Redundancy of protein disulfide isomerases in the catalysis of the inactivating disulfide switch in A Disintegrin and Metalloprotease 17

A Disintegrin and Metalloprotease 17 (ADAM17) can cause the fast release of growth factors and inflammatory mediators from the cell surface. Its activity has to be turned on which occurs by various stimuli. The active form can be inactivated by a structural change in its ectodomain, related to the pattern of the formed disulphide bridges. The switch-off is executed by protein disulfide isomerases (PDIs) that catalyze an isomerization of two disulfide bridges and thereby cause a disulfide switch. We demonstrate that the integrity of the CGHC-motif within the active site of PDIs is indispensable. In particular, no major variation is apparent in the activities of the two catalytic domains of PDIA6. The affinities between PDIA1, PDIA3, PDIA6 and the targeted domain of ADAM17 are all in the nanomolar range and display no significant differences. The redundancy between PDIs and their disulfide switch activity in ectodomains of transmembrane proteins found in vitro appears to be a basic characteristic. However, different PDIs might be required in vivo for disulfide switches in different tissues and under different cellular and physiological situations.

The differences in the activity between the tested PDIs might be a general characteristic. An insulin turbidity assay was performed to establish whether the differences in activity are specific for the disulfide switch within the MPD or whether the purified PDIA1 comprises, in general, a higher isomerase activity than PDIA3 and PDIA6. In this assay, PDIs catalyze the reduction of disulfide bridges in insulin leading to its precipitation. This was detectable by an increase in turbidity recorded at λ = 650 nm, as presented for four different concentrations of PDIA1 in Fig. 2A. Higher amounts of PDIA3 and PDIA6 than of PDIA1 had to be applied in order to obtain a maximum of precipitation (Fig. 2B). The displayed concentrations were added to an insulin solution at a concentration of 1 mg/ml. The units of activity were calculated relative to 1 mg PDI, with 1 unit representing a change of 0.01 per minute in turbidity. Again, in this assay, PDIA1 was more active than PDIA3 and PDIA6, by 1.5-fold and 2.9-fold, respectively (Fig. 2C). These differences in activities are in agreement with previous reports 39, 40 and comparable with those with the opMPD as substrate. In order to catalyze the disulfide switch the PDIs have to be in their reduced state. Noteworthy, the differences in their activities were not due to incomplete reduction, which was exemplary verified by Ellman's test ( Table 1).
The affinities of PDIs for their substrates are in a comparable nanomolar range. High concentrations of PDIs are necessary to obtain measurable isomerization of the opMPD to its closed isomer. This implicates that PDIs interact with their substrates only with moderate affinities. To proof this implication the interactions between PDIs towards the MPD were measured by microscale thermophoresis (MST) experiments. The binding of PDIA1, PDIA3, and PDIA6 (see respective subsection below for details) to opMPD and clMPD was analyzed ( Table 2 and Fig. 3). The dissociation constants K D were, in all cases, in the same order of magnitude ranging from 50 to 200 nM and thus lower as previously reported K D values of 1-10 µM in isomerization processes during protein folding 41 or, in the case of PDIA6 and β3-integrin 21 µM, as an example for an extracellular disulfide switch 36 . No significant difference was observed for the affinity of PDIs towards the opMPD and the clMPD, indicating that the structural change attributable to the disulfide switch did not modify or block the interaction side for PDIs within the opMPD. These moderate affinities might be required for transient interactions with a fast release upon catalysis. Noteworthy, they are in the same order of magnitude as the affinity between MPD and GRP78, which protects the opMPD from the isomerization catalyzed by the PDIs 42 . Extracellular PDIA1, PDIA3, and PDIA6 share overlapping and complementary functions as shown for the activation of integrins during coagulation 34,36-38 . All three PDIs tested showed comparable affinities for the MPD of ADAM17, independently of its isoform. Hence, the local situation in the cells and/or tissues might determine which PDI is secreted in order to (B) All three PDIs are able to catalyze the reduction of disulfide bonds in insulin, but higher amounts PDIA3 and PDIA6 than PDIA1 must be applied to obtain maximal precipitation. The indicated concentrations of PDIs were made up in a solution of 1 mg/ml insulin, and the enzyme activity was calculated as described in the methods section. (C) Display of the quantification and comparison of the capacities of PDIs to precipitate insulin.  Table 1. Ellman's tests confirmed the successful reduction of analysed PDIs. Upon DTT treatment and buffer exchanges, the percentage of reduced cysteine residues was tested by Ellman's test. The concentrations of detected free disulfide groups were taken into the relation to the calculated concentration, which was taken from the protein concentration (measured by the absorption at 280 nm) multiplied by the numbers of cysteine residues in the individual PDIs. These concentrations were set to 100% of reduced cysteine residues. PDIA6 was tested in two experimental serials, ones together with PDIA1 and PDIA3, and ones with the indicated TRAP mutants. All tests were performed at least three times independently.  Table 2. Dissociation constants (K D ) of the PDI-MPD interaction with 68% confidence interval (CI) obtained by fitting the data from at least 3 independent MST analyses each with sixteen dilutions (1:2 dilution series) of the respective PDI at 25 °C. K D 's were calculated from the concentration-dependent ligand induced fluorescence intensity changes according to Eq. 1. Additional parameters resulting from the fit: Response amplitude of the fluorescence intensity shift between bound and unbound state, standard error (SE) of regression, reduced χ² and signal to noise. *Replicates increased to 5 due to variations at low concentration. catalyze disulfide switches within extracellular targets. The complex task of identifying individual PDIs, which are secreted under particular situations, lies outside the scope of this report and will be addressed in future work.
The order of the PDI domains. We further focused on the active sites of the PDIs, which contain a typical CGHC-motive. The N-terminal cysteine residue is thought to be responsible for the reduction and isomerization of disulfide bonds through intermediate intermolecular disulfide bond formation. The second cysteine residue is thought to resolve mixed disulfides of PDIs and substrates 43 . PDIA1 and PDIA3 have four domains arranged in the order: the first catalytic domain a, two catalytic inactive b and b' domains, followed by the second catalytically active a' domain ( Fig. 4A). In contrast, PDIA6 consists of only three domains, arranged in the order a, a′, and b domain, and lacks the b′ domain 28 (see Fig. 4B-D). PDIA1 (56.71 ± 0.57 kDa, theoretical M w monomer: 57.6 kDa) and PDIA3 (57.7 ± 4.6 kDa, theoretical M w monomer: 56.6 kDa) appear to be monomeric in solution as determined by size-exclusion chromatography (SEC)-Multi angle laser light scattering (MALS) analysis (Fig. 4E). The in vivo dimerization of PDIA1 supposedly functions as a regulatory mechanism in the ER and occurs through a bb' dimer 44 . In vitro, dimerization is dependent on the presence of divalent cations 45 and leads to a decreased activity 46 . In contrast, PDIA6 exist as a dimer in solution (103.7 ± 8.3 kDa, theoretical M w monomer: 48.8 kDa) and is most likely active as such. This suggests, that the PDIA6 lacking the b' and being present as a dimer, is likely differently regulated than PDIA1 and presumably PDIA3.
The catalytic domains of PDIA6 contribute equally to the disulfide switch and require full integrity. So-called TRAP mutants were generated to analyze the properties of the catalytic sites within PDIA6. In these mutants, the cysteine residues are exchanged by serine residues (Fig. 4B). Consequently, a disulfide bridge covalently traps mixed disulfide intermediates of PDI and substrate, which are normally unlinked by the second resolving cysteine residue. Isomerization processes in which the intermediates are unlinked by the substrates should be unaffected. The latter situation appears to be likely for the disulfide switch in ADAM17. In this case, the classic TRAP mutants should comprise a comparable isomerization activity to that of the wild-type enzyme. The replacement of the second cysteine residue in the CGHC-motive in the a or a' domains results in the classic C58S or C193S TRAP mutants. Additional to the single TRAP mutants, the double TRAP mutant (2 TRAP) with an exchange in C58S and C193S was generated (Fig. 4B). Both single mutants showed comparable reduced activities in catalyzing the disulfide switch of approximately 60% of the wild-type PDIA6 (Fig. 4F). In addition, the 2 TRAP mutant, with the C58S and C193S exchange, was completely inactive. Moreover, the a and a' domain catalyzes the isomerization with the same efficiency, and so no major difference might occur in their interaction with the opMPD. A residual activity of the N-terminal attacking cysteine residue can be excluded because the activity of the single TRAP mutants was on the same level as the activity of a PDIA6 variant in which both cysteine residues of the a domain were exchanged (C55S C58S, Fig. 4C). Because of the inactivity of the classic TRAP mutants, an inverse TRAP mutant was generated to ensure that the C-terminal resolving cysteine residue did not catalyze the disulfide switch (Fig. 4D). This mutant contains an exchange of C55S and C190S and is completely inactive, like the 2 TRAP mutant. Remarkably, all N-and C-terminal exchanges of the cysteine residues in the CGHC-motive result in inactive enzymes with regard to the disulfide switch within the opMPD. Thus, a definite identification of the attacking residue is not possible; a change in the redox potential of the active site caused by the mutations appears as the most probable cause for the inactivity of the TRAP mutants. Again, to exclude the possibility that the reduced activity of the TRAP mutants is due to an incomplete reduction, Ellman's tests were performed exemplary for the wild type PDIA6 and all mutants, whereby no significant differences in the percentage of reduction was detected (Table 1). Noteworthy, the cysteine to serine exchanges did not alter the ability of the isomerase to bind the MPD, since all variants and the wild type isomerase comprise comparable K D -values (Fig. 5). Hence, the cysteine residues of the active sites have no notable impact on the affinity between PDIA6 and the MPD.
In conclusion, all PDIs tested here have basic affinities for potential substrate proteins in a similar nanomolar range. However, in order to produce a general statement, more interactions and affinities have to be tested. Moderate binding affinities to certain proteins allow rapid catalysis and the release of these proteins, which then follow their fate. Regarding ADAM17, the moderate affinity might permit guiding interactions such as its transport to the cell surface by iRhoms [47][48][49][50][51] . The integrity of catalytic domains is essential for activity, possibly to ensure the correct redox potential. The finding of no notable difference in the activities of the catalytic domains, at least for PDIA6, further supports the hypothesis of a moderate general interaction. Thus, we consider it reasonable that there is no unique principal isomerase of ADAM17. Instead, the type of PDI released to the cell surface probably decides which isomerase catalyzes the disulfide switch, and the underling regulatory mechanism of this transport acts to fine-tune the activity of ADAM17.  (C58S, C193S, 2 TRAP: C58SC193S, C55SC58S, inverse TRAP C55SC190S) were generated by overlapping extension PCR cloning. Protein expression was performed in E. coli BL21 (DE3) by inducing the culture at about OD 600 = 0.8 with 1 mM isopropyl-β-D-1-thiogalactopyranoside at 37 °C for three hours. Bacteria were pelleted and resuspended in phosphate buffer saline (PBS). A protease inhibitor cocktail (cOmplete; Roche Diagnostics) and Benzonase (Santa Cruz Biotechnologies) were added before cell lysis and freezing at −20 °C. Bacteria were lysed by sonification, and PDIs were purified via Ni-affinity and size-exclusion chromatography (SEC) by using HisTrap and Superdex 75 or 200 16/60 columns (GE Healthcare). Recombinant open membrane-proximal domain (opMPD) (9.9 kDa) was obtained in a comparable procedure but was further purified via reverse-phase (RP) HPLC as described earlier 11 . The ectodomain of ADAM17 was obtained from Sigma-Aldrich. Human interleukin-6 (IL-6) was expressed and purified as described previously 52 .

Reduction of PDIs.
PDIs were incubated with 50 mM dithiothreitol (DTT) for 120 minutes at room temperature in order to reduce the proteins prior to isomerization experiments. Subsequently, DTT was removed by using NAP TM −5 and NAP TM -10 gel filtration columns (GE Healthcare) according the manufacturers' instructions by using PBS. Gel filtration was performed twice, and the removal of DTT was exemplarily tested by Ellman's test. The protein concentration was determined by absorbance at 280 nm.
Ellmann's test. To verify that the disulfide bridges were reduced upon DTT treatment and buffer exchange Ellman's tests were performed. 200 µl Ellman's reagent (50.0 mM sodium acetate, 2.0 mM 5,5′-Dithio-bis (2-nitrobenzoic acid)) were mixed with 500 µl ddH 2 O, 100 µl 1 M Tris pH 8.0 and 200 µl of either standard or sample probes. Standard curve includes a blank sample (PBS) and samples containing β-mercaptoethanol from 1 to 16 µM in PBS. After preparation of samples, they were incubated for 5 minutes at room temperature before measuring the absorption at 412 nm. To calculate the concentration of free thiol groups an extinction of 13 600 M −1 cm −1 was taken. The theoretical concentrations of free disulfide groups were calculated from the molar concentrations calculated from the absorptions of the protein solutions at 280 nm, which was multiplied by the theoretical number of cysteine residues (PDIA1: 6; PDIA3 7; PDIA6: 6; C58S and C193S: 5, 2 TRAP, C55SC58S and inverse TRAP: 4). The calculated number of free disulfide groups was set to 100% reduced disulfide groups and from the actual measured concentration the percentage of reduced disulfide groups were calculated. All PDIs and variants of PDIs were tested at least three times.
Isomerization Experiments. Ten micrograms of opMPD were incubated together with freshly reduced PDIs in PBS with molar ratios as indicated in the results section. If not otherwise indicated, the isomerization was stopped after 170 minutes at 37 °C by addition of 0.1% trifluoroacetic acid (TFA) and analyzed by RP HPLC (Bio-200-C18 5 µ, MultoHigh, CS-Chromatography Service GmbH). Time kinetics were measured for PDIA6 at a ratio of 1:1, 1:5, and 1:10 PDIA6 to opMPD. For comparison of the activities of PDIs, PDIA1 was used in ratios 1:10, 1:20, and 1:30, PDIA3 in ratios 1:5, 1:10, and 1:20, and PDIA6 in ratios 1:5 and 1:10. Peak areas of the opMPD and the closed MPD (clMPD) were determined, and the sum was set to 100% total MPD. The percentage of the closed isoform was taken as a measure of PDI activity. The means and the standard deviations of at least three independent experiments were calculated. Since PDIA6 tends to degrade, the integrity of PDIs were routinely tested by a SDS-PAGE after incubation at 37 °C for the longest time of incubation.
Insulin turbidity assay. The activity of PDIs to catalyze the reduction of disulfide bridges within insulin was analyzed in a solution of 1 mg/ml insulin in 100 mM Tris-HCl pH 7.5, 5 mM MgCl 2 , and 0.5 mM ATP, which was incubated with the indicated amounts of PDI in the presence of 1 mM DTT at 25 °C. Turbidity was detected at 650 nm against reference samples without PDIs. One unit represents the change of 0.01 OD 650 per minute per 1 mg protein.
SEC-MALS analysis. The SEC-multi angle light scattering (MALS) measurements were performed using an online MALS detector (miniDAWN Treos, Wyatt Technology Corp.) coupled to an Agilent Technologies 1100 series HPLC system equipped with a refractive index detector (differential refractometer, RID, G1362A, Agilent Technologies) and a multiple wavelength detector (DAD, G1315B, Agilent Technologies). Protein samples (100-500 µg/analysis) were separated on a Superdex 200 10/300 GL SEC column (GE Healthcare Life Sciences) at a flow rate of 0.5 ml/minute in PBS pH 7.4. The MALS detector was normalized with bovine serum albumin (BSA) (Sigma-Aldrich) prior PDI analysis. The molecular weight was calculated based on the MALS and refractive index data by using 0.185 as refractive index increment (dn/dc) and a second virial coefficient of zero by using the Zimm plot. Data acquisition and analysis was performed with the software ASTRA version 5. shift. Next, the supernatant was incubated for 5 minutes at 95 °C including 2% (w/v) sodium dodecyl sulfate and 20 mM DTT. The fluorescence intensity of these samples was measured with the Monolith NT.115 instrument as described above, resulting in the same fluorescence intensity (≤±10%) in all tested samples. Unspecific binding to the RED-NHS dye was tested in triplicates by measuring PDIA1, PDIA3 and PDIA6 (each at 3 dilutions of 3750, 37.5 and 0.375 nM) against 20 nM of the dye alone with the same instrument settings as those described above in PBS-P. The reactive NHS-ester of the dye was inactivated by incubation with a fourfold molar excess of ethanolamine for 30 minutes at room temperature. The MST analysis and the calculation of the K D values based on the change of the initial fluorescence intensity were performed by using MO. affinity analysis software v2.2.4 (NanoTemper Technologies) to fit the following equation: with C PDI = concentration of PDI, C MPD = concentration of MPD, R U = response value of unbound state, R B = response value of bound state, and K D = dissociation constant.