New Insights on the Glyphosate-Degrading Enzymes C-P Lyase and Glyphosate Oxidoreductase Based on Bioinformatics

Abstract

Bioremediation, the degradation of environmental pollutants by living organisms, has immense potential to lead to a greener planet. Bioinformatics analysis can contribute to the identification of novel microorganisms, which biodegrade contaminants, or of participating proteins and enzymes, and the elucidation of the complex metabolic pathways involved. In this study, we focus on C-P lyase and glyphosate oxidoreductase (Gox), two enzymes which degrade glyphosate, a widely used pesticide. Amino acid sequences of the two enzymes were collected from a broad range of microorganisms using the KEGG database and BLAST. Based on this, we identified additional lineages, with putative glyphosate-degrading activity, for which no glyphosate-degrading species have been reported yet. The conserved residues in each enzyme were identified via multiple alignments and mapped onto the 3D structures of the enzymes, using PyMOL, leading to novel insights into their function. As the experimental structure of Gox is still unknown, we created structural models based on three different programs and compared the results. This approach can be used to yield insights into the characteristics of potential glyphosate-degrading enzymes. Given the limited information available, such a step is important to gain further knowledge about them, which can contribute to their application in bioremediation in the future.

1. Introduction

Glyphosate, N-(phosphonomethyl)glycine, is an active ingredient in many widely used, broad-spectrum herbicides (e.g., Roundup^TM). It is a synthetic amino acid analogue of glycine incorporating a phosphonic acid [1]. It inhibits enolpyruvylshikimate-3-phosphate synthase (EPSP synthase), resulting in the inhibition of the shikimate pathway, which is essential for the biosynthesis of aromatic amino acids in plants [2].

Glyphosate has been characterized as a safe compound by the European Food Safety Authority (EFSA) for human health, animals, and the environment, as “no critical areas of concern” were confirmed. Additionally, the European Chemicals Agency (ECHA) does not classify glyphosate as carcinogenic or possibly carcinogenic [3]. On the other hand, it is considered as “probably carcinogenic” to humans (Group 2A) according to the International Agency for Research on Cancer (IARC) [1] and it is labeled as “toxic to aquatic life with long lasting effects” as per the Hazard Statements of the Globally Harmonized System of Classification and Labelling of Chemicals (GHS) [1]. Glyphosate was an approved pesticide until December 2023 in the E.U. [3], where its use was recently extended for 10 more years [4].

Nevertheless, there is increasing concern about the accumulation of pesticide residues and their metabolites in soils and food crops. In soil sampled from agricultural sites throughout the E.U. and other cooperating countries, glyphosate and its metabolite alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) were among the most frequently detected substances and also among those with the highest concentrations [5]. A case study in central Germany yielded similar results, as glyphosate and AMPA were detected at the highest concentrations in surface water and topsoil samples [6]. Concerns have also been raised about the persistence of glyphosate and its metabolite AMPA in soil, especially in fields where glyphosate-based herbicides have been extensively used [3,7], its potential to bioaccumulate in terrestrial organisms [8], and its impact on crop health [3], while it has also led to glyphosate-resistant weeds [9]. Both compounds are transported off site with rainfall, contaminating surface water, hedgerows, and adjacent non-treated fields [7,8].

In particular, glyphosate can affect the microbial population, and hence, have an impact on plant-beneficial microbes [10] and cause metabolic changes in plants, even at low doses [11]. Glyphosate and its salt form, isopropylamine (IPA), also act as chelating agents by binding to metal cations necessary for plant nutrition [12]. Additionally, studies have shown potential negative effects on insects; i.e., rendering insects more prone to infections, inhibiting the production of melanin, disturbing their gut microbiome [13,14,15,16], and leading to a low responsiveness to sucrose [17] and impaired aversive learning [18]. Aquatic organisms are also affected: in zebrafish and other commercial fish, such as Nile tilapia, glyphosate has been linked to developmental problems [19], cardiovascular toxicity [20], oxidative stress and anxiety [21], and kidney disfunction [22]. Glyphosate impacts aquatic ecosystems via changes in the abundance and composition of freshwater picoplanktonic populations [23] and certain species of marine phytoplankton [24]. Glyphosate can have a negative impact on mammals by altering gene expression and affecting DNA methylation in human cells [25,26,27]. The glyphosate-based herbicide Roundup^TM, glyphosate, and AMPA can indirectly cause DNA damage by inducing reactive oxygen species (ROS) production in human peripheral blood mononuclear cells [28]. In addition, exposure of human lymphocytes to glyphosate in higher concentrations (200 μmol/L) than those normally found in the environment (20–40 μmol/L) demonstrated genotoxic effects [29]. It has also been associated with increased urinary glyphosate and AMPA levels in humans [30], which, in turn, have been associated with longer telomere length in leukocytes [31]. Low chronic maternal exposure to glyphosate has also been associated with reduced immune response in mice [32].

The degradation of glyphosate can be either biotic, by microorganisms in both aerobic and anaerobic conditions, or abiotic, by Mn-containing minerals [33], and depends on many factors, e.g., whether it is found in water or soil, soil properties, weather conditions, and cultivated plants in the area. Regarding its degradation rate, half-life estimates range from 5 to 180 days [34].

There is limited current knowledge about the pathways involved in phosphonate degradation, including glyphosate, their regulation, and the relevant enzymes [35]. Three enzymes have been reported to degrade glyphosate: C-P lyase, which cleaves the bond after phosphorus; and glyphosate oxidoreductase (Gox) and glycine oxidase (GO), which both cleave the C-N bond. The last two enzymes produce the same products but their function is based on a different mechanism (Figure 1).

C-P lyase is a protein complex that consists of two copies each of four proteins (PhnG, PhnH, PhnI, PhnJ) and one ATP-binding subunit, PhnK. The degradation of glyphosate by C-P lyase, which takes place under inorganic phosphate (Pi) deficiency [35], results in the production of sarcosine and phosphate. The C-P lyase complex is part of the C-P lyase pathway, which converts phosphonate into PRPP (5-phosphoribosyl-a-1-diphosphate) using ATP [35]. According to the literature, it is widely present across bacteria, and shows broad substrate specificity, as it cleaves the C-P bond of organophosphonates [36]. At first, the tertiary structure of the complex was determined without PhnK (PDB accession number: 4XB6) [37]. A few years later, PhnK was also included in the complex via single-particle cryogenic electron microscopy, which provided valuable insights into the interactions of PhnK with the core complex and the conformational changes before and after utilizing ATP (PDB accession number: 7Z19, with ATP: 7Z18), and also, with PhnL in various conformations (PhnK/PhnL dual ABC dimer, PDB accession numbers: 7Z15, 7Z16) [38]. PhnJ is the active site of the enzyme (Figure S1). It contains an Fe₄S₄ complex and a Zn molecule, which is stabilized by three conserved Cys residues (Cys241, Cys244, and Cys266), found within the conserved motif CX2CX21C. Two more residues are important for the reaction mechanism of the complex: Gly32 and Cys272 [39]. Due to the presence of the fourth conserved Cys residue, it can be included as part of the aforementioned motif, namely, as CX2CX21CX5C [39]. Cys272 is responsible for the generation of a thiyl radical, which cleaves the C-P bond of 5-phosphoribosyl-1-phosphonate. Notably, PhnJ does not contain the signature radical SAM enzyme motif (CX3CX2C), which stabilizes the Fe₄S₄ complex [39]. A second potential active site is located between PhnJ and PhnI, which includes the residues His328 and His333 (PhnI) and His108 (PhnJ). Four further domains (Figure S1) have been described [37]:

• CMD: C-terminal mini domain (PhnJ);
• CID: Central insertion domain (PhnJ), which could participate in the binding of PhnK to PhnJ;
• BBD: Beta barrel domain (PhnI);
• NTD: N-terminal domain (PhnI).

It has been proposed that not all C-P lyases are specific to glyphosate, and that C-P lyase of Eschericha coli is an example of an (amino)alkylphosphate-specific enzyme without glyphosate-degrading activity [35,40]. For this reason, there is the hypothesis that two different types of independently induced C-P lyases exist [35]. However, later research on Ochrobactrum anthropi and Achromobacter sp. [41] could not provide evidence for the hypothesis of two distinct C-P lyases (Sviridov A, personal communication).

The glyphosate oxidoreductase (Gox) gene was first characterized from Achromobacter sp. LBAA, a bacterial strain isolated in a glyphosate waste treatment facility, as part of a Monsanto patent [42], in order to construct glyphosate-resistant plants. Gox is a monomeric FAD-dependent enzyme, cleaving the C-N bond of glyphosate and yielding glyoxylate and the metabolite AMPA. It is speculated that the reduction of FAD takes place in the active site of the enzyme by glyphosate [42]. There is no known experimental structure of Gox, although a computed structural model is available [43].

Glycine oxidase (GO) is the third enzyme known to degrade glyphosate. Several GO structures are available in the PDB (e.g., 3IF9, 1NG4 and 1RYI from Bacillus subtilis). It is a homotetrameric enzyme and its substrates include glycine, neutral D-amino acids, and sarcosine. Similar to Gox, it is an FAD-dependent enzyme and it cleaves the C-N bond, yielding the same degradation products: AMPA and glyoxylate. However, Gox and GO show low sequence identity [2] and, based on oxygen utilization, appear to have a different reaction mechanism: oxygen reduction in Gox does not result in the production of hydrogen peroxide, which is the case for GO [2,42]. Several studies have focused on engineering GO in order to enhance its glyphosate-degrading activity [2,44,45,46,47].

Given the negative consequences of glyphosate, in this study we look at the characteristics of the proteins of the glyphosate-degrading enzymes at a sequence level as a way to understand and optimize their function, by utilizing bioinformatics tools. As GO has been extensively studied and improved through mutagenesis, we focus on C-P lyase and Gox. Further insights into the enzymatic degradation of glyphosate by bacteria can contribute to the eventual application of bioremediation techniques for a faster removal of glyphosate from affected areas.

2. Materials and Methods

Information on the bacterial species reported to degrade glyphosate and the pathways or enzymes involved in this activity was collected from relevant references in PubMed [48] and ScienceDirect [49] (Table 1). These databases, along with PDB [50], were also used to find the already-available information on enzymes responsible for glyphosate degradation, namely, Gox, C-P lyase, and GO.

For C-P lyase, the amino acid sequences for each subunit of the complex were collected using the KEGG Orthology (KO) database [51] and the following entries: PhnJ: K06163, PhnI: K06164, PhnH: K06165, PhnG: K06166, and PhnK: K05781. The available sequences were parsed in such a way that a wide variety of representatives from across the bacterial kingdom were included; the selection was performed manually, using the taxonomy option of the KO entry and selecting at least one representative per family. The selected amino acid sequences are available in the Supplementary Dataset in folder “C-P lyase”. The sequence selected from each taxonomical group was random and we also included the sequence of E. coli. MUSCLE [52] was used for the multiple alignments, and the highly conserved residues were marked onto the 3D structure using PyMOL (linux version 2.5.2) [53]. The absolutely conserved residues were selected based on Jalview [54] (scoring 11); residues in positions with a Jalview score of 10 were also selected, which reflects that, despite variation, physicochemical properties are conserved [54]. The ScanProsite tool (https://prosite.expasy.org/scanprosite/, accessed on 15 March 2022) was used to search for motifs in the PhnJ sequences of the C-P lyase complex.

As there is no KEGG Orthology (KO) group for Gox, BLASTp [55] searches for orthologs in other species were performed using as a query the Gox amino acid sequence of Ochrobactrum sp. G-1 (ACZ58378.1). Two different databases were used:

• The nr database (BLASTp_nr), which includes all non-redundant GenBank CDS translations and also entries from PDB, SwissProt, PIR, and PRF, excluding environmental samples from WGS projects.
• RefSeq (BLASTp_ref), which contains only curated entries, i.e., one representative or “Select” transcript for every protein-coding gene, based on specific criteria [55].

Two different BLASTp searches were performed, with default settings, for each database (Table S1), one without any filters (BLASTp_nr_NOGROUP, BLASTp_ref_NOGROUP) and one for each bacterial group (BLASTp_nr_GROUP, BLASTp_ref_GROUP), as defined by Jun et al. [56], aiming at better sampling of more bacterial representatives. The selection of the resulting sequences for BLASTp_nr_NOGROUP and BLASTp_ref_NOGROUP was based on the first 100 sequences with the lowest E-value for each protein query. For the BLASTp_nr_NOGROUP, the highest E-value was 3 × 10⁻¹²⁸ and for the BLASTp_ref_NOGROUP 1 × 10⁻¹²⁷ (available in Supplementary Dataset). For BLASTp_nr_GROUP and BLASTp_ref_GROUP, the results with the lowest E-value for each protein query from each bacterial group were selected. MUSCLE [52] was used for the multiple alignments.

The Gox 3D structure was computed using I-TASSER [57] and Phyre2 [58], which are based on homology modeling, as well as AlphaFold Colab, which performs ab initio prediction [59]. The default settings were used. A general comparison of the results was made by aligning the structures with PyMOL. The results were assessed using QMEANDisCo, which is part of SWISS-MODEL [60].

3. Results and Discussion

3.1. Distribution of Bacterial Species Previously Reported to Degrade Glyphosate

In the literature, 42 different bacterial species have been reported to degrade glyphosate (Table 1). The majority of the species belong to α-, β-, and γ-proteobacteria, a smaller number to bacilli and actinobacteria, and some to bacteroidetes (Figure 2). The enzymes or genes involved in glyphosate degradation have been identified only in some research papers. In most cases, the possible mechanism was inferred based on the products resulting after glyphosate consumption by the bacteria, specifically, based on the presence of sarcosine (indicating C-P lyase activity) or AMPA (indicating Gox activity). In addition to the three known glyphosate-degrading proteins, the presence of a class II EPSPS synthase-encoding gene (aroA gene) has also been reported, although the experimental structure of the protein is not yet known and no further experiments were carried out to test and validate the activity of aroA [61]. Consequently, the knowledge about the genes and the structure or mechanism of the enzymes involved in glyphosate degradation is still limited.

Table 1. Bacterial species reported to degrade glyphosate. Depending on the evidence of glyphosate degradation in each paper, “G” represents the discovery of the responsible gene, “P” the presence of the degradation product, and hyphen (-) a decrease in the amount of glyphosate.

Group	Genus, Species	Evidence	Products/Genes Detected	Reference
actinobacteria	Arthrobacter sp. GLP-1	P	Sarcosine	[62]
actinobacteria	Streptomycete sp. StC	P	Sarcosine	[63]
actinobacteria	Streptomyces sp. StA	P	Sarcosine	[63]
actinobacteria	Rhodococcus soli G41	G	soxB gene	[64]
actinomycetes	Arthrobacter atrocyaneus ATCC 13752	P	AMPA	[65]
bacilli	Geobacillus caldoxylosilyticus T20	P	AMPA	[66]
bacilli	Bacillus cereus CB4	P	AMPA, glyoxylate, sarcosine, formaldehyde	[67]
bacilli	Lysinibacillus sphaericus	P	Free phosphorus concentration, glycine	[68]
bacilli	Bacillus aryabhattai FACU	G	goxB gene	[69]
bacilli	Bacillus. cereus 6 P	-	Conversion of glyphosate to polyphosphate	[70]
bacteroidetes	Flavobacterium sp. GD1	P	AMPA	[33]
bacteroidetes	Chryseobacterium sp. Y16C	P	AMPA	[71]
αctinobacteria	Arthrobacter atrocyaneus ATCC 13752	P	AMPA	[65]
α-proteobacteria	Agrobacterium radiobacter	P	AMPA	[33]
α-proteobacteria	Ochrobactrum anthropi GPK 3	P	AMPA	[72]
α-proteobacteria	Ochrobactrum sp. GDOS	P	AMPA	[73]
α-proteobacteria	Ochrobactrum sp. G1	G	gox gene	[42]
α-proteobacteria	Agrobacterium radiobacter SW9	P	AMPA	[74]
α-proteobacteria	Rhizobiaceae meliloti 1021	G	Homology to part of the phn gene cluster of E. coli	[75]
α-proteobacteria	Ochrobactrum sp. GDOS	P	AMPA	[73]
α-proteobacteria	Agrobacterium tumefaciens CHLDO	G	phn gene cluster	[76]
α-proteobacteria	Ochrobactrum haematophilum SR	-	-
α-proteobacteria	Ochrobactrum intermedium Sq20	G, P	Sarcosine, glycine, aroA gene (class II EPSPS)	[61]
β-proteobacteria	Comamonas odontotermitis P2	G	gox and phnJ genes	[77]
β-proteobacteria	Achromobacter sp. strain MPK 7A	P	Sarcosine	[72]
β-proteobacteria	Achromobacter sp. Kg 16	P	AcGP(N-acetylglyphosate), possibly AMPA	[72]
β-proteobacteria	Achromobacter Group V D (Agrobacterium sp. LW9)	P	AMPA	[74]
β-proteobacteria	Achromobacter sp. MPS 12A	P	Methane, sarcosine, glycine	[41]
β-proteobacteria	Alcaligenes sp. GL	P	Sarcosine	[78]
β-proteobacteria	Burkholderia vietnamiensis strain AQ5-12	-	-	[79]
β-proteobacteria	Achromobacter denitrificans SOS5	-	-	[76]
β-proteobacteria	Achromobacter insolitus SOR2	-	-	[76]
β-proteobacteria	Achromobacter xylosoxidans SOS3	-	-	[76]
β-proteobacteria	Achromobacter insolitus str Kg 19 (VKM B-3295)	G	phnJ gene	[80]
γ-proteobacteria	Pseudomonas sp. LBr	P	AMPA, a very small amount of glycine 5%	[81]
γ-proteobacteria	Providencia rettgeri GDB 1	P	AMPA	[82]
γ-proteobacteria	Pseudomonas sp. 4ASW	P	Sarcosine	[83]
γ-proteobacteria	Enterobacter cloacae K7	P	Sarcosine, glycine	[84]
γ-proteobacteria	Enterobacter sp. Bisph2	-	-	[85]
γ-proteobacteria	Pseudomonas pseudomallei 22	G, P	Phosphotransferase genes glpA and glpB	[86]
γ-proteobacteria	Pseudomonas sp. GLC11	-	-	[87]
γ-proteobacteria	Pseudomonas sp. PG2982	P	Sarcosine	[88]
γ-proteobacteria	Pseudomonas sp. SG-1		Glyphosate concentration decreased	[89]
γ-proteobacteria	Pseudomonas nitroreducens TR3	-	-	[76]

Table 1. Bacterial species reported to degrade glyphosate. Depending on the evidence of glyphosate degradation in each paper, “G” represents the discovery of the responsible gene, “P” the presence of the degradation product, and hyphen (-) a decrease in the amount of glyphosate.

Group	Genus, Species	Evidence	Products/Genes Detected	Reference
actinobacteria	Arthrobacter sp. GLP-1	P	Sarcosine	[62]
actinobacteria	Streptomycete sp. StC	P	Sarcosine	[63]
actinobacteria	Streptomyces sp. StA	P	Sarcosine	[63]
actinobacteria	Rhodococcus soli G41	G	soxB gene	[64]
actinomycetes	Arthrobacter atrocyaneus ATCC 13752	P	AMPA	[65]
bacilli	Geobacillus caldoxylosilyticus T20	P	AMPA	[66]
bacilli	Bacillus cereus CB4	P	AMPA, glyoxylate, sarcosine, formaldehyde	[67]
bacilli	Lysinibacillus sphaericus	P	Free phosphorus concentration, glycine	[68]
bacilli	Bacillus aryabhattai FACU	G	goxB gene	[69]
bacilli	Bacillus. cereus 6 P	-	Conversion of glyphosate to polyphosphate	[70]
bacteroidetes	Flavobacterium sp. GD1	P	AMPA	[33]
bacteroidetes	Chryseobacterium sp. Y16C	P	AMPA	[71]
αctinobacteria	Arthrobacter atrocyaneus ATCC 13752	P	AMPA	[65]
α-proteobacteria	Agrobacterium radiobacter	P	AMPA	[33]
α-proteobacteria	Ochrobactrum anthropi GPK 3	P	AMPA	[72]
α-proteobacteria	Ochrobactrum sp. GDOS	P	AMPA	[73]
α-proteobacteria	Ochrobactrum sp. G1	G	gox gene	[42]
α-proteobacteria	Agrobacterium radiobacter SW9	P	AMPA	[74]
α-proteobacteria	Rhizobiaceae meliloti 1021	G	Homology to part of the phn gene cluster of E. coli	[75]
α-proteobacteria	Ochrobactrum sp. GDOS	P	AMPA	[73]
α-proteobacteria	Agrobacterium tumefaciens CHLDO	G	phn gene cluster	[76]
α-proteobacteria	Ochrobactrum haematophilum SR	-	-
α-proteobacteria	Ochrobactrum intermedium Sq20	G, P	Sarcosine, glycine, aroA gene (class II EPSPS)	[61]
β-proteobacteria	Comamonas odontotermitis P2	G	gox and phnJ genes	[77]
β-proteobacteria	Achromobacter sp. strain MPK 7A	P	Sarcosine	[72]
β-proteobacteria	Achromobacter sp. Kg 16	P	AcGP(N-acetylglyphosate), possibly AMPA	[72]
β-proteobacteria	Achromobacter Group V D (Agrobacterium sp. LW9)	P	AMPA	[74]
β-proteobacteria	Achromobacter sp. MPS 12A	P	Methane, sarcosine, glycine	[41]
β-proteobacteria	Alcaligenes sp. GL	P	Sarcosine	[78]
β-proteobacteria	Burkholderia vietnamiensis strain AQ5-12	-	-	[79]
β-proteobacteria	Achromobacter denitrificans SOS5	-	-	[76]
β-proteobacteria	Achromobacter insolitus SOR2	-	-	[76]
β-proteobacteria	Achromobacter xylosoxidans SOS3	-	-	[76]
β-proteobacteria	Achromobacter insolitus str Kg 19 (VKM B-3295)	G	phnJ gene	[80]
γ-proteobacteria	Pseudomonas sp. LBr	P	AMPA, a very small amount of glycine 5%	[81]
γ-proteobacteria	Providencia rettgeri GDB 1	P	AMPA	[82]
γ-proteobacteria	Pseudomonas sp. 4ASW	P	Sarcosine	[83]
γ-proteobacteria	Enterobacter cloacae K7	P	Sarcosine, glycine	[84]
γ-proteobacteria	Enterobacter sp. Bisph2	-	-	[85]
γ-proteobacteria	Pseudomonas pseudomallei 22	G, P	Phosphotransferase genes glpA and glpB	[86]
γ-proteobacteria	Pseudomonas sp. GLC11	-	-	[87]
γ-proteobacteria	Pseudomonas sp. PG2982	P	Sarcosine	[88]
γ-proteobacteria	Pseudomonas sp. SG-1		Glyphosate concentration decreased	[89]
γ-proteobacteria	Pseudomonas nitroreducens TR3	-	-	[76]

3.2. Protein Conservation across Prokaryotes

The entries in KEGG for the proteins of the C-P lyase complex belong to a wide range of bacterial phyla (Figure S2A,C). The bacterial species expected to express C-P lyase based on KEGG (Figure S2) belong to more bacterial groups than the bacterial species reported in the literature as degrading glyphosate (Table 1, Figure 2). This suggests that there are more bacterial species with the potential to degrade glyphosate that have not been studied yet; it may also simply show conserved function against phosphonates, which may or may not include glyphosate. More specifically, the additional bacterial groups found after our KEGG search are actinobacteria, chloroflexi, clostridia, cyanobacteria, δ-proteobacteria, and ε-proteobacteria.

For Gox, as there were no relevant KEGG entries, we performed BLASTp and NCBI [90] keyword searches. We were able to find only 13 annotated Gox proteins. Those belong to α-proteobacteria, β-proteobacteria, and bacilli, but many of them are partial and the organism is not yet known (Table S2). The BLASTp search results included mostly FAD-dependent oxidoreductases, D-amino acid dehydrogenases, and glycine oxidases. Based on the percent identity of the BLASTp results, the sequences with the general term “FAD-dependent oxidases” showed higher sequence similarity to the annotated Gox sequences than the “D-amino acid dehydrogenases” and GOs; GOs demonstrated the lowest average percent identity (Figure S3).

The BLASTp searches for Gox without filtering per group showed fewer bacterial groups than those reported in the literature, namely, sequences only from α-, β-, and γ-proteobacteria (Figures S2B, S4 and S5). However, when we performed the searches per bacterial group, the diversity of the results increased significantly (Figure S2B,D). In addition to the aforementioned groups, we found entries belonging to verrucomicrobia, fusobacteria, aquificae, and planctomycetes. Figures S4 and S5 show distinctive changes in the phylogenetic distribution of the BLASTp results. Although there are no known motifs of Gox responsible for glyphosate degradation, these BLASTp results are worth investigating further for such a potential.

3.3. Conserved Residues in C-P Lyase

Multiple alignments of each of the proteins of the C-P lyase complex revealed a large number of conserved residues (Figure S6). The results were compared with the C-P lyase complex of E. coli (4X6B), which has been extensively studied previously. Although E. coli cannot degrade glyphosate, PhnI and PhnJ proteins show a high degree of conservation with glyphosate-degrading bacteria (Sinorhizobium meliloti) [75]. A detailed description of the discovered conserved residues, previous mentions in the bibliography, and the secondary structure of each region are found in

Table 2. Conserved residues of the C-P lyase complex of E. coli, identified in alignments from this study, compared to those described in the literature, for the proteins PhnJ, PhnI, and PhnG. The blue highlighted residues are the unique ones found in this study, which are not mentioned in the literature. The green highlighted residues are those mentioned in the literature which were not highly conserved in our alignments (i.e., scores 10 or 11 based on Jalview, see methods). The asterisk (*) denotes that P36 is at the border of loop2 and a b-sheet that precedes the loop.

PhnJ
Secondary Structure	MUSCLE Alignment Residues Identified in Alignments from This Study		Conserved Residues as Given by Seweryn et al. [37]	Residues Predicted In Silico to Participate in Interaction with AMPA and GLP [43]. HB: H Bond HI: Hydrophobic Interaction
Residues Surrounding the Fe4S4 Cluster
loop 1	A207, G208, R209, E210		-	AMPA: G208 (HB), R209 (HB)
loop 2	P36 *, P40, P43 Near the cluster: G47, G49		-	AMPA: G47 (HB), G49 (HB)
loop 3	Q124, V125, P126		-	AMPA: V125 (HB)
helix 1	T50, G51, G52 More distant: Q54		-	AMPA: T50 (HB), G51 (HB)
Residues surrounding Zn
loop 4	G245		C241, C244	Y250 (HI); more distant to Zn, S267 (HB)
β sheet loop			C266
helix6			C272
Residues surrounding universally conserved residue G32
loop 5	P31, Q34		G32
CID—Central insertion domain—contacts both of the central PhnI molecules
loop 3	L131, E135		129–169
helix 2	H145
helix 3	Y150, L157
CMD—C-terminal mini domain—stabilized by Zn ion and conserved C residues
β hairpin	E253 (loop), G260 (β sheet)		236–281
His108 coordinate Zn ion with 2 His from PhnI
loop 6	R107, R109 More distant: P111, F112, L115		H108	GLP: H108 (HB), R109 (HI)
loop 7	P187		-
helix 5	P189, D192		-	GLP: P189 (HI)
loop 8	D70, Q71, G72		-	GLP: Q71 (HB)
Two of the residues (PhnI His328 and His333) coordinate the zinc ion directly PhnJ His108, PhnI His328, and PhnI His333 are located in a cavity between PhnI and PhnJ
β strand 1	K67, V68, I69, D70
loop	Q71, G72
β strand 2	I104, Q105, R107
loop	R109		(H108)
loop (Bhatt)	P174, F232			Y171 (HI)
PhnI
Secondary structure	MUSCLE alignment Residues not reported by [37], identified in alignments from this study			Conserved residues [37]
Residues surrounding the Fe4S4 cluster
loop 1	A55	G278		-
loop 2
NTD—N-terminal domain
β strand 1	Y2, A4			1–88
helix 2	G7, G8, A11
helix 3	E51
loop 1	A55
helix 4	A60, A63, Q66
loop 4	G69
helix 5	E73, A74, R82, T84
β strand 2	K87
His108 coordinate Zn ion with 1 His from PhnJ
loop 5	P332			-
helix 6	Y334			H333
helix 7	F325, K330 Further away: D318, G324			H328
C-terminal
helix 8	R189, R198, Y209, R213			188–318
loop 6	G214, G216, H219
β strand 3	G223, E224, R226, G228
β strand 4	G246, E252
β strand 5	G272, G274
loop 7	G278
helix 9	E281, K283, M287
helix 10	F312, H316
helix 11	D318
PhnG
Secondary structure	MUSCLE alignment Residues not reported by [37], identified in alignments from this study			Conserved residues [37]
β-hairpin and C-terminal helix form a molecular clamp that connects to a groove in PhnI
β strand 1	G63	-		-
β strand 2	G80	-		-
helix	A96, D99, A100	-		-
loop	V139, F141	-		-

. The position of the highly conserved residues on the tertiary structure of the enzyme is highlighted in Figure S7.

From our alignments, many previously unreported conserved residues were discovered, which could be significant for the function of the enzyme. Some of these residues are close to the active site of PhnJ, or are part of the domains CMD, CID (PhnJ), BBD, and NTD (PhnI). The function of certain conserved residues has been suggested previously and, by proximity, we can deduce the function of several of the newly identified conserved residues. For example, residues G208, R209, G47, G49, V125, T50, and G51 were previously found, using molecular docking, to interact with AMPA via hydrogen bonds [43]; we find conservation of these residues, which lends support to their important role, and we suggest that they also surround and may help stabilize the Fe₄S₄ cluster. Also, in addition to the Cys residues surrounding Zn, Gly245 is conserved right next to Cys244, suggesting that this may also contribute to stabilizing Zn. The loop containing the universally conserved Gly32 also contains two more conserved residues (Pro31, Gln34), which could participate in receiving the intermediate radical while catalyzing the substrate formed by SAM. The fact that the CMD domain only contains three conserved residues matches the hypothesis that it might detach from the core of PhnJ during interaction [37]. Conserved His residues in the interface of PhnJ and PhnI also coordinate a second Zn molecule. From our alignments, more residues were found to be conserved in this region in both the PhnJ and PnhI proteins, suggesting that they contribute to the overall structure around Zn. The large number of highly conserved residues supports the assumption that this could be a potential second active site of the C-P lyase complex [37]. PhnG is generally not highly conserved. However, there are a small number of conserved residues in the C-terminal part, which is reported to interact with PhnI [47]. Notably, there is a set of conserved residues, remote from the rest of the C-P lyase core complex, which does not seem to interact with any substrate or other proteins of the complex.

Most residues which are reported to be conserved in the literature and play a significant role in the function of the enzyme were also conserved in our alignments. However, five residues of PhnJ that are expected to interact with glyphosate and AMPA in silico [43], and three of the residues participating in the conserved Cys motif of PhnJ, are not absolutely conserved in our alignment. This discrepancy may be because we used more sequences in the alignments compared to Seweryn et al. [37]. Cys244 and Cys266 are present in 96% of the sequences, but they are replaced by amino acids with the same properties (Ser or Thr and Ile or Thr, respectively). The most important change is noticed for Cys272, which plays a significant role in the function of the enzyme as it is the last Cys residue of the conserved motif CX2CX21CX5C [37]. According to our results, position 310 of the multiple alignment, which corresponds to Cys272 of E. coli, was not absolutely conserved in all sequences and it was replaced by either Met or Leu. The sequences that did not correspond to Cys in this position were:

• cyb—Synechococcus sp. JA-2-3B’a(2-13) (bacteroidota)—Met269;
• muh—Mucilaginibacter celer HYN0043 (bacteroidota)—Leu276;
• shv—Salinicoccus halodurans H3B36 (bacilli)—Met269;
• crn—Carnobacterium sp. 17-4 (bacilli)—Met27.

This raised the question of whether the motif is present in another part of the sequences, and whether the rest of the sequences actually have the conserved motif. Although Cys was conserved in this position for the rest of the entries, in some cases the number of residues between the conserved Cys was different than that determined for the CX2CX21CX5C motif. For this reason, we searched for the motif CX2CX21CX5C in all of our sequences: it was found in only 77 out of 93 PhnJ sequences in total. The 16 sequences in which the motif was not present are shown in Table S3. However, as the typical motif for SAM enzymes is CX3CX2C, and variations of this motif (CX4CX2C, CX5CX2C, CX3CX3C, CX14CX2C) can also be observed [91], we also searched for these motifs in our sequences. The CX3CX2C SAM typical motif was not found in any of the sequences, and none of the sequences without the CX2CX21CX5C motif contain any alternative motif. Only two sequences contained the alternative motif CX5CX2C and, interestingly, both have the motif of PhnJ too: hbe—Halomonas beimenensis NTU-111 (γ-proteobacteria), and stea—Sporolactobacillus terrae DRG1 (bacilli).

3.4. Structure of Gox

There is no available experimental structure for Gox in PDB. The only attempt at a structural approach used Phyre2, based on the amino acid sequence of Ochrobactrum sp. G1 (ACZ58378.1) [43]. We created structural models based on AlphaFold Colab and I-TASSER, which present similarities with Phyre2 (Figure 3). According to the evaluation based on QMEANDisCo (Figure S8), the best model is produced by AlphaFold, a conclusion that is supported by the recent assessment of structure prediction models (CASP14), referring to the performance of AlphaFold [92]. Phyre2 and AlphaFold show significant differences in specific regions of the predicted structure, which end up around the surface of the structure (Figure 3 and Figure S9). Based on the docking results of Bhatt et al. [43], glyphosate is expected to bind to the center of the protein. There are slight differences between the three models; for example, R358 is part of a loop in I-TASSER but on a beta sheet in AlphaFold and Phyre2, and M388 is in a helix in Phyre2, but part of a loop in I-TASSER and AlphaFold. As glyphosate is a small molecule and these accumulative small changes may alter the shape of the binding site, this can have an impact on the functionality of the proposed binding site on each calculated structure. It is also not known yet if Gox acts as a monomer or dimer/polymer; in the latter case, the peripheral structure may also play a role in subunit interactions.

The residues hypothesized by Bhatt et al. [43] to interact with glyphosate were not consistently conserved in our multiple alignments (

Table 3. Conservation of residues hypothesized by Bhatt et al. [43] to interact with glyphosate in our multiple alignments. Green crosses show the residues of the computed binding site that are highly conserved in each of our own multiple alignments (corresponding to a Jalview score of 10 or 11).

Bhatt et al., 2021	BLASTp_nr_NOGROUP	BLASTp_ref_NOGROUP	BLASTp_nr_GROUP	BLASTp_ref_GROUP
G14		+		+
I15
V16
A43		+
S44
N47	+	+
R358	+	+
G384
G387	+			+
M388

). None of the residues were conserved for all datasets, and only N47 and R358 were highly conserved for the results of both BLASTp_nr_NOGROUP and BLASTp_ref_NOGROUP. Residues like V16, A43, and M388 were conserved in less than 50% of the sequences in all 4 alignments. Therefore, further structural and docking studies are needed in order to obtain a more precise model of the Gox active site.

4. Conclusions

In this study, based on BLAST and KEGG searches, additional lineages were identified, which are worth investigation for glyphosate-degrading activity, as no glyphosate-degrading species have been reported from these lineages yet. Furthermore, additional conserved residues have been identified for each protein of the C-P lyase complex. Further studies would be useful to determine their role in the structure and function of the complex. Interestingly, Cys272 of PhnJ from E. coli, an important residue for the function of the active site of C-P lyase, is overall conserved but not present in all PhnJ sequences, raising the question of whether those PhnJ sequences have an alternative mechanism or if they have a different activity. Given the limited information on glyphosate degradation through the C-P lyase pathway [35], further studies of the C-P lyase pathway and the participating genes are needed, e.g., regarding the efficiency of the enzyme and the dependence of the phn gene expression on Pi deficiency.

Oxidoreductases are a large group of versatile enzymes [93] with conserved domains of low sequence similarity. There is a big research gap in the amino acid sequences of proteins known to degrade glyphosate, as most studies of glyphosate-degrading activity are based on detecting the degradation products with no further analysis at sequence level or identification of the responsible protein (Table 1). Bioinformatics analysis of already available biological data can contribute to the knowledge about Gox proteins. There are only a few annotated Gox proteins and we lack information about their reaction mechanism, the participating residues, and even the biological function in the bacteria which express them. The BLAST results show that Gox sequences are more similar to “FAD-dependent oxidases”. An experimentally determined Gox structure is not yet available and the binding site of glyphosate to the protein is not known yet. Here, we compare models based on three programs. The majority of the residues previously proposed to interact with glyphosate are not highly conserved in our alignments. Therefore, it would be interesting to create a more detailed structure in silico, which can be used for docking, in order to study the interacting residues and to search for enzymes with similar active sites. This could lead to the discovery of enzymes with higher efficiency, being better candidates to be applied in situ for the remediation of contaminated sites.