Role of glycine 221 in catalytic activity of hyaluronan-binding protein 2

HABP2 (hyaluronan-binding protein 2) is a Ca2+-dependent serine protease with putative roles in blood coagulation and fibrinolysis. A G221E substitution, known as the Marburg I polymorphism, reportedly affects HABP2 function and has been associated with increased risk for cardiovascular disease. However, the importance of Gly-221 for HABP2 activity is unclear. Here, we used G221E, G221A, and G221S mutants to assess the role of Gly-221 in HABP2 catalysis. The G221E variant failed to activate the single-chain urokinase-type plasminogen activator, and the G221A and G221S variants displayed moderately reduced single-chain urokinase-type plasminogen activator activation. Activity toward the peptide substrate S-2288 was markedly decreased in all HABP2 variants, with G221E being the most defective and G221A being the least defective. In the absence of Ca2+, S-2288 cleavage by wild-type HABP2 was Na+-dependent, with Km decreasing from 3.0 to 0.6 mm upon titration from 0 to 0.3 m Na+. In the presence of 5 mm Ca2+, Km was further reduced to 0.05 mm, but without an appreciable contribution of Na+. At physiological concentrations of Na+ and Ca2+, the three HABP2 variants, and particularly G221E, displayed a major Km increase for S-2288. Chemical footprinting revealed that Ile-16 is significantly less protected from chemical modification in G221E than in wild-type HABP2, suggesting impaired insertion of the N terminus into the G221E protease domain, with a concomitant impact on catalytic activity. Homology modeling suggested that the Glu-221 side chain could sterically hinder insertion of the N terminus into the HABP2 protease domain, helping to explain the detrimental effects of Glu-221 substitution on HABP2 activity.

HABP2 (hyaluronan-binding protein 2) is a Ca 2ϩ -dependent protease that belongs to the family of serine proteases. The single-chain zymogen is activated by a single cleavage, resulting in a light chain consisting of the catalytic domain with the typical chymotrypsin fold and a heavy chain that comprises three EGF-like domains and a kringle domain (1,2). HABP2 is also known as FSAP (factor seven-activating protease). This, however, proved to be a misnomer, because coagulation factor VII is a remarkably poor substrate for FSAP/HABP2 (3). In contrast, HABP2 does effectively activate single-chain urokinase-type plasminogen activator (scuPA), 2 suggesting a role in fibrinolysis (3)(4)(5). HABP2 has further been reported to cleave tissuefactor pathway inhibitor (6). Interestingly, HABP2/FSAP knockout mice display a mild bleeding phenotype, suggesting a role in hemostasis in relation to tissue-factor pathway inhibitor cleavage (7).
As a serine protease with the chymotrypsin fold, the protease domain of HABP2 contains a dual antiparallel ␤-barrel architecture (8,9). In the ␤-barrel cleft resides the catalytic triad comprising the residues Ser-195, His-57, and Asp-102 (chymotrypsin numbering is used throughout this paper). In extension of the protein core, there are eight unique surface loops. These loops appear more variable between serine proteases because they are involved in substrate recognition and allosteric regulation of the catalytic activity. For instance, the so-called 70-loop of HABP2 may interact with Ca 2ϩ , which is indispensable for allosteric regulation of the catalytic activity in coagulation factor IX and X (2, 10 -12). In contrast, the 220-loop of thrombin, protein C, factor IX, and factor X is involved in Na ϩ binding (13)(14)(15)(16). The 220-loop is particularly important because it, together with the 180-loop, forms the primary S1 specificity pocket involved in recognition of the principle (P1) substrate residue (Arg in scuPA for HABP2). Moreover, the 220-loop is closely situated closely to the N terminus insertion pocket (8,9).
Whether or not the 220-loop has similar implications in HABP2 remains unclear. The HABP2 220-loop is considerably shorter than that of the typical Na ϩ -binding proteases (see Fig.  7A). On the other hand, a polymorphism in the HABP2 gene, which encodes the amino acid substitution G221E in the 220loop (G534E in HABP2 numbering), has been associated with an increased risk for cardiovascular disease or venous thrombosis (17)(18)(19). This mutation has been designated as the Marburg I polymorphism and has been reported to cause reduced activity (17). Although the role of Gly-221 remains unclear, the above-mentioned findings indicate that G221E might not merely be a polymorphism and attributes a specific function to the 220-loop. In the present paper, we address the role of Gly-221 with particular reference to the regulation of HABP2 catalytic activity.

Construction of HABP2 derivatives
To elucidate the role of residue 221 for HABP2 function, we initially constructed three HABP2 variants in a WT-HABP2 background. In these variants, Gly-221 was replaced by either the small apolar alanine, the small polar serine, or the negatively charged glutamic acid. Protein expression studies revealed that HABP2 G221A and HABP2 G221S were, like WT-HABP2, highly sensitive to autocatalytic inactivation (data not shown). Purified HABP2 G221E was, however, fully processed into a twochain form (Fig. 1A). N-terminal sequencing identified Ile-16 as the N terminus of the 25-kDa light chain, demonstrating that two-chain HABP2 G221E had been cleaved at the authentic activation site.
We have previously demonstrated that the inactivation of HABP2 can be controlled by mutating the natural site of activation into a cleavage site for the protease thermolysin (HABP2 R15Q ) (3). We therefore constructed thermolysin cleavage sites in all the HABP2 variants. After introduction of this mutation, the variants were purified as a single-chain molecule. There was, however, a differential effect of thermolysin on these HABP2 derivatives. Whereas HABP2 R15Q was fully activated by thermolysin within 90 min, HABP2 R15Q/G221A and HABP2 R15Q/G221S required a prolonged incubation time with thermolysin to obtain the two-chain molecule. An even longer incubation with thermolysin was required to proteolytically process HABP2 R15Q/G221E . However, this also led to additional cleavages within the light chain of this variant. We were therefore unable to obtain two-chain HABP2 R15Q/G221E (Fig. 1B). These observations show that the nature of the amino acid at position 221 affects the activation of the HABP2 variants by thermolysin. To examine the role of amino acid substitution at 221, we employed in this study HABP2 R15Q/G221S , HABP2 R15Q/G221A , and HABP2 R15Q , which can be specifically activated by thermolysin, as well as the constitutively active HABP2 G221E .

HABP2 activity toward macromolecular substrate scuPA
Each HABP2 variant was incubated with increasing concentrations of scuPA. The amount of uPA generated after 10 min was assessed employing the uPA-specific substrate S-2444 (Glu-Gly-Arg-p-nitroanilide). The data revealed that there is no detectable cleavage of scuPA in the presence of HABP2 G221E under these conditions (Fig. 2). HABP2 R15Q/G221A and HABP2 R15Q/G221S displayed slower scuPA activation than HABP2 R15Q . The data in Fig. 2 suggest that these mutants display a combination of lower K m and k cat , making the net effect on catalytic efficiency activity relatively small (see Table  2). Nevertheless, these data point to the possibility of changes in scuPA binding to HABP2 associated with mutations in position 221.

Effect of Na ؉ on HABP2 R15Q activity in the presence and absence of Ca 2؉
The catalytic activity of HABP2 R15Q toward the peptide substrate S-2288 (Ile-Pro-Arg-p-nitroanilide) was assessed at varying Na ϩ concentrations in the presence or absence of Ca 2ϩ (Fig. 3). In the absence of Ca 2ϩ , the activity of HABP2 R15Q increased with increasing Na ϩ concentrations (Fig. 3A). However, the effect of Na ϩ was small in the presence of Ca 2ϩ (Fig.  3B). Calculation of kinetic parameters showed that Na ϩ causes a 5-fold drop in K m in the absence of Ca 2ϩ , whereas K m proved independent of Na ϩ when Ca 2ϩ was present (Fig. 3C). As summarized in Table 1, Na ϩ and Ca 2ϩ did predominantly affect K m , but not k cat . In terms of catalytic efficiency (k cat /K m ), Na ϩ stimulated HABP2 activity by 5-fold, whereas the effect of Ca 2ϩ was 8-fold higher (Table 1). Thus, in comparison with the  magnitude of the effect of Ca 2ϩ , the effect of Na ϩ seems limited. Control experiments (data not shown) indicated that in the absence of Ca 2ϩ , Li ϩ had no effect on the amidolytic activity of HABP2. The same holds for choline chloride, which was used to keep the ionic strength constant in the Na ϩ -titration studies.

Amidolytic activity of the HABP2 variants
The activity of HABP2 variants was examined by Na ϩ -and Ca 2ϩ -titration experiments. The activity of the mutants was dependent on the amino acid in position 221, in the order Gly Ͼ Ala Ͼ Ser Ͼ Glu (Fig. 4, A and B). Like HABP2 R15Q , increasing Na ϩ concentrations enhanced the activity of the HABP2 variants (Fig. 4, A and C). Similarly, all variants displayed similar response to Ca 2ϩ in the presence of Na ϩ (Fig. 4, B and D), the optimal rate of substrate hydrolysis being reached at ϳ1 mM for all variants.
The kinetic parameters for S-2288 hydrolysis were then assessed in the presence of 5 mM CaCl 2 and 150 mM NaCl ( Fig.  5 and Table 2). These data show that replacement of Gly in position 221 greatly reduces the catalytic efficiency. The most affected variant was HABP2 R15Q/G221E , which proved 400-fold less efficient than HABP2 R15Q . Because substitution of Gly-221 predominantly affects K m , these data are compatible with perturbation of the interaction between active site of HABP2 and the substrate S-2288. Thus, Gly-221 in the 220-loop is beneficial for HABP2 activity toward both the peptide substrate S-2288 and the macromolecular substrate scuPA ( Table 2).

Insertion of Ile-16 in the protease domain of HABP2 G221E
Following proteolytic activation of HABP2 R15Q and HABP2 G221E , insertion of the newly formed N terminus (Ile-16) was probed by incubating with sulfo-NHS-LC-biotin that specifically reacts with free, solvent-exposed amino groups (20,21). The biotinylated proteins were proteolytically processed with trypsin and analyzed by mass spectrometry. After 5 min of incubation with sulfo-NHS-LC-biotin, a biotinylated N-terminal peptide of HABP2 was identified (i.e. biotin-IYGGFK, m/z 512.271) for both activated HABP2 R15Q and activated HABP2 G221E . The area under the curve of the reconstructed ion chromatograms shows that the degree of biotinylation at Ile-16 was at each time point higher for HABP2 G221E than for HABP2 R15Q (Fig. 6).

Discussion
The aim of the present study was to address the role of Gly-534 in the serine protease HABP2, and in particular the functional implication of the G534E substitution. This naturally occurring variant is encoded by the Marburg I polymorphism in the HABP2 gene and has been associated with increased cardiovascular risk. In chymotrypsin numbering, the Marburg I polymorphism affects position 221 and thus is part of the characteristic 220-loop that participates in allosteric regulation of the catalytic activity in a variety of serine proteases (13)(14)(15)(16)22).

on S-2288 hydrolysis
The table below displays the kinetic constants of HABP2 R15Q for hydrolysis of S-2288 in either the presence or the absence of Na ϩ and Ca 2ϩ . The constants were obtained from fitting the data of Fig. 3 using the Michaelis-Menten equation.

The role of Gly-221 in HABP2 catalytic activity
This raised the possibility that this loop, and in particular residue 221 therein, would serve a similar role in HABP2.
In related serine proteases such as activated protein C, factor IXa, and factor Xa, allosteric regulation involves interplay between the S1 site in the specificity pocket, the Ca 2ϩ -binding site in the protease domain, and Na ϩ binding that involves the 220-loop (14 -16). In HABP2, the 220-loop differs from that in other mammalian serine proteases, in that it is two amino acid residues shorter (Fig. 7A). Moreover, HABP2 has Pro in position 225, which is typical for trypsin-like proteases, whereas most Na ϩ -binding serine proteases have Tyr in this position (22). In view of these differences, it is surprising that the amidolytic activity of activated HABP2 is sensitive not only to Ca 2ϩ , but also to Na ϩ (Figs. 3 and 4). The catalytic efficiency (k cat /K m ) was ϳ5-fold increased by Na ϩ alone and 42-fold by Ca 2ϩ alone (Table 1). In the presence of Ca 2ϩ , the contribution of Na ϩ proved virtually absent, presumably because it is small in comparison with the Ca 2ϩ response (Fig. 3B and Table 1). In this regard, HABP2 differs from for instance factor IXa, wherein Na ϩ and Ca 2ϩ contribute to a similar extent (14), or activated protein C, wherein the Na ϩ response predominates (15). The limited response to Na ϩ in the presence of Ca 2ϩ makes it difficult to conclude whether or not HABP2, like activated protein C, factor IXa, and factor Xa (14 -16), displays any thermodynamic linkage between the putative Na ϩ -and the Ca 2ϩ -binding sites. It is evident, however, that Na ϩ induces a 5-fold reduction of K m for the peptide substrate S-2288 (Fig. 3C). Because k cat remains essentially unchanged, the drop in K m suggests that Na ϩ substantially facilitates the interaction of HABP2 with the peptide substrate. This effect seems negligible, however, in the

Table 2 Effects of Gly-221 substitutions on scuPA activation and S-2288 hydrolysis
Given are the calculated kinetic constants of the HABP2 variants for the proteolytic cleavage of scuPA and hydrolysis of S-2288. These constants were obtained by fitting data of Figs. 2, 3, and 5, respectively, to the Michaelis-Menten equation. Because of the low activity of HABP2 G221E towards scuPA, kinetic constants thereof could not be determined (ND).

Hydrolysis of S-2288
Activation of scuPA presence of Ca 2ϩ , which reduces K m by a further 12-fold (Fig.  3B). The Na ϩ sensitivity of HABP2 implies that it is a Na ϩbinding protease, in a manner that is highly reminiscent to related serine proteases that display Na ϩ binding to the 220loop. Because the effect of Na ϩ is no longer apparent at physiological Ca 2ϩ concentrations ( Fig. 3C and Table 1), this is unlikely to be of physiological significance. In this respect HABP2 may be classified as being Na ϩ -binding but not Na ϩdependent in the strict sense.
As for the role of Gly in position 221, it is evident that the substitution into Ala or Ser leads to a moderate reduction of proteolytic activity of HABP2 toward scuPA, whereas mutation into Glu, the Marburg I variant, displays by far the most prominent defect (Fig. 2 and Table 2). Similar conclusions can be drawn with respect to amidolytic activity toward S-2288 (Fig.  5), although in terms of catalytic efficiency, the difference between mutants seems more pronounced ( Table 2). The G221A and G221S substitutions predominantly increased K m , which seems indicative for perturbed substrate interaction, whereas the effect of the G221E variant seems more complex, affecting both K m and k cat . We considered the possibility that the substitutions may severely affect the response to Na ϩ and Ca 2ϩ , but titration experiments did not provide any appreciable evidence in this direction (Fig. 4). Thus, the severely impaired activity of HABP2 G221E is unlikely to be due to a major disruption in the interaction between the S1 specificity pocket and the binding sites for Na ϩ and/or Ca 2ϩ . Another interaction with the S1 pocket in serine proteases occurs upon proteolytic activation via an electrostatic interaction between Asp-194 and the nascent new N terminus Ile-16 (9). The observation that in HABP2 G221E the N terminus is significantly more exposed than in normal HABP2 (Fig. 6) suggests that HABP2 G221E lacks the stabilization of the S1 pocket that normally results from N terminus insertion. As such, the activated form of HABP2 G221E may predominantly be inactive, presumably because of a shift of the equilibrium between closed (E*) and open (E) forms (23) or in the continuum between zymogen-and proteinase-like states (24).
In absence of a crystal structure of HABP2, it remains difficult to fully appreciate the role of Gly-221 and the effect of the mutations described in the present paper. Of all human serine proteases, hepatocyte growth factor activator (HGFA) exhibits the highest primary sequence homology with full-length HABP2 (1). To assess the structural role Gly-221, we constructed a model of the catalytic domain of HABP2 based on that of HGFA by employing comparative homology modeling (25,26). Fig. 7B shows the structure of the active site region of HABP2, with the putative position of the Ca 2ϩ and Na ϩ ions. As for Na ϩ -binding, five of six coordinating ligands could be provided by the carbonyl O-atoms of residues Gly-221, Lys-221a, and Arg-224 from the 220-loop and Leu-185 and Asn-185a from the opposite 180-loop (Fig. 7B). Although in other Na ϩ -binding serine proteases, the coordination for Na ϩ binding is disrupted by Pro in position 225 (22), HABP2 may be an exception because its 220-loop is shorter and thereby less sensitive to destabilization by the side chain at position 225. The model suggests that in HABP2 G221E the putative Na ϩ coordination may be conserved (Fig. 7C), which seems compatible with the observed Na ϩ response of this variant (Fig. 4C). The model further suggests that Glu in position 221 could introduce a steric clash with Tyr-17, thus impeding the insertion of the new N terminus Ile-16 after HABP2 activation (Fig. 7C) and the concomitant conversion of the zymogen-like form into the proteinase-like conformation.
The findings of this study show that Gly-221 is of major importance for the catalytic activity of HABP2. As such, the significance of position 221 is shared with other serine proteases like protein C, factor IX, and thrombin. For thrombin, mutation of the Asp-221 into a glutamic acid, which is highly similar in properties to the former amino acid residue, already leads to a severe defect in fibrinogen clotting (27). Similarly, the G221R mutation in protein C leads to an impaired biological Peptides of the HABP2 derivatives were obtained and separated on a C18 liquid chromatography column that was connected to a mass spectrometer as described under "Experimental procedures." A, shown are the reconstructed ion chromatograms of the eluted peptide representing the modified N terminus of HABP2 R15Q and of HABP2 G221E after 15 min of incubation with sulfo-NHS-LCbiotin. B, shown are the total number of modified N-terminal peptide ions of HABP2 R15Q (squares) and HABP2 G221E (diamonds) normalized to the maximum number of modified ions that were obtained for HABP2 G221E after 60 min of incubation with sulfo-NHS-LC-biotin (see "Experimental procedures"). The data show the averages of two independent experiments. The error bars show the deviation between the two experiments.
activity of this protease (28). In contrast, alanine to aspartic acid exchange in coagulation factor IX does not affect its clotting activity or its binding to macromolecular substrates at all (29). However, mutation of the same alanine into a valine severely impairs activity, which has been attributed to obstruction of N terminus insertion by the branched side chain (29). Our study proposes a similar role of position 221 in the 220-loop on N terminus insertion in HABP2.
In view of the numerous substrates that have been proposed for HABP2 (3)(4)(5)(6)(7)10), it remains difficult to predict the phenotype associated with Gly-221 substitution. It remains plausible that HABP2 is a promiscuous enzyme that activates more substrates than scuPA alone and thus may have more roles than stimulation of fibrinolysis. However, irrespective of the physiological function of HABP2, it is evident that the G221E substi-tution as encoded by the Marburg I polymorphism results in a severely dysfunctional protein.

Chemicals
All chemical were from Sigma-Aldrich unless otherwise stated. Sulfo-NHS-LC-biotin was from QB Perbio (Tattenhall, Cheshire, UK). Acetonitrile and ultra-pure water, which were employed for mass spectrometry, were obtained from Biosolve (Valkenswaard, The Netherlands).

Recombinant HABP2 derivatives and their activation by thermolysin
Recombinant HABP2 R15Q is described by Stavenuiter et al. (3). The thermolysin HABP2 variants carrying amino acid A, the 220-loop (delineated by a box) of HABP2 is shorter than that of other serine proteases as visualized by the primary sequence alignment. Except for Cys-220, the 220-loop residues of HABP2 are poorly conserved with those of factor IX, X, thrombin, and protein C. B, models of the HABP2 protease domain were constructed with Modeler 9v7 employing the crystal structure of HGFA as a template (Protein Data Bank code 1YC0) (26). The presence of the Na ϩ ion was modeled according to its position in the FXa crystal structure (Protein Data Bank code 2BOK) (32). A representative image is shown that also comprises the Ca 2ϩ -binding site, 220-loop, S1 pocket, and N terminus insertion site. The modeled Na ϩ ion is shown as a light blue sphere, and the Ca 2ϩ ion is shown as an orange sphere. A close-up is given of the 180-loop and 220-loop in interaction with a Na ϩ ion. Putative coordination of the Na ϩ ion by amino acid carbonyl groups of Leu-185, Asn185-a, Gly-221, Lys-221a, and Arg-224 is indicated by dashed lines. Tyr-17 at the N terminus of the protease domain is indicated in cyan, and Gly-221 is shown in a green-colored stick representation. C, the model of HABP2 G221E displays the potential steric and electrostatic repulsion between Glu-221 and Tyr-17 at the N terminus insertion site. On the right, a close-up is shown of Glu-221 and Tyr-17 in respectively green-and cyan-colored stick representations. All images were generated using PyMOL (Schrödinger). substitution at position 221 were constructed employing QuikChange mutagenesis (Stratagene, La Jolla, CA) using HABP2 R15Q DNA as a template and the appropriate primers. Recombinant proteins were produced, purified, activated, and quantified as described (3). The HABP2 variants were activated by incubating 0.7 M of the protein with 3 nM thermolysin for 2 h at 30°C in buffer containing 10 mM MES (pH 5.0), 150 mM NaCl, 5% (v/v) glycerol, 0.01% (v/v) Tween 80, 100 M CaCl 2 , and 50 nM ZnCl 2 . Thermolysin was inhibited by the addition of phosphoramidon disodium salt to a final concentration of 10 M. Because of its inherent instability, HABP2 was freshly activated for all individual experiments. SDS/PAGE was performed to check whether activation was complete (Ͼ90 -95%). Molar concentrations of activated HABP2 were based on protein concentrations prior to activation.

Hydrolysis of S-2288 by HABP2 derivatives and the determination of uPA concentration
The initial rates of p-nitroaniline release from S-2288 were measured by following the absorbance at 405 nm using a Rosys-Anthos Lucy 3 photometer (Anthos Labtec Instruments GmbH, Wals, Austria). To calculate the concentration of p-nitroaniline, an extinction coefficient of 9.9 mM Ϫ1 cm Ϫ1 was used. The concentration of uPA, obtained from scuPA cleavage by the HABP2 derivatives, was assessed by adding S-2444 to the reaction mixture (final concentration, 375 M) (3). The initial rates of p-nitroaniline release were assessed by monitoring the absorbance at 405 nm in time as described above. The results were corrected for the background levels of p-nitroaniline in the absence of the HABP2 derivatives. Absorbance changes were converted into increase in p-nitroaniline concentration and divided by the enzyme concentration to yield S-2288 hydrolysis rates in min Ϫ1 . The concentration of uPA was determined using a calibration curve of the HMW uPA international standard 87/594 (National Institute for Biological Standards and Control (NIBSC), South Mimms, UK).

Determination of steady states kinetic constants and inhibition kinetics
Kinetic constants of S-2288 hydrolysis and scuPA activation by the HABP2 derivatives were assessed by assuming Michaelis-Menten-like behavior for the enzymatic reactions (30). GraphPad software was employed to fit the data to the Michaelis-Menten equation employing non-linear regression.

Biotinylation of the free amino group of Ile-16
350 nM of activated HABP2 R15Q or HABP2 G221E was incubated with 25 mM sulfo-NHS-biotin in 100 mM NaCl, 1 mM CaCl 2 , 0.02% (v/v) Tween 80, 50 mM HEPES (pH 7.5). At different time points, the reactions were stopped by the addition of lysine (125 mM). Proteins were subsequently separated on precast NuPage 4 -12% Bis-Tris gels (Invitrogen) under reducing conditions. The protein bands were subsequently visualized employing Coomassie Brilliant Blue staining. Bands of interest were excised and processed for in-gel digestion according to the method of Shevchenko et al. (31). Briefly, the bands were washed with a buffer containing 50 mM ammonium bicarbonate (pH 7.9) followed by a wash with 50% (v/v) acetonitrile. This step was repeated three times. Cysteine bonds were subsequently reduced with 10 mM dithiothreitol for 1 h at 56°C and alkylated with 50 mM iodoacetamide for 45 min at room temperature in the dark. After two subsequent wash/dehydration cycles, the bands were dried for 10 min in a vacuum centrifuge (Thermo Fisher Scientific Inc.) and incubated overnight with 0.06 g/l trypsin at 37°C. Peptides were eluted from the gel with 1% (v/v) formic acid and twice with 50% (v/v) acetonitrile and 5% (v/v) formic acid. The three elution fractions were pooled and concentrated to 20 l in a vacuum centrifuge.

Mass spectrometry and data analysis
The above-described peptides were separated using a reversed-phase C18 column (50 m ϫ 40 cm, 5-m particles) (Nanoseparations, Nieuwkoop, The Netherlands) running at 100 nl/min with a 1-h gradient from 0% to 35% (v/v) acetonitrile with 0.1% (v/v) acetic acid. The peptides were sprayed directly from the column into the LTQ Orbitrap mass spectrometer (Thermo Fisher Scientific Inc.) using a nanoelectrospray source with a spray voltage of 1.9 kV. The LTQ was operated in a data-dependent mode by performing collision induced dissociation in the ion trap (35% normalized collision energy) for the five most intense precursor ions selected from each full scan in the Orbitrap (350 -2000 m/z, resolving power 30.000). An isolation width of 2 Da was used for the selected ions (charge Ն 2) and an activation time of 30 ms. Dynamic exclusion was activated for the MS/MS scan with a repeat count of 1 and exclusion duration of 30 s. Peptides were identified employing a Sequest search against the human entries in the NCBI database utilizing Proteome Discoverer 1.0 software (Thermo Scientific). The same software was used to obtain the peak area of the reconstructed ion chromatograms of the peptide precursor ions for quantification purposes. Direct comparison of the tryptic digests obtained from gel is not possible because of a differential loss of the peptides during the peptide purification procedure. Therefore, the areas of the reconstructed ion chromatograms of control peptides present in each sample were used as a reference to correct for peptide loss.