CORE Metadata, citation and similar papers at core.ac.uk

Provided by Elsevier - Publisher Connector

FEBS Letters 587 (2013) 3021–3026

journal homepage: www.FEBSLetters.org

Additivity-based design of the strongest possible turkey ovomucoid third domain inhibitors for porcine pancreatic (PPE) and Streptomyces griseus protease B (SGPB) ⇑ Mohammad A. Qasim ,1, Lixia Wang 2, Sabiha Qasim, Stephen Lu 3, Wuyuan Lu 4, Richard Wynn 5, Zheng-Ping Yi 6, Michael Laskowski Jr. 7

Department of Chemistry, Purdue University, 1393 Brown Building, West Lafayette, IN 47907-1393, United States

article info abstract

Article history: We describe here successful designs of strong inhibitors for porcine (PPE) and Received 20 May 2013 Streptomyces griseus protease B (SGPB). For each two inhibitor variants were designed. In Revised 15 July 2013 one, the reactive site residue (position 18) was retained and the best residues were substituted at Accepted 16 July 2013 contact positions 13, 14, and 15. In the other variant the best residues were substituted at all contact Available online 23 July 2013 positions except the reactive site where a Gly was substituted. The four designed variants were: for PPE, T13E14Y15-OMTKY3 and T13E14Y15G18M21P32V36-OMTKY3, and for SGPB, S13D14Y15-OMTKY3 and 13 14 15 18 19 21 0 Edited by Robert B. Russell S D Y G I K -OMTKY3. The free energies of association (DG ) of expressed variants have been measured with the proteases for which they were designed as well as with five other serine prote- Keywords: ases and the results are discussed. Inhibitor design Ó 2013 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. Kazal inhibitor Protease inhibitor Additivity

1. Introduction form a stable non-covalent complex with the serine protease [1– 4]. Based on sequence homologies and disulfide bond topologies, Standard mechanism serine protease inhibitors bind to serine eighteen families of standard mechanism serine protease inhibi- proteases like substrates but instead of getting hydrolyzed they tors have been recognized [3,5]. Despite differences in sequences, size, and disulfide bonding patterns, all of the eighteen families follow the same mechanism of inhibition commonly called the Abbreviations: SRA, sequence to reactivity algorithm; OMTKY3, turkey ovomu- coid third domain; SGPA and SGPB, Streptomyces griseus protease A and B. In standard mechanism of inhibition [1]. MEROPS database and recent literature these are listed as Streptogrisin A and B; PPE, We have been involved in the research work aimed at develop- porcine pancreatic elastase; HLE, human leukocyte elastase; CARL, ing a sequence to reactivity algorithm (SRA) for the Kazal family of Carlsberg standard mechanism inhibitors. In the first part of our research ⇑ Corresponding author. project, ovomucoid third domains (a Kazal family inhibitor) were E-mail address: [email protected] (M.A. Qasim). 1 Present address: Department of Chemistry, Indiana U Purdue U Fort Wayne, IN prepared and purified from egg whites of a large number of species 46805, United States. of birds. The ovomucoid third domains were sequenced [6–8] and 2 Present address: Institute of Plant Protection, Chinese Academy of Agricultural free energy changes of their association (DG0) were measured with Sciences, Beijing 100193, PR China. 3 Present address: AbbVie BioResearch Center, 100 Research Drive, Worcester, MA a panel of six serine proteases [9–12]. In the second part of the pro- 01605, United States. ject all single amino acid variants at ten of the twelve consensus 4 Present address: Institute of Human Virology, University of Maryland School of contact positions of turkey ovomucoid third domain (OMTKY3) Medicine, 725 West Lombard Street, Baltimore, MD 21201, United States. (see Fig. 1) were prepared and their DG0 values were measured 5 Present address: Incyte Corporation, Experimental Station, E336/241B, Route 141 & Henry Clay Road, Wilmington, DE 19880, United States. against the same set of six serine proteases [13–15]. The culmina- 6 Present address: Department of Pharmaceutical Sciences, Eugene Applebaum tion of these two projects produced an SRA for the Kazal family of College of Pharmacy/Health Sciences, Wayne State University, Detroit, MI 48201, inhibitors, in addition to providing a large and unbiased set of United States. inhibitors for testing the algorithm. 7 Deceased.

0014-5793/$36.00 Ó 2013 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.febslet.2013.07.029 3022 M.A. Qasim et al. / FEBS Letters 587 (2013) 3021–3026

SRA. The substitution of Gly at P1 is based on overwhelming data [14,24–27] that show strong additivity of substitutions involving

the P1 position of inhibitors as well as substrates. Most of the pre- dicted and the measured values were in excellent agreement. The success of these studies emphasizes the importance of including more serine proteases in further developing the sequence to reac- tivity algorithm.

2. Materials and methods

2.1. Chemicals

Four of the six serine proteases used in this research namely, TLCK treated bovine a- (Worthington), human leuko- cyte elastase (HLE) (Elastin Products), porcine pancreatic elastase (PPE) (Sigma), and subtilisin Carlsberg (CARL) (Sigma) were ob- tained from the commercial sources listed in parentheses. The other two serine proteases, Streptomyces griseus protease A and B, (SGPA and SGPB) were purified from a commercially obtained preparation of (Sigma) as described [28]. The identity and the purity of the two proteases were established by amino acid

Fig. 1. Primary structure of expressed wtOMTKY3. The first five residues are not analysis and by analytical ion exchange chromatography. shown because they are not part of the expressed OMTKY3 (and its variants) and The chromogenic and fluorogenic synthetic substrates of the have been found to have no effect on the inhibitory activity of OMTKY3. The type succinyl-ala-ala-pro-Xxx-pNA and succinyl-ala-ala-pro-Xxx- residues are numbered sequentially as well as in Schechter–Berger notation [30].In AMC were purchased from BACHEM. Other chemicals used in this Schechter–Berger notation, the reactive site residue (shown by the arrow) is labeled work were all analytical grade. as P1. Residues towards the N-terminal of the P1 residue are sequentially labeled as P2, P3, ..., Pn whereas residues C-terminal to P1 residue are labeled as P10,P20, ..., Pn0. The amino acids shown as filled circles are the consensus contact residues in inhibitor–protease complexes. The two filled circles shown in grey color represent 2.2. Construction and expression of variants the residues which in addition to being the contact residues also play a structural role. Site-directed mutagenesis was carried out to introduce amino acid substitutions in the recombinant OMTKY3. For the variant S13D14Y15, the plasmid of variant Y15 was used as template, and An important assumption in our proposal of SRA was the addi- the following primers were used to create the indicated changes: tivity of DG0 values when substitutions at the contact positions of S13D14Y15-forward primer: 50-GAC TGT AGT GAG TAC CCT AGC OMTKY3 are made. In principle, a substitution at an inhibitor con- GAT TAC TGC ACG CTG-30;S13D14Y15-reverse primer: 50-CAG CGT tact position is additive if that position is independent of other GCA GTA ATC GCT AGG GTA CTC ACT ACA GTC-30. The variant positions in the inhibitor and it also does not produce alterations plasmid could be easily distinguished from the parental plasmid through protease contact residues [16,17]. Thus, the additivity de- by the digestion with Pst I. For the mutant S13D14Y15G18I19K21, pends both on the contact position of the inhibitor as well as on the the plasmid of the variant S13D14Y15 was further used as template, serine protease being investigated. We presented extensive (400) and the following primers were used: S13D14Y15G18I19K21-forward additivity tests in our SRA paper [15]. These tests were based on primer: 50-C TGC ACG GGG ATC TAC AAA CCT CTC TGT GGA natural ovomucoid third domains that differed from OMTKY3 at TC-30;S13D14Y15G18I19K21-reverse primer: 50-GA TCC ACA GAG two or more contact positions [6,14]. Since that time we have per- AGG TTT GTA GAT CCC CGT GCA G-30. formed many more additivity tests [18]. The general consensus in For the variant T13E14Y15, the plasmid of variant Y15 was used as all additivity tests is that most contact positions, with the excep- template, and the following primers were used to create the indi- 0 13 14 15 0 tion of the contact positions P2 and P1, are additive with the six ser- cated changes: T E Y -forward primer: 5 -GAC TGT AGT GAG ine proteases that we have used [15,16,19,20]. The two important TAC CCT ACG GAG TAT TGC ACG CTG-30;T13E14Y15-reverse primer: applications of additivity-based SRA are: (i) the prediction, with 50-CAG CGT GCA ATA CTC CGT AGG GTA CTC ACT ACA GTC-30. The few restrictions, of the free energy of association of any Kazal variant plasmid could also be easily distinguished from the inhibitor of known protein or gene sequence with any of the six parental plasmid by the digestion with Pst I. For the variant serine proteases we have used, and (ii) the design of strong, spe- T13E14Y15G18M21, the plasmid of the variant T13E14Y15 was further cific, or non-specific inhibitors for the six serine proteases. used as template, and the following primers were used: T13E14 Structure based design of strong and specific drugs and ligands Y15G18M21-forward primer: 50-G TAT TGC ACG GGG GAA TAC for target proteins is an area of great academic and practical inter- ATG CCT CTC TG-30;T13E14Y15G18M21-reverse primer: 50-CA GAG est [21–23]. In this communication, we describe the design and AGG CAT GTA TTC CCC CGT GCA ATA C-30. For the variant T13E14 expression of the strongest possible OMTKY3-based inhibitors for Y15G18M21P32V36, the plasmid of the variant T13E14Y15G18M21 was PPE and SGPB. We also measure the free energy changes in the further used as template, and the following primers were used: association of the designed inhibitors with the target serine prote- T13E14Y15G18M21 P32V36-forward primer: 50-CA TAT CCA AAC AAG ase as well as with the other five serine proteases in the panel and TGC GTC TTC TGC AAT G-30;T13E14Y15G18M21 P32V36-reverse compare them with the predicted free energy changes. The pre- primer: 50-C ATT GCA GAA GAC GCA CTT GTT TGG ATA TG-30. dicted free energy changes of association of the strongest possible All the substitutions were confirmed by DNA sequencing. Each inhibitors were outside our reliable measurement range variant plasmid was then transformed into Escherichia coli strain (4.0–17.5 kcal/mol). Therefore, to bring these numbers into the RV308 for protein expression. An engineered Z domain of protein measureable range we introduced a Gly at the P1 position of A was used as a fusion protein in the construction of variant the designed inhibitor instead of the best residue dictated by the plasmids [14]. The expressed protein inhibitors were purified by M.A. Qasim et al. / FEBS Letters 587 (2013) 3021–3026 3023

Table 1 3. Results and discussion Free energy changes of inhibitor–protease association. Standard free energy changes for the association of some OMTKY3 single variants with six serine proteases at pH 8.3 0 and 22 ± 1 °C. 3.1. DG values of inhibitor variants

OMTKY3 variants CHYM HLE PPE SGPA SGPB CARL The amino acid sequence of recombinant wild type OMTKY3 is DG0 (kcal/mol) shown in Fig. 1. The consensus contact residues of OMTKY3 deter- wtOMTKY3 15.23 13.21 14.34 15.49 14.51 14.22 mined from the X-ray crystallographic structures of its complexes S13OMTKY3 15.65 12.75 14.18 17.31 17.07 14.52 T13OMTKY3 14.85 13.04 14.43 17.30 16.44 13.34 with different serine proteases are shown as filled circles. There are D14OMTKY3 16.07 13.70 14.96 16.02 15.53 16.71 12 such contact residues in OMTKY3. Our lab, in collaboration with E14OMTKY3 15.94 14.62 15.05 16.12 15.13 16.44 Anderson’s lab at Rutgers, constructed and expressed all single 15 Y OMTKY3 16.09 13.39 15.41 16.82 16.14 15.05 amino acid variants at ten of the twelve contact positions shown 18 G OMTKY3 15.96 9.91 12.09 10.33 9.56 11.89 in Fig. 1. The two contact positions which were not subjected to I19OMTKY3 12.61 12.49 12.15 15.26 15.13 13.49 16 33 0 K21OMTKY3 14.57 12.61 13.61 15.16 14.78 14.31 substitution are Cys (P3) and Asn (P15 ). These residues, in addi- M21OMTKY3 12.64 13.96 14.84 15.57 14.38 13.78 tion to being contact residues, also serve as structural residues and P32OMTKY3 14.91 12.95 15.28 14.47 13.43 13.15 their mutation generally produces dramatic changes in the confor- 36 V OMTKY3 15.19 12.87 15.19 16.51 14.65 13.56 mation and stability of OMTKY3. In all, 190 inhibitor variants (corresponding to single amino acid changes at 10 contact posi- tions) and the wild type OMTKY3 were expressed, purified, and the DG0 values for the interaction of these variants with six serine affinity chromatography on an IgG-sepharose 6 fast flow column. proteases were measured [15]. The DG0 values of some of the After affinity separation the fusion protein was cleaved at an engi- variants that are relevant to the research work described in this neered methionine placed at the junction of the Z domain and the paper are listed in Table 1. ovomucoid third domain variant. The inhibitor variants were then separated from cleaved fusion protein by size exclusion column chromatography on Bio-gel P-10 column and purified by ion ex- 3.2. Design of the strongest possible inhibitors for PPE and SGPB change column chromatographies on SP-sepharose and Q-sephar- ose columns. The variants were characterized by size exclusion The strongest OMTKY3-based inhibitor for PPE and SGPB will be HPLC, amino acid analysis, and by mass spectral analysis by MALDI the one that has a residue at each of the contact positions that pro- TOF. duces the highest association equilibrium constant for these prote- ases. The sequences of the strongest possible inhibitors for all six serine proteases have been described [15]. For PPE and SGPB, these 2.3. Measurement of free energy changes in the association of 13 14 15 21 32 36 13 14 15 19 21 inhibitors with proteases sequences are: T E Y M P V and S D Y I K , respec- tively. Here, only the residues that are different from wtOMTKY3 (Fig. 1) are shown. Assuming full additivity at each contact position The free energy changes in the association of the inhibitors with 0 the panel of six serine proteases were calculated from experimen- one can calculate the free energy change (DG predicted) in the asso- ciation of these variants with any of the six serine proteases by tally determined values of association equilibrium constants, Ka,by 0 using the following relation [15]: using the equation, DG = RTlnKa. Association equilibrium con- stants for the binding of the inhibitor variants with the serine pro- 0 0 0 0 teases were determined by a procedure perfected in this lab [9,14]. DGpredicted ¼ DGTKY3 þ RDDG ðXTKY3iX Þð1Þ

The Ka measurements, except in those cases where they were ex- 13 1 0 0 pected to be >10 M , were performed in 0.1 M Tris–HCl buf- The term DDG (XTKY3 i X ) represents the change in the free en- fer + 0.02 M CaCl2 + 0.005% Triton X-100, pH 8.3. The technical ergy of association when a contact residue in OMTKY3, X, at posi- difficulties such as long incubation times (several weeks) and tion i is changed to X0. The summation extends to all contact non-availability of sensitive enough substrates to accurately deter- positions. The free energies of association of OMTKY3 and single mine picomolar concentrations of the protease used in these mea- variants used for the design of these inhibitors are shown for all 13 1 surements, prevent us from measuring large Ka values (>10 M ) six in Table 1. If the above equation is used to calculate at pH 8.3. However, we have found that the Ka measurement range free energy of association for the two inhibitors that are designed can be increased by about a factor of 10 for some enzymes (such as for PPE and SGPB, their predicted free energies of association will 13 1 SGPA, SGPB and chymotrypsin) by performing the Ka measure- come out to be 18.5 kcal/mol (Ka = 5.1 10 M ) for inhibitor ments at pH 5.0 and then converting these values to pH 8.3 by sequence T13E14Y15M21P32V36 (for PPE) and 20.6 kcal/mol 15 1 13 14 15 19 21 using an appropriate conversion factor. As part of our studies on (Ka = 1.8 10 M ) for S D Y I K (for SGPB). Our range of 0 pH-dependencies of Ka, we measured Ka values of a number of P1 accurate DG determinations ranges from 4.0 kcal/mol to 3 1 13 1 variants of OMTKY3 with different serine proteases in the pH range 17.5 kcal/mol (Ka:10 M to 10 M ). These numbers, however, 4.0–10.0. The pH dependence for variants having non-ionizable represent the lower and the upper limit of our measurements amino acid residues at P1 was found to be identical, within exper- when all six serine proteases are considered. For individual en- imental error, for a given protease ([29] and Qasim and Laskowski zymes these ranges are slightly different. For example, for PPE

– unpublished). This means that the ratio of Ka for P1L variant (or the upper limit is 16.2 kcal/mol, whereas for SGPB it is 17.5 kcal/ 0 any other non-ionizable P1 variant) at pH 8.3 and at any given mol. We were able to extend the upper limit of DG value for SGPB 0 pH is constant. Such a ratio at pH 8.3 and pH 5.0 comes out to be to 18.9 kcal/mol (a 10-fold increase in Ka) by determining DG at 0 115 for SGPB. This factor can be used for estimating Ka value at pH 5.0 and converting it to the DG at pH 8.3 by using our knowl- 0 0 pH 8.3 from a measured Ka at pH 5.0. This method works well edge of pH dependence of DG . The predicted DG values for the 13 1 for measurements in which the Ka value is P10 M . Such mea- strongest possible inhibitor of PPE and SGPB are clearly outside surements are more easily done at pH 5.0 because the drop in Ka our measurement range. Therefore, although we could make the with pH is steeper than the drop in the enzyme activity. The Ka strongest inhibitors using the above sequences, we would not be measurements at pH 5.0 were performed in 0.1 M acetic acid–ace- able to measure them in order to see whether they are actually tate buffer + 0.02 M CaCl2 + 0.005% Triton X-100, pH 5.0. as strong as our predictions suggest. In order to overcome this 3024 M.A. Qasim et al. / FEBS Letters 587 (2013) 3021–3026

problem, we made the strongest inhibitors in which the best P1 of association and the measured free energy of association 0 0 0 residue was replaced by the simplest of the amino acids, i.e. Gly. (DG predicted DG measured) is represented as DG I (see Eq. (2)). 0 The choice of P1 and the choice of the amino acid Gly for substitu- The term DG I is in effect the sum of any non-additivity and any tion in the designed inhibitor were made on the basis of clear mer- experimental error inherent in the acquisition of the data set that 0 its. The P1 is the most additive of all of the contact residues. The is used as the predictive tool. The error in the measurement of DG data on the additivity of the P1 position are overwhelming value for OMTKY3 for all six enzymes is ±100 cal/mol. The error is [3,5,24–27]. The choice of Gly at P1 is based on the simplicity of generally higher at the lower and the upper ends of our measure- this amino acid (no side chain) and its uniformly lower DG0 value ment range. Thus, in our overall calculation of error, we use a 0 for all serine proteases. 2r level to decide whether the DG I is due to experimental error or whether it should be attributed to non-additivity. For details

3.3. Strongest inhibitors with a Gly at P1 of error analysis, readers are referred to our SRA paper [15]. The 0 calculation of DG I is performed by using the following equation: The design strategy for these inhibitors was exactly the same as qffiffiffiffiffi qffiffiffiffiffi described above except that the best amino acid residue at P was 1 0 2 2 replaced by a Gly. The two designed inhibitors are designated as DGI experimental ¼2r k k þ 2 ¼200cal=mol k k þ 2 ð2Þ T13E14Y15G18M21P32V36-OMTKY3 and S13D14Y15G18I19K21-OMTKY3. For each of these designed inhibitors, we also produced an inter- Here, k represents the number of amino acid changes from 0 mediate designed inhibitor (see Tables 2 and 3) that had the best OMTKY3 sequence. The DG I value for three changes comes out residues at P6,P5, and P4 positions but having the rest of the con- to be 0.57 kcal/mol and for six and seven changes 1.13 kcal/mol tact residues left the same as in wtOMTKY3. Having inhibitors with and 1.33 kcal/mol, respectively. According to this criterion, 17 of fewer substitutions is important particularly in situations where the 24 (71%) DG0 values are in good to excellent agreement with the measured DG0 values for the designed inhibitor do not match the predicted DG0 values (see Tables 2 and 3). This is slightly better 0 with the predicted DG values. The measured and predicted Kas than the value of 63% found for the much larger set of 400 compar- and the free energies of association of designed inhibitors for PPE isons [15]. Of the seven measured values that do not agree with the and SGPB are given in Tables 2 and 3. The DG0 values shown in Ta- predicted values, the one for S13D14Y15G18I19K21-OMTKY3 for SGPB bles 2 and 3 are at pH 8.3. With the exception of T13E14Y15 and is only marginally outside the allowed error range. For the other six S13D14Y15 values for SGPA and SGPB all other values were directly a combination of different factors may be responsible for the dis- measured at pH 8.3. The DG0 values for T13E14Y15 and S13D14Y15 for agreement. It is worthwhile to mention here that of the six en- SGPA and SGPB were measured at pH 5.0 and converted to DG0 val- zymes we use, the two most non-additive enzymes are HLE and ues at pH 8.3 by using the extensive DG0 data that we have ac- CARL [15,31]. Three of the six predominantly non-additive num- quired at both of these pH values. A cursory look at the bers reported here are for HLE and CARL. The structural explana- measured and predicted DG0 values shows that most of these are tion for the greater non-additivity in CARL has been provided by close to each other. The question, how close the measured and the X-ray crystallographic structure determination of the OMT- the predicted DG0 values should be to each other, in order to be KY3–CARL complex [32]. On the other hand, the greater degree considered within experimental error, was addressed in detail of non-additivity in HLE is at least in part due to inherent difficul- in our SRA paper [15]. The difference in the predicted free energy ties in accurate determinations of DG0 for this protease. In spite of

Table 2 Equilibrium constants and free energies of association for the designed inhibitors. Association equilibrium constants and free energies of associations of the two inhibitors designed for PPE. The values listed here are at pH 8.3 and 22 ± 1 °C. Predicted values were calculated using the data given in Table 1 as described in the text.

OMTKY3 variants CHYM HLE PPE SGPA SGPB CARL

1 Ka (M ) T13E14Y15 Measured 9.4 1011 2.2 1010 6.8 1011 1.4 1014 3.8 1013 1.5 1011 Predicted 1.5 1012 7.0 1010 1.0 1012 1.8 1014 6.9 1013 1.4 1012 T13E14Y15G18M21P32V36 Measured 9.2 106 3.1 108 3.0 1011 3.1 1010 2.5 109 5.8 108 Predicted 3.4 105 3.2 108 1.1 1012 3.0 1010 2.8 109 6.4 108 DG0 (kcal/mol) T13E14Y15 Measured 16.16 13.96 15.97 19.10 18.33 15.09 Predicted 16.44 14.64 16.20 19.24 18.68 16.40 T13E14Y15G18M21P32V36 Measured 9.40 11.46 15.49 14.16 12.69 11.83 Predicted 7.47 11.48 16.26 14.14 12.75 11.89

Table 3 Equilibrium constants and free energies of association for the designed inhibitors. Association equilibrium constants and free energies of associations of the two inhibitors designed for SGPB. The values listed here are at pH 8.3 and 22 ± 1 °C. Predicted values were calculated using the data given in Table 1 as described in the text.

OMTKY3 variants CHYM HLE PPE SGPA SGPB CARL

1 Ka (M ) S13D14Y15 Measured 1.1 1012 7.6 109 5.3 1011 9.2 1013 6.1 1013 8.9 1011 Predicted 7.2 1012 8.7 109 5.7 1011 1.6 1014 1.6 1014 1.7 1013 S13D14Y15G18I19K21 Measured 9.0 105 6.8 106 1.4 108 4.6 109 4.8 1010 7.8 109 Predicted 9.4 105 3.4 106 8.3 107 8.9 109 3.9 1011 1.1 1011 DG0 (kcal/mol) S13D14Y15 Measured 16.26 13.34 15.82 18.85 18.61 16.13 Predicted 17.36 13.42 15.87 19.18 19.18 17.86 S13D14Y15G18I19K21 Measured 8.04 9.22 11.00 13.04 14.41 13.40 Predicted 8.06 8.82 10.69 13.43 15.65 14.91 M.A. Qasim et al. / FEBS Letters 587 (2013) 3021–3026 3025 some failures, the results in general are highly encouraging. First, [8] Apostol, I., Giletto, A., Komiyama, T., Zhang, W. and Laskowski Jr., M. (1993) despite six and seven changes in the contact position of designed Amino acid sequences of ovomucoid third domains from 27 additional species of birds. J. Protein Chem. 12, 419–434. inhibitors, the additivity works very well. Second, the measured [9] Empie, M.W. and Laskowski Jr., M. (1982) Thermodynamics and kinetics of DG0 values for the designed inhibitors against the proteases for single residue replacements in avian ovomucoid third domains: effect on which they were designed were reasonably good. Third, quite inhibitor interactions with serine proteinases. Biochemistry 21, 2274–2284. [10] Laskowski, M. Jr., Park, S. J., Tashiro, M. and Wynn, R. (1989) In Protein astonishingly the designed strongest inhibitor for SGPB has signif- recognition of immobilized ligands: UCLA Symposium on Molecular and icantly higher DG0 value for SGPB than that for SGPA. The two en- Cellular Biology (Hutcgens, T. W. Ed.) p. 149–168, Alan R. Liss Inc., New York. zymes are sequentially and structurally very closely related [11] Park, S. J. (1985) Effect of amino acid replacements in ovomucoid third domains upon their association with serine proteinases. Ph. D. thesis. Purdue [33,34]. Enzymatically SGPA is more active [35] and also is inhib- University, West Lafayette, IN. ited about 2- to 20-fold more strongly by most of the naturally [12] Wynn, R. (1990) Design of a specific human leukocyte elastase inhibitor based on occurring inhibitors that we have tested in our lab. The fact that ovomucoid third domains. Ph. D. thesis. Purdue University, West Lafayette, IN. 13 14 15 18 19 21 [13] Bigler, T.L., Lu, W., Park, S.J., Tashiro, M., Wieczorek, M., Wynn, R. and the S D Y G I K OMTKY3 inhibits SGPB 10 times more Laskowski Jr., M. (1993) Binding of amino acid side chains to preformed strongly than SGPA is important and suggests that our SRA can cavities: interaction of serine proteinases with turkey ovomucoid third be used for the design of specific inhibitors for closely related ser- domains with coded and non-coded P1 residues. Protein Sci. 2, 786–799. ine proteases. [14] Lu, W., Apostol, I., Qasim, M.A., Warne, N., Wynn, R., Zhang, W.L., Anderson, S., Chiang, Y.W., Ogin, E., Rothberg, I., Ryan, K. and Laskowski Jr., M. (1997)

The findings of the increasing roles of serine proteases in cancer Binding of amino acid side-chains to S1 cavities of serine proteinases. J. Mol. [36], in inflammations [37], and in many viral infections [38] Biol. 266, 441–461. makes them an important target for the design of strong and spe- [15] Lu, S.M., Lu, W., Qasim, M.A., Anderson, S., Apostol, I., Ardelt, W., Bigler, T., Chiang, Y.W., Cook, J., James, M.N.G., Kato, I., Kelly, C., Kohr, W., Komiyama, T., cific inhibitors. Such designed inhibitors can be used as molecular Lin, T.-Y., Ogawa, M., Otlewski, J., Park, S.-J., Qasim, S., Ranjbar, M., Tashiro, M., tools in the investigation of the structure and properties of target Warne, N., Whatley, H., Wieczorek, A., Wieczorek, M., Wilusz, T., Wynn, R., serine proteases as well as can also be used as therapeutics. SRAs Zhang, W. and Laskowski Jr., M. (2001) Predicting the reactivity of proteins from their sequence alone: Kazal family of protein inhibitors of serine similar to the one developed for the six serine proteases and used proteinases. Proc. Natl. Acad. Sci. USA 98, 1410–1415. here can in principle also be developed for other serine proteases [16] Laskowski Jr., M., Qasim, M.A. and Yi, Z. (2003) Additivity-based prediction of using our OMTKY3 variant set and can then be used for inhibitor equilibrium constants for some protein–protein associations. Curr. Opin. Struct. Biol. 13, 130–139. design against them. [17] Lee, T.-W., Qasim, M.A., Laskowski Jr., M. and James, M.N.G. (2007) Structural insights into the non-additivity effects in the sequence-to-reactivity algorithm Author contribution for serine peptidases and their inhibitors. J. Mol. Biol. 367, 527–546. [18] Qasim, M.A., Lu, W., Lu, S.M., Ranjbar, M., Yi, Z., Chiang, Y.-W., Ryan, K., Anderson, S., Zhang, W., Qasim, S. and Laskowski Jr., M. (2003) Testing of the Mohammad Qasim performed measurements of DG0 values, additivity-based protein sequence to reactivity algorithm. Biochemistry 42, interpretations of data, and writing of the manuscript. Lixia Wang 6460–6466. [19] Li, J., Yi, Z., Laskowski, M.C., Laskowski Jr., M. and Bailey-Kellogg, C. (2005) performed all molecular biology operations and wrote the section Analysis of sequence-reactivity space for protein–protein interactions. on ‘Construction and Expression of Variants’. Sabiha Qasim ex- Proteins: Struct., Funct., Genet. 58, 661–671. pressed and purified variants and determined amino acid composi- [20] Yi, Z., Vitek, O., Qasim, M.A., Lu, S.M., Lu, W., Ranjbar, M., Li, J., Laskowski, M.C., Bailey-Kellogg, C. and Laskowski Jr., M. (2006) Functional evolution within a tions. Stephen Lu, Wuyuan Lu, Richard Wynn and Zhen-Ping Yi . Proteins: Struct., Funct., Genet. 63, 697–708. contributed in the development of SRA. All authors contributed [21] Smith, A.T.J., Zhang, X., Leach, A.G. and Houk, K.N. (2009) Beyond picomolar in the discussion of results and in the reading of the manuscript. affinities: Quantitative aspects of non-covalent and covalent binding of drugs to proteins. J. Med. Chem. 52, 225–233. Michael Laskowski Jr. directed the research work. [22] Jorissen, R.N., Kiran, G.S., Reddy, K., Ali, A., Altman, M.D., Chellappan, O.S., Anjum, S.G., Tidor, B., Schiffer, C.A., Rana, T.M. and Gilson, M.K. (2009) Additivity in the analysis and design of HIV protease inhibitors. J. Med. Chem. Acknowledgments 52, 737–754. [23] Greving, M.P., Belcher, P.E., Diehnelt, C.W., Gonzalez-Moa, M.J., Emery, J., Fu, J., The research work was supported at Purdue by a National Insti- Johnston, S.A. and Woodbury, N.W. (2010) Thermodynamic additivity of tute of Health grant GM10831. We thank Professor Michael James, sequence variations: an algorithm for creating high affinity peptides without large libraries or structural information. PLoS ONE 5, e15432. Department of Biochemistry, University of Alberta, for careful [24] Qasim, M.A., Ganz, P.J., Saunders, C.W., Bateman, K.S., James, M.N.G. and reading of the manuscript and for suggesting many corrections. Laskowski Jr., M. (1997) Interscaffolding additivity. Association of P1 variants of eglin c and of turkey ovomucoid third domain with serine proteinases. Biochemistry 36, 1598–1607. References [25] Qasim, M.A., Lu, S.M., Ding, J., Bateman, K.S., James, M.N.G., Anderson, S., Song, J., Markley, J.L., Ganz, P.J., Saunders, C.W. and Laskowski Jr., M. (1999)

[1] Laskowski Jr., M. and Kato, I. (1980) Protein inhibitors of proteinases. Annu. Thermodynamic criterion for the conformation of P1 residues of substrates Rev. Biochem. 49, 593–626. and of inhibitors in complexes with serine proteinases. Biochemistry 38, [2] Read, R.J. and James, M.N.G. (1986) Introduction to the Protein Inhibitors: X- 7142–7150. ray Crystallography in: Proteinase Inhibitors (Barrett, A.J. and Salvesen, G., [26] Qasim, M.A., Van Etten, R.L., Yeh, T., Saunders, C., Ganz, P.J., Qasim, S., Wang, L. Eds.), pp. 301–336, Elsevier Science Publishers B.V., Amsterdam. and Laskowski Jr., M. (2006) Despite having a common P1 leu, eglin C inhibits [3] Laskowski Jr., M., Qasim, M.A. and Lu, S. (2000) Interaction of standard a-lytic proteinase a million-fold more strongly than does turkey ovomucoid mechanism, canonical protein inhibitors with serine proteinases in: Protein– third domain. Biochemistry 45, 11342–11348. Protein Recognition (Kleanthous, C., Ed.), pp. 228–279, Oxford University [27] Qasim, M.A., Song, J., Markley, J.L. and Laskowski Jr., M. (2010) Cleavage of Press, Oxford. peptide bonds bearing ionizable amino acids at P1 by serine proteases with [4] Otlewski, J., Jelen, F., Zakrzewska, M. and Oleksy, A. (2005) The many faces of hydrophobic S1 pocket. Biochem. Biophys. Res. Commun. 400, 507–510. protease–protein inhibitor interaction. EMBO J. 24, 1303–1310. [28] Qasim, M.A. (2013) Streptogrisin B in: Handbook of Proteolytic Enzymes [5] Laskowski Jr., M. and Qasim, M.A. (2000) What can the structures of enzyme– (Rawlings, N.D. and Woessner, J.F., Eds.), 3rd ed, pp. 2549–2555, Oxford inhibitor complexes tell us about the structures of enzyme substrate Academic Press. complexes? Biochim. Biophys. Acta, Protein Struct. Mol. Enzymol. 1477, [29] Qasim, M.A., Ranjbar, M.R., Wynn, W., Anderson, S. and Laskowski Jr., M. 324–337. (1995) Ionizable P1 residues in serine proteinase inhibitors undergo large pK [6] Laskowski Jr., M., Kato, I., Ardelt, W., Cook, J., Denton, A., Empie, M.W., Kohr, shifts on complex formation. J. Biol. Chem. 270, 27419–27422. W.J., Park, S.J., Parks, K., Schatzley, B.L., Schoenberger, Oe.L., Tashiro, M., Vichot, [30] Schechter, I. and Berger, A. (1967) On the size of the in proteases I G., Whatley, H.E., Wieczorek, A. and Wieczorek, M. (1987) Ovomucoid third Papain. Biochem. Biophys. Res. Commun. 27, 157–162. domains from 100 avian species: isolation, sequences, and hypervariability of [31] Lu, S. (2000) The specificity of ovomucoid third domain at residues other than enzyme–inhibitor contact residues. Biochemistry 26, 202–221. P1. Ph. D. thesis. Purdue University, West Lafayette, IN. [7] Laskowski Jr., M., Apostol, I., Ardelt, W., Cook, J., Giletto, A., Kelly, C.A., Lu, W., [32] Maynes, J.T., Cherney, M.M., Qasim, M.A., Laskowski Jr., M. and James, M.N.G. Park, S.J., Qasim, M.A., Whatley, H.E., Wieczorek, A. and Wynn, R. (1990) Amino (2005) Structure of the subtilisin Carlsberg–OMTKY3 complex reveals two acid sequences of ovomucoid third domain from 25 additional species of birds. different ovomucoid conformations. Acta Crystallogr., Sect. D: Biol. J. Protein Chem. 9, 715–725. Crystallogr. 61, 580–588. 3026 M.A. Qasim et al. / FEBS Letters 587 (2013) 3021–3026

[33] James, M.N.G., Delbaere, L.T.G. and Brayer, G.D. (1978) Amino acid sequence [36] Obiezu, C.V. and Diamandis, E.P. (2005) Human tissue family: alignment of bacterial and mammalian pancreatic serine proteinases based on applications in cancer. Cancer Lett. 224, 1–22. topological equivalences. Can. J. Biochem. 56, 396–402. [37] Pham, C.T.N. (2006) Neutrophil serine proteases: specific regulators of [34] Delbaere, L.T., Brayer, G.D. and James, M.N.G. (1979) The 2.8 Å resolution inflammation. Nat. Rev. 6, 541–550. structure of Streptomyces griseus protease B and its homology with alpha- [38] Di Marco, S., Rizzi, M., Volpari, C., Walsh, M.A., Narjes, F., Colarusso, S., De chymotrypsin and Streptomyces griseus protease A. Can. J. Biochem. 57, 135–144. Francesco, R., Matassa, V.G. and Sollazzo, M. (2000) Inhibition of the Hepatitis [35] Bauer, C.-A. (1978) Active centers of Streptomyces griseus protease I, C virus NS3/4A protease. The crystal structure of two protease–inhibitor Streptomyces griseus protease 3, and chymotrypsin: enzyme–substrate complexes. J. Biol. Chem. 275, 7152–7157. interactions. Biochemistry 17, 375–380.