The fifth epidermal growth factor-like domain of thrombomodulin does not have an epidermal growth factor-like disulfide bonding pattern

Combined proteomic and biochemical analyses redefine the consensus sequence requirement for epidermal growth factor-like domain hydroxylation

Epidermal growth factor-like domains (EGFDs) have important functions in cell-cell signaling. Both secreted and cell surface human EGFDs are subject to extensive modifications, including aspartate and asparagine residue C3-hydroxylations catalyzed by the 2-oxoglutarate oxygenase aspartate/asparagine-β-hydroxylase (AspH). Although genetic studies show AspH is important in human biology, studies on its physiological roles have been limited by incomplete knowledge of its substrates. Here, we redefine the consensus sequence requirements for AspH-catalyzed EGFD hydroxylation based on combined analysis of proteomic mass spectrometric data and mass spectrometry-based assays with isolated AspH and peptide substrates. We provide cellular and biochemical evidence that the preferred site of EGFD hydroxylation is embedded within a disulfide-bridged macrocycle formed of 10 amino acid residues. This definition enabled the identification of previously unassigned hydroxylation sites in three EGFDs of human fibulins as AspH substrates. A non-EGFD containing protein, lymphocyte antigen-6/plasminogen activator urokinase receptor domain containing protein 6B (LYPD6B) was shown to be a substrate for isolated AspH, but we did not observe evidence for LYPD6B hydroxylation in cells. AspH-catalyzed hydroxylation of fibulins is of particular interest given their important roles in extracellular matrix dynamics. In conclusion, these results lead to a revision of the consensus substrate requirements for AspH and expand the range of observed and potential AspH-catalyzed hydroxylation in cells, which will enable future study of the biological roles of AspH.

Aspartate/asparagine-β-hydroxylase crystal structures reveal an unexpected epidermal growth factor-like domain substrate disulfide pattern

AspH is an endoplasmic reticulum (ER) membrane-anchored 2-oxoglutarate oxygenase whose C-terminal oxygenase and tetratricopeptide repeat (TPR) domains present in the ER lumen. AspH catalyses hydroxylation of asparaginyl- and aspartyl-residues in epidermal growth factor-like domains (EGFDs). Here we report crystal structures of human AspH, with and without substrate, that reveal substantial conformational changes of the oxygenase and TPR domains during substrate binding. Fe(II)-binding by AspH is unusual, employing only two Fe(II)-binding ligands (His679/His725). Most EGFD structures adopt an established fold with a conserved Cys1–3, 2–4, 5–6 disulfide bonding pattern; an unexpected Cys3–4 disulfide bonding pattern is observed in AspH-EGFD substrate complexes, the catalytic relevance of which is supported by studies involving stable cyclic peptide substrate analogues and by effects of Ca(II) ions on activity. The results have implications for EGFD disulfide pattern processing in the ER and will enable medicinal chemistry efforts targeting human 2OG oxygenases.

AspH catalyses hydroxylation of asparagine and aspartate residues in epidermal growth factor-like domains (EGFDs). Here, the authors present crystal structures of AspH with and without substrates and show that AspH uses EFGD substrates with a non-canonical disulfide pattern.

Structural determinants in protein folding: a single conserved hydrophobic residue determines folding of EGF domains

The epidermal growth factor (EGF) domain is evolutionarily conserved despite hypervariability in amino acid sequences. They fold into a three-looped conformation with a disulfide pairing of C(1)-C(3), C(2)-C(4) ,and C(5)-C(6). To elucidate the structural determinants that dictate the EGF fold, we selected the fourth and fifth EGF domains of thrombomodulin (TM) as models; the former domain folds into the canonical conformation, while the latter domain folds with alternate disulfide pairing of C(1)-C(2), C(3)-C(4), and C(5)-C(6). Since their third disulfide (C(5)-C(6)) is conserved, we examined the folding tendencies of synthetic peptides corresponding to truncated domain four (t-TMEGF4) and five (t-TMEGF5), encompassing the segment C(1) to C(4). These peptides fold into their respective disulfide isoforms indicating that they contain all the required structural determinants. On the basis of the folding tendencies of these peptides in the absence and presence of 6 M Gn·HCl or 0.5 M NaCl, we determined that hydrophobic interactions are needed for the canonical EGF fold but not for the noncanonical fold. Sequence alignment of extant EGF domains and examination of their three-dimensional structures allowed us to identify a highly conserved hydrophobic residue in intercysteine loop 3 as the key contributor, which nucleates the hydrophobic core and acts as the lynch pin. When this hydrophobic residue (Tyr25) was substituted with a more hydrophilic Thr, the hydrophobic interactions were disrupted, and t-TMEGF4-Y25T folds similar to t-TMEGF5. Taken together, our results for the first time demonstrate that a single conserved hydrophobic residue acts as the key determinant in the folding of EGF domains.

Structural determinants of protein folding

The last several decades have seen an explosion of knowledge in the field of structural biology. With critical advances in spectroscopic techniques in examining structures of biomacromolecules, in maturation of molecular biology techniques, as well as vast improvements in computation prowess, protein structures are now being elucidated at an unprecedented rate. In spite of all the recent advances, the protein folding puzzle remains as one of the fundamental biochemical challenges. A facet to this empiric problem is the structural determinants of protein folding. What are the driving forces that pivot a polypeptide chain to a specific conformation amongst the vast conformation space? In this review, we shall discuss some of the structural determinants to protein folding that have been identified in the recent decades.

Allosteric disulfide bonds in thrombosis and thrombolysis

Allosteric disulfide bonds control protein function by mediating conformational change when they undergo reduction or oxidation. The known allosteric disulfide bonds are characterized by a particular bond geometry, the -RHStaple. A number of thrombosis and thrombolysis proteins contain one or more disulfide bonds of this type. Tissue factor (TF) was the first hemostasis protein shown to be controlled by an allosteric disulfide bond, the Cys186-Cys209 bond in the membrane-proximal fibronectin type III domain. TF exists in three forms on the cell surface: a cryptic form that is inert, a coagulant form that rapidly binds factor VIIa to initiate coagulation, and a signaling form that binds FVIIa and cleaves protease-activated receptor 2, which functions in inflammation, tumor progression and angiogenesis. Reduction and oxidation of the Cys186-Cys209 disulfide bond is central to the transition between the three forms of TF. The redox state of the bond appears to be controlled by protein disulfide isomerase and NO. Plasmin(ogen), vitronectin, glycoprotein 1balpha, integrin beta(3) and thrombomodulin also contain -RHStaple disulfides, and there is circumstantial evidence that the function of these proteins may involve cleavage/formation of these disulfide bonds.

Structural and functional consequences of methionine oxidation in thrombomodulin

Thrombomodulin (TM) is an endothelial cell surface glycoprotein that is responsible for switching the catalytic activity of thrombin away from fibrinogen cleavage (pro-coagulant) and towards protein C cleavage (anticoagulant). Although TM is a large protein, only the fourth and fifth epidermal growth factor-like (EGF-like) domains are required for anticoagulant function. These two domains must work together, and the linker between the two domains contains a single methionine residue, Met 388. Oxidation of Met 388 is deleterious for TM activity. Structural studies, both X-ray and NMR, of wild type and variants at position 388 show that Met 388 provides a key linkage between the two domains. Oxidation of the methionine has consequences for the structure of the fifth domain, which binds to thrombin. Oxidation also appears to disrupt the interdomain contacts resulting in structural and dynamic changes. The functional consequences of oxidation of Met 388 include decreased anticoagulant activity. Oxidative stress from several causes is reflected in lower serum levels of activated protein C and a higher thrombotic tendency, and this is thought to be linked to the oxidation of Met 388 in TM. Thus, TM structure and function are altered in a subtle but functionally critical way upon oxidation of Met 388.

Structural stability of human α-thrombin studied by disulfide reduction and scrambling

Human alpha-thrombin is a very important plasma serine protease, which is involved in physiologically vital processes like hemostasis, thrombosis, and activation of platelets. Knowledge regarding the structural stability of alpha-thrombin is essential for understanding its biological regulation. Here, we investigated the structural and conformational stability of alpha-thrombin using the techniques of disulfide reduction and disulfide scrambling. alpha-Thrombin is composed of a light A-chain (36 residues) and a heavy B-chain (259 residues) linked covalently by an inter-chain disulfide bond (Cys(1)-Cys(122)). The B-chain is stabilized by three intra-chain disulfide bonds (Cys(42)-Cys(58), Cys(168)-Cys(182), and Cys(191)-Cys(220)) (Chymotrypsinogen nomenclature). Upon reduction with dithiothreitol (DTT), alpha-thrombin unfolded in a 'sequential' manner with sequential reduction of Cys(168)-Cys(182) within the B-chain followed by the inter-chain disulfide, generating two distinct partially reduced intermediates, I-1 and I-2, respectively. Conformational stability of alpha-thrombin was investigated by the technique of disulfide scrambling. alpha-Thrombin denatures by scrambling its native disulfide bonds in the presence of denaturant [urea, guanidine hydrochloride (GdmCl) or guanidine thiocyanate (GdmSCN)] and a thiol initiator. During the process, cleavage of the inter-chain disulfide bond and release of the A-chain from B-chain was the foremost event. The three disulfides in the B-chain subsequently scrambled to form three major isomers (designated as X-Ba, X-Bb, and X-Bc). Complete denaturation of alpha-thrombin was observed at low concentrations of denaturants (0.5 M GdmSCN, 1.5 M GdmCl, or 3 M urea) indicating low conformational stability of the protease.

Identification of the disulfide bonds in the recombinant somatomedin B domain of human vitronectin

The NH(2)-terminal somatomedin B (SMB) domain (residues 1-44) of human vitronectin contains eight Cys residues organized into four disulfide bonds and is required for the binding of type 1 plasminogen activator inhibitor (PAI-1). In the present study, we map the four disulfide bonds in recombinant SMB (rSMB) and evaluate their functional importance. Active rSMB was purified from transformed Escherichia coli by immunoaffinity chromatography using a monoclonal antibody that recognizes a conformational epitope in SMB (monoclonal antibody 153). Plasmon surface resonance (BIAcore) and competitive enzyme-linked immunosorbent assays demonstrate that the purified rSMB domain and intact urea-activated vitronectin have similar PAI-1 binding activities. The individual disulfide linkages present in active rSMB were investigated by CNBr cleavage, partial reduction and S-alkylation, mass spectrometry, and protein sequencing. Two pairs of disulfide bonds at the NH(2)-terminal portion of active rSMB were identified as Cys(5)-Cys(9) and Cys(19)-Cys(21). Selective reduction/S-alkylation of these two disulfide linkages caused the complete loss of PAI-1 binding activity. The other two pairs of disulfide bonds in the COOH-terminal portion of rSMB were identified as Cys(25)-Cys(31) and Cys(32)-Cys(39) by protease-generated peptide mapping of partially reduced and S-alkylated rSMB. These results suggest a linear uncrossed pattern for the disulfide bond topology of rSMB that is distinct from the crossed pattern present in most small disulfide bond-rich proteins.

The disulfide structure of denatured epidermal growth factor: preparation of scrambled disulfide isomers

The conformational stability of human epidermal growth factor (EGF) and the structure of denatured EGF were investigated using the technique of disulfide scrambling. Under denaturing conditions and in the presence of a thiol catalyst, the native EGF denatures by shuffling its three native disulfide bonds and converts to a mixture of scrambled isomers. Analysis by HPLC reveals that the denatured EGF is composed of about 10 fractions of scrambled isomers. The heterogeneity varies under different denaturing conditions, with the heat-denatured samples exhibiting the highest degree of heterogeneity. The disulfide structures of eight major scrambled isomers of EGF were determined. The most predominant isomer adopts the bead-form structure with disulfide bonds bridged by three pairs of neighboring cysteines: Cys6-Cys14, Cys20-Cys31, and Cys33-Cys42. The denaturation curve of EGF is determined by the relative yield of the scrambled and native species of EGF. EGF is a highly stable molecule and can be effectively denatured only by guanidine chloride at a concentration of greater than 4-5 M. At 8 M urea, less than 16% of the native EGF was denatured. The unusual conformational stability of EGF was compared with that of eight different disulfide proteins that were similarly characterized by the method of disulfide scrambling.

NMR structure and backbone dynamics of a concatemer of epidermal growth factor homology modules of the human low-density lipoprotein receptor

The ligand-binding region of the low-density lipoprotein (LDL) receptor is formed by seven N-terminal, imperfect, cysteine-rich (LB) modules. This segment is followed by an epidermal growth factor precursor homology domain with two N-terminal, tandem, EGF-like modules that are thought to participate in LDL binding and recycling of the endocytosed receptor to the cell surface. EGF-A and the concatemer, EGF-AB, of these modules were expressed in Escherichia coli. Correct protein folding of EGF-A and the concatemer EGF-AB was achieved in the presence or absence of calcium ions, in contrast to the LB modules, which require them for correct folding. Homonuclear and heteronuclear 1H-15N NMR spectroscopy at 17.6 T was used to determine the three-dimensional structure of the concatemer. Both modules are formed by two pairs of short, anti-parallel beta-strands. In the concatemer, these modules have a fixed relative orientation, stabilized by calcium ion-binding and hydrophobic interactions at the interface. 15N longitudinal and transverse relaxation rates, and [1H]-15N heteronuclear NOEs were used to derive a model-free description of the backbone dynamics of the molecule. The concatemer appears relatively rigid, particularly near the calcium ion-binding site at the module interface, with an average generalized order parameter of 0.85+/-0.11. Some mutations causing familial hypercholesterolemia may now be rationalized. Mutations of D41, D43 and E44 in the EGF-B calcium ion-binding region may affect the stability of the linker and thus the orientation of the tandem modules. The diminutive core also provides little structural stabilization, necessitating the presence of disulfide bonds. The structure and dynamics of EGF-AB contrast with the N-terminal LB modules, which require calcium ions both for folding to form the correct disulfide connectivities and for maintenance of the folded structure, and are connected by highly mobile linking peptides.

Disulfide bond plasticity in epidermal growth factor

Epidermal growth factor (EGF) has a (1-3,2-4,5-6) disulfide-bonding pattern. This pattern is found in nearly all EGF-like domains, despite wide variation in sequences. Biological data from EGF and at least one EGF-like domain show that disulfide bond isomers have significant bioactivity and suggests that the EGF fold can accommodate alternate disulfide-bonding patterns. The disulfide bonds in murine EGF were altered to seven different patterns and structures were calculated incorporating all the restraints from the highest resolution restraint set available (Tejero et al., 1996). Results showed that besides the native (1-3,2-4,5-6), two other disulfide-bonding patterns: (1-2,3-4,5-6) and (1-3,2-5,4-6) satisfied the restraints as well as the native. The results for these two patterns were indistinguishable from the native on the basis of distance and dihedral violations, XPLOR energies, Procheck statistics, and RMSDs of the final set of structures. Two other disulfide bond patterns, (1-2,3-5, 4-6) and (1-4,2-3,5-6) were able to satisfy all the distance restraints but had one or more cysteine dihedral violations. For all seven isomers, the final calculated structures were highly similar to EGF with all-atom RMSD's in the 1. 5-2 A range. These results suggest that the EGF backbone fold has the unique property of accommodating several different disulfide-bonding patterns.

Study on complex formation between recombinant human thrombomodulin fragment and thrombin using surface plasmon resonance

Human thrombomodulin (hTM) is a newly described endothelial cell associated protein that functions as a potent natural anticoagulant by converting thrombin from a procoagulant protease to an anticoagulant. The affinity constant of recombinant human soluble TM (rhs-TM) and a peptide containing active site of hTM fragment (f-hTM) with thrombin were determined using the surface plasmon resonance. The interaction of f-hTM with thrombin could be analyzed by a simple model, whereas the association and the dissociation steps of rhs-TM with thrombin consisted of at least two kinds of interaction phases. The dissociation constant for complex (K(D)) of f-hTM and thrombin was determined to be 205 nM, which was more than twice as high as that of rhs-TM (6.7 and 75 nM). The lower affinity of f-hTM was not due to the slow association rate but to the rapid dissociation rate. It comes clear that f-hTM interacts with thrombin rapidly.

The interaction of thrombomodulin with Ca2+

Thrombomodulin (TM) is a cofactor for protein C activation by thrombin and each residue of a consensus Ca2+ site in the sixth epidermal growth factor domain (EGF6) is essential for this cofactor activity [Nagashima, M., Lundh, E., Leonard, J.C., Morser, J. & Parkinson, J.F. (1993) J. Biol. Chem. 268, 2888-2892]. Three soluble analogs of the extracellular domain of TM, solulin (Glu4-Pro490), TME1-6 (Cys227-Cys462) and TMEi4-6 (Val345-Cys462) were prepared for equilibrium dialysis experiments by exhaustive dialysis against Ca2+-depleted buffer. However, all three analogs still contained one tightly bound Ca2+ (Kd approximately 2 microm), which could only be removed by EDTA. Epitope mapping with Ca2+-dependent monoclonal antibodies to EGF6 provided further localization of this tight Ca2+ site. Equilibrium dialysis of the soluble TM analogs in [45Ca2+] between 10 and 200 microm revealed a second Ca2+ site (Kd = 30 +/- 10 microm) in both solulin and TME1-6, but not in TMEi4-6. Ca2+ binding to this second site was unaffected by bound thrombin and we attribute it to the consensus Ca2+ site in EGF3. A 75-fold decrease in the binding affinity of thrombin to TM was observed with immobilized solulin treated with EDTA to remove the high affinity Ca2+ by measuring kassoc and kdiss rates in a BIAcoretrade mark instrument. Ca2+-dependent conformational transitions detected by CD spectroscopy in the far UV indicate a more ordered structure upon Ca2+ binding. Bound Ca2+ stabilized soluble TM against protease digestion at a trypsin-like protease-sensitive site between Arg456 and His457 in EGF6 compared with protease treatment in EDTA. Finally, TM containing EGF domains 4-6, but lacking the interdomain loop between EGF3 and 4 (TME4-6), has an identical Ca2+ dependence for the activation of protein C as found for TMEi4-6, indicating this interdomain loop is not involved in Ca2+ binding.

Production of large quantities of isotopically labeled protein in Pichia pastoris by fermentation

Heterologous expression in Pichia pastoris has many of the advantages of eukaryotic expression, proper folding and disulfide bond formation, glycosylation, and secretion. Contrary to other eukaryotic systems, protein production from P. pastoris occurs in simple minimal defined media making this system attractive for production of labeled proteins for NMR analysis. P. pastoris is therefore the expression system of choice for NMR of proteins that cannot be refolded from inclusion bodies or that require post-translational modifications for proper folding or function. The yield of expressed proteins from P. pastoris depends critically on growth conditions, and attainment of high cell densities by fermentation has been shown to improve protein yields by 10-100-fold. Unfortunately, the cost of the isotopically enriched fermentation media components, particularly 15NH4OH, is prohibitively high. We report fermentation methods that allow for both 15N-labeling from (15NH4)2SO4 and 13C-labeling from 13C-glucose or 13C-glycerol of proteins produced in Pichia pastoris. Expression of an 83 amino acid fragment of thrombomodulin with two N-linked glycosylation sites shows that fermentation is more cost effective than shake flask growth for isotopic enrichment.

Role of the 6-20 disulfide bridge in the structure and activity of epidermal growth factor

Two synthetic analogues of murine epidermal growth factor, [Abu6, 20] mEGF4-48 (where Abu denotes amino-butyric acid) and [G1, M3, K21, H40] mEGF1-48, have been investigated by NMR spectroscopy. [Abu6, 20] mEGF4-48 was designed to determine the contribution of the 6-20 disulfide bridge to the structure and function of mEGF. The overall structure of this analogue was similar to that of native mEGF, indicating that the loss of the 6-20 disulfide bridge did not affect the global fold of the molecule. Significant structural differences were observed near the N-terminus, however, with the direction of the polypeptide chain between residues four and nine being altered such that these residues were now located on the opposite face of the main beta-sheet from their position in native mEGF. Thermal denaturation experiments also showed that the structure of [Abu6, 20] mEGF4-48 was less stable than that of mEGF. Removal of this disulfide bridge resulted in a significant loss of both mitogenic activity in Balb/c 3T3 cells and receptor binding on A431 cells compared with native mEGF and mEGF4-48, implying that the structural changes in [Abu6, 20] mEGF4-48, although limited to the N-terminus, were sufficient to interfere with receptor binding. The loss of binding affinity probably arose mainly from steric interactions of the dislocated N-terminal region with part of the receptor binding surface of EGF. [G1, M3, K21, H40] mEGF1-48 was also synthesized in order to compare the synthetic polypeptide with the corresponding product of recombinant expression. Its mitogenic activity in Balb/c 3T3 cells was similar to that of native mEGF and analysis of its 1H chemical shifts suggested that its structure was also very similar to native.

Structure of the fifth EGF-like domain of thrombomodulin: An EGF-like domain with a novel disulfide-bonding pattern

The structure of the fifth EGF-like domain (residues Q387 to E426) of thrombomodulin (TMEGF5) has been determined by two-dimensional NMR. TMEGF5 binds to thrombin with a Ki of 1.9 microM and has been shown to have a novel disulfide bonding pattern in a fully active fragment of TM. In EGF, the disulfide bonding pattern is (1-3,2-4, 5-6), while TMEGF5 has an uncrossed (1-2,3-4,5-6) pattern. The structure of this novel domain, determined from 483 NOE-derived distance restraints, appears to have diverged from the common EGF-like structure. Superposition of the 14 lowest-energy structures of TMEGF5 gives an overall r.m.s.d. of 1.09 A for the backbone atoms. The central two-stranded beta-sheet common to all EGF-like domains is not present in TMEGF5. The A loop, residues C390 to C395, is twisted away from interacting with the B loop, residues C399 to C407, as in EGF, and is close to the C loop, residues C409 to C421. This twist causes the N and C termini to be closer together in TMEGF5 than in EGF. Most of the residues that are important for activity lie on one face of the molecule, which is likely to be the thrombin-binding surface of the domain. The structure of the C loop within the domain, which is a beta-hairpin similar to EGF, is similar to the structure of a synthetic version of the loop bound to thrombin as determined by transferred NOE experiments. Despite the similarity in the structures of the loops, the residues immediately following C421 are in different positions in the two structures suggesting that these "tail" residues may change conformation upon thrombin binding.