Thimet Oligopeptidase: Fluorescent Labeling to Investigate Structural Changes

Saint Mary’s College of California Summer Research Program 2010 1 Thimet Oligopeptidase: Fluorescent Labeling to Investigate Structural Changes

Gabrielle B. Diaz, and Jeffrey A. Sigman.

Chemistry Department, Saint Mary’s College of California, 1928 St. Mary’s Road, Moraga, CA 94556, U.S.A.

The intent of the present study was to optimize and fractions were collected; currently they are awaiting procedures to monitor structural changes in the enzyme mass spectrometry analysis at a facility in UCSF. Once the thimet oligopeptidase (TOP). Wild-type rat TOP was cova- location of attachment is identified by mass spectrometry, lently labeled with the cysteine-specific, fluorescent probe further assays will be done to analyze the sensitivity of the N,N’-dimethyl-N-(iodoacetyl)-N’-(7-nitrobenz-2-oxa-1,3-di- fluorophore on site. Depending on the results, the enzyme azol-4-yl)ethylenediamine (IANBD). Adding this fluorescent will be mutated to rearrange the site of IANBD attachment tag is analogous to adding an antennae for detection. This to more plausible sensitive sites. will allow structural changes to be monitored and analyzed Activity assays were done and confirmed that the under fluorescence spectroscopy due to the tag’s sensitiv- modification did not affect enzymatic performance. Ligand ity to the polarity of its environment. binding assays were also done with the inhibitor angio- The procedure done to modify the enzyme with tensin (1-7) to detect initial changes in fluorescence when IANBD involved covering surface cysteines with iodo- the enzyme is open or closed. acetate, leaving cysteines within the substrate binding Conducting this research will provide more knowl- cleft available for IANBD attachment. A tryptic digest and edge on the structure and function of TOP in an outcome HPLC was done to isolate the peptide segment contain- to provide drugs and inhibitors that will bind strongly to ing the fluorophore. The peptide fragments were eluted the enzyme.

256 J. A. Sigman and others In t r o d u c t i o n Recombinant TOP preparation Thimet oligopeptidase (TOP) (EC 3.4.24.15) is a TOP (accession number P24155) was expressed and purified as metallo endopeptidase that is ubiquitous throughout described previously [15]. The enzyme concentration was deter- 1 the body with concentrated presence in the brain and mined using the molar absorption coefficient ε280 73.11 mM− 1 = · cm− , calculated based on the amino acid content of the protein gonads. It is homologous to several metallopeptidases using the automated ProtParam Tool on the SWISS ExPASyserver across species of which all share a common HEXXH [5,16]. metal-binding motif1-2 i.e., a recurring structure. TOP functions to cleave an array of peptides which in Kinetic assays effect involve immune responses and regulation Kinetic assays were performed using a Cary Eclipse spectro- of physiological activities such as lowering blood fluorimeter or PerkinElmer luminescence spectrometer LS 50 B. 1-2 The cleavage of the fluorogenic MCA [17] or mca-Bk substrate pressure and reproduction . Another function that was monitored by the increase in emission at 400 nm over time has intrigued further research is the enzyme’s ability using an λexcitation of 325 nm. Substrate concentration was calcu- 1 lated based on the molar absorption coefficient ε365 (17.3 mM− to degrade Aβ amyloid plaques present in Alzheimer’s 1 · cm− ) of the DNP. Product formation was determined to be linear 3 disease . with time under all conditions monitored, and less than 10% X-ray diffraction studies have provided of the substrate was consumed during the assay. Assays were performed in duplicate at 23 ◦C in 25 mM Tris/HCl at pH 7.8, invaluable information with regards to its structure adjusted to a conductivity of 12 mS/cm2 with KCl and containing and confirmed structural homology to its closest 10% glycerol. The temperature was chosen so as to minimize relative, neurolysin1-2. Figure 1 shows the overall breakdown of both enzyme and substrate and to allow for sensitivity at low substrate concentrations; the pH represents the opti-

structure; TOP is primarily α-helical with two domains mal value for kcat/Km for MCA [5]. TCEP (1 mM) was also added separated by a deep cleft where the substrate binds. to the buffer to prevent protein dimerization [29]. For assays in urea, the above buffer and an identical buffer containing approx. Domain II contains the Zn(II) active-site and residues 1 FigureFigure 1 1: Overall Tertiary tertiary structure structure of TOP showing of TOP, the indicating location ofthe the domains. active site 10 M urea were mixed in the appropriate ratio. The final urea zinc (white sphere) and the tryptophan residues (space filled) present in concentration was determined based on the refractive index of domains I and II the solution [18]. The change in fluorescence intensity over time was converted into rate of product formation using a standard curve calculated for the peptide products. Individual stan- TOP contains seven tryptophan residues distributed unevenly be- dard curves were prepared at each urea concentration. Although tween the two domains. Trp335 and Trp614 reside in domain I, and the intercept of the standard curve changed, the slope was found to tryptophan at positions 26, 124, 389, 511 and 513 are found in be independent of urea concentration. The kinetic parameters V max domain II. We here report on use of the denaturing agent urea and Km were determined using a hyperbolic fit rate V max[S]/ { = as a tool to study partial unfolding of TOP, as monitored by fluor- (Km [S]) to the plot of substrate concentration (µM) versus escence emission, collisional quenching and anisotropy. Unfold- rate+ of product} formation (µmol/s per µmol of enzyme) under ing occurs as a two-step process, with loss of the catalytic zinc conditions in which [S] is above and below Km. occurring only with the second unfolding event. Complete loss of activity towards a 5-residue quenched fluorescent substrate MCA (7-methoxycoumarin-4-acetyl-Pro-Leu-Gly-Pro-Lys-di- HPLC analysis nitrophenol) is apparent at even low urea concentrations, but activ- Products of the enzymatic reactions with mca-Bk were analysed ity towards a 9-residue Bk-derived substrate is enhanced at low by HPLC on a Hewlett Packard 1090 apparatus. The reaction urea levels. These consistent results show that TOP undergoes mixture, 200 µl total volume in Tris buffer or Tris buffer with conformational changes that differentially affect substrate recog- 2 M urea, contained mca-Bk (0.4 mg/ml) and 0.05 µM enzyme. nition, and suggest a mechanism by which TOP can accommodate A sample was taken at 0 min (before initiation of the reaction) and substrate variation. after reacting for 30 min at 23 ◦C. The reaction was terminated with an equal volume of 1% trifluoroacetic acid in methanol. A 20-µl aliquot was subjected to reverse-phase HPLC using EXPERIMENTAL aC183µ column (150 mm 4.6 mm; Alltech). Solvent A was acetonitrile and solvent B 0.1×% trifluoroacetic acid in water. A Materials linear gradient of 10–82% solvent A was applied, and the pro- Glutathione–Sepharose, Sephacryl S-200 and PD-10 columns ducts were detected by absorbance monitored at 330 nm. were obtained from Amersham Pharmacia Biotech (Piscataway, NJ, U.S.A.). Tris/glycine polyacrylamide gels (12%) were from Invitrogen (Carlsbad, CA, U.S.A.), and stained with Gelcode DSC (differential scanning calorimetry) Blue from Pierce (Rockford, IL, U.S.A.). TCEP [tris-(2-carboxy- All experiments were performed using a VP-DSC microcalori- ethyl)phosphine] was also purchased from Pierce. The quenched meter (MicroCal Inc., Northampton, MA, U.S.A.) at an upscan fluorescent substrates MCA and modified bradykinin, mca- rate of 60 ◦C per h over the range 20◦ to 110 ◦C. Degassed buffer, Bk 7-methoxycoumarin-4-acetyl-[Ala7, Lys(DNP)9]-bradykinin, identical with that diluting the sample, was used as the reference. where{ DNP is 2,4-dinitrophenyl , were obtained from Bachem The injection volume was 0.51 ml. Raw heat data were trans- (King of Prussia, PA, U.S.A.). All} other reagents were purchased formed and plotted as heat capacity as a function of temperature from Sigma Chemical Co (St. Louis, MO, U.S.A.). using the Origin for DSC software supplied by the manufacturer.

c 2005 Biochemical Society � Saint Mary’s College of California Summer Research Program 2010 G. B. Diaz 2

necessary for peptide cleavage. Domain I has a are located on the surface5. However, two residues, putative role for limiting substrate accessibility to Cys175 and Cys425, are located within the substrate- the active site1. Between the two domains, a hinge binding cleft and are the target of attachment5. movement has been speculated due to the flexible Tagging the enzyme will demonstrate if the sites are loop regions connecting the domains. This movement sensitive to structural changes when monitored by poses curiosity with regards to substrate recognition, ligand binding assays. If the expected data and results specificity and selectivity which can be connected to are contrary, these two cysteines or other amino acids the function and properties of homologous enzymes.1 can be mutated into other residues to change the site In the present research, a method has been of attachment to a more sensitive location. modified and produced to attach a fluorescent tag to the enzyme and to identify the location of Ex p e r i m e n t a l Pr o c e d u r e s attachment. The purpose of tagging the enzyme is to investigate and monitor structural changes during Materials activity, particularly the hinge movement collaborated Wild-type rat thimet oligopeptidase was by the two domains. provided by M.J. Glucksman (FUHS/Chicago Medical This investigation will lead to better drug School, Midwest Proteome Center and Department design and understanding of enzymatic activity. of Biochemistry and Molecular Biology, Chicago, The fluorescent tag used was N,N’-dimethyl-N- IL 60064, USA). The reductant tris(2-carboxyethyl) (iodoacetyl)-N’-(7-nitrobenz-2-oxa-1,3-diazol-4-yl) phosphine (TCEP) was obtained from Pierce. PD-10 ethylene-diamine (IANBD); it is cysteine-specific columns were obtained from Amersham Pharmacia and environmentally sensitive. The fluorescence of Biotech (Piscataway, NJ, U.S.A.). PepCleanTM C-18 IANBD increases as polarity of its solvent environment Spin Columns and In-Solution Tryptic Digestion and decreases4. Figure 2 conveys this information; the Guanidination Kit were purchased from Thermo fluorescence of IANBD changes in different solvents. Fisher Scientific Inc. (Waltham, MA. U.S.A.). The Proteins that have been labeled can then be 7-methoxycoumarin-4-acetyl-Pro-Leu-Gly-Pro- monitored by fluorescence spectroscopy to analyze Lys-dinitrophenol (MCA) fluorescent substrate conformational changes and protein-substrate was obtained from Bachem (King of Prussia, PA). interactions. Iodoacetate and N,N’-dimethyl-N-(iodoacetyl)-N’- TOP has 14 cysteine residues, most of which (7-nitrobenz-2-oxa-1,3-diazol-4-yl)ethylenediamine 28270 Fluorescent Labeling of the ␤2 Adrenergic Receptor (IANBD) and all other reagents were purchased from Sigma Chemical Co (St. Louis, MO, U.S.A.).

Enzyme Modification The following protocol was referenced from Sigman et al., (2003)5. A concentrated stock of wild- type rat TOP was obtained. A PD10 column was equilibrated with prepared buffer containing Tris-HCl and KCl with added reductant, TCEP (1mM), all at a pH of 7.8. Fractions of the enzyme were collected from the PD10 and checked via UV-vis. This step separated the DDT from the enzyme stock and transferred it into buffer with TCEP. The enzyme was Figure 24: Fluorescence of IANBD in various solvents. This data from then recombined from the fractions and concentrated literature was referenced to the solvents IANBD was used in the assays in a centricon for 20mins at 4000rpm, then 1000rpm for the set of experiments. The solvents used were acetonitrile and the for one minute. A 50mM iodoacetate (IA) solution was prepared buffer in the Experimental Section. Both conditions were prepared in 25mM phosphate and 5mM EDTA, all at comparable to this data. a final pH of 8.03. The enzyme was retrieved from the

FIG. 2. Competition binding profiles of unlabeled and IANBD- 3 labeled, purified ␤2 receptor. a and b, competition binding of [ H]dihydroalprenolol (1.2 nM) with isoproterenol (a) and alprenolol (b) to 3 unlabeled, purified ␤2 receptor. c and d, competition binding of [ H]dihydroalprenolol (1.2 nM) with isoproterenol (a) and alprenolol (b) to Downloaded from

IANBD-labeled, purified ␤2 receptor. Data are expressed as percent of maximum bound [3H]dihydroalprenolol (mean Ϯ S.E., n ϭ 3).

FIG. 1. Fluorescence properties of IANBD and IANBD-labeled www.jbc.org ␤2 receptor. a, emission spectra of cysteine-reacted IANBD (0.3 ␮M) in solvents of different polarity. Excitation was set at 481 nm. b, emission spectrum of IANBD-labeled ␤2 receptor (0.15 ␮M receptor, 1.2 mol IANBD per mol receptor). Control is emission spectrum of 0.15 ␮M ␤2 receptor “labeled” with IANBD prebound to free cysteine instead of free IANBD to asses possible nonspecific attachment of the probe to the by guest, on July 20, 2010 receptor during labeling. Insert, 10% SDS-polyacrylamide gel electro- phoresis of IANBD-labeled ␤2 receptor. Lane 1, 150 pmol of IANBD- labeled ␤2 receptor; lanes 2 and 3, 150 pmol of ␤2 receptor preincubated before exposure to IANBD with iodoacetamide (lane 2) and N-ethylma- leimide (lane 3). Inset: left panel, Coomassie Blue staining of gel; right panel, gel photographed under UV light. The weak band with an apparent molecular mass of 32.5 kDa is a degradation product of the receptor.

ceptor with IC50 values of 1.3 Ϯ 0.5 ␮M (n ϭ 3, mean Ϯ S.E.) and 2.8 Ϯ 0.6 nM (n ϭ 3, mean Ϯ S.E.), respectively, as compared to 0.8 Ϯ 0.2 ␮M (n ϭ 3, mean Ϯ S.E.) and 2.9 Ϯ 0.9 nM (n ϭ 3, mean Ϯ S.E.) for the unlabeled receptor. These data are in agreement with previous binding data on purified ␤2 receptor (23). Labeling of the ␤2 receptor with IANBD did not affect the total number of binding sites as assessed from binding assays using a saturating concentration of [3H]dihydroalprenolol (10 nM) (data not shown). Stereospecificity, Dose Dependence, and Reversibility of Iso- proterenol-mediated Decrease in Fluorescence from IANBD-labeled ␤2 Receptor—Binding of the full agonist, isoproterenol, to IANBD-labeled ␤2 receptor caused a decrease in fluorescence FIG. 3. Stereospecificity of isoproterenol induced decrease in intensity without detectable change in the wavelength at which fluorescence from IANBD-labeled ␤2 receptor. a, emission spectra maximal emission occurred (Fig. 3). The decrease was ste- of IANBD-labeled ␤2 receptor obtained immediately (t ϭ 0) and after 15 reospecific as illustrated in Fig. 3 by comparing the effect of 30 min (t ϭ 15) following addition of 30 ␮M of the less active (ϩ)-isomer of ␮M of the (Ϫ)-isomer of isoproterenol with 30 ␮M of the less the agonist, isoproterenol ((ϩ)ISO). b, Emission spectra of IANBD- labeled ␤ receptor obtained immediately (t ϭ 0) and after 15 min (t ϭ active (ϩ)-isomer. In addition, no response to isoproterenol was 2 15) following addition of 30 ␮M of the active (Ϫ)-isomer of the agonist, observed with IANBD-labeled receptor denatured in guani- isoproterenol ((Ϫ)ISO). The experiments shown is representative of dinium chloride (data not shown). To investigate the kinetics of three identical experiments. Fluorescence measurements were done as this agonist-mediated change, the fluorescence intensity was described under “Experimental Procedures” with excitation set at 481 nm. measured as a function of time (Figs. 4 and 5). Prior to adding ligand we observed a slight but constant decline in base-line fluorescence is due to denaturation of the protein, since a fluorescence (Figs. 4 and 5). This loss of fluorescence over time similar loss of fluorescence also was observed with labeled is likely due to factors such as bleaching and hydrolysis of the receptor that was intentionally denatured in guanidinium chlo- probe during the experiments. It is unlikely that this decline in ride (see Fig. 7d). It should also be noted that the decrease over Saint Mary’s College of California Summer Research Program 2010 G. B. Diaz 3 centricon and the concentration determined using analysis, C-18 reversed-phase resin columns were -1 -1 its molar absorptivity ε280 = 73.11mM cm , which was used. The procedure used was provided by the kit. The calculated based on the amino acid content of TOP washes from each step was collected and saved to using the automated ProtParam Tool on the SWISS undergo HPLC analysis. ExPASy server1. An inhibitor solution was prepared using HPLC Analysis 100µM of angiotensin 1-7 and added at twice the An Agilent HPLC 1200 series was used along volume to maintain inhibitor concentration. The with a size exclusion column. The mobile phase was enzyme was incubated with the inhibitor for 5-10mins 100% acetonitrile HPLC reagent grade. The flow rate at room temperature; this kept the enzyme in a closed was set to 0.500mL/min and peaks were detected at confirmation, blocking the inner cysteines from the IA 481nm and 530nm by fluorescence along with the reaction. typical protein identifying wavelengths on the diode After incubation, an equivalent volume of IA array detector. was added to the enzyme and reacted for 90mins at 4ºC then 15mins at RT. The PD10 column was Activity Assays re-equilibrated with buffer without TCEP and the A fluorimeter was used to monitor reaction was run through to separate the excess IA enzyme activity; the instrument was Perkin Elmer solution. Fractions were collected and concentration Luminescence Spectrometer LS 50B. Activity assays determined as described above. were done to assess if modification of the enzyme A 15mM IANBD solution was prepared. The affected activity. MCA was the substrate used at high volume of the modified enzyme was brought up to and low concentrations. The assay required enzyme 400µL. IANBD was titrated slowly to the enzyme and in prepared buffer described above. Activity was reacted at 4ºC overnight. monitored under fluorescence spectroscopy at an The reaction was run through the PD10. The excitation of 325nm and emission of 400nm. enzyme was concentrated down and stored at -80ºC. The stoichiometry of protein to IANBD was calculated using the molar absorptivity of IANBD at Ligand Binding Assays 481nm4. The ligand binding assays had similar requirements as the activity assays. Fluorescent changes were compared between IANBD-modified Tryptic Digest TOP with and without the inhibitor angiotensin(1-7). The procedure for the digest was based on the Changes were monitored under varying method provided from the kit, along with reagents concentrations of enzyme and inhibitor. Excitation used. Digestion and reducing buffers were added to was at 481nm and emission wavelengths were 0.025-10µg of modified enzyme and incubated at monitored from 490nm to 625nm4 95ºC for 5 minutes. Alkylation buffer was added and reacted at room temperature for 20mins. The provided A summary of the procedure is depicted in Figure 3. trypsin was added and incubated at 37ºC for 3 hours. Another aliquot of trypsin was added and the reaction was incubated at 30ºC overnight. Re s u l t s & Di s c u s s i o n An optional guanidination procedure Optimization of HPLC usage was done followed digestion; this step converted lysines extensively to obtain favorable separation of distinct into homoarginines to enhance downstream mass peaks. Numerous buffers and flow rates were tested spectrometry analyses. on various peptides such as MCA and various forms of MCA i.e., MCA-bradykinin. Running the peptides and PepCleanTM C-18 Spin Columns undigested enzyme gave estimates on the retention To further purify the digested enzyme for time for the actual experiment. Figures 4 and 5 show Saint Mary’s College of California Summer Research Program 2010 G. B. Diaz 4

1 2 3 4

Folded Modified Unfolded Digested enzyme; Enzyme peptide fragments Figure 3: 1: Denaturation. 2: Digestion. The digest procedure was modified numerous times, to assure complete digestion of enzyme. Twice as much trypsin was added and overnight digestion was necessary. Also a higher amount (µg) of enzyme gave a better fluorescent intensity and compromised sample loss downstream. 3: Downstream purification. 4: HPLC. The chromatograph is enlarged in Figure 8.

HPLC chromatograms of a peptide and undigested enzyme. Evidence of enzyme modification was given by UV-vis spectroscopy. Figure 6 shows the spectra of TOP before and after modification. Activity assays were then done to verify that the modification did Fluorescence Intensity Fluorescence not affect enzymatic performance; Figure 7 depicts this. All enzyme concentrations were kept at 0.1µM. The substrate concentration used for WT and IA was 40.688µM and 81.376µM for IANBD-TOP. In the graph, min Figure 4: HPLC chromatogram of MCA-GnRH (Gonadotropin it shows that the activity of IANBD-TOP is twice as releasing hormone attached to MCA for fluorescent properties). After much than WT-TOP due to the difference of substrate numerous sample runs, a flow rate of 0.500m/min in acetonitrile concentration. However the activity of IA-TOP gave the best results. This data shows that peptide fragments were increased compared to WT-TOP. This is most likely due estimated to have a retention time of ~30mins when running the 5 digested TOP. to dimerized enzyme in WT-TOP where it would not be a problem for IA-TOP since the cysteine residues are covered. HPLC analysis followed the tryptic digest and samples were run to determine retention times of the peptide fragments. The results are shown on Figure 8. Fractions at the resulting retention times were Fluorescence Intensity Fluorescence collected. About 20-25 fractions were collected over the period of time the HPLC was running. The fractions were sampled under fluorescence spectrometry to locate the fluorescent tag. Wherever the tag was located via fluorimetry, the fraction was set aside to be min Figure 5: HPLC chromatogram of undigested, modified TOP. This sent for mass spectrophotometry analysis. data is a reference to running the digested TOP, if there is undigested Ligand binding assays were then done to enzyme left, the digest procedure needs to be modified by adding observe changes in fluorescence intensity when the more trypsin or longer incubation period. The retention time was enzyme would open and close. An initial change was ~10mins. The retention time shown at 27.587min is likely contamina- seen in sample 1 with 125µM angiotensin(1-7). This is tion. depicted as the red curve in Figure 9. This complied to the characteristic of IANBD; fluorescent intensity increases as polarity of solvent decreases. There must Saint Mary’s College of California Summer Research Program 2010 G. B. Diaz 5 /s) int Rate (F Absorbance (AU) Absorbance

Figure 6: UV-vis spectra of TOP before and after modification. The Figure 7: Activity rate comparison of TOP at different stages of modi- blue peak curve represents unmodified TOP at 280nm. The red curve is fication. WT for wild-type state. IA for iodoacetate attachment to TOP. modified TOP. The peak at 350nm is iodoacetate and IANBD at 481nm. And IANBD for the final step. Fluorescence Intensity Fluorescence Fluorescence Intensity Fluorescence

Figure 8: Three prominent peaks were seen after running multiple nm samples for consistency. The major retention times were these three as shown: 29.042mins, 31.231mins, and 37.033 minutes. Figure 9: Ligand Binding Assays of wild-type TOP modified with IANBD (WT-IANBD). 0.25µM was the initial concentration of enzyme used and inhibitor (Ang (1-7) was titrated. have been an increase in fluorescence because TOP is bound to angiotensin(1-7), which is a competitive Co n c l u s i o n inhibitor. As angiotensin was bound, TOP was in A method has been developed to successfully a closed conformation, pushing out solvent from modify TOP with a fluorescent tag without altering the tag and caused a change in the fluorescence of enzyme functional integrity. The assays done in IANBD. Another observation seen was a leg in the assessment of the enzyme needs to be repeated curve forming as opposed to trailing off to baseline for further consistency. Different substrates can also which was the case in WT-IANBD without angiotensin. be used for the ligand binding assays to observe if Samples following this varied in both substrate they have various affects in fluorescence intensity. and enzyme concentrations. The trend seen was a Any changes in intensity convey if TOP undergoes decrease in fluorescence to an intensity similar to WT- conformational changes. Data and information IANBD without inhibitor. However the leg in the curve obtained from this experiment can also be used to was still present. assess relative enzymes across species, particularly Currently the identification of the peptide neurolysin. fragment bearing the fluorescent tag is awaiting Once results from mass spectrometry are results from mass spectrometry from the facility in received, more information will be provided on the UCSF. location of the tag and structural changes in relation to where the tag is attached. Further experiments will be done as necessary. Saint Mary’s College of California Summer Research Program 2010 G. B. Diaz 6

Re f e r e n c e s 1. Sigman, J.A., et al. (2005) Flexibility in substrate recognition by thimet oligopeptidase as revealed by denaturation studies. Biochem. J. 388, 255-261. 2. Shrimpton, C.N., et al (2000) Biochem. Soc. Trans. 28, 430-434. 3. Koike, H., et al. (1999) Thimet Oligopeptidase Cleaves the Full-Length Alzheimer Amyloid Precursor Protein at a β-Secretase Cleavage Site in COS Cells. J. Biochem. 126, 235-242. 4. Gether, U., et al. (1995) Fluorescent Labeling of

Purified 2β Adrenergic Receptor. J. Bio. Chem. 270, 28268-28275. 5. Sigman, J.A., et al. (2003) Involvement of surface cysteines in activity and multimer formation of thimet oligopeptidase. Prot. Engineering. 16, 623-628.