Solution NMR structures of Immunoglobulin-like domains 7 and 12 from Obscurin-like protein 1 contribute to the structural coverage of the human cancer protein interaction network

Abstract

High-quality solution NMR structures of immunoglobulin-like domains 7 and 12 from human Obscurin-like protein 1 were solved. The two domains share 30 % sequence identity and their structures are, as expected, rather similar. The new structures contribute to structural coverage of human cancer associated proteins. Mutations of Arg 812 in domain 7 cause the rare 3-M syndrome, and this site is located in a surface area predicted to be involved in protein-protein interactions.

Introduction

Cytoskeletal adaptor proteins function by linking cytoskeletal proteins to maintain the structural integrity of cells. Obscurin-like protein 1 (OBSL1; UniProtKB accession number O75147) represents such an adapter protein and constitutes an integral part of the muscle sarcomere, where it is localized at both M- and Z-bands. It interacts with the proteins titin and myomesin [1, 2] and functions in some respects similarly to the protein obscurin (UniProt Q5VST9; hence its name), another very large cytoskeletal adaptor protein [1, 2]. Moreover, OBSL1 controls Golgi and dendrite morphogenesis through regulation of Cullin7 (CUL7) and Fbox/WD repeat containing protein 8 (FBXW8) via ubiquitin ligase signaling [3]. Mutations of OBSL1 [4, 5], CUL7 [6] and the coiled-coil domain containing protein 8 (CCDC8) [7] have been associated with the rare ‘3-M syndrome’ [8] which is manifested in postnatal growth restriction, recognizable facial features and skeletal abnormalities [5, 9]. OBSL1 also interacts with Mitogen-activated protein kinase 11 (MAPK11) and MAPK14 [10]. Both MAPK11 and MAPK14 are involved in cancer associated MAPK and TLR (Toll-like receptor) pathways.

The genes encoding OBSL1 and obscurin are considered to belong to a novel family of homologous genes derived from the Unc-89/OBSL family [1, 2]. OBSL1 is expressed in wide range of tissues as three major isoforms: A (~210 kDa), B (~110 kDa) and C (~150 kDa). While isoforms A and C are present in non-cardiac tissues, isoform B is highly expressed in heart and placental tissues [1]. According to UniProt entry O75147, the full length OBSL isoform A comprises 14 immunoglobulin-like (Ig-like) domains and a single fibronectin type III domain inserted after the fourth Ig-like domain [1]. However, a multiple sequence alignment (Supplementary Material) shows that UniProt Ig-like domain 12 (residues 1357–1534) actually comprises of two Ig-like domains. Furthermore; the NMR structure of a construct comprising residues 615–713 (Protein Data Bank ID: 2E6Q) reveals the presence of another Ig-like domain not listed in UniProt entry O75147. Taken together, protein OBSL1 thus comprises 16 Ig-like domains.

All Ig-like domains belong to Pfam [11] protein family PF07679 (‘Intermediate-set’), which currently contains 1617 members from bacteria and eukaryotic organisms participating in 3030 different unique domain organizations (architectures). Remarkably, the sequence identity among the 16 Ig-like domains varies extensively (between 9 to 67 %, Supplementary Material). Previously, structures were obtained for Ig-like domains 1, 5, 6, and 8 by solving, respectively, the X-ray crystal structure of OBSL1(1–106) in complex with Titin- (M10) (PDB [12] IDs: 2WP3, 2WWK, 2WWM, 3KNB) [13, 14] and the three solution NMR structures of OBSL1(615–713) (PDB ID: 2E6Q), OBSL1(714–804) (PDB ID: 2E6P) and OBSL1(893–992) (PDB ID: 2CPC). These domains exhibit the expected Ig-like fold [15] in which ~8 antiparallel β-strands are arranged in two β-sheets forming a β-sandwich [15].

OBSL1 Ig-like domains 7 and 12 share 30 % sequence identity and were selected by the Northeast Structural Genomics (NESG) Consortium (http://www.nesg.org; target IDs: HR8578K, and HR8578D, respectively) for structure determination with the aim to provide extensive structural coverage for human cancer-associated proteins and protein complexes constituting the ‘Human Cancer Protein Interaction Network’ (HCPIN) [16]. Here we report the high-quality solution NMR structures of OBSL1(805–892) and OBSL1(1277–1357) which comprise Ig-like domains 7 and 12, respectively.

Material and methods

OBSL1(805–892) and OBSL1(1277–1357) were cloned, expressed, and purified following protocols [17–19] established by NESG (see Supplementary Material for details). The corresponding pET expression vectors (NESG HR8578K-805–892-Av6HT.12 and HR8578D-1277–1357-Av6HT.12) have been deposited in the PSI Materials Repository (http://psimr.asu.edu/). [U-¹³C,¹⁵N]-labeled protein samples were prepared at ~1 and ~0.6 mM concentrations for OBSL1(805–892) and OBSL1(1277–1357) in 90% H₂O/10% D₂O containing, respectively, 20 mM MES, along with 100 mM NaCl, 5 mM CaCl₂, 10 mM DTT, 50 µM DSS and 0.02% NaN₃ at pH 6.5 and 10 mM Tris-HCl, 100 mM NaCl, 5 mM DTT, 0.02 % NaN₃, 50 mM DSS at pH 7.5. [5% ¹³C, U-¹⁵N]-labeled samples enabled stereospecific assignments of the methyl groups of Val and Leu residues [20]. Isotropic overall rotational correlation times around 6 ns were inferred for both proteins from backbone ¹⁵N spin relaxation times for both proteins (see Supplementary Material; http://www.nmr2.buffalo.edu/nesg.wiki), indicating that the two proteins are monomeric in solution. This finding was confirmed by static light scattering (Supplementary Figs. S1 and S2).

NMR data were acquired at 25 °C on Varian INOVA 600 and 750 spectrometers equipped with cryogenic ¹H{¹⁵N,¹³C} probes. The total measurement times for OBSL1(805–892) and OBSL1(1277–1357) were ~8 days for both proteins. Nearly complete sequence-specific ¹H, ¹³C and ¹⁵N resonance assignments (Table 1; Supplementary Figs. S3 and S4) were obtained from conventional triple-resonance NMR experiments (see Supplementary Material for details) using the programs AutoAssign [21, 22], followed by manual assignment of side-chain resonances. Resonance assignments were validated using AVS software suite [23]. For OBSL1(805–892), backbone ¹⁵N-¹H RDCs were measured for two alignment media (Supplementary Material). Chemical shifts, NOESY peak lists and all time domain NMR data were deposited in the BioMagResBank [accession numbers 18563 and 18511 for OBSL1(805–892) and OBSL1(1277–1357), respectively].

Table 1.

Statistics^a for NMR Structures of OBSL1(805–892) and OBSL1(1277–1357)

		OBSL1 (805–892)		OBSL1 (1277–1357)
Completeness of resonance assignmentsb
Backbone (%)		99.5		94.8
Side chain (%)		99.1		99.4
Aromatic (%)		92.2		100
Stereospecific methyl (%)		100		100
Conformationally-restricting constraintsc
Distance constraints
Total		1380		1270
intra-residue (i = j)		228		170
sequential (|i−j| = 1)		389		310
medium range (1 < |i − j| < 5)		198		173
long range (|i − j| ≥ 5)		565		617
Dihedral angle constraints		68		64
RDC constraints (15N-1H)		147
No. of constraints per residue		15.8		16.9
No. of long range constraints per residue		6.7		8.2
Residual constraint violationsc
Average no. of distance violations per structure:
0.1 – 0.2 Å		4.45		13.9
0.2 – 0.5 Å		0.7		3.85
> 0.5 Å		0		0
Average no. of dihedral angle violations per structure:
1 – 10°		7.6		11.7
> 10°		0		0
Model Qualityc
RMSD backbone atoms (Å) d		0.6		0.5
RMSD heavy atoms (Å) d		1.1		0.9
RMSD bond lengths (Å)		0.021		0.021
RMSD bond angles (°)		1.1		1.1
RMSD heavy atoms of molecular core (Å) e		0.9		0.8
MolProbity Ramachandran statistics c,d
Most favored regions (%)		96.5		94.4
Allowed regions (%)		3.0		5.3
Disallowed regions (%)		0.6		0.3
Global quality scores (Raw / Z-score)c
Verify3D	0.32	−2.25	0.43	−0.48
ProsaII	0.48	−0.70	0.44	−0.87
Procheck (phi-psi) d	−0.68	−2.36	−0.85	−3.03
Procheck (all) d	−0.45	−2.66	−0.51	−3.02
MolProbity clash score	13.75	−0.83	18.56	−1.66
RPF Scoresf
Recall / Precision	0.993	0.876	0.966	0.832
F-measure / DP-score	0.931	0.860	0.894	0.829
BMRB accession number		18563		18511
PDB ID		2LVC		2LU7

^aStructural statistics computed for the ensemble of 20 deposited structures.

^bComputed using AVS software [23] from the expected number of resonances, excluding: highly exchangeable protons (N-terminal, Lys, and Arg amino groups, hydroxyls of Ser, Thr, Tyr), carboxyls of Asp and Glu, non-protonated aromatic carbons, and the C-terminal His₆ tag.

^cCalculated using PSVS 1.4 [32]. Average distance violations were calculated using the sum over r⁻⁶.

^dBased on ordered residue ranges [S(phi) + S(psi) > 1.8], Residues 807–891 for OBSL1(805–892) and residues 1278–1356 for OBSL1(1277–1357)

^eCalculated using MOLMOL [35] for residues 808,809,811,815,817,819, 825,827–829,831,840,847,854,855, 864, 865,867,869,871,877,879–881,886, 889–892 for OBSL1(805–892) and residues 1277,1279,1281–1283,1287–1291, 1293,1300,1302–1304,1308–1311, 1315,1317,1324–1326,1328,1331,1336, 1338–1340,1342,1343,1347,1349–1354 for OBSL1(1277–1357)

^fRPF scores [24] reflecting the goodness-of-fit of the final ensemble of structures (including disordered residues) to the NOESY data and resonance assignments.

Structure calculations were performed using standardized methods of the NESG consortium [24, 25] (http://www.nmr2.buffalo.edu/nesg.wiki) and consensus analysis of automated NOESY cross peak assignments provided by the programs CYANA [26, 27] and AutoStructure [28]. This analysis was based on ¹H-¹H NOE-derived upper limit distance constraints, and back-bone dihedral angle constraints derived from chemical shifts using the program TALOS+ [29] for residues located in well-defined regular structural elements; orientational constraints were derived for a similar set from polypeptide backbone ¹⁵N-¹H RDCs for OBSL1(805–892) [30]. Stereospecific assignments for methylene protons were performed with the GLOMSA module of CYANA. The final structure calculations were performed on CYANA, followed by refinement in an ‘explicit water bath’ using the program CNS [31]. Structure validation was accomplished using the Protein Structure Validation Software (PSVS) server 1.4 [32], and agreement of structures and NOESY peak lists was verified using the AutoStructure RPF 2.2.2 package [24].

Results and discussion

High-quality solution NMR structures of OBSL1(805–892) and OBSL1(1277–1357) containing OBSL1 Ig-like domains 7 and 12, respectively, were obtained (Table I) and the coordinates were deposited in the PDB [12] on 06/30/2012 (accession code 2LVC) and 06/08/2012 (accession code 2LU7). The domain structures exhibit the expected Ig-like fold in which two antiparallel β-sheets form a sandwich. In both domains, sheet I contains strands β1, β2, β5, and β4, and sheet II contains strands, β3, β6 and β7) (Fig. 1b, 1g) In addition to these β-strands, domain 7 exhibits a β-strand 1` (808–810) contributing to β-sheet-II as well as a 3<sub>10</sub>-helix (871–873) located in the loop connecting β-strands 5 and 6, and domain 12 exhibits an additional β-strand 8 (Fig. 1b, 1g) added to β-sheet-I. Strand 1` in domain 7 arises from a β-bulge formed by Arg 812 and Glu 813, while a twist at cis-Pro 811 results in the two short strands β1 and β1` participating in different sheets.

Fig. 1.

a Stereo view of the 20 conformers representing the solution structure of OBSL1(805–892) obtained after superposition of the C^α atoms of β-strand residues for minimal RMSD. Terminal, flexibly disordered residues 803–805 and 892 were omitted for clarity, and the termini are labeled as ‘N’ and ‘C’. b Ribbon diagram of the lowest-energy conformer of OBSL1(805–892) containing Ig-like domain 7. The location of secondary structure elements are: β-strand 1` (808–810), β1(815–818), β2(825–831), β3(839–842), β4(854–858), β5(861–866), 3₁₀-helix (871–873), β6(876–880), β7(885–891). β-strands are depicted in cyan, the 3₁₀ α-helix is shown in red and yellow, and other polypeptide segments are in gray. c Sausage representation of backbone and superposition of the conformation of the best defined side chains (Table 1). A spline curve was drawn through the mean positions of C^α atoms of residues 806–891 with the thickness proportional to the mean global displacement of C^α atoms in the 20 conformers superimposed in (a). d The electrostatic surface potential of lowest energy NMR conformer. Positively and negatively charged regions are colored in blue and red, respectively, and the neutral regions are colored in white. e Surface and space filling representations of the lowest-energy conformer of OBSL1(805–892) colored according to scores from PredUs prediction of regions potentially involved in protein–protein interactions; the default PredUs color scheme for residues with scores larger than zero are shown from light red to red with increasing score. f–j Show the same as in a–e but for OBSL1(1277–1357) containing Ig-like domain 12. Terminal, flexibly disordered residues 1277–1279 and 1357 were excluded for clarity. The location of secondary structure elements are: β1(1279–1281), β2(1287–1290), β3(1301–1304), β8(1307–1308), β5(1315–1319), β6(1324–1328), β7(1337–1342), and β8(1348–1354). Due to the presence of both the cis- and trans-conformation for Pro 1299, residues 1282, 1294–1302, 1319–1320, 1323–1325, 1340–1351 and 1354–1355 exhibit two sets of signals in 2D [¹⁵N,¹H] HSQC (Fig. S4). These polypeptide segments are highlighted in grey.in (f)

For domain 12, two distinct sets of signals with approximately equal intensities are observed in 2D [¹⁵N,¹H] HSQC (Fig. S4) for the polypeptide backbone NH moieties of residues 1282, 1294–1302, 1319–1320, 1323–1325, 1340–1351 and 1354–1355 (Fig. 1f). The lack of exchange broadening of these signals reveals that the two conformations are in slow exchange on the chemical shift time scale, and the analysis of sequential NOEs between Gly 1298 and Pro 1299 revealed that this is due to the Pro being present in both cis and trans conformation (note that the structure in the PDB contains Pro 1299 in the cis conformation). As expected for 30 % sequence identity, the two structures are rather similar: the root mean square deviation (RMSD) calculated between the mean coordinates of the backbone heavy atoms N, C^α, and C’ of the β-strands present in both domains is 1.9 Å.

Since Ig-like domains are highly abundant in vertebrates [15], a search of the PDB for structurally similar proteins using the DALI server [33] yielded a very large number of hits spanning a quasi-continuum of Z-scores > 4.0 (i.e., > 953 and > 984 for domains 7 and 12, respectively), which included the structures of OBSL1 Ig-like domains 1, 5, 6 and 8 with Z-scores ranging from 8 to 14 and corresponding backbone RMSD values between 1.4 and 2.5 Å (Table SI). Electrostatic potential calculations show that domains 7 and 12 both exhibit a mixed, albeit rather different surface charge distribution (Fig. 1d, 1i). This observation indicates that the two domains interact with different epitopes on other cytoskeletal proteins. We thus used the PredUs server [34, 35] to predict potential protein-protein interaction sites. Inspection of Figs 1e, 1j shows that indeed quite different regions were identified for the two domains. Consistent with the prediction for domain 7, mutations of Arg 812 causing the 3-M syndrome [4] are located in the predicted protein-protein interaction site. Inspection of OBSL1 Ig-like domain 1 in a complex with titin Ig-like domain M10 [14] reveals that different surface regions are involved in the protein-protein complex formation for this domain. Overall, it might well be that each of the OBSL1 Ig-like domains is involved in distinctly different protein-protein interactions.

Abbreviations

CUL7

Cullin7

CCDC8

Coiled-coil domain containing protein 8

DSS

4,4-dimethyl-4-silapentane-1-sulfonate sodium salt

DTT

Dithiothreitol

FBXW8

Fbox and WD repeat containing protein 8

HCPIN

Human Cancer Pathway Interaction Network

Ig-like

Immunoglobulin-like

MES

2-(N-morpholino)ethanesulfonic acid

MAPK

Mitogen-activated protein kinase

NESG

Northeast Structural Genomics Consortium

NOESY

Nuclear Overhauser enhancement spectroscopy

OBSL1

Obscurin-like protein 1

PDB

Protein Data Bank

RDC

Residual dipolar coupling

RMSD

Root mean square deviation

TLR

Toll-like receptor

Tris

Tris(hydroxymethyl)aminomethane