1. Introduction
To date, recombinant toxins from various biological sources (bacteria, yeast, scorpions, snakes, spiders and other living organisms) are widely used as: (i) antimicrobial agents for medical purposes, as well as antimicrobial additives for the food and biotechnological industries, (ii) groundwork for the creation of drugs with anticancer activity and the treatment of neurodegenerative diseases and (iii) the basis to develop vaccines, etc. Multiple works have been performed to study the mechanisms of action of genetically modified toxins and their applications
[1][2][3][4][5][6] (
Figure 1).
Figure 1. Various applications of recombinant toxins.
The protein/polypeptide nature of most of these natural toxins allows them to obtain their recombinant forms. The potential for developing these biomolecules in high enough quantities is the basis for further advancements in developing vaccines and drugs with reduced cost and their widespread use, on the one hand. On the other hand, the production of recombinant toxins avoids the need to work directly with the natural sources of these biomolecules (animals and microbial pathogens). Obtaining genetic constructs encoding the synthesis of recombinant toxins expands the possibilities of their synthesis in special modified forms. Like many recombinant proteins, recombinant toxins can be obtained in high yields using different expression systems, including extracellular secretion, and further isolated and finely purified using affine carriers
[7][8].
2. Spectrum of Recombinant Toxins and Their Origins
Most of proteinaceous toxins well-studied to date are produced by various bacteria. However, toxins that are found in yeast, snake, scorpion and spider venoms and other living organisms are also actively studied by various scientific groups today. Recombinant toxins obtained from various origins and purposes of their obtaining are presented in
Table 1 [9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37][38][39][40][41][42][43][44][45][46][47][48][49][50][51][52][53][54][55][56][57][58][59][60][61][62][63][64].
2.1. Bacterial Recombinant Toxins
Bacterial cells are capable of synthesizing endo- and exotoxins. Endotoxins, as a rule, are cell-bound lipopolysaccharides that are released after cell destruction, while exotoxins are protein toxins that are synthesized inside cells and released into the environment. Thus, the recombinant forms of these exotoxins are discussed next (Table 1).
Botulinum neurotoxin (BoNT) and tetanus toxin (TeNT) produced by cells of the Clostridium botulinum and C. tetani, respectively, are among the most dangerous and therefore the most well-studied bacterial toxins. Botulism and tetanus diseases caused by these toxins are among the most severe neurological diseases that cause flaccid paralysis and spastic paralysis, respectively. In addition, BoNT is widely used to treat a number of diseases. Consequently, recombinant forms of these toxins have been actively created and researched for many years with the aim of both developing effective antidotes and obtaining drugs based on them.
A double-blind, placebo-controlled study evaluated the safety, tolerability and pharmacodynamics (PD) of the recombinant botulinum toxin serotype E (rBoNT-E) compared with commercial botulinum toxin type A (ABO, Dysport
®)
[9]. All doses of the recombinant toxin were well tolerated, and rBoNT-E had a faster onset of action, a greater peak effect and a shorter duration of effect at the highest tested doses compared with ABO.
To solve an opposite task and neutralize BoNT and other toxins, various antibodies are usually used. Special interest is afforded to single-domain camel antibodies (sdAb, VHH or nanobody) possessing unique structure and characteristics and their chimeras with usual human immunoglobulins. As a result, such immunotherapeutic agents could have up to 1000 times increased protective activity against
C. botulinum and prolonged circulation in blood
[45].
Different subtypes of BoNT have a varying toxicity, and BoNT/A is more potent toward the human neuroblastoma cell line as compared to BoNT/B
[51]. At the same time, genetic modification of the latter to BoNT/BY resulted in improved affinity for human synaptotagmin and BoNT/B receptor, as well as increased toxicity toward this cell line.
C3 protein toxin from
C. botulinum (C3bot) cells is a mono-ADP-ribosyltransferase that selectively intoxicates macrophages, osteoclasts and dendritic cells by cytosolic modification of Rho GTPases (Rho-A, Rho-B and Rho-C). Thus, C3bot and, even better, its nontoxic variant C3bot
E174Q have been proven as perspective transporters for selective delivery of small molecules, peptides and proteins to the cytosol of macrophages and other cells
[23][24].
Proteolytically activated separate binding/transport subunit C2IIa of C2 toxin from
C. botulinum has been found
[25] to be a specific inhibitor of chemotaxis of polymorphonuclear neutrophils (PMN), allowing selective suppression of excessive and harmful PMN recruitment to organs as a result of trauma. The enzymatically inactive N-terminal part of the
C. botulinum C2 toxin (C2IN) when fused to Rho-inhibiting C3 toxin from
C. limosum (C3lim) significantly improves the toxic action of the latter
[52]. In a clinically significant mouse model, the in vivo introduction of C2IN-C3lim into the lungs after a blunt chest injury prevented injury-induced recruitment of monocytes into the lungs. Thus, such combinatorial fusion chimeras can be of practical interest due to great variability of available toxin modules.
Until now, vaccination has been the best way to combat diseases associated with many bacterial strains, including
C. perfringens cells and α-, β- and
ε-toxins of the bacteria. However, commercially available vaccines are based on inactivated toxins and have many manufacturing disadvantages that can be overcome using recombinant antigens. Recombinant α-, β- and ε-toxins were synthesized in
E. coli cells to create a trivalent vaccine and evaluated on rabbits, cattle, sheep and goats. The levels of produced antibodies in all animals exceeded the minimum values recommended by international protocols
[46][47], thus proving the viability of the approach. Even more, nonvirulent species of the same bacteria can be modified to bear a specific toxin or its part and safely modulate strong immune response, e.g.,
Vibrio cholerae cells expressing the β-subunit of cholera toxin (CTB)
[48].
Another major group of proteinaceous toxins is produced by the members of the genus
Bacillus.
Bacillus cereus cells causing foodborne diseases secrete various pore-forming pathogenicity factors, including Hemolysin II (HlyII). As above mentioned, it can be specifically neutralized by antibodies
[27], thus preventing mortality in vivo
[28].
B. anthracis cells cause one of the most dangerous infectious diseases, Anthrax. The use of the Anthrax protective antigen (PA) is considered the most promising approach to the development of an Anthrax vaccine. However, the instability of the recombinant PA complicates the production of stable recombinant vaccines. Thus, a number of modification methods have been applied in recent years to design a stable recombinant Anthrax PA. For example, proteolytic-sensitive sites simultaneously with deamidation-prone amino acids can be genetically modified
[56][57]. Alternatively, additional stabilizers, e.g., spherical particles (SPs) of tobacco mosaic virus, can be added
[58]. Joint application of both methods gives even better results in terms of stability, immunogenicity and protectiveness of the final product, including in vivo tests with a fully virulent
B. anthracis strain.
B. thuringiensis cells produce δ-endotoxins (Cry), which are toxic to a wide variety of insect pests and currently used widely in agriculture. Insertion of the gene encoding Cry1Ia toxin into a bacterial strain inhibiting fungal growth results in combined fungistatic and insecticidal activity as well as ability to induce plant resistance
[29].
A lot of bacterial toxins in their structures contain metal ions performing various purposes. First of all, metal ions can be located in the active sites of metalloproteases such as BFT toxin from
B. fragilis leading to damage and necrosis of the intestinal epithelium
[30]. In addition, such metals can contribute to toxin structural stabilization and even promote recognition of the target receptor like in the case of staphylococcal enterotoxin-like protein P (SElP) from
Staphylococcus aureus binding to major histocompatibility complex class II (MHCII)
[59]. It should be noted that some virulence factors secreted by bacteria may be toxic to the microorganisms themselves. To prevent collateral damage and to additionally protect active components, they can be secreted in nanovesicles, which are able to be modeled in silico
[30].
The diphtheria toxin (DT) from
Corynebacterium diphtheriae kills mammalian cells by inactivating the elongation factor EF-2. The translocation domain in DT plays a critical role in allowing the catalytic domain to pass to the cytosol from endosomal compartments and can be used as a functional vector for active transport of protein drugs
[31].
Some mammalian species are resistant to DT. The DT receptor, proHB-EGF, in resistant and sensitive species differs by amino acid sequence and therefore by secondary structure; however, there is no consensual opinion on how the difference in the structure of primary receptors changes the process of internalization of DT by resistant cells compared to sensitive ones. According to some publications
[32], there can be even very little difference of binding constants of DT subunit B (which includes receptor-binding and translocation domains) to resistant and sensitive cells, while there was huge difference of intracellular concentrations of toxin within model cells. It means that multiple mechanisms of resistance to DT may exist in mammalian cells.
Several approved drugs, e.g., denileukin diftitox, which is fusion of DT with interleukin-2, are commercially available and actively used to date. However, research to improve their efficiency, producibility and safety as well as to obtain new therapeutics with DT are constantly continued
[54][55].
Listeria monocytogenes cells apply internalins InlA and InlB to attach and penetrate into mammalian cells. Curiously, hepatocyte growth factor receptors (HGFR) together with other multiple variants are also affected by InlB
[33]. This is important since HGF/HGFR play crucial role in liver restoration after its acute toxic damage. Thus, truncated bacterial InlB was implemented as a functional analogue of HGF to obtain novel drugs with hepatoprotective activity.
Bacterial toxins can interact not only with receptors themselves but with complexes of receptor and signal molecules. One of such examples is LcrV from
Yersinia pestis [34]. It is a strong virulence factor having multiple functionalities, one of which is specific activation of human receptor-bound interferon-γ (hIFN-γ), which resulted in immune cell death via apoptosis. It became possible only after hIFN-γ binding to receptor and presentation of its
138GRRA
141 site, which specifically interacts with
32LEEL
35 and/or
203DEEI
206 sites of LcrV. Thus, inactivation of these sites by specific antibodies completely prevents any harmful effects of LcrV.
Protein biosynthesis can be targeted by bacterial toxins, as well. For example, bacteria can utilize multiple enzymes from Gcn5-related N-acetyltransferase (GNAT) superfamily to acetylate and thus inactivate specific aminoacyl tRNAs, including transporters of Met, Ile, Gly, etc.
[35].
2.2. Yeast Recombinant Toxins
Killer yeasts are able to produce proteins named “killer toxins” that are often glycosylated and bind to specific receptors on the surface of the sensitive microorganism, which is then destroyed by a target-specific mechanism of action (
Table 1). They are widespread among yeasts and attract a lot of attention of researchers. To date, more than 100 types of killer yeasts have been described
[65]. The most well-characterized killer toxins in terms of their genetic determinants, biochemical characteristics, molecular targets on sensitive cells and mechanisms of their destruction are toxins K1, K2 and K28 from
Saccharomyces cerevisiae; zymocin from
Kluyveromyces lactis; PMKT and PMKT2 from
Pichia membranifaciens; PaKT from
Wickerhamomyces anomalus; HM-1 from
Cyberlindnera mrakii and Kpkt from
Tetrapisispora phaffii. Due to their properties and spectrum of action, which is aimed at pathogenic microorganisms, recombinant killer toxins are being actively investigated in order to develop therapeutic agents based on them. However, the lack of research on their effects on humans and animals limits their use in the food and feed industry. Another drawback is that additional information about the mechanisms underlying the formation of killer toxins in yeast is required. Without solving these issues, it is not possible to successfully implement killer toxins in practice
[65][66].
A study of
S. paradoxus revealed a new K1-like toxin (K1L) being active against sensitive competing yeast cells
[10]. It is encoded by double-stranded RNA (dsRNA) and satellite dsRNA, which may also be of virus origin. Its homologues have been identified in other six yeast species not belonging to
Saccharomyces and are likely to be acquired by horizontal gene transfer via dsRNA and/or DNA with subsequent diversification of their structure and toxicity profile.
Genetic fusion of toxins with fluorescent proteins allowed researchers to study the binding of the toxin to the cell envelope of affected yeast
[11]. However, intracellular translocation of labeled recombinant toxin K28 was not observed then, in spite of the presence of toxicity. It means there are gaps in our understanding of the true mechanism of killer toxin action and transport even among best-investigated ones. Further research is required to visualize intracellular transfer of toxins using high-resolution imaging techniques of individual molecules.
Killer toxin K1 is secreted by
S. cerevisiae strains in a heterodimeric form. After binding to the primary receptor (β-1,6 glucans) in the cell wall, K1 is transported to the plasma membrane and is initially supposed to interact with its secondary receptor Kre1p, which ultimately leads to an ionophoric disruption of the membrane function. However, expression of recombinant K1α in resistant yeasts lacking Kre1p resulted in profound toxic effect
[12], thus excluding role of the receptor. At the same time, co-expression of toxin precursor(s) in sensitive cells eliminated any negative effects. Thus, resistance to killer toxins is a part of adaptive (acquired) immune system.
Some killer toxins, e.g., Kpkt from
T. phaffii (formerly
Kluyveromyces phaffii), have antimicrobial activity not only on yeast but also on bacteria
[13]. Interestingly, activity of Kpkt was not detected toward all tested mycelial fungi. Meanwhile, Kpkt has a β-1,3-glucanase activity
[14][36] and thus can be combined, for example, with chitinases to synergistically improve their antifungal effects. At concentrations effective again yeasts, recombinant Kpkt has no effect on immortalized human epidermal keratinocyte cell line HaCaT
[36]. That makes it promising for further investigations.
2.3. Recombinant Toxins of Various Animals
Venoms of snakes, scorpions and spiders are used by animals as their own defensive and offensive means by immobilizing victim and blocking the functional activity of their cardiovascular, respiratory and/or nervous systems. Proteinaceous toxins are the main components of these systems and modulate important ion channels and receptors after introduction into the body. Today, powerful databases of poisons and protein toxins with improved properties have been assembled already for more selective action, resistance to the effects of proteases, less immunogenicity and improved characteristics, in terms of pharmacokinetic properties. These characteristics can be improved by genetic modification of amino acid sequences, addition of disulfide and ion bridges, etc. After all, animal venom toxins are of great interest for applications in medicine as a basis for drug development
[5] (
Table 1).
The β/δ agatoxin-1 of the spider
Agelena orientalis was obtained in recombinant form in the entomopathogenic fungus
Lecanicillium muscarium with a special secretory signal peptide
[16]. Further toxin was fused with eGFP to simplify the screening procedure. Unfortunately, toxic activity of the fusion protein was not investigated in the work.
Another fusion protein of GFP with agitoxin-2 from scorpion
Leiurus quinquestriatus hebraeus was more useful
[62]. That allowed researchers to visualize the binding of toxins to their receptor as well as to determine dissociation constants of various toxins competing for the same Kv1.3 channel.
Purotoxin-1 (PT1) from the venom of the Central Asian spider
Geolycosa sp. selectively inhibits the purinergic receptor P2X3 and is a potent analgetic. It can be produced in pilot scale as self-cleavable fusion protein with mini-intein DnaB
[17]. However, its purification is multistage and labor-intensive with modest yield at the end.
Interestingly, Tbo-IT2 toxin was identified in the spider
Tibellus oblongus by cDNA analysis of the transcriptome of its venom glands
[40]. Its amino acid sequence has a 41% identity match with the closest protein toxin, while its spatial structure folds into a well-known inhibitory cysteine knot (ICK). The first main difference is the formation of five disulfide bonds instead of the typical three that should result in extreme stability of the toxin. The second and the most puzzling difference is that Tbo-IT2 did not have inhibitory activity on the tested panel of available ion channels and neuroreceptors, while it is still toxic to the housefly,
Musca domestica. Further research may elucidate the target(s) of the Tbo-IT2.
Another attempt to apply mini-intein DnaB was a little bit more successful
[21], although the target toxin, APHC3 from the anemone
H. crispa, which has analgesic activity, was produced in inclusion bodies and multistage purification was still required.
Fusion with His-tag and Smt3-leader peptide was shown to be a much more efficient method
[22]. The resulting inhibitor of the TRPV1 ion channel (HCRG21 peptide from the sea anemone
H. magnifica) was easier to purify and after cleavage was obtained at comparable yield to APHC3.
As stated previously with bacterial toxins, antibodies are used almost exclusively in antivenoms
[20]. Combining several recombinant toxins in simple mixture
[15][49] or even in fusion protein
[63] often leads to improved efficiency of antivenoms, including comparing to commercial ones. Furthermore, it was found that rationally selected toxin-specific single-stranded DNA aptamers can exhibit broad cross-reactivity in vitro and ex vivo against isoforms of toxins found in various snake venoms
[50].
Computer modeling provides powerful tools to thoroughly solve even complicated issues. For example, interaction of proteinaceous toxins KTx from scorpion
M. eupeus with potassium channels (KV) was simulated and explained, followed by modulation of their activity using genetic modification
[38][39]. In other work
[43], authors have investigated binding of TFTs with novel receptors. Secondary structures of multiple actinoporins Hct from the sea anemone
Heteractis crispa were generated
[67], followed by analysis and successful structure–activity hypothesis testing.
Venoms contain a large number of biologically active compounds with diverse activities. Shorter peptides, e.g., azemiopsin acting on neuroreceptors
[18] and bradykinin-potentiating peptides (i.e., affecting blood pressure)
[19], could be prepared by a solid phase synthesis using a general Fmoc-method, while larger polypeptides are the most rational to produce using common expression systems
[42][43].
A modulatory effect of some proteinaceous toxins on neuroreceptors is worth mentioning. Well-known three-finger toxins (TFTs) and their analogues
[42][43] as well as azemiopsin bind mostly nicotinic acetylcholine receptors (nAChRs), but γ-aminobutyric acid receptors (GABARs) can also be affected
[43].
Pore-forming toxins, e.g., Hct from the sea anemone
H. crispa [44], have a wide nonspecific action and are almost equally cytotoxic to normal and malignant cells. However, fusing them with targeting partners, such as site-specific ligands, toxins or antibodies, could result in new drug platform development.
Recombinant toxins can be easily genetically modified and truncated to help researchers investigate their toxic action in a more detailed way. For example, peptide Ms 9a-1 from sea anemone
Metridium senile causes significant analgesic and anti-inflammatory effects by desensitization of TRPA1-expressing sensory neurons, and it was thought to be a positive modulator of TRPA1 channel. However, truncation of its unordered domains on the N- or C-terminus resulted in complete loss of analgesic and anti-inflammatory activities in vivo
[64]. Thus, another target receptor(s) is likely present in neurons.
2.4. Recombinant Prions
Prions (Pr) are infectious agents that cause devastating and incurable disorders known as transmissible spongiform encephalopathies (TSE). With the advent of innovative technologies, such as protein misfolding cyclic amplification (PMCA) and real-time quaking-induced conversion (RT-QuIC), in vitro amplification of prions has become possible. There is evidence suggesting that prion complexes can acquire high-order assemblies in vivo, which may look structurally ordered. However, the biophysical nature of these structures and their role in amyloid biology are still unclear. Despite the fact that the amyloid collected in vitro has some biochemical similarities with the ex vivo amyloid of the same protein, it often does not reproduce the biological activity of the latter. For example, preparations of prion protein (PrP), which are resistant to proteinase K and obtained exclusively from one recombinant PrP (rPrP), may not have any detectable infectious activity both in cell cultures and in animal bioassays. However, the proteinase K-resistant PrP obtained from rPrP is infectious if it is placed in the homogenate of a diseased brain ex vivo using the PMCA assay
[68].
To study the rPrP, mechanisms of the development of toxicity and pathogenicity of prion diseases as well as their role in the development of pathologies of the nervous system is an important task of the world scientific community (
Table 2,
[69][70][71][72][73][74][75]).
Recent studies have shown that the infectivity of prions and their neurotoxicity may not be related to each other. Therefore, it is important to distinguish directly infective prions and those with a toxic effect, since the current hypothesis suggests that it is not the prions themselves that are toxic but another type of protein responsible for the toxicity of the disease. This species may be a by-product of prion formation, in a non-pathway amyloid PrP structure or even a non-protein whose formation is catalyzed by a prion
[76]. Thus, using highly purified infectious prions, it was demonstrated that prions are not directly neurotoxic and that the toxicity presented in infected brain tissue may be different from infectious prions
[74].
rPrP was obtained using the insect baculovirus cell expression system (Bac-rPrP)
[69] to determine whether pathogenic Bac-pathogenic PrP (PrP
Sc) is produced spontaneously in intermittent ultrasound reactions. No spontaneous formation of Bac-rPrP
Sc was observed at 37 °C, but when the reaction temperature increased to 45 °C, Bac-rPrP
Sc was formed in all samples studied. Some variants of Bac-rPrP
Sc were transmitted to mice, but when the reaction was repeated for 40 cycles, transmissibility was lost. It is noteworthy that various variants of Bac-rPrP
Sc, including nontransmissive ones, were characterized by resistance to proteinase K and were dependent on the presence of cofactors during amplification. However, their characteristics also disappeared after 40 reaction cycles, and the variety converged on one variant. These results show that different variants of Bac-rPrP
Sc are generated with different transmissivity to mice and structural properties; variants of Bac-rPrP
Sc compete with each other and gradually converge to a variant with a slightly higher amplification rate.
To understand the role of the hydrophobic region in the formation of an infectious prion at the molecular level, X-rays of crystal structures of mouse PrP (MoPrP, residues 89–230) in complex with a nanobody (Nb484) were obtained
[70]. Using a rPrP reproduction system, it has been shown that binding of Nb484 to the hydrophobic region of MoPrP effectively inhibits the reproduction of proteinase-resistant PrP
Sc and the infectivity of prions. In addition, when added to cultured mouse brain slices in high concentrations, Nb484 did not exhibit neurotoxicity, which is sharply different from other neurotoxic antibodies against PrP. Thus, Nb484 may be a potential therapeutic agent against prion disease.
Five groups of transgenic mice expressing elk PrP (TgElk) were vaccinated with either one CpG adjuvant or one of four rPrP immunogens: deer dimer (Ddi); deer monomer (Dmo); mouse dimer (Mdi) and mouse monomer (Mmo)
[71]. Then mice were intraperitoneally infected with prions of chronic wasting disease (CWD). All vaccinated mice developed anti-PrP antibody titers detected by ELISA. It is important to note that all four vaccinated groups survived longer than the control group, while in the group immunized with Mmo, the average survival time increased by 60% compared to the control group (183 vs. 114 days after inoculation).
Thus, the use of recombinant forms of prions allows researchers to study their immunogenicity and to develop novel vaccines.
In order to establish how various cofactors modulate the formation and selection of prion strains, PMCA was used to generate a variety of infectious rPrP strains by multiplication in the presence of brain homogenate
[75]. It is known that brain homogenate contains certain cofactors whose identity is only partially known and which facilitate the transformation of normal PrP (PrP
C) into PrP
Sc. A mixture of various infectious prion strains was obtained and introduced into the brain homogenate, where various polyanionic cofactors were present. These cofactors could control the evolution of mixed prion populations toward the development of specific strains (types of conformations). As a result, it has been shown that various infectious rPrP can be obtained in vitro. Their specific conformation (strain) depends on the cofactors available during reproduction.
These observations are very important for understanding the pathogenesis of prion diseases and their ability to reproduce in various tissues and hosts.
The RT-QuIC method can be used to detect pathogenic PrP in various biological tissues of humans and animals. However, this method requires a continuous supply of freshly purified PrP and thus is not available in a diagnostic laboratory. To solve the issue, a method for obtaining a rPrP has been developed
[72]. Lyophilized rPrP from bank vole (BV rPrP) can be stored for a long time before use, as well as be transported at certain temperatures to appropriate diagnostic laboratories, which can facilitate implementation of the RT-QuIC method as a diagnostic tool
[72].
Nucleic acids have been shown in recent studies to act as potential cofactors of protein aggregation and prionogenesis. For example, RNAs, regardless of their sequence, source and size, modulate rPrP aggregation in a bimodal manner, affecting both the degree and the rate of rPrP aggregation depending on the concentration
[73].
This entry is adapted from the peer-reviewed paper 10.3390/ijms24054630