NMR Solution Conformations and Interactions of Integrin αLβ2 Cytoplasmic Tails*

The integrins are bi-directional signal transducers. Devoid of enzymatic activity, the integrin cytoplasmic tail serves as a hub for the recruitment of cytosolic proteins, and many of these are signaling molecules. The leukocyte-restricted integrin αLβ2 is essential for the adhesion, migration, and proliferation of leukocytes. Here we report solution conformations and interactions of the αLβ2 cytoplasmic tails by NMR analyses. The αL tail is characterized by three helical segments in the order of helix 1-3 that are connected by two loops with helix 3 having a number of nuclear Overhauser effect contacts with helix 1 and helix 2. The conformation of the β2 tail is less defined with only a helical segment restricted at its N terminus. Acidic residues from the helix 2-loop-helix 3 motif of αL were found to be responsible for its binding to calcium ion. There were detectable interactions between αL and β2 tails, involving helix 1 and helix 3 of the αL tail and the N-terminal helix of the β2 tail. Talin head domain that contains the FERM domain showed binding affinity of Kd ∼ 0.5 μm with the β2 tail. The binding affinity of αL and β2 tails is Kd ∼ 2.63 μm. These data are in line with the activating property of talin head domain on αLβ2 by which binding of talin head domain to β2 tail disrupts the interface of the αL and β2 tails that constrains αLβ2 in a resting state.

Integrins are cell adhesion molecules that are essential for maintaining the integrity and physiology of metozoans by promoting cell-cell and cell-matrix interactions (1). Integrins are noncovalently associated ␣␤ heterodimeric cell surface glycoproteins. Both the ␣ and ␤ subunits are type I membrane proteins. Generally, each subunit has a large extracellular region, a single transmembrane domain, and a short C-terminal cytoplasmic tail. Integrins are bi-directional signal transducers, and they provide connectivity between the interior of the cell and its external environment. Despite the absence of intrinsic enzy-matic activity, the cytoplasmic tails of the integrins provide docking sites for an expanding list of cytosolic proteins, many of which are signaling molecules (2). The cytoplasmic tails also contain phosphorylation sites that impinge on the activity of the integrins (3). The importance of conformational changes in the extracellular regions of the integrins for their functions is well reported (4,5). The cytoplasmic tails of the integrins may be perceived either as the trigger point or the end receiver of these conformational changes in the context of integrin insideout or outside-in signaling, respectively (4,6). Therefore, the interactions between the integrin ␣ and ␤ cytoplasmic tails and the structural changes in these tails when they associate with cytosolic proteins require investigation.
The cytoplasmic tails of the platelet integrin ␣IIb␤3 are extensively analyzed. Intermolecular interactions of the ␣IIb and ␤3 tails have been reported by biochemical (7,8) and NMR studies despite differences in observations with regards to the conformations of the tails that could be attributed to a difference in the lengths of the tails analyzed (9,10). The relatively weak interactions between the isolated ␣IIb and ␤3 tails, which could present difficulty in detection (11,12), are proposed to restrain the integrin in a resting state (10). A notable feature in the ␣IIb␤3 cytoplasmic complex is the potential for ␣IIb Arg 995 and ␤3 Asp 723 to form a salt bridge that was demonstrated functionally by charge-reversal mutations to maintain the resting conformation of ␣IIb␤3 (13). The association of the N-terminal region of the large cytoskeletal protein talin, referred to as talin head domain, with the integrin ␤ cytoplasmic tails induces integrin conformational changes leading to its activation (14 -18). This is attributed primarily to the separation of the cytoplasmic tails by talin head domain binding as evident from fluorescence resonance energy transfer analyses of integrin ␣L␤2 tails (19). Based on NMR analyses of talin F3 subdomain with the ␤3 tail, it was proposed that the separation of the integrin cytoplasmic tails is a culmination of events in discrete steps involving talin docking to integrin ␤ tail, stabilization of the membrane proximal region in the ␤ tail, and the electrostatic interactions between a positively charge surface of talin with the acidic head groups of the inner membrane phospholipids (20).
Many integrins share a common ␤ subunit but different ␣ subunits, and it is becoming clear that the sequence variations in the ␣ cytoplasmic tails determine signaling specificity in these integrins. Integrin ␣2␤1 (VLA-2) showed different cell migration on collagen and laminin when the ␣2 cytoplasmic tail was replaced by that from integrin ␣5 or ␣4 (21). Integrin ␣4␤1 (VLA-4) exhibited different adhesion properties to ligand VCAM-1 in shear flow when ␣4 cytoplasmic tail was exchanged with that from integrin ␣2 or ␣5 (22). The different ␣ tails of integrin ␣L␤2 and ␣M␤2 confer to these integrins distinct chemokine-induced activation kinetics (23), and confer selective recruitment of the Src kinase Hck to ␣M␤2 but not ␣L␤2 (24). Although the membrane proximal regions of the ␣ tails are highly conserved, the membrane distal regions vary in lengths and sequences. The latter allows for structural variations and can also serve as distinct sites for the docking of specific cytosolic molecules. Examples include the specific association of nischarin with ␣5 (25), paxillin with ␣4 (26), calcein integrinbinding protein with ␣IIb (27), and CD45 cytoplasmic domain with ␣L (28). By contrast, the integrin ␤ cytoplasmic tails, with the exception of ␤4, share many similarities. A notable feature in these tails is the presence of two NPX(Y/F) motifs that can serve as internalization signals (29), and each of these motifs serves as separate docking sites for talin and the kindlins, and the latter are co-activators of integrins (14, 30 -32).
The integrin ␣L␤2, also referred to as leukocyte functionassociated antigen-1, is expressed only in leukocytes, and it serves major roles in leukocyte physiology, including diapedesis, immune synapse formation, and killer cell cytotoxicity (33)(34)(35). Ectopic expression of talin head domain induces ␣L␤2 activation possibly via association of talin head domain with the membrane proximal NPXF motif in the ␤2 tail (16,19). Another actin-binding protein ␣-actinin binds to the membrane proximal sequence His 706 -Ser 723 of the ␤2 tail of intermediate affinity ␣L␤2 (36,37). Interestingly, the binding of filamin to Thr 736 in the triplet Thr motif of the ␤2 tail has an inhibitory effect on ␣L␤2-mediated T cell adhesion (38). The regulator of adhesion and cell polarization enriched in lymphoid tissues RAPL associates with Rap1-GTP, and the activating effect of this complex on ␣L␤2 requires the membrane proximal Lys 1097 and Lys 1099 in the ␣L tail (39). Collectively, a multifaceted (positive and negative) regulatory network of molecules at the cytoplasmic face of the ␣L␤2 allows fine-tuning of ␣L␤2 activity in cells under different conditions and in different regions of a polarized and migrating cell. An impediment toward the understanding of the molecular basis of these regulatory events is a lack of structural information of the complete ␣L␤2 cytoplasmic tails. In this study, we have characterized solution conformations and interactions between the cytoplasmic tails of ␣L␤2 integrin. Taken together, the ␣ tails of integrins may have significant conformational diversities that could be important in the regulation and specific activities of integrins in various cell types.

EXPERIMENTAL PROCEDURES
Protein Expression and Purification-The numbering of the ␣L and ␤2 cytoplasmic tails is based on the work of Barclay et al. (40). The cytoplasmic tails of ␣L (Lys 1088 -Asp 1145 ) and ␤2 (Lys 702 -Ser 747 ) investigated herein are denoted as ␣L (Lys 1 -Asp 58 ) and ␤2 (Lys 1 -Ser 46 ), respectively (Fig. 1A), for ease of reference in the text and figures. The ␣L and ␤2 cytoplasmic tails were PCR-amplified from ␣L and ␤2 expression vectors (41) with relevant forward and reverse primers containing the AlwNI site. PCR products were digested with AlwNI and subcloned into the AlwNI site of pET-31b(ϩ) vector (Novagen EMD, San Diego) to generate the ketosteroid isomerase (KSI) 4integrin tail-His 6 fusion construct. The AlwNI cut sites employed for the subcloning inherently generates Met coding sequences in the connecting sequences of KSI-integrin tail and integrin tail-His 6 . The Met located between integrin tail and His 6 was mutated to Gly. In addition, ␣L Met 13 and ␤2 Met 39 were mutated to Ile. These procedures generated KSI-integrin tail-His 6 fusion proteins that contain a single Met residue located between the KSI and integrin tail. Instead of removing the His 6 tag from the integrin tail, it was retained because we found that the cleavage at Met between the integrin tail and His 6 tag was incomplete, which generated two species of integrin tails (with and without His 6 tag) that were poorly resolved by reverse-phase HPLC. To prevent complication in downstream analyses, we have therefore retained the His 6 tag by substituting the Met between the integrin tail and the tag with Gly as mentioned previously. Amino acid substitutions were made using the QuikChange TM site-directed mutagenesis kit (Stratagene, La Jolla, CA). All constructs were verified by DNA sequencing (First Base Sequencing Service, Singapore).
Fusion proteins were expressed in Escherichia coli BL21(DE3) either in LB medium or in M9 minimal medium supplemented with [ 15 N]ammonium chloride and/or [ 13 C]glucose. Protein productions were induced by 0.8 mM isopropyl ␤-D-thiogalactopyranoside at 28°C. Harvested cell pellet was resuspended in binding buffer (5 mM imidazole, 500 mM NaCl, 40 mM Tris-HCl, pH 7.9), sonicated on ice, and centrifuged at 20,000 ϫ g for 20 min at 4°C. The pellet was resuspended in binding buffer containing 8 M urea followed by affinity purification on a nickel nitrilotriacetic acid resin (Qiagen) column. Fusion protein was eluted in elution buffer (300 mM imidazole, 40 mM Tris-HCl, pH 7.9, 500 mM NaCl, 8 M urea) and dialyzed overnight at 4°C against distilled water in dialysis tubing (3.5-kDa cutoff) (Pierce). The majority of the protein formed a white precipitate and was pelleted by centrifugation at 4,000 ϫ g for 15 min at 4°C. The pellet was dissolved in 70% (v/v) formic acid, and cyanogen bromide (CNBr) (37.5 mg per 1 mg of fusion protein) was added and incubated overnight at room temperature. The solution was evaporated to dryness at 28°C in a rotary evaporator. The remaining protein gel was dissolved in deionized water, and the insoluble KSI protein was removed by centrifugation at 12,000 ϫ g for 10 min at 4°C. The cleaved tail was subjected to further purification by reverse-phase HPLC (Waters) using a C 18 column (300 Å pore size, 5 m particle size) by a linear gradient of actonitrile/water mixture. The major peak fractions were collected and lyophilized. The molecular masses of the tails were confirmed by mass spectrometry.
The human talin 1 head domain (Met 1 -Gln 435 ) was amplified from the expression plasmid pXJ40-HA-talin head domain (16) with relevant forward and reverse primers and cloned into expression vector pET-24a(ϩ) (Novagen) to generate a fusion construct containing a talin head domain with a C-terminal His 6 tag. The fusion protein was expressed in E. coli BL21(DE3) in LB medium containing 0.8 mM isopropyl ␤-D-thiogalactopyranoside for 4 h at 30°C. Harvested cells were resuspended in phosphate buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, 20 mM imidazole, pH 8.0), sonicated on ice, and centrifuged at 20,000 ϫ g for 20 min at 4°C. The fusion protein was recovered from the supernatant by nickel nitrilotriacetic acid affinity purification. Fusion protein was eluted in phosphate elution buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, 300 mM imidazole, pH 8.0). The eluted protein was further purified by size exclusion chromatography in 50 mM Tris buffer, pH 7.5, containing 150 mM NaCl on an FPLC system (Amersham Biosciences).
NMR Spectroscopy-All NMR experiments were performed at 25°C on a Bruker DRX 600-MHz instrument equipped with an actively shielded cryoprobe. Triple resonance experiments HNCA, HN(CO)CA, CBCA(CO)NH, HNCACB were carried out to achieve sequence-specific backbone resonance assignments using 15 N/ 13 C-labeled 0.8 mM protein samples in 10 mM sodium phosphate buffer, pH 6.2, 150 mM NaCl, and 2 mM tris(2-carboxyethyl) phosphine (TCEP) as reducing agent. The buffer conditions used for the NMR experiments were selected to be close to the cytoplasmic environment. Side chain aliphatic protons were assigned from three-dimensional 1 H-15 N HSQCtotal correlation spectroscopy spectra with mixing time of 70 ms. Three-dimensional 1 H-15 N NOESY experiments were carried out in the above-mentioned buffer in the absence and in presence of 50 mM calcium chloride. NMR data were processed using the Topspin program suite (Bruker) and analyzed by Sparky (T. D. Goddard and D. G. Kneller, University of California, San Francisco). All the chemical shifts were indirectly referenced to 2,2-dimethyl-2-silapentanesulfonic acid.
NMR-derived Structure Calculations and Modeling of ␣L-␤2 Tail Complex-Inter-proton distance restraints were obtained from three-dimensional 1 H-15 N HSQC-NOESY spectra, mixing time of 200 ms, and were categorized as strong, medium, and weak with upper bound distances of 3, 4, and 5 Å, respectively, whereas all the lower limits were set to 2 Å. The pseudoatom corrections were applied to the upper bound distances for methyl and methylene group protons. In addition, backbone dihedral angle ( and ) restraints were derived from TALOS (42). These predicted dihedral angle constraints were used for structure calculation with a variation of Ϯ25°from the average values. An ensemble of solution structures of ␣L tail was calculated using the program DYANA 1.5 (43). Out of 100 structures calculated, 10 conformers with the lowest energy values were selected to present the NMR ensemble. The stereochemical quality of the structures was determined using the program PROCHECK-NMR (44). A molecular model of ␣L-␤2 tails complex was constructed by iterative docking utilizing chemical shift perturbation data. The salt bridge interactions between membrane proximal helices, i.e. helix 1 of ␣L and N-terminal helix of ␤2, are initially optimized by rotation of side chain dihedral angles. The model structure was further energyminimized to relieve short inter-atomic contacts using conjugate gradient protocol, force field cvff, Insight II program (Accelrys Inc.).
NMR Interaction Studies by 1 H- 15 N HSQC Experiments-To identify binding between ␣L and ␤2 tails, 1 H-15 N HSQC experiments were carried out using either 15 N-labeled ␣L or ␤2 tails and unlabeled ␤2 or ␣L tails, respectively, at 298 K. In the tailtail complexes, concentrations of 15 N-labeled tails were fixed to 0.3 mM, and concentrations of unlabeled tails were 0.45 mM, resulting in a 1:1.5 molar ratio. 1 H-15 N HSQC spectra were also obtained after addition of talin head domain (0.5 mM) into a solution of 15 N-labeled ␣L tail and unlabeled ␤2 tail. The protein samples were prepared in 10 mM sodium phosphate buffer, pH 6.2, 150 mM NaCl, 10 mM CaCl 2 , and 2 mM TCEP. Under similar buffer conditions, isotope-filtered ( 15 N/ 14 N and 13 C/ 12 C) three-dimensional NOESY experiments were recorded using a mixture of 15 N/ 13 C-labeled ␣L tail and unlabeled ␤2 tail at a 1:1.5 molar ratio. The interactions between ␤2 tail and talin head domain were studied by acquiring 1 H-15 N HSQC spectra of 15 N-labeled ␤2 tail as a function of concentrations of tail head domain. Titrations were carried out by addition of aliquots of unlabeled talin head domain (1 mM) into 0.3 mM 15 N-labeled ␤2 tail in 10 mM phosphate buffer, pH 6.2, 150 mM NaCl, and 2 mM TCEP at 298 K. The chemical shift changes of backbone 1 H and 15 N resonances were calculated using the following equation: ((⌬ 1 H ) ϩ (⌬ 15 N )) (10). The calcium binding to ␣L tail was performed by additions of aliquots of CaCl 2 (1 M stock solution) into 15 N-labeled ␣L (0.3 mM) in 10 mM sodium phosphate buffer, pH 6.2, 150 mM NaCl, and 2 mM TCEP. All samples were allowed to equilibrate for 20 min before twodimensional 1 H-15 N HSQC spectra were recorded on a Bruker DRX 600-MHz instrument, equipped with cryo-probe and at 298 K.
Isothermal Titration Calorimetric Studies-Isothermal titration calorimetry (ITC) experiments were performed on a VP-ITC Micro Calorimeter (MicroCal LLC, Northampton, MA). All measurements were made in 10 mM phosphate buffer containing 150 mM NaCl, 2 mM TCEP, and pH 6.2 at 293 K. A fixed concentration of 0.01 mM talin head domain placed in the sample cell was titrated by sequential additions of a 5-l aliquot of a 0.2 mM ␤2 tail from the injection syringe, except for the first injection of 2.5 l. Typically, 40 injections were performed at an interval of 4 min while the reaction cell was constantly stirred at 300 rpm. Similar titrations were carried out using a fixed concentration of 0.01 mM ␣L tail placed in the sample cell with 25 injections of 5-l aliquots from 0.2 mM ␤2 tail placed in the injection syringe. All the reaction conditions were the same as that of the former one. For calcium binding titration, 0.01 mM ␣L placed in the sample cell was titrated by sequential addition of 5-l aliquots of 100 mM CaCl 2 from the injection syringe. Raw data were collected and integrated using MicroCal Origin 5.0 supplied with the instrument. A single set of binding site model, provided with the software, was fitted to the data to obtain association constant, K a , and enthalpy change ⌬H. Free energy (⌬G) and entropy (⌬S) changes were calculated using the equations ⌬G ϭ ϪRT lnK a and ⌬S ϭ (⌬H Ϫ ⌬G)/T, respectively.

Resonance Assignments of the Cytoplasmic Tails of ␣L␤2
Integrin-The ␣L and ␤2 tails were overexpressed into inclusion bodies in E. coli as a fusion protein with KSI. This strategy prevented undesirable proteolytic degradation of the target proteins. The integrin tails were released from KSI by CNBr cleavage and subjected to further purification by HPLC (see under "Experimental Procedures"). A sequence comparison among the ␣ tails from other integrins revealed that the ␣ tails are heterogeneous in length and in amino acid compositions except for the first seven conserved amino acids at the N terminus (Fig. 1A). In all subsequent figures, the amino acids are numbered as Lys 1 -Asp 58 for Lys 1088 -Asp 1145 in the ␣L tail and Lys 1 -Ser 46 for Lys 702 -Ser 747 in the ␤2 tail. It is also noteworthy that the ␣L tail is much longer as compared with other ␣ tails,  including the well investigated 20-residue tail of the ␣IIb integrin (9,10,45). By contrast, the ␤2 and ␤3 tails are similar in length and contain a number of conserved residues. Assignments for the backbone HN, 15 N, 13 C ␣ , and 13 C ␤ resonances of the ␣L and ␤2 tails were achieved by a combined analysis of triple resonance HNCA, HN(CO)CA, HNCACB, and CBCA(CO)NH spectra. The 1 H-15 N heteronuclear single quantum coherence (HSQC) spectrum of ␣L tail with backbone HN and 15 N correlations of individual residues is shown (Fig.  1B). All the backbone 1 H-15 N correlations were identified except for four Pro residues. The 15 N resonances of Gly and Ser/Thr residues are found to be the most upfield-shifted (46). The primary amino acid sequence of ␣L contains 10 Gly and 4 Ser residues distributed along the sequence (Fig. 1A). Gly residues resonating at 108 -113 ppm at the 15 N dimension and Ser residues resonating at 115-117 ppm were identified in the HSQC spectrum of the ␣L tail (Fig. 1B). 1 H-15 N HSQC crosspeaks of the 46-residue ␤2 tail were also assigned except for residues Phe 15 , Glu 16 , Leu 31 , and Lys 42 (Fig. 1C). Interestingly, we found two 1 H-15 N cross-peaks for residues Ala 34 , Thr 35 , Thr 36 , Phe 43 , Ala 44 , Glu 45 , and Ser 46 , presumably as a result of cis-trans-isomerization of the Asn 40 -Pro 41 peptide bond.
Conformational Characteristics of ␣L and ␤2 Tails-Secondary structures of ␣L and ␤2 tails were obtained from the deviation of 13 C ␣ chemical shifts from random coil values and short and medium range NOEs (Fig. 2). 13 C ␣ chemical shift deviation is a reliable indicator for identification of helix and ␤-sheet secondary structures (47). The 13 C ␣ atom experiences a downfield shift in helical structure and an upfield shift in ␤-strand. A stretch of at least four contiguous residues or a stretch of at least three adjacent residues with helical or strand-type chemical shift deviations can be assigned as a stable helix or ␤-strand conformation, respectively (47). Based on this, three helical segments (helix 1, Lys 6 -Ile 13 ; helix 2, Ala 25 -Glu 36 ; and helix 3, Lys 44 -Gly 55 ) were derived for the ␣L tail ( Fig. 2A). Analysis of 15 N-edited three-dimensional HSQC-NOESY spectra of ␣L reveals diagnostic short and medium range NOEs, sequential HN/HN and C ␣ H/HN (i to i ϩ 3/i ϩ 4), consistent with the helical structures for these segments. Residues between helix 1 and helix 2 (Glu 14 -Pro 24 ) and residues between helix 2 and helix 3 (Gly 38 -Leu 43 ) are characterized by predominantly sequential NOEs, indicating a lack of regular secondary structures. It is noteworthy that there are a number of Gly and/or Pro residues in these two segments that may be responsible for the disordered or loop-like conformations for these segments. The first five amino acid residues at the N terminus (Lys 1 -Phe 5 ) also appeared to be flexible in the ␣L tail. Interestingly, we have identified several NOE connectivities among the three helices of the ␣L tail (Table 1). These long range NOEs, e.g. F4C ␤ Hs/ L43HN (Fig. 3A, right panel), N21C ␤ H/G55HN, and I13C ␥ H 3 / E52HN (Fig. 3, B and C) suggest a close proximity (Յ5 Å) between helix 3/helix 1 and helix 2/helix 3.  Table 1.
By contrast, the analyses of 13 C ␣ chemical shift that deviates from random coil values and NOE of the ␤2 tail suggest that a large part of the polypeptide chain is devoid of any regular secondary structures (Fig. 2B). Only a short region at the N terminus, residues Leu 6 -Arg 14 , showed a propensity for a helical conformation. However, we could not unambiguously detect any medium range helical NOEs among these residues, suggesting a probable nascent helical conformation (48).
A Compact Conformation of the ␣L Tail-An ensemble of solution structures of the ␣L tail was determined based on short/medium and long range NOEs and deduced backbone dihedral (, ) constraints ( Fig. 4 and Table 2). The superpositions of backbone atoms (C ␣ , N, and CЈ) of 10 lowest energy structures of the ␣L tail are shown (Fig. 4, A-D). The individual helical structure of ␣L appears to show a close superposition with root mean square deviation of 0.14, 0.25, and 0.28 Å for helix 1, helix 2, and helix 3, respectively. The superposition of the backbone atoms of the overall tertiary fold of the ␣L tail showed relatively higher root mean square deviation values of backbone and all heavy atoms at 0.77 and 1.25 Å, respectively ( Fig. 4D and Table 2). The compact structure of ␣L tail is largely defined as the fold back of helix 3 between helix 1 and helix 2 (Fig. 4E). We also examined plausible side chain-side chain interactions that may be involved in defining the orientations of these helices. There are a number of potential polar and/or ionic interactions between helix 1 and helix 3 and between helix 2 and helix 3 that may impart stabilization to the folded conformation of the ␣L tail. For example, residues Lys 6 and Lys 10 from helix 1 may form salt bridges or hydrogen bonds with residues Glu 48 and Glu 52 from helix 3 (Fig.  4F). Furthermore, side chains of residues Lys 44 , His 47 , and Lys 57 from helix 3 were found to be in close proximity with residues Glu 36 , Ser 33 , and Glu 29 of helix 2 (Fig. 4G), indicating plausible charge-charge interactions or hydrogen bond formations among these residues. NOE interactions observed among these residues indeed suggest probable occurrence of such ionic and/or hydrogen bond formations in ␣L structure (Table 1). A hydrophobic packing interaction between residues Phe 4 and Leu 43 could also be involved in maintaining the tertiary fold of ␣L in solution (Table 1). Intra-helical salt bridge interactions can also be present between residues Arg 7 -Glu 11 in helix 1 and residues Lys 44 -Glu 48   residues (Fig. 1A). Few hydrophobic packing interactions that stabilize well folded structures of proteins were found in the ␣L conformation.
Binding of Calcium to the ␣L Tail-Divalent cations like Ca 2ϩ , Mg 2ϩ , and Mn 2ϩ have been shown to interact with the ␣IIb tail (7). We examined probable Ca 2ϩ binding to the ␣L tail by monitoring changes in 1 HN and 15 N chemical shifts in 1 H-15 N HSQC spectra acquired at different concentrations of calcium chloride. Sections of 1 H-15 N HSQC spectra are shown (Fig. 5, A-C). 1 H-15 N HSQC cross-peaks of residues Glu 26 , Asp 27 , Ser 28 , Glu 29 , Gln 35 , and Glu 36 from helix 2, residues Gly 38 , Asp 39 , Gly 41 , and Cys 42 from the loop between helix 2 and helix 3, and residues His 47 , Glu 48 , Lys 49 , Asp 50 , Ser 51 , Glu 52 , Ser 53 from helix 3 showed considerable changes upon addition of CaCl 2 . Many of these affected residues are negatively charged. They are likely to coordinate with Ca 2ϩ and possibly with other divalent cations. Interestingly, these residues also appear to be congregated in the folded structure of ␣L tail forming a negatively charged surface (Fig. 5, D and E). The binding of Ca 2ϩ with ␣L tail was examined by ITC experiments (Fig. 5F). There was an upward trend of ITC titration peaks (Fig. 5F, top  panel), and the integrated heats are found to be positive (bottom panel) suggesting that the Ca 2ϩ binding to the ␣L tail is driven by hydrophobic interactions (Table 3). However, one would expect a predominant role of ionic interactions or an exothermic heat released as a result of binding of Ca 2ϩ ions to the negatively charged surface of ␣L (Fig. 5D). The dominance of hydrophobic interactions presumably underscores an overwhelming release of water molecules from the binding site(s). We also observed a higher molar ratio of Ca 2ϩ /protein that was necessary to saturate all the binding sites of ␣L (Fig. 5F, bottom  panel). This may arise because of the dynamic nature of the interaction between metal ions and ␣L. We have also analyzed three-dimensional NOESY spectra of the ␣L tail in the presence of excess CaCl 2 (see under "Experimental Procedures"). There were no significant differences in the NOE connectivities between the apo-and holo-forms of protein, indicating a similar helical fold even after Ca 2ϩ binding (data not shown).
Interactions between ␣L and ␤2 Tails-The 1 H-15 N HSQC spectra of 15 N-labeled free ␣L tail alone (red contour) and in the presence of unlabeled ␤2 tail (black contour) at 1:1.5 (␣L:␤2) ratio were obtained (Fig. 6A). Chemical shift changes of HSQC cross-peaks were detected for a number of residues of the ␣L tail upon addition of the ␤2 tail, suggesting interactions between the two tails. Chemical shift changes of ␣L tail as a function of amino acid residues are also shown (Fig. 6B). Chemical shift perturbations, although in different magnitude, were observed for many residues throughout the sequence of the ␣L tail. However, chemical shift changes were found to be significantly higher (Ն75 Hz) only for residues from helix 1 (Lys 6 , Arg 7 , Asn 8 , Lys 10 , and Arg 17 ) and helix 3 (Lys 44 , His 47 , and Lys 49 ) (Fig. 6, B and C). By contrast, only one residue, Glu 36 , from helix 2 of the ␣L tail showed a change in chemical shift of equivalent level (Ն75 Hz). These data suggest that residues from helix 1 and helix 3 of the ␣L tail are involved in interacting with the ␤2 tail. Inclusion of talin head domain appeared to disrupt the interactions between the ␣L and ␤2 tails because no significant chemical shift perturbations in the 1 H-15 N HSQC spectra of ␣L tail were detected in the presence of talin head domain (supplemental Fig. 1).
In the reverse 1 H-15 N HSQC titration, chemical shifts of the 15 N-labeled ␤2 tail (red contour) were detected in the presence of unlabeled ␣L tail (black contour) (Fig. 6D). However, the changes in chemical shift were restricted only to the N-terminal residues of the ␤2 tail, suggesting their involvement in ␣L tail interaction. Residues His 5 , Leu 6 , Asp 8 , Glu 11 , and Tyr 12 of the ␤2 tail could be directly involved in the complex formation with ␣L tail (Fig. 6, E and F). The observations of a downward trend of ITC titration peaks (Fig. 7, top panel) and the resultant negative integrated heats (Fig. 7, bottom panel) suggest that the association of ␣L and ␤2 tails is exothermic. The thermodynamic parameters of this association are provided ( Table 3). The interaction appears to be driven by a favorable change in enthalpy that may indicate predominant role(s) of ionic and/or hydrogen bonding toward the formation of the ␣L-␤2 tails complex. To determine a plausible orientation of the ␣L-␤2 tails complex, 15 N/ 14 N-and 13 C/ 12 C-filtered NOESY experiments were carried out (see under "Experimental Procedures"). However, we could not detect any intermolecular NOE contacts between the tails. The lack of NOE contacts may result from a fast dissociation of the complex or a low affinity binding between the tails (Table 3). Therefore, the orientation of the ␣L-␤2 tails complex was determined from an iterative docking followed by energy minimization procedure (Fig. 8). The positively charged residue Arg 7 from helix 1 of the ␣L tail (Fig. 8,  red) is engaged to form salt bridges with acidic residues Asp 8 and Glu 11 from the N-terminal helix of the ␤2 tail (orange). The importance of these salt bridge interactions, in particular Arg 7 -Asp 8 , in maintaining inactive states of integrins was determined in ␣IIb␤3 and ␣L␤2 (13,16,49). In the docked structure, potential ionic interactions were also found between acidic residues Glu 11 , Glu 14 in helix 1 of ␣L tail with residue Arg 14 of ␤2 tail. Interestingly, we also found plausible stabilizing interactions  40 , and Leu 43 of the ␣L tail were found to be in close proximity with Ile 4 of the ␤2 tail.
Interactions of the ␤2 Tail with Talin Head Domain-Talin is a FERM domain-containing and actin-binding protein that connects integrins to the actin cytoskeleton (50,51). The head domain of talin is involved in the activation of several integrins, including ␣L␤2 (14 -16, 18, 19). Here we examined the binding of ␤2 tail with the talin head domain (Met 1 -Gln 435 ) by 1 H-15 N HSQC and ITC studies (Fig. 9). The 1 H-15 N HSQC spectra of ␤2 tail at various concentrations of unlabeled talin head domain was determined (Fig. 9A). Talin head domain induced marked chemical shift changes or broadening of a number of 1 H-15 N HSQC cross-peaks of the ␤2 tail. This suggests talin head domain interacting with the ␤2 tail. Of note, significant changes in chemical shifts were detected for residues located at the N-terminal membrane proximal and central regions of the ␤2 tail (Fig. 9B). These suggest a substantial portion of the ␤2 tail is involved in the interaction with talin head domain. Similar results were reported from 1 H-15 N HSQC titration of ␤3 tail with the F3 subdomain of the talin head (52).
A recent NMR structure of a complex between a chimeric peptide, containing N-terminal membrane proximal region of ␤3 and C-terminal residues of PIPKI␥, with F3 subdomain of talin revealed extended interfacial contacts with residues of membrane proximal and more distal regions (20). Binding of the ␤2 tail with talin head domain was further investigated by ITC measurements (Fig. 9C and Table 3). The binding of ␤2 tail to talin head domain was endothermic as suggested by an upward trend of the ITC titration peaks (Fig. 9C,  top panel), and the resultant positive integrated heats (Fig. 9C,  bottom panel). The ITC results suggest that the talin head domain and ␤2 tail forms a 1:1 complex with an estimated K d  ϳ0.5 M, and the complex formation is driven by a positive change in entropy with a free energy (⌬G) change of Ϫ8.45 kcal/mol (Table 3), indicating dominant role of hydrophobic interaction in the complex formation.

DISCUSSION
The integrin ␣L␤2 mediates activation-dependent adhesion and migration of the leukocytes (34). Distinct populations of ␣L␤2 conformers representing different activation states are localized at specific regions of polarized and migrating T cells (37,53,54). To generate a dynamic equilibrium of ␣L␤2 conformers in these cells, cytosolic signaling and cytoskeleton-bridging molecules play indispensable roles (33,55,56). Hence, the ␣L␤2 cytoplasmic tails that serve as a hub for the recruitment of cytosolic proteins warrant investigations. Despite increasing number of studies examining ␣L␤2 cytoplasmic tails by biochemical, mutational, and functional analyses, the structures of the ␣L␤2 cytoplasmic tails remain unknown. This is in contrast to the platelet integrin ␣IIb␤3 cytoplasmic tails that have been subjected to many structural analyses (9 -12, 45, 57). Although the ␤2 and ␤3 tails share considerable sequence homology, the ␣L and ␣IIb tails are highly divergent in terms of sequence composition and length.
Therefore, we have investigated the solution structures and interactions of the ␣L␤2 cytoplasmic tails. In solution the isolated ␣L tail adopts a folded conformation that is characterized by three well defined helical segments. These helices are oriented into a novel compact tertiary structure whereby helix 3 is engaged in interactions with helix 1 and helix 2 (Fig. 4). The specific arrangement of the three helices is presumably strengthen by salt bridges and hydrogen bonds among the helices. The well conserved N-terminal or membrane proximal region of the ␣L tail forms part of helix 1 (residues Lys 6 -Ile 13 ). This region was also found to be helical in the structure of the ␣IIb tail in lipid micelles (45,57) and when in complex with the ␤3 tail (9,10). The packing between helix 1 and helix 3 showed plausible ionic/hydrogen bond interactions between positively charged residues Lys 6 and Lys 10 with the negatively charged residues Glu 48 and Glu 52 , respectively (Fig. 4F). Interestingly, the folded structure of ␣IIb tail obtained in DPC micelle was also found to contain salt bridges between conserved residue Lys 994 (Lys 6 in ␣L tail) with acidic residues Asp 1004 and Glu 1005 from the loop (45). However, ionic interactions between helix 2 and helix 3 found here are unique to the ␣L tail (Fig. 4G). The binding of metal ions (Ca 2ϩ , Mg 2ϩ , and Mn 2ϩ ) were characterized for the ␣IIb tail (7), suggesting plausible role of intracellular cations in the regulation of integrin activity. These divalent cations coordinate, at micromolar affinity, with the ␣IIb tail utilizing a stretch of acidic residues, Glu 1001 -Glu 1008 , at the C terminus (Fig. 1A). These acidic residues are largely nonconserved among the amino acid sequences of integrin ␣ tails (Fig. 1A). Here we were able to detect calcium binding of the ␣L tail. The calcium binding to the ␣L tail appeared to be stabilized by the acidic residues primarily located in the helix 2-loop-helix 3 region of the ␣L tail (Fig. 5D). In the folded topol-ogy of the ␣L tail, these residues form a negatively charged surface (Fig. 5E). We also surmise that apart from binding cations, this negatively charged surface may be available for interactions with other positively charged ligands.
In solution the isolated ␤2 cytoplasmic tail is largely unstructured except for the N-terminal residues Leu 6 -Arg 14 similar to the conformation of the free ␤3 tail (Fig. 2B) (11). We were unable to detect any turn conformations for 28 NPLF 31 sequence motif in the ␤2 tail as a result of limited NOEs and spectral overlap. The corresponding sequence motif in ␤3 tail adopts a ␤-turn conformation and is involved in the interaction with talin head domain (11). With regard to tails complex formation, the association of ␣L tail with ␤2 tail predominantly involved residues located at the membrane proximal helix 1 and helix 3 of the ␣L tail, and the membrane proximal helix of the ␤2 tail (Fig. 6). These interactions may form an interface in the ␣L-␤2 tails complex as depicted in a docked model (Fig. 8).
The NMR structure of the ␣IIb-␤3 tails complex revealed an interface stabilized by two membrane proximal helices, each from one tail (10). A notable feature in the interface of the ␣IIb-␤3 tails complex is the potential salt bridge interactions involving ␣IIb Arg 995 and ␤3 Asp 723 and Glu 726 . Nonpolar packing between ␣IIb Val 990 and Phe 992 and ␤3 Leu 718 and Ile 719 was also determined. These interactions that potentially stabilize the interface of the ␣IIb-␤3 tails complex are plausible in the docked structure of ␣L-␤2 tails complex. These include potential salt bridges between Arg 7 of ␣L tail with Asp 8 and Glu 11 of ␤2 tail and the hydrophobic packing between residues Phe 4 of ␣L and Ile 4 of ␤2 (Fig. 8). Ionic interactions specific to the ␣L-␤2 tails complex because of the close proximities of acidic residues Glu 11 and Glu 14 (helix 1) of the ␣L tail with the basic residue Arg 14 of the ␤2 tail are permissible (Fig. 8). All these residues experienced changes in chemical shift as a result of mutual interactions between ␣L and ␤2 tails (Fig. 6). The dominant role of ionic and/or polar interactions in the complex formation between ␣L and ␤2 tails are also supported by the detection of exothermic heat release in the ITC analyses (Fig. 7).
The cytoskeleton protein talin binds directly with ␤1, ␤2, and ␤3 integrins (2). An in vivo study demonstrated that the head domain of talin interacts with the ␤2 tail that leads to the activation of ␣L␤2 by disrupting the ␣L-␤2 tails complex (19). In this study, we detected direct interaction between talin head domain with the ␤2 tail by NMR and ITC measurement. The binding of talin head domain to the ␤2 tail induces substantial changes in the chemical shifts of a large number of residues, indicating a considerable region of the ␤2 tail is engaged in interactions, including the membrane proximal helix. As mentioned previously, the membrane proximal helix of ␤2 is also involved in the interaction with the  ␣L tail. However, binding affinity between ␣L and ␤2 tails estimated from ITC studies was found to be significantly lower, K d ϳ 2.63 M, as compared with binding affinity, K d ϳ 0.5 M, of talin head domain to ␤2 tail (Table 3). Therefore, it is likely that a high affinity interaction between talin head domain with the ␤2 tail may disrupt the association of the ␣L and ␤2 tails and trigger the activation of ␣L␤2.
The ␣L␤2 cytoplasmic tails can undergo phosphorylations that impinge on the activity of ␣L␤2 (58). In the ␣L tail, Ser 1140 (Ser 53 herein) was found to be abundantly phosphorylated in resting T cells, and mutation of Ser 1140 to Ala abrogated ␣L␤2 affinity up-regulation by the small GTPase Rap1 (59). Of note, the substitution of Ser 1140 with Ala did not affect heterodimerization of ␣L and ␤2 (59), and in this study, Ser 1140 is not perturbed when the ␣L tail interacts with the ␤2 tail (Fig. 6B). This suggests that Ser 1140 phosphorylation may serve to modulate interactions of cytosolic factors with the ␣L tail. In this regard, it is interesting to note that Ser 1140 in helix 3 is located at the negatively charged surface of the ␣L tail, and its phosphorylation can enhance the negative charge of this surface (Fig. 5, D  and E). In the ␤2 tail, phosphorylation of Thr 736 (Thr 35 herein) but not Ser 734 (Ser 33 herein) allows the docking of 14-3-3 proteins, thereby modulating the activity of ␣L␤2 (59 -61). Thr 736 is also one of the threonine triplets in the ␤2 tail that was reported to undergo phosphorylation, two threonines at a time, in T cells stimulated with phorbol ester (62).
In a series of studies that examined the interaction of ␣IIb and ␤3 tails, differences in observations were made. These are possibly attributed to different approaches employed that included coiled-coil fusion proteins, tails with transmembrane domains, and fragments of tails other than full-length tails of the ␣IIb and ␤3 subunits (9 -12). In vivo, the orientation of the tails and their juxtaposition when in complex may be affected by the positions of the transmembrane domains. Recent NMR studies suggest that the first charged residue Lys on the intracellular side of ␣IIb and ␤3 does not necessarily demarcate the C-terminal end of the transmembrane domains because of the tilting of transmembrane domains in the lipid bilayer (63,64). Interestingly, the ␣IIb cytoplasmic tail residues Phe 992 and Phe 993 after the Lys 989 insert back into the membrane. These Phe residues are conserved in the ␣L tail (Phe 4 and Phe 5 herein, see Fig. 1A). Whether the insertion of these residues into the membrane as in ␣IIb may change the conformation of the ␣L tail and its mode of interaction with the ␤2 tail is not known at present. Furthermore, the ␣IIb tail adopts a "closed" conformation with its C-terminal region folded back and interacting with its N-terminal helix (45). It was shown that membrane-permeable myristoylated wild-type ␣IIb tail peptide inhibited fibrinogen binding to platelets, whereas a mutant peptide, with P998A/P999A mutations that disrupt the turn between the N-terminal helix and C-terminal region, had no inhibitory effect (45). This demonstrates the relevance and importance of the ␣IIb tail conformation in ␣IIb␤3 function. In the ␣L tail conformation, a turn is observed between helix 1 and helix 2 and another between helix 2 and helix 3. Hence, it will be interesting in future studies to determine the physiological significance of the ␣L tail conformation in ␣L␤2 functional assays using, for example, a T cell-based system and a series of myristoylated wild-type and fold-disrupting mutant ␣L tail peptides.