Báo cáo khoa học: Limited mutagenesis increases the stability of human carboxypeptidase U (TAFIa) and demonstrates the importance of CPU stability over proCPU concentration in down-regulating fibrinolysis doc

Limited mutagenesis increases the stability of human carboxypeptidase U (TAFIa) and demonstrates the importance of CPU stability over proCPU concentration in down-regulating fibrinolysis Wolfgang Knecht 1 , Johan Willemse 3 , Hanna Stenhamre 1 , Mats Andersson 2 , Pia Berntsson 1 , Christina Furebring 2 , Anna Harrysson 1 , Ann-Christin Malmborg Hager 2 , Britt-Marie Wissing 1 , Dirk Hendriks 3 and Philippe Cronet 1 1 AstraZeneca R & D Mo ¨ lndal, Mo ¨ lndal, Sweden 2 Alligator Bioscience AB, Lund, Sweden 3 Laboratory of Medical Biochemistry, University of Antwerp, Wilrijk, Belgium The fragile balance between the activities of the coagu- lation cascade (thrombin generation) and the fibrino- lytic system (plasmin generation) is essential to prevent excessive blood loss upon damage of a blood vessel, while maintaining the blood flow in parts of the body distant from the injury. Procarboxypeptidase U [proCPU, thrombin-activatable fibrinolysis inhibitor (TAFI), EC 3.4.17.20, MEROPS M14.009] belongs to the metallocarboxypeptidase family and is a human plasma zymogen, which is also known as thrombin-activatable fibrinolysis inhibitor (TAFI), plasma procarboxypeptidase B and procarboxypepti- dase R [1,2]. ProCPU has been proposed to be a molecular link between coagulation and fibrinolysis [3,4]. The physiological role of proCPU and its activa- ted form, carboxypeptidase U (CPU) is outlined in Fig. 1. ProCPU is synthesized in the liver and secreted into the plasma following the removal of its signal Keywords carboxypeptidase; coagulation; directed evolution; fibrinolysis; protease Correspondence W. Knecht, Molecular Pharmacology – Target Production, AstraZeneca R & D Mo ¨ lndal, 431 83 Mo ¨ lndal, Sweden Fax: + 46 317763753 Tel: + 46 317065341 E-mail: wolfgang.knecht@astrazeneca.com (Received 5 November 2005, accepted 19 December 2005) doi:10.1111/j.1742-4658.2006.05110.x Procarboxypeptidase U [proCPU, thrombin-activatable fibrinolysis inhib- itor (TAFI), EC 3.4.17.20] belongs to the metallocarboxypeptidase family and is a zymogen found in human plasma. ProCPU has been proposed to be a molecular link between coagulation and fibrinolysis. Upon activation of proCPU, the active enzyme (CPU) rapidly becomes inactive due to its intrinsic instability. The inherent instability of CPU is likely to be of major importance for the in vivo down-regulation of its activity, but the under- lying structural mechanisms of this fast and spontaneous loss of activity of CPU have not yet been explained, and they severely inhibit the structural characterization of CPU. In this study, we screened for more thermostable versions of CPU to increase our understanding of the mechanism underly- ing the instability of CPU’s activity. We have shown that single as well as a few 2–4 mutations in human CPU can prolong the half-life of CPU’s activity at 37 °C from 0.2 h of wild-type CPU to 0.5–5.5 h for the mutants. We provide evidence that the gain in stable activity is accompanied by a gain in thermostability of the enzyme and increased resistance to proteo- lytic digest by trypsin. Using one of the stable mutants, we demonstrate the importance of CPU stability over proCPU concentration in down-regu- lating fibrinolysis. Abbreviations BEVS, Baculovirus expression vector system; CLT, clot lysis time; CPB, carboxypeptidase B; CPU, carboxypeptidase U; EPP, error prone PCR; Hip-Arg, hippuryl- L-arginine; ORF, open reading frame; PTCI, potato tuber carboxypeptidase inhibitor; TAFI, thrombin-activatable fibrinolysis inhibitor; WT, wild type. 778 FEBS Journal 273 (2006) 778–792 ª 2006 The Authors Journal compilation ª 2006 FEBS peptide (prepeptide, see Fig. 2). It can be activated from its zymogen form to CPU by thrombin, plasmin or most efficiently the thrombin–thrombomodulin complex by cleavage after R114 [1,5,6](Fig. 2). In con- tact with a fibrin clot, CPU attenuates fibrinolysis by removing carboxy-terminal lysines from partially cleaved fibrin molecules, thereby diminishing its cofac- tor activity for activation of plasminogen to plasmin [7–9]. Following its activation, CPU’s activity is unsta- ble both in vivo and in in vitro experiments (as an isolated protein), with reported half-lives at 37 °C from 8 to 15 min, hence the U in its name stands for unsta- ble [10,11]. The inherent and irreversible decay of CPU’s activity is believed to be of major importance for its in vivo down-regulation of activity and has been linked to structural changes of the enzyme [3,12,13]. In vivo, CPU can also be inactivated by proteolytic degradation, indicating more accessible and flexible parts of the molecule exist. It was therefore suggested that the instability of CPU’s activity is due to intrinsic structural lability of the enzyme, priming its inactiva- tion [14]. Because of its prominent bridging function between coagulation and fibrinolysis, the development of CPU inhibitors as pro-fibrinolytic agents is an attractive concept [15,16]. But the instability of the enzyme has prevented crystallization of CPU and the use of struc- turally based drug design methods. A three-dimen- sional model of human proCPU based on the structure of human pancreas procarboxypeptidase B, a closely related protease exhibiting a higher stability, has been published recently by Barbosa Pereira et al. [17]. Recently, it was reported independently by two separate groups that CPU prevents clot lysis from proceeding into the propagation phase through a threshold-dependent mechanism [18,19]. The study of this threshold phenomenon and, more generally, the study of the effect of CPU on fibrinolysis, are also severely complicated by its intrinsic instability of activity. ‘Directed evolution’ approaches allow the random generation of a large number of mutants followed by selection for the desired features. Several proteins have been changed towards more desired properties using this approach. Some examples include deoxyribonucleo- side kinases for changed substrate specificities [20,21], phosphotriesterase for improved catalytic rates [22], haem peroxidase for exotic environments (for example, inside a washing machine) [23], or amylase and sub- tilisin for improved thermostability [24,25]. In this study, we present the generation of CPU mutants with highly stable activity obtained by molecular evolution techniques and selection for decreased thermo-inactivation. To achieve this we used a directed evolution approach comprising the genera- tion of random libraries and recombination of advan- tageous mutations by Fragment-INduced Diversity (FIND TM ) technology, as well as site-directed muta- genesis. A high-throughput screen based on mamma- lian cells expressing proCPU mutants was developed to select CPU variants with more thermostable activ- ity. Seven proCPU mutants were selected and purified. After activation by thrombin–thrombomodulin, three showed a remaining activity of more than 80% after 24-h incubation at 22 °C versus 20% for the wild type (WT), and two of these three showed a more than 25-fold increase in half-life of activity at 37 °C. Using one of the stable mutants, we have demonstrated the importance of CPU stability over proCPU concentra- tion in down-regulating fibrinolysis. Results To investigate the role of exposed hydrophobic resi- dues on the stability of CPU’s activity, 13 point muta- tions were introduced in proCPU by site-directed mutagenesis and expressed in 3T3 cells (F135Q, I147S, F201T, I204Y, I205E, I204Y ⁄ I205E, L214N, F244T, L281S, L335S, L376Q, T347I). Based on the alignment of CPU sequence to the structure of carboxypeptidase Fig. 1. Physiological role of CPU. CPU attenuates fibrinolysis by removing C-ter- minal exposed lysines from partially degra- ded fibrin. W. Knecht et al. Stable human CPU mutants FEBS Journal 273 (2006) 778–792 ª 2006 The Authors Journal compilation ª 2006 FEBS 779 B (CPB) [26] (Fig. 2), these mutants were chosen to replace hydrophobic amino acids of human CPU with more hydrophilic residues located on the surface of porcine CPB. In addition, the T347I naturally occur- ring variation in CPU was reported to double the half-life (T1 ⁄ 2) of its activity at 37 °C [11] and was therefore included. We found that the T347I mutant, when tested in cell culture supernatant, was only 50% more stable than our WT CPU with threonin at posi- tion 347 (Table 1). Recently, Barbosa Pereira et al. [17] proposed, on the basis of their model of human CPU, that the two consecutive I at positions 204 and 205 are exposed to the surface, and because they are quite unique to CPU, might be of importance for the process of CPU’s activity destabilization. When we changed these two amino acids to their counterpart in porcine CPB (I204Y ⁄ I205E), the T1 ⁄ 2 of the mutants’ activity was unchanged compared with WT CPU (data not shown). In order to create a high number of mutants, ran- dom mutagenesis was done using either error prone PCR (EPP) or creating a library of mutants with the Genemorph PCR mutagenesis kit (GMK, Stratagene, La Jolla, CA, USA). Sequencing of the full open read- ing frame (ORF) of randomly picked clones from these two approaches revealed a base mutation frequency of 0.41 ± 0.22% and 0.55 ± 0.23% per clone in 19 clones from the EPP library and in 17 clones from the Fig. 2. Multiple alignment of human preproCPU, human preproCPB and porcine proCPB. The amino acid sequences of human preproCPU (accession number AAP35582.1), human preproCPB (accession number P15086) and porcine proCPB (accession number 1NSA) were aligned using CLUSTAL W [40]. The pre- and the propeptide in preproCPU are shaded in black and grey, respectively. Amino acid exchanges found in mutants with increased thermostability of CPU’s activity are marked in yellow. ‘*’ means that the residues that column are identical in all sequences in the alignment. ‘:’ means that conserved substitutions have been observed ‘.’ means that semiconserved substitutions are observed. Stable human CPU mutants W. Knecht et al. 780 FEBS Journal 273 (2006) 778–792 ª 2006 The Authors Journal compilation ª 2006 FEBS GMK library, respectively. On the amino acid level this corresponded to an average of 3.7 or 5.2 exchanges per enzyme for the error-prone PCR or the Genemorph kit, respectively. In total, 24 600 clones, 14 600 from the EPP library and 10 000 from the GMK library were screened for improved thermostability of CPU activity in the super- natant of mammalian cells in an HTS format. The best clones selected were more thoroughly analyzed using an HPLC-based activity assay for CPU. The most sta- ble clone, GMK1, is five times more stable than WT CPU. From both libraries, about 1 in every 5000 clones exhibited a more than doubled T1⁄ 2 of activity compared with WT. The best clones selected from the random mutagenesis approach, as well as the site-direc- ted mutagenesis, are summarized in Table 1 and consti- tute the basis for the first round of FIND TM treatment. It was also noted that all clones displayed fewer mutations than the average number of mutations present in randomly selected clones from both libraries. To explore further combinations of activity stabil- izing mutations identified in the first screening step (Table 1) FIND TM was used. For the first round of FIND TM approach, the following clones from Table 1 were used in two different combinations: in F1.1: EPP1, EPP2, GMK1, GMK2 and, in F1.2: all clones in Table 1 except WT. FIND TM libraries were expressed and 5000 clones of each library screened for improved thermostability. Table 2 summarizes clones derived from this step. As shown in Table 2, six clones with improved T1 ⁄ 2 of their activity compared to the parental clones could be found in the F1.1 treatment, while only two clones were found in the F1.2 treat- ment with improved or equal properties, despite the higher number of clones put into this library. It should also be mentioned here that the FIND TM treatment not only recombined existing mutations, but also intro- duces new mutations as observed in six out of the eight selected clones (Table 2). To ascertain the combination of mutations that are very close in sequential space, the GMK2 clone (Table 1) was modified by site-directed mutagenesis to create the mutants YQ and YP, and the T1⁄ 2 of their activity was determined (Table 2). These combinations increased the stability of CPU’s activity, especially the YQ mutant. Following the first round of FIND TM treatment, 50% of the mutants with improved thermal stability of their activity appeared to bear mutations in the region encompassing residues 327–357. New mutants were made by site-directed mutagenesis, trying to combine the mutations leading to the strongest decrease in thermo-inactivation by site-directed muta- genesis. The stability of their activity was evaluated either after expression in 3T3 cells or in insect cells using the Baculovirus expression vector system (BEVS) (Table 3) as an alternative expression system. The S327P mutation was introduced because P is the corresponding amino acid to S327 in porcine CPB (Fig. 2). A second round of FIND TM treatment (F2) then included the clones: GMK2 + T347I, F1.1.C + R315H, F1.1.F + S327P, F1.1.A and YQ (see Tables 2 Table 1. Half-life (T1 ⁄ 2) of different CPU mutants’ activity at 37 °C created by site-directed or random mutagenesis. WT and mutant CPU were expressed in 3T3 cells and their stability was accessed in the cell culture supernatant. The remaining enzymatic activity after incubation of CPU or its mutants at 37 °C was determined using a HPLC assay. Clone Amino acid changes in CPU T1 ⁄ 2at 37 °C (min) Method of generation EPP1 K166N, H357Q 31 Error prone PCR EPP2 I251T, H357P 31 Error prone PCR EPP3 I180F a , H357Q 55 Error prone PCR GMK1 H315R, S327C 60 Genemorph GMK2 H355Y 47 Genemorph A L376Q 16 Site-directed B T347I 18 Site-directed WT – 12 a This mutation was not present in all PCR products derived from this clone. Table 2. Half-life (T1 ⁄ 2) of different CPU mutants’ activity at 37 °C derived from the first round of FIND TM treatment and site-directed mutagenesis. WT and mutant CPU were expressed in 3T3 cells and their stability was accessed in the cell culture supernatant. The remaining enzymatic activity after incubation of CPU or its mutants at 37 °C was determined using a HPLC assay. New mutations, not present in the parental clones are underlined. Clone Amino acid changes in CPU T1 ⁄ 2at 37 °C(h) F1.1.A I251T, H315R, S327C, N350S, H357Q 2.2 F1.1.B K166N, H315R, S327C, N350S, H357Q 1.5 F1.1.C K166N, H315R, S327C, H357P 4.4 F1.1.D H315R, S327C, R352K 1.6 F1.1.E H315R, S327C, N350S, H357Q 2.4 F1.1.F S327C, S348N, H357Q 2.9 F1.2.A H315R, S327C, H355Y 2.2 F1.2.B V219A, H315R, S327C 1 YP a H355Y, H357P 1.5 YQ a H355Y, H357Q 3 WT – 0.2 a These mutants were generated by site-directed mutagenesis from GMK2. W. Knecht et al. Stable human CPU mutants FEBS Journal 273 (2006) 778–792 ª 2006 The Authors Journal compilation ª 2006 FEBS 781 and 3). Libraries created from these clones by FIND TM technology were expressed, screened and characterized as described in material and methods. A total of about 14 200 clones were screened. Table 4 summarizes clones derived from this second round of FIND TM . The same mutation combination as in the best clone made by site-directed mutagenesis (YQ + S327C) was also generated by this second round of FIND TM treatment and identified by the screening. The subsequently increased activity stabilization during the different steps of directed evolution and screening is illustrated in Fig. 3, displaying the most stable clones found in each step. From the mutants created, seven clones (F1.2.A; F1.1.F, YQ, YQ + S327C, F1.2.A + R315H, F1.1.F + N348S, F1.1.F + H355Y) were chosen for expression using the BEVS and purification of the mutants for analysis as purified protein. WT proCPU and mutants were expressed in Sf9 insect cells as C-terminal His-tagged proteins and purified from the supernatant of a 1-L culture using IMAC. Figure 4 shows as examples the homogenity of the WT proCPU-CHis and YQ proCPU-CHis preparations (0-min samples). The parameters determined for these mutants are summarized in Table 5. In contrast to the screening and previous characterization in crude cell superna- tants, assays were now carried out in a defined buffer of 50 mm Hepes, pH 7.4. The T1 ⁄ 2 of CPU activity at 37 °C increased from 0.2 h for WT CPU to more than 5 h for the two most stable clones (Table 5). It appears that the T1 ⁄ 2 of activity measured directly in the supernatant of the cell cultures deviates from the T1 ⁄ 2 of the purified proteins in a defined buffer system. It is likely that cell culture medium components influence the thermo-inactivation of the mutants. This was con- firmed by putting purified YQ + S327C back into insect cell culture medium, which prolonged the T1 ⁄ 2 of activity at 37 °C (data not shown). A second estimation of the thermal stability of activity of each mutant was measuring activity after Table 3. Half-lives (T1 ⁄ 2) of different CPU mutants’ activities at 37 °C made from clones in Tables 1 and 2 by site-directed muta- genesis. WT and mutant CPU were expressed in 3T3 cells or in insect cells (as indicated) and their stability was accessed in the cell culture supernatant. The remaining enzymatic activity after incuba- tion of CPU or its mutants at 37 °C was determined using a HPLC assay. The T1 ⁄ 2 of the parental clone is shown in brackets for easy comparison. Clone Amino acid changes in CPU T1 ⁄ 2at37°C (h) GMK2 + T347I a T347I, H355Y Not done (0.8) F1.1.C + R315H K166N, S327C, H357P 1.6 (4.4) F1.1.A + R315H I251T, S327C, N350S, H357Q 0.7 (2.2) F1.2.A + R315H b S327C, H355Y 4.3 (2.2) F1.1.F + N348S b S327C, H357Q 2.4 (2.9) F1.1.F + H355Y b S327C, S348N, H355Y, H357Q 4 (2.9) F1.1.F + S327P S327P, S348N, H357Q 0.3 c (2.9) YQ + S348N S348N, H355Y, H357Q 2.4 (3) YQ + T347I T347I, H355Y, H357Q 3.5 (3) YQ + S327P S327P, H355Y, H357Q 1.1 (3) YQ + N350S N350S, H355Y, H357Q 1.4 (3) YQ + S327C b,d S327C, H355Y, H357Q 26 (3) WT – 0.2 a Very low expression level did not allow T1 ⁄ 2 determinations for GMK2 + T347I. b These mutants were expressed in insect cells and have an 8xHis tag as described in Experimental procedures. c Activ- ity was determined using the Hippuricase assay. d The same combi- nation was independently found within the second FIND TM treatment (see Table 4). Fig. 3. Subsequent increase in stability of activity during the directed evolution process of CPU. T1 ⁄ 2 data at 37 °C for the most stable clones as determined in the supernatant of 3T3 cells are presented. More results for the different steps are presented in the correspond- ing tables: Random mutagenesis (Table 1), first FIND TM (Table 2), second FIND TM ⁄ site-directed mutagenesis (Tables 3 and 4). Table 4. Half-life (T1 ⁄ 2) of different CPU mutants’ activity at 37 °C derived from the 2nd round of FIND TM treatment. WT and mutant CPU were expressed in 3T3 cells and their stability was accessed in the cell culture supernatant. The remaining enzymatic activity after incubation of CPU or its mutants at 37 °C was determined using a HPLC assay. Mutations not found in the parental clones are underlined. Clone Amino acid changes in CPU T1 ⁄ 2at 37 °C (h) F2.A I251T, H355Y, H357Q 2 F2.B I204T, Y230C, S348N, H357Q 2.9 F2.C a S327C, H355Y, H357Q 6.8 WT – 0.2 a Identical to YQ + S327C (see Table 3). Stable human CPU mutants W. Knecht et al. 782 FEBS Journal 273 (2006) 778–792 ª 2006 The Authors Journal compilation ª 2006 FEBS incubation at 22 °C for 24 h (Table 5). Mutants YQ + S327C, F1.2.A and F1.1.F + H355Y were again the most stable and only one mutant, F1.1.F + N348S, lost more than 50% of its activity. In order to exclude profound effects of the muta- tions on enzymatic activity and inhibitor binding affin- ity, the K m for hippuryl-l-arginine (Hip-Arg), the specific activity at 24 mm Hip-Arg and the IC 50 for the specific inhibitor PCI were determined. As can be seen in Table 5, the K m values of the mutants shift to lower values, while all mutants except F1.1.F show an increased specific activity. To detect any changes in the positioning of the propeptide, or, in other words, to see if the contact region between the catalytic domain and the prodomain was changed by the mutations, we also measured the residual activity without activation by thrombin–thrombomodulin. A correct positioning of the propeptide should keep the residual activity on Fig. 4. Tryptic digest of WT and YQ proCPU-CHis. (A) SDS PAGE of a bovine trypsin digest of WT proCPU-CHis (1.3 lgÆlane )1 ) and YQ pro- CPU-CHis (2 lgÆlane )1 ). Two proCPU-CHis to bovine trypsin ratios (w ⁄ w) were used: (i) 1 : 20 and (ii) 1 : 100. Digests were run at 26 °Cfor the times indicated and then separated by SDS ⁄ PAGE and the gel was Coomassie stained. Two major degradation products of WT- and YQ-proCPU-CHis became visible and are indicated by arrows in the figure. (B) WT and YQ proCPU-CHis were digested by bovine trypsin as described under (A) (i) for the times indicated. Fifteen micrograms per lane were separated by SDS ⁄ PAGE and transferred to a polyvinylid- ene difluoride membrane for N-terminal sequencing (Amidoblack staining). The bands indicated by numbers were identified as starting at the N-terminus with (i) a mixture of A115 and F23, (ii) a mixture of A115 and F23, (iii) Y353 and (iv) A115. Table 5. Kinetic and stability parameters for purified WT and mutant CPUs. The T1 ⁄ 2 of activity at 37 °C in cell culture medium is shown in brackets for easy comparison. Specific activity was determined at 24 m M Hip-Arg and the IC 50 of PCI at 4 mM Hip-Arg. The specific activity for 24 m M Hip-Arg without activation by thrombin–thrombomodulin is given in brackets. H315 S327 S348 H355 H357 T1 ⁄ 2 at 37 °C(h) Activity left after 24 h at 22 °Cin % (mean ± SD) K m (mM) Specific activity (UÆmg )1 ) IC 50 PTCI (l M) WT 0.2 (0.2) 20 ± 11 2.2 53 (1.9) 0.2 F1.2.A R C Y 5.2 (2.2) 89 ± 8.9 3.7 98 (2.4) 0.04 F1.1.F C N Q 2.2 (2.9) 56 ± 6.3 0.7 41 (1.8) 0.13 YQ Y Q 1.5 (3) 78 ± 7.9 0.9 88 (1.6) 0.16 YQ + S327C C Y Q 5.5 (26; 6.8) 81 ± 7.8 1.1 121 (2.3) 0.16 F1.2.A + R315H C Y 1.3 (4.3) 63 ± 2.2 1.5 150 (2.5) 0.06 F1.1.F + N348S C Q 1 (2.4) 45 ± 3.3 0.6 64 (3.7) 0.12 F1.1.F + H355Y C N Y Q 4.9 (4) 86 ± 6.6 1 89 (4.4) 0.18 W. Knecht et al. Stable human CPU mutants FEBS Journal 273 (2006) 778–792 ª 2006 The Authors Journal compilation ª 2006 FEBS 783 the same level as for WT-proCPU until it is cleaved away and proCPU activated to CPU. The activities ranged from 1.6 to 4.4 UÆmg )1 , with 1.9 UÆmg )1 for WT-proCPU-CHis, or, as a percentage of the specific activity after activation, from 1.7 to 5.8% with 3.6% for WT-proCPU-CHis. At 4 mm Hip-Arg as substrate in the assay, inhibition of all mutants is achieved at somewhat lower concentrations of potato tuber carb- oxypeptidase inhibitor (PTCI). The YQ mutant, cho- sen because it had the most stable activity of the purified mutants having only two mutations, was used for further extensive characterization. To determine if the increased thermal stability of CPU activity is connected to an increased thermosta- bility of the protein itself, we monitored the thermal unfolding of WT and YQ proCPU-CHis. Compared with the WT, the midpoint temperature (T m ) of the protein-unfolding transition has increased for YQ proCPU-CHis by 10.4 °C (Fig. 5a). Because in YQ proCPU-CHis, H355 and H357 are replaced by nonio- nizable amino acids, we monitored thermal unfolding also at different pH values (Fig. 5b). Approaching low pH values, when histidines become fully protonated, a pronounced drop of T m was seen with WT proCPU- CHis, while only a marginal one was recorded with YQ proCPU-CHis. The drop in T m from pH 7.4 to pH 4.5 was 12.8 °C for WT proCPU-CHis but only 2.3 °C for YQ proCPU-CHis. This indicates a role of H355 and ⁄ or H357 in the thermal stability of proCPU. Furthermore, we digested WT proCPU-CHis and YQ proCPU-CHis with bovine trypsin (Fig. 4), which resulted in the case of WT proCPU-CHis in one prominent degradation product of approximately 25 kDa and a weak, probably intermediate band at about 38 kDa (arrows in Fig. 4A), while for YQ proCPU-His, a strong band at 38 kDa became visible but none was visible at about 25 kDa. Subsequently, N-terminal sequencing of these bands identified a clea- vage site between R352 and Y353 in WT proCPU- CHis, but not in the YQ mutant. Consequently, the two mutations of YQ make the mutant less sensitive to tryptic digestion close to the positioning of its two mutations. Next, we compared the affinity of the enzyme for synthetic and physiological substrates, and determined the K m constants of native CPU from plasma, recom- binant WT CPU and YQ CPU for Hip-Arg and bra- dykinin using an arginine kinase-based kinetic assay [27]. Data are presented in Table 6. No differences were seen in the K m values of the three CPUs for bra- dykinin and Hip-Arg when the kinetic assay was used, proving that the mutations in the YQ proCPU did not alter the affinity of the carboxypeptidase for synthetic and physiological substrates. However, when the K m for Hip-Arg was measured using HPLC (Table 5), YQ shows K m value similar to the kinetic assay, while WT CPU does not. Fig. 5. Thermal unfolding of WT proCPU-CHis and YQ proCPU- CHis. The thermal unfolding of WT and YQ proCPU-CHis was mon- itored using the fluorescent dye Sypro orange. The unfolding pro- cess results in increase in fluorescence, which was monitored. (A) shows the means of three independent unfolding curves in 50 m M Hepes pH 7.4 and the solid line present the best fit of equation 1 to all data. (B) shows the T m of thermal unfolding curves at differ- ent pH values (best fit of equation 1 to all data ± SEM of the fit). d, YQ proCPU-CHis; s, WT proCPU-CHis. Buffers used were 50 m M sodium acetate, pH 4.5, 50 mM Mes pH 5.6–6.5, 50 mM Hepes, pH 7.4. Table 6. Comparison of K m constants of native, WT and YQ CPU for Hip-Arg and bradykinin using an continuous enzyme assay. K m values are expressed in lMÆL )1 and are the mean ± SEM of a duplicate measurement. Native CPU WT YQ Bradykinin 39 ± 2 44 ± 6 35 ± 5 Hip-Arg 840 ± 21 825 ± 44 774 ± 39 Stable human CPU mutants W. Knecht et al. 784 FEBS Journal 273 (2006) 778–792 ª 2006 The Authors Journal compilation ª 2006 FEBS The hypothesis that CPU down-regulates fibrinolyis by a threshold dependent mechanism was recently pub- lished [18,19]. As long as the CPU activity remains above this threshold (reported to be 8 UÆL )1 ), fibrinoly- sis does not accelerate but stays in its initial phase [19]. The study of this threshold phenomenon is severely complicated by the intrinsic instability of CPU’s activ- ity. YQ proCPU-CHis was consequently tested for its antifibrinolytic potential in an in vitro clot lysis assay and used for confirmation of the threshold hypothesis. We reconstituted proCPU-depleted plasma with increasing amounts of the activated stable YQ mutant or with WT CPU (CPU activities ranging from 0 to 237 UÆL )1 ) and used these in clot lysis experiments, as described previously [19,28]. Recovery of the added CPU was in the range of 96–103%, as measured with a kinetic plasma assay [27]. The final t-PA concentration used was 40 ngÆmL )1 . The stable YQ mutant was able to prolong the in vitro clot lysis time (CLT) in a way that can be theoretically expected based on its stability. The decay of CPU can be expressed using the fol- lowing simplified equation N ¼ N 0 · e –k ⁄ t where k ¼ ln(2) ⁄ T, T ¼ half life of CPU. Rearrangement of this formula gives the equation: t ¼ [T log(2) )1 ] · [log(No ⁄ N)], where t is the time above the threshold, N 0 the initial CPU activity and N the threshold activity value. This equation indicates that the time above the threshold is linearly related with the CPU half life and only logarithmically with the initial CPU activity (gen- erated from proCPU by first order kinetics). The hypo- thesis that this time above the threshold determines the CLT is strongly confirmed and illustrated in Figs 6 and 7. Figure 6 shows representative clot lysis profiles at different YQ CPU concentrations. Increasing the enzyme activity below the ‘threshold value’ did not show a significant increase in CLT. However, each doubling of the CPU activity in excess of the ‘thresh- old value’ increased CLT with one CPU mutant half life. Plotting log (CPU activity added) versus CLT clearly confirms the CPU threshold hypothesis. The estimated threshold value in our experiments was 12 UÆL )1 which corresponds very well with the 8UÆL )1 described by Leurs et al. [19]. Figure 7 illustrates the linear relationship between CPU stability and CLT. Adding 40 UÆL )1 WT CPU to proCPU-depleted plasma increases CLT by 22 min. However, the addition of 40 UÆL )1 YQ CPU (with a 7.5- fold increased stability) increases CLT by 153 min, which corresponds very well with the increase one theoretically can expect (i.e. 7.5 · 22 min). When the selective CPU inhibitor PTCI (20 lgÆmL )1 ) was added from the start, no significant prolongation of CLT was seen by adding YQ or WT CPU. 1 YQ t 1/2 Fig. 6. Threshold hypothesis confirmation. The graph shows representative clot lysis profiles of proCPU depleted plasma reconstituted with increasing concentrations of activated YQ mutant (concentrations ranging from 0 UÆL )1 to 237 UÆL )1 ). The threshold value is estimated by plotting log of the CPU activity added versus the clot lysis time (inset). Each doubling of the enzyme activity above the threshold value increases clot lysis time with one CPU mutant half-life. W. Knecht et al. Stable human CPU mutants FEBS Journal 273 (2006) 778–792 ª 2006 The Authors Journal compilation ª 2006 FEBS 785 Discussion Due to its physiological role and the need for a very tight regulation in the blood coagulation cascade, it is likely that CPU has been selected for intrinsic instabil- ity, which ensures rapid inactivation of its activity at the site of action. The irreversible decay in activity has been shown to be accompanied by structural changes of CPU [12,13], it is therefore very likely that the loss of activity is caused by structural changes of the enzyme triggered upon activation. This instability is a serious challenge when dealing with overexpression and purification of the protein. The mechanism behind CPU’s activity inactivation is still not fully understood, but several aspects contributing to CPU’s instability are illuminated by our work. CPB is a close homologue to CPU, but has a signifi- cantly higher stability. Aligning the CPU sequence onto the CPB structure [26] reveals the presence of numerous potentially exposed hydrophobic amino acids in CPU. Exposed hydrophobic residues lead to aggregation, and replacing exposed hydrophobic resi- dues with more polar residues has been reported to stabilize proteins [29,30]. Of the 12 hydrophobic to hydrophilic point mutations carried out in CPU, only one, L376Q (clone A), had a stabilizing effect, in this case, of about 33%. All the other mutants either did not change the T1 ⁄ 2 of CPU’s activity more than ± 20%, or, in the case of I147S, did not express at all (data not shown), suggesting that the instability does not result from hydrophobically driven aggregation of the protein. This is further confirmed by the existence of a natural variant of CPU, where T347 is subsituted by an I. Although accentuating the hydrophobic character of the protein surface, the mutation induces a stabilization of the protein (Table 1 and [11]). Random evolution of the enzyme has allowed us to identify mutants of 2.5 to five-fold increased T1 ⁄ 2in activity (Table 1), with one or two mutations per clone. The following first round of FIND TM treatment pro- longed T1⁄ 2 from 12 min for the WT to 4.4 h for clone F1.1.C. Further combination by rational site-directed deletion or addition of mutations (Table 3) resulted in more than half of the cases in a decrease of T1 ⁄ 2. A fur- ther round of FIND TM treatment did not improve T1 ⁄ 2 further compared with a combination of mutations pre- viously found by site-directed mutagenesis, but inde- pendently produced the same combination of mutations that were also determined to display the most stable activity (YQ + S327C ¼ F2.C). An overall view of the evolution process is presented in Fig. 3. A number of mutants with modifications in this region of the polypeptide chain were expressed in insect cells, purified and characterized (Table 5). The mutant displaying the most stable activity at 37 °C had mutations at the positions S327, H355 and H357, and this is also reflected by the selection of proteins to be purified, that all have at least two mutations at these positions. The T1 ⁄ 2 of activity of the purified mutants determined in a defined buffer system, as used during purification procedures, differed significantly from T1 ⁄ 2 determined in mammalian or insect cell cul- ture supernatant. From a practical point of view, to allow for high-throughput mutant screening, thermo- stability had to be measured in cell culture superna- tants. The corresponding values obtained from purified proteins show that cell medium itself and ⁄ or unknown substances secreted by the cells sometimes strongly Fig. 7. CPU stability versus proCPU concentration in influencing clot lysis time. The graph shows the effect of adding increasing activities of WT CPU and YQ CPU on the clot lysis time, clearly showing the importance of the CPU stability over proCPU concentration. Stable human CPU mutants W. Knecht et al. 786 FEBS Journal 273 (2006) 778–792 ª 2006 The Authors Journal compilation ª 2006 FEBS prolonged or decreased the stability of the activity of the CPU mutants (Table 5). This is most striking for YQ + S327C, with T1⁄ 2 of 5.5 h in Hepes buffer, 6.8 h in mammalian cell culture medium and 26 h in insect cell culture medium. For mutants containing the S327C mutation, conditions, such as pH, determining how fast oxidation of the cystein might occur, may play a role. Because most of the purification and in vitro assays are carried out at room temperature, we also determined the activity after 24-h incubation at 22 °C. Of all puri- fied mutants, three showed a remaining activity of more than 80% after 24-h incubation at 22 °C versus 20% for the WT, and two of these three showed a more than 25- fold increase in T1 ⁄ 2 of activity at 37 °C. The decreased K m and mostly increased specific activity may partially reflect the improved stability of activity, especially at low substrate concentrations during K m determinations, resulting in a higher velocity than for WT CPU and thereby decreasing the observed K m in comparison to WT CPU. This hypothesis is supported by the use of a newly developed continuous coupled enzyme assay instead of the discontinuous HPLC assay that demon- strated similar K m values of the native and WT, and the YQ mutant CPU with a synthetic and physiological substrate of CPU. There seem to be no major changes in the positioning of the propeptide, as indicated by residual activities of the mutants close to WT-proCPU. IC 50 values for the inhibition by a specific inhibitor PTCI [16] are maximally five-fold lower than for the WT. With the exception of stability of activity, the CPU mutants appear surprisingly similar to the WT in their enzymatic properties. Marx et al. [31] also described the generation of forms of CPU with a highly stable activity, but in con- trast to the work presented here, this refers to a hybrid of CPU ending at position 314 (Fig. 2) fused to the following C-terminal part of human CPB. This chi- mera had a half-life of 1.5 h at 37 °C. We therefore show here that a stabilization of CPU’s activity that is more than that which naturally occurs can already be achieved with only one or a few mutations in the region following position 314 in CPU. Fifty per cent of the residues mutated in the clones selected from the first round of FIND TM are located in a distinct region encompassing residues 327–357 (Table 2 and Fig. 2), as well as the naturally occurring and activity stabilizing mutation T347I. The mutants with the most stable activity are achieved by combina- tions of few conservative mutations, S327C, H355Y and H357Q. Can the effects of the mutations reported here and the reasons for the increased stability of activity if connected to structural changes be rationally explained? The three residues correspond to P300, Y327 and P329 in porcine CPB (numbering according to Fig. 2). Keeping a strict orientation of the side chains, replacing P300 with a serine would leave the H-bond to the OH group of the side-chain nonsatis- fied, thereby destabilizing the protein. Based on the CPB structure, H355 lies in close proximity to a cluster of charged residues: R324, K326, H330 and E360. Introducing a Q at position 355 is likely to favour the formation of H-bonds with one or several of these resi- dues, attenuating the charge repulsions between some of the basic amino acids. The stabilization induced by the replacement of H357 by a Y is more difficult to explain, but the aromatic nature of the side chain is likely to interact favourably with the hydrophobic clus- ter made up of I316, F318, A337 and V341. Another contribution to the low stability of the WT proCPU is the close spatial proximity of the three His residues at 330, 355 and 357. In the YQ mutant, two histidines are replaced by nonionizable amino acids. Although not very pronounced at physiological pH, partial charges on the His could induce a destabilizing charge–charge repulsion effect. This hypothesis is sup- ported by the findings that WT proCPU-CHis is less stable in thermal unfolding at low pH, when H330, H355 and H357 would be protonated, while the drop of stability of YQ proCPU-CHis is a lot less pro- nounced (Fig. 5b). These observations suggest that our mutations improve residue interactions in this region, leading to an improved structural stability of the protein. Limited trypsinolysis of WT and YQ proCPU-CHis further corroborate this scenario, as trypsin cleavage occurs at R352 in WT CPU, but not in the mutant harbouring the H355Y ⁄ H357Q mutations (Fig. 4). Recently, the hypothesis was put forward that CPU can down-regulate fibrinolysis through a threshold- dependent mechanism [19]. We used the stable YQ CPU mutant to test this hypothesis. The antifi- brinolytic potential of the stable mutant was tested in an in vitro clot lysis assay. The YQ mutant was able to prolong in vitro clot lysis time in a way that can be expected based on the stability of its activity. Thus YQ is the first described stable CPU form with conserved antifibrinolytic potential. This threshold hypothesis [19] could be confirmed by adding activated YQ pro- CPU-CHis to proCPU depleted plasma and plotting CLT versus the log of the CPU activity added. The threshold value in our experiments was 12 UÆL )1 , which is in good agreement to the value reported by Leurs et al. [19] of 8 UÆL )1 . As long as CPU remains above this activity value, fibrinolysis does not proceed into the acceleration phase. The threshold hypothesis W. Knecht et al. Stable human CPU mutants FEBS Journal 273 (2006) 778–792 ª 2006 The Authors Journal compilation ª 2006 FEBS 787 [...]... according to the manufacturer’s instructions Reverse transcription-PCR using preproCPU-specific primers (CPU_ fwd_XhoI and CPU_ rev_NotI) were performed with the Titan RT-PCR kit (Roche, Basel, Switzerland) according to the manufacturer’s instructions The PCR products were subcloned into pGEM-T for sequencing Expression of WT proCPU in insect cells To express WT proCPU, the ORF of preproCPU was amplified in. .. sigma plot 8 F(T) is the fluorescence intensity at temperature T N-Terminal amino acid sequencing Characterization of purified WT and mutant proCPUs SDS ⁄ PAGE was carried out using 4–12% Bis-Tris Gels (NuPAGETM, Invitrogen) according to the manufacturer’s instructions The concentration of proCPU (mutants) in cell culture supernatants or purified samples was determined using a proCPU ELISA as described... System (Invitrogen), according to the manufacturer’s instructions Expression of mutant proCPUs in insect cells Determination of the half-life (T1/2) of CPU s activity The T1 ⁄ 2 of the activity of the best mutants secreted in the supernatant of the cells in both screens was then determined as follows: Activated (mutant-) CPU (activation as described above) was incubated at a constant temperature and samples... differences between mutant and WT CPU, except for a prolongation of clot lysis time proportional to the increase in T1 ⁄ 2 of activity of the mutant The YQ mutant was also used to demonstrate the importance of CPU stability over proCPU concentration in down-regulating fibrinolysis It is therefore very likely that the mutants presented here constitute a relevant model system for structural studies of the enzyme... stability of activity of purified WT and mutant CPUs were determined with the same protocol, but under defined buffer conditions of 50 mm Hepes, pH 7.4 Determination of the ORF of mutated preproCPU stably expressed in 3T3 cells After selection of more stable mutant proCPUs (see above), RNA was purified from selected stable 3T3 cell lines using The ORF of selected mutant preproCPUs were amplified by PCR using the. .. procedures Cloning of human preproCPU cDNA The cloning of human preproCPU, e.g pAM245, has been described by Stromqvist et al [32] ¨ Directed nucleotide substitutions were introduced into the preproCPU cDNA with the Quikchange XL site-directed mutagenesis kit (Stratagene, La Jolla, CA, USA) according to the manufacturer’s instructions Error-prone PCR was performed according to Cadwell and Joyce [33,34] The. .. by increased CPU stability (e.g related to the 347 Thr ⁄ Ile polymorphism), the study of the naturally occurring functional polymorphism at position 347 should be included in clinical settings evaluating proCPU as a thrombotic risk factor In summary, of seven selected and purified mutants, three showed a remaining activity of more than 80% after 24 h incubation at 22 °C versus 20% for the WT; two of these... Random recombination of mutated preproCPU cDNAs was performed using in vitro molecular evolution of protein function procedure (now known as Fragment-INduced Diversity (FINDTM) technology) according to the methods disclosed in UK Patent Publication No GB 2370 038 A (UK Patent Of ce, London, UK) Generation of stable mouse cell lines expressing proCPU and mutant proCPUs A retroviral gene delivery and expression... described Further characterization of a selected mutant (YQ) Km constants of YQ CPU- CHis for Hip-Arg and bradykinin were also determined using a coupled enzyme assay for CPU activity [27] and compared with WT CPU- CHis and native CPU (purified according to the protocol described by Schatteman et al [6]) Thermal unfolding of WT and YQ proCPU- CHis was monitored using the fluorescent dye Sypro orange (Molecular... pFASTBac1 (Invitrogen) The primers C-HIS1rev and C-HIS2rev introduced the coding sequence for an octa-His tag at the C-terminus of proCPU (amino acid sequence of the tag: LEPGDDDDKHHHHHHHHSGS) The resulting plasmid was named pAM1079 Recombinant Baculovirus for expression of recombinant proCPU with C-terminal octaHis tag (proCPU- CHis) was generated starting from pAM1079 with the Bac-to-BacÒ Baculovirus Expression . Limited mutagenesis increases the stability of human carboxypeptidase U (TAFIa) and demonstrates the importance of CPU stability over proCPU concentration in. to the manufacturer’s instructions. The concentration of proCPU (mutants) in cell culture supernatants or purified samples was determined using a proCPU