Structures and Interaction Analyses of Integrin αMβ2 Cytoplasmic Tails*

Background: Cytosolic tails of integrins are critical for activation. Results: Structures of cytosolic tail of αM and phosphorylated αM integrin were determined. The interactions between αM and β2 tails were investigated. Conclusion: Structures of αM are characterized by N-terminal helix and loop at the C terminus with helix-loop packing. αM tail interacted with β2. Significance: Structures and interactions provide insights into activation of integrins. Integrins are heterodimeric (α and β subunits) signal transducer proteins involved in cell adhesions and migrations. The cytosolic tails of integrins are essential for transmitting bidirectional signaling and also implicated in maintaining the resting states of the receptors. In addition, cytosolic tails of integrins often undergo post-translation modifications like phosphorylation. However, the consequences of phosphorylation on the structures and interactions are not clear. The leukocyte-specific integrin αMβ2 is essential for myeloid cell adhesion, phagocytosis, and degranulation. In this work, we determined solution structures of the myristoylated cytosolic tail of αM and a Ser phosphorylated variant in dodecylphosphocholine micelles by NMR spectroscopy. Furthermore, the interactions between non-phosphorylated and phosphorylated αM tails with β2 tail were investigated by NMR and fluorescence resonance energy transfer (FRET). The three-dimensional structures of the 24-residue cytosolic tail of αM or phosphorylated αM are characterized by an N-terminal amphipathic helix and a loop at the C terminus. The residues at the loop are involved in packing interactions with the hydrophobic face of the helix. 15N-1H heteronuclear single quantum coherence experiments identified residues of αM and β2 tails that may be involved in the formation of a tail-tail heterocomplex. We further examined interactions between myristoylated β2 tail in dodecylphosphocholine micelles with dansylated αM tail peptides by FRET. These studies revealed enhanced interactions between αM or phosphorylated αM tails with β2 tail with Kd values ∼5.2 ± 0.6 and ∼4.4 ± 0.7 μm, respectively. Docked structures of tail-tail complexes delineated that the αM/β2 interface at the cytosolic region could be sustained by a network of polar interactions, ionic interactions, and/or hydrogen bonds.

Integrins are cell surface adhesion molecules formed by specific non-covalent associations between different ␣ and ␤ subunits (1). Each subunit has a large extracellular region that is required for ligand binding, a single transmembrane domain, and a cytoplasmic tail (2,3). Integrins are bidirectional signaling receptors that allow communication between the interior of a cell and its external microenvironment (1). Integrin occupation and clustering induce cytoplasmic signaling by recruiting cytoplasmic adaptor and cytoskeletal proteins to its cytoplasmic tails, ultimately forming an integrin adhesome (4,5). In the immune system, integrins mediate many leukocyte functions such as the migration across endothelium, phagocytosis, antigen presentation, cytotoxic cell killing, and selective homing of lymphocytes into specialized tissues (6 -8). The four leukocyterestricted ␤2 integrins have different ␣ subunits pairing with a common ␤2 subunit. They are ␣L␤2 (LFA-1, CD11aCD18), ␣M␤2 (Mac-1, CD11bCD18), ␣X␤2 (p150,95, CD11cCD18), and ␣D␤2 (CD11dCD18). The functional importance of the ␤2 integrins is underscored by the high susceptibility to microbial infections of leukocyte adhesion deficiency I and III patients (9). The adhesive and migratory properties of the leukocytes are compromised in these patients because of disrupting mutations in the ␤2 subunit (leukocyte adhesion deficiency I) or in the integrin cytoplasmic regulator kindlin-3 (leukocyte adhesion deficiency III) that regulates the activation of the ␤2 integrins (10,11). The integrin ␣M␤2 is expressed on myeloid, natural killer, and ␥␦ T cells (12)(13)(14). It is a promiscuous receptor that binds to many ligands including intercellular adhesion molecules, junctional adhesion molecule-3, iC3b, fibrinogen, microbial saccharides, and denatured proteins (15)(16)(17)(18)(19). In addition to its well documented roles as an adhesion molecule, as a phagocytic receptor, and in the regulation of degranulation and apoptosis, integrin ␣M␤2 has been shown to be involved in the maintenance of tolerance and the control of inflammation (14,20,21). Furthermore, ␣M␤2 has been shown to be involved in the regulation of monocyte differentiation (22)(23)(24). Although ␣M has a short cytoplasmic tail as compared with ␣L, ␣X, and ␣D, it serves important functions in regulating ␣M␤2 ligand binding and outside-in signaling. It is well established that ␣M␤2 outside-in signaling regulates neutrophil survival. In particular, ␣M␤2 with a truncated ␣M cytoplasmic tail or with the ␣M tail replaced by a ␣L tail have an impaired ability to protect transfected K562 cells from apoptosis as a result of inactivation of Akt and Erk1/2 (25). Notably, ␣M␤2, but not the ␣M␤2 mutant with its ␣M cytoplasmic tail replaced with that of ␣L or ␣X, recruits the Src family kinase Hck (26). Interestingly, a genome-wide association study found an association between systemic lupus erythematosus and a non-synonymous SNP, rs1143678 (C3 T), that substitutes amino acid Pro 1130 with Ser in the ␣M tail (27). Phosphorylation of integrin cytoplasmic tails is an important post-translational modification that is necessary for integrin functions (28). In polymorphonuclear leukocytes, the ␣M cytoplasmic tail is constitutively phosphorylated, whereas the phosphorylation of the ␤2 tail is dependent on cellular activation (29,30). There are two phosphorylation sites in the ␣M cytosolic tail, Tyr 1121 and Ser 1126 . Interestingly, it was found that mutation of Ser 1126 in the ␣M cytosolic tail disrupts the ability of ␣M␤2 to bind intercellular adhesion molecule-1 and intercellular adhesion molecule-2 but not iC3b and denatured BSA (31), suggesting that ␣M cytosolic tail Ser 1126 phosphorylation plays an important role in ␣M␤2 functional regulation. Indeed, the mutant ␣MS1126A␤2 failed to react with the activation reporter mAb KIM127 and bound poorly to intercellular adhesion molecule-1 (32). In addition, cells expressing ␣MS1126A␤2, but not wild-type ␣M␤2, when injected into mice showed reduced extravasation into the spleens and lungs (32). Collectively, these data suggest that the ␣M cytoplasmic tail is pivotal in the regulation of ␣M␤2 ligand binding and outside-in signaling. To understand the mechanism of ␣M␤2 regulation and its cytoplasmic signaling, detailed analyses of its cytoplasmic tails are required. In this study, solution structures of myristoylated ␣M tail in two forms, phosphorylated (Ser(P) 1126 ) and non-phosphorylated, were determined in solution containing dodecylphosphocholine (DPC) 3 micelles. Interactions studies between the ␣M and ␤2 tails using NMR and fluorescence resonance energy transfer (FRET) demonstrated a low affinity binding for the non-myristoylated tails in solution. However, FRET experiments carried out in detergentcontaining solutions showed high affinity interactions between the tails when one of the tails was myristoylated.

EXPERIMENTAL PROCEDURES
Synthesis and Purification of Synthetic Cytoplasmic Tails-Synthetic cytoplasmic tail peptides of ␣M and ␤2 used in this study are given in Table 1. All of the cytoplasmic tail peptides (myristoylated, non-myristoylated, and other derivatives) were purchased from GL Biochem (Shanghai, China) and were further purified using reverse phase HPLC (Waters, Milford, MA) connected to a C 18 column (300-Å pore size, 5-m particle size). A linear gradient of acetonitrile/water with a flow rate of 2 ml/min was used to elute the peptides, and the major peak fractions were lyophilized. The mass of the peptides was confirmed by mass spectrometry.
Expression Plasmids-The numbering of the amino acids in the ␣M and ␤2 tails is based on the work of Buyon et al. (29). The full-length ␤2 tail (Lys 702 -Ser 747 ) that was subcloned into the AlwNI site of pET-31b(ϩ) vector (Novagen, EMD, San Diego, CA) containing an N-terminal ketosteroid isomerase (KSI) has been reported previously (33). The cDNA encoding the full-length ␣M tail (Lys 1113 -Gln 1136 ) with the terminating stop codon was PCR-amplified from ␣M-pcDNA3.0 expression plasmid (31) and cloned into the AlwNI site of the pET-31b(ϩ) vector. The forward primer used for the PCR was designed such that formic acid cleavage GGGGSDP sequence (34) was introduced between the KSI and the ␣M tail sequences. The DP site allows cleavage of the ␣M tail from the KSI using formic acid in subsequent protein expression and purification procedures. A stop codon was introduced immediately after Gln 1136 in the ␣M tail that disrupts the expression of a C-terminal His 6 tag in the pET-31b(ϩ) vector; therefore, we constructed a His 6 tag at the N terminus of the KSI to generate the final expression plasmid containing His 6 -KSI-GGGG-SDP-␣M tail. This allowed affinity purification of the fusion protein, cleavage of the fusion protein by formic acid, and recovery of full-length ␣M tail without additional tag sequences. Furthermore, the cytoplasmic tails of ␣M (Lys 1113 -Gln 1136 ) and ␤2 (Lys 702 -Ser 747 ) are numbered as ␣M (Lys 1 -Gln 24 ) and ␤2 (Lys 1 -Ser 46 ), respectively.
Protein Expression and Purification-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 [ 15 N]ammonium chloride/[ 13 C]glucose as described in our previous study (33). For the induction of protein expression, isopropyl ␤-D-thiogalactopyranoside (0.8 mM) was added, and the bacteria culture was incubated at 25°C. The harvested cell pellet was resuspended in binding buffer (5 mM imidazole, 500 mM NaCl, 40 mM Tris, pH 7.9), sonicated on ice, and centrifuged at 20,000 rpm for 20 min at 4°C. The protein pellet was then resuspended in binding buffer containing 8 M urea, and His 6 -tagged fusion protein was purified on a nickelnitrilotriacetic acid resin (Qiagen) column. The fusion protein was eluted in elution buffer (300 mM imidazole, 500 mM NaCl, 8 M urea, 40 mM Tris-HCl, pH 7.9) and dialyzed against distilled water overnight at 4°C. The protein precipitate was pelleted by centrifugation at 4000 ϫ g for 15 min at 4°C. CNBr in 70% (v/v) formic acid was used for the cleavage of the ␤2 tail from the KSI fusion partner as described previously (33). For the cleavage of the ␣M tail from the KSI fusion partner, the fusion protein precipitate was dissolved in 90% (v/v) formic acid and incubated overnight at room temperature. The solution was then evaporated to dryness at 28°C in a rotary evaporator. The protein gel was dissolved in deionized water. The KSI protein precipitate was removed by centrifugation at 12,000 ϫ g for 10 min at 4°C. The cleaved ␣M or ␤2 tails were subjected to reverse phase HPLC (Waters) purification on a C 18 column (300-Å pore size, 5-m particle size) by a linear gradient of an acetonitrile/water mixture. NMR Experiments-All NMR spectra were recorded on a Bruker DRX 600-MHz instrument equipped with an actively shielded cryoprobe. NMR data were processed using TOPSPIN 2.1 and analyzed with SPARKY. 4 Two-dimensional 1 H-1 H total correlation spectroscopy (TOCSY) and two-dimensional 1 H-1 H nuclear Overhauser effect spectroscopy (NOESY) experiments were carried out for myristoylated ␣M (Myr-␣M) and myristoylated and phosphorylated ␣M (Myr-P␣M) at a concentration of 0.7 mM in 10 mM sodium phosphate buffer, pH 5.6, containing 200 mM perdeuterated DPC (DPC-d 38 ) (Cambridge Isotope Laboratories, Andover, MA) at 308 K. NOESY spectra were acquired for myristoylated peptides (200 M) in 10 mM sodium phosphate buffer, pH 5.6 at 308 K.
TOCSY and NOESY spectra were acquired for non-myristoylated ␣M tail peptide at a concentration of 0.7 mM in 10 mM sodium phosphate buffer, pH 5.6 at 278 K. The phosphorylated or non-phosphorylated ␣M tail peptides were also dissolved in 200 mM perdeuterated DPC to record both TOCSY and NOESY spectra. The mixing time for TOCSY and NOESY experiments was fixed to 50 and 200 ms, respectively. Natural abundance 13 C-1 H heteronuclear single quantum coherence (HSQC) experiments were carried out for Myr-␣M and Myr-P␣M in DPC solution under similar D 2 O buffer conditions to obtain 13 C ␣ chemical shifts. Paramagnetic perturbation studies of the myristoylated or non-myristoylated ␣M tails with or without 200 mM DPC were carried out in 10 mM sodium phosphate buffer, pH 5.6 by acquiring two-dimensional NOESY and TOCSY spectra with 200-and 50-ms mixing times, respectively, in the absence or presence of 1 mM MnCl 2 . The changes in the intensity of intraresidue C ␣ H/HN cross-peaks were normalized to the peak intensity of the respective unperturbed spectra. Standard three-dimensional HNCA, HN(CO)CA, HNCACB, and CBCA(CO)NH experiments were performed with 0.5 mM 15 N/ 13 C-labeled samples of ␣M and ␤2 tails at 298 K to obtain backbone resonance assignments. The interactions between the tails of ␤2 and ␣M or P␣M were determined by 1 H-15 N HSQC experiments. HSQC spectra of 15 N-labeled ␤2 or ␣M tail peptides were acquired in the presence of unlabeled ␣M or phosphorylated ␣M or of unlabeled ␤2 tail peptides, respectively. Samples were prepared by dissolving lyophilized powder of tail peptides in 10 mM sodium phosphate buffer, pH 6.5. The chemical shift changes, after addition of the unlabeled binding partners, of the backbone amide nitrogen and amide proton of each residue were calculated using the following equation: ⌬ 1 H ϩ ⌬ 15 N where ⌬ 1 H is the absolute value of the change in amide proton chemical shift and ⌬ 15 N is the absolute value of the change in backbone amide nitrogen chemical shift.
Circular Dichroism (CD) Studies-Circular dichroic measurements were performed by a Chirascan CD spectrophotometer (Applied Photophysics, Leatherhead, UK) using a cuvette of 0.1-cm path length (Hellma, Müllheim, Germany). CD spectra of 0.2 mM Myr-␣M and Myr-P␣M in 10 mM sodium phosphate buffer, pH 6.5, containing 60 mM DPC and of 0.3 mM synthetic ␣M in 10 mM sodium phosphate buffer, pH 6.5 were acquired from wavelengths of 190 -240 nm at 0.1-s time steps. CD spectra were averaged over three scans and subtracted from the corresponding background buffer contributions. The final CD spectra were converted to molar ellipticity.
Structure Determination and Docking-Three-dimensional structures of Myr-␣M and Myr-P␣M were determined using CYANA 2.1 (35). Interproton distance restraints were derived from strong, medium, and weak NOE intensities translated to upper bound distance limits of 2.5, 3.5, and 5.0 Å, respectively. Backbone dihedral angle (, ) restraints were determined based on 13 C ␣ chemical shifts using PREDITOR (36). Of the 100 structures, the 20 lowest energy structures were selected for evaluation and analyses. PROCHECK-NMR (37) was used to evaluate the stereochemical quality of the structure ensembles, and structures were visualized using PyMOL, MOLMOL, and Discovery Studio 2.0. RosettaDock (38) was used to generate low energy models of the complexes between the ␣M tails and ␤2 tail.
FRET Studies-All fluorescence studies were conducted using a 0.5-cm path length luminescence spectroscopy microcell (PerkinElmer Life Sciences) on a Varian Cary Eclipse spectrophotometer. Peptide samples were prepared in 10 mM sodium phosphate buffer, pH 6.5 containing 1.2 mM DPC. FRET experiments were typically conducted by collecting fluorescence emission spectra of intrinsic Trp residue of Myr-␤2 tail mutated (␤2H5W) or Myr-␤2 at 1 M concentration in the presence of various concentrations (ranging from 1 to 5 M) of dansylated and unphosphorylated ␣M (Dan-␣M) and phosphorylated ␣M (Dan-P␣M) tail peptides. Samples were excited at a wavelength of 295 nm, and emission was monitored from 310 to 400 nm. The extent of quenching of Trp emission (resulting from FRET) of Myr-␤2H5W at 355 nm was used to determine the dissociation constants (K d ) for the interactions between the ␤2 and ␣M tail peptides following a standard nonlinear least square fit. Additional FRET studies were also performed for the non-myristoylated ␤2H5W in 10 mM sodium phosphate buffer, pH 6.5 and myristoylated ␤2H5W in 50% acetonitrile, buffer mixture with dansylated ␣M peptides.

Effect of Myristoylation on Conformational Stabilization and
Localization of ␣M Cytosolic Tails-We used a myristoylation strategy, i.e. covalently attaching a myristic acid to the N terminus of the ␣M cytoplasmic tail, to determine three-dimensional structures in the DPC lipid micelles. The long acyl chain (C 14 ) of myristic acid, a probable mimic of the transmembrane domain, may stabilize conformations of peptides (39,40) in the presence of detergent lipids as demonstrated for the cytoplasmic tail of the ␣IIb of platelet-specific ␣IIb␤3 integrin (39). Furthermore, the acylated cytosolic tails of integrins were found to be biologically active (40 -42). Fig. 1 shows two-dimensional 1 H-1 H NOESY spectra correlating low field resonances (6.5-9.0 ppm) with upfield resonances (0.8 -4.5 ppm) of the 24-residue synthetic ␣M tail peptides (Table 1) in Myr-␣M (Fig. 1A) and Myr-P␣M (Fig. 1B) in DPC micelles at 308 K and non-myristoylated ␣M in aqueous solution at 278 K (Fig. 1C). As evident from the figure, there are dramatic differences in the NOE connectivities observed for the Myr-␣M (Fig. 1A), Myr-P␣M (Fig. 1B), and ␣M (Fig. 1C). There were lesser NOE connectivities in the NOESY spectrum of ␣M tail in free solution even at 278 K, particularly among the backbone/side chain and side chain/side chain resonances (Fig. 1C). By contrast, the NOESY spectra of the Myr-␣M and Myr-P␣M demonstrated an overwhelming number of NOE connectivities in solutions containing DPC micelles (Fig. 1, A and B). The paucity of NOE interactions observed for the non-myristoylated ␣M tail in aqueous solution indicates a lack of significant populations of folded conformations. The structural transitions of the ␣M cytosolic tail peptides were also examined by far-UV CD spectroscopy (supplemental Fig. S1). The myristoylated peptides Myr-␣M and Myr-P␣M in DPC micelles showed negative CD bands at 225 and 208 nm diagnostic of helical structure (43). By contrast, far-UV CD spectra of the non-myristoylated ␣M peptide delineated an intense CD band of negative ellipticity at ϳ200 nm, indicating random conformations (43).
To ascertain conformational stabilization of ␣M tail peptides by myristoylation and micelle incorporation, NOESY spectra were also acquired for the non-myristoylated and myristoylated ␣M tail peptides either in DPC-containing solutions (supplemental Fig. S2) or in aqueous solutions, respectively (supplemental Fig. S3). Again, the myristoylated ␣M tail peptides in the presence of DPC micelles yielded a greater number of NOE connectivities in comparison with the non-myristoylated tail peptides in DPC or myristoylated tail peptides in aqueous solutions. The NMR spectra of the myristoylated tail peptides in aqueous solutions appeared to be significantly broadened (supplemental Fig. S3). Such broad resonances may potentially arise as a result of aggregations of the myristoylated peptides in aqueous solutions. In other words, the myristoylated peptides may be improperly folded in the absence of DPC detergent solution. Therefore, the presence of a large number of NOE connectivities for the Myr-␣M and Myr-P␣M establishes that the cytoplasmic tails of ␣M, either non-phosphorylated or phosphorylated at residue Ser 14 , acquire well defined conformations in the presence of DPC micelles.
We further probed whether the myristoylated peptides Myr-␣M and Myr-P␣M were inserted into the DPC micelles using NMR paramagnetic relaxation enhancement (PRE) experiments. In these studies, two-dimensional 1 H-1 H NOESY spectra were acquired for the myristoylated peptides in DPC micelles either in the absence or in the presence of 1 mM MnCl 2 , and changes in the intensity of NH/C ␣ H cross-peaks were measured (see "Experimental Procedures"). It should be noted that MnCl 2 would exhibit a PRE effect on those amino acid residues that are exposed to the aqueous solution (44). As can be seen, there was a dramatic diminution of the intensity of NH/C ␣ H correlations for most of the residues of Myr-␣M and Myr-P␣M peptides, indicating their proximity to the paramagnetic Mn 2ϩ ions (Fig. 2). However, a relatively restricted diminution of intensity (Յ40%) of NH/C ␣ H cross-peaks was detected for a few N-terminal residues, Leu 2 -Phe 5 , suggesting a limited exposure of these residues to the paramagnetic ions (Fig. 2). The relatively low PRE effect of the N-terminal residues  DECEMBER 23, 2011 • VOLUME 286 • NUMBER 51 may occur as a consequence of a potential insertion of the ␣M tails into DPC lipid micelles. Note that the N-terminal region of the ␣M tail contains a stretch of conserved hydrophobic and aromatic residues that may be inserted into the non-polar region of the lipid micelles (Table 1). Therefore, PRE studies were performed for the non-myristoylated tail peptides in lipidfree aqueous solutions and DPC-containing solutions (supplemental Fig. S4). Notably, the N-terminal residues of the nonmyristoylated ␣M tails also demonstrated a lower PRE effect in comparison with the C-terminal residues in lipid-free and DPC-containing solutions (supplemental Fig. S4). These observations indicated that the N-terminal non-polar/aromatic residues of the tail peptides are intrinsically less exposed to the paramagnetic ions, plausibly as a result of structure formation or the presence of bulky aromatic side chains. The overlapping and broad resonances observed for Myr-␣M and Myr-P␣M peptides in aqueous solution precluded residue-specific analyses of PRE studies. Regardless, the marked PRE effect experienced by the majority of the residues of Myr-␣M and Myr-P␣M clearly establishes that these residues are placed into the aqueous milieu. Consequently, the N-terminal myristic acid moiety appears to act as a lipid anchor akin to a transmembrane domain, imparting conformational stabilization and aqueous localization for the cytosolic tail of ␣M integrin in DPC micelles.

Structure and Interactions of ␣M␤2 Cytosolic Tails
Analyses of Secondary Chemical Shifts and NOEs-C ␣ H and 13 C ␣ chemical shift deviations from random coil are an indicator of secondary structures of proteins and peptides (45) . Fig. 3 shows the chemical shift deviations of the 13 C ␣ and C ␣ H resonances from their random coil values for Myr-␣M (Fig. 3A), Myr-P␣M (Fig. 3B), and ␣M (Fig. 3C). The chemical shifts were corrected for the nearest neighbor (i Ϫ 2, i Ϫ 1, i ϩ 1, and i ϩ 2) effects (46). In particular, the 13 C ␣ atom would experience a downfield shift in helical structure and an upfield shift in ␤-strand, whereas a reverse trend has been demonstrated for the C ␣ H proton resonances (45). A stretch of at least four or three continuous residues with helical or strand type chemical shift deviations can be assigned as a stable helix or ␤-strand conformation, respectively (45). The 13 C ␣ and C ␣ H resonances of the Myr-␣M and Myr-P␣M tails were found to experience a positive deviation and a negative deviation, respectively, for the N-terminal residues Phe 4 -Ser 14 (Fig. 3, A and B). By contrast, the chemical shift deviations for 13 C ␣ and C ␣ H resonances at the C-terminal half, residues Glu 15 -Gln 24 , of Myr-␣M and Myr-P␣M were largely limited (Fig. 3, A and B). Therefore, the N-terminal segment, residues Phe 4 -Ser 14 , of the cytoplasmic tail irrespective of the phosphorylation at residue Ser 14 (Ser 1127 ) acquires a stable helical structure when inserted into DPC micelles through myristoylation. On the other hand, the C-terminal region, residues Glu 15 -Gln 24 , of the cytoplasmic tail appears to be devoid of regular secondary structures. The chemical shift deviations for the non-myristoylated ␣M tail in free solution were less pronounced in comparison with the myristoylated tails in DPC micelles, indicating a lack of stable secondary structures (Fig. 3C). However, a population of transient helical structures could be inferred for the N-terminal residues of ␣M tail in solution because the C ␣ H and 13 C ␣ resonances showed, albeit limited, negative and positive deviations, respectively (Fig. 3C).
The myristoylated tail peptides Myr-␣M and Myr-P␣M demonstrated a large number of NOE contacts involving backbone/backbone, backbone/side chain, and side chain/side chain resonances in DPC micelles. Analyses of NOESY spectra revealed medium range C ␣ H/NH NOEs (i to i ϩ 2, i ϩ 3, and i ϩ 4) and NH/NH NOEs (i to i ϩ 1and i ϩ 2) for the N-terminal residues both for Myr-␣M (supplemental Fig. S5) and Myr-P␣M peptides (supplemental Fig. S5), indicating stable helical  conformation for this segment. The cytoplasmic domain of ␣M in free solution yielded fewer NOE contacts in comparison with myristoylated peptides in DPC micelles (supplemental Fig. S5). However, the N-terminal residues of ␣M tail showed a number of medium range C ␣ H/NH NOEs evidencing the plausible existence of a population of helical conformations (supplemental Fig. S5). Furthermore, several side chain/side chain NOEs were identified for the Myr-␣M and Myr-P␣M peptides in DPC micelles. The N-terminal helical regions of Myr-␣M and Myr-P␣M in DPC micelles were defined by medium range interside chain NOE contacts between aromatic/aromatic and aliphatic/ aromatic residues (supplemental Fig. S6). Notably, the aromatic ring protons of Phe 5 of Myr-␣M (supplemental Fig. S6) and Myr-P␣M (supplemental Fig. S6) showed NOE interactions with the ring proton resonances of residue Tyr 9 . In addition, side chain/side chain NOEs were present for Myr-␣M and Myr-P␣M peptides between residues Leu 2 /Phe 5 and residues Tyr 9 / Met 12 /Met 13 (supplemental Fig. S6). Interestingly, a number of long range NOEs could be detected among residues Tyr 9 /Gln 24 , Met 12 /Gln 24 , Met 13 /Gln 24 , Met 12 /Pro 18 , Ser 14 /Gln 24 , and Met 12 /Glu 22 for the Myr-␣M peptide (Fig. 4, top panel), implying packing interactions between the C-terminal loop and N-terminal helical region. There were also long range NOEs, e.g. Met 12 /Gln 24 , Met 13 /Gln 24 , and Ser(P) 14 /Gln 24 , observed for the Myr-P␣M tail peptide. However, more such long range NOEs could be identified for Myr-␣M tail in comparison with Myr-P␣M peptide (supplemental Table S1), plausibly indicating diminished interactions between the helix and the loop region.

Solution Structures of Myr-␣M and Myr-P␣M Cytoplasmic Tail Peptides-Three-dimensional structures of Myr-␣M and
Myr-P␣M were determined by use of NOE-driven distance and backbone dihedral angel (, ) restraints (Table 2). Fig. 5 shows a superposition of all backbone atoms (C ␣ , N, and CЈ) of the 20 lowest energy structures of Myr-␣M (Fig. 5A) and Myr-P␣M (Fig. 5B). The root mean square deviation values from the mean  Table 2. The three-dimensional structure of the Myr-␣M tail peptide revealed a helical conformation encompassing residues Phe 4 -Ser 14 at the N terminus, whereas residues Glu 15 -Gln 24 at the C terminus acquire a looplike structure (Fig. 5C). The C-terminal loop folds back onto the helical structure, yielding a compact conformation of the Myr-␣M tail (Fig. 5D). The N-terminal helical structure of the Myr-␣M tail appears to be amphipathic in nature whereby the polar face is determined by the side chains of residues Lys 6 , Arg 7 , Lys 10 , Asp 11 , and Ser 14 (Fig. 5C). There are potential ionic interactions and/or hydrogen bonding between residues Arg 7 and Asp 11 and residues Lys 10 and Ser 14 at the polar face of the helix (Fig. 5C). On the other hand, the non-polar face of the N-terminal helix is characterized by the mutual packing interactions among the side chains of aromatic and aliphatic residues, e.g. Phe 4 , Phe 5 , Tyr 9 , Met 12 , and Met 13 (Fig. 5D). The specific positioning of the C-terminal region of the Myr-␣M tail in the three-dimensional structure essentially created a non-polar surface involving van der Waals packing interactions among residues Met 12 and Met 13 from the helix and residues Pro 18 and Pro 23 and methylene groups of residue Gln 24 from the C-terminal loop (Fig. 5D). The N-terminal helical structure, residues Phe 4 -Ser 14 , also was demonstrated to be preserved in the Myr-P␣M (Fig. 5E). The disposition of polar and non-polar side chains in the helical structure of Myr-P␣M was found to be amphipathic, akin to Myr-␣M (Fig. 5E). In the context of the phosphorylation of Ser 14 , there could be ionic interactions between the cationic side chain of residue Lys 10 and the negatively charged phosphate group of Ser(P) 14 in the helical structure of Myr-P␣M (Fig. 5F). However, the tertiary packing interactions between the C-terminal loop and N-terminal helix observed in the Myr-␣M tail appear to be largely disrupted in the Myr-P␣M tail (Fig. 5F). We found only a few NOE interactions between the residues at the loop and in the helix for the Myr-P␣M tail, resulting in poorly defined structural organization of the loop.
Interactions between ␣M and ␤2 Tails by 15 6A and 7, A and B), indicating interactions between the cytosolic tails of ␣M and ␤2. However, the binding affinity between the tail peptides appeared to be low as conspicuous spectral changes in 15 N-1 H HSQC were observed only in the presence of a higher molar concentration of unlabeled tails (supplemental Fig. S7). Similar observations were also made for the NMR interaction studies between tails of ␣IIb/␤3 and ␣L/␤2 integrins (33,47).
Combined chemical shift changes for 1 HN and 15 N resonances of ␣M are shown as a function of residue (Fig. 6B). As can be seen, a number of residues of the ␣M tail, namely Leu 2 , Arg 7 , Lys 10 , Asp 11 , Met 12 , Met 13 , and Ser 14 , showed higher changes in chemical shifts (Ն10 Hz) as compared with the others, indicating their plausible involvement in interactions with the ␤2 cytoplasmic tail (Fig. 6B). Notably, all of these residues are located in the helical region of the ␣M tail (Fig. 6C). To map the interacting residues of the ␤2 tail, 15 N-labeled ␤2 tail was titrated with either unlabeled ␣M (Fig. 7A) or phosphorylated ␣M (Fig. 7B) tail peptides. Residues H 5 , Leu 6 , Glu 11 , and Tyr 12 of the cytoplasmic tail of ␤2 demarcated binding-induced chemical shift changes for the ␣M tail (Fig. 7C) and its phosphorylated variant (Fig. 7D).
Interactions between ␤2 and ␣M Tails by FRET in DPC Micelles-It is likely that membrane anchoring of the cytoplasmic tails by N-terminal myristoylation increases binding affin- ity between the tails as a result of structural stabilization. We utilized FRET using a dansyl group (as acceptor) and intrinsic Trp (as donor) to detect binding between the tails in the presence of DPC micelles. For these assays, the ␣M tail and the phosphorylated ␣M tail peptides were conjugated at their N termini with a dansyl group, whereas a ␤2 tail analog, termed Myr-␤2H5W, was prepared with an N-terminal myristoylation and replacement of two amino acid residues ( Table 1). The residues His 5 and Trp 24 of the native ␤2 were substituted with Trp and Ala, respectively, in the Myr-␤2H5W analog to observe FRET unequivocally. tail, demonstrating a plausible FRET between the Trp and dansyl group of ␤2 and ␣M peptides, respectively. In other words, the dansylated ␣M tail peptides are in close proximity (Յ30 Å) with the lipid-anchored ␤2 tail peptide. Note that to achieve an efficient FRET between the tail peptides the concentrations of DPC were maintained slightly above the critical micelle concentrations (see "Experimental Procedures"). A lower detergent concentration may enhance a close proximity between the donor and acceptor peptides (48).  FRET experiments were also carried out with Dan-␣M or Dan-P␣M tail peptides and a non-myristoylated analog of ␤2H5W in detergent-free aqueous solutions. The quenching of Trp fluorescence of the non-myristoylated analog of ␤2H5W peptide was found to be highly limited upon additions of either Dan-␣M or Dan-P␣M peptides, indicating a lack of FRET under such conditions (supplemental Fig. S8). This observation could be attributed to a rather low affinity binding between the tails in the absence of lipid tethering. Therefore, the quenching of Trp fluorescence emission intensity observed for the micelleinserted Myr-␤2H5W peptide might occur because of an increase in binding affinity between the tails. The changes in Trp fluorescence intensity were used to determine equilibrium dissociation constant (K d ) values of the interactions between  Myr-␤2H5W and dansylated ␣M tail peptides (Fig. 8, C and D).  Fig. S8). FRET experiments were also performed for Myr-␤2H5W and dansylated ␣M tails in aqueous solution without DPC micelles. The Myr-␤2H5W had a highly limited solubility in aqueous buffer and was soluble only after addition of organic solvents. There was no detectable FRET between Myr-␤2H5W and Dan-␣M or Dan-P␣M under such solution conditions, suggesting that myristoylation per se does not favor a tail-tail complex formation (supplemental Fig.  S9).
Collectively, the FRET experiments delineated that the binding interactions between the cytoplasmic tails might be enhanced while they are inserted into membrane lipids. In addition, FRET studies established that the N-terminal regions of ␤2 and ␣M tails are likely to be in close proximity in the tail-tail heterocomplex.
Mode of Interactions of ␣M and ␤2 Tails-Molecular models of complexes between unphosphorylated and phosphorylated ␣M tails and the ␤2 tail were obtained by iterative docking of ␣M and ␤2 structures based on 15 N-1 H chemical shift changes. These models were further energetically refined using the RosettaDock protocol ("Experimental Procedures"). The membrane-proximal N-terminal region of ␤2 or ␤3 cytoplasmic tails had been shown to adopt helical conformations in previous studies (33,47,49,50). The ␣M-␤2 tail complexes are characterized by intimate interactions between the helical structures, which are orientated in a parallel manner (Fig. 9). A parallel orientation of the two interacting helices is consistent with the FRET studies between Myr-␤2H5W and dansylated ␣M tails that indicated proximity between the N termini of the tails. In the full-length native integrins, the two subunits of the heterodimer assume a parallel orientation (1). In the structures of the model complexes, the residues at the loop of the ␣M tails are distally placed from the interface and therefore do not  DECEMBER 23, 2011 • VOLUME 286 • NUMBER 51 occlude the interactions with the ␤2 tail (Fig. 9, A and B). As can be seen, the tail-tail complexes may be sustained through a number of side chain-side chain interactions that are predominantly ionic and/or polar in nature (Fig. 9, A and B). In addition, hydrophobic packing interactions are plausible among the conserved N-terminal aromatic residues Phe 4 and Phe 5 of the ␣M or phosphorylated ␣M tail and residues Leu 3 and Ile 4 of the ␤2 tail (Fig. 9, A and B). At the interfacial regions, the side chain of residue Asp 8 of ␤2 and residues Arg 7 and Lys 10 of ␣M or phosphorylated ␣M are found to be in close proximity, implying a potential salt bridge and/or hydrogen bonding interactions between the tails (Fig. 9, A and B). The carboxylate side chain of residue Glu 11 of ␤2 is also in close contact with the cationic side chain of residue Lys 10 of ␣M (Fig. 9A) and phosphorylated ␣M (Fig. 9B). Furthermore, residues Arg 10 and Arg 14 of ␤2 appear to be engaged in a network of multiple ionic and hydrogen bonding interactions with residues Asp 11 and Glu 15 of the ␣M tail in the ␤2-␣M tail complex (Fig. 9A). For the phosphorylated ␣M and ␤2 tail complex, the phosphate moiety of Ser(P) 14 is found to be in close proximity with the guanidinium group of residue Arg 14 of the ␤2 tail, demarcating ionic and/or hydrogen bond interactions (Fig. 9B). However, the side chains of Arg 10 and Arg 14 of the ␤2 tail are only in close contact with the side chains of residues Asp 11 and Glu 15 , respectively (Fig. 9B). It may be noteworthy that the introduction of a bulky phosphate group at the side chain of residue Ser 14 may render a rearrangement of potential ionic interactions between the phosphorylated ␣M and ␤2 tails.

DISCUSSION
In humans, there are 24 different integrins formed by specific combinations of 18 ␣ subunits and eight ␤ subunits (1-3). Many integrins share a common ␤ subunit but different ␣ subunits. Conceivably, signaling specificity in these integrins is governed by their ␣ cytoplasmic tails. The membrane-proximal regions of the ␣ tails are well conserved, but the membranedistal regions significantly vary in lengths and sequences. The latter may permit for structural variations and can serve as distinct docking sites for cytoplasmic molecules. Notably, integrin ␣2␤1 (VLA-2) and integrin ␣4␤1 (VLA-4) exhibited altered functional behaviors when their ␣ tails were replaced by tails from other integrins (51,52). The ␣ tails of integrin ␣L␤2 and ␣M␤2 might be responsible for distinct chemokine-induced activation kinetics and selective interactions of the Src kinase Hck with ␣M␤2 (26). Different ␣ tails are also known to exhibit specific associations with cytoplasmic proteins, e.g. ␣5 with nischarin (53), ␣4 with paxillin (54), ␣IIb with calcein integrinbinding protein (55), and ␣L with CD45 cytoplasmic domain (56). By contrast, the ␤ tails with the exception of ␤4 have high sequence conservation with two NPX(Y/F) motifs that are docking sites for talins, DOK, and kindlins (supplemental Fig.  S10) (57)(58)(59)(60).
To date, the two ␣ tails with known structures are ␣IIb (40) and ␣L (33). The NMR structure of an N-terminal Myr-␣IIb tail peptide embedded in DPC micelles was determined previously (40). The ␣IIb tail has a short N-terminal helical segment (Val 990 -Arg 997 ) followed by a loop in its C-terminal half that packs onto the surface of this helix. This conformation is stabi-lized by salt bridges formed by residues Lys 994 and Arg 997 of the N-terminal helix and the C-terminal acidic residues 1001 EED-DEEGE 1008 (40). Unlike the ␣IIb tail, NMR analyses have revealed that the ␣L tail in aqueous solution is characterized by a compact structure maintained by mutual interactions among three helical segments (33). Notably, the folded structure of the ␣L tail contains a large negatively charged surface that can bind divalent metal ions, e.g. calcium, that is similar to the metal ion binding capacity of the negatively charged loop of the ␣IIb tail (33). In this study, we determined NMR structures of the Myr-␣M tail in DPC micelles. The Myr-␣M tail acquires an amphipathic helical structure for the N-terminal residues Leu 1114 -Glu 1127 (Leu 2 -Glu 15 in the ␣M tail peptide used herein) and a loop for residues Gly 1128 -Gln 1136 (Gly 16 -Gln 24 in the ␣M tail peptide used herein) at the C-terminal region (Fig.  5). The C-terminal loop folds back onto the helical structure via non-polar packing. The overall structure of the ␣M tail is similar to that of ␣IIb despite a lack of sequence homology except for their N-terminal membrane-proximal regions (supplemental Fig. S10). A difference between the two structures is that the C-terminal loop fold-back of the ␣M tail is mediated by hydrophobic packing, whereas for the ␣IIb tail, helix-loop stabilizations are mediated by electrostatic interactions. In addition, the N-terminal helix of the ␣M tail is longer (residues Leu 1114 -Glu 1127 ) than the ␣IIb tail (residues Val 990 -Pro 998 ).
The Ser 1126 (Ser 14 in the ␣M tail peptide herein) of the ␣M tail has been reported to be constitutively phosphorylated, and it is required for inside-out activation of ␣M␤2 (31). Thus, we also determined the structure of the Myr-␣M tail with Ser(P) 1126 in DPC micelles (Fig. 5). Phosphorylation of Ser 1126 did not induce large conformational changes in the ␣M tail, but there was lesser packing of the loop region with the N-terminal helix (Fig. 5). Phosphorylation is often found to be a "conformational switch" in proteins and peptides involved in signaling and other cellular functions (61)(62)(63). In particular, structural disruptions (either global or local) have been reported for signaling proteins upon phosphorylation of Ser or Thr (64 -66). We did not observe significant perturbations in the ␣M tail with phosphorylated Ser 1126 as compared with the unmodified ␣M tail, and there was no detectable difference in their binding affinities to the ␤2 tail. Thus, Ser 1126 phosphorylation is unlikely to modulate directly the conformation of the ␣M tail or the packing of the ␣M and ␤2 tails. However, it may present a docking site for cytoplasmic protein that can perturb either the structure of the ␣M tail or the interaction of the ␣M, ␤2 tails, or both that regulates the activation of ␣M␤2.
The interaction between the ␣ and ␤ tails of the integrin and the packing of the TMs are required to maintain integrin in a resting state (47,(67)(68)(69)(70). Surface plasmon resonance analyses showed that the ␣IIb and ␤3 tails form a complex with a K d of 7.7 or 50 M in the presence or absence of calcium, respectively (71,72). In the presence of calcium, the NMR-derived structure of the ␣IIb and ␤3 tails in association revealed ionic/polar interactions involved in the packing of their N-terminal helices (47). However, the interactions between ␣IIb and ␤3 tails were not detected by others (49,73). What accounts for this disparity in observations is unclear. In our previous study, a K d of ϳ2.6 M was determined for the complex formation of ␣L and ␤2 tails in the presence of calcium, and a larger surface of the ␣L trihelical fold contributes to its interactions with the ␤2 tail (33). Collectively, these studies suggest that the interactions between integrin ␣ and ␤ tails may be dependent on the structure of the ␣ tail, which can be significantly different among the ␣ tails because of length, sequence divergence, and modulation by metal ions.
In this study, non-Myr-␣M tail with or without Ser 1126 phosphorylation in aqueous buffer solution interacted much more weakly (K d values greater than mM) with the ␤2 tail in comparison with the ␣L␤2 and ␣IIb␤3 systems. 15 N and 1 HN chemical shift changes in HSQC spectra could only be seen at higher concentrations of unlabeled binding partners (Figs. 6 and 7). Regardless, these data provided information on the residues of the ␣M and ␤2 tails that are perturbed by complex formation. We also examined whether the ␣M tail can interact with divalent metal ions. However, there was no detectable binding to calcium (data not shown). Interestingly, FRET studies demonstrated that the binding affinity between cytoplasmic tails could be increased significantly when one of the cytoplasmic tails was inserted into lipid micelles. This suggests that the interactions between the ␣ and ␤ tails of an integrin can be stabilized when they are anchored into lipid membranes through the TMs. The energy-minimized molecular models of the ␣M-␤2 and P␣M-␤2 were obtained by RosettaDock in the absence of lipid micelles. The exact orientations of the tail-tail complexes within DPC micelles are not clear at this moment. However, the docked structures provide molecular insights into integrin activation and regulation of cytosolic tails.
Based on the docked structures of the ␣M and ␤2 tails (Fig.  9), ionic and/or van der Waals interactions of the tails are possible. A comparison with the interfacial contacts of the ␣IIb␤3 (supplemental Fig. S11) and ␣L␤2 (supplemental Fig. S11) with the ␣M␤2 tails demarcated a conserved salt bridge interaction at the membrane-proximal region. However, the docked structure of the ␣M␤2 tails (and ␣L␤2 tails) showed more interactions distal to the membrane-proximal region as compared with that of ␣IIb␤3 (supplemental Fig. S11). It is likely that the interactions between the TMs of ␣ and ␤ subunits of integrins in lipid membranes may potentially impart stabilizing interactions between the cytosolic tails. An NMR study using peptides of ␣IIb␤3 TMs with membrane-proximal tail sequences revealed interactions between the tails (70). We conjecture that in an intact ␣M␤2 the packing of its TMs and these interactions of its tails together constrain the receptor in a resting state. To better understand the detailed mechanism of ␣M␤2 inside-out activation, future NMR studies using ␣M and ␤2 TM-tail peptides in lipid micelles can be performed. In addition, it will be interesting to identify a cytoplasmic protein(s) that preferentially binds the Ser 1126 phosphorylated ␣M tail.