Prediction of caspase cleavage sites using Bayesian bio-basis function neural networks

MOTIVATION: Apoptosis has drawn the attention of researchers because of its importance in treating some diseases through finding a proper way to block or slow down the apoptosis process. Having understood that caspase cleavage is the key to apoptosis, we find novel methods or algorithms are essential for studying the specificity of caspase cleavage activity and this helps the effective drug design. As bio-basis function neural networks have proven to outperform some conventional neural learning algorithms, there is a motivation, in this study, to investigate the application of bio-basis function neural networks for the prediction of caspase cleavage sites. RESULTS: Thirteen protein sequences with experimentally determined caspase cleavage sites were downloaded from NCBI. Bayesian bio-basis function neural networks are investigated and the comparisons with single-layer perceptrons, multilayer perceptrons, the original bio-basis function neural networks and support vector machines are given. The impact of the sliding window size used to generate sub-sequences for modelling on prediction accuracy is studied. The results show that the Bayesian bio-basis function neural network with two Gaussian distributions for model parameters (weights) performed the best and the highest prediction accuracy is 97.15 +/- 1.13%. AVAILABILITY: The package of Bayesian bio-basis function neural network can be obtained by request to the author.     

Subsite preferences of pepstatin-insensitive carboxyl proteinases from prokaryotes: kumamolysin, a thermostable pepstatin-insensitive carboxyl proteinase

Kumamolysin, a carboxyl proteinase from Bacillus novosp. MN-32, is characterized by its thermostability and insensitivity to aspartic proteinase inhibitors such as pepstatin, diazoacetyl-DL-norleucine methylester, and 1,2-epoxy-3-(p-nitro-phenoxy)propane. Here, its substrate specificity was elucidated using two series of synthetic chromogenic substrates: P(5)-P(4)-P(3)-P(2)-Phe*Nph (p-nitrophenylalanine: *cleavage site)-P(2)'-P(3)', in which the amino acid residues at the P(5)-P(2), P(2)' and P(3)' positions were systematically substituted. Among 74 substrates, kumamolysin was shown to hydrolyze Lys-Pro-Ile-Pro-Phe-Nph-Arg-Leu most effectively. The kinetic parameters of this peptide were K(m) = 41+/-5 microM, k(cat) = 176+/- 10 s(-1), and k(cat)/K(m) = 4.3+/-0.6 mM(-1) x s(-1). These systematic analyses revealed the following features: (i) Kumamolysin had a unique preference for the P(2) position. Kumamolysin preferentially hydrolyzed peptides having an Ala or Pro residue at the P(2) position; this was also observed for the pepstatin-insensitive carboxyl proteinase from Bacillus coagulans J-4 [J-4; Shibata et al. (1998) J. Biochem. 124, 642-647]. Other carboxyl proteinases, including Pseudomonas sp. 101 pepstatin-insensitive carboxyl proteinase (PCP) and Xanthomonas sp. T-22 pepstatin-insensitive carboxyl proteinase (XCP), preferred peptides having hydrophobic and bulky amino acid residue such as Leu at the P(2) position. (ii) Kumamolysin preferred such charged amino acid residues as Glu or Arg at the P(2)' position, suggesting that the S(2)' subsite of kumamolysin is occupied by hydrophilic residues, similar to that of PCP, XCP, and J-4. In general, the S(2)' subsite of pepstatin-sensitive carboxyl proteinases (aspartic proteinases) is hydrophobic in nature. Thus, the hydrophilic nature of the S(2)' subsite was confirmed to be a distinguishing feature of pepstatin-insensitive carboxyl proteinases from prokaryotes.     

Substrate specificity of beta-collagenase from Clostridium histolyticum

The substrate specificity of beta-collagenase from Clostridium histolyticum has been investigated by measuring the rate of hydrolysis of more than 50 tri-, tetra-, penta-, and hexapeptides covering the P3 to P3' subsites of the substrate. The choice of peptides was patterned after sequences found in the alpha 1 and alpha 2 chains of type I collagen. Each peptide contained either a 2-furanacryloyl (FA) or cinnamoyl (CN) group in subsite P2 or the 4-nitrophenylalanine (Nph) residue in subsite P1. Hydrolysis of the P1-P1' bond produces an absorbance change in these chromophoric peptides that has been used to quantitate the rates of their hydrolysis under first order conditions ([S] much less than KM) from kcat/KM values have been obtained. The identity of the amino acids in all six subsites (P3-P3') markedly influences the hydrolysis rates. In general, the best substrates have Gly in subsites P3 and P1', Pro or Ala in subsite P2', and Hyp, Arg, or Ala in subsite P3'. This corresponds well with the frequency of occurrence of these residues in the Gly-X-Y triplets of collagen. In contrast, the most rapidly hydrolyzed substrates do not have residues from collagen-like sequences in subsites P2 and P1. For example, CN-Nph-Gly-Pro-Ala is the best known substrate for beta-collagenase with a kcat/KM value of 4.4 X 10(7) M-1 min-1, in spite of the fact that there is neither Pro nor Ala in P2 or Hyp nor Ala in P1. These results indicate that the previously established rules for the substrate specificity of the enzyme require modification.     

Competitive inhibition of the dengue virus NS3 serine protease by synthetic peptides representing polyprotein cleavage sites

The NS3 serine protease of dengue virus is required for the maturation of the viral polyprotein and consequently represents a promising target for the development of antiviral inhibitors. However, the substrate specificity of this enzyme has been characterized only to a limited extent. In this study, we have investigated product inhibition of the NS3 protease by synthetic peptides derived from the P6-P1 and the P1'-P5' regions of the natural polyprotein substrate. N-terminal cleavage site peptides corresponding to the P6-P1 region of the polyprotein were found to act as competitive inhibitors of the enzyme with K(i) values ranging from 67 to 12 microM. The lowest K(i) value was found for the peptide representing the NS2A/NS2B cleavage site, RTSKKR. Inhibition by this cleavage site sequence was analyzed by using shorter peptides, SKKR, KKR, KR, AGRR, and GKR. With the exception of the peptide AGRR which did not inhibit the protease at a concentration of 1mM, all other peptides displayed K(i) values in the range from 188 to 22 microM. Peptides corresponding to the P1'-P5' region of the polyprotein cleavage sites had no effect on enzymatic activity at a concentration of 1mM. Molecular docking data of peptide inhibitors to a homology-based model of the dengue virus type 2 NS2B(H)-NS3p co-complex indicate that binding of the non-prime site product inhibitors is similar to ground-state binding of the corresponding substrates.     

Residues surrounding Arg336 and Arg562 contribute to the disparate rates of proteolysis of factor VIIIa catalyzed by activated protein C

Activated Protein C (APC) inactivates factor VIIIa by cleavage at Arg(336) and Arg(562) within the A1 and A2 subunits, respectively, with reaction at the former site occurring at a rate approximately 25-fold faster than the latter. Recombinant factor VIII variants possessing mutations within the P4-P3' sequences were used to determine the contributions of these residues to the disparate cleavage rates at the two P1 sites. Specific activity values for 336(P4-P3')562, 336(P4-P2)562, and 336(P1'-P3')562 mutants, where indicated residues surrounding the Arg(336) site were replaced with those surrounding Arg(562), were similar to wild type (WT) factor VIII; whereas 562(P4-P3')336 and 562(P4-P2)336 mutants showed specific activity values <1% the WT value. Inactivation rates for the 336 site mutants were reduced approximately 6-11-fold compared with WT factor VIIIa, and approached values attributed to cleavage at Arg(562). Cleavage rates at Arg(336) were reduced approximately 100-fold for 336(P4-P3')562, and approximately 9-16-fold for 336(P4-P2)562 and 336(P1'-P3')562 mutants. Inhibition kinetics revealed similar affinities of APC for WT factor VIIIa and 336(P4-P3')562 variant. Alternatively, the 562(P4-P3')336 variant showed a modest increase in cleavage rate ( approximately 4-fold) at Arg(562) compared with WT, whereas these rates were increased by approximately 27- and 6-fold for 562(P4-P3')336 and 562(P4-P2)336, respectively, using the factor VIII procofactor form as substrate. Thus the P4-P3' residues surrounding Arg(336) and Arg(562) make significant contributions to proteolysis rates at each site, apparently independent of binding affinity. Efficient cleavage at Arg(336) by APC is attributed to favorable P4-P3' residues at this site, whereas cleavage at Arg(562) can be accelerated following replacement with more optimal P4-P3' residues.     

Cleavage specificity of the subtilisin-like protease C1 from soybean

The cleavage specificity of protease C1, isolated from soybean (Glycine max (L.) Merrill) seedling cotyledons, was examined using oligopeptide substrates in an HPLC based assay. A series of peptides based on the sequence Ac-KVEKEESEEGE-NH2 was used, mimicking a natural cleavage site of protease C1 in the alpha subunit of the storage protein beta-conglycinin. A study of substrate peptides truncated from either the N- or C-terminus indicates that the minimal requirements for cleavage by protease C2 are three residues N-terminal to the cleaved bond, and two residues C-terminal (i.e. P3-P2'). The maximal rate of cleavage is reached with substrates containing four to five residues N-terminal to the cleaved bond and four residues C-terminal (i.e. P4 or P5 to P4'). The importance of Glu residues at the P1, P1', and P4 positions was examined using a series of substituted nonapeptides (P5-P4') with a base sequence of Ac-KVEKEESEE-NH2. At the P1 position, the relative ranking, based on kcat/Km, was E>Q>K>A>D>F>S. Substitutions at the P1' position yield the ranking E congruent withQ>A>S>D>K>F, while those at P4' had less effect on kcat/Km, yielding the ranking F congruent with S congruent with E congruent withD>K>A congruent withQ. These data show that protease C1 prefers to cleave at Glu-Glu and Glu-Gln bonds, and that the nature of the P4' position is less important. The fact that there is specificity in the cleavage of the oligopeptides suggests that the more limited specific cleavage of the alpha and alpha' subunits of beta-conglycinin by protease C1 is due to a combination of the sequence cleavage specificity of the protease and the accessibility of appropriate scissile peptide bonds on the surface of the substrate protein.     

Switching the substrate specificity of the two-component NS2B-NS3 flavivirus proteinase by structure-based mutagenesis

The flavivirus NS2B-NS3(pro)teinase is an essential element in the proteolytic processing of the viral precursor polyprotein and therefore a potential drug target. Recently, crystal structures and substrate preferences of NS2B-NS3pro from Dengue and West Nile viruses (DV and WNV) were determined. We established that the presence of Gly-Gly at the P1'-P2' positions is optimal for cleavage by WNV NS3pro, whereas DV NS3pro tolerates well the presence of bulky residues at either P1' or P2'. Structure-based modeling suggests that Arg(76) and Pro(131)-Thr(132) limit the P1'-P2' subsites and restrict the cleavage preferences of the WNV enzyme. In turn, Leu(76) and Lys(131)-Pro(132) widen the specificity of DV NS3pro. Guided by these structural models, we expressed and purified mutant WNV NS2B-NS3pro and evaluated cleavage preferences by using positional scanning of the substrate peptides in which the P4-P1 and the P3'-P4' positions were fixed and the P1' and P2' positions were each randomized. We established that WNV R76L and P131K-T132P mutants acquired DV-like cleavage preferences, whereas T52V had no significant effect. Our work is the first instance of engineering a viral proteinase with switched cleavage preferences and should provide valuable data for the design of optimized substrates and substrate-based selective inhibitors of flaviviral proteinases.     

P2-P3 conformationally constrained ketoamide-based inhibitors of cathepsin K

An orally bioavailable series of ketoamide-based cathepsin K inhibitors with good pharmacokinetic properties has been identified. Starting from a potent inhibitor endowed with poor drug properties, conformational constraint of the P(2)-P(3) linker and modifications to P(1') elements led to an enhancement in potency, solubility, clearance, and bioavailability. These optimized inhibitors attenuated bone resorption in a rat TPTX hypocalcemic bone resorption model.     

Heterotrimeric collagen peptides as fluorogenic collagenase substrates: synthesis, conformational properties, and enzymatic digestion

The collagenase cleavage site of collagen type I, i.e., the sequence portions 772-784 (P(4)-P(9)') and 772-785 (P(4)-P(10)') of the two alpha1-chains and the sequence portion 772-784 (P(4)-P(9)') of the alpha2-chain, were assembled in an alpha1alpha2alpha1' register by C-terminal cross-linking of these peptides with an artificial cystine knot. The triple-helical conformation of the construct was stabilized by N-terminal extensions with (Gly-Pro-Hyp)(5) repeats. The gaps in the sequence alignment were filled up, and the alpha1-chain was dansylated and the alpha1'-chain was acylated with a tryptophan residue to place in spatial proximity the two chromophores for an efficient fluorescence resonance energy transfer. Although the incorporation of the two N-terminal chromophores leads to partial destabilization of the overall triple-helical fold, the heterotrimer behaved as a collagen-like substrate of the matrix metalloproteinases MMP-1 and MMP-13. Cleavage of the fluorogenic heterotrimer leads to a 6-fold increase in fluorescence intensity, thus making it a useful fluorogenic substrate for interstitial collagenases. With this folded heterotrimeric collagen molecule it was shown that fluorescence resonance energy transfer, as applied so far only for the design of linear fluorogenic enzyme substrates, can also be exploited in conformation dependency.     

Inhibition of adenosine and thymidylate kinases by bisubstrate analogs

Potential bisubstrate analogs, in which the 5'-hydroxyl group of adenosine was joined to the phosphoryl group acceptor by polyphosphoryl bridges of varying length (ApnX, where n is the number of phosphoryl groups and X is the nucleoside moiety of the acceptor), were tested as inhibitors of human liver adenosine kinase and of thymidylate kinase from peripheral blast cells of patients with acute myelocytic leukemia. Adenosine kinase was most strongly inhibited by P1,P4-(diadenosine 5')-tetraphosphate (Kd = 30 nM) and P1,P5-(diadenosine 5')-pentaphosphate (Kd = 73 nM). Thymidylate kinase was most strongly inhibited by P1-(adenosine 5')-P5-(thymidine 5')-pentaphosphate (Kd = 120 nM) and by P1(adenosine 5')-P6-(thymidine 5')-hexaphosphate (Kd = 43 nM). In these enzymes, as in adenylate and thymidylate kinases, strongest inhibition was achieved in compounds containing one or two more phosphoryl groups than the substrates combined. These results support the view that nucleoside and nucleotide kinases mediate direct transfer of phosphoryl groups from ATP to acceptors, rather than acting by a double displacement mechanism.     

Cleavage-site specificity of prolyl endopeptidase FAP investigated with a full-length protein substrate

Fibroblast activation protein (FAP) is a prolyl-cleaving endopeptidase proposed as an anti-cancer drug target. It is necessary to define its cleavage-site specificity to facilitate the identification of its in vivo substrates and to understand its biological functions. We found that the previously identified substrate of FAP, α(2)-anti-plasmin, is not a robust substrate in vitro. Instead, an intracellular protein, SPRY2, is cleavable by FAP and more suitable for investigation of its substrate specificity in the context of the full-length globular protein. FAP prefers uncharged residues, including small or bulky hydrophobic amino acids, but not charged amino acids, especially acidic residue at P1', P3 and P4 sites. Molecular modelling analysis shows that the substrate-binding site of FAP is surrounded by multiple tyrosine residues and some negatively charged residues, which may exert least preference for substrates with acidic residues. This provides an explanation why FAP cannot cleave interleukins, which have a glutamate at either P4 or P2', despite their P3-P2-P1 sites being identical to SPRY2 or α-AP. Our study provided new information on FAP cleavage-site specificity, which differs from the data obtained by profiling with a peptide library or with the denatured protein, gelatin, as the substrate. Furthermore, our study suggests that negatively charged residues should be avoided when designing FAP inhibitors.     

Heparin enhances the specificity of antithrombin for thrombin and factor Xa independent of the reactive center loop sequence. Evidence for an exosite determinant of factor Xa specificity in heparin-activated antithrombin

Heparin activates the primary serpin inhibitor of blood clotting proteinases, antithrombin, both by an allosteric conformational change mechanism that specifically enhances factor Xa inactivation and by a ternary complex bridging mechanism that promotes the inactivation of thrombin and other target proteinases. To determine whether the factor Xa specificity of allosterically activated antithrombin is encoded in the reactive center loop sequence, we attempted to switch this specificity by mutating the P6-P3' proteinase binding sequence excluding P1-P1' to a more optimal thrombin recognition sequence. Evaluation of 12 such antithrombin variants showed that the thrombin specificity of the serpin allosterically activated by a heparin pentasaccharide could be enhanced as much as 55-fold by changing P3, P2, and P2' residues to a consensus thrombin recognition sequence. However, at most 9-fold of the enhanced thrombin specificity was due to allosteric activation, the remainder being realized without activation. Moreover, thrombin specificity enhancements were attenuated to at most 5-fold with a bridging heparin activator. Surprisingly, none of the reactive center loop mutations greatly affected the factor Xa specificity of the unactivated serpin or the several hundred-fold enhancement in factor Xa specificity due to activation by pentasaccharide or bridging heparins. Together, these results suggest that the specificity of both native and heparin-activated antithrombin for thrombin and factor Xa is only weakly dependent on the P6-P3' residues flanking the primary P1-P1' recognition site in the serpin-reactive center loop and that heparin enhances serpin specificity for both enzymes through secondary interaction sites outside the P6-P3' region, which involve a bridging site on heparin in the case of thrombin and a previously unrecognized exosite on antithrombin in the case of factor Xa.     

Kinetic and modeling studies of S3-S3' subsites of HIV proteinases.

Kinetic analysis and modeling studies of HIV-1 and HIV-2 proteinases were carried out using the oligopeptide substrate [formula: see text] and its analogs containing single amino acid substitutions in P3-P3' positions. The two proteinases acted similarly on the substrates except those having certain hydrophobic amino acids at P2, P1, P2', and P3' positions (Ala, Leu, Met, Phe). Various amino acids seemed to be acceptable at P3 and P3' positions, while the P2 and P2' positions seemed to be more restrictive. Polar uncharged residues resulted in relatively good binding at P3 and P2 positions, while at P2' and P3' positions they gave very high Km values, indicating substantial differences in the respective S and S' subsites of the enzyme. Lys prevented substrate hydrolysis at any of the P2-P2' positions. The large differences for subsite preference at P2 and P2' positions seem to be at least partially due to the different internal interactions of P2 residue with P1', and P2' residue with P1. As expected on the basis of amino acid frequency in the naturally occurring cleavage sites, hydrophobic residues at P1 position resulted in cleavable peptides, while polar and beta-branched amino acids prevented hydrolysis. On the other hand, changing the P1' Pro to other amino acids prevented substrate hydrolysis, even if the substituted amino acid had produced a good substrate in other oligopeptides representing naturally occurring cleavage sites. The results suggest that the subsite specificity of the HIV proteinases may strongly depend on the sequence context of the substrate.     

Sequence signals for generation of antigenic peptides by the proteasome: implications for proteasomal cleavage mechanism

Proteasomal cleavage of proteins is the first step in the processing of most antigenic peptides that are presented to cytotoxic T cells. Still, its specificity and mechanism are not fully understood. To identify preferred sequence signals that are used for generation of antigenic peptides by the proteasome, we performed a rigorous analysis of the residues at the termini and flanking regions of naturally processed peptides eluted from MHC class I molecules. Our results show that both the C terminus (position P1 of the cleavage site) and its immediate flanking position (P1') possess significant signals. The N termini of the peptides show these signals only weakly, consistent with previous findings that antigenic peptides may be cleaved by the proteasome with N-terminal extensions. Nevertheless, we succeed to demonstrate indirectly that the N-terminal cleavage sites contain the same preferred signals at position P1'. This reinforces previous findings regarding the role of the P1' position of a cleavage site in determining the cleavage specificity, in addition to the well-known contribution of position P1. Our results apply to the generation of antigenic peptides and bare direct implications for the mechanism of proteasomal cleavage. We propose a model for proteasomal cleavage mechanism by which both ends of cleaved fragments are determined by the same cleavage signals, involving preferred residues at both P1 and P1' positions of a cleavage site. The compatibility of this model with experimental data on protein degradation products and generation of antigenic peptides is demonstrated.     

Cleavage preference distinguishes the two-component NS2B-NS3 serine proteinases of Dengue and West Nile viruses

Regulated proteolysis of the polyprotein precursor by the NS2B-NS3 protease is required for the propagation of infectious virions. Unless the structural and functional parameters of NS2B-NS3 are precisely determined, an understanding of its functional role and the design of flaviviral inhibitors will be exceedingly difficult. Our objectives were to define the substrate recognition pattern of the NS2B-NS3 protease of West Nile and Dengue virises (WNV and DV respectively). To accomplish our goals, we used an efficient, 96-well plate format, method for the synthesis of 9-mer peptide substrates with the general P4-P3-P2-P1-P1'-P2'-P3'-P4'-Gly structure. The N-terminus and the constant C-terminal Gly of the peptides were tagged with a fluorescent tag and with a biotin tag respectively. The synthesis was followed by the proteolytic cleavage of the synthesized, tagged peptides. Because of the strict requirement for the presence of basic amino acid residues at the P1 and the P2 substrate positions, the analysis of approx. 300 peptide sequences was sufficient for an adequate representation of the cleavage preferences of the WNV and DV proteinases. Our results disclosed the strict substrate specificity of the WNV protease for which the (K/R)(K/R)R/GG amino acid motifs was optimal. The DV protease was less selective and it tolerated well the presence of a number of amino acid residue types at either the P1' or the P2' site, as long as the other position was occupied by a glycine residue. We believe that our data represent a valuable biochemical resource and a solid foundation to support the design of selective substrates and synthetic inhibitors of flaviviral proteinases.     

Alteration of serpin specificity by a protein cofactor. Vitronectin endows plasminogen activator inhibitor 1 with thrombin inhibitory properties.

Serine protease inhibitors ("serpins") are highly homologous proteins which inhibit selected "target" serine proteases by acting as a pseudo-substrate. Their specificity is primarily determined by the amino acid sequence around the carboxyl-terminally located reactive center (P1-P1'). In addition, the association rate constant between a serpin and a serine protease can be dramatically increased by non-protein cofactors, such as heparin in the case of thrombin inhibition by antithrombin III. In an attempt to alter the specificity of PAI-1 from an inhibitor of the fibrinolytic system to an inhibitor of coagulation, we replaced P1-P1' or P3 through P3' of the reactive center of PAI-1 by the corresponding residues of antithrombin III and assessed whether the mutant proteins, purified from lysates of transformed Escherichia coli cells, had acquired thrombin inhibitory properties. The experiments were performed in the presence and absence of vitronectin, a multifunctional protein which has been shown to bind PAI-1 in plasma and in the matrix of endothelial cells. The second-order rate constants for t-PA inhibition of "wild-type" PAI-1 and PAI P1-P1'ATIII, irrespective of the presence of vitronectin, were similar, whereas replacing P3-P3' resulted in a 40-fold decrease of the second-order rate constant towards t-PA, again independent of vitronectin. In the absence of vitronectin, reactivity of PAI-1 and its "antithrombin III-like" variants towards thrombin was slow; however, PAI-1 P3-P3' ATIII had a 10-fold higher k1 than wild-type PAI-1 (1.3 x 10(4) M-1 s-1 versus 1.1 x 10(3) M-1 s-1). In contrast, in the presence of vitronectin, PAI-1 and even more rapidly PAI-1 P3-P3'ATIII were found to be effective thrombin inhibitors, with k1 values of 2.2 x 10(5) M-1s-1 and 1.8 x 10(6) M-1 s-1, respectively. Thus, in the presence of vitronectin, PAI-1 P3-P3'ATIII displays a 3-fold higher k1 with thrombin than with t-PA. It is shown that vitronectin enhances, in a dose-dependent manner, the formation of sodium dodecyl sulfate-resistant complexes between PAI-1 or mutants thereof and thrombin. Therefore, vitronectin is the first protein described to function as a cofactor for serpin specificity. PAI-1 is proposed to be a versatile inhibitor which, in the presence of vitronectin, can modulate both coagulation and fibrinolysis.     

Pepstatin-insenstive acid proteases from Scytalidium lignicolum. Kinetic study with synthetic peptides.

A kinetic study was conducted on the acid proteases A-1 and A-2 from Scytalidium lignicolum using synthetic peptides as substrates. Almost maximum activity was attained with N-acylated tetrapeptides as the molecular size of substrates was increased. Suitable amino acid residues were required at the P1-P2 and P1'-P2' positions [notation of Schechter and Berger (14)]. Hydrophobic or bulky residues such as leucine were specifically required at the P1 and P1' positions, with the specificity at the latter position being considerably lower than that at the former. For catalysis, the presence of certain amino acid residues at the P2 and P2' positions was essential, mainly in relation to kcat. An inhibition study supported this view. Stringent stereospecificity was observed at the P2 and P2' positions, but the side chain specificity was low. Study of the B enzyme from the same organism was very difficult owing to its low activity against the peptides used. The Scytalidium acid proteases A-1, A-2, and B showed considerably different behavior against peptide substrates in comparison with usual acid proteases, which are senstive to pepstatin.     

Inductive learning approaches to rainfall-runoff modelling

Trying to model the rainfall-runoff process is a complex activity as it is influenced by a number of implicit and explicit factors--for example, precipitation distribution, evaporation, transpiration, abstraction, watershed topography, and soil types. However, this kind of forecasting is particularly important as it is used to predict serious flooding, estimate erosion and identify problems associated with low flow. Inductive learning approaches (e.g. decision trees and artificial neural networks) are particularly well suited to problems of this nature as they can often interpret underlying factors (such as seasonal variations) which cannot be modelled by other techniques. In addition, these approaches can easily be trained on the explicit factors (e.g. rainfall) and the inexplicit factors (e.g. abstraction) that affect river flow. Inductive learning approaches can also be extended to account for new factors that emerge over a period of time. This paper evaluates the application of decision trees and two artificial neural network models (the multilayer perceptron and the radial basis function network) to river flow forecasting in two flood prone UK catchments using real hydrometric data. Comparisons are made between the performance of these approaches and conventional flood forecasting systems.     

Specificity and inhibition of proteases from human immunodeficiency viruses 1 and 2

Highly purified, recombinant preparations of the virally encoded proteases from human immunodeficiency viruses (HIV) 1 and 2 have been compared relative to 1) their specificities toward non-viral protein and synthetic peptide substrates, and 2) their inhibition by several P1-P1' pseudodipeptidyl-modified substrate analogs. Hydrolysis of the Leu-Leu and Leu-Ala bonds in the Pseudomonas exotoxin derivative, Lys-PE40, is qualitatively the same for HIV-2 protease as published earlier for the HIV-1 enzyme (Tomasselli, A. G., Hui, J. O., Sawyer, T. K., Staples, D. J., FitzGerald, D. J., Chaudhary, V. K., Pastan, I., and Heinrikson, R. L. (1990) J. Biol. Chem. 265, 408-413). However, the rates of cleavage at these two sites are reversed for the HIV-2 protease which prefers the Leu-Ala bond. The kinetics of hydrolysis of this protein substrate by both enzymes are mirrored by those obtained from cleavage of model peptides. Hydrolysis by the two proteases of other synthetic peptides modeled after processing sites in HIV-1 and HIV-2 gag polyproteins and selected analogs thereof demonstrated differences, as well as similarities, in selectivity. For example, while the two proteases were nearly identical in their rates of cleavage of the Tyr-Pro bond in the HIV-1 gag fragment, Val-Ser-Gln-Asn-Tyr-Pro-Ile-Val, the HIV-1 protease showed a 64-fold enhancement over the HIV-2 enzyme in hydrolysis of a Tyr-Val bond in the same template. Accordingly, the HIV-2 protease appears to have a different specificity than the HIV-1 enzyme; it is better able to hydrolyze substrates with small amino acids in P1 and P1', but is variable in its rate of hydrolysis of peptides with bulky substituents in these positions. In addition to these comparisons of the two proteases with respect to substrate specificity, we present inhibitor structure-activity data for the HIV-2 protease. Relative to P1-P1' statine or Phe psi [CH2N]Pro-modified pseudopeptidyl inhibitors, compounds having Xaa psi[CH(OH)CH2]Yaa inserts were found to show significantly higher affinities to both enzymes, generally binding from 10 to 100 times stronger to HIV-1 protease than to the HIV-2 enzyme. Molecular modeling comparisons based upon the sequence homology of the two enzymes and x-ray crystal structures of HIV-1 protease suggest that most of the nonconservative amino acid replacements occur in regions well outside the catalytic cleft, while only subtle structural differences exist within the active site.(ABSTRACT TRUNCATED AT 400 WORDS)     

Reduced bio-basis function neural networks for protease cleavage site prediction

This paper presents a new neural learning algorithm for protease cleavage site prediction. The basic idea is to replace the radial basis function used in radial basis function neural networks by a so-called bio-basis function using amino acid similarity matrices. Mutual information is used to select bio-bases and a corresponding selection algorithm is developed. The algorithm has been applied to the prediction of HIV and Hepatitis C virus protease cleavage sites in proteins with success.