Engineering of the substrate-binding region of the subtilisin-like, cell-envelope proteinase of Lactococcus lactis

Comparative Structure Analysis of the Multi-Domain, Cell Envelope Proteases of Lactic Acid Bacteria

Lactic acid bacteria (LAB) have an extracellular proteolytic system that includes a multi-domain, cell envelope protease (CEP) with a subtilisin homologous protease domain. These CEPs have different proteolytic activities despite having similar protein sequences. Structural characterization has previously been limited to CEP homologs of dairy- and human-derived LAB strains, excluding CEPs of plant-derived LAB strains. CEP structures are a challenge to determine experimentally due to their large size and attachment to the cell envelope. This study aims to clarify the prevalence and structural diversity of CEPs by using the structure prediction software AlphaFold 2. Domain boundaries are clarified based on a comparative analysis of 21 three-dimensional structures, revealing novel domain architectures of CEP homologs that are not necessarily restricted to specific LAB species or ecological niches. The C-terminal flanking region of the protease domain is divided into fibronectin type-III-like domains with various structural traits. The analysis also emphasizes the existence of two distinct domains for cell envelope attachment that are preceded by an intrinsically disordered cell wall spanning domain. The domain variants and their combinations provide CEPs with different stability, proteolytic activity, and potentially adhesive properties, making CEPs targets for steering proteolytic activity with relevance for both food development and human health.

Modeled Structure of the Cell Envelope Proteinase of Lactococcus lactis

The cell envelope proteinase (CEP) of Lactococcus lactis is a large extracellular protease covalently linked to the peptidoglycan of the cell wall. Strains of L. lactis are typically auxotrophic for several amino acids and in order to grow to high cell densities in milk they need an extracellular protease. The structure of the entire CEP enzyme is difficult to determine experimentally due to the large size and due to the attachment to the cell surface. We here describe the use of a combination of structure prediction tools to create a structural model for the entire CEP enzyme of Lactococcus lactis. The model has implications for how the bacterium interacts with casein micelles during growth in milk, and it has implications regarding the energetics of the proteolytic system. Our model for the CEP indicates that the catalytic triad is activated through a structural change caused by interaction with the substrate. The CEP of L. lactis might become a useful model for the mode of action for enzymes belonging to the large class of S8 proteinases with a PA (protease associated) domain and a downstream fibronectin like domain.

LAB-Secretome: a genome-scale comparative analysis of the predicted extracellular and surface-associated proteins of Lactic Acid Bacteria

Background

In Lactic Acid Bacteria (LAB), the extracellular and surface-associated proteins can be involved in processes such as cell wall metabolism, degradation and uptake of nutrients, communication and binding to substrates or hosts. A genome-scale comparative study of these proteins (secretomes) can provide vast information towards the understanding of the molecular evolution, diversity, function and adaptation of LAB to their specific environmental niches.

Results

We have performed an extensive prediction and comparison of the secretomes from 26 sequenced LAB genomes. A new approach to detect homolog clusters of secretome proteins (LaCOGs) was designed by integrating protein subcellular location prediction and homology clustering methods. The initial clusters were further adjusted semi-manually based on multiple sequence alignments, domain compositions, pseudogene analysis and biological function of the proteins. Ubiquitous protein families were identified, as well as species-specific, strain-specific, and niche-specific LaCOGs. Comparative analysis of protein subfamilies has shown that the distribution and functional specificity of LaCOGs could be used to explain many niche-specific phenotypes.

A comprehensive and user-friendly database LAB-Secretome was constructed to store, visualize and update the extracellular proteins and LaCOGs http://www.cmbi.ru.nl/lab_secretome/. This database will be updated regularly when new bacterial genomes become available.

Conclusions

The LAB-Secretome database could be used to understand the evolution and adaptation of lactic acid bacteria to their environmental niches, to improve protein functional annotation and to serve as basis for targeted experimental studies.

Characteristic features in the structure and collagen-binding ability of a thermophilic collagenolytic protease from the thermophile Geobacillus collagenovorans MO-1

A collagen-degrading thermophile, Geobacillus collagenovorans MO-1, extracellularly produces a collagenolytic protease with a large molecular mass. Complete nucleotide sequencing of this gene after gene cloning revealed that the collagenolytic protease is a member of the subtilisin family of serine proteases and consists of a signal sequence for secretion, a prosequence for maturation, a catalytic region, 14 direct repeats of 20 amino acids at the C terminus, and a region with unknown function intervening between the catalytic region and the numerous repeats. Since the unusual repeats are most likely to be cleaved in the secreted form of the enzyme, the intervening region was investigated to determine whether it participates in collagen binding to facilitate collagen degradation. It was found that the mature collagenolytic protease containing the intervening region at the C terminus bound collagen but not the other insoluble proteins, elastin and keratin. Furthermore, the intervening region fused with glutathione S-transferase showed a collagen-binding ability comparable to that of the mature collagenolytic protease. The collagen-binding ability was finally attributed to two-thirds of the intervening region which is rich in beta-strands and is approximately 35 kDa in molecular mass. In the collagenolytic protease from strain MO-1, hydrogen bonds most likely predominate over the hydrophobic interaction for collagen binding, since a higher concentration of NaCl released collagen from the enzyme surface but a nonionic detergent could not. To the best of our knowledge, this is the first report of a thermophilic collagenolytic protease containing the collagen-binding segment.

Identification and genetic characterization of a novel proteinase, PrtR, from the human isolate Lactobacillus rhamnosus BGT10

A novel proteinase, PrtR, produced by the human vaginal isolate Lactobacillus rhamnosus strain BGT10 was identified and genetically characterized. The prtR gene and flanking regions were cloned and sequenced. The deduced amino acid sequence of PrtR shares characteristics that are common for other cell envelope proteinases (CEPs) characterized to date, but in contrast to the other cell surface subtilisin-like serine proteinases, it has a smaller and somewhat different B domain and lacks the helix domain, and the anchor domain has a rare sorting signal sequence. Furthermore, PrtR lacks the insert domain, which otherwise is situated inside the catalytic serine protease domain of all CEPs, and has a different cell wall spacer (W) domain similar to that of the cell surface antigen I and II polypeptides expressed by oral and vaginal streptococci. Moreover, the PrtR W domain exhibits significant sequence homology to the consensus sequence that has been shown to be the hallmark of human intestinal mucin protein. According to its alpha(S1)- and beta-casein cleavage efficacy, PrtR is an efficient proteinase at pH 6.5 and is distributed throughout all L. rhamnosus strains tested. Proteinase extracts of the BGT10 strain obtained with Ca(2+)-free buffer at pH 6.5 were proteolytically active. The prtR promoter-like sequence was determined, and the minimal promoter region was defined by use of prtR-gusA operon fusions. The prtR expression is Casitone dependent, emphasizing that nitrogen depletion elevates its transcription. This is in correlation with the catalytic activity of the PrtR proteinase.

Determination of the domain of the Lactobacillus delbrueckii subsp. bulgaricus cell surface proteinase PrtB involved in attachment to the cell wall after heterologous expression of the prtB gene in Lactococcus lactis

Belonging to the subtilase family, the cell surface proteinase (CSP) PrtB of Lactobacillus delbrueckii subsp. bulgaricus differs from other CSPs synthesized by lactic acid bacteria. Expression of the prtB gene under its own promoter was shown to complement the proteinase-deficient strain MG1363 (PrtP(-) PrtM(-)) of Lactococcus lactis subsp. cremoris. Surprisingly, the maturation process of PrtB, unlike that of lactococcal CSP PrtPs, does not require a specific PrtM-like chaperone. The carboxy end of PrtB was previously shown to be different from the consensus anchoring region of other CSPs and exhibits an imperfect duplication of 59 amino acids with a high lysine content. By using a deletion strategy, the removal of the last 99 amino acids, including the degenerated anchoring signal (LPKKT), was found to be sufficient to release a part of the truncated PrtB into the culture medium and led to an increase in PrtB activity. This truncated PrtB is still active and enables L. lactis MG1363 to grow in milk supplemented with glucose. By contrast, deletion of the last 806 amino acids of PrtB led to the secretion of an inactive proteinase. Thus, the utmost carboxy end of PrtB is involved in attachment to the bacterial cell wall. Proteinase PrtB constitutes a powerful tool for cell surface display of heterologous proteins like antigens.

Cell-wall proteinases PrtS and PrtB have a different role in Streptococcus thermophilus/Lactobacillus bulgaricus mixed cultures in milk

The manufacture of yoghurt relies on the simultaneous utilization of two starters: Streptococcus thermophilus and Lactobacillus delbrueckii subsp. bulgaricus (Lb. bulgaricus). A protocooperation usually takes place between the two species, which often results in enhanced milk acidification and aroma formation compared to pure cultures. Cell-wall proteinases of Lactococcus lactis and lactobacilli have been shown to be essential to growth in milk in pure cultures. In this study, the role of proteinases PrtS from S. thermophilus and PrtB from Lb. bulgaricus in bacterial growth in milk was evaluated; a negative mutant for the prtS gene of S. thermophilus CNRZ 385 was constructed for this purpose. Pure cultures of S. thermophilus CNRZ 385 and its PrtS-negative mutant were made in milk as well as mixed cultures of S. thermophilus and Lb. bulgaricus: S. thermophilus CNRZ 385 or its PrtS-negative mutant was associated with several strains of Lb. bulgaricus, including a PrtB-negative strain. The pH and growth of bacterial populations of the resulting mixed cultures were followed, and the Lactobacillus strain was found to influence both the extent of the benefit of Lb. bulgaricus/S. thermophilus association on milk acidification and the magnitude of S. thermophilus population dominance at the end of fermentation. In all mixed cultures, the sequential growth of S. thermophilus then of Lb. bulgarius and finally of both bacteria was observed. Although proteinase PrtS was essential to S. thermophilus growth in milk in pure culture, it had no effect on bacterial growth and thus on the final pH of mixed cultures in the presence of PrtB. In contrast, proteinase PrtB was necessary for the growth of S. thermophilus, and its absence resulted in a higher final pH. From these results, a model of growth of both bacteria in mixed cultures in milk is proposed.

The autoproteolysis of Lactococcus lactis lactocepin III affects its specificity towards beta-casein

The effect of autoproteolysis of Lactococcus lactis lactocepin III on its specificity towards beta-casein was investigated. beta-Casein degradation was performed by using either an autolysin-defective derivative of L. lactis MG1363 carrying the proteinase genes of L. lactis SK11, which was unable to transport oligopeptides, or autoproteolyzed enzyme purified from L. lactis SK11. Comparison of the peptide pools by high-performance liquid chromatography analysis revealed significant differences. To analyze these differences in more detail, the peptides released by the cell-anchored proteinase were identified by on-line coupling of liquid chromatography to mass spectrometry. More than 100 oligopeptides were released from beta-casein by the cell-anchored proteinase. Analysis of the cleavage sites indicated that the specificity of peptide bond cleavage by the cell-anchored proteinase differed significantly from that of the autoproteolyzed enzyme.

Streptococcus thermophilus cell wall-anchored proteinase: release, purification, and biochemical and genetic characterization

Streptococcus thermophilus CNRZ 385 expresses a cell envelope proteinase (PrtS), which is characterized in the present work, both at the biochemical and genetic levels. Since PrtS is resistant to most classical methods of extraction from the cell envelopes, we developed a three-step process based on loosening of the cell wall by cultivation of the cells in the presence of glycine (20 mM), mechanical disruption (with alumina powder), and enzymatic treatment (lysozyme). The pure enzyme is a serine proteinase highly activated by Ca(2+) ions. Its activity was optimal at 37 degrees C and pH 7.5 with acetyl-Ala-Ala-Pro-Phe-paranitroanilide as substrate. The study of the hydrolysis of the chromogenic and casein substrates indicated that PrtS presented an intermediate specificity between the most divergent types of cell envelope proteinases from lactococci, known as the PI and PIII types. This result was confirmed by the sequence determination of the regions involved in substrate specificity, which were a mix between those of PI and PIII types, and also had unique residues. Sequence analysis of the PrtS encoding gene revealed that PrtS is a member of the subtilase family. It is a multidomain protein which is maturated and tightly anchored to the cell wall via a mechanism involving an LPXTG motif. PrtS bears similarities to cell envelope proteinases from pyogenic streptococci (C5a peptidase and cell surface proteinase) and lactic acid bacteria (PrtP, PrtH, and PrtB). The highest homologies were found with streptococcal proteinases which lack, as PrtS, one domain (the B domain) present in cell envelope proteinases from all other lactic acid bacteria.

Deletion of various carboxy-terminal domains of Lactococcus lactis SK11 proteinase: effects on activity, specificity, and stability of the truncated enzyme

The Lactococcus lactis SK11 cell envelope proteinase is an extracellular, multidomain protein of nearly 2,000 residues consisting of an N-terminal serine protease domain, followed by various other domains of largely unknown function. Using a strategy of deletion mutagenesis, we have analyzed the function of several C-terminal domains of the SK11 proteinase which are absent in cell envelope proteinases of other lactic acid bacteria. The various deletion mutants were functionally expressed in L. lactis and analyzed for enzyme stability, activity, (auto)processing, and specificity toward several substrates. C-terminal deletions of first the cell envelope W (wall) and AN (anchor) domains and then the H (helix) domain leads to fully active, secreted proteinases of unaltered specificity. Gradually increasing the C-terminal deletion into the so-called B domain leads to increasing instability and autoproteolysis and progressively less proteolytic activity. However, the mutant with the largest deletion (838 residues) from the C terminus and lacking the entire B domain still retains proteolytic activity. All truncated enzymes show unaltered proteolytic specificity toward various substrates. This suggests that the main role played by these domains is providing stability or protection from autoproteolysis (B domain), spacing away from the cell (H domain), and anchoring to the cell envelope (W and AN domains). In addition, this study allowed us to more precisely map the main C-terminal autoprocessing site of the SK11 proteinase and the epitope for binding of group IV monoclonal antibodies.

Genetic characterization of a cell envelope-associated proteinase from Lactobacillus helveticus CNRZ32

A cell envelope-associated proteinase gene (prtH) was identified in Lactobacillus helveticus CNRZ32. The prtH gene encodes a protein of 1,849 amino acids and with a predicted molecular mass of 204 kDa. The deduced amino acid sequence of the prtH product has significant identity (45%) to that of the lactococcal PrtP proteinases. Southern blot analysis indicates that prtH is not broadly distributed within L. helveticus. A prtH deletion mutant of CNRZ32 was constructed to evaluate the physiological role of PrtH. PrtH is not required for rapid growth or fast acid production in milk by CNRZ32. Cell surface proteinase activity and specificity were determined by hydrolysis of alpha(s1)-casein fragment 1-23 by whole cells. A comparison of CNRZ32 and its prtH deletion mutant indicates that CNRZ32 has at least two cell surface proteinases that differ in substrate specificity.

Role of calcium in activity and stability of the Lactococcus lactis cell envelope proteinase

The mature lactococcal cell envelope proteinase (CEP) consists of an N-terminal subtilisin-like proteinase domain and a large C-terminal extension of unknown function whose far end anchors the molecule in the cell envelope. Different types of CEP can be distinguished on the basis of specificity and amino acid sequence. Removal of weakly bound Ca2+ from the native cell-bound CEP of Lactococcus lactis SK11 (type III specificity) is coupled with a significant reversible decrease in specific activity and a dramatic reversible reduction in thermal stability, as a result of which no activity at 25 degrees C (pH 6.5) can be measured. The consequences of Ca2+ removal are less dramatic for the CEP of strain Wg2 (mixed type I-type III specificity). Autoproteolytic release of CEP from cells concerns this so-called "Ca-free" form only and occurs most efficiently in the case of the Wg2 CEP. The results of a study of the relationship between the Ca2+ concentration and the stability and activity of the cell-bound SK11 CEP at 25 degrees C suggested that binding of at least two Ca2+ ions occurred. Similar studies performed with hybrid CEPs constructed from SK11 and Wg2 wild-type CEPs revealed that the C-terminal extension plays a determinative role with respect to the ultimate distinct Ca2+ dependence of the cell-bound CEP. The results are discussed in terms of predicted Ca2+ binding sites in the subtilisin-like proteinase domain and Ca-triggered structural rearrangements that influence both the conformational stability of the enzyme and the effectiveness of the catalytic site. We argue that distinctive primary folding of the proteinase domain is guided and maintained by the large C-terminal extension.

Subtilases: the superfamily of subtilisin-like serine proteases

Subtilases are members of the clan (or superfamily) of subtilisin-like serine proteases. Over 200 subtilases are presently known, more than 170 of which with their complete amino acid sequence. In this update of our previous overview (Siezen RJ, de Vos WM, Leunissen JAM, Dijkstra BW, 1991, Protein Eng 4:719-731), details of more than 100 new subtilases discovered in the past five years are summarized, and amino acid sequences of their catalytic domains are compared in a multiple sequence alignment. Based on sequence homology, a subdivision into six families is proposed. Highly conserved residues of the catalytic domain are identified, as are large or unusual deletions and insertions. Predictions have been updated for Ca(2+)-binding sites, disulfide bonds, and substrate specificity, based on both sequence alignment and three-dimensional homology modeling.

Isolation and characterization of the hyperthermostable serine protease, pyrolysin, and its gene from the hyperthermophilic archaeon Pyrococcus furiosus

The hyperthermostable serine protease pyrolysin from the hyperthermophilic archaeon Pyrococcus furiosus was purified from membrane fractions. Two proteolytically active fractions were obtained, designated high (HMW) and low (LMW) molecular weight pyrolysin, that showed immunological cross-reaction and identical NH2-terminal sequences in which the third residue could be glycosylated. The HMW pyrolysin showed a subunit mass of 150 kDa after acid denaturation. Incubation of HMW pyrolysin at 95 degrees C resulted in the formation of LMW pyrolysin, probably as a consequence of COOH-terminal autoproteolysis. The 4194-base pair pls gene encoding pyrolysin was isolated and characterized, and its transcription initiation site was identified. The deduced pyrolysin sequence indicated a prepro-enzyme organization, with a 1249-residue mature protein composed of an NH2-terminal catalytic domain with considerable homology to subtilisin-like serine proteases and a COOH-terminal domain that contained most of the 32 possible N-glycosylation sites. The archaeal pyrolysin showed highest homology with eucaryal tripeptidyl peptidases II on the amino acid level but a different cleavage specificity as shown by its endopeptidase activity toward caseins, casein fragments including alphaS1-casein and synthetic peptides.

A new cell surface proteinase: sequencing and analysis of the prtB gene from Lactobacillus delbruekii subsp. bulgaricus

Investigation of the chromosomal region downstream of the lacZ gene from Lactobacillus delbrueckii subsp. bulgaricus revealed the presence of a gene (prtB) encoding a proteinase of 1,946 residues with a predicted molecular mass of 212 kDa. The deduced amino acid sequence showed that PrtB proteinase displays significant homology with the N termini and catalytic domains of lactococcal PrtP cell surface proteinases and is probably synthesized as a preproprotein. However, the presence of a cysteine near the histidine of the PrtB active site suggests that PrtB belongs to the subfamily of cysteine subtilisins. The C-terminal region strongly differs from those of PrtP proteinases by having a high lysine content, an imperfect duplication of 41 residues, and a degenerated sequence compared with the consensus sequence for proteins anchoring in the cell walls of gram-positive bacteria. Finally, the product of the truncated prtM-like gene located immediately upstream of the prtB gene seems too short to be involved in the maturation of PrtB.

Modelling and engineering of enzyme/substrate interactions in subtilisin-like enzymes of unknown 3-dimensional structure

Homology modelling was used to predict enzyme-substrate interactions in three entirely different subtilisin-like enzymes of unknown three-dimensional structure. i.e. (a) cell-envelope proteinase of Lactococcus lactis, (b) putative leader peptidase for pre-nisin from L. lactis, and (c) human furin. Models were based on known three-dimensional structures of subtilisins and thermitase in complex with inhibitors. Detailed analysis of interactions of the P1-P4 residues of model substrates with the S1-S4 binding sites in each enzyme suggest that electrostatic interactions at all four binding sites can contribute to binding and hence to specificity. In particular, one or more negative charges in the S1 or S4 pockets can lead to a high selectivity for Arg residues in the substrate. Many of the predicted interactions have been confirmed by engineering of either enzyme, substrate or both.

Involvement of enzyme-substrate charge interactions in the caseinolytic specificity of lactococcal cell envelope-associated proteinases

Three series of oligopeptides were synthesized to investigate the proposal that a major factor in determining the differences in specificity of the lactococcal cell surface-associated proteinases against caseins is the interactions between charged amino acids in the substrate and in the enzyme. The sequences of the oligopeptides were based on two regions of kappa-casein (residues 98 to 111 and 153 to 169) which show markedly different susceptibilities to PI- and PIII-type lactococcal proteinases. In each series, one oligopeptide had an identical sequence to that of the kappa-casein region, while in the others, one or more charged residues were substituted by an amino acid of opposite charge, i.e., His<-->Glu. Generally, substitution of His by Glu in the oligopeptides corresponding to residues 98 to 111 of kappa-casein resulted in reduced cleavage of susceptible bonds by the PI-type proteinase and increased cleavage of susceptible bonds by the PIII-type proteinase. In the case of the oligopeptide corresponding to residues 153 to 169 of kappa-casein, one major cleavage site was evident, and the bond was hydrolyzed by both types of proteinase (even though this sequence in kappa-casein itself is extremely resistant to the PI-type enzyme). Substitution of Glu by His in this oligopeptide, even in the P7 position, resulted in increased cleavage of the bond by the PI-type proteinase and reduced cleavage by the PIII-type proteinase. C-terminal truncation of this oligopeptide resulted in a 100-fold decrease in the rate of hydrolysis of the susceptible bond and a change in the pattern of cleavage.(ABSTRACT TRUNCATED AT 250 WORDS)

Structural basis of substrate specificity in the serine proteases

Structure-based mutational analysis of serine protease specificity has produced a large database of information useful in addressing biological function and in establishing a basis for targeted design efforts. Critical issues examined include the function of water molecules in providing strength and specificity of binding, the extent to which binding subsites are interdependent, and the roles of polypeptide chain flexibility and distal structural elements in contributing to specificity profiles. The studies also provide a foundation for exploring why specificity modification can be either straightforward or complex, depending on the particular system.

Prevention of C-terminal autoprocessing of Lactococcus lactis SK11 cell-envelope proteinase by engineering of an essential surface loop

The catalytic domain of the cell-envelope proteinase from Lactococcus lactis SK11 has various inserts, situated in external loops of the catalytic domain, compared with the related subtilisins. Protein engineering was employed to analyse the necessity and function of one of these extra loops (residues 205-219), that is predicted to be located in close proximity to the substrate-binding region and is susceptible to autoproteolysis. We constructed a deletion mutant which lacks 14 residues of this surface loop and subsequently introduced various insertion cassettes coding either for the original loop with three mutations (E205S/E218T/M219S: triple-mutant proteinase) or for neutral spacers (1, 4, 7 and 16 serine residues). Engineered proteinases were analysed for activity, (auto)processing, and cleavage specificity. The presence of residues 205-219 is shown to be essential for proteolytic activity, as only triple-mutant proteinase retained activity towards casein substrates. The triple-mutant proteinase was found to be defective in C-terminal autoprocessing, and subsequent release from the lactococcal cell envelope in a calcium-free medium, indicative of either an altered proteolytic specificity or altered accessibility of the processing site. The specificity change appears to be subtle, as only small differences were found between wild-type and triple-mutant proteinase in the breakdown of casein substrates.

The sequence of a subtilisin-type protease (aerolysin) from the hyperthermophilic archaeum Pyrobaculum aerophilum reveals sites important to thermostability

The hyperthermophilic archaeum Pyrobaculum aerophilum grows optimally at 100 degrees C and pH 7.0. Cell homogenates exhibit strong proteolytic activity within a temperature range of 80-130 degrees C. During an analysis of cDNA and genomic sequence tags, a genomic clone was recovered showing strong sequence homology to alkaline subtilisins of Bacillus sp. The total DNA sequence of the gene encoding the protease (named "aerolysin") was determined. Multiple sequence alignment with 15 different serine-type proteases showed greatest homology with subtilisins from gram-positive bacteria rather than archaeal or eukaryal serine proteases. Models of secondary and tertiary structure based on sequence alignments and the tertiary structures of subtilisin Carlsberg, BPN', thermitase, and protease K were generated for P. aerophilum subtilisin. This allowed identification of sites potentially contributing to the thermostability of the protein. One common transition put alanines at the beginning and end of surface alpha-helices. Aspartic acids were found at the N-terminus of several surface helices, possibly increasing stability by interacting with the helix dipole. Several of the substitutions in regions expected to form surface loops were adjacent to each other in the tertiary structure model.

Homology modelling of the catalytic domain of human furin. A model for the eukaryotic subtilisin-like proprotein convertases

A model is presented for the three-dimensional structure of the catalytic domain of the human serine proteinase furin and its interaction with model substrates. This homology model is based on the crystal structures of subtilisin BPN' and thermitase in complex with the inhibitor eglin, and it also applies to other members of the eukaryotic subtilisin-like proprotein convertases. Predictions are made of the general protein fold, inserted loops, disulfide bonds, Ca(2+)-binding sites and salt bridges. A detailed prediction of the substrate-binding region attempts to explain the basis of specificity for multiple basic residues preceding the cleavage site. Specific acidic residues in the S1, S2 and S4 subsites of the substrate-binding region of furin are identified which appear to be of particular importance, while residues of the S2', S3, S5 and S6 subsites may also contribute to substrate binding. Based on this model, protein engineering can be employed not only to test the predicted enzyme-substrate interactions, as demonstrated for human furin, but, equally importantly, to design proprotein convertases with a desired specificity, or to design novel substrates or inhibitors.