Prokaryote-derived protein inhibitors of peptidases: a sketchy occurrence and mostly unknown function

Abstract

In metazoan organisms protein inhibitors of peptidases are important factors essential for regulation of proteolytic activity. In vertebrates genes encoding peptidase inhibitors constitute up to 1% of genes reflecting a need for tight and specific control of proteolysis especially in extracellular body fluids. In stark contrast unicellular organisms, both prokaryotic and eukaryotic consistently contain only few, if any, genes coding for putative peptidase inhibitors. This may seem perplexing in the light of the fact that these organisms produce large numbers of proteases of different catalytic classes with the genes constituting up to 6% of the total gene count with the average being about 3%. Apparently, however, a unicellular life-style is fully compatible with other mechanisms of regulation of proteolysis and does not require protein inhibitors to control their intracellular and extracellular proteolytic activity. So in prokaryotes occurrence of genes encoding different types of peptidase inhibitors is infrequent and often scattered among phylogenetically distinct orders or even phyla of microbiota. Genes encoding proteins homologous to alpha-2-macroglobulin (family I39), serine carboxypeptidase Y inhibitor (family I51), alpha-1-peptidase inhibitor (family I4) and ecotin (family I11) are the most frequently represented in Bacteria. Although several of these gene products were shown to possess inhibitory activity, with an exception of ecotin and staphostatins, the biological function of microbial inhibitors is unclear. In this review we present distribution of protein inhibitors from different families among prokaryotes, describe their mode of action and hypothesize on their role in microbial physiology and interactions with hosts and environment.

I - Overview of mechanisms regulating proteolysis

Regardless of the complexity of the organism, peptidases in general are essential at every stage in the life of every individual cell. Apart from non-specific protein degradation, peptidases are involved in highly sophisticated essential biological processes, both in prokaryotic and eukaryotic cells. Therefore, keeping in mind that proteolysis in physiological conditions is an irreversible reaction, the need for precise control of peptidases is obvious. Indeed, the proteolysis regulation occurs at multiple levels, including protease gene transcription and translation, and, most importantly, at the protein level. Some of protein level systems operate in all domains of life including prokaryotes, unicellular eukaryotes, plants and metazoans. This includes production of peptidases as zymogenic proenzymes, which get activated after reaching a designated location, which, in case of prokaryotes, is the periplasm or extracellular environment. In prokaryotes secretion of active peptidases can be compared to compartmentalization of proteolytic activity. Other peptidases require cofactors for activity, or have very unique specificities, or active sites are confined in multimeric superstructures of which the proteasome is the best example. Finally, the proteolytic activity is regulated by inhibitors of peptidases. This mean of regulation of proteolysis is common in metazoa, essential for homeostasis in vertebrates but, with few exceptions, basically absent in prokaryotes and unicellular eukaryotes. Nevertheless, as described in the following sections, some Bacteria and Archaea have acquired genes encoding different types of peptidase inhibitors.

II - Distribution of known families of protein inhibitors of peptidases among prokaryotes

Based on the evolutionary and structural relationship among protein inhibitors of peptidases, inferred from the comparison of tertiary structure and/or amino acid sequence all modern-day inhibitors are grouped in clans and families. Clans comprise one or more families of inhibitors which have arisen from a single evolutionary origin, the evidence being the similarity of their tertiary structures. This method, introduced by Rawlings and colleagues [1] and currently implemented in the MEROPS data server (www.merops.ac.uk), is a powerful tool, allowing the logical classification of peptidase inhibitors.

Out of 67 families of proteinase inhibitors listed in the MEROPS database (updated on Jan. 25, 2010) only 18 families were recognized in prokaryotes (Table 1). Significantly, with some remarkable exceptions, the occurrence of individual types of inhibitors is limited to few bacterial species scattered among phylogenetically distinct orders or even phyla of microbiota. The most abundant peptidase inhibitors in prokaryotic cells are homologous to alpha-2-macroglobulin (family I39), serine carboxypeptidase inhibitor (family I51), alpha-1-peptidase inhibitor (family I4) and ecotin (family I11) but even the genes encoding homologues of alpha-2-macroglobulin are found in only 26% of prokaryotic species with fully sequenced and analyzed genomes.

Table 1.

Distribution of protein inhibitors of proteases belonging to different clans and families among prokaryotes

Clan	Family	Type of Inhibitor	# of homologous genes
			Bacteria (out of 662 fully sequenced genomes)	Archaea (out of 55 fully sequenced genomes)
IA	I1	Ovomucoid	11	1
IB	I2	Aprotinin	8	0
ID	I4	Alpha-1-peptidase inhibitor	88	17
JC	I9	Peptidase B inhibitor	6	0
I-	I10	Marinostatin	10	0
IN	I11	Ecotin	76	0
IY	I16	Streptomyces subtilisin inhibitor	43	0
IX	I31	Equistatin inhibitor	1	0
IU	I36	Streptomyces metallopeptidases inhibitor	1	0
IK	I38	Metallopeptidases inhibitor	17	0
IL	I39	Alpha-2-macroglobulin	184	1
I-	I42	Chagasin	35	13
JE	I51	Serine carboxypeptidase Y inhibitor	138	19
IK	I57	Staphostatin B	2	0
IK	I58	Staphostatin A	2	0
JB	I63	Pro-eosinophil major basic protein	4	0
I-	I75	Bacteriophage lambda CIII protein	2	0
I-	I78	Aspergillus elastase inhibitor	33	0

III - Characterization of prokaryote-derived protein inhibitors of peptidases

A – Ovomucoid family (family I1)

Family I1 is frequently referred to as the Kazal family of inhibitors which are abundant in animals including invertebrates but fairly infrequent in protozoa and plants and absent in fungi and viruses. They function as inhibitors of serine proteases from both chymotrypsin (S1) and subtilisin (S8) families of serine peptidases. Despite the common occurrence of these types of peptidases among prokaryotes, genes encoding homologues of family I1 inhibitors were found only in eleven bacterial species in the Proteobacteria phylum. Interestingly in Sorangium cellulosum, a bacterium playing an important role in soil ecology due to its ability to degrade cellulose, four Kunitz family inhibitor homologues were found but their function remains unknown [2].

B – Aprotinin family (family I2)

Numerous inhibitors belonging to this family play important roles in regulating serine peptidase activities in animals. Strikingly, only a few homologous were found from other kingdoms of life. In Bacteria genes encoding homologous proteins were identified only in 8 species scattered in 3 different phyla, including Bacteroidetes, Cyanobacteria and Proteobacteria, and nothing is known with respect to their function.

C – Alpha-1-peptidase inhibitor (family I4)

Members of the alpha-1-peptidase inhibitor (alpha-1-antitrypsin) family, also known as the serpin (serine proteinase inhibitor) superfamily, until fairly recently have been known only to be present in higher multicellular eukaryotes (plants and animals) and viruses [3]. The advent of genome sequencing and improvement of database searching methods totally changed the perception that serpins are absent in unicellular microorganisms, including prokaryota, protozoa and fungi [4].

The first twelve serpin-like sequences in the genome of prokaryotic organisms were identified in 2002 by James Whisstock and colleagues [5]. In this seminal paper it was predicted that bacterial and archaeal serpins are functional peptidase inhibitors. Interestingly, many serpins are found in thermophilic or even hyperthermophilic organisms living at temperature incompatible with a metastable fold of native inhibitory serpins. Of note, serpins utilize a conformational switch to inhibit target peptidases of both serine (families S1 and S8) and cysteine (family C1) catalytic type. The price paid for this conformational flexibility is thermolability of serpins, therefore it was of great interest to have an insight into function and structure of a serpin from the thermophilic bacterium Thermobifida fusca. The protein referred to as thermopin was shown to inhibit chymotrypsin by formation of a covalent complex and to have a high degree of stability at 60°C, a temperature at which alpha-1-peptidase inhibitor loses the activity within minutes. The increased stability was attributed to a C-terminal extension of thermopin interacting with highly conserved residues important for serpin stability (Fig. 1) [6]. Furthermore, comparison of a high resolution crystal structure of native and reactive-site-loop-cleaved thermopin revealed that the native state of thermopin is relatively flexible and loosely packed and the transition to the cleaved conformation results in a dramatic increase of salt bridges. The difference in stability is considered as a driving force critical for rapid conformational change and inhibitory function of thermostable serpins [7]. That such a mechanism functions in serpins from hyperthermophiles such as the crenarchaeon Pyrobaculum aerophilum, which lives in boiling marine water, still needs to be verified experimentally.

Figure 1. Family I4 – serpin inhibitors. Structures of two forms of thermostable Thermoanaerobacter tengcongensis serpin.

The native, active form (blue) with N-terminal fragment (green) and latent, inactive form (light gray). The RSL sequence is marked red in both structures. In the latent form, the RSL region is inserted into the β-sheet formed by four antiparallel β-strands. The bottom panel shows detailed image of N-terminal fragment filling the hydrophobic pocket formed by Leu 159, Ile 162 and Ile170, which interaction supports the native conformation and prevents spontaneous switch to the latent form. The structures were obtained from RCSB database (www.pdb.org) with accession codes 2PEE and 2PEF for native and latent form, respectively [129].

Analysis of prokaryotic genomes available in the MEROPS database indicates that serpins are the third most abundant family of protease inhibitors in microbiota. They are found in 17 out of 55 and in 88 out of 662 fully sequenced genomes of Archaea and Bacteria, respectively. That means serpins are far more prevalent in the former superkingdom (31% of total gene count) than the latter (13%). Serpins are scattered among three (Korarchaeota, Crenarchaeota, Euryarchaeota) out of five evolutionary lineages of Archaea and in twelve phyla (out of 28) of Bacteria, including Acidobacteria, Aquificae, Bacteroidetes, Chlamydiae, Chloroflexi, Cyanobacteria, Dictyoglomi, Gemmatimonadetes, Firmicutes, Proteobacteria, Thermotogae, and Verrucomicrobia. Interestingly, with the exception of three species of the Bacteroidetes phylum (Bacteroides ovatus, Bacteroides uniformis and Prevotella copi) which are human commensal bacteria, serpins are almost exclusively present in benign environmental microbiota.

The sporadic distribution of serpin genes in prokaryotes suggests that serpins may have evolved first within eukaryotes and then on several occasions, the ancestral gene has been passed by lateral transfer to prokaryotes. Alternatively, serpins have an ancient origin preceding the time when the two major lineages of life, Bacteria and Archaea diverged. In Bacteria serpins were mostly lost but persisted in Archaea and were passed to eukaryotes, where they diversified to fulfill a variety of different functions as essential regulators of intracellular and extracellular proteolysis in metazoan organisms. Finally, it is possible that serpins emerged in Archaea, were passed to eukaryotes and then laterally transferred to Bacteria. It is likely that detailed analysis of serpin sequences from organisms living at the boundaries of life, when they become available in large numbers, will resolve the mystery of the evolutionary origin of serpins.

Sporadic distribution of serpins in prokaryotes and almost non-existing information about their expression and inhibitory specificity makes it difficult to predict their physiological function. Taking into account the versatile environment and metabolism of bacterial species bearing serpin genes it is apparent that they must have evolved to perform very specialized functions. So far there is only one report supporting this contention. Clostridium thermocellum, the Firmicutes bacterium capable of degrading cellulose, produces a multiprotein superstructure referred to as a cellulosome which is responsible for glycoside hydrolase assembly, bacterial adhesion to cellulose fibers and polysaccharide hydrolysis. Cellulosomal proteins are directed to their destination by a sequence motif called dockerin. In an elegant work Mark Morrison and colleagues [8] have shown that two serpins expressed by C. thermocellum bearing dockerin modules are secreted and inserted into the cellulosome structure. Furthermore, they have shown that not only recombinant serpin, but also the entire cellulosome inhibits bacterial subtilisin but not mammalian or plant proteases. Serpins bearing the dockerin motif are also present in Ruminococcus albus and Thermobifida fusca, another Gram-positive cellulose-degrading bacteria. Taking into account that these three bacteria possess genes encoding serine proteases it can be justifiable claimed that serpins in the Gram-positive cellulolytic microbes play a physiological role in regulation of proteolysis and protection of proteins in the cellulosome structure from proteolytic damage. Interestingly, Sorangium cellulosum, a gram-negative cellulolytic bacterium (Class: Deltaproteobacteria) possesses seven putative serpin genes. Taken together it is tempting to speculate that serpins are somehow important in protection of the cellulose-degrading apparatus from proteolysis. Nevertheless, the physiological role of serpins in prokaryotes is largely unknown.

D – Peptidase B inhibitor family (family I9)

Inhibitors of this family inhibit serine peptidases of the subtilisin family (S8) and occur exclusively in fungi and Bacteria. In Bacteria only nine proteins belonging to family I9 have been identified to date. They are found in just six species from genus Bacillus (seven sequences) and Geobacillus (one sequence) of the Bacillaceae family of Firmicutes. The I9 inhibitors are compact alpha/beta proteins with a fold unlike that of any other protein [9; 10]. Interestingly, while fungus-derived inhibitors are encoded as separate polypeptides chains being inhibitory molecules on their own, prokaryotic versions of the inhibitors are the propeptides of subtilisin-like protease precursors. During enzyme maturation they assist in the folding of the enzyme working as intramolecular chaperones [11]. The released profragment retains the function of a chaperone. It binds the protease through a large interface involving subtilisin’s two parallel surface helices and insertion of the C-terminus of the propeptide into the enzyme active site in a product-like manner (Fig. 2) [12]. Although initially the complex is very stable, the potency of inhibition is reduced following cleavage of the inhibitor by bound subtilisin leading to release of the free enzyme.

Figure 2. Family I9 – complex of subtilysin propeptide (green) with inactive mutant of subtilisin BPN (red).

The right panels present interaction of enzyme with propeptide in the distal region (top) and in the active site cleft (bottom). Panels on the right present detailed view on the C-terminal of inhibitor filling the active cleft of the enzyme. Catalytic triad and interacting residues of inhibitor are labelled. The bottom right panel shows inhibitor Tyr 77 and Ala 76 interacting with P1 and P1’ enzyme subsites. The surface is coloured by electrostatic potential (blue – positive values, red negative). The structure was obtained from the www.pdb.org with accession code 1SPB [130].

Subtilisin-like proteases (S8) are widespread among prokaryotic species and the vast majority of these enzymes are produced with a profragment. Nonetheless, none of them share any significant sequence similarity with I9 inhibitors. Do other profragments of S8 family proteases serve as protease inhibitors? There is currently no answer to this question.

E – Marinostatin family (family I10)

The occurrence of marinostatin family inhibitors is limited to ten species of marine organisms from three different phyla of Bacteria. Altogether 24 gene-encoded sequences have been identified. The inhibitor is produced as a precursor polypeptide of 45 to 100 residues with inhibitory domain(s) located at the C-terminus. During maturation the C-terminally located 11- or 12-residue inhibitory peptides are released by proteolytic cleavage and become post-translationally modified by two ester linkages between Thr3 - Asp9 and Ser8 - Asp11. The ester bonds stabilize the peptide structure and are essential for the inhibitory activity [13–15]. Therefore it is no surprise that the amino acid residues forming the ester bonds are conserved in all marinostatin-like molecules (Fig. 3).

Figure 3. I10 family – marinostatin, a serine protease inhibitor from Alteromonas sp..

A schematic representation of inhibitor structure, dashed lines denote ester bonds (A). The short aminoacid chain of inhibitor is stabilised by two ester bonds (marked as violet). P1 Met4 and P1’ Arg5 are labelled (B). Structure was obtained from www.pdb.org with accession code 1IXU [131].

Despite its small size marinostatin is a very potent inhibitor of subtilisin BPN’ (Ki = 10⁻⁹ M) and also inhibits chymotrypsin and pancreatic elastase but does not affect the activity of trypsin [16–19]. This selectivity is in agreement with the presence of Met in the P1 position. Interestingly, several marinostatin-like inhibitors apparently present Arg or Lys at the P1 site (Fig. 3) and most likely they are effective inhibitors of serine proteases with trypsin-like activity. The biological role of marinostatin and marinostatin-like peptides as protease inhibitors is unknown but pharmaceutical and industrial application of these molecules has been suggested [20].

F – Ecotin family (family I11)

The occurrence of genes encoding inhibitors belonging to the ecotin family is limited to Bacteria and Protozoa. In Bacteria, apart from few exceptions (Acidobacteria, Cyanobacteria and Planctomycetes) ecotin-like proteins are found exclusively in different species from the Proteobacteria phylum, especially the gamma subdivision. Although they are particularly abundant among species of Burkholderia, Rickettsia, and Shewanella there are always closely related species which are missing the ecotin-like gene. The only exception is genus Yersinia where ecotin is found in all the genomes sequenced so far.

Ecotin was isolated from Escherichia coli cell extracts and characterized as a homodimeric protein of 38 kDa resistant to boiling and stable at pH 1.0 which potently inhibited trypsin, chymotrypsin, rat mast cell chymase and pancreatic elastase. Clearly, the inhibitory spectrum of ecotin is limited to serine proteases of the trypsin/chymotrypsin fold (S1) but, nevertheless, it is astonishingly broad potently targeting such diverse enzymes as factor Xa [21], human neutrophil elastase [22], kallikrein, urokinase, factor XIIa [23], fiddler crab collagenase, and granzyme B [24]. This pan-specificity of inhibition is due to formation of a heterotetrameric complex with target proteases. In a complex each ecotin monomer in the homodimer binds a protease via primary and secondary interactions [25–27]. The primary binding site, a protruding surface loop, interacts with the active cleft of a protease in a substrate-like manner typical for canonical inhibitors using the Laskowski mechanism of inhibition. The secondary binding site is smaller in size and interacts in a unique way with a protease region distant from the active site (Fig. 4). This feature together with the dimerization of ecotin acting as a hinge allows the binding of different S1 peptidases [28–30].

Figure 4. Family I11 - The 3D structure of ecotin, E. coli in two forms, free (red), and from the complex with chymotrypsin (blue).

The rectangles mark the primary (top) and secondary (bottom) trypsin-interaction sites. The panels on the right present reoriented zoom-in on respective loop. Green aminoacid side chains are derived from complexed ecotin, while yellow are from the free form.

Notable differences between both forms can be observed in the primary interaction region. Both, backbone and side chains are in altered conformations, indicating forced conformation of the loop empowered by interaction with enzyme active site. The P1 residue Met 84 is labeled. The secondary interaction site does show only minor alterations. Structures were obtained from the RCSB database (www.pdb.org) with accesion codes 1N8O [132] (chymotrypsin-ecotin complex) and 1ECY (free ecotin) [133].

The unique architecture and inhibition mechanism exerted by ecotin make this protein an attractive target for protein engineering. A point mutation exchanging the P1 Met residue for Arg (M85R) made ecotin an efficient inhibitor of thrombin and factor XI [31; 32]. Modified in this way ecotin was found useful in the isolation and characterization of proteins with a trypsin/chymotrypsin fold [33; 34]. On the other hand a phage display approach [35–37] and sophisticated combination of synthetic DNA shuffling, fragmentation-based DNA shuffling and phage display allowed for selection of ecotin with increased specificity for distinct serine proteases [38; 39]. Such ingeniously engineered inhibitors potentially have a very broad spectrum of biomedical applications to control physiologically essential processes involving serine proteases including among others, blood clotting, fibrinolysis and contact activation.

In E. coli and probably in all other bacteria expressing ecotin the protein is translocated into the periplasmatic space. Based on the broad inhibitory specificity against digestive tract serine proteases and lack of inhibition of bacterial endogenous enzymes, it was suggested that ecotin plays a role protecting the bacterium from external proteolytic attack in the mammalian gut [40]. This still may hold true for ecotin expressed by other species of Enterobacteriaceae which are exposed to pancreatic proteases but does not explain ecotin function in many other species, including human pathogens such as Pseudomonas aeruginosa, and Burkholderia, Rickettsia and Yersinia species. Therefore it was hypothesized that ecotin may yield protection against host proteases abundant in neutrophils, the immune cells which constitute the first line of defense against bacterial invasion. This hypothesis was supported by the finding that ecotin protects E. coli from the bactericidal activity exerted by neutrophil elastase [41]. Elastase permeabilizes the outer membrane by degrading OmpA [42] and probably enters the periplasmic space where its activity interferes with damage repair leading to loss of cell viability. It was concluded that inhibition of this intraperiplasmic elastase activity by ecotin facilitated E. coli recovery from elastase insult in concordance with the experimental model. A similar mechanism was postulated for other pathogens in which killing by neutrophils depends on serine proteases. Although many of these are not directly killed by neutrophil proteases, it is likely that in phagosomes outer membrane damage by antibacterial peptides permits the proteases entry into the periplasmic space where their activity interferes with bacterial vital processes. Recently, this hypothesis was strongly supported by an elegant study showing that ecotin orthologues are important for the early stages of infection of the mammalian host by the proteozoan parasite Leishmania major [43]. Apparently the parasite ecotins target macrophage-derived serine proteases. This explains the regulated expression of ecotin genes during the life cycle of L. major, because the organism lacks genes encoding serine proteases of the chymotrypsin/trypsin fold (family S1A).

Apart from defense against bactericidal activity of the host, ecotin may have other functions since the genes encoding orthologous proteins are also present in plant pathogens and symbionts, as well as in benign environmental bacterial species. Most likely, inhibitory activity of ecotin is not directed against the bacterium’s own proteases since many of species bearing the ecotin gene do not express proteases of the chymotrypsin/trypsin fold. From an evolutionary point of view it is likely that the ecotin gene appeared in the Protobacteria lineage and was transferred to other phyla by occasional lateral gene transfer. In these new hosts ecotin may have no function. Interestingly, the genus Rickettsia appeared to be in the process of removing the gene because several Rickettsia sp. contain an ecotin pseudogene. As obligatory intracellular pathogens these bacteria are not exposed to the proteolytic enzymes produced by host immune cells.

G – Streptomyces subtilisin inhibitor family (family I16)

A Streptomyces subtilisin inhibitor (SSI) was first described from Streptomyces albogiseolus by Murao & Sato [44] and up to data several inhibitors from many different species of the Streptomyces genus have been isolated and characterized. All seemingly functional inhibitors are secretory proteins with a mature inhibitor molecule encompassing circa 110 residues and tendency to form a homodimmer. Characterized SSI proteins are potent inhibitors of serine proteases of the subtilisin (family S8) and chymotrypsin/trypsin (family S1) fold, including bacterial, fungal and eukaryotic subtilisins (kexins) and trypsin, plasmin and SAM-P26, a chymotrypsin-like endogenous protease from Streptomyces [45–48]. In addition, some SSIs strongly inhibit grisilysin, the Streptomyces griseus enzyme belonging to family M4 of metallopeptidases but none inhibits thermolysin. The inhibitory interaction occurs via the same reactive site as for subtilisin inhibition [49]. Also the SSI-like inhibitor from S. caespitosus, as well as subtilisin, strongly inhibit the endogenous metalloprotease (ScNP) belonging to the Metzincin family [50]. The homodimeric double-headed inhibitor formed a stable complex with 2:2:2 stoichiometry with ScNP and subtilisin BPN’. In this case, however, the interaction site with ScNP was mapped to a protruding loop located on the pole opposite to the subtilisin reactive site where the conserved Pro residue was substituted with Tyr.

SSI-like proteins inhibit targeted enzymes by inserting the reactive site loop in a substrate-like manner into the reactive site cleft of the protease. As in other inhibitors obeying the standard mechanism of inhibition [51] the P1 residue is the main determinant of SSI specificity (Fig. 5). Interestingly, the reactive site loop shows little variation with Lys, Arg and Met prevailing at the P1 site. The analysis of phylogenetic trees suggest that SSI-like proteins with the latter two residues arose multiple times on independent lineages from ancestral proteins possessing Lys at the P1 site [52]. The limited variation of the P1 residue argues that a similar set of proteases in different bacterial species is targeted by SII inhibitors. Of note, the amino acid sequence and protein fold around the reactive site bond are similar to those seen in the Kazal family (I1) [53; 54] but the different general folds of these proteins argues against a common evolutionary origin of these two families.

Figure 5. Family I16 - Streptomyces Subtilisin Inhibitor (red) in the complex with subtilisin BPN (blue).

The left panel presents general view on the complex, where only limited regions of both molecules are interacting. The figure shows a monomer of inhibitor. SSI forms a dimer by interaction in the top β-sheet region, inhibiting two molecules of enzyme (E₂I₂). Right panel presents detailed view on the interaction region in the enzyme active site cleft. Inhibitor residues are in red and subtilisin sidechains are coloured green. Disulphide bridge stabilising the RSL of inhibitor is marked yellow. Interacting residues and P1 Met 73 are labeled. The structure was obtained from www.pdb.org with accesion code 2SIC [134].

In contrast to other families of bacterial protease inhibitors where homologues are scattered among different phylogenetic lineages and are of mostly obscure biological function, distribution of SSI proteins is strictly limited to the Actinomycetales order of the Actinobacteria phylum and their physiological role is fairly well defined. Based on an analysis of phenotype of mutant strains devoid of SSI or its target protease, Taguchi and his colleagues hypothesized that SSI-like proteins play an important role(s) in physiological and/or morphological regulation in Streptomycetes [55; 56]. The hypothesis was fully corroborated by recent findings [57] and is in keeping with the widespread occurrence of SSI in Streptomyces species. The SSI coding gene is present in 31 out of the 33 species with fully sequenced genomes. Apparently, one way of wielding the physiological effect by SSI is the control of the proteolytic activation of transglutaminase, an extracellular enzyme that cross-links proteins to high molecular weight aggregates, by endogenous proteases [58–60].

H – Equistatin (thyropin) family of cysteine proteases (family I31)

Equistatin is an inhibitor isolated from the sea anemone Actinia equine [61]. With exception of a putative protein encoded by locus CBU0898 of Coxiella burnetii equistatin-like inhibitors are limited to eukaryotic organisms. Since the inhibitor shares similarity with thyroglobulin type-1 repeats the family is also referred to as thyropins [62]. Thyropins primarily inhibit papain-like cysteine peptidases (family C1) but inhibition of the aspartic protease cathepsin D and metallopeptidases has also been described. C. burnetii, an obligatory intracellular pathogen, presumably acquired the gene by lateral transfer from its eukaryotic host. The bioinformatics analysis suggests that the product of the CBU0898 locus may have an inhibitory activity. Unfortunately, nothing is known about the biological role of this potential inhibitor in the life cycle and the pathogenicity of this etiologic agent of Q fever that is considered a potential biological weapon [63].

I – Streptomyces metallopeptidases inhibitor family (family I36)

This family is represented by a single protein isolated in 1979 from Streptomyces nigrescens TK-23 and named as streptomyces metalloprotease inhibitor (SMPI) [64]. Up to now no orthologous proteins have been identified. SMPI contains 102 residues and two small disulfide loops of seven and six residues [65]. The SMPI polypeptide chain is folded into two beta-sheets, each consisting of four antiparallel beta-strands [66].

SMPI inhibits metalloproteases from family M4 (gluzincins) forming most stable complexes with Pseudomonas aeruginosa elastase and griselysin (Ki approximately 2 pM) and then with thermolysin (Ki = 1.14 × 10⁻¹⁰ M) [67; 68]. The reactive site was identified as the Cys64-Val65 peptide bond which is on a protruding surface loop. The substitution of the P1’ Val residue with residues which are not consistent with the thermolysin substrate specificity like Ile, Leu, Phe or Tyr resulted with weakening of SMPI inhibitory activity and rapid protein degradation [69].

Interestingly, unlike other proteinaceous inhibitors of metallopeptidases, the interaction of SMPI follows the standard mechanism of inhibition, including a tetrahedral reaction intermediate and the resynthesis of the hydrolyzed Cys64-Val65 reactive site peptide [70; 71] (Fig. 6), as described for inhibitors of serine peptidases [51]. From modeling studies and molecular dynamics simulation of the complex formation it is apparent that the ridged active site loop of SMPI is complementary in shape and surface charge distribution to the active site cleft of thermolysin [72]. The complex stability is further enforced by contact sites outside the loop [73; 74].

Figure 6. Family I36 - Streptomyces nigrescens metalloprotease inhibitor.

The structure is folded in the 2 “greek key” motifs with additional loop expanding from the barrel structure, containing the inhibitor activity of the molecule. P1 Cys 64 and P1’ Val 65 are marked orange. Two disulphide bridges stabilising the structure are marked red. The atom coordinates were obtained from www.pdb.org with accession code 1BHU [135].

J – Metallopeptidases inhibitor family (family I38)

The first member of the I38 family was isolated from the periplasm of Erwinia chrysanthemi [75]. The family encompasses relatively small proteins (ca. 11 kDa) that tightly inhibit metallopeptidases (K_d in picomolar range) in subfamily M10B of metzincins (family M10), including serralysin of Serratia marcescens, aeruginolysin (alkaline protease) of Pseudomonas aeruginosa and extracellular metalloproteases of Erwinia chrysanthemi; [76–78]. Genes encoding the aprin (alkaline protease inhibitor)-like proteins are found exclusively in 17 different species in Enterobacteriaceae and Pseudomonadaceae of the gamma division of Proteobacteria. Since they are located in the periplasmic space it is suggested that they protect bacteria from adventitious proteolysis during secretion of serralysin-like proteases. This conclusion is supported by the grouping of the protease and its inhibitor in one transcriptional unit. Similar gene organization is found in the case of staphopain and staphostatin, cysteine proteases and their inhibitors in staphylococci (see section N). Furthermore, at least in Pseudomonadaceae, the presence of the aprin encoding gene(s) coincides with the presence of gene(s) coding for secretory serralysin-like protease(s).

Aprin is folded into an eight-stranded, disulfide-stabilized, antiparallel beta-barrel with the reactive site located at the N-terminal strand linked to the barrel by a single-turn alpha-helix [79]. Interestingly, the aprin fold resembles the structure of staphostatins and therefore these families of inhibitors are included into the same clan, clan IK of proteinase inhibitors. In the free molecule the N-terminal strand is apparently folded back on the core of the inhibitor [80] (Fig. 7). It is therefore likely that interaction with the target protease displaces this strand which like a trunk inserts into the reactive site cleft to occupy the extended substrate binding site (S’-sites) of the enzyme. The precise length of the trunk allows the N-terminal residue to chelate the zinc via the NH₂ group and the carbonyl oxygen. Interaction of the trunk with S’-sites of the enzyme almost solely contributes to affinity of binding with the beta-barrel body of the inhibitor preventing the N-terminal trunk from being inserted too far into the active site cleft and being cleaved. In keeping with this model truncation of the N-terminus results in precipitous reduction of the binding affinity [81]. Interestingly, a similar mode of inhibition has been observed between matrix metalloproteases (family M10A) and their cognate inhibitors, TIMPs. Nevertheless, aprin does not inhibit MMPs, nor do TIMPs have any inhibitory effect on proteases targeted by aprins. Furthermore, metallopeptidases other than those from the M10B subfamily are not inhibited by aprins and the inhibitor is a substrate for pseudolysin (elastase B, LasB) (family M4) from P. aeruginosa, which cleaves off the N-terminal pentapeptide [82]. Taken together these data suggest that aprins have evolved to specifically inhibit bacterial metallopeptidases from M10B family.

Figure 7. Family I38 - Serratia marcescens metalloprotease (deep blue) in complex with Erwinia chrysanthemi inhibitor (green).

General view on the structure (left panel) and detailed zoom-in on the interaction sites (right panel). The interacting locations in inhibitor are highlighted as red (N-terminal tail and 3-strand β-sheet). The hydrogen bond between inhibitor Ser 2 and enzyme Glu 177 is marked as green spheres and respective residues are labelled. The enzyme α-helix interacting with inhibitor β-sheet is marked as steel blue. The structure was obtained from www.pdb.org with accesion code 1SMP [136].

K – Alpha-2-macroglobulin family (family I39)

This family encompasses large homologous proteins, occurring as monomers, homodimers or homotetramers. Working as “molecular traps” alpha-macroglobulins are universal inhibitors of endopeptidases regardless of the catalytic class or family. They offer proteases a so-called bait region for a proteolytic attack. Proteolysis of any peptide bond in this region triggers change in the alpha-macroglobulin structure leading to protease “entrapment”. In tetrameric alpha-macroglobulins an encaged protease can still cleave small peptides but not larger molecules which cannot penetrate into the cage [83].

Alpha-macroglobulins are present on all metazoan organisms but are absent in unicellular eukaryotes (Protozoa and Fungi) and are considered to play an important part in the system for defence against bacterial infections [84]. Surprisingly, members of the alpha-2-macroglobulin family are the most abundant inhibitors in Bacteria, but not in Archaea. In the latter superkingdom of organisms only one homologue gene was identified in the Methanococcoides burtonii genome. In Bacteria phylogenetic distribution of alpha-macroglobulin homologous sequences is patchy and they are found in 14 out of 29 recognized evolutionary lines (phyla), including such diverse clades as Acidobacteria, Actinobacteria, Bacteroidetes, Chloroflexi, Cyanobacteria, Firmicutes Fusobacteria, Gemmatimonadetes, Planctomycetes, Proteobacteria, Spirochaetes, Synergistetes, Thermotogae, Verrucomicrobia, and Xenobacteria. Nevertheless, genes encoding alpha-macroglobulin-like proteins are most abundant in Proteobacteria, where they occur in representatives of five evolutionary divisions (alpha-, beta-, gamma-, delta-, and epsilon-proteobacteria). Bacteroidetes is the second phylum rich in species bearing alpha-2-macroglobulin homologues. Interestingly, some species in the Bacteroidetes phylum carry many copies (up to 56 in Pedobacter heparinus) of putative alpha-macroglobulin genes.

In general, no correlation can be found between the life style of bacteria and the presence of alpha-2-macroglobulin homologues, which are found with the same frequency in free-living autotrophic species, as in saprophytic environmental species, commensals or pathogens of metazoan organisms. Furthermore, macroglobulin genes are also present in genomes of early-branching, evolutionary ancient species of hyperthermophilic chemolitotrophic Thermotogae, anoxygenic Chloroflexi and oxygenic phototrophic Cyanobacteria phyla. This observation undermines the finding (based on the analysis of limited numbers of fully sequenced bacterial genomes available in 2003) that the alpha-macroglobulin genes found among bacterial species which colonize metazoan hosts originated from several horizontal gene transfers [85]. Also the assumption that alpha-macroglobulin was acquired by any bacterium by horizontal gene transfer from a metazoan host may require a revision. Taking into account of the wide spread of alpha-macroglobulin homologues among deep-branching bacterial species it is tempting to suggest a different scenario. In our opinion it is very likely that the ancestral alpha-macroglobulin gene predates the divergence of ancient Bacteria species. During evolution it was lost from some bacterial lineages but at one or more times it was laterally transferred to a primordial metazoan species. If our hypothesis is confirmed by rigorous bioinformatics analysis, it would indicate that the key complement proteins, C3, C4, and C5, closely related to alpha₂M, may also have they origin in bacteria before evolving into a potent antimicrobial system in metazoan animals [86].

In many bacteria the alpha-macroglobulin gene is found in tandem with a gene for an enzyme involved in a cell wall peptidoglycan repair. This correlation supported by predicted periplasmic localization of alpha-macroglobulin led Budd and colleagues [85] to suggest that alpha-macroglobulin plays a role in bacteria protection from proteolytic enzymes which get access to the periplasmic space via an outer membrane compromised by the attack of complement complex and/or cationic antibacterial peptides and proteins. The recent study showing that E. coli alpha-macroglobulin is, indeed, the periplasmic protein which gets cleaved within a bait region and efficiently inhibits human neutrophil elastase lends credit to this attractive hypothesis [87]. In this respect, alpha-macroglobulin function resembles the role of ecotin-like inhibitors. However, in view of the alpha-macroglobulin occurrence in many free-living bacteria this hypothesis needs to be expanded. To this end it can be easily envisioned that the physiological role of bacterial alpha-macroglobulin is general protection of the periplasmic compartment from damage by the bacterium’s own proteases.

L – Chagasin family (family I42)

Chagasin was discovered as a protein form Trypanosoma cruzi, the causative agent of Chagas” heart disease [88] and then characterized from other protozoa [89]. Now, it is apparent that the occurrence of the chagasin family proteins is limited to protozoa and prokaryotes, including both Bacteria and Archaea. Chagasin-like inhibitors target exclusively protozoan and mammalian family C1 (papain-like) cysteine proteases [90–92]. In parasites chagasins regulate the functions of endogenous cysteine proteases which are essential for the life cycles of the protozoan pathogens [93; 94]. With respect to the structure chagasin has a unique variant of the immunoglobulin fold with homology to human CD8a and inhibits target proteases in a manner reminiscent to cystatins [95; 96].

In Bacteria genes encoding chagasin-like proteins are scattered among six divergent evolutionary clades and found both in free-living organisms and pathogens. In Archaea they are limited to methanotrophic organisms. Sequence analysis suggests that majority of bacterial chagasins are secretory proteins. In some cases (e.g. Pseudomonas aeruginosa and Dehalococcoides sp.), the presence of a “lipobox” consensus sequence of LAG-C at the junction between the signal peptide and the N-terminal cysteine of a mature protein suggests that some chagasins are subject to post-translational modification and are inserted into the outer membrane of gram-negative bacteria as lipoproteins [97]. Finally, sequence alignment suggests that prokaryotic chagasins should also adopt folds similar to those of their protozoan orthologues and may have inhibitory activity [98; 99]. Unfortunately, no prokaryotic chagasin has been characterized on the protein level and their function can be only the matter of speculation. In prokaryotic species in which chagasin coexist with papain-like proteases (Bacillus cereus, Chlorobium phaeobacteroides, Clostridium acetobutylicum, Clostridium cellulolyticum, Dickeya dadantii, Klebsiella pneumoniae, Legionalla pneumophila, Microcystis aeruginosa, Pseudomonas syringae, Spirosoma linguale and several Archaea species) the inhibitor may protect microbial cell proteins from degradation by its own enzyme. Conversely, in human pathogens such as Coxiella burnetii, Klebsiella pneumoniae, Legionalla pneumophila and P. aeruginosa chagasin may a have a role in protecting the pathogen from host cathepsins in a manner analogous to that suggested for ecotin and microbial alpha-macroglobulin.

M – Serine carboxypeptidase Y inhibitor family (family I51)

This family encompasses inhibitors of serine carboxypeptidases, a large group of exopeptidases present in all forms of live. The founder of the family, a protein inhibiting yeast carboxypeptidase Y (CPY) was isolated from yeast in 1974 [100] (Fig. 8). CPY shares a high degree of sequence similarity to phosphatidylethanolamine-binding proteins (PEBP) that inhibit various kinases [101]. Therefore numerous, evolutionary diverse sequences indentified in Bacteria and Archaea species as homologues to CPY may likely represent PEBPs. Recently, it was shown that archaeal BEPB from Sulfolobus solfataricus inhibits bovine alpha-chymotrypsin and elastase with K_i values of 80 and 100 nM, respectively [102]. This suggests that prokaryotic CPY/BEPB-like proteins, in addition to inhibition of carboxypeptidases may also efficiently control serine proteases of S1 family. Nevertheless, the biological functions of CPY/BEPB-like proteins in prokaryotes are obscure. However, in the context of the importance of inhibition of kinase activity by PEBP in modulation of signaling cascades in host cells, [103], it is tempting to speculate that bacterial BEPB-like proteins in bacterial pathogens were adopted to function as virulence factors via interference with kinase activity-dependent signaling pathways, including the NFκB signaling cascade.

Figure 8. Family I51 – E. coli homologue of Saccharomyces cerevisiae inhibitor of serine carboxypeptidase Y [137].

Secondary structure is presented as orange ribbon and electrostatic potential was mapped onto the molecular surface (blue (+) and red (−)). The structure was obtained from RCSB protein database (www.pdb.org) with accesion code 1VI3 [138].

N – Staphostatin families (families I57 and I58)

Staphostatins are inhibitors of staphopains (family C47), cysteine peptidases from Staphylococcus aureus. Staphostatin A (family I58) specifically inhibits staphopain A and not any other peptidase, whereas staphostatin B (family I57) specifically inhibits staphopain B [104]. Although the protein sequences of staphostatins A and B are too divergent for both to be included in the same family, the similarity of the tertiary structures revealed by crystallographic studies [105; 106] indicates that the proteins are distant homologues (Fig. 9). The fold is also similar to that of the Erwinia metallopeptidase inhibitor (family I38; see section J above) and families I38, I57 and I58 are included in the same clan (clan IK). The genes encoding staphostatin A and staphopain A are located in a single operon in S. aureus, and in vitro, the two have to be expressed together for efficient expression [107]. It is likely that the biological role of staphostatin A is the regulation of the activity of staphopain A. On the basis of the crystal structures of staphostatin B in complex with both an inactive mutant of staphopain B and with the active enzyme, Filipek et al. [108] contrasted the mechanism of inhibition of staphopains by the staphostatins with that of serine peptidases by the Laskowski-mechanism inhibitors. Staphopain B spans the active site cleft in the same orientation as a substrate, with residues Ile-Gly-Thr-Ser in the P2 to P2’ positions, but there is no sign of peptide bond cleavage. The P1 glycine residue of the inhibitor is in a conformation only possible for glycines [109]. Besides staphostatins A and B, S. aureus also produces staphostatin C (family I58), which is as yet uncharacterized. A single, uncharacterized staphostatin A-like protein is known from Staphylococcus epidermidis. Only one homologue of staphostatin B is known, from Staphylococcus warneri.

Figure 9. Family I56 and I57 – staphostatins.

Staphostatin A (red) and staphostatin B (violet) are presented as coloured ribbons (top panels). Bottom left panel presents the interaction between staphopain B (green) and staphostatin B (violet), with rectangle marking the active site. Bottom right panel shows detailed view on interaction in the active site of the enzyme. Catalytic residues are presented as green side chains and inhibitor P1 Gly 98 and P1’ Thr 99 are violet. The P1 residue clearly points out of the active site, making the hydrolysis of P1-P1’ bond impossible. The structures were obtained from RCSB database (www.pdb.org) with accession codes 1OH1 [139] (staphostatin A) and 1Y4H [140] (staphopain B-staphostatin B complex).

O – Pro-eosinophil major basic protein family (family I63)

Pro-eosinophil major basic protein (proMBP) associates with the metallopeptidase pappalysin-1 (family M43) in an extracellular 2:2 tetrameric complex, the formation of which involves changes in the redox state of six cysteine residues of the proMBP component and two cysteine residues of the PAPP-A subunit from pappalysin-1. Pappalysin-1 promotes cell growth by cleaving insulin-like growth factor binding proteins-4 and -5, which leads to the release of bound insulin-like growth factors [110]. However, the cysteine residues important for the disulfide exchange are not well conserved in family I63, and it is not clear if any protein other than that from human functions in the same way or is inhibitory. No homologues are known from Archaea, and only six from Bacteria, including three from Haliangium ochraceum (Deltaproteobacteria) and one each from Desulfovibrio salexigens (Deltaproteobacteria), Thioalkalivibrio sp. (Gammaproteobacteria) and Cyanothece sp. (Cyanobacteria). These proteins are entirely hypothetical, and none has been biochemically characterized.

P – Bacteriophage lambda CIII protein family (family I75)

Family I75 contains the bacteriophage lambda CIII protein, which has been shown to be an inhibitor of the metalloendopeptidase FtsH (family M41). Degradation of the phage CII protein by FtsH pushes the phage towards lysis rather than lysogeny, so inhibition of FtsH retains the phage in the lysogenic state [111]. Homologues are known from E. coli and Salmonella enterica (Enterobacteriaceae). The sequences are extremely closely related to the phage CIII protein and presumably represent phage genes retained in the bacterial genomes prior to the DNA being harvested and sequenced.

Q – Aspergillus elastase inhibitor family (family I78)

The elastase inhibitor from Aspergillus has been shown to inhibit the elastinolytic enzymes oryzin (family S8) and neutrophil elastase (family S1), but not elastinolytic metallopeptidases [112]. Over thirty homologues have been identified in bacterial genomes, the majority from Proteobacteria (of the alpha, beta and gamma groups). A single homologue is known from the Gram-positive bacterium Kineococcus radiotolerans (class: Actinobacteria). None of these homologues has been biochemically characterized and it is not known how many, if any, are peptidase inhibitors.

R – Other possible inhibitors

There are a number of prokaryote proteins that have been claimed to be peptidase inhibitors, but have not been fully characterized, and so the possibility that they are competing substrates has not been eliminated. An “intracellular proteinase inhibitor” from Bacillus subtilis that inhibited the intracellular subtilisin ISP-I from the same organism at nanomolar K_i was described by Nishino & Murao [113], but the inhibition of intracellular subtilisins ISP-II and ISP-III was affected by the substrate casein. The sequence was unrelated to that of any other peptidase inhibitor [114].

An extracellular inhibitor of trypsin, chymotrypsin and subtilisin was isolated and sequence from Bacillus brevis and named BbrPI [115]. The sequence contained no cysteine residues and was unrelated to that of any other peptidase inhibitor. Multiple forms of the inhibitor exist and it is synthesized as a precursor. A bacterial mutant strain deficient in BbrPI had a higher extracellular proteinases activity than the wild type but grew normally. The deficient mutant showed higher sensitivity to trypsin which suggests that the inhibitor may have a protective role from attack by exogenous proteinases [116].

The propeptide of pseudolysin (P. aeruginosa) is removed in the periplasm but remains associated with the metallopeptidase, which is inactive [117]. The propeptide also acts as a molecular chaperone, assisting in the correct folding of the peptidase [118]. It is not clear, however, whether the propeptide acts as an inhibitor when it is dissociated from the peptidase.

Sporulation in B. subtilis is governed by sigma factors, in particular sigma factor K which is synthesized as a precursor and processed by the membrane metallopeptidase SpoIVFB (family M50). By associating with two other proteins, SpoIVFA and BofA, the peptidase is inactive. SpoIVFB can cleave SpoIVFA at multiple sites, which leaves the possibility that BofA is a peptidase inhibitor [119].

Besides inhibitors that are proteins, some bacteria synthesize peptides and derivatives of peptides that are efficient peptidase inhibitors, often ones that affect peptidases from different families and different catalytic types. Perhaps the best known of these is leupeptin (N-acetylL-leucyl-L-leucyl-D,L-argininaldehyde), which was originally isolated from Streptomyces exfoliates [120]. This small molecule inhibits a wide range of serine, cysteine and threonine-type peptidases, including trypsin [121], PACE4 [122], calpain [123], clostripain [124] and the trypsin-like activity of the proteasome [125]. Other small molecule inhibitors produced by actinomycetes include bestatin and amastatin (inhibitors of aminopeptidases) [126; 127] and tyrostatin, which inhibits sedolisin (family S53) [128]. Because these inhibitors are not encoded by genes, it is not possible to predict whether or not they are present in prokaryotes, whether or not the genome has been sequenced. Presence of small molecule inhibitors may partially explain the paucity of protein inhibitors in prokaryotes.