NMR Structure of F-Actin Binding Domain of Arg/Abl2 from Homo sapiens

INTRODUCTION

The Northeast Structural Genomics Consortium has used bioinformatics methods to construct a Human Cancer Pathway Interaction Network (HCPIN),¹ a comprehensive 3D structure-function database of human-cancer-associated proteins and protein complexes in the context of their interaction networks. The FABD domain of Arg (Abl-related gene; Abl2) was selected as NESG HCPIN target HR5537. Arg and Abl (Abelson tyrosine kinase; Abl1), the Abl non-receptor tyrosine kinases, link diverse cell surface receptors to the regulation of cytoskeletal dynamics and regulate cytoskeletal reorganization, cell proliferation, survival, adhesion, migration and stress responses in multiple cells types.^2–7 Abl and Arg kinases are multi-domain proteins with highly conserved Src kinase homology 3 (SH3), SH2, kinase (SH1) domains in the N-terminal half. The C-terminal halves of these kinases are more divergent, however, the functions encoded by the C-terminus are critical for the overall functions of these proteins.⁴ Although Abl and Arg exhibit overlapping expression in many tissues, Arg is most highly expressed in brain, thymus, spleen, and muscle.⁸ Differences in regulation of cell migration by Abl and Arg have also been reported.⁷ Consistent with the nuclear and cytoplasmic localization of Abl and the predominant cytoplasmic localization of Arg, three nuclear localization signals (NLS), one nuclear export signal NES motifs and a DNA-binding domain are found in Abl but not in Arg.⁴ Abl and Arg share a C-terminal calponin homology F-actin-binding domain (FABD) with ~44% sequence identity, which distinguishes Abl family kinases from other non-receptor tyrosine kinases.^4,9,10 Preceding this shared FABD, Arg has a microtubule-binding (MT) domain and a second talin-like F-actin-binding domain that is characterized by an I/LWEQ sequence while Abl kinase has a globular (G)-actin binding domain.^4,9,10,11 Arg uses its FABD to anchoring actin and other cytoskeletal partner for signal transfer and other biological functions.⁴ Both human Abl FABD and Arg FABD belong to the F_actin_bind protein domain family (Pfam¹² entry PF08919) comprised of 21 sequences. The NMR structure of the human Abl FABD, the only available structure in F_actin_bind family, has been reported recently¹³ ; it forms a compact left-handed four helix bundle in solution. The Arg FABD was selected as HCPIN target by NESG for structure determination.¹ The NMR structure of human Arg FABD reported here can serve as a structural basis for elucidating the molecular mechanism of Arg pathway, for studies of protein complex formation, and potentially in small molecule drug design.

MATERIALS AND METHODS

The Arg (Abl2) F-actin binding domain (FABD) from Homo sapiens (UniProtKB/Swiss-Prot ID P42684/ABL2_HUMAN, residues 1058–1182) was cloned, expressed and purified following standard, largely automated NESG protocols to produce a uniformly ¹³C, ¹⁵N-labeled protein sample.¹⁴ Briefly, the truncated ABL2_HUMAN (1058–1182) gene was cloned into a pET14-15C (Novagen) derivative, yielding the plasmid pHR5537A-14.12. The resulting construct contains 10 nonnative residues at the N-terminus (MGHHHHHHSH) to facilitate protein purification and one single mutation T1062A was introduced. Escherichia coli BL21 (DE3) pMGK cells were transformed with pHR5537A-14.12, and cultured in MJ9 minimal medium¹⁵ containing (¹⁵NH₄)₂SO₄ and U -¹³C-glucose as sole nitrogen and carbon sources. U-¹³C, ¹⁵N Arg FABD was purified using an ÄKTAxpress™ (GE Healthcare) based two step protocols consisting of IMAC (HisTrap HP) and gel filtration (HiLoad 26/60 Superdex 75) chromatography. The final yield of purified U-¹³C, ¹⁵N Arg FABD (> 98% homogeneous by SDS-PAGE; 15.2 kDa by MALDI-TOF mass spectrometry) was ~100 mg/L. In addition, a U-¹⁵N and 5% biosynthetically directed fractionally ¹³C-labeled sample¹⁶ was generated for stereo-specific assignment of isopropyl methyl groups. Both U-¹³C,¹⁵N and 5%¹³C,U-¹⁵N Arg FABD were dissolved, respectively, at concentrations of ~1.2 mM and 1.4 mM in 95% H₂O/5% D₂O (20 mM MES, 200 mM NaCl, 10 mM DTT, 5 mM CaCl₂, 0.02% NaN₃) at pH 4.5. Analytical gel filtration and static light scattering data¹⁴ indicates that this domain is monomeric in solution under the conditions used for these NMR studies.

All NMR spectra were recorded at 25 °C. Triple resonance NMR data (3D HNCO, 3D HNCACB, 3D CBCAcoNH, and (4,3)D GFT H^αβC^αβcoNHN¹⁷) were collected on Varian INOVA 600 MHz, a simultaneous 3D ¹⁵N/¹³C^aliphatic/¹³C^aromatic-edited NOESY ¹⁸ spectrum (mixing time 100 ms) in H2O and a 3D ¹³C-edited NOESY in D2O were acquired on a Bruker AVANCE 800 MHz spectrometer. 2D constant-time [¹³C, ¹H]-HSQC spectra with 28 ms and 42 ms constant-time delays were recorded for the 5% biosynthetically directed fractionally ¹³C-labeled sample on a Varian INOVA 600 MHz spectrometer equipped with a cryogenic probe in order to obtain stereo-specific assignments for isopropyl groups of valines and leucines.¹⁶ All NMR data were processed using the program NMRPipe ¹⁹ and analyzed using the program XEASY.²⁰ Spectra were referenced to external DSS. Resonance assignments were achieved as described previously.²¹ Sequence specific assignments (H^N, H^α, N, C^α) and H^β/C^βassignments were obtained largely automated with the program AUTOASSIGN.²² These assignments were then used to simulate a NOESY peak list that was used to aid interactive side-chain assignments. The simultaneous 3D ¹⁵N/¹³C^aliphatic/¹³C^aromatic-edited NOESY and CCH-TOCSY spectra were then analyzed manually to obtain nearly complete side-chain assignment. Assignments were obtained for 90% of backbone and side-chain chemical shifts assignable with the NMR experiments listed above (excluding N-terminal NH₃⁺, Lys NH₃⁺, Arg NH₂, OH of Ser, Thr and Tyr, ¹³C^γ of Asp and Asn, ¹³C^δ of Glu and Gln, and aromatic ¹³C^γ shifts). Chemical shifts were deposited in the BioMagResBank on 06/14/2009 with accession code 16349.

Based on chemical shifts, the locations of regular secondary structure elements were identified.²³ A NOESY peak list containing expected intra-residue, sequential and α-helical medium range NOE peaks was initially generated and was manually edited by visual inspection of the simultaneous NOESY spectra, and subsequent manual peak picking was pursued to identify remaining, primarily long-range NOEs.²¹ The programs CYANA ^24,25 and AUTOSTRUCTURE²⁶ were used in parallel to automatically assign long-range NOEs. Assignments identically obtained by both programs (‘consensus assignments’) were retained and established the starting point for iterative cycles of noise/artefact peak removal, peak picking, NOE assignment and structure calculation.^{21 1}H - ¹H upper distance limit constraints for structure calculations obtained from both NOESY were summarized in Table I. In addition, backbone dihedral angle constraints were derived from chemical shifts using the program TALOS+²⁷ for residues located in well-defined secondary structure elements (Table 1). The final structure calculation was performed with CYANA 3.0, and the 20 conformers with the lowest target function value were refined in an ‘explicit water bath’ ²⁸ using the program CNS.²⁹ The coordinates were deposited in the Protein Data Bank on 06/14/2009 (accession code 2KK1).

Table 1.

Statistics of Arg FABD NMR Structure Determination.

Conformationally-restricting distance constraintsa
Distance constraints
Intra-residue (i = j)	634
Sequential (|i– j| = 1)	1017
Medium range (1 < |i – j| < 5)	1270
Long range (|i – j| ≥ 5)	961
Total	3882
Hydrogen bond constraintsb	84
Dihedral angle constraints	172
φ	86
ϕ	86
Number of constraints per residuec	31.3
Number of long range constraints per residuec	7.3
CYANA target function [Å2]	0.51 ± 0.06
Average r.m.s.d. to the mean CNS coordinates [Å]
regular secondary structure elements d backbone heavy atoms N, Cα, C’	0.36 ± 0.06
regular secondary structure, d all heavy atoms	0.68 ± 0.04
Residues 1078–1182, backbone heavy atoms N, Cα, C’	0.52 ± 0.09
Residues 1078–1182, all heavy atoms	0.79 ± 0.08
AutoQF33 R/P/F/DPe	97/94/96/82
Ramachandran plot summary for all ordered (by PSVS) f
most favorable regions	95.0
additionally allowed regions	4.9
generously allowed regions	0.0
disallowed regions	0.0
Global quality scores (raw / Z - score) f
VERIFY3D	0.30 / −2.57
PROSAII	0.55 / −0.41
PROCHECK G-factor ϕ and φ34	0.30 / 1.49
PROCHECK G-factor all dihedral angles	0.14 / 0.83
MOLPROBITY35 Clash Score	18.20 / −1.60

^aAnalysed for all residues included HisTag .

^bHelical (i, i+4) hydrogen bond constraints were inferred from structures by using program CYANA, and were applied only in the final refinement.

^cThere are 132 residues with conformationally restricting constraints, first three residues of HisTag was excluded.

^dRegular secondary element: alpha residues 1086–1099, 1106–1123, 1130–1152, 1165–1181

^eRPF scores³³ (Recall, Precision, F-measurement and DP) reflecting the goodness-of-fit of the final ensemble of structures (including disordered residues) to the NMR data.

^fCalculated using the protein structure validation software suite (PSVS) 1.3 program, order residues 1065–1066, 1071–1154, 1159–1181, defined based on dihedral angle order parameters S(φ) and S(ψ) > 0.90. Z-scores were computed relative to corresponding structure quality measures for high resolution X-ray crystal structures ³⁶.

RESULTS AND DISCUSSIONS

The solution NMR structure of Arg FABD consists of four α-helices α1–α4 (residues 1086–1099, 1106–1123, 1130–1152, 1165–1181) and shows a typical four-helical up-and-down bundle (Fig. 1a & 1b). Structural statistics are given in Table 1; the resulting ensemble of structures exhibits high structure quality assessment scores. The four helices are packed anti-parallel and connected by short “underhand” loops. The N-terminus region from residues 1058–1070 are flexible and disordered, which is consistent with the disorder prediction analysis³⁰ and hetero nuclear NOE data. Some conformers also show a short one turn helix α1’ at residues 1080–1083. Most residues buried in the bundle are hydrophobic. Very few inter-helical salt-bridges are observed and the helix bundle is mainly stabilized by the hydrophobic core. The helical bundles, including the hydrophobic core, are well defined by the extensive inter-helical NOE interactions network observed for both neighbor helices and cross-diagonal helices.

Fig. 1.

Arg FABD shows 44% sequence identity to Abl FABD. The structures of Arg and Abl FABD (PDB 1zzp) are very similar with DALI³¹ Z score of 12.1, and backbone RMSD of 1.9 Å and 1.2 Å for all 93 aligned residues and all consensus helical residues, respectively (Fig.1c).¹³ Residues in the hydrophobic core, important for helical packing, are also highly conserved between Arg and Abl FABDs. On the molecular surface, the residues on N-terminal half of the helix α3 are highly conserved between Arg and Abl FABDs. (Fig1d). These conserved surface residues on the N-terminal half of helix α3 and some adjacent surface residues on helix α2 and α4 have been identified to be responsible for F-actin binding and cytoskeletal association.¹³

Both Arg and Abl FABDs have similar basic charged surface distribution near these conserved F-actin binding area (Fig 1e). The conserved F-actin binding is a potential target area for drug development aimed at inhibiting Arg/Abl function. These structural similarities suggest Arg FABD may bind to F-actin using a similar mechanism as Abl FABD. There are also some obvious differences between Arg and Abl FABD structures. In Arg FABD, the loop between α1 and α2 is proline rich, and four residues shorter comparing to the same loop in Abl FABD. Therefore, the loop between α1 and α2 in Arg FABD is less flexible than the same loop in Abl FABD. The loop between α3 and α4 in Abl FABD is disordered; however, it is shorter and actually ordered in Arg FABD (supported by the hetero nuclear NOE data, data not shown). These different loop regions are not conserved between Arg and Abl FABDs. The helix α3 in Arg FABD extendes two or three more residues on both N- and C-terminal, longer than the helix α3 in Abl FABD. Furthermore, the titled angle of the C-terminal α3 relative to the other three helices in Arg FABD is larger than the corresponding titled angle in Abl FABD.

The helix bundle architecture is emerging as a common interaction module in a number of proteins involved in cytoskeletal regulation.¹³ Structure alignment using the program DALI³¹ reveals significant structural similarity between Arg FABD and several other cytoskeletal protein domains that bind F-actin directly or indirectly, including protein domains with low sequence identity (i.e. most of the DALI ‘hits’ have < 16% sequence identity, except human Abl FABD). These include vinculin head domain (Fig.1f, PDB code 1zvz with DALI Z score 11.6, RMSD 1.9 Å), alpha-catenin (Fig.1f, PDB code 1dow with DALI Z score 11.5, RMSD 2.5 Å), focal adhesion targeting domain (FAT) of focal adhesion kinase (FAK)³² (Fig.1f, PDB code 1ktm with DALI Z score 11.5, RMSD 2.1 Å), talin-1 (Fig.1f, PDB code 2b0h with DALI Z score 10.6, RMSD 2.3 Å). The activated Abl kinases link diverse cell surface receptors to reorganization of the actin cytoskeleton and regulation of chemotaxis, migration, and invasion.⁴ Arg regulate focal adhesion dynamics, and Arg requires both its kinase activity and its cytoskeleton-binding C-terminal half to fully inhibit focal adhesions.^5,7 Similar to the role of FABD in Abl tyrosine kinase, FAT domain is responsible for FAK’s localization.³² The structural similarity of Arg C-terminal FABD and the FAK C-terminal FAT domain indicates a possibility that FABD may compete with FAT domain for binding partners to inhibit focal adhesion.