Enhancement through Mutagenesis of the Binding of the Isolated Kringle 2 Domain of Human Plasminogen to ω-Amino Acid Ligands and to an Internal Sequence of a Streptococcal Surface Protein*

In the background of the recombinant K2 module of human plasminogen (K2Pg), a triple mutant, K2Pg[C4G/E56D/L72Y], was generated and expressed inPichia pastoris cells in yields exceeding 100 mg/liter. The binding affinities of a series of lysine analogs, viz.4-aminobutyric acid, 5-aminopentanoic acid, ε-aminocaproic acid, 7-aminoheptanoic acid, andt-4-aminomethylcyclohexane-1-carboxylic acid, to this mutant were measured and showed up to a 15-fold tighter interaction, as compared with wild-type K2Pg (K2Pg[C4G]). The variant, K2Pg[C4G/E56D], afforded up to a 4-fold increase in the binding affinity to these same ligands, whereas the K2Pg[C4G/L72Y] mutant decreased the same affinities up to 5-fold, as compared with K2Pg[C4G]. The thermal stability of K2Pg[C4G/E56D/L72Y] was increased by approximately 13 °C, as compared with K2Pg[C4G]. The functional consequence of up-regulating the lysine binding property of K2Pg was explored, as reflected by its ability to interact with an internal sequence of a plasminogen-binding protein (PAM) on the surface of group A streptococci. A 30-mer peptide of PAM, containing its K2Pg-specific binding region, was synthesized, and its binding to each mutant of K2Pg was assessed. Only a slight enhancement in peptide binding was observed for K2Pg[C4G/E56D], compared with K2Pg[C4G] (K d = 460 nm). A 5-fold decrease in binding affinity was observed for K2Pg[C4G/L72Y] (K d = 2200 nm). However, a 12-fold enhancement in binding to this peptide was observed for K2Pg[C4G/E56D/L72Y] (K d = 37 nm). Results of these PAM peptide binding studies parallel results of ω-amino acid binding to these K2Pg mutants, indicating that the high affinity PAM binding by plasminogen, mediated exclusively through K2Pg, occurs through its lysine-binding site. This conclusion is supported by the 100-fold decrease in PAM peptide binding to K2Pg[C4G/E56D/L72Y] in the presence of 50 mm 6-aminohexanoic acid. Finally, a thermodynamic analysis of PAM peptide binding to each of these mutants reveals that the positions Asp56 and Tyr72 in the K2Pg[C4G/E56D/L72Y] mutant are synergistically coupled in terms of their contribution to the enhancement of PAM peptide binding.

binding site in a kringle module that does not contain such a site or, more modestly, to up-regulate this site in a kringle motif that possesses a weaker site. This latter feature characterizes K2 Pg , and it was our aim to redesign this kringle unit to contain a stronger -amino acid-binding site. The results of this investigation are reported herein.

EXPERIMENTAL PROCEDURES
Construction of pPIC9k[K2 Pg ]-A construct containing K1 Pg linked through a factor Xa-sensitive cleavage site (IEGR) to K2 Pg was inserted into the yeast expression bacterial shuttle vector pPIC9k (Invitrogen, San Diego, CA) (pPIC9k[K1 Pg XaK2 Pg ]) and used as the template DNA for PCR-based mutagenesis. The plasmid was provided by Hui Wang of this laboratory. The K2 Pg was amplified by PCR for preparation of the mutants used in the work described herein.
For PCR-based amplification of K2 Pg , the forward primer (1), 5Ј-CGATGAATTCTCCGAGGAATGTATGCATTGC, consisted of the 5Ј end of the K2 Pg gene, engineered with a 5Ј EcoRI site (in bold).
The reverse primer (P2), 5Ј-ATTCGCGGCCGCTTACTATGCTGCG-CAGCGGGGG, contained the 3Ј end of the K2 Pg gene, engineered with a 3Ј NotI site. This primer also coded for two Ala residues (mutated bases underlined) immediately following the final residue, Cys 78 , of K2 Pg . Two Thr residues are found at these positions in the natural hPg molecule, but they have been replaced by Ala residues to avoid the possibility of O-linked glycosylation at these sites during the subsequent expression in yeast. This primer also encodes two stop codons (in italics) immediately following the two Ala residues and are immediately upstream of the NotI site.
By using primers P1 and P2, a 276-bp fragment (F1) was amplified, containing the wild-type K2 Pg gene flanked by an upstream EcoRI site and a downstream NotI site, which were used to ligate F1 into plasmid pPIC9k, generating the pPIC9k[K2 Pg ] construct.
The ultimate expression of the isolated wtr-K2 Pg gene product yielded little intact properly folded target material. This is most likely due to the existence of Cys 4 , which is normally disulfide-linked to the Cys 42 of K3 Pg in hPg. Therefore, the Cys 4 of K2 Pg was mutated to Gly, which is the amino acid found in all ligand-binding kringles of hPg at this position. This is considered to be the wild-type K2 Pg .
Construction of pPIC9k[K2 Pg /C4G]-This plasmid was generated with a two-round PCR strategy, using the pPIC9k[K2 Pg ] as the template.
For round 1, the forward primer (P3) was 5Ј-ACTACTATTGCCAG-CATTGC, which is positioned approximately 86 bases upstream of the 5Ј terminus of the K2 Pg gene, at the ␣-factor signal sequence of the pPIC9k vector.
The reverse primer (P4) employed was 5Ј-CTCCACTGCCATG-CATAC, which contains the sequence within the K2 Pg gene coding for the C4G mutation (the codon for G is in bold, with the mutated base underlined).
The forward primer (P5), 5Ј-GTATGCATGGCAGTGGAG, is the complement of P4 and also contains the sequence within the K2 Pg gene which codes for the C4G mutation (the codon for Gly is in bold, with the mutated base underlined).
The reverse primer (P6), 5Ј-GGCAAATGGCATTCTGACATCCTC, is positioned approximately 130 bases downstream from the 3Ј end of the K2 Pg gene, at the 3Ј AOX1 gene sequence region of the pPIC9k vector.
Because fragments F2 and F3 are complementary around the region of the C4G mutation in K2 Pg , they were employed as overlapping templates in round 2 of a PCR reaction, using primers P3 and P6 to amplify a 457-bp fragment (F4). This latter fragment is composed of the following genes: ␣-factor signal sequence Reverse primer P9, 5-GATGTCGCAATATTCCCAGCGCTT, and forward primer P10, AAGCGCTGGGAATATTGCGACATC, are complementary primers at the region of the Leu 72 of the K2 Pg gene, and both contain the L72Y mutation (in bold).
Forward primer P3 and reverse primer P9 were used to amplify a 312-bp fragment (F8) containing the genes encoding for the aminoterminal region of K2 Pg , including the C4G and L72Y mutations.
Forward primer P10 and reverse primer P6 were used to amplify a 169-bp fragment F9 containing the gene encoding for the carboxylterminal region of K2 Pg , including the L72Y mutation.
Because fragments F8 and F9 are complementary around the region of the L72Y mutation, they were used as overlapping templates in a subsequent PCR reaction, using primers P3 and P6 to amplify a 457-bp fragment (F10). Fragment F10 is comprised of the genes for the following: ␣-factor signal sequence EcoRI site-SEE[K2 Pg /C4G, L72Y]AA-NotI site-3ЈAOX1, which was ligated into pPIC9k through the upstream EcoRI and downstream NotI sites as above.
The final fragment (F11) was a 457-bp product comprised of the genes for the following: ␣-factor signal sequence EcoRI site-SEE[K2 Pg / C4G, E56D, L72Y]AA-NotI site-3ЈAOX1, which was ligated into pPIC9k as above.
Expression and Purification of the K2 Pg Variants-For each of these variants, 35 transformed clones were screened for expression on a small scale. The best expressing clones from plasmids linearized with either BglII or SalI, and transformed into the GS115 strain of Pichia pastoris, were fermented with a 20-h methanol induction. The media were then collected and removed by centrifugation. After adjusting the supernatant to pH 7.8, the precipitated salts were removed by centrifugation. The resulting supernatant was directly percolated over a lysine-Sepharose column that was equilibrated in 50 mM Tris-HCl, 50 mM NaCl, pH 7.8.
In the case of the K2 Pg [C4G] mutant, the protein bound weakly to the column and was eluted with extensive application of the wash buffer, viz. 50 mM Tris-HCl, 50 mM NaCl, pH 7.8. Elution with the competing ligand, EACA, was not necessary. The retarded protein, however, bound to the column to a sufficient degree to be separated from any contaminating yeast proteins produced in the fermentation. No further purification was necessary. The yield was Ͼ100 mg/liter after purification.
For K2 Pg [C4G/E56D], purification on lysine-Sepharose column was attained by chromatography with lower ionic strength buffer than above, viz. 10 mM Tris-HCl, 10 mM NaCl, pH 7.8. Under these conditions, this peptide was retained more tightly than was the case with K2 Pg /C4G but could still be eluted from the column in a purified form with extensive washing. However, a large fraction was eluted with 0.15 M EACA. The final yield was approximately 100 mg/liter.
In the case of K2 Pg [C4G/L72Y], the lysine-Sepharose step afforded no purification since this protein was not retarded during this process. Therefore, the media were dialyzed extensively against a solution of 25 mM NaOAc, pH 4.5, and percolated over a Mono S column pre-equilibrated in the same buffer, using fast protein liquid chromatography. A 0 -0.5 M linear gradient of NaCl in 25 mM NaOAc, pH 4.5, was applied to the column at 1 ml/min over 60 min. The peptide was eluted in a homogeneous state at 0.5 M NaCl. The purified yield was Ͼ100 mg/liter.
The variant, K2 Pg [C4G/E56D/L72Y], did adsorb to the lysine-Sepharose column in a buffer of 50 mM Tris-HCl, 50 mM NaCl, pH 7.8. After extensive washing with this same buffer, the peptide was batch-eluted with 0.15 M EACA. A homogenous product was obtained in a yield of Ͼ160 mg/liter.

Synthesis and Characterization of the Streptococcal Surface Protein
(PAM)-based Peptide-The 30-mer PAM peptide, VEK30, that interacts with the K2 domain of hPg (17), viz. Ac-[VEKLT ADAEL QRLKN ERHEE AELER LKSEY]-CONH 2 , was synthesized on a 0.1 mmol scale by automated solid state methodology on a PAL resin support (Perspective Biosystems, Framingham, MA), using an Applied Biosystems (Foster City, CA) model 433A Peptide Synthesizer. Each residue was coupled using the manufacturer's protocols for FastMoc (N-(9fluorenyl)methoxycarbonyl) chemistry. Deprotection of blocking groups and cleavage from the resin was effected using trifluoroacetic acid/H 2 O/ dithiothreitol/triisopropylsilane (88/5/5/2, v/v/w) for 2 h at room temperature. The resin was separated from the peptide deprotection mixture by filtration, and the trifluoroacetic acid was removed in vacuo by rotary evaporation. Cold diethyl ether was added to the filtrate to precipitate the peptide. Following filtration and several ether washes, the crude peptide products were dissolved in water and lyophilized. Peptide purification was accomplished by ion exchange chromatography on a Bio-Rad DEAE-20 column equilibrated in 10 mM Tris-HCl, pH 7.4. A 500-ml linear gradient from 10 mM Tris-HCl, pH 7.4 (start solvent), to 10 mM Tris-HCl, 350 mM NaCl, pH 7.4 (limit solvent), was applied. Fractions corresponding to the major peak were pooled and lyophilized. Desalting was carried out on a 1.5 ϫ 100-cm bed of Sephadex G-15 (Amersham Pharmacia Biotech) equilibrated in 0.1% NH 4 OH. Pooled peptide fractions were lyophilized to constant weight and dissolved in water at final concentrations of 15-20 mM. These stock solutions were retained at Ϫ20°C until needed.
The purity of the peptide was assessed by reverse-phase high pressure liquid chromatography using a 5 mm Vydac 218TP column (4.6 ϫ 250 mm) equilibrated in a solution containing 95% of 0.1% trifluoroacetic acid in H 2 O, 5% of 0.1% trifluoroacetic acid in CH 3 CN at a flow rate of 1.0 ml/min. At 3 min after injection onto the column, a 60-min linear gradient was implemented to a limiting value of 60% of 0.1% trifluoroacetic acid in H 2 O, 40% of 0.1% trifluoroacetic acid in CH 3 CN. Absorbance detection was performed at 215 nm. Further characterization of the peptide was accomplished by DE-MALDI-TOF mass spectrometry on a Voyager-DE spectrometer (Perspective Biosystems) as described previously (40).
Calorimetry-Differential scanning calorimetry experiments were accomplished with a Nano Differential Scanning Calorimeter (Calorimetric Sciences Corp., Provo, UT). The peptides were dialyzed against a buffer containing 100 mM sodium phosphate, 100 mM NaCl, pH 7.4, or 100 mM sodium phosphate, 50 mM NaCl, 50 mM EACA, pH 7.4. The peptide concentrations were approximately 1 mg/ml. Temperature scans were recorded between 20 and 95°C at a scan rate of 1°C/min. The temperatures of maximum heat capacity (T m ) of the deconvoluted peaks were determined from the software that accompanied the calorimeter.
Binding isotherms of -amino acids and the PAM-based peptide (VEK30) to K2 Pg were obtained by isothermal titration calorimetry (ITC) with an OMEGA titration calorimeter (Microcal, Inc., Northhampton, MA) at 25°C in a buffer of 100 mM sodium phosphate, pH 7.4. Peptide samples ranging in concentrations from 30 to 90 M in a total volume of 1.4 ml were placed in the reaction cell. After equilibration, an appropriate concentration of -amino acids or peptide VEK30 in matching buffer was delivered at discrete intervals in 5-l aliquots. The observed heat was measured after each injection. The total observed heat effects were corrected for the heats of dilution of ligand by performing control titrations in the absence of peptide. The resulting titration curves were deconvoluted for the best-fit model using the ORIGIN for ITC software package supplied by Microcal and yielded values for stoichiometry of binding (n), K d , and ⌬H.
Intrinsic Fluorescence Titrations-The binding of ligands to K2 Pg and its variants was determined by titration of the change in intrinsic fluorescence in the kringle that occurred upon ligand binding. The experiments were carried out at 25°C in a buffer containing 50 mM Tris-OAc, 150 mM NaOAc, pH 8.0. The K d values that characterize the ligand-kringle interaction were calculated from the fluorescence titrations by nonlinear least squares iterative curve fitting of the titration data (20).
Analytical Methods-Molecular weight analysis of the K2 tPA samples was measured by time-of-flight matrix-assisted laser-desorption-ionization with delayed-extraction mass spectrometry (TOF-MALDI-DE-MS) on a Voyager-DE spectrometer (Perspective Biosystems, Framingham, MA) under conditions reported previously (41). Amino-terminal amino acid sequences of the mutant proteins were conducted as described (42). All methods employed for manipulations of the DNA samples have been published earlier (20,34,36). Extinction coefficients of the kringles were calculated by the General Protein Mass Analysis for Windows software, based on the amino acid and disulfide bond composition.

RESULTS
To examine the importance of specific amino acid residues in the lysine-binding site of the K2 Pg , a series of mutations, viz. C4G, C4G/E56D, C4G/L72Y, and C4G/E56D/L72Y, were incorporated into this isolated module that would be predicted to up-regulate its LBS. To obtain sufficient amounts of the variant recombinant kringles, a P. pastoris expression system was employed that has been successfully used by this laboratory to obtain large amounts of other kringles from hPg, as well as from tPA (23,26,41,(43)(44)(45). Large scale fermentation of the recombinant yeast GS115 cells produced these peptides at levels Ͼ100 mg/liter. Purification was accomplished by a combination of affinity chromatography on Sepharose-lysine and fast protein liquid chromatography and yielded very high purity (Ͼ98%) materials in all cases. Analyses of the peptides by TOF-MALDI-DE-MS indicated that several molecular weight forms were present for each of the K2 Pg mutants. However, each of these forms is readily identifiable as a variant of K2 Pg that lacks one or more amino acids at its amino-terminal stretch of non-kringle residues. More specifically, the K2 Pg parent form, from which all mutants were generated, was constructed to express the following amino acids at its amino terminus: NH 2 -YVEFSEE immediately preceding Cys 1 of K2 Pg (the YVEF originates from the polylinker site). Minor heterogeneity in cleavage by yeast enzymes that liberated the ␣-factor signal sequence fully account for the different molecular weight species listed in Table I. Amino-terminal sequence analysis of the mutants was consistent with this conclusion since this same heterogeneity was observed. However, since all of these cleavages were upstream of the first Cys residue of K2 Pg , they would have no effect on subsequent studies. Thus, no further attempts were made to resolve these forms.
The effect of the mutations on the thermal stability of K2 Pg was assessed by differential scanning calorimetric analysis. A typical example of a thermogram is shown in Fig. 1 for K2 Pg [C4G/E56D/L72Y]. The temperature of maximum constant pressure (Cp) (T m ) for the single endothermic transition is 70.9°C, and the calorimetric ⌬H is 80.2 kcal/K mol. The T m value differs significantly from that of the wild-type K2 Pg (K2 Pg [C4G]), which possesses a T m value approximately 10°C lower. Thus, the particular mutations inserted at Glu 56 and Leu 72 have substantially increased the thermal stability of this mutant. Furthermore, the ⌬(⌬H) of denaturation between these two peptides is seen to be approximately 12 kcal/K mol, a change consistent with the possibility that additional interactions, possibly involving hydrogen bonds, in K2 Pg [C4G/E56D/ L72Y] need to be overcome for thermal denaturation to occur. As also noted from the data of Table II, the single change at Glu 56 (E56D) did not significantly influence these thermody- namic properties of K2 Pg [C4G], but the mutation at Leu 72 (L72Y) appears to be the major basis for the increased thermal stability of K2 Pg [C4G/E56D/L72Y]. The addition of high levels of EACA increases the T m values of each of these peptides by approximately 7-13°C ( Fig. 1 and Table II), suggesting that EACA interacts with all of these mutant K2 Pg domains. Of particular relevance is the fact that the T m of wild-type K2 Pg can be increased by more than 20°C through the two additional mutations illustrated and by the addition of EACA. It is well known that the binding of -amino acids to kringle domains induces a large change in their intrinsic fluorescence, and this property has been employed extensively to determine the steady state thermodynamic properties of these interactions (e.g. Ref. 36). In the case of K2 Pg , for all of the ligands tested, maximal fluorescence changes of 7%, at the lowest, to 75%, at the highest, were observed. All fluorescence changes were saturable and were fit to simple binding isotherms for a single binding site. K d values for a series of lysine analogues, viz. 4-aminobutyric acid (4-ABA), 5-aminopentanoic acid (5-APA), 6-aminohexanoic acid (EACA), 7-aminoheptanoic acid (7-AHA), and trans-4-aminomethyl-1-cyclohexanoic acid (t-AM-CHA), were determined by this method. An example titration of K2 Pg [C4G/E56D/L72Y] with EACA is illustrated in Fig. 2, which yielded a K d value of 60 M. All K d values obtained by this method are listed in Table III. To obtain an independent measure of these binding constants, K d values of several of these ligands with K2 Pg [C4G/E56D/L72Y] have been measured by ITC titrations. An example of one of these titrations, that of K2 Pg [C4G/ E56D/L72Y] with t-AMCHA, is shown in Fig. 3. The K d value obtained was 11 M, which is in good agreement with the same value of 7 M obtained by fluorescence titrations (Table III).
Finally, the relationship between the LBS of K2 Pg and the binding of this kringle domain to bacterial cell surfaces has been examined. For these studies, a 30-mer peptide (VEK30) was synthesized, which consists of 6 amino acids upstream of a1 region of PAM, the 13 amino acid a1 region, and 10 amino acids from the a2 region of this protein, followed by a Tyr. This peptide has been shown to constitute the K2 Pg -binding site on the group A Streptococcus pyogenes surface protein, PAM (17). Thermodynamic properties describing the binding of this peptide to K2 Pg variants have been determined by ITC titrations, and an illustration of the data obtained for binding of VEK30 to K2 Pg [C4G/E56D/L72Y] is provided in Fig. 4

DISCUSSION
Kringle domains of proteins are mainly involved in interactions with other soluble or insoluble proteins or with proteins on cell surfaces. Many of these interactions are mediated by the LBS of kringle modules that are able to interact with lysine and its structural analogues. Important examples of this are the binding of hPg with proteins containing carboxyl-terminal lysine residues, such as ␣-enolase (46), plasmin-digested fibrin (47), and ␣ 2 -antiplasmin (48). Because of these important functional roles of LBS on kringle domains of hPg, and other proteins containing kringle domains capable of interacting with lysine, many studies have appeared that have effectively characterized the mode of binding of lysine analogues to these structural motifs. In the case of hPg, the kringles containing an LBS are K1 Pg , K2 Pg , K4 Pg , and K5 Pg .
X-ray crystal structures and solution structures of of K1 Pg (30,33), K4 Pg, (49), and K5 Pg (23) complexed to ligands, along with site-directed mutagenesis investigations (22,37), have identified various residues as essential for binding of -amino acids to the LBS, which include (K1 Pg numbering) Asp 54 and Asp 56 , Arg 70 , Tyr 63 , and Tyr 71 , as essential to such binding interactions. An aromatic residue (preferably a Tyr) at position 73 is also a characteristic of kringles with a strong LBS but not through direct participation in ligand binding (39). K2 Pg also binds these types of ligands (24,50) but much more weakly than its companion kringles. In examining the amino acid sequence of K2 Pg , in light of known determinants of kringle/ ligand binding, it appeared that changes of Glu 56 to Asp and Leu 72 to Tyr would provide all primary structural elements for strong ligand binding to occur.  The T m for K2 Pg is 60.7°C, which is within the range of those found for other isolated kringle modules. The T m values for K1 tPA (51) and K2 tPA (34) were 61.3 and 75.4°C, respectively. Those same values for K1 Pg (20), K4 Pg (52), and K5 Pg (52) were 67.7, 57.8, and 50.4°C, respectively. Thus K2 Pg is in about the mid-range of kringles in its thermal stability properties. However, insertion of the Tyr at position 72, a residue present in these other kringles (except for K1 tPA , in which a Phe is found) and the presence of which is known to lead to substantial stabilization of the native structure of K2 tPA (39), also increases the T m of K2 Pg by approximately 8°C. Additional placement of an Asp at position 54 lends further stability to K2 Pg (Table II). The final engineered K2 Pg [C4G/E56D/L72Y] displays an approximate 10°C increase in its T m value. Addition of -amino acids, such as EACA, to kringle modules with which this ligand interacts typically leads to increases in the T m value of 9 -15°C. This also occurs with K2 Pg and all of the mutants (Table II), showing that a similar stabilization occurs in this case. The increase in the ⌬H for thermal denaturation in the presence of EACA also suggests that additional bonds need to be disrupted for destabilization of the native conformation to occur. Thus, the thermal denaturation properties of K2 Pg , characterized in this report, appear to be typical of kringle motifs. With this additional characterization of another hPg structural domain, a time is approaching when all of the properties of the individual modules can be compared with their same properties in the intact protein and the information employed to assess rigorously their structural independence.
The binding of a series of -amino acids to wild-type-K2 Pg (K2 Pg [C4G]) demonstrated an order of binding of t-AMCHA [mt] 5-APA Ͼ EACA [mt ]4ABA Ն 7AHA, in general agreement with a similar study that examined this same binding (50). The K d of EACA for K 2 Pg[C4G], of 560 M, is considerably weaker than that of K1 Pg (12 M (20)), K4 Pg (26 mM (21)), and K5 Pg (140 M (23)). Single substitutions of E56D or L72Y led to a 5-fold increased affinity for EACA in the former case and considerably diminished this binding in the latter case. A double mutation of E56D/R57G resulted in a similar increase in the affinity of K2 Pg for EACA (50) as in our K2 Pg [C4G/E56D] mutant. Changes in both of these critical residues, generating K2 Pg [C4G/E56D/L72Y], resulted in an enhancement in binding of up to 12-fold in the series of ligands investigated (Table III). The K d value for EACA in this case, of 60 M, places this mutant kringle nearest to the K d for K4 Pg .
Finally, with the provision of the up-regulated derivative of K2 Pg designed in this study, a point is reached that enables a determination as to whether binding of this module to selected bacterial cell surfaces is directly coupled to its -amino acidbinding site. Bacterial surface-associated plasmin can provide the proteolytic activity required for invasiveness, and the a1a2 repeats of PAM, a M-like surface protein present in group A streptococci, are known to contain a strong binding site for hPg (53), mediated by the K2 Pg domain (17). This interaction, coupled with the presence of the hPg activator, streptokinase, in these bacteria leads to generation of plasmin on the cell surface (54). Because of this important functional role of K2 Pg , we investigated the binding properties of the K2 Pg mutants gen-   (Table IV) suggest that both enthalpic and entropic processes contribute substantial binding energies to this process. The consistent and somewhat larger enthalpic contribution to the binding suggests that actual bond formation, likely hydrogen bonds, is more important to the binding than reorientation phenomena. These data also suggest that binding sites are largely preformed in the K2 Pg domain, as is the case with other -amino acidbinding kringles. Since a free carboxyl-terminal lysine group does not exist in the peptide, the nature of its interaction with the LBS of K2 Pg is of great interest. It is unlikely that a side chain lysine residue could provide the energy for such tight binding in the absence of the added stability that would be provided by the ligand carboxylate group. However, there are several side chain glutamates present in this peptide that may form a pseudo-lysine when appropriately positioned with the side chain lysine. Examination of the amino acid sequence of peptide VEK30 shows that Lys 14 and Lys 27 are flanked by Glu residues and, in combination, might provide a suitable ligand for the LBS. As seen from the thermodynamic box of Fig. 5, alterations of Glu 56 and Leu 72 are not additive in their effects but synergistic. Furthermore, the coupling energy between E56D and L72Y of Ϫ2.43 kcal⅐mol is equal to that of approximately the generation of a single net hydrogen bond. Interestingly, we proposed earlier in the case of K2 tPA (55) that the OH of Tyr 76 (equivalent to Tyr 72 of K2 Pg [C4G/E56D/L72Y]) is sufficiently close to the OD1 atom of Asp 59 (equivalent to Asp 56 of K2 Pg [C4G/E56D/L72Y], such that a hydrogen bond is possible between these side chains, which may stabilize the binding pocket. Whether distance considerations are appropriate to allow this hydrogen bond to form in K2 Pg awaits more detailed structural information.
In conclusion, the results of this study have shown that strategic rational and knowledge-based minimal mutagenesis of K2 Pg can be employed to up-regulate its LBS, which then can be employed to engineer enhanced functionality into this kringle domain, and likely then into the parent molecule, hPg. This shows the power of approaches that investigate properties of isolated regions of proteins of this type to redesign the characteristics of these proteins.