Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

Proteolysis is probably the most widespread mechanism of biological regulation. By controlling protein synthesis, turnover and activity, it is involved in fundamental physiological processes including cell differentiation, growth and signaling, apoptosis, fertilization, blood coagulation, immune response and digestion. Yet, uncontrolled proteolysis can be harmful for organisms, causing amongst others such diseases as emphysema, inflammation, cancer, neurodegenerative, immunological, and cardiovascular disorders. Proteolytic enzymes can be divided into at least seven families: serine, aspartic, cysteine, glutamic, metallo, asparagine and threonine proteases. Serine proteases (SPs) are among the most widely studied proteins; they are widespread in nature and are involved in numerous and diverse physiological and pathological processes. Over 82 thousand sequences of SPs have been collected in the MEROPS database (http://merops.sanger.ac.uk/), whereas the second largest family is the family of metalloproteinases (over 75 thousands). SPs originate from extracellular matrix, mast cells, cytotoxic cells, and neutrophils. Some of the most studied SPs are those involved in blood coagulation. There are several mechanisms found in mammalian systems that strictly control the activities of proteases. Like other proteins, their ribosomal synthesis can be controlled at several stages, followed in many cases by post-translational modifications. Compartmentalization and the manner in which proteinases are released in the pericellular environment are additional important mechanisms for the regulation of their activity. Release and folding of the active enzyme from its inactive precursor is one of the predominant methods of protease control. A second method is effected by the ubiquitous presence of protease inhibitors. Naturally occurring inhibitors are proteins or peptides except in the case of non-proteinaceous ones isolated from microorganisms.

According to recent data brought together in the MEROPS database almost 21 thousand inhibitor amino acid sequences have been identified. Based on their sequence homology, they have been grouped into 91 families and further assigned to 36 clans according to their similar tertiary structures. There are also a large number of small-molecule inhibitors, both naturally occurring and synthetic. These peptidomiometics, or simply small organic molecules, are outside the scope of this mini-review. Since more than one third of identified proteases are serine proteinases, it is no surprise that their inhibitors (SPIs) also represent the largest group. They are found in all living organisms. In fact, the first described were trypsin inhibitors. In 1936 Kunitz and Northrop isolated and crystallized a bovine pancreatic trypsin inhibitor (BPTI). Later, Kunitz identified another, named STI, from soybean. In the meantime, another inhibitor, the Bowman-Birk inhibitor (BBI), was identified from the same source and characterized. These proteins inhibited trypsin and at neutral pH formed stable complexes which were dissociated at low pH. Progress in bioanalytical techniques has confirmed earlier suggestions that these inhibitors interact with enzymes in a substrate-like manner. They are probably the best studied proteins and their complexes with cognate enzymes are among the best known examples of protein—protein interactions. This was also one of the reasons that knowledge about inhibitors of other proteolytic enzymes was, until the last decade, significantly less well advanced.

2 Inhibition Mode of Serine Proteinase Inhibitors

On the basis of their mechanism of action, SPIs are classified as: canonical (standard mechanism), serpins and non-canonical. Except for the last group, SPIs interact with target enzymes in a substrate-like manner. Canonical inhibitors are small proteins or peptides with molecular weights varying from 1.5 to 2.1 kD, corresponding to 14−200 amino acid residues. They are characterized by rather compact, thermally resistant structures, and although they possess different folding motives, they are usually stabilized by several disulfide bridges. These native structures are also stable in denaturizing agents. A common structural element in canonical inhibitors is a short fragment comprising less than 10 amino acid residues which is directly involved in interaction with cognate enzymes. This exposed fragment, named the “binding loop” or the “primary binding segment”, has an extended conformation with a similar anti-parallel β-strand. Bode and Huber [1] proposed to call it “canonical”. Some inhibitors contain a single binding loop, some others consist of multiple domains where such elements are located. Thus, such inhibitors interact with proteinases with stoichiometry other than 1:1. Alternatively, they are able to inhibit more than one enzyme simultaneously. It transpires that all standard mechanism inhibitors exhibit this canonical conformation of binding loop, and are therefore named “canonical inhibitors”. This mode of inhibition was studied in detail by the group of Laskowski [2]. Upon enzyme—inhibitor interaction, a stable complex C is formed rapidly and then (unlike in substrates) very slowly dissociated into a free enzyme and modified inhibitor. As indicated by X-ray data, the majority of contacts with enzymes are made by the inhibitor binging loop [1] that fit into the active site of a cognate enzyme. The intermolecular contact area is relatively small, restricted to 6−9 nm2. In the middle of this loop is located a peptide bond P1-P1′ (reactive site). According to the Schechter and Berger nomenclature [3], P1-P1′ is the scissile peptide bond and both amino acid residues P1 and P1′ interact with the corresponding enzyme—S1 and S1′- subsites, respectively. The hyper-exposed amino acid residue in position P1 interacts with the S1 cavity of the enzyme accounting for the largest part (up to 50 %) of the total association energy released during complex formation with the target enzyme. Therefore, P1 residue determines inhibitor specificity and is often referred to as the primary specificity residue. As in substrates, the reactive site P1-P1′ is selectively cleaved by proteinase yielded modified inhibitor. After sufficient incubation time, an equilibrium between intact and modified inhibitor is formed. The determined values of the hydrolysis constant (Khyd = [I]/[I*]) for most canonical inhibitors are, in neutral pH, close to unity. It is worth mentioning that in a number of 3D structures of proteinase-inhibitor complexes determined by X-ray crystallography the reactive site P1-P1′ remains intact. The mechanism of inhibition of canonical inhibitors with cognate enzymes is shown in Fig. 12.1. One of the most eminent researchers in the field of serine proteinase inhibitors was undoubtedly Michael Laskowski Jr. who devoted his whole scientific life to the elucidation of this mode of inhibition. In recognition of his contribution, MEROPS suggests the adoption of the new name, the “Laskowski mechanism”, for what has been accepted as the standard mechanism.

Fig. 12.1
figure 1

Enzyme—inhibitor interaction according to Laskowski Jr

Full size image

The acronym “Serpin” (serine proteinase inhibitors) was introduced in the mid 1980s [4] to describe a superfamily of relatively large proteins (45–55 kD); some of them are serine and cysteine protease inhibitors, others play non-inhibitory functions. Despite a low sequence homology, they adopt a similar metastable conformation comprising three β-sheets, eight to nine α-helical segments and an exposed binding loop, also called the “reactive center loop” (RCL). This large family (I4) consists of over 2500 members, found in all kingdoms of organisms including 44 in humans. Serpins interact with target proteinases based on a unique mechanism that is described as the suicide substrate-like inhibition. Inhibition by serpins is irreversible. Proteinase is inactivated, kinetically trapped in the complex and can be cleared from the circulation by binding to members of the low density lipoprotein receptor family. Irreversibility is also reinforced by the induced proteolytic susceptibility of proteases in complex with serpins.

The first non-canonical inhibitor, hirudin, was isolated from the saliva of the medicinal leech in the 1950s and a decade later its primary structure was determined. This polypeptide, consisting of 65 amino acid residues, displays anticoagulant activity and is a very potent and specific thrombin inhibitor. Hirudin and its variants isolated from leech species share a common 3D structure, N-terminal globular domain cross-linked by three disulfide bridges and an acidic C-terminal fragment, which displays flexible high conformational freedom. Both structural elements are directly involved in the interaction with proteinase. Negatively charged C-terminal segment directs hirudin towards a positively charged thrombin exosite I (fibrinogen recognition exosite), followed by their tight binging. This induces conformational changes in the enzyme allowing the interaction of N-terminal tripeptide of the inhibitor with the thrombin active site. In contrast to canonical inhibitors, this fragment does not form an anti-parallel but rather a parallel β-sheet with side chains of a thrombin Ser214—Gly219 segment and a negatively charged side chain of Asp189 located in the enzyme S1 subsite does not interact with the inhibitor. Another thrombin inhibitor isolated from leeches (Hirudinaria manillensis)—named haemadin, although it does not share sequence homology with hirudin, displays a similar tertiary structure and similar mode of thrombin inhibition. This was the reason that hirudin and haemadin were assigned to the same family (I14). Nevertheless, the analysis of its complex with thrombin revealed several differences. One is the direct contact of the positively charged guanidine group of inhibitor’s Arg2 with carboxylate of enzyme’s Asp189. This resembles the P1—S1 interaction of canonical inhibitors. The second is the interaction of the C-terminal segment not with exosite I but with exosite II. Several Cys-rich inhibitors including antistatin, therostatin, tessulin, bdellastatin, guamerin and the most potent thrombin inhibitor theromin were assigned to this family (I15) and this is collectively called the antistatin family. These inhibitors display a common folding motif consisting of several repeats of N- and C-domains. On the basis of limited structural data, it is postulated that inhibitors of the I15 family interact with target enzymes in a canonical (substrate-like) manner with P1 position located in the C-domain, while (in at least some inhibitors) the highly acidic N-domain may interact with thrombin exosites. A more detailed mechanism of interaction between non-canonical inhibitors and proteinases has been presented in a recently published review [5].

3 Classification of the Canonical Serine Proteinase Inhibitors

Extensive studies on the isolation, characterization and determination of the 3D structures of inhibitors obeying a common, so-called standard mechanism (later named canonical) led to the conclusion that not all inhibitors are homologous. They were found in almost all tissues where they were looked for—animal, plant and microbial. Laskowski and Kato proposed grouping them into families. Originally, nine families were recognized, but information on three of them was rather marginal. Subsequently, with progress in the identification and characterization of new canonical inhibitors more members of each family were found and, in addition, more families were added to the list. Finally, based on results obtained over the following two decades, Laskowski Jr’s group [2] recognized at least 16 different families (Table 12.1). More recently, Rawlings et al. [6] proposed a new classification system for the protease inhibitors that allows them to be added to the MEROPS database. They introduced the term “inhibitor unit” which is a segment of the inhibitor containing a single reactive site. An inhibitor containing a single inhibitor unit is named a simple inhibitor, whereas one containing multiple units is termed a compound inhibitor. Based on sequence homology, inhibitors were grouped into 48 families (currently, there are 79 distinct families). In addition, inhibitors which display the same type of protein fold were assigned to the same clan. To date 39 clans have been recognized, and this means that some clans contain representatives of more than one family. In the case of the canonical inhibitors of serine proteinases 18 families have been recognized and grouped in 14 clans. Table 12.1 presents a summary of the Laskowski and MEROPS classifications of these inhibitors. The size of these proteins varies from 14 to 200 amino acid residues. As shown in Table 12.1, canonical inhibitors are widely distributed in the plant kingdom. To date, at least eight families have been classified as being of plant origin. They are especially ubiquitous in tissues responsible for storage (seeds and drupes). It has been proved that the biosynthesis of inhibitors in plants is stimulated by infection or damage of vegetative tissue. Thus, they are important components of plant defense systems targeting exogenous proteinases.

Table 12.1 Families and clans of canonical serine proteinase inhibitors
Full size table

4 Squash Family of Serine Proteinase Inhibitors

The first members of this family were isolated from squash (Cucurbita maxima) seeds and characterized at the beginning of the 1980s independently by Polanowski et al. from University of Wroclaw (Poland) and Hojima et al. from NIH, Bethesda (USA). These polypeptides inhibiting several serine proteinases (such as bovine β-trypsin, human Hageman factor XIIa and kallikreins) were named ITD I, ITD III and PHFI, respectively. It turned out that ITD III is identical to PHFI. Subsequently, more homologues inhibitors were isolated from Cucurbitaceae, and in 1985 Wieczorek et al. [7] postulated that these low-molecular mass inhibitors could be assigned to a new family of serine proteinase inhibitors (I7). In addition, the Polish group introduced a new nomenclature using the scientific names of the plant which is consequently applied to all members of the squash family. Thus, the acronym ITD was replaced by CMTI (Cucurbita maxima trypsin inhibitor). To date over 40 inhibitors assigned to this family have been identified in seeds of Cucurbitaceae. Their amino acid sequences, sources and years of discovery are summarized in Table 12.2. A unique feature of the discussed family of inhibitors is its unusually small size, compact structure and high resistance to denaturizing conditions. Squash inhibitors consisting of 28–34 amino acid residues and their structure are stabilized by three disulfide bridges. Two disulfide bridges form a ring that is threaded by a third one. Such a structural motive is called the cystine knot (CK) [8]. Interestingly, two members of the discussed family called MCoTI-I and MCoTI-II isolated from Momordica cochinchinensis possess an additional cyclic element—head-to-tail cyclization and are the first macrocyclic inhibitors of the squash family [9]. At present over 160 such cyclic peptides found in nature are collectively named cyclotides [10]. They display a characteristic folding path where the cystine knot is embedded within a macrocyclic backbone, defining a motif referred to as a cyclic cystine knot (CCK). The topology of disulfide bridges, cystine knots and CCK motives is illustrated in Fig. 12.2.

Table 12.2 Amino acid sequences of squash family inhibitors
Full size table
rFig. 12.2
figure 2

The cystine knot motifs

Full size image

4.1 Cystine Knot Inhibitors

In squash inhibitors, the P1-P1’ corresponds to positions 5 and 6 in CMTI inhibitors. In substrate specificity P1 position Arg or Lys is present; therefore, they are potent inhibitors of trypsin and other serine proteinases with a negatively charged S1 subside. The exceptions are MCEI inhibitors with Leu5 which therefore inhibit elastase. Also, P1’ position is conservative and is occupied in all in squash inhibitors by Ile residue.

The unusually small size of squash inhibitors as compared with other proteinaceous serine proteinase inhibitors makes them readily accessible by means of chemical synthesis. In contrast to biological methods of synthesis where introduction into the protein molecule other than by proteinogenic amino acids is rather difficult, chemical methods allow practically all modifications to be achieved. CMTI-III and -I, the first members of the family were synthesized by us in the mid 1980s [11]. This definitely confirmed the correctness of their sequences. A few years later, the French group isolated, synthesized and determined solution structure [12] homologues EETI-II from seeds of Ecballium elaterium. The authors showed that deletion of its C-terminal dipeptide Ser29-Pro30 retained its inhibitory activity. Both CMTI-III and EETI-II (shortened 28 amino acid residue variant) were used for structure- activity-relationships. Since the first synthesis, we have synthesized over 60 analogues of CMTI/EETI analogues. The obvious modification introduced by several groups into the squash inhibitors was the substitution of Arg5 by aromatic or aliphatic amino acid residues. This yielded analogues with specific chymotrypsin and elastase inhibitory activity, respectively. On the other hand, we have shown that the requirements for P1 position are very strict. The introduction of Arg or Lys derivatives in this position dramatically lowered trypsin inhibitory activity. It is worth mentioning that derivatives differ from proteiongenic amino acids by one methylene group in the side chains (l-homoarginine or l-homolysine) introduced in the P1 position reducing the affinity of such analogues towards trypsin by 2–3 fold. We were also able to considerably simplify the structure of CMTI. The analogue with a truncated N-terminal dipeptide, and C-terminal Gly and contains 7 amino acid residues replaced by Gly or Pro residues retaining high inhibitory activity. All three disulfide bridges are important in maintaining the tertiary structure of squash inhibitors. On the other hand, we have recently shown [13] the substitution of each disulfide bridge by diselenide yielded potent trypsin inhibitors and more redox-resistant EETI-II analogues. Le-Nguyen et al. [14] reported the synthesis of hybrids of EETI-II and carboxypeptidase inhibitors. The rationale of their work was the fact that both peptides display the same fold as the cystine knot motive. It turned out that such an analogue inhibited both proteases used. Analogues of EETI-II were also synthesized by Hilpert et al. [15]. They replaced an N-terminal heptapeptide in EETI-II by an optimized binding loop of the turkey ovomucoid inhibitor (OMTKY3). The hybrid inhibitor HEI-TOE I retained inhibitory activity against porcine pancreatic elastase with considerably higher proteolytic stability. Inhibitors of this family were also successfully obtained using biological methods. Sato et al. [16] obtained recombinants of CMTI-II by expression of its gene in silkworms, whereas Grzesiak et al. [17] expressed series of CMTI-I analogues modified within a binding loop in E. coli. Since plant inhibitors play an important role in defense systems against insects or micro-organism predators, Milner et al. [18] fused CMTI-1 to the coat protein of potato potyvirus Y and bacterial β-glucuronidase in order to check whether such an insert could protect a protein against proteolysis. They showed that inhibitory activity was retained when the inhibitor was fused to the first protein only. All the above mentioned synthetic studies on squash inhibitors proved that their analogues with possible therapeutic applications can be accessible for both chemists and biochemists.

4.2 Cyclic Cystine Knot Inhibitors

It has already been mentioned that MCoTI-I/II are members of a family of miniproteins named cyclotides. Peptides displaying the CCK motive possess a diverse range of biological activities including antimicrobial, insecticidal and pesticidal activity. Cyclotides represent attractive scaffolds for the design of peptide-based drugs. This was the main reason that MCoTI-I/II became the most studied inhibitors of the squash family. Both were isolated from the seeds of the squash Momordica cochinchinensis in 2000, and in the following year the tertiary structure of MCoTI-II was determined. It was independently shown by two groups that its structure is similar to that of the other members of the squash family. Cysteine-rich proteins are attractive models to study oxidative folding pathways. Bearing in mind that the discussed inhibitors contain 6 Cys residues, 15 different sets of disulfide bridges can be formed. It turned out that the pathways of oxidation folding of the reduced precursors of cyclic and acyclic squash inhibitors (EETI-II and MCoTI-II were studied) are similar, yielding the native species. In both cases, only one major intermediate with two native disulfide bridges was present [19].

In contrast to other inhibitors of the discussed family, MCoTIs contain macrocycle formed by a polypeptide chain. Therefore, chemical synthesis of such inhibitors requires an additional step. One of the options is generation of a thioester at the C-terminus of the fully protected precursor followed by removal of the side chain protecting groups and a ring-closing macrolactam formation by native chemical ligation. Oxidative refolding of homodetic peptides gives the desired inhibitor [20]. MCoTIs were also biosynthesized in E. coli cells using a biomimetic approach that involves the use of a modified protein splicing unit in combination with an in-cell intramolecular native chemical ligation reaction [21]. Briefly, a MCoTI-I linear precursor was inserted into a modified intein yielding Met-Cys dipeptide at the N-terminus. After the enzymatic removal of Met and generation of thiester at the C-terminus of the MCoTI precursor, native chemical ligation was used to accomplish the main chain cyclization. In the final step, oxidative folding produced MCoTI. Finally, the inhibitor was purified on HPLC or using trypsin-immobilized Sepharose beads. A biochemical approach was applied to obtain the library of MCoTI-I analogues and affinity on trypsin-immobilized Sepharose beads was utilized to select active ones. Leatherbarrow’s group [22] reported on modification (including deletion of fragment 3–6) within the binding loop of MCoTI-II that gave potent inhibitors of physiologically important tryptase and human leukocyte elastase (HLE). Substitution of Lys10 located in substrate specificity P1 position by Arg increased inhibitory activity towards the first enzyme, whereas the introduction in this position of Val or Ala gave analogues possessing potent inhibitory activity against HLE. Moreover, considering the fact that a shortened analogue (28 amino acid residues) preserved tryptase inhibitory activity without significantly altering the three-dimensional structure, this proved the impressive adaptability of this cyclotide scaffold. These results correlate well with high proteolytic stability in the stomach and in plasma of cyclic hybrid of MCoTI-II(6-21) and EETI-II(16-28) named McoEeTI [23]. It was pointed out that this inhibitor can be considered a novel pharmacophoric carrier for oral peptide drug delivery.

Perhaps the most promising feature of MCoTIs is their membrane permeability. MCoTI-II biotin labeled at all three Lys residues was internalized into macrophages RAW264.7 and breast cancer MCF-7 cells at a non-toxic concentration [24]. This was the first reported cyclic cell-penetrating peptide. In this first report Craik’s group has shown that translocation into cells is mediated through macropinocytosis. By contrast, another cyclotide—kalata B1 analogue (with Thr16 replaced by the biotin labeled Lys)—is not only significantly more toxic but is also not able to cross the cell membrane. This finding opens a new possibility to use MCoTI-II as a carrier of bioactive peptides to intracellular targets. The above mentioned internalization of MCoTI-II was analyzed in fixed cells. Recently, this process was studied on MCoTI-I in live HeLa cells by Camarero’s group [25]. Since this inhibitor contains one Lys residue only (corresponding to inhibitor P1 position), its acylation by AlexaFluor488 N-hydroxysuccinimide yielded a mono-labeled analogue. With the use of various endocytic markers and inhibitors, the authors proved the uptake of AF488-MCoTI-I in HeLa cells following multiple endocytic pathways. Moreover, a significant amount of MCoTI-I was found in lysosomes, and also in late endosomes or other types of acidic organelles. These results were confirmed very recently by Craik’s group who re-evaluated their studies; this time on live cells [26]. They observed that Alexa-labeled MCoT-II and also kalata B1 analogue (Thr20 replaced by labeled Lys) were present inside macrophages and breast cancer cells. The first peptide penetrated cells mainly by macropinocytosis, whereas the second did through direct membrane interaction. Interestingly, uptake in MCF-7 cells of the mono-labeled analogue of MCoT-II (at Lys10 only with the remaining two replaced by Ala) was increased by 20 % as compared with the native sequence.

5 Sunflower Trypsin Inhibitor 1 (SFTI-1)

Since its discovery in 1999, trypsin inhibitor isolated from sunflower seeds (SFTI-1) has become the object of study for several groups. To date, SFTI-1 is the smallest naturally occurring circular proteinase inhibitor. SFTI-1 is a member of the Bowman-Birk group of inhibitors (I12) retaining the nonapepide binding loop of other inhibitors of this family, spanning the disulfide bridge. It consists of 14 amino acid residues and, in addition to the circular polypeptide chain, contains the second cycle—the disulfide bridge formed by Cys residues. The P1-P1’ reactive site is located at Lys5-Ser6. Despite the fact that this is much smaller than other BBIs which are composed of 60–90 amino acids, SFTI-1 is the most potent trypsin inhibitor of this family. The elimination of one of the cycles retained inhibitory activity and the three dimensional structures of such analogues. Structure-activity-relationships studies on SFTI-1 have shown that substitution in the P1 position caused dramatic changes in specificity, opening the possibility to design inhibitors of physiologically important proteinases. Several groups have shown that modifications introduced within the SFTI-1 reactive site and its vicinity led to potent physiologically important enzymes including: chymotrypsin, human neutrophil elastase, cathepsin G, matriptase, β-tryptase, proteinase K, and human kallikrein-related peptidase 4. Nevertheless, none of the analogues displaced inhibitory activity higher than the wild inhibitor. This might indicate that the structure of SFTI-1 is already optimized by nature. Most of the results obtained for SFTI-1 in the first decade of its study are already summarized in review papers [27, 28]; therefore, here we wish to discuss only those published recently. Pereira et al. [29] proposed the use of SFTI-1 analogues with different specificity immobilized on Sepharose as tools to identify proteinases in biological samples. Authors covalently attached SFTI-1 analogues through hydrazone bonds formed by the hydrazine group of Sepharose and the aldehyde moiety of a side chain of N-terminal amino acid of the inhibitor. The aldehyde group was introduced by the oxidation of the hydroxyl group of additional Ser coupled to N-termini of inhibitors. Such affinity resins display high capacity and stability under experimental conditions. An interesting issue discussed in recent papers was the cyclization mechanism of SFTI-1. Craik’s group [30] determined that in the case of SFTI-1, the ratio between intact and modified inhibitors reached 9:1, regardless of whether or not intact or modified inhibitor was incubated with trypsin. More recently, the same group studied cyclization and degradation of series SFTI-1 analogues with cleaved peptides bonds at Arg2, Lys5 [31]. In the case of substrates, proteolysis of a single peptide bond usually gives two products, whereas in these cyclic inhibitors a single product is formed and re-synthesis of cleaved bonds is observed. Interestingly, when an analogue lacking a disulfide bridge was incubated with an enzyme, two peptide bonds were cleaved (after Arg2 and Lys5), tripeptide Arg-Abu-Lys was removed and peptide bond Arg2-Ser6 was formed yielding a 11 amino acid residue cyclic analogue. This corresponds with our results on proteolysis of double-sequence of SFTI-1 analogues. Following this mechanism, incubation of these analogues with trypsin or chymotrypsin yielded monocyclic SFTI-1 or [Phe5]SFTI-1, depending on the amino acid residue present in P1 position. We proposed to name this re-arrangement of the peptide chain peptide splicing [32].

Recently, Cascales et al. [26] reported that fluorescent Alexa-labeled SFTI-1 penetrates MCF-7 cells and is located in endosomal compartments. The cellular uptake of SFTI-1 is comparable with that determined for Alexa-labeled MCoTI-II, but in contrast to the latter inhibitor, no affinity of SFTI-1 to phospholipids was observed. Taking into consideration that all three cyclic peptides tested (MCoTI, kalata B1 and SFTI-1) enter cells, the authors proposed that they constitute a new family of cyclic cell-penetrating peptides. Moreover, in our recent work with the help of new fluorescent labels, we have shown that also monocyclic SFTI-1 analogues (with disulfide bridge only) are able to enter cancer cells. The work is in progress, and only preliminary results have been published [28].

6 Biosynthesis of Plant Cyclic Peptides

Cyclic peptides discussed herein were found in the plant kingdom. All are composed of proteinogenic amino acids. Their precursors are synthesized on ribosomes that undergo post-translational processing to generate cyclized inhibitors. It was observed that Asp or Asn is a conservative C-terminal residue present in cyclotides (including MCoTIs) and also in SFTI-1 precursors. This led to the suggestion that an asparaginyl endoprotease is involved in head-to tail-cyclization in plants. According to the proposed mechanism, in the first step a peptide bond after Asp (or Asn) is cleaved, followed by the formation of the peptide bond with N-terminal residue. This mechanism is also supported by experiments showing in vitro cyclization of linear precursors. Although other cyclization mechanisms are not excluded (e.g. the peptide splicing discussed above) the one mediated by asparaginyl endoprotease seems to be the most accepted [33].

7 Conclusions and Future Perspectives

Inhibitors of serine proteinase isolated from the seeds of Cucurbitaceae and sunflowers are a very exciting group of peptides. They are the smallest naturally occurring proteinase inhibitors composed of proteinogenic amino acids. They display well-defined, compact structures stabilized by disulfide bridges. Some of them are characterized by a head-to-tail cyclic backbone. These cyclic peptides are able to cross cell membranes. Unlike most peptides discussed here, low-molecular plant inhibitors are exceptionally resistant to denaturizing conditions and proteolysis. The observed rearrangement of the polypeptide chain (peptide splicing) upon incubation of some SFTI-1 analogues with serine proteinases is another feature worth considering for further studies. It might be possible to utilize such compounds for transfer and release inside the cell of certain peptidic or peptidomimetic drugs. All the information discussed above makes these plant serine proteinase inhibitors valuable scaffolds for designing compounds with potential commercial possibilities for medical and agrochemical applications. The size of these inhibitors makes them readily accessible by chemical synthesis. In contrast to biological methods, the possible modifications introduced to the chemically synthesized compounds are almost unlimited.