Plant amino acid analogues as antimicrobial agents

Abstract

Plants are known as a source of different biologically active compounds, which are uncommon for other kingdoms of life. Among them are different amino acid analogues, which are synthesized and accumulated in certain plants as a passive defense mechanism against herbivorous insects and grazing mammals. As a rule, cell protein synthesis machinery of herbivores cannot effectively differentiate between standard proteinogenic amino acids and their specific plant analogues, resulting in misincorporation of the latter into nascent proteins and their malfunctioning, which constitutes a mechanism of plant defense. Examples of such amino acids are analogues of arginine (canavanine, indospicine), proline (azetidine-2-carboxylic acid), and cysteine/lysine (thialyasine). This review summarizes existing knowledge on these and other related amino acids as potential antibacterial and antifungal agents, including their possible targets and known resistance mechanisms. We also discuss the possibility of using amino acid analogues as sole antimicrobial agents or in combination with known antibacterials and antifungals. We also propose a strategy of enhancing the antimicrobial activity of amino acid analogue by concomitant starvation for the corresponding standard amino acid, which has been proven efficient in anticancer studies. Such an approach might potentially help to overcome, at least partially, microbial resistance to known antibiotics, especially when such resistance relies on increased protein synthesis in pathogen cells.

Introduction

Plants have been well known as a source of different biologically active substances since antiquity. Due to their specific metabolic pathways, plants often produce unique substances uncommon to other kingdoms (animals, bacteria, and fungi) and sometimes toxic to the latter. Such substances are often structural analogues of known compounds that mimic their basic structure but also exhibit certain structural changes that significantly alter their biophysical properties and biological activity.

Such structural analogues or their derivatives are often of interest in the search for new anticancer, antimicrobial, and antifungal agents. Among them, of special interest are amino acid analogues (AAAs) that are not used in protein biosynthesis in their plant hosts, i.e., are non-proteinogenic, but mimic natural so-called standard amino acids in their structure, and thus become conditionally (organism-dependent) proteinogenic when consumed by other organisms such as herbivores (Pass et al. 1996). These compounds possess various chemical modifications, such as altered side chains, backbone modifications, or non-standard functional groups. Plant AAAs can exhibit different mechanisms of toxicity against grazing animals, bacteria, or fungi. Besides interfering with protein synthesis and stability, they can also act as enzyme inhibitors, serving as potential therapeutic agents for diseases involving specific enzymatic activities and providing insights into the catalytic regulatory mechanisms of enzymes.

As follows, studying AAAs deepens our understanding of physiological processes and drives innovation in drug design, protein engineering, synthetic biology, and biotechnology, expanding our understanding of protein function and enabling novel biomedical applications (Fuertes et al. 2023).

Classification of AAAs

By definition, amino acids are organic compounds containing amino and carboxyl groups and are monomeric subunits of proteins, in which amino acid residues are connected by peptide bonds. There are 20 so-called standard (canonical) proteinogenic L-amino acids that are translated by the universal genetic code and participate in protein biosynthesis. Another two, selenocysteine and pyrrolysine, are also incorporated into proteins, but in very rare cases in eukaryotes, archaea, and bacteria (Rother and Krzycki 2010; Yuan et al. 2010). Additionally, more than 500 AAAs are found in nature, which carry chemical modifications that affect their recognition by protein translation machinery, biochemical, and functional properties.

There are several major categories of AAAs that have to be mentioned, such as:

Stereoisomeric analogues (L- and D-amino acids). While all naturally occurring amino acids in proteins are in L-configuration relative to their α-carbon atom, their stereoisomeric D-counterparts exist in bacterial cell walls and specific peptides. Incorporating D-amino acids into peptides and proteins can increase resistance to enzymatic degradation and alter biological activity, making them valuable for antimicrobial peptide research and drug design. The following sections provide descriptions for L-AAAs, unless indicated otherwise.

Chain-modified analogues feature modifications to the amino acid backbone, such as replacing methylene groups or introducing heteroatoms. Such modifications can enhance proteolytic stability and alter secondary structural preferences, as seen in β-amino acids and their oligopeptide derivatives. Azetidine-2-carboxylic acid (Aze), a proline (Pro) analogue, disrupts protein folding and has been used to study misfolding-related diseases (Roest et al. 2018). Its capacity to substitute for Pro in protein synthesis underlies its teratogenic effects. It is a valuable tool for creating non-native proteins to investigate proteotoxic and endoplasmic reticulum stresses (Rodgers et al. 2025).

Isosteric analogues possess substitutions of atoms with chemically similar elements. In these AAs, specific atoms in the amino acid structure are replaced with elements of comparable size and electronic configuration. For example, selenium-containing selenomethionine serves as an isosteric substitute for methionine and is widely used in X-ray crystallography for phase determination (Hendrickson et al. 1990; Strub et al. 2003). This approach has revolutionized protein structure determination, with over two-thirds of new protein structures being solved using selenomethionine derivatization (Sheng and Huang 2008). Other isosteric analogues include arginine (Arg) mimetics of plant origin, canavanine (Cav) and indospicine (Isp), described in detail below.

R-group modified analogues retain the core structure of natural amino acids but exhibit alterations in their side chains (or R-groups). For example, due to their unique electronic properties, fluorinated amino acids, such as 4-fluorophenylalanine, have been used in ¹⁹F NMR to study protein stability and enzyme catalysis. ¹⁹F NMR provides valuable insights into such processes as protein folding, enzymatic activity, and interactions involving protein–protein, protein-lipid, or protein–ligand, as well as aggregation and fibrillation in both soluble and membrane proteins (Kitevski-LeBlanc and Prosser 2012; Meyer and Knapp 2015). Another example, difluoromethylornithine (DFMO), is an irreversible inhibitor of ornithine (Orn) decarboxylase, a key enzyme of polyamine biosynthesis. By depleting polyamines, DFMO exhibits activity against malignant tumors, in particular neuroblastoma (Gandra et al. 2024).

Amino acid analogues as antimicrobial agents

One of the still unsolved challenges for modern biomedical sciences is the spread of resistance to existing antimicrobial drugs and the apparent deficit of novel, effective antibiotics. The annual global cost of treating infections caused by drug-resistant bacteria is currently around $66 billion and is projected to rise to $160 billion per year by 2050. At the same time, up to 10 million additional deaths per year due to antibiotic resistance are predicted (Poudel et al. 2023). Also, the use of newly approved drugs to treat resistant bacterial infections is up to 60 times higher compared to old ones (Yahav et al. 2021). The development of a novel antibiotic is a complex, time-consuming, and resource-intensive process. The last discovery of a molecule belonging to a novel antibiotic class was daptomycin as late as 1987 (entered the market in 2003) (Heidary et al. 2018; Gray and Wenzel 2020).

The development of alternatives to classical antibiotics based on secondary plant metabolites is a promising and strategically important direction, because these compounds are less susceptible to the development of resistance compared to existing antibiotics (Keita et al. 2022). At least in part, this is due to their unique mechanisms of action (e.g., disruption of signaling pathways, membrane dysfunction, inhibition of adhesion). Many plant compounds exhibit synergistic effects with antibiotics, restoring the sensitivity of bacteria to existing drugs (Atta et al. 2023). Secondary metabolites can be used in the form of topical agents (gels, creams, ointments), inhalations, local antiseptics, or additives to dressings, combined with drugs (e.g., enzymes or probiotics).

Another advantage of plant metabolites is their relatively affordable cost. However, several challenges and limitations are associated with their studies and introduction into clinical use, including standardization of concentrations of bioactive substances and their stability in pure form under physiological conditions (Nishad 2022). The development of drugs based on plant secondary metabolites, AAAs in particular, is therefore a sound strategy for curbing the spread of antibiotic resistance and expanding the arsenal of antimicrobial agents.

Among many AAAs found in different plant species, we identified several AAAs that are conditionally proteinogenic and exhibit proven antibacterial and/or antifungal activity. Their antimicrobial activity, mechanisms of acquired resistance, and possible use in novel antimicrobial strategies are discussed below.

Arginine analogues

Leguminous plants of the genera Canavalia and Indigofera caught the attention of scientists as a source of many bioactive compounds. Among others, they synthesize analogues of Arg, a conditionally essential amino acid in humans, Cav and Isp (Bueno Pérez et al. 2013).

Cav and Isp are naturally occurring, organism-dependent proteinogenic analogues of Arg. Cav is the delta-oxo-analog of Arg, where the terminal methylene group of Arg is replaced by an oxygen atom, forming an oxygen-containing guanidine group (Staszek et al. 2017). Isp contains a modified guanidine group, making it an isosteric analogue of Arg (Fig. 1a). In nature, these compounds are not toxic to the plant organism, accumulate in all tissues, predominantly in leaves and seeds, and protect the plant from pathogenic bacteria, herbivorous insects, and other animals (Hegarty and Pound 1970; Rosenthal 1990a).

Fig. 1.

Structures of selected AAAs described in the review. A Arginine analogues—Canavanine, Indospicine; B Ornithine analogue—Canaline; C Proline analogue—Azetidine-2-carboxylic acid; D Cysteine/Lysine analogue—Thialysine

Such AAAs specifically target rapidly dividing cells, such as certain cancer cells, especially those dependent on the elevated exogenous Arg uptake (Ho Jang et al. 2002; Vynnytska et al. 2011; Nurcahyanti and Wink 2017). Still, studies on Cav anticancer activity were essentially abandoned due to observed high cytotoxic doses in the presence of Arg, insufficient specificity, hepatotoxicity, and autoimmune concerns associated with misincorporation of Cav into nascent proteins (Rosenthal 1977; Fletcher et al. 2015; Karatsai et al. 2020; Hauth et al. 2022b). However, Cav can also misregulate several Arg-dependent signaling pathways and may serve as a precursor of another cytotoxic non-proteinogenic AAA, canaline (Can) (Rosenthal 1997; Bobak et al. 2016; Fried et al. 2021).

Similarly to Cav, Isp is incorporated into and destabilizes nascent proteins and acts as a competitive inhibitor of arginase, leading to liver degeneration and abortions in feral livestock consuming Isp-containing legumes or domestic animals fed feral meat-containing food (Fletcher et al. 2015). However, our group recently proposed using Cav or Isp in combination with Arg deprivation (e.g., using recombinant human arginase I (rhARG1) and irradiation (Vynnytska-Myronovska et al. 2016; Kurlishchuk et al. 2016; Karatsai et al. 2020; Shuvayeva et al. 2021). These studies produced some encouraging results and are in progress (For review, see (Stasyk et al. 2015)).

Canavanine

Studies on the antimicrobial effects of Cav were conducted on gram-negative bacteria, mainly Escherichia coli and, to a lesser extent, on Pseudomonas spp. Cav was toxic to some strains of E. coli, especially those characterized by upregulated Arg and Orn uptake (Schwartz and Maas 1960; Celis 1977), with minimum inhibitory concentration (MIC) against different E. coli strains ranging from 32 to 256 µg/mL (0.18–1.45 M). Cav also reduced the biofilm-forming potential of these bacteria from 68.42% to 31.58%, respectively (Abouzeid et al. 2023). In other studies, Cav inhibited the growth of different gram-positive Lactobacilli, but only in Arg-deficient conditions (Volcani and Snell 1948; Kihara and Snell 1955).

In contrast to the above-mentioned data, Cav in a broad range of concentrations (1 µM-10 mM) significantly activated the growth of Streptococcus (Enterococcus) faecalis var. liquefaciens in the presence of Arg in the medium (0.05 M), with the maximum effect observed at 0.1 mM. The proposed mechanism underlying this effect was hypothetically associated with Cav-mediated inhibition of Arg-degrading enzymes and longer bioavailability of Arg molecules for bacterial cells (Hammel and Zimmerman 1963).

The competitive nature of into-protein incorporation between Cav and Arg and growth inhibition by Cav (Richmond 1959a) mentioned above, partially explains the elevated Cav toxicity in Arg-free medium, also in the context of Cav studies as an anticancer agent. Cav (in micromolar range) became almost two orders of magnitude more toxic for cultured human cancer cells in Arg-deficient medium relative to Arg-sufficient conditions (in millimolar range) (Vynnytska et al. 2011).

An interesting finding was that D-Cav, produced by P. putida broad-spectrum racemase from L-Cav, has dramatically higher antibacterial effects. It altered the peptidoglycan structure and affected the cell division process of Rhizobiales (e.g., Agrobacterium tumefaciens, Sinorhizobium meliloti). A point mutation in the penicillin-binding protein 3a could overcome this effect by making it insensitive to D-Cav and, thus, reducing its incorporation into peptidoglycan (Aliashkevich et al. 2021).

Besides incorporating into the polypeptide chain, Cav can also interact with various microbial proteins, thereby affecting the corresponding metabolic pathways. For example, Cav incorporation resulting in loss of function has been shown for various bacterial enzymes, including alkaline phosphatase, beta-galactosidase, hyaluronidase, and phosphatase (Richmond 1959a; Attias et al. 1969). 10 mM Cav significantly decreased the activity of Arg deiminases (ADI) isolated from bacteria (P. putida, P. aeruginosa, E. coli, Burkholderia mallei, Bacillus cereus) (Li et al. 2008; Patil et al. 2019). This process was associated with the formation of a modestly stable S-alkylthiouronium intermediate and a stable Cys-alkylthiocarbamate adduct (Li et al. 2008). However, Cav concentration used in those experiments far exceeds the known physiological levels for AAAs.

Interestingly, intrinsic ADI expression increased bacterial sensitivity to Cav, as demonstrated in Tetragenococcus halophilus. The possible mechanism could be associated with Arg depletion from the medium due to ADI activity, and/or, alternatively, with concomitant cytotoxicity of Cav hydrolytic catabolite produced by ADI (Lu et al. 2005). Cav also inhibited polyamine (putrescine and spermidine) synthesis in T4-infected E. coli, but the detailed inhibitory mechanism has not been deciphered. The mechanisms involved in bacterial response can also be associated with the inhibitory effect of Cav on Arg anabolic enzymes (Bolin and Cummings 1975).

As for alternative enzymes, combined treatment of E. coli cells with Mg⁺⁺ ions and Arg or Cav resulted in changes in the activity of acetylornithine delta-transaminase (Leisinger et al. 1969). Cav (0.05–0.1 mM) repressed E. coli Orn transcarbamylase (Faanes and Rogers 1968). Cav (0.1–1.0 mM) was also an inhibitor of several enzymes not associated with Arg metabolism in E. coli: alcohol dehydrogenase, beta-glucosidase, and oxynitralase. The proposed mechanism was associated with the Cav interaction through its oxo-group with -SH groups in protein active sites (Tschiersch 1966), but this hypothesis requires further clarification.

Notably, exposure to Cav led to the accumulation of Cav-containing proteins in E. coli cells, which formed agglomerates with bacterial DNA (Simonnet and Paresys-Richli 1972). These agglomerates, in the form of membrane-associated canavanyl-protein bodies, affected transcription and translation, resulting in cell death (Schachtele and Rogers 1965, 1968; Schachtele et al. 1970).

Different mechanisms of resistance to Cav have been described for various bacterial species. Firstly, some microorganisms, predominantly associated with the rhizosphere of leguminous plants (e.g., Pseudomonas canavaninivorans and Rhisobiales spp.), can degrade Cav, and thus use this compound as a sole source of nitrogen and carbon (Hauth et al. 2022a). Such specific substrate utilization is mediated (at least partially) by Cav-gamma-lyase, which removes the hydroxyguanidine moiety from Cav (Hauth et al. 2022b).

Other mechanisms of Cav metabolism by bacteria include its decarboxylation to γ-guanidinoxy-propylamine by some E. coli strains (Makisumi 1961) and Cav degradation to homoserine and guanidine by some S. faecalis strains (Richmond 1959a, b), supporting S. faecalis growth as indicated above.

Since Cav toxicity is mainly manifested via disturbed Cav-containing proteins, another mechanism of resistance to Cav is associated with the activation of proteases. In E. coli, it especially concerns protease La, which degrades Cav-containing proteins. However, this process may not be Cav-specific, but rather be part of the cell's overall response to stress (Schachtele and Rogers 1968; Goff and Goldberg 1985; Lee et al. 1991). Such proteolysis can be inhibited in vitro by antibiotic rifampicin, which blocks protein translation in many bacteria (Krzyzek and Rogers 1972; George et al. 1980).

Cav misincorporation into proteins could also be overcome by mutations in bacterial Arg-tRNA synthetase, causing lower affinity to Cav (Williams 1973). This leads to a lower frequency of Cav misincorporation into proteins. Such a resistance mechanism was also found in the bruchid beetle (Caryedes brasiliensis), which has evolved to feed on Cav-producing plants. This insect has evolved tRNA synthetase, which is highly selective for Arg, i.e., can effectively differentiate between Arg and Cav (Rosenthal 1990b).

P. canavaninivorans possesses a unique protein, canavanyl-tRNA^Arg deacylase, which is upregulated by Cav and is controlled by guanidine riboswitches, revealing a functional connection of Cav-response and guanidine metabolism. This enzyme thus acts as a standalone editing protein, specifically deacetylating canavanylated tRNA^Arg and preventing its incorporation into proteins. Knockout strains in CtdA show severe growth defects in Cav-containing media and incorporate higher amounts of Cav into their proteome (Hauth et al. 2023).

Yet another alternative mechanism of Cav resistance is associated with impaired expression or functioning of its permeases, e.g., Arg and Orn transporters, which can become less affine to Cav due to acquired mutations. For example, Cav-resistant strains of E.coli were characterized by a high number of mutations in the ArgP locus (encoding LysR-type transcriptional regulator protein), resulting in aberrant expression of permeases involved in Arg and Lys-Arg-Orn transport, higher Arg efflux, and reduced ability to accumulate Cav inside cells (Rosen 1973; Celis 1977; Nandineni and Gowrishankar 2004). Mesorhizobium tianshanense, a Gram-negative bacterium, expresses MsiA protein as a specific Cav exporter, which allows this organism to form a symbiosis with the Cav-producing legume plant Glycyrrhiza uralensis (Zhong et al. 2015).

Thus, Cav can be used as a selective antimicrobial agent, though different resistance mechanisms to this compound exist. It should be emphasized that Cav, a toxic Arg analogue, showed the highest antimicrobial effects in Arg-depleted conditions (Richmond 1959b). Thus, combined exposure to Cav and Arg deprivation (evoked, for instance, by commercially available enzymes like rhARG1) might have beneficial synergic antibacterial effects. To confirm this hypothesis and investigate effective antimicrobial Cav doses and possible synergy, additional studies are needed on different bacterial species, especially on clinical isolates with acquired antibiotic resistance.

Cav was also studied on different yeast strains and species, predominantly being used in mutagenic assays and as a reporter for transformation with foreign DNA (Gocke and Manney 1979; Fantes and Creanor 1984; Cunha et al. 2006; Nishimura et al. 2023). This was due to the identification of three genes associated with Cav resistance in different unicellular fungi (Saccharomyces cerevisiae, Schizosaccharomyces pombe, Candida glabrata):

• CAT1 gene or its orthologs, encoding Arg permease responsible for Arg and Cav uptake;

• VHC1 (syn. CAN1), encoding vacuolar membrane cation-chloride co-transporter;

• ANY1, encoding α-arrestin (ubiquitin ligase adaptor)—a transcriptional repressor of plasma membrane amino acid transporters (Gocke and Manney 1979; Kaufman and Livingston 1983; Yang et al. 2022; Ait Saada et al. 2022).

The primary mechanism of Cav toxicity in yeasts is essentially the same as in bacterial or mammalian cells: proteotoxic stress after Cav incorporation into proteins instead of Arg (Polacheck and Kwon-Chung 1986). This corresponds to the fact that sensitivity to Cav was markedly decreased when ubiquitin was overexpressed in S. cerevisiae cells (Chen and Piper 1995), indicating the involvement of proteolytic enzymes in the clearance of misfolded ubiquitinated Cav-containing proteins, thus reducing proteotoxic stress (Pazlarová et al. 1993; Galego et al. 1993; Druseikis et al. 2019).

Another yeast-specific mechanism of Cav toxicity is associated with the induction of mitochondrial stress due to a significant decrease in mitochondrial protein synthesis and respiratory activity, for example, in certain S. cerevisiae strains (Wilkie 1970), and affected intracellular Arg transport in these yeasts upon exposure to Cav (Boller et al. 1975). Interestingly, Cav toxicity increased when yeasts were forced to respire, suggesting a possible connection between respiration and the development of proteotoxic stress (Druseikis et al. 2019).

Sensitivity to Cav is often associated with high Arg demand as observed in S. pombe mutants that lacked farnesyl transferase beta subunits and S. cerevisiae mutants that lacked Rheb GTPase (Yang et al. 2000; Urano et al. 2000; Panepinto et al. 2002). This resulted in abnormal signal transduction from Arg sensors to TOR Complex 1, placing yeasts in a state of apparent Arg deprivation.

Interestingly, high sensitivity to Cav was also observed in riboflavin-deficient Pichia guilliermondii mutants (Sibirnyĭ et al. 1977). The exact underlying mechanism has not been described, suggesting yet unidentified Cav-resistance mechanisms involving riboflavin or enzymes dependent on riboflavin-derived cofactors, such as flavin mononucleotide or flavin adenine dinucleotide.

As in bacteria, an association of Cav resistance in yeasts with its decomposition and degradation to an unspecified non-toxic product was proposed but not deciphered in detail (Polacheck and Kwon-Chung 1986). Analogues of bacterial Cav-detoxifying enzymes in yeasts have not yet been identified or described.

In laboratory practical diagnostics, Cav is used to differentiate between pathogenic basidiomycetes Cryptococcus species complexes, namely Cryptococcus neoformans and Cryptococcus gattii. These tests are based on differences in utilization of glycine as a sole source of carbon and nitrogen, along with Cav resistance (Polacheck and Kwon-Chung 1986; Min and Kwon-Chung 1986). Upon growth on Cav-glycine-bromothymol blue (CGB) agar, C. gattii (but not C. neoformans) can degrade Cav to non-toxic metabolites using metabolic pathways that are different from those found in bacteria and are not fully described yet (Polacheck and Kwon-Chung 1986). These reactions shift the pH of the medium and trigger the blue color reaction of bromothymol indicator. This is usually observed after 48 h of incubation (Nakamura et al. 1998; McTaggart et al. 2011). This test is more accurate than previously used creatinine-dextrose-bromthymol blue or glycine-cycloheximide-phenol red media (Kwon-Chung et al. 1982). However, it cannot be confidently used as a single test only (Klein et al. 2009). Besides using CGB agar in differentiation tests for Cryptococcus spp., studies on yeasts were mainly performed on non-pathogenic yeasts. Thus, more knowledge in this field is needed on pathogenic and semi-pathogenic species to consider Cav as a potential antifungal agent.

Indospicine (Isp)

Isp is another, but less-known, Arg analogue found in leguminous plants of the genus Indigofera (Hegarty and Pound 1970). It is considered a component of metabolic anticancer therapy based on Arg deprivation (Shuvayeva et al. 2021). Unlike Cav, it lacks a true guanidinium moiety (Fig. 1a). This is why Isp is not degraded by known Arg-degrading enzymes, which likely leads to its higher stability relative to Cav in mammals (FitzGerald et al. 2011; Tan et al. 2016a, b and our unpublished data). Historically, interest in studies on Isp was mainly associated with its accumulation in the meat of grazing animals and subsequent intoxication of pet animals that consumed contaminated wild meat (Fletcher et al. 2015). Thus, in vitro studies with Isp were often conducted on hepatocytes or other relevant hepatotoxicity models (Sultan et al. 2018).

Recently, however, it was shown that grazers’ rumen-derived microbes can slowly degrade Isp to 2-aminopimelamic and 2-aminopimelic acids, thus partially protecting animals from Isp-induced toxicity. The exact mechanisms of this transformation and the proteins involved remain to be further elucidated (Tan et al. 2017; Gilbert et al. 2021). Another mechanism of Isp degradation to 2-aminopimelamic acid and 2-aminopimelic acid is heating of Isp-containing solution with sodium bicarbonate in the microwave oven or autoclaving for 15 min (Tan et al. 2016b). Thus, slow Isp degradation in the rumen may involve both chemical and biochemical mechanisms.

Few studies addressed the antimicrobial effects of Isp. It was shown that the addition of Isp to the growth medium in low-Arg conditions inhibited the growth of E. coli and P. aeruginosa (Leisinger et al. 1972). The proposed reason was that, similarly to Cav, Isp could charge bacterial tRNA^Arg and incorporate into proteins instead of Arg, resulting in their misfolding and malfunctioning. Also, Isp at millimolar concentrations inhibited the activity of certain enzymes involved in Arg biosynthesis (namely, N-acetylglutamate-5-phosphotransferase from P. aeruginosa) (Leisinger et al. 1972).

Thus, Isp may be presumed to evoke antimicrobial effects that mimic those of Cav, simultaneously being substantially more resistant to degradation by microbial and animal enzymes.

Ornithine analogue—Canaline (Can)

Can is a non-proteinogenic analogue of Orn (Fig. 1b), a product of Cav hydrolysis by arginases (Rosenthal 1997). Similarly to Cav, Can is also present in many legumes, but its toxicity mechanisms are less elucidated. Can is the only naturally occurring free amino acid with an aminooxy group in the side chain, replacing the δ-amino group in Orn, and forming stable oximes between Can and carbonyls (Rosenthal 1978; Hauth et al. 2022a). Can is capable of inhibiting Orn-dependent enzymatic activities and functions also as a Lys antagonist (Rosenthal 1977). However, as mentioned, Can is not incorporated into proteins (Bence et al. 2002).

Can also reacts with pyridoxal phosphate (coenzyme B6), forming a stable, ninhydrin-positive complex. This complex inhibits B6-containing enzymes, especially bacterial L-tyrosine decarboxylase (Rosenthal 1981). Can's irreversible oximes with pyridoxal phosphate also block the activity of Orn aminotransferase (OAT) and other aminotransferases. This AAA was shown to inhibit vaccinia virus replication in HeLa cells. The addition of Orn did not reverse the inhibition by Can. However, the inhibitory effect of Can was gradually reversed by the addition of increasing amounts of pyridoxal phosphate. These observations suggested that Can, in this case, does not act as a structural analogue of Orn and inhibits Orn aminotransferase activity by its interaction with pyridoxal phosphate, which results in a deficiency of this cofactor (Archard and Williamson 1974). Also, Can evoked inhibition of aminotransferases in Mycobacterium spp., although the efficacy was lower than that of other aminooxy compounds (Kito et al. 1978).

Can had a potent inhibitory effect on Plasmodium falciparum with IC₅₀ of 300 nM. Synergy with DFMO (Orn decarboxylase inhibitor) was also observed, indicating a disruption of polyamine biosynthesis in this parasitic organism (Berger 2000).

Unfortunately, there is an apparent lack of studies on the effects of Can on bacteria and yeasts, including pathogenic and semi-pathogenic, to discuss its use as an antimicrobial agent.

Proline analogue—Azetidine-2-carboxylic acid (Aze)

Aze is found in Convallaria majalis (also known as Lily of the Valley), where it comprises up to 7% of the leaf mass (Fowden 1956; Gross et al. 2008), in some leguminous Fabaceae plants and, in smaller quantities, in Beta vulgaris (sugar beets) (Rubenstein et al. 2006). Aze, the Pro analogue (Fig. 1c), participates in a plant defense mechanism against herbivores and phytopathogens, similarly to Cav and Isp (Klaubert et al. 2025). Aze can directly inhibit Pro biosynthesis pathways, be incorporated into polypeptides, and disrupt their secondary structure due to different bond angles compared to Pro (Fowden and Richmond 1963; Couty and Evano 2006; Thives Santos et al. 2024).

Available data on Aze antibacterial activity, resistance mechanisms, and human toxicity are extensively summarized in (Rodgers et al. 2025). Briefly, the antibacterial activity of Aze was mainly studied on E. coli. Similarly to Arg analogues described above, profound antibacterial effects were observed only when growth occurred on a Pro-deficient medium. Adding Pro to the medium fully protected bacteria and mammalian cells from Aze’s cytotoxicity (Piper et al. 2023).

Aze was previously studied as a possible antibiotic, but the studies were terminated due to unacceptable side effects (mainly teratogenicity and neurotoxicity). This compound is currently being studied as a potential anticancer agent because its ability to modify tumor proteins by being incorporated instead of Pro enhances tumor immunogenicity, potentially improving the outcomes of tumor immunotherapy procedures (Piper et al. 2023). This strategy may be promising, but considering the reported data on Aze teratogenicity, neurotoxicity, and autoimmune effects, rational means of reducing its overall toxicity to normal human cells are strongly needed.

One can assume that the above-mentioned antimicrobial strategy (simultaneous application of a given AAAs and corresponding standard amino acid-depleting enzymes) may indeed be effective. Nevertheless, none of the identified Pro-degrading enzymes (Pro oxidases, Pro dehydrogenases, etc.) have even entered pre-clinical trials. Considering that Aze in general has a worse safety profile than those of Arg analogues, it can be predicted that the probability of success of such a strategy may be rather low.

A number of resistance mechanisms to Aze have been reported in bacteria and fungi. Aze can be detoxified by acetylation in S. cerevisiae Σ1278b, mediated by MPR1/2 gene products, members of the N-acetyltransferase superfamily (Takagi et al. 2000). Fungus Aspergillus nidulans can utilize Aze as a source of nitrogen due to the activity of AzhA hydrolase and activation of GABA catabolic pathways (Biratsi et al. 2021). Some bacteria can even use this substance as a sole source of nitrogen (Enterobacter, Agrobacterium, Rhizobium genera, and some Pseudomonas sp.) due to transaminase activity and opening of the azetidine ring (Yeung et al. 1998; Gross et al. 2008). This reaction typically occurs in the bacterial periplasm to further protect bacterial cells from Aze-mediated toxicity, similar to the activity of beta-lactamases, which hydrolytically open the beta-lactam rings of antibiotics in the periplasm.

Cys/Lys analogue—Thialysine (ThLys)

Thialysine ((S)-2-aminoethyl cysteine, ThLys) is a toxic AAA, in which the second carbon atom in the side chain of Lys has been replaced with sulfur. ThLys can also be considered a structural cysteine S- (2-aminoethyl) analogue (Fig. 1d).

Studies on Lys auxotrophic E. coli mutants showed that ThLys (and, to a lesser extent, selenolysine) can be effectively incorporated into proteins in Lys-low conditions, but not in the case of total Lys depletion (Coccia et al. 1983). Interestingly, bacteria could substitute up to 60% of the total Lys in their proteins to ThLys without any significant changes in viability (Cinia et al. 1984), indicating a high level of baseline resistance to the compound. Further incubation led to the instability of ThLys-containing proteins, their rapid degradation, and cell division arrest (Di Girolamo et al. 1988). In another study, up to 42.5% of the Lactobacillus and 83.3% of the yeast isolates were capable of Lys production even in the presence of ThLys in the medium (Odunfa et al. 2001).

Artificial point mutations in E. coli Lys-tRNA synthetase (Y280F and F426W) resulted in the development of resistance to ThLys because of lower binding affinity for ThLys compared to Lys, and, therefore, higher selectivity of aminoacylation (Ataide et al. 2007). Resistance to ThLys in S. cerevisiae was associated with activation of the Lys production pathway, especially in industrial Lys-overproducing strains.

As for eukaryotes, the ability to incorporate ThLys into proteins was also demonstrated in vitro on CHO and Jurkat cells, resulting in cell growth inhibition, arrest in S and G2/M phases of the cell cycle, and apoptosis (Blarzino et al. 1987; Jun et al. 2003). This suggests a low selectivity of ThLys between prokaryotic and eukaryotic cells, resulting in a high risk of unwanted toxic effects. Considering data on baseline resistance levels, even in low-Lys conditions, we consider the possible risk/benefit ratio for ThLys to be lower than for Aze.

Safety concerns

It should be acknowledged that there is a very limited amount of data in the literature regarding the pharmacokinetics of the aforementioned AAAs. The common safety concern for AAAs is their low selectivity between microbial and human cells, due to nonspecific main mechanisms of toxicity, namely, misincorporation into primary polypeptide structures, resulting in protein misfolding and the development of proteotoxic stress. However, in case of topical applications, such a risk factor should greatly diminish.

To date, several studies have observed toxic effects for humans and animals associated with the consumption of Can, Isp, Can, Aze, and ThLys. A general summary of these findings is provided in the table (Table 1).

Table 1.

Summary of existing data on in vivo toxicity of discussed AAAs

AAA	Animals, route of administration	Doses, effects	References
Cav	Rats, adult, s/c	LD50: 5.9 ± 1.8 g/kg (single injection) 2.0 g/kg—serum amylase and lipase levels were elevated	(Thomas and Rosenthal 1987)
	Rats, 10-day-old, s/c	LD50: 5.0 ± 1.0 g/kg (single injection)
Isp	Cattle, sheep, camels, goats, etc	Evidence from small studies, focused on feeding different animals with Isp-containing herbs (like I. spicata) A diet containing more than 25% of I. spicata caused toxic effects: loss of appetite, transient neurological impairment, liver and kidney lesions, which could lead to death. These findings and their intensity were not uniform between different species Isp consumption in a daily dose of 0.2 mg/kg for 52 days was considered safe for the Angora goat Long-term I. spicata-containing diet during pregnancy caused abortion or stillbirths in cows and horses	Reviewed in (Fletcher et al. 2015)
Can	n/a	n/a	n/a
Aze	Mice, s/c	LD50 = 1000 mg/kg	(Bérdy et al. 1980)
	Mice, p/o, i/p	Single 600 mg/kg caused oligodendropathy by altering myelin structure	(Sobel et al. 2022)
	Different animal models	Toxic effects were registered in rats after a single s/c injection of 0.1 g/kg Aze Toxic changes were generally associated with collagen misfolding and resulting embryonal, skeletal, and vascular malformations	Reviewed in (Rodgers et al. 2025)
ThLys	Mice, i/p	LD50 = 300 mg/kg	Mentioned in the MSDS from various vendors

The toxicological profile for each of the mentioned AAAs is incomplete. For example, many essential indices, such as LOAEL and NOAEL, have not yet been established. There is phenomenological evidence that the compounds mentioned can be toxic to animals and humans; however, cut-off values for safe exposure or dose–effect relationships remain unknown. Several comprehensive studies have been published, discussing the in vivo toxicology and pharmacokinetics of Cav (Thomas and Rosenthal 1987), Isp (Fletcher et al. 2015) and Aze (Rodgers et al. 2025). Unfortunately, data on Can and ThLys toxicity, as well as on the toxicity of plant extracts containing AAAs, are scarce, and further studies in the field are needed. Currently, these compounds are actively studied alone or in combination with other agents as anticancer agents; however, these studies are mainly performed in vitro, with limited contribution to the understanding of toxicology and pharmacokinetics.

Also, there are safety concerns associated with the inability of the human organism to metabolize certain AAAs (especially, Isp). Phenomena of Isp accumulation in the meat of herbivorous (ponies, camels, horses) and carnivorous animals (dogs eating camel meat enriched in Isp) were confirmed in several studies (Tan et al. 2016a; Fletcher et al. 2018; Netzel et al. 2019). Still, no safe/toxic doses, exposure parameters, or dose–response relationships were elucidated.

Based on the data presented above, we suggest strategies aimed at limiting the systemic exposure of model organisms to AAAs. For example, considering the topical application of these agents rather than injections or oral administration could be a viable approach, as this method is known to reduce unwanted systemic exposure and side effects. Other approaches might aim to lower the effective concentrations of AAAs using targeted delivery systems, or in combination with other medications. In our previous studies, we demonstrated that the simultaneous application of Cav and rhARG1 reduces the effective anticancer concentration of Cav by more than an order of magnitude (Vynnytska et al. 2011; Kurlishchuk et al. 2016). We assume that this approach (application of AAAs and enzymatic depletion of corresponding amino acids) can be hypothetically applied to other AAAs and significantly improve their safety profile.

Future perspectives

In this review, we provide a brief overview of the limited available information on the potential application of AAAs from plants as antimicrobial and antifungal agents. It was established that Cav, Isp, Aze, and ThLys can be incorporated into protein structure in place of Arg, Pro, and Lys, respectively, resulting in protein malfunction and the development of proteotoxic stress of different strengths. Among other AAAs, Cav is better studied as an antimicrobial, exhibiting broad-spectrum antimicrobial activity against various bacteria and yeasts. Compared to Aze and ThLys, Arg analogues possess several unique features that may be of added value in the context of their antimicrobial action. For instance, Isp does not possess a true guanidinium moiety and is not subject to degradation by the known Arg-degrading enzymes such as arginases or ADI. Cav, in turn, is degraded by arginase to Can, another toxic AAA, which is not incorporated into protein structure but can interact with carbonyl groups and disrupt protein translation.

The common feature of the mentioned conditionally proteinogenic AAAs is that they evoke their most potent cytotoxic effects in the environment with low amounts of the corresponding standard amino acid counterparts. We propose, therefore, the simultaneous application of AAAs and corresponding standard amino acid-degrading enzymes as a potential antimicrobial approach. Such combinations should conform to the requirement that given AAA, contrary to a corresponding standard amino acid, is not a subject of degradation by a given hydrolytic enzyme. For Arg AAAs, PEGylated rhARG1 (pegzilarginase, Loargys) recently received market authorization for treatment of arginase 1 deficiency, also known as hyperargininemia; PEGylated arginine deiminase (ADI-PEG 20) is currently in late-stage clinical trials for cancer treatment (reviewed in (Feng et al. 2025)). There is limited data from preclinical studies on Lys-alpha-oxidase as an anticancer agent, and almost no data on Pro-degrading enzymes as therapeutic agents. Thus, other strategies for amino acid depletion in combination modalities may be considered for Aze or ThLys.

The primary concern for practical use of AAAs is their presumable low selectivity between bacterial, fungal, and human cells, which can result in tissue accumulation, liver demage and autoimmune reactions after prolonged or repeated exposure. As discussed above, we propose limiting exposure to AAAs by depleting their standard amino acid counterparts and developing them rather as local (for example, topical) rather than systemic agents.

Alternatively, AAAs can be evaluated as antimicrobial agents also in combination with known antibiotics against, e.g., clinical isolates of pathogens with acquired antibiotic resistance that rely on de novo protein synthesis as a resistance mechanism. Hypothetically, AAAs could help overcome this type of antibiotic resistance (at least in part) by destabilizing nascent proteins, including those responsible for acquired resistance. Future research should establish whether this approach can provide a cost-effective way to overcome microbial drug resistance.

Author contributions

D.D.—the idea for the article, literature search, data analysis and manuscript writing, Olena S.—literature search, data analysis, manuscript writing, Oleh S.—conceptualization, supervision, manuscript writing and editing. All authors read and approved the final version of the manuscript.

Funding

This study was supported by the National Research Fund of Ukraine grant №2023.04/0048 “Development of a broad-spectrum antimicrobial agent based on the extract of the leguminous plant Indigofera spicata”, state registration number 0124U003831. This study was partially supported by “Presidential Discretionary-Ukraine Support Grants” from Simons Foundation, Award No 1030281 to O.G.S. and O.V.S.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.