chloro-1-naphthalene sulfonamide (W7), the anti-psychotic trifluoperazine (TFP), and the anti-estrogen tamoxifen (TMX) induce apoptosis through a Fas-related mechanism in cholangiocarcinoma and other cancer cell lines (12,14,34). CaM antagonists also inhibit tumor cell invasion in vitro (35) and metastasis in vivo (36), suggesting that these are promising chemotherapeutic agents for malignancies. The molecular mechanism by which CaM antagonists manipulate the Fas signaling pathway is still unclear.
In summary, the Fas-CaM interaction appears to be an inhibitory component of DISC and may play a vital role in obstruction of caspases activation. Elucidation of the structural determinants of Fas-CaM interaction and mechanism of inhibition will be critical to understanding the precise molecular mechanism of Fas-mediated apoptosis, which may help in the development of new anticancer therapeutic strategies.
Here we employ NMR and biophysical techniques to identify the structural determinants of FasDD-CaM interactions. We show that CaM binds directly to FasDD with a dissociation constant of ϳ2 M and 2:1 CaM:FasDD stoichiometry. Our NMR data show that both of the N-and C-terminal lobes of CaM are important for FasDD binding. In addition, we show that CaM antagonists block interaction with FasDD, providing a structural basis for their role in the activation of Fas-mediated apoptosis.

EXPERIMENTAL PROCEDURES
Sample Preparation-A plasmid encoding full-length (amino acids 1-148) Norvegicus rattus calmodulin was a kind gift from Dr. Madeline Shea (University of Iowa). The Rattus CaM protein sequence is identical to that of human CaM (Swiss-port code: P62158). Plasmids encoding for CaM-N (residues 1-80) and CaM-C (residues 76 -148) were constructed using the pT7-7 vector. CaM protein expression and purification were conducted as described (37). CaM-N and CaM-C proteins were expressed as described for CaM (37) and purified by ion exchange (Q-column) and gel filtration chromatography methods. CaM samples were stored in a buffer containing 50 mM HEPES or Tris at pH 7.0, 100 mM NaCl, and 5 mM CaCl 2 .
A vector (pET28) harboring the FasDD gene encoding for residues 191-335 fused to a His 6 -SUMO tag on the N terminus was kindly provided by Dr. Jay McDonald (University of Alabama at Birmingham). Consistent with the NMR studies of FasDD, we used the numbering of FasDD amino acids as described (3,29). Thus, we numbered residues 191-335 as 175-320. This plasmid was used to construct two new Fas plasmids encoding for residues 205-305 (Fas wt ). The first plasmid encodes for an N-terminal His 6 -SUMO tag fused to Fas wt in a pET28 vector. The second plasmid contains Fas gene encoding for residues 205-305 cloned into the pET-11a vector at its NdeI and BamHI sites in-frame with the C-terminal His 6 tag (His 6 -Fas). The His 6 -Fas clone was used to make Fas D244A mutant (Fas monomer (Fas m )) via QuikChange site-directed mutagenesis.
FasDD proteins were expressed in Escherichia coli BL21 (DE3) codonϩ RIL cells. To make uniformly 15 N-and 13 C, 15 Nlabeled Fas samples, cells were grown in 4 liters of LB media at 37°C until the A 600 was ϳ0.6. Next, cells were spun down and washed with 1ϫ M9 salt before transferring them to 1 liter of M9 minimal media containing 15 NH 4 Cl and/or [ 13 C]glucose as the sole sources to produce 15 N-and/or 13 C-labeled proteins, respectively. Cells were induced with isopropyl ␤-D-1-thiogalactopyranoside, grown at 37°C for ϳ12 h, spun down, and stored overnight at Ϫ80°C. Cells were then lysed by sonication methods. The FasDD proteins were purified by nickel or cobalt affinity chromatography (Thermo Scientific) and gel filtration chromatography (GE Healthcare). For His 6 -SUMO-Fas wt , the His 6 -SUMO tag was cleaved via SUMO protease, and the Fas wt protein was subsequently purified by gel filtration chromatography. Fas m protein was stored in 50 mM HEPES or Tris (pH 7.0), 100 mM NaCl, and 5 mM CaCl 2 , whereas Fas wt was stored in 50 mM MOPS (pH 6.5), 100 mM NaCl, and 5 mM CaCl 2 . Because of the high tendency of Fas wt to form large and insoluble aggregates, the protein was kept at very low concentrations (ϳ10 M). Protein molecular weights were confirmed by electrospray ionization mass spectrometry. Because of the zero extinction coefficient for CaM-N, protein concentrations were measured using bicinchoninic protein assay (Thermo Scientific). TFP, W7 and TMX were purchased from Sigma. TFP and W7 were dissolved in 20:80 DMSO:H 2 O at concentration of ϳ20 mM and TMX was dissolved in 100% DMSO at concentrations of ϳ0.1 mM.
Gel Filtration Assay-The mobility of Fas wt , Fas m , and CaM and their complexes was analyzed by gel filtration assay. Briefly, 0.5-2 ml of CaM, Fas wt , Fas m , or their complexes (ϳ20 -400 M) were run through a HiLoad 16/60 Superdex 75 column (GE Healthcare) in a buffer containing 50 mM MES or MOPS (pH 6.5-7.0), 100 mM NaCl, and 5 mM CaCl 2 . Protein fractions were analyzed by SDS-PAGE. Molecular weight calibration kits (GE Healthcare) were used to determine the approximate molecular weights of loaded proteins.
Analytical Ultracentrifugation Experiments-Sedimentation velocity measurements were collected on a Beckman XL-I Optima system equipped with a 4-hole An-60 rotor (Beckman Coulter). Protein samples were prepared in 50 mM HEPES, 100 mM NaCl, and 5 mM CaCl 2 . Loading concentrations were at 60 and 20 M for CaM and Fas m , respectively. Three CaM-Fas m complexes have been prepared at 0.5:1, 1:1, and 2:1 CaM:Fas m stoichiometries with Fas m concentration set constant at 20 M. Rotor speed was set at 40,000 rpm (20°C), and scans were acquired at 280 nm. Partial specific volumes (v-bar) and molar extinction coefficients were calculated by using the program SEDNTERP, and buffer densities were measured pycnometrically. Sedimentation velocity data analysis were performed by using SEDFIT (38 -41).
Isothermal Titration Calorimetry (ITC)-ITC experiments were conducted on protein samples in a buffer containing 50 mM HEPES (pH 7.0) or MOPS (pH 6.5), 100 mM NaCl, and 5 mM CaCl 2 . Thermodynamic parameters of CaM binding to Fas wt and Fas m were determined using an Auto-iTC 200 microcalorimeter (MicroCal Corp.). CaM at 410 -430 M was titrated into the cell sample containing 17 M Fas m or Fas wt . In the reciprocal ITC titrations, Fas m at 400 M was titrated into the cell sample containing 80 M CaM. Heat of reaction was measured at 35°C for 19 injections. Heat of dilution was measured by titrating CaM or FasDD proteins into buffers under identical conditions. Data analysis was performed using the Microcal Origin package. Base-line corrections were performed by subtracting heat of dilution from the raw Fas-CaM titration data. Binding curves were analyzed, and dissociation constants (K d ) were determined by nonlinear least-square fitting of the baseline-corrected data. The formula used to fit the data as single set of identical sites is, where ⌬Q(i) is the heat released at ith injection, Q(i) is the total heat content of the solution, dV i is injection volume, and V o is total volume. NMR Spectroscopy-NMR data were collected at 35°C on a Bruker Avance II (700 MHz 1 H) spectrometer equipped with a cryogenic triple-resonance probe, processed with NMRPIPE (42), and analyzed with NMRVIEW (43). All NMR samples were prepared in a buffer containing 50 mM Tris-d11 (pH 7.0), 50 mM NaCl, and 5 mM CaCl 2 . Sample concentrations used for NMR titration data were at 100 M. 1 H, 13 C, and 15 N NMR chemical shifts for CaM have been reported (44,45). Fas m back-bone signals were assigned using standard triple resonance (HNCA, HNCO, HNCOCA, HNCACB, HNCOCACB) and 15 N-edited HSQC-NOESY and HSQC-TOCSY (two-dimensional total correlation spectroscopy) datasets collected on ϳ0.5-0.8 mM samples (see Refs. 46 and 47 and citations therein).
Surface Plasmon Resonance (SPR) Experiments-SPR experiments were performed on a BIAcore 2000 system (GE Healthcare) at 20°C using HBS running buffer (10 mM HEPES (pH 7.4), 150 mM NaCl, 0.005% surfactant P20, and 5 mM CaCl 2 ). CaM was immobilized by amine-coupling chemistry on CM5 sensor chip (GE Healthcare) that had been activated with a 1:1 mixture of 0.1 M N-hydroxysuccinimide and 0.1 M 3-(N,Ndimethylamino)propyl-N-ethylcarbondiimide. Surfaces were blocked by the injection of 1 M ethanolamine for 7 min and regenerated by injecting HBS running buffer containing 10 mM EDTA for 1 min at a flow rate of 20 l/min. Fas wt or antagonist binding to CaM were evaluated at increased concentrations (0.5-20 M) at a flow rate of 15 l/min. Experiments were performed in duplicate. Results were analyzed using BIAevaluation software Version 3.2 (GE Healthcare), and the response curves were globally fit to a one-site Langmuir binding model.

RESULTS
Mutants Inhibit Fas Self-association-Structural studies of FasDD and characterization of its interactions with other proteins have been hampered by the high propensity to self-associate and form large aggregates in solution (8,29). Previous studies have shown that solubility of FasDD is as low as ϳ20 M at pH 7 (29). Consequently, the NMR structure was determined at pH 4 ( Fig. 1) (29). Mutagenesis studies have shown that selfassociation of FasDD is mediated by several charged residues localized in helix ␣3 (29). These include Asp-244, Glu-245, and Lys-247. Recent solution studies on FasDD and its interaction with FADD were conducted on a mutant (K247A) that inhibits FasDD aggregation (48). NMR data have also shown that D244A mutant inhibits FasDD self-association without altering the Fas structure (29,49). For initial characterization of its binding to CaM, we have made a protein construct of wild type FasDD encompassing residues 205-305 (Fas wt ). Consistent with previous studies (29), solubility of Fas wt was very low (ϳ20 M). Because of these complications, we have made a D244A mutant (Fas m ) and examined the effect of this mutant on CaM binding by gel filtration, ITC, and NMR methods.
Mobility Characteristics of FasDD-CaM Complexes as Determined by Gel Filtration Assay-We first analyzed the solution properties of Fas wt , CaM, and Fas wt -CaM complex by gel filtration mobility assay. Samples were run on a size exclusion column (Superdex 75) under similar buffer conditions. Fas wt protein concentration was at ϳ20 M. As shown in Fig. 2A, the elution volumes of Fas wt and CaM were at 76 and 65 ml, respectively. Interestingly, we found that the elution volume of the Fas wt -CaM complex is dependent on the stoichiometry of the proteins. At a 1:1 molar ratio, the complex eluted at 63.5 ml, whereas a complex formed by mixing a slight excess of CaM (1.5:1 CaM:Fas wt ) eluted at 59 ml ( Fig. 2A). Despite the presence of excess amount of CaM, a peak indicating free Fas wt was also observed at 76 ml. At 2:1 CaM:Fas wt , the elution volume of the complex had changed to 56 ml, suggesting formation of a larger molecular weight complex ( Fig. 2A). A further increase of CaM concentration (2.5:1 CaM:Fas wt ) led to only a minor change of the elution volume to 55 ml (Fig. 2B, red). No further changes in the elution volume of the complex were observed at higher ratio, indicating complete formation of the complex at molar ratio 2-2.5:1 CaM:Fas wt . The observation of a single peak for the complex with a variable elution volume may suggest an equilibrium between 1:1 and ternary CaM-Fas complexes.
One possible explanation of the results described above is that Fas wt self-associates and forms high order complexes. This has been shown previously for Fas wt -FADD complexes (8,10). This scenario was examined by characterizing CaM interaction with Fas m . Gel filtration data show that the elution volumes of free Fas m and Fas wt are identical (Fig. 2B), indicating that Fas wt does not self-associate at 20 M. A complex formed between CaM and Fas m at 2.5:1 also eluted at 55 ml, indicating that both Fas wt and Fas m form similar complexes with CaM and that D244A mutation has no detectable effect on CaM binding. Fractions of the complexes were collected and analyzed by SDS-PAGE. As shown in Fig. 2C, two bands representing CaM and FasDD proteins were observed. The band representing the CaM protein is more intense than that of Fas wt or Fas m , indicating that CaM is more abundant in both complexes. A gel filtration mobility assay with known protein standards revealed that the molecular mass of unbound Fas m or Fas wt is ϳ14 kDa, consistent with a monomer. The migration behavior of CaM (ϳ27 kDa) is attributed to its elongated dumbbell shape (45,50). Both CaM:Fas wt and CaM:Fas m complexes with 2.5:1 molar ratio migrated as ϳ45-48-kDa complexes, an apparent molecular mass that is significantly higher than a 1:1 complex (Fig. 2D). At a 1:1 molar ratio, the calculated molecular mass of the CaM-Fas m complex was ϳ35 kDa, an average value of the molecular masses of 1:1 and 2:1 CaM:Fas m complexes, suggesting an equilibrium between the 1:1 and 2:1 complexes. Altogether, our gel filtration results provide compelling evidence for direct binding of CaM to FasDD and indicate that formation of the complex is dependent on stoichiometry and that two molecules of CaM probably bind to FasDD. We also show that D244A mutation has no effect on the formation of the CaM-FasDD complex.
Sedimentation Studies of FasDD-CaM Interactions-Gel filtration results provided insights on the mobility of the Fas m and CaM proteins as well as their complex. Here, we examined the oligomerization and equilibrium properties and stoichiometry of binding using analytical ultracentrifugation. As shown in Fig.  3, sedimentation velocity profiles of Fas m and CaM exhibited a single sedimentation boundary. Analysis of the data using SEDFIT yielded peaks at 1.4 and 1.9 S for Fas m and CaM, respectively. Molecular weight distribution analysis gave 11 and 17 kDa for Fas m and CaM, respectively. The sedimentation velocity data obtained on 3 sample mixtures of the CaM-Fas m complex with variable molar ratios (0.5:1, 1:1, and 2:1 CaM: Fas m ) are shown in Fig. 3. Sedimentation velocity profiles of the complex exhibited two sedimentation boundaries, of which one represented unbound Fas m at 1.4 S and a second represented the complex at 3 S. Upon increasing the ratio of CaM, the intensity of the peak representing the complex was increased, whereas the intensity of the peak denoting the unbound Fas m was decreased. The apparent molecular mass for the CaM-Fas m complex was calculated to be ϳ42 kDa, consistent with the formation of a 2:1 CaM-Fas m complex. These results have been recapitulated for the CaM-Fas wt complex (data not shown).
Thermodynamic Properties of CaM Binding to FasDD as Detected by ITC-The two hydrophobic regions on the N-and C-terminal lobes of CaM have been widely implicated in CaM cellular activity (24). Additional electrostatic interactions have also been shown to stabilize CaM-protein interactions (24,51,52). The ITC method is widely used to assess the relative contribution of hydrophobic versus electrostatic factors. ITC data provide values for K d , stoichiometry (n), and enthalpy change (⌬H°). The K d value is then used to calculate the change in Gibbs energy (⌬G°), which together with ⌬H°allows the calculation of the entropic term T⌬S°. The entropic and enthalpic components of the free energy reveal the nature of the forces that drive the binding reaction.
Thermodynamic parameters of CaM binding to Fas wt and CaM-Fas m Interactions as Detected by NMR Spectroscopy-The binding sites in CaM and FasDD can be identified through NMR resonance perturbations as detected in two-dimensional 1 H, 15 N HSQC (heteronuclear single quantum coherence) spectra. These experiments allow for identification of residues that are involved in the interaction and/or accompanying conformational changes and provide an effective method for examining the folding of the protein. Because of the high propensity of Fas wt to self-associate, which often complicates the NMR studies, and because mutation of Asp-244 does not affect CaM binding, we conducted our NMR experiments with the Fas m protein. Consistent with previous studies, mutation of Asp-244 to alanine has increased solubility of Fas by 10 3 -fold to Ͼ2 mM. We assessed whether FasDD undergoes significant conformational changes upon substituting a single amino acid (D244A). Asp-244 is located at the center of ␣3 (Fig. 1). 1   13 C, and 15 N NMR signals of Fas m have been assigned using standard triple resonance methods (see "Experimental Procedures"). NMR chemical shifts of Fas m were very similar to those described for Fas K247A (48) except for residues in the proximity of the mutation site. No new intramolecular NOEs were detected that would be indicative of an altered protein conformation. Three-dimensional 15 N-edited HSQC-NOESY data of Fas m confirmed that the helical character of ␣3 is retained (data not shown). These NMR observations indicate that D244A mutation inhibits FasDD self-association without altering the helical character of ␣3 or the overall structure of the protein.

H,
To identify the binding interface, we first conducted NMR titrations on a uniformly 15 N-labeled Fas m (100 M) as a function of added unlabeled CaM (Fig. 5). The addition of substoichiometric amounts of unlabeled CaM to Fas m (0.5:1) led to a decrease in intensity for the majority of 1 H, 15 N resonances (Fig. 5A, red). A steady decrease in intensity for the vast majority of the 1 H, 15 N signals in the HSQC spectra was clearly observed with further addition of CaM (Fig. 5A, green). At 2.5:1 CaM:Fas m , the vast majority of 1 H, 15 N resonances disappeared, but many new others were observed (Fig. 5B). No further changes in the HSQC spectrum were observed above 2.5:1 CaM:Fas m .
Severe broadening and/or loss of NMR signals can be attributed to one or more of the following events; first, an intermediate chemical exchange between the free and bound states of Fas m . To determine if this was the case, we collected a set of two-dimensional HSQC data on a 15 N-labeled Fas m bound to CaM at different temperatures (15-40°C). No major improvement in signal intensity in the HSQC spectra was observed as a function of temperature. Second, the line-width (short transverse relaxation time, T2) of signals was increased as a result of increasing the molecular size of the CaM-Fas m complex. This is the mostly likely factor as the complex is considered large (ϳ45 kDa) and tumbles significantly slower than unbound proteins. Third, binding of CaM induces a conformational change, leading to alteration or unfolding of the tertiary structure of Fas m . In the two-dimensional HSQC spectrum of CaM:Fas m at 2.5:1 (Fig. 5B), about 40 of the new resonances appeared strong and are clustered in a narrow chemical shift range, which may suggest that a portion of the Fas m protein became unfolded upon binding to CaM. The alternative explanation is that these signals correspond to residues localized in highly dynamic regions or loops in the Fas m protein. The number of new amide resonances is significantly higher than the number of residues localized within flexible regions (25 residues) in the free FasDD protein (PDB ID 1DDF), which may suggest that CaM induced a conformational change in FasDD, rendering a portion of the protein more flexible.
Next, we sought to identify the FasDD binding region on CaM. We collected the reciprocal HSQC experiment by which a uniformly 15 N-labeled sample of CaM was titrated with unlabeled Fas m (Fig. 6). NMR signal assignments of CaM are described elsewhere (44,45). The addition of substoichiometric amounts of Fas m has led to dramatic changes in the HSQC spectra of CaM. A significant number of 1 H, 15 N resonances  have disappeared, and many others have become very broad. Changes in the HSQC spectra ceased at 0.5:1 Fas m :CaM, indicating complete complex formation at this molar ratio (Fig. 6). This result is consistent with formation of a 1:2 Fas m :CaM complex. As evident from the NMR titration data, 1 H, 15 N resonances that exhibited substantial chemical shift changes correspond to residues spread throughout the CaM protein. These results suggest engagement of a wide interacting interface or induction of significant conformational changes in CaM upon binding to Fas m .
Both N-and C-terminal Lobes of CaM Are Required for FasDD Binding-To identify which domains of CaM are critical for FasDD binding, we obtained ITC data on Fas m as titrated with the isolated N-and C-terminal domains of CaM (CaM-N and CaM-C, respectively). ITC data clearly show that the binding affinity is dramatically reduced upon titration of CaM-N or CaM-C into Fas m (Fig. 7). Although it was virtually impossible to fit the titration data of CaM-C due to the extremely weak binding, fitting of the CaM-N data yielded a K d of 20 M, which is 10-fold higher than that of full-length CaM. These results indicate that both N-and C-terminal lobes of CaM are important for FasDD binding.
CaM Antagonists Block Binding of FasDD-A variety of small organic molecules with distinct chemical structures interact directly with CaM and inhibit CaM-mediated processes (12). As described above, W7, TFP, and TMX were shown to induce apoptosis through a Fas-related mechanism in cholangiocarcinoma and other cancer cell lines (12,14,34). The molecular mechanism by which CaM antagonists manipulate the Fas signaling pathway is still unclear. Previous structural studies revealed that W7 and TFP bind to CaM with different modes. Whereas two W7 molecules bind to both hydrophobic pockets on the N and C termini of CaM (53), controversy still surrounds the mode of TFP binding to CaM. Three x-ray structures revealed that one, two, or four TFP molecules are capable of binding to CaM (54 -56). NMR studies have also suggested that four TFP molecules can bind to CaM (57). The binding affinity of the first TFP molecule is, however, suggested to be significantly higher than the additional molecules (54).
One aim of the current study is to examine the effect of these antagonists on CaM-FasDD interactions. To do so, we employed NMR and SPR methods to assess whether antagonists block FasDD binding to CaM. First, we performed twodimensional HSQC NMR titrations on a uniformly 15 N-labeled sample of CaM as a function of added TFP. As shown in Fig. 8A, the vast majority of 1 H, 15 N resonances exhibited substantial chemical shift changes upon the addition of TFP. Likewise, the majority of amide signals in the HSQC spectrum of CaM exhibited substantial chemical shift changes upon titration of W7 (Fig. 8B); these changes are similar to those reported previously (53), confirming that two molecules of W7 are packed in the N and C termini hydrophobic pockets of CaM. After saturating the CaM protein with TFP or W7, unlabeled Fas m was added at 1:1 molar ratio followed by acquisition of two-dimensional HSQC spectra (Fig. 8, C and D). Interestingly, no chemical shift changes were detected, indicating that TFP and W7 blocked binding of Fas m . TMX, which was also shown to induce apoptosis through a Fas-related mechanism in cholangiocarcinoma cells, is poorly soluble, which precluded studying its effect on FasDD-CaM interactions by NMR. Thus, we resorted to SPR methods to assess whether TMX also disrupts binding of CaM to FasDD. In a control experiment, injection of Fas wt over the CaM surface resulted in sensorgrams (Fig. 9) that could be globally fit to yield association (k on ) and dissociation rates (k off ) of 5.3 ϫ 10 4 M Ϫ1 s Ϫ1 and 0.0477 s Ϫ1 , respectively. Using these kinetic constants, the equilibrium dissociation constant (K d ϭ k off /k on ) for the interaction between CaM and Fas wt is 0.9 M, which is in good agreement with the binding constant obtained from the ITC data. Despite the compelling evidence that CaM binds to FasDD in a 2:1 model, the SPR data were best fit to a one-site model. It has been shown that if the k on rates for both sites are within a factor of five, one-site fitting yields the best fit (58). Injection of either TMX or TFP at increased concentrations (5-20 M) led to a significant decrease in the response units, indicating disruption of the CaM-Fas interactions. Altogether, consistent with previous studies (12,14,34), our data suggest that TFP, W7, and TMX probably induce apoptosis in cholangiocarcinoma cells by directly interfering with Fas-CaM interactions.

DISCUSSION
It is established that the Fas receptor undergoes major conformational changes upon binding to FasL, allowing its cytoplasmic DD to interact with the DD of FADD (5)(6)(7)(8). Fas-FasL engagement initiates a cascade of downstream interactions that lead to activation of caspases. In the extrinsic apoptotic pathway, Fas-FasL interaction leads to formation of DISC, a prerequisite for activation of caspases 8, 3, 6, and 7 (8 -10). There is growing evidence that CaM interacts with FasDD upon recruitment into DISC in cholangiocarcinoma cells, suggesting a novel role of CaM in Fas-mediated apoptosis (11)(12)(13)(14)(15)(16)(17). In this report we have sought to identify the molec- Typically, the CaM binding domain of target proteins is a short peptide (15-20 residues) that is hydrophobic-basic in nature and has the propensity to form an ␣-helix (22). In many of the classical CaM binding targets, hydrophobic residues usually occupy conserved positions at 1-5-10 or 1-8-14, which point to one face of the helix (22). Although these patterns are typically found in many CaM-binding proteins, other unclassified examples were also identified (22). Perhaps one of the most surprising results is the finding that two CaM molecules bind to FasDD. Analysis of the protein sequence using a web-based tool provides scores from 0 to 9 based on multiple criteria including hydropathy, ␣-helical propensity, residue weight, residue charge, hydrophobic residue content, helical class, and occurrence of particular residues (59). The most likely binding site in a protein sequence is given a score of repeated 9s. Analysis of the FasDD protein sequence using this method yielded the highest score (8 -9) for residues 224 -238 (MTLSQVKGFVRKNGV), which form helix ␣2 and a short loop. This region contains a classical 1-5-10 motif (underlined residues). A second potential CaM-binding site with a repeated score of 7 has also been detected for residues 283-297 (KKANLCTLAEKIQTI), which are localized in ␣6 and a preceding loop. This region does not contain any of the known CaM binding motifs. Based on mutagenesis studies, previous studies suggested that the CaM-binding site in FasDD is located within helices ␣1 and ␣2 and the connecting loop ( Fig.  1) (13, 60). Co-immunoprecipitation studies conducted on truncated Fas constructs also suggested a single CaM binding domain (13). However, NMR, ITC, gel filtration, and analytical ultracentrifugation data described above indicate a 2:1 CaM: FasDD model. The contradiction between the results described here and the previous mutagenesis and co-immunoprecipitation data is probably a consequence of using different methods. One possible explanation for the 2:1 model is the induction of CaM dimerization upon binding to FasDD. Non-covalent dimerization of CaM has been observed in many cases (61)(62)(63). FasDD-induced CaM dimerization, however, is ruled out based on the HSQC NMR data obtained for CaM, which revealed only one set of 1 H, 15 N resonances upon titration of Fas m (Fig. 6). Thus, the most likely scenario is that CaM binds to two regions on FasDD. The precise binding mode of CaM to FasDD is under investigation.
Mutations that may alter the structure and/or function of Fas have been detected in many types of cancer (30,31). In particular, mutation of valine 238 to asparagine (V238N, named V254N in other studies; Refs. 13 and 32) is analogous to the identified mutant allele of Fas in lpr-autoimmune mice that have a deficiency in Fas-mediated apoptosis (33). Immunoprecipitation data have shown that Fas V238N mutant has a lower binding affinity to CaM (13). Val-238, which is located at the end of helix ␣2, is an essential hydrophobic residue in the 1-5-10 motif and is probably involved in binding to CaM. In a previous NMR study, it was reported that V238N mutation caused significant chemical shift changes in 1 H, 15 N resonances, suggesting a significant change in the FasDD structure (29). Although it has yet to be elucidated whether the effect of V238N mutant on CaM-Fas interactions is direct or allosteric, analysis of the FasDD protein sequence with V238N mutation using the web-based tool yielded a repeated score of 4 for residues 224 -237 (MTLSQVKGFVRKNG) and a score of 9 for residues 283-297 (KKANLCTLAEKIQTI). These results may suggest that CaM may still bind to Fas V238N but with either a lower affinity or an altered stoichiometry. Studies are currently focused on understanding, at the structural level, the effect of V238N mutation on FasDD structure and CaM binding properties.
The CaM protein adopts a collapsed structure when bound to peptides with a 1-5-10 motif (24). Previous molecular modeling studies suggested a collapsed structure for CaM when bound to helix ␣2 of FasDD (13). As shown in the three-dimensional structure of FasDD (Fig. 1), helix ␣2 is packed against helices ␣3, ␣4, and ␣5. The accessibility of helix ␣2 of FasDD to generate a collapsed structure of CaM probably requires protein unfolding. This unfolding event also releases helix ␣6 and make it accessible to CaM. The ability of CaM to disrupt protein targets has been recently observed in studies of the interactions between CaM and the human immunodeficiency virus type-1 matrix protein (37,45,64). CaM-induced unfolding of protein targets is not unusual. Partial unfolding is considered critical for the biological function of target proteins as in enzymatically driven cleavage reactions (65).
Recent structural analysis of the FasDD and CaM proteins has revealed interesting and unexpected findings (66). Despite the absence of sequence similarity, FasDD and CaM share significant structural features. During a signaling event and upon binding to target proteins, CaM undergoes a conformational change involving bending of the central helix to form a closed form (21,22). On the other hand, FasDD undergoes a transition from a closed state to open state when it is bound to FADD (8). Thus, at the resting state CaM and FasDD are in open and closed conformations, respectively. It was hypothesized that both proteins possess a "designed mobility" or instability (66). This molecular effect is governed by a capability of the intermolecular interactions to dominate in a complex and disrupt the original structure, causing a structural transformation that leads to a signaling event that is frequently associated with a hinge motion. Thus, binding of CaM to FasDD may involve a conformational switch upon recruitment to DISC.
The current findings provide new insights into understanding the precise molecular mechanism of CaM-Fas interaction and will help in identification of the functional role of CaM in Fas-mediated apoptosis. CaM is known as a primary transducer of Ca 2ϩ -dependent signals. Elevated levels of CaM have been linked to diseases characterized by pathological, unregulated cell growth such as cancer (19). CaM antagonists have long been known as anti-proliferative agents (67) and have been identified as inhibitors of tumor cell invasion in vitro (35) and metastasis in vivo (36). We have shown that CaM antagonists (TFP, W7, and TMX) inhibit binding of FasDD, providing a potential molecular basis for the role of these antagonists in inducing Fas-mediated apoptosis in cholangiocarcinoma and Jurkat cells (12,14,34). Previous structural studies have shown that binding of TFP and W7 not only block the hydrophobic lobe(s) that are important for protein/peptide binding but also induce structural changes in CaM and likely alter the binding interface. Combined with interferon (IFN-␥), CaM antagonists have been also been shown to induce apoptosis in Fas-low apoptosis-resistant cholangiocarcinoma cells (12). CaM antagonists can also inhibit interactions with other DISC components such as FLICE-like inhibitory protein (FLIP). Overexpression of FLIP lacking the CaM binding region increased both spontaneous and Fas-stimulated apoptosis in cholangiocarcinoma cells, suggesting that FLIP-CaM interaction is important in mediating survival signals by FLIP (11,14,15,60,68).
In summary, we have provided new insights on the molecular determinants of Fas-CaM interactions and mechanism of inhibition. Molecular characterization of signaling pathways emanating from Fas is critical for identifying novel molecular targets that are crucial in switching between the death versus survival signals in response to the same ligand, which may help in the development of new anticancer therapeutic strategies.