Metabolism of Aldoximes and Nitriles in Plant-Associated Bacteria and Its Potential in Plant-Bacteria Interactions

In plants, aldoximes per se act as defense compounds and are precursors of complex defense compounds such as cyanogenic glucosides and glucosinolates. Bacteria rarely produce aldoximes, but some are able to transform them by aldoxime dehydratase (Oxd), followed by nitrilase (NLase) or nitrile hydratase (NHase) catalyzed transformations. Oxds are often encoded together with NLases or NHases in a single operon, forming the aldoxime–nitrile pathway. Previous reviews have largely focused on the use of Oxds and NLases or NHases in organic synthesis. In contrast, the focus of this review is on the contribution of these enzymes to plant-bacteria interactions. Therefore, we summarize the substrate specificities of the enzymes for plant compounds. We also analyze the taxonomic and ecological distribution of the enzymes. In addition, we discuss their importance in selected plant symbionts. The data show that Oxds, NLases, and NHases are abundant in Actinobacteria and Proteobacteria. The enzymes seem to be important for breaking through plant defenses and utilizing oximes or nitriles as nutrients. They may also contribute, e.g., to the synthesis of the phytohormone indole-3-acetic acid. We conclude that the bacterial and plant metabolism of aldoximes and nitriles may interfere in several ways. However, further in vitro and in vivo studies are needed to better understand this underexplored aspect of plant-bacteria interactions.


Introduction
Oximes (RR -C=N-OH) are metabolites that cross the boundaries between the kingdoms of life. They are widely distributed, especially in higher plants. Knowledge about plant oximes has been recently summarized [1]. Briefly, the plant oximes are largely aldoximes (R-CH=N-OH) or their methylated analogs (R-CH=N-OMe). They are derived from a few amino acids (Val, Leu, Ile, Met, Trp, Phe, and Tyr) and analogs of some of these acids. Aldoximes are intermediates in the biosynthesis of defense compounds (cyanogenic glucosides (CGs), glucosinolates (GSLs), and camalexin), but also per se have some biological activities. The functions of plant oximes are numerous, but not all of them are well understood. They include defense against pathogens and herbivores, growth control, radical scavenging, interaction with pollinators, interspecies communication, and nitrogen supply [1,2].
Oximes are rarely produced in bacteria. An exception is Bacillus sp. OxB-1, which forms phenylacetaldoxime (PAOx) from N-hydroxy-L-phenylalanine [4]. The latter is likely a metabolite of L-Phe but the enzyme that catalyzes the conversion of L-Phe is not known. The same strain is able to metabolize PAOx and other oximes (see below).
The industrial potential of the above enzymes has been summarized many times (e.g., [6][7][8][9]) but reviews on their natural roles are scarce. The in vivo functions of NLases were discussed by Piotrowski [3] and Howden and Preston [2], with the former work focusing on plant NLases. Aldoxime metabolism was briefly tackled in these works.
The metabolism of the phytohormone indole-3-acetic acid (IAA), which, under certain conditions, proceeds via indole-3-acetaldoxime (IAOx) or indole-3-acetonitrile (IAN), was reviewed previously [10]. However, the focus has been on other pathways to IAA that are more common (Section 4). The role of IAA in plant-bacteria interactions was also discussed in the study [10] and another review [11].
The above-mentioned review of natural, mainly plant oximes focused on the structures, distribution, biosynthesis, and biological activities of these compounds [1].
However, a review focusing on the role of Oxds and nitrile-converting enzymes in plant-bacteria interactions is lacking. The metabolism of oximes and nitriles is evidently important for plant-bacterial cross-talk, resulting in metabolites found in both microbes and plants. Therefore, in this survey, we will summarize what is known about the activities of bacterial Oxds, NLases, and NHases toward plant compounds.
We will also analyze the taxonomic and ecological distribution of these enzymes. In addition, we will discuss their role in some plant-growth-promoting bacteria and phytopathogens. Finally, we propose possible topics for future studies in this field.

Substrate Specificity
The transformation of the aldoxime IAOx into the corresponding nitrile in microbes (fungi) was noticed in the 1960s [12]. However, the discovery of the first Oxds was reported only about two decades ago. These were Oxds from Rhodococcus sp. YH3-3 [13], Bacillus sp. OxB-1 [14], and Pseudomonas chlororaphis B23 [15]. NLases have been known since the 1960s and their spectrum has been expanded since then [6,16]. The first report on NHase appeared in 1980 [17]. Previous reviews largely focused on the use of the enzymes for industrially important reactions [5,7,9]. In the following, we analyze the data on the substrate specificity of these enzymes for plant oximes and nitriles.

Aldoxime Dehydratases
Oxds are unique heme enzymes that catalyze the dehydration of aldoximes in an aqueous environment. The catalytic Fe(II) atom is essential for the correct binding of the oxime via the N-atom of the oxime ( Figure 2); Fe(III) is inactive. Figure 2. The binding of the substrate, phenylacetaldoxime, in the catalytic center of aldoxime dehydratase and release of the product, phenylacetonitrile; according to [18], modified. Amino acid numbering as in OxdA from Pseudomonas chlororaphis B23. Substrate and product in red.

OxdB
OxdB of Bacillus sp. OxB-1 [14], OxdK of Pseudomonas sp. K-9 [21], and OxdRG of Rhodococcus globerulus A-4 [22] transformed IAOx, PAOx, PPOx, and at least one aliphatic oxime (Table 2). However, their preferred substrates are different: arylaliphatic aldoximes (PPOx, PAOx) in OxdB and OxdRG and the aliphatic compound iVaOx in OxdK. IAOx is an acceptable substrate for OxdB and OxdRG, but OxdK has a very low activity for it. The K M values are relatively high (≈0.5−5.5 mM) in all cases. OxdRG showed lower V max values [22] than the other Oxds. However, the activity increased under altered conditions [21]. In particular, anaerobic conditions were important to prevent the oxidation of the Fe(II) cofactor. Thus, the specific activity increased from less than 0.1 U/mg to 633 U/mg for Z-PAOx [21]. However, kinetic data were not reported for these conditions. OxdBr1 of Bradyrhizobium sp. LTSPM299 was found to be similar to OxdB and OxdRG in terms of substrate specificity [24]. The Oxd of Rhododoccus sp. YH3-3 was tested with whole cells. iBuOx was dehydrated, but not IAOx, PAOx, or PPOx [28]. Kinetic data were not determined for these enzymes.
Two hypothetical genes were found upstream of the oxd gene in Bacillus and Bradyrhizobium [23,24]. They probably encode an NLase and a transcriptional regulator NitR, which belongs to the AraC family. The clusters in Rhodococcus and Pseudomonas are more complex. Both encode two NHase subunits, an amidase, and the NHase activator P47K [14,15]. The NHases in P. chlororaphis B23 and R. erythropolis N-771 are of Fe-type (with an Fe(III) cofactor) [29,30]. The NHase activator likely incorporates Fe into the active site [31]. However, its homolog in Pseudomonas sp. was found to be not necessary for NHase activity [32].
The substrate specificities of the NHases linked to Oxds are partially known. The NHase of P. chlororaphis transforms n-alkyl nitriles but not iVaN or PAN [29]. In contrast, the NHase from R. erythropolis N-771 was found to have a good activity for iBuN [30]. The NHase in Rhododoccus sp. YH3-3 was also characterized [28]. It belongs to the Co-type (with a Co(III) cofactor). It showed high activity for PAN, but lower activities for iBuN, IAN, and 3-phenylpropionitrile (PPN). Thus, some of the nitriles produced by the Oxds are substrates of NHases.
In contrast, less is known about the substrate specificity of the above NLases. The NLase of Bacillus sp. hydrolyzed PAN [14]. The NLase of Bradyrhizobium sp. LTSPM299 was not studied but it has a close homolog with activity for mandelonitrile [24]. This homolog was expressed from metagenomic DNA [33]. In addition, another homolog was characterized in Bradyrhizobium japonicum USDA 110 [34], which was reclassified as Bradyrhizobium diazoefficiens [26]. The best substrates of this homolog were found to be mandelonitrile and PAN [34].

Taxonomic and Ecological Distribution
The distribution of Oxds in bacteria was assessed based on activity and gene screening [35,36] and on data from the GenBank database [26] ( Table 3).
The screening of Oxd activities included 540 bacterial strains and 435 fungal strains including yeasts [35]. The substrates tested were PAOx and pyridine-3-aldoxime. The latter was not one of the plant oximes [1]. Therefore, we focus on the activity for PAOx. This was present at significant levels in over twenty bacteria and over thirty fungi including two yeasts. The study also examined the hydrolysis of the corresponding nitriles. These activities were generally present in the strains with Oxd activities. The Oxd activities were most abundant in rhodococci.
In addition, the distribution of Oxds in some bacteria selected by the above screening was investigated by Southern hybridization and PCR amplification [36]. The hybridization procedure used probes based on oxd genes from R. erythropolis N-771. This allowed for the identification of oxd genes in Rhodococcus sp. NCIBM 11215 and NCIBM 11216, R. erythropolis strains JCM 3201, BG13, and B16, Brevibacterium butanicum ATCC 21196, and Pseudomonas sp. K-9. These strains and Bacillus sp. OxB-1 also gave positive results in PCR amplification with primers based on conserved oxd regions. PCR was also performed with primers for Fe-type and Co-type NHases. The nha1 genes ( Figure 4) were found in almost all strains of rhodococci shown to have oxd genes, in Pseudomonas sp. K-9, and in B. butanicum ATCC 21196. All of these NHases were identified as Fe-type.
According to GenBank searches, Oxds are most abundant in the phyla Actinobacteria and Proteobacteria. The phyla Firmicutes and Bacteroidetes contain few Oxds.
Among the Actinobacteria, the greatest number of putative Oxds was found in Rhodococcus, in accordance with the above-mentioned activity screening. A few were also found in Arthrobacter, Corynebacterium, Microbacterium, or Streptomyces. These bacteria are widely distributed in water and soils, typically in the rhizosphere [37][38][39]. They also include endophytic bacteria belonging to Rhododoccus [40] or Streptomyces [41].   1 Retrieved from [26]. Searches for hypothetical Oxds were performed using OxdBr1 (WP_044589203 [24]) as a template. Searches for hypothetical NLases and NHases were performed using the NLase from Bradyrhizobium diazoefficiens (NP_770037 [34]) and the α-subunit of the NHase from Rhododoccus erythropolis N-771 (2AHJ_A [42]) as templates, respectively. 2 According to [35]. The active strains exhibited both aldoxime dehydratase and nitrile-hydrolyzing activity for phenylacetaldoxime and phenylacetonitrile, respectively; n.a., not assayed; n.d., not detected. 3 Other four Oxds were identified in a bacterium of the Chitinophagaceae family in Bacteroidetes.
In the Firmicutes, most of the Oxds were found in Bacillus, Brevibacillus and the related genus Siminovitchia, while some other genera contain a single Oxd. Bacilli have a number of different habitats including the rhizosphere and also include endophytic bacteria [45].
In general, the genera with oxd genes also contain nitrilase (nit) genes. NLases are much more diverse than Oxds in terms of substrate specificity. There are several types of NLases: aromatic NLases, arylacetoNLases, plant NLase homologs, thermostable NLase, etc. [2,50]. In addition, some putative NLases differ significantly in sequence from characterized NLases, and thus may have other substrate specificities. NLases are distributed throughout bacterial phyla. The highest number of NLase sequences was found in Rhododoccus, Streptomyces, Paraburkholderia, Pseudomonas, Rhizobium, Variovorax, and Flavobacterium. The typical habitats of these bacteria are soils.
NHases are more conserved; Fe-type and Co-type NHases are about 50% identical. They are less widely distributed than NLases, as only a few were found in Firmicutes and Bacteroidetes. In Actinobacteria and Proteobacteria, their distribution is similar to that of NLases, with a few exceptions.

Potentiality of the Aldoxime-Nitrile Pathway in Plant-Associated Bacteria
The bacterial ability to detoxify oximes and nitriles appears to be advantageous in both mutualistic and parasitic lifestyles. Bacterial symbionts may also benefit from this ability in different ways, such as modulating plant growth through interactions with plant hormone synthesis.
The interaction of plant-growth-promoting bacteria with the host plant is complex. The beneficial effects are associated with the bacterial ability to produce IAA, nutrients (ammonia, phosphate), surfactants, siderophores, or enzymes that attack the cell wall polysaccharides in fungal pathogens [51]. Other mechanisms strengthen the plant cell wall, prevent an excessive production of the plant hormone ethylene [51], or change the distribution of the plant hormone abscisic acid [52].
In the pathway to ethylene (Figure 5), the intermediate 1-aminocyclopropane-1carboxylate (ACC) can be degraded by the bacterial ACC deaminase [52] (EC 3.5.99.7). If not degraded, ACC is transformed to ethylene and cyanoformic acid by the plant ACC oxidase (EC 1.14.17.4); cyanoformic acid decomposes to CO 2 and HCN. The latter is eliminated in the plant by reaction with L-Cys to form β-cyano-L-alanine (AlaCN). The enzyme that catalyzes this reaction (AlaCN synthase, EC 4.4.1.9) is widespread in plants but has been found in few bacteria [53]. The next step is the hydrolysis of AlaCN [2]. Plant NLases of the NIT4 type are highly specific for AlaCN [2,3]. The presence of similar NLases in bacteria such as Pseudomonas fluorescens SBW25 [2,54] and fungi [55] suggests the possibility of horizontal gene transfer (HGT). HGT from bacteria to fungi was already demonstrated for ACC deaminase (acdS) genes [56]. The substrate specificity distinguishes NIT4 from NLases typical of the aldoxime-nitrile pathway.  [2,52], modified). The synthesis of AlaCN proceeds in plants and, rarely, in bacteria. The hydrolysis of AlaCN proceeds in plants but also in some bacteria and fungi. NIT4, nitrilase specific for AlaCN (plant enzymes in black, bacterial enzymes in red).
The Trp-dependent pathways to the phytohormone IAA involve the indole-3-acetamide (IAm), indole-3-pyruvate (IPyA), tryptamine (TAm), Trp side-chain oxidase (TSO), and IAN pathway ( Figure 6). The preferential pathways in plants and microbes are different. Plants have been assumed to mainly synthesize IAA via IPyA, tryptamine, and probably also IAM and IAN. Currently, the most studied metabolic pathway is the Trp-IPyA-IAA pathway catalyzed by an amino transferase and a monooxygenase of the YUCCA family [57,58]. IAAld has been found not to be an intermediate of the IPyA pathway, as previously thought [59]. Bacteria have been assumed to largely produce IAA via the IAM and IPyA pathway, with some rare examples of the IAN, TAM, and TSO pathway [10]. It is still largely unclear to what extent IAA is formed in bacteria or plants from IAOx or IAN. Plants probably hydrolyze IAN to IAA under certain conditions (lack of sulfur, infection) [3]. Both bacteria and plants may also synthesize IAA by the Trp-independent pathways from indole or indole-3-glycerolphosphate [10].
IAA has versatile biological activities. It supports germination and growth, triggers plant resistance mechanisms, and also has nematocidal activity [60]. However, IAA only has plant-promoting effects within a restricted range of concentrations [51]. The regulation of IAA production in bacteria is also sensitive to environmental factors [10]. The production of IAA by microbes can interfere with plant defense and physiology in many ways.
The endophyte P. fluorescens BRZ63 showed biocontrol activity against some fungal pathogens (Rhizoctonia, Sclerotinia, Colletotrichum, and Fusarium) in Brassica napus (oilseed rape). The bacterium improved plant fitness by a complex mechanism that also included the production of significant concentrations of IAA (0.06 mg/mL) [51]. Another strain of P. fluorescens protected the host plant (Medicago trunculata, a legume) against the fungus Botrytis cinerea at lower IAA concentrations (0.022 mg/mL). The strains with lower IAA production (<0.001 to ≈0.01 mg) did not promote shoot growth, notwithstanding the fungal infection [61].
Production of IAA from IAN or IAOx in P. fluorescens is possible, as the corresponding genes are present. In addition, IAN can be produced from glucobrassisin in Brassicaceae [3]. On the other hand, P. fluorescens may also transform Trp to IAA via IAAld produced by direct oxidation of Trp under specific conditions (stationary phase) [10].
P. fluorescens has several tens of putative Oxds. Some of them are most similar to OxdK from Pseudomonas sp. K-9 (up to 97% identity) and some to OxdRG from R. globerulus A-4 (up to 78% identity). OxdK has a non-zero activity for IAOx, and OxdRG has a good activity for this substrate (Section 2). P. fluorescens also has dozens of NLases. One of them (AAW79573) is an arylacetoNLase with significant activities for PAN, IAN, or PPN, and also for mandelonitrile. In this strain, P. fluorescens EBC191, genes corresponding to the mandelate pathway were found in the proximity of the nit gene but no oxd gene was found [62]. Thus, it is an example of an NLase that is not part of the ANP. However, a putative indole-3-acetamide (IAm) hydrolase gene was present upstream of the nit gene. IAm hydrolase is an enzyme of the IAm pathway [10] (Figure 6). Functional expression of the genes for the mandelate pathway was confirmed by demonstrating the growth of the strain on mandelate as the sole source of N; this pathway, together with nitrilase, is thought to allow for bacterial utilization of mandelonitrile released from plants [62]. The rhizobacterium Variovorax paradoxus 5C-2 supports plant growth in a variety of ways, including ACC deamination ( Figure 5) and abscisic acid distribution between plant tissues in Pisum sativum [52]. It also produces IAA (e.g., 0.075 mg/L [52]) in media containing Trp [52,63]. However, Trp concentrations in plant exudates (the Trp source for rhizobacteria) may not be sufficient [64]. This allows us to consider alternative pathways of IAA synthesis such as the aforementioned Trp-independent pathway [10]. However, we also cannot exclude the synthesis of IAA from the plant-produced IAOx or IAN. V. paradoxus has 12 putative Oxds, all of which are relatively similar (≈50% identical) to the Oxds active for IAOx (OxdBr1, OxdRG). The species V. paradoxus also contains over 40 NLases. Most of them are closely related to the NLase Nit6803 from Synechocystis sp. PCC6803 [65], whose homologs are collectively referred to as the Nit6803 NLase family. The best substrate of this NLase is fumaronitrile, but a variety of structurally diverse substrates including IAN are also accepted [66].
Fungi are also able to transform oximes to carboxylic acids, which may add complexity to plant-microbial interactions. One of the oilseed rape (Brassica napus) pathogens, Sclerotinia sclerotiorum clone #33 [51], has an Oxd with a narrow specificity for a few aldoximes, while IAOx is its best substrate. However, the activity of the purified enzyme is low (0.014 U/mg protein) [19]. Another Oxd was characterized in Fusarium graminearum MAFF305135, which causes head blight in some crops. This enzyme showed high activity for IAOx (V max of 19.3 U/mg protein), but also for PAOx, PPOx, and aliphatic nitriles [20]. These are the only two fungal Oxds that have been characterized, but Ascomycota contain many more Oxds. Hundreds of fungal Oxd sequences have been deposited in GenBank, while less than ten were found in Basidiomycota. Ascomycota are also remarkably rich in NLases with several thousands of sequences, while approximately two hundred NLases were found in Basidiomycota [55,67]. Fusarium and its teleomorph Gibberella have dozens of NLases, some of which were characterized. They include several substrate-specificity types [6]. They are generally active towards PAN, whereas their specificity for other natural nitriles is still largely unclear.
The transformation of mutualistic bacteria to pathogens is relatively simple. For example, the pathogenicity of Rhodococcus fascians D188 depends on the production of cytokinins. This was enabled in the strains that had acquired the virulent plasmid [68,69]. R. fascians increases the IAA concentration in the plant by about tenfold, but the role of this effect in pathogenesis is largely unclear. No intermediates corresponding to the IAN, IAM, or TAm pathways were found at significant levels. The study suggested that the indole-3-pyruvate pathway ( Figure 6) is the main route to IAA in this organism [70]. Thus, the role of the approximately ten oxd genes found in this taxon is still poorly understood.

Conclusions
The long coexistence of plants and microbes has led to a variety of close relationships (symbiosis). The bacterial symbionts must have pathways that detoxify plant defenses. One of these pathways is ANP, which likely confers detoxification and assimilation capabilities to the bacteria. It may also have a signaling function. Here, we summarized the data that shed light on the role of this pathway in plant-associated bacteria and the interaction with their host plants. Moreover, NLases are much more diverse than Oxds in terms of sequence and substrate specificity. They can also act outside the ANP, e.g., in the degradation of the toxic plant metabolites AlaCN and mandelonitrile.
First, we analyzed data on the substrate specificity of the characterized bacterial Oxds. It has been shown that some plant oximes are Oxd substrates in vitro. We have also investigated whether the nitrile-converting enzymes in the same organisms are capable of converting the Oxd products-nitriles. This is likely for most nitriles, as indicated by the data on these enzymes or their homologs.
We then examined the distribution of Oxds using data from activity screening and database mining. It became clear that there are hundreds of oxd genes in bacteria, but they are unevenly distributed. Oxds are only abundant in four genera-Bradyrhizobium, Rhodoccus, Pseudomonas, and Variovorax-along with NLases or NHases. NLases are the most abundant. Oxds and nitrile-converting enzymes are common in both plant-growthpromoting bacteria and plant pathogens, and can alter plant growth or defense via, e.g., IAA production. However, studies that could explain the mechanisms of this plant-bacterial cross-talk are largely lacking. Funding: This work was funded by the Czech Science Foundation, grant number GF20-23532L, the Charles University Grant Agency, grant number 452120, and by the Austrian Science Fund FWF, grant number I 4607.