Discovery of a unique Ni2+-dependent heterohexameric metformin hydrolase

doi:10.21203/rs.3.rs-3656883/v1

Download PDF

Article

Discovery of a unique Ni²⁺-dependent heterohexameric metformin hydrolase

https://doi.org/10.21203/rs.3.rs-3656883/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 20 Jul, 2024

Read the published version in Nature Communications →

Version 1

posted

You are reading this latest preprint version

The biguanide drug metformin is prescribed worldwide as a first-line blood glucose-lowering medication for type 2 diabetes, leading to its presence in the environment around the world. Despite mounting evidence indicating that metabolic interactions of metformin with environmental or intestinal microbes affect ecological and human health, little is known about the fate of metformin by microbial catabolism. Here, we characterized a Ni²⁺-dependent bacterial enzyme that catalyzes the hydrolysis of metformin to form guanylurea and dimethylamine. The metformin hydrolase MetCaCb is a heterohexamer with an uneven α₂β₄ stoichiometry. Both subunits are from the arginase protein family, members of which are typically homomultimers. Either subunit alone is catalytically inactive, but together they work as an active enzyme highly specific for metformin. The crystal structure of the MetCaCb complex clearly shows the coordination of the binuclear metal cluster only in MetCa, following a geometry typical for arginase family enzymes. A unique pseudoenzyme MetCb, which evolved without a binuclear metal cluster, contributes to hydrolase activity as a protein binder of its active cognate. An in-silico search and functional assay led to the discovery of a group of MetCaCb-like protein pairs exhibiting metformin hydrolase activity in the environment. Our findings not only establish the genetic and biochemical foundation for metformin catabolism but also provide new insights into the adaption of the ancient arginase family proteins toward newly occurred substrate.

Biological sciences/Biochemistry/Enzymes/Hydrolases

Biological sciences/Microbiology/Microbial genetics/Bacterial genes

Arginase protein family

hydrolase

metformin

nickel

pseudoenzyme

Active pharmaceutical ingredients (APIs) are specifically designed to benefit humans or livestock. These compounds are continuously discharged into the environment, inadvertently becoming a source of contamination in diverse ecosystems^1,2, and therefore eliciting underestimated effects by interacting with animals, plants and microbes. Microbial catabolism of APIs has sparked significant interest, as the environmental microbial degradation of APIs offers a shield to mitigate the environmental impact of API contaminants. Additionally, gut microbes catabolizing APIs have been found to have a profound impact on drug efficacy, bioavailability, and potential side effects^3,4. However, the microbial catabolism mechanisms for many APIs are only beginning to be uncovered.

Metformin (1,1-dimethylbiguanide) is an anthropogenic biguanide derivative with a history spanning over 100 years⁵. It is an extremely hydrophilic compound with a pKa value of 11.5, and usually exists as an organic cation. Metformin is known for its therapeutic effects on type 2 diabetes (T2D) and has potential effects against cancer, cardiovascular diseases, aging, and inflammation⁶. A substantial amount of metformin is consumed each year due to its use at a dose of 0.5–2.5 g daily by T2D individuals, estimated to grow to 783 million people by 2045⁷. However, the orally administrated metformin is mainly excreted by the human body as an unmetabolized prototype compound, making it one of the most prevalent APIs in the environment globally^2,8,9.

Evidence are now available on the metformin-microbiota interactions^6,10,11 and biodegradability of metformin¹². However, whether gut microbiota can catabolize metformin and thereby affect metformin-microbiota interactions remains an unanswered question. Nonetheless, metformin biodegradation has been identified in many soil and aquatic bacterial consortia over the past decade^13,14. The analysis of its catabolic intermediates has unveiled several potential catabolic pathways, with guanylurea emerging as the most commonly detected metabolite accumulating in the environment¹². Only recently the genes responsible for guanylurea catabolism were identified in Pseudomonas spp^15,16, in which a hydrolase encoded by guuH initiated guanylurea catabolism with the production of guanidine and ammonia¹⁵. However, to our knowledge, enzymes mediating the cleavage of metformin have not been characterized.

Recently, we identified a metformin utilizer, Aminobacter sp. strain NyZ550, that catabolized metformin through hydrolysis to produce of guanylurea and dimethylamine¹⁷. Meanwhile several other metformin utilizers were also isolated^16,18, providing further support for our predicted metformin catabolic pathway. However, the molecular and biochemical mechanism of metformin catabolism in these strains remain elusive. In this study, we report that two paired arginase genes encoding a two-subunit hydrolase are responsible for metformin catabolism in Aminobacter sp. strain NyZ550. Together with crystal structure and mutagenesis studies, we demonstrated that MetCa and MetCb composed a unique heterohexameric enzyme through intermolecular interactions. We also revealed the divergent functional evolution of the proteins from the ancient arginase family, in which MetCa retained a metal-containing catalytic subunit but MetCb evolved to be a pseudoenzyme as metal-free binding partner.

A Ni²⁺-dependent metformin hydrolase

When examining the genomes of several metformin utilizers recently characterized^16–19, a gene cluster with seven genes (designated metR2TR1ABCaCb) was found to be present in all of them (Extended Data Fig. 1). This gene cluster consists of two transcriptional regulators belonging to the MerR family (MetR2) and AcrR family (MetR1), a nucleobase-cation-symport-1 (NCS1) like importer (MetT), two genes annotated as Ni/Fe hydrogenase nickel incorporation-associated proteins (MetA and MetB), as well as two arginase/agmatinase family proteins (MetCa and MetCb). The differential omics analyses revealed that the expression levels of the genes metCaCb encoding arginase family proteins are upregulated under metformin treatment condition^17,18. Arginase family enzymes are metalloenzymes, usually preferring Mn²⁺ for the catalysis of cleavage of C-N bond²⁰. In this regard, the met gene cluster encoding metallohydrolases, metallochaperones, a putative uptake system, and transcriptional regulators might encode metformin catabolism (Fig. 1a).

When the two arginase family proteins MetCaCb from strain NyZ550 were coexpressed in E. coli, low activity was detected in soluble protein extracts without the addition of transition metal salts (Fig. 1b). This activity was absent in extracts from E. coli cells carrying an empty vector. The addition of Ni²⁺ to the cell extracts strongly increased the activity and Co²⁺ slightly increased the activity, but the activity was inhibited by the addition of Zn²⁺ and Cu²⁺. Comparable activity was detected when coexpression of MetCaCb with or without the two metal chaperones MetAB, either in the presence or absence of metal ions during the assay (Fig. 1b). Additionally, when MetCa or MetCb was individually expressed in the absence or presence of MetAB, no metformin hydrolase activity was detected in the soluble cell extracts (Extended Data Fig. 2a). It was observed that MetCa had low solubility when expressed alone in E. coli, while coexpression with MetCb improved its solubility. To rule out the possibility that the low solubility of MetCa in the cell extracts led to a failure to detect activity, the recombinant MetCa and MetCb were purified individually for activity assays. As shown in Fig. 1c, neither protein exhibited activity in vitro, even in the presence of excess Ni²⁺. Likewise, no activity was observed when equal amounts of MetCa and MetCb were mixed and preincubated at 30°C for 1 h before the addition of 10 mM metformin. However, activity appeared when MetCa and MetCb were co-expressed and purified (Fig. 1c, d; and see below), suggesting that metCa and metCb together encode metformin hydrolase activity.

Since six additional genes encoding arginase family proteins were present in the genome of strain NyZ550, gene deletion was performed to verify whether the metformin hydrolase activity of MetCaCb was exclusive to strain NyZ550. With the deletion of coding sequence of either metCa or metCb following a two-step homologous recombination strategy, the mutants were no longer able to utilize metformin as the sole carbon and nitrogen source for growth, whereas their capacity to grow with dimethylamine was unaffected (Fig. 1e). Moreover, a metAB mutant also lost the ability to grow on metformin, indicating that MetAB was required for metformin utilization in vivo (Fig. 1e). It was then concluded that MetCa and MetCb together function as a Ni²⁺-dependent metformin hydrolase mediating the utilization of metformin.

MetCa and MetCb form a heterocomplex

To examine whether MetCa and MetCb bind each other to form a heteromultimeric metformin hydrolase, MetCaCb was purified from cell extracts containing maltose binding protein (MBP)-tagged MetCa and tag free MetCb. MetCb was co-purified with MBP-MetCa, as evidenced by two bands visible on the SDS-PAGE, and metformin hydrolase activity was only detected in co-purified MetCaCb (Fig. 2a). This result was also repeatable when the MBP tag was replaced with a His tag purified by Ni-nitrilotriacetic acid (NTA) agarose (Fig. 2b). MetCa and MetCb comprise 357 (theoretically 40 kDa) and 348 (theoretically 38 kDa) amino acids, respectively. Size exclusion chromatography (SEC) coupled with multi-angle light scattering (MALS) analysis showed that the molecular weight of the co-purified active MetCaCb complex was determined to be 223.7 ± 0.6 kDa in solution (Fig. 2c), approximately a hexamer. These results confirmed that the interaction of MetCa with MetCb to form a heteromultimeric complex was essential for its metformin hydrolase activity.

Characterization of the metformin hydrolase

¹³C-NMR and HPLC-MS analyses of the reaction mixture containing the ¹³C¹⁵N-labelled metformin as the substrate indicated that MetCaCb catalyzed the conversion of metformin into guanylurea (Fig. 1d; Extended Data Fig. 2b and c). The oxygen atom of guanylurea was derived from water, as evidenced by H₂¹⁸O label assays (Extended Data Fig. 2d). The other product from the hydrolysis of metformin by MetCaCb was determined to be dimethylamine, which is stoichiometrically formed with guanylurea (Extended Data Fig. 2e). Dimethylamine is a productive product used as a carbon and nitrogen resource by strain NyZ550 (Fig. 1e).

The MetCaCb’s dependence on Ni²⁺ was confirmed by the significant inhibition of its metformin hydrolase activity in the presence of the Ni-specific chelator dimethylglyoxime (DMG) (Fig. 1c). With the addition of Ni²⁺, the metformin hydrolase MetCaCb performed under optimal reaction conditions similar to other Mn²⁺-dependent arginase family enzymes²⁰. The optimum temperature for the reaction was ~ 50°C (Fig. 2d). MetCaCb hydrolyzed metformin most efficiently at pH 9.0 (Fig. 2e), and exhibiting a Michaelis constant (Km) of 6.84 mM for metformin with a catalytic efficiency of 1.88 mM^− 1 s^− 1 (Fig. 2f; Extended Data Table 1). The hydrolase activity was inhibited by the arginase inhibitor 2(S)-amino-6-boronohexonic acid (ABH)²¹ (Fig. 1c), indicating the possible formation of a tetrahedral intermediate during MetCaCb-catalyzed metformin hydrolysis.

MetCaCb also catalyzed the hydrolysis of biguanide and its derivatives, including 1-butylbiguanide, 1-methylbiguanide, and phenformin with the production of guanylurea, albeit with markedly reduced activities (Extended Data Fig. 2f). The low affinity and efficiency of MetCaCb toward 1-methylbiguanide compared to that for metformin (Fig. 2g; Extended Data Table 1) was consistent with the previously observed weak growth of strain NyZ550 on 1-methylbiguanide¹⁷. No hydrolase activity of MetCaCb was observed toward other guanidinium moiety-containing substrates, including guanidine, dimethylguanidine, agmatine, L-arginine, guanidinobutyrate, and guanidinopropionate (Extended Data Fig. 2f).

Architecture of the MetCaCb complex

To gain insight into the structure of metformin hydrolase and its recognition of metformin, a crystal structure analysis of the enzyme was initiated using X-ray crystallography. The co-purified MetCaCb was subjected to SEC before crystallization, and the active fractions of the MetCaCb protein complex (Fig. 2b) were used for crystallizing by the sitting-drop method. Finally, the atomic structure of the substrate-free MetCaCb complex was resolved at a resolution of 1.84 Å (PDB NO. 8X3G). The asymmetric unit (AU) of the P2₁2₁2₁ crystal showed an uneven stoichiometry of the two subunits, which contained two MetCa (α subunit) and four MetCb (β subunit) molecules (Fig. 3a and Extended Data Fig. 3a). This uneven composition of the MetCaCb complex was also roughly inferred from different intensities of SDS-PAGE bands (Fig. 2b). The hexamer was composed as a centrosymmetric dimer of trimers in which a MetCa and two MetCb molecules were arranged in a head-to-tail manner (Extended Data Fig. 3). To observe the polymeric state of the MetCaCb complex approximating physiological states, we performed single-particle cryoEM using the same sample for crystallization. This analysis demonstrated a remarkable degree of uniformity among the enzyme particles, and the resulting cryoEM map evidently unveiled the hexameric configuration of the MetCaCb complex (Extended Data Fig. 4).

MetCa and MetCb share 34% identity in their amino acid sequences, whereas their overall architectures were found to be basically identical (rms deviation of 1.94 Å for 321 Cα atoms), and exhibited the canonical three-layer alpha-beta-alpha fold of the arginase protein superfamily (Fig. 3b). The major structural differences between MetCa and MetCb lay in their N- and C-termini. MetCa had an additional N-terminal loop (T8–G26) contributing to the interaction with another MetCa subunit through a hydrophobic interface (Extended Data Fig. 5a and b). In contrast, an extended C-terminal loop of MetCb was located on the MetCa-MetCb interface, and it also stabilized the N-terminal structure of MetCb that capped the substrate-binding cavity of MetCa (Extended Data Fig. 5c).

Another difference between MetCa and MetCb was found in the arginase family active center region. Although the corresponding regions in both MetCa and MetCb were folded in a similar conformation, only the active site of MetCa contained a bimetallic center coordinated by four Asp (D183, D187, D276 and D278) and two His (H158 and H185) residues (Fig. 3c and Extended Data Fig. 5d). This metal ion coordination pattern is conserved among bacterial arginase family enzymes. A glycerol molecule from the crystal buffer was bound in the active site and coordinated to the two metal ions. Variants that disrupted the binuclear metal ion cluster in MetCa were introduced by substituting the conserved coordinating residues with Ala, which led to a loss of activity of the enzyme (Extended Data Table 1). In contrast, no electron density for metal ion was observed in the corresponding region of the subunit MetCb. This was likely due to the mutations in five of the six conserved metal ion coordinating residues in MetCb (Fig. 3c, d and Extended Data Fig. 5e). Analysis of the metal content of purified metformin hydrolase by inductively coupled plasma-mass spectrometry (ICP-MS) indicated there were 0.8 manganese and 0.05 nickel atoms per α₁β₂ trimer, while the metal content was 1.9 manganese and 0.06 nickel atoms per trimer when the enzyme was incubated with excess Mn²⁺ before ICP-MS analysis. In contrast, metal ions were not detected in the recombinant MetCb protein. The results are consistent with the aforementioned observations from the protein complex structure of MetCaCb. Taken together, these results inferred MetCa as the catalytic center of this metalloenzyme.

Active-site architecture of MetCaCb

An unexpected discovery from the crystal structure of the holoenzyme is that the active site cavity within the head domain of MetCa is capped by the tail domain of a MetCb subunit (Extended Data Fig. 6a). This configuration is distinct from typical arginase family proteins, which feature an open, solvent-exposed substrate-binding pocket²². The N-terminal 23 amino acids of MetCb adopted a hollowed-out structure, predicted to serve as the substrate entrance connecting to the active site cavity of MetCa (Fig. 3e). Notably, the truncation of these 23 amino acids in the N-terminal of MetCb resulted in a failure to achieve soluble expression of the MetCaCb complex, suggesting its contribution to stabilizing the protein complex.

Next, a metformin molecule was computationally docked into the metal ion-containing active center of MetCa, which revealed that metformin was bound in a highly negative pocket buried by the aromatic amino acids Y22 and F82 from MetCb and W192 from MetCa (Fig. 3f). The noncoordinating residues in the active site of MetCa were mostly hydrophilic, including Q80, N199, N288, S289, and E320 (Fig. 3e; Extended Data Fig. 6b). The N199 and S289 residues of MetCa appeared to be unique among arginase family enzymes (Extended Data Fig. 6c). The equivalent residue of N199 in the arginase family proteins is a conserved noncoordinating histidine which is catalytically important during the reaction process²³. The complete loss of activity of the N199A variant was observed (Extended Data Table 1), implying that N199 serves as a proton shuttle like its equivalent histidine. Additionally, S289 was identified as a substrate-binding residue and as being essential for metformin hydrolase activity (Extended Data Table 1).

MetCaCb and paired arginase family proteins

Arginase family proteins are commonly known as single gene-encoding enzymes that function as monomer or homomultimer²⁰. However, the heterohexameric metformin hydrolase encoded by the arginase gene pair represents a unique member in this family (Extended Data Fig. 7). To investigate the possible distribution of the MetCaCb-like arginase pairs, their homologs with at least 25% amino acid sequence identity were searched in the NCBI nr database²⁴. As a result, 56 gene pairs were identified from four phyla of bacteria including Proteobacteria (44), Actinomycetota (10), Acidobacteriota (1) and Planctomycetota (1) (Supplementary Data 1). The occurrence of these paired arginases in soil, aquatic, and plant-associated bacteria underlies their environmental ubiquity. A phylogenetic tree analysis showed that MetCa and MetCb were located in two well-separated clades that evolved from a common origin (Fig. 4a). Homologs within the MetCa clade retained all six conserved metal ion coordinating residues, whereas the MetCb clade contained a group of unique arginase family proteins exhibiting three to five mutations of the six conserved coordinating residues (Fig. 4a). The genetic context analysis indicated that most encoding genes within the MetCaCb clades were found in close physical proximity to hydrogenase maturation nickel metallochaperones genes that shared homology with MetAB in this study, indicating their likely dependence on Ni²⁺ ion as well. Furthermore, there was a set of arginase pairs that contained the intact metal ion coordinating residues in both members of the pairs. Nevertheless, their specific functions, metal ion-dependence, and whether they collaborate in their enzymatic activities remain open questions.

To test whether these MetCaCb homologous pairs were functional, three codon-optimized DNA sequence pairs (AcMetCaCb, HmMetCaCb, and RsMetCaCb, as shown in Fig. 4a) from the MetCaCb clades were synthesized for the expression of recombinant proteins in E. coli. The α subunits within these three pairs exhibited AA identities of 93.8%, 71.2%, and 61.7% with MetCa. All the protein pairs were successfully expressed and purified to be heterohexamers as the positive control MetCaCb (Fig. 4b). In the presence of Ni²⁺, conversion of metformin into guanylurea by AcMetCaCb and HmMetCaCb was confirmed by LC-MS. However, the hydrolase activities were approximately 20,000-fold lower in comparison to that of MetCaCb. In contrast, no detectable enzymatic activity against metformin was observed for RsMetCaCb (Fig. 4c).

The frequent occurrence of metformin and the accumulation of its metabolite guanylurea in the environment has raised interest in their transformation by microorganisms. The recent discovery that metformin is utilized by several bacteria isolated from different continents further supports the notion that microorganisms have developed catabolic enzymes specifically tailored to the synthetic metformin molecule. Here, we characterized a unique bacterial enzyme capable of hydrolyzing metformin into guanylurea and dimethylamine. Biochemical and structural studies demonstrated that the metformin hydrolase was a nickel metalloenzyme derived from the ancient arginase family. Enzymes from this family are metallohydrolases that typically use two Mn²⁺ ions to activate a water molecule for nucleophilic attack of the guanidinium group, although there are several exceptions, such as the Fe²⁺-dependent agmatinase (SpeB) from Methanocaldococcus jannaschii²⁵, Zn²⁺-dependent guanidinobutyrases from Aspergillus niger²⁶ and Arthrobacter sp. KUJ 8602²⁷, and Co²⁺-preferring arginase from Helicobacter pylori²⁸. Notably, the guanidine hydrolase (GdmH) from Synechocystis sp. PCC 6803 is a Ni²⁺-dependent enzyme²⁹. Like other nickel-dependent enzymes, such as [NiFe]-hydrogenase and urease, activation of GdmH requires two nickel incorporation chaperones (GhaAB). In this study, the genes (metAB) encoding metal chaperones were found adjacent to metCaCb. It seems that they were not essential for the in vitro activity of MetCaCb (Fig. 1b). However, the inability of the metAB knockout strain to grow on metformin suggested that MetAB was necessary for the function of metformin hydrolase in an intracellular environment in which the concentration of nickel ion is far lower than that used in the in vitro assays.

The crystal structure and metal content analyses were consistent with the conclusion that the MetCa subunit contained a binuclear metal center. We were able to clearly assign two metal ions to the active site of MetCa from the structure of the MetCaCb complex. Despite Mn²⁺ can no longer activate the enzyme, ICP-MS analysis of recombinant MetCaCb revealed that Mn²⁺ remained the preferentially bound metal ion, suggesting that the Mn²⁺-binding capacity of MetCa was inherited from the Mn²⁺-dependent arginase family enzymes. Nonetheless, the mechanism by which nickel is inserted into MetCa and subsequent activation of the metformin hydrolase needs further investigation. In contrast, the MetCb subunit of metformin hydrolase lost its ability to bind metal ions due to variations in the conserved coordinating residues (Fig. 3d). Notably, noncanonical arginase proteins without a binuclear metal center have been observed in both eukaryotes^30,31 and prokaryotes³², however, their biochemical and physiological functions have been elusive. These proteins are known as pseudoenzymes that are incapable of catalyzing metal ion-dependent hydrolysis. However, their metal binding ability and hydrolase activity can be restored by reconstructing the binuclear metal center, indicating that they are evolutionarily close to the active arginase family enzymes. In this study, we uncovered a precedent in which the inactive arginase-family protein MetCb mediated the hydrolase activity by interacting with its catalytically active partner MetCa, which physiologically was essential for the utilization of metformin by the strain. Thus, a new member has been now added to the emerging group of pseudoenzyme-enzyme pairs with a characterized function.

Arginase family enzymes are commonly characterized as homomultimers²⁰; however, their catalytic activity does not seem to depend on the multimeric state. For instance, disruption of the hexameric structure of arginase from Bacillus caldovelox by introducing an E278A variant exhibited similar kinetic parameters as those of the wild-type enzyme³³. Additionally, eukaryotic arginases from humans and rats are naturally trimeric, whereas their monomers exhibited comparable activities as trimeric holoenzymes^34,35. In this study, however, metformin hydrolase was identified as the first two-subunit arginase family enzyme composing a heterohexamer. The unique interaction of MetCa with MetCb to form a protein complex was essential to the metformin hydrolase activity. This interaction enhanced the stability of MetCa in solution, and contributed to both the substrate tunnel and fine-tuning of the active site.

The metCa and metCb genes were found as an adjacent gene pair in the same clade, as shown by the sequence similarity network analysis of the arginase family proteins (Extended Data Fig. 7). It is likely that metCaCb evolved via a gene duplication event, and the loss of catalytic activities of one of the proteins might have led it into a new evolutionary trajectory to become specialized as a protein binder. For the bacterial cell, the loss of the nickel-binding ability of the non-catalyzing MetCb subunit relieves the intracellular demand for the highly toxic nickel ion and also the energy consumption for nickel insertion. Notably, inactive cognates of enzymes are commonly found, but their functions are not well understood³⁶, as is the cases with the carboxyguanidine deiminase (CdgAB) in the DUF1989 protein family³⁷ and the 2,4-dinitroanisole demethylase (DnhAB) in the metallo-beta-lactamase (MBL) family³⁸. Nonetheless, these cases imply that the presence of heterocomplexes within the protein families extend the catalytic diversity of enzymes by using a limited number of available gene products.

The available evidence clearly indicated that metformin functions as its prototype structure within the human body, although the mechanisms underlying the effect of metformin on reducing blood glucose levels in humans are intricate and still not fully understood⁶. Metformin has been extensively used to treat diabetes and subsequently has been found in the environment for only a few decades. However, our results revealed the emergence of a set of metformin hydrolases in environmental bacteria that are rapidly evolving to become efficient. This raises concerns about whether metformin hydrolase may potentially be present in gut microbes, resultantly leading to the breakdown of the active prototype structure of metformin. A similar situation has been observed regarding acarbose, another antidiabetic drug, which is rendered inactive by enzymes present in the human gut and oral microbes^39,40. Considering the significant secondary failure rate of metformin monotherapy in clinical practice⁴¹, the potential presence of metformin hydrolase in the human gut microbiome should be carefully investigated due to its implications for this important medication.

Growth conditions of Aminobacter sp. strain NyZ550

Aminobacter sp. strain NyZ550, a metformin utilizer, was isolated as previously descripted¹⁷. The cells were cultivated in 1/5 strength tryptic soy broth (TSB) in Erlenmeyer flasks at 30°C in a shaking incubator at 180 rpm. When testing the substrate-utilizing ability of the wildtype strain NyZ550 and its derivatives, the cells were cultivated in nitrogen-free minimal salts medium (MSM, pH 7.0)⁴² supplemented with 5 mM metformin or dimethylamine. Ammonium sulfate (0.2%, w/v) was used as a nitrogen source if necessary. Biomass accumulation was estimated by measurement of the optical density at 600 nm (OD_600nm).

Cloning, heterologous expression, and purification of recombinant proteins

The coding sequences of metA, metB, metCa, and metCb were amplified from genomic DNA of strain NyZ550 and subcloned into vector pUC19 containing a lac promoter. For recombinant protein expression and purification, pMAL-c2X containing a maltose-binding protein (MBP) tag and pET-28a(+) containing a 6×His tag were used. Different combinations of the genes were generated by overlap PCR amplification and inserted into desired vectors by Gibson assembly to generate expression constructs. The DNA sequences of MetCaCb homologs were synthesized by a commercial company and inserted into a pET-28a(+) vector. Recombinant proteins were produced in E. coli Rosetta(DE3) harboring the expression constructs. The E. coli Rosetta(DE3) cells were grown in lysogeny broth (LB) with appropriate antibiotics at 37°C to an OD_600nm of 0.6. Then IPTG was added to a final concentration of 0.1 mM, and cells were induced at 16°C with shaking for 12 h. After harvesting the cells by centrifugation, they were suspended in an elution buffer containing 50 mM Tris-HCl (pH 8.0), 100 mM NaCl and 5% glycerol and lysed by ultrasonication. Cell lysates were centrifuged at 16,000 g for 40 min at 4°C. The soluble protein fractions were used for enzymatic assays or purification by loading onto affinity columns. For the His-tagged proteins, the lysates were loaded onto a nickel-nitrilotriacetic acid column (GE Healthcare) with elution buffer containing 25 mM imidazole, and eluted in elution buffer supplemented with 250 mM imidazole. For the MBP-tagged proteins, the lysates were loaded onto a 1-ml dextrin column (GE Healthcare). After washing with elution buffer, proteins were eluted with 10 mM maltose in elution buffer. The proteins were then subjected to size-exclusion chromatography with a Superdex 200 Increase 10/300 GL column (GE Healthcare) using elution buffer for column equilibration and elution. The resulting fractions were concentrated by ultrafiltration using a 10-kDa Amicon Ultra tube (Millipore) and stored at − 80°C for further use.

Molecular weight determination

The recombinant MetCaCb protein solution underwent dialysis against a glycerol-free buffer containing 20 mM Tris-HCl (pH 8.0) and 200 mM NaCl. The molecular weight was determined in a protein concentration of 6.0 mg/mL by size exclusion chromatography (SEC) coupled with multi-angle light scattering (MALS). Bovine serum albumin (Thermo Fisher Scientific, Waltham, MA) with a molecular weight of 66 kDa was used to normalize the static light scattering detector.

Metformin hydrolase activity assays

Standard metformin hydrolase activity assays were carried out in 500 µLof a reaction mixture containing 50 mM Tris-HCl buffer (pH 9.0), 0.2 mM Ni₂SO₄, and 0.1–5 µg of the purified enzyme or 300 mg of crude bacterial extracts. The reaction was initiated by adding 10 mM metformin or ¹³C¹⁵N-labelled metformin and incubated at 50°C for 60 min. The reaction was terminated by the addition of equal volumes of acetonitrile, and centrifugated at 15,000 g for 20 min before further high-performance liquid chromatography-mass spectrum (HPLC-MS) analysis. Enzyme assays were performed at an appropriate enzyme concentration for the linear production of guanylurea during the incubation period. For the reaction conditions and kinetic analyses, the metformin concentration, temperature, pH, buffer, and transition metal ions were modified as indicated in the respective figure legends. Guanylurea and dimethylamine release was measured by HPLC-MS and ion chromatography, respectively, using methods described previously¹⁷.

NMR spectroscopy

The reaction mixtures were prepared as standard metformin hydrolase assays with 10 mM ¹³C¹⁵N-labelled metformin (Sigma-Aldrich). Then, 0.4 mL of each reaction mixture was transferred to NMR tubes to record the ¹³C-NMR spectra on a Quantum-I Plus 400 MHz spectrometer (Q.One Instruments Ltd., China).

Genetic manipulation

Construction of markerless mutants in strain NyZ550 was performed using a two-step homologous recombination procedure. Briefly, a downstream fragment and an upstream fragment of the target gene were amplified from the genomic DNA of strain NyZ550, and the resulting fragments were assembled together by overlap PCR. The recombinant fragment was then ligated to EcoRI- and BamHI-digested pEX18Gm, derived from a suicide vector pEX18Tc by replacing the tetracycline resistance gene cassette with a gentamicin resistance gene cassette. The resulting pEX18Gm derivatives were transformed into diaminopimelic acid auxotroph E. coli WM3064, which was then conjugated to strain NyZ550. The transconjugants bearing suicide plasmids were subjected to double-crossovers on an LB agar plate containing 10 µg/mL gentamicin (for the first recombination) or 15% sucrose (for the second recombination) at 30°C. The in-frame deletions of the target genes were confirmed via PCR and sequencing.

Metal analysis of metformin hydrolase by ICP-MS

Proteins were buffer-exchanged into 20 mM Tris-HCl by performing gel filtration chromatography with a Superdex 200 Increase 10/300 GL column. The proteins were concentrated to 5–10 mg/mL and analyzed by ICP-MS. One milliliter of the flow-through buffer was collected and used as a blank.

Crystallization and structure determination

The coding sequences of MetCaCb were amplified from genomic DNA and cloned into a pET28a(+) vector by Gibson assembly for coexpression of MetCaCb, in which MetCa contained an N-terminal 6×histidine tag. MetCaCb was stored in a buffer containing 20 mM Tris-HCl (pH 8.0), 100 mM NaCl, and 5% (v/v) glycerol. The sitting-drop vapor-diffusion method was used for crystallization of MetCaCb with equal volumes (0.8 µL each) of protein (16 mg/mL) and reservoir solutions at 20°C. The crystal grew from only one liquor condition: 0.15 M calcium chloride dihydrate, 0.1 M MES buffer system (pH 6.2), 5% (v/v) glycerol and 15% (v/v) PEG smear broad.

For data collection, the crystal was soaked in the reservoir solution supplemented with 20% (v/v) glycerol as a cryoprotectant, and subsequently vitrified by flash freezing in liquid nitrogen. X-ray diffraction data were collected at 100 K using a wavelength of 1.0 Å on the BL17U beamline at the Shanghai Synchrotron Radiation Facility. Diffraction data were processed using the XDS program package⁴³, and phasing was solved by molecular replacement in PHASER⁴⁴. The initial structures of MetCa and MetCa were produced by Alphafold2⁴⁵ with pLDDT values > 90, and then employed as search models. Automated model building and refinement were performed in PHENIX⁴⁶. The program COOT⁴⁷ was used for visualization of the electron density map, as well as for manual adjustments and refinements of the model. The model quality was assessed by MolProbity⁴⁸. The metformin molecule was docked into MetCaCb using AutoDock Vina v1.1.2⁴⁹ and the top model with an AutoDock score of − 4.7 is shown.

Sequence comparison and phylogenetic analysis

To explore the co-occurrence of MetCa and MetCb homologs in bacteria, amino acid sequences of MetCa and MetCb were used as queries to search against the NCBI nr database in June 2023 using Cblaster (v1.3.18) with a threshold of > 25% identity and > 70% coverage. All sequences of the retrieved homologous protein pairs were extracted through the extract_clusters module and manually curated with a gap threshold of less than 2000 bp between genes. The resulting sequence pairs were used to construct a phylogenetic tree by the maximum likelihood method with human arginase I (hARGI, NCBI accession number: NP_000036.2) as the outgroup.

Multiple sequence alignment of MetCaCb with characterized sequences was performed by MUSCLE⁵⁰. The sequences used included guanidine hydrolase (GdmH), agmatinase (SpeB), guanidinobutyrase (GbuA), guanidinopropionase (GpuA), arginase from Thermus thermophilus (ttARG), human arginases I and II (hARGI and hARGII), and arginase from Schistosoma mansoni (smARG).

CryoEM data acquisition and image processing

MetCaCb was purified following the aforementioned procedure, and then 3.5 µL of the sample was applied to Quantifoil 1.2/1.3 copper grids and mounted onto a Vitrobot Mark IV. Subsequently, the sample was blotted with a filter paper for a duration of 2 s using a blot force of − 1. The grids were promptly immersed into liquid ethane to ensure their rapid cryopreservation.

Data acquisition was carried out at the Instrumental Analysis Center of the Shanghai Jiao Tong University with a Thermo Scientific Glacios transmission electron microscope operated at 200 kV and equipped with a Falcon 3 camera. Automated data collection was carried out using EPU software in linear mode. In total, 1068 movies were taken at a defocus range between − 1.5 and − 2.5 µm, calibrated pixel size of 1.6 Å/pixel, and a total dose of 40 electrons/Å² was applied to the movies over an exposure time of 2.54 s.

Image processing was performed by cryoSPARC (v3.1). For pre-processing, beam-induced motion and the contrast transfer function (CTF) was corrected on the movies using cryoSPARC’s own implementations Patch Motion Correction and Patch CTF Estimation, respectively. Particles were selected from pre-processed micrographs using the blob picker. A total of 2,052,983 particles were picked and extracted from the micrographs. Subsequently, a subset of particles was selected after several rounds of 2D classification to build the initial model using Ab initio reconstruction. Then, several rounds of heterogeneous refinement (n = 2) were used to improve the initial map quality. The final map of 3D classification was further processed using non-uniform (NU) refinement and resulted in a map of approximately 5.6 Å resolution based on a Fourier Shell Correlation (FSC) cutoff at 0.143.

Data availability

Atomic coordinate and structure factor for the crystal structure of MetCaCb from Aminobacter sp. strain NyZ550 have been deposited to the Protein Data Bank (PDB) under accession number 8X3G (for immediate release on publication). The cryoEM movies and the refined model are available from the corresponding author upon request.

Acknowledgements This work was supported by the National Natural Science Foundation of China (NSFC) (32230001). We thank Kaixuan Gao for the assistance with the CryoEM data processing.

Author information

Authors and affiliations:

State Key Laboratory of Microbial Metabolism, Joint International Research Laboratory of Metabolic and Developmental Sciences, and School of Life Sciences and Biotechnology, Shanghai Jiao Tong University, Shanghai, China

Tao Li, Zhi-jing Xu, Shu-Ting Zhang, Jia Xu, Piao-Piao Pan & Ning-Yi Zhou

Author contributions:

T.L. and N.Y.Z developed the research plan and experimental design. Z.J.X. and T.L. performed the molecular and biochemical experiments. J.X. constructed the gene-knockout variants. T.L. and Z.J.X. isolated and crystallized proteins. T.L. and S.T.Z. collected and analyzed crystallographic data. P.P.P. performed the bioinformatic analysis. T.L., Z.J.X. and N.Y.Z wrote the manuscript, which was read and approved by all authors.

Corresponding author: Correspondence to Ning-Yi Zhou.

Competing interests: The authors declare no competing interests.

Additional Information:

Correspondence and requests for materials should be addressed to Ning-Yi Zhou.

Dos SGR et al (2022) Medicines as an emergent contaminant: the review of microbial biodegration potential. Folia Microbiol (Praha) 67:157–174
Wilkinson JL et al (2022) Pharmaceutical pollution of the world's rivers. Proc. Natl. Acad. Sci. U S A 119, e2113947119
Lindell AE, Zimmermann-Kogadeeva M, Patil KR (2022) Multimodal interactions of drugs, natural compounds and pollutants with the gut microbiota. Nat Rev Microbiol 20:431–443
Zimmermann M, Zimmermann-Kogadeeva M, Wegmann R, Goodman AL (2019) Mapping human microbiome drug metabolism by gut bacteria and their genes. Nature 570:462–467
Werner EA, Bell J (1922) CCXIV.—The preparation of methylguanidine, and of ββ-dimethylguanidine by the interaction of dicyanodiamide, and methylammonium and dimethylammonium chlorides respectively. J Chem Soc Trans 121:1790–1794
Foretz M, Guigas B, Viollet B (2023) Metformin: update on mechanisms of action and repurposing potential. Nat Rev Endocrinol 19:460–476
IDF. In: IDF Diabetes Atlas, ten ed. IDF Diabetes Atlas, 10th ed. Brussels, Belgium (2021)
Balakrishnan A, Sillanpää M, Jacob MM, Vo DN (2022) Metformin as an emerging concern in wastewater: Occurrence, analysis and treatment methods. Environ Res 213:113613
Elizalde-Velazquez GA, Gomez-Olivan LM (2020) Occurrence, toxic effects and removal of metformin in the aquatic environments in the world: Recent trends and perspectives. Sci Total Environ 702:134924
Cabreiro F et al (2013) Metformin retards aging in C. elegans by altering microbial folate and methionine metabolism. Cell 153:228–239
Forslund K et al (2015) Disentangling type 2 diabetes and metformin treatment signatures in the human gut microbiota. Nature 528:262–266. 10.1038/nature15766
He Y, Zhang Y, Ju F (2022) Metformin contamination in global waters: biotic and abiotic transformation, byproduct generation and toxicity, and evaluation as a pharmaceutical indicator. Environ Sci Technol 56:13528–13545
Briones RM, Zhuang WQ, Sarmah AK (2018) Biodegradation of metformin and guanylurea by aerobic cultures enriched from sludge. Environ Pollut 243:255–262
Poursat BAJ et al (2019) Biodegradation of metformin and its transformation product, guanylurea, by natural and exposed microbial communities. Ecotox Environ Saf 182:109414
Tassoulas LJ, Robinson A, Martinez-Vaz B, Aukema KG, Wackett LP (2021) Filling in the gaps in metformin biodegradation: a new enzyme and a metabolic pathway for guanylurea. Appl Environ Microbiol 87:e03003–e03020
Martinez-Vaz BM et al (2022) Wastewater bacteria remediating the pharmaceutical metformin: Genomes, plasmids and products. Front Bioeng Biotechnol 10:1086261
Li T, Xu Z-J, Zhou N-Y (2023) Aerobic Degradation of the antidiabetic drug metformin by Aminobacter sp. strain NyZ550. Environ Sci Technol 57:1510–1519
Chaignaud P et al (2022) A Methylotrophic bacterium growing with the antidiabetic drug metformin as its sole carbon, nitrogen and energy source. Microorganisms 10:2302
Hillmann KB, Niehaus TD (2022) Genome sequences of two Pseudomonas isolates that can use metformin as the sole nitrogen source. Microbiol Resour Announc 11:e0063922
Hernández VM, Arteaga A, Dunn MF (2021) Diversity, properties and functions of bacterial arginases. FEMS Microbiol Rev 45:fuab034
Baggio R et al (1997) Inhibition of Mn^2 + ₂-arginase by borate leads to the design of a transition state analogue inhibitor, 2(S)-amino-6-boronohexanoic acid. J Am Chem Soc 119:8107–8108
Di Costanzo L et al (2005) Crystal structure of human arginase I at 1.29-Å resolution and exploration of inhibition in the immune response. Proc. Natl. Acad. Sci. U S A 102, 13058–13063
Velázquez-Libera JL, Caballero J, Tuñón I, Hernández-Rodríguez EW, Ruiz-Pernía JJ (2020) On the nature of the enzyme–substrate complex and the reaction mechanism in human arginase I. A combined molecular dynamics and QM/MM study. ACS Catal 10:8321–8333
O'Leary N et al (2015) Reference sequence (RefSeq) database at NCBI: current status, taxonomic expansion, and functional annotation. Nucleic Acids Res 44:D733–D745
Miller D, Xu H, White RH (2012) A new subfamily of agmatinases present in Methanogenic Archaea Is Fe(II) dependent. Biochemistry 51:3067–3078
Saragadam T, Kumar S, Punekar N (2019) Characterization of 4-guanidinobutyrase from Aspergillus niger. Microbiology 165:396–410
Arakawa N, Igarashi M, Kazuoka T, Oikawa T, Soda K (2003) D-arginase of Arthrobacter sp. KUJ 8602: characterization and its identity with Zn²⁺-guanidinobutyrase. J biochem 133:33–42
McGee D et al (2004) Purification and characterization of Helicobacter pylori arginase, RocF: unique features among the arginase superfamily. Eur J biochem 271:1952–1962
Funck D et al (2022) Discovery of a Ni²⁺-dependent guanidine hydrolase in bacteria. Nature 603:515–521
Romero N et al (2017) Mammalian agmatinases constitute unusual members in the family of Mn²⁺-dependent ureahydrolases. J Inorg Biochem 166:122–125
Hai Y, Kerkhoven EJ, Barrett MP, Christianson DW (2015) Crystal structure of an arginase-like protein from Trypanosoma brucei that evolved without a binuclear manganese cluster. Biochemistry 54:458–471
Mothersole RG et al (2022) Sequence divergence in the arginase domain of ornithine decarboxylase/arginase in Fusobacteriacea leads to loss of function in oral associated species. Biochemistry 61:1378–1391
Garcia D, Uribe E, Salgado M, Martínez M (2014) Mutagenic and kinetic support for an allosteric site in arginase from the extreme thermophile Bacillus caldovelox, which allows activation by arginine. Biochimie 108:8–12
Lobos M et al (2020) Insights on the participation of Glu256 and Asp204 in the oligomeric structure and cooperative effects of human arginase type I. J Struct Biol 211:107533
Lavulo L et al (2001) Subunit-subunit interactions in trimeric arginase. Generation of active monomers by mutation of a single amino acid. J biol Chem 276:14242–14248
Adrain C, Freeman M (2012) New lives for old: evolution of pseudoenzyme function illustrated by iRhoms. Nat rev Mol cell biol 13:489–498
Schneider NO et al (2020) Solving the conundrum: widespread proteins annotated for urea metabolism in bacteria are carboxyguanidine deiminases mediating nitrogen assimilation from guanidine. Biochemistry 59:3258–3270
Fida TT, Palamuru S, Pandey G, Spain JC (2014) Aerobic biodegradation of 2,4-dinitroanisole by Nocardioides sp. strain JS1661. Appl Environ Microbiol 80:7725–7731
Balaich J et al (2021) The human microbiome encodes resistance to the antidiabetic drug acarbose. Nature 600:110–115
Tian J et al (2023) Inactivation of the antidiabetic drug acarbose by human intestinal microbial-mediated degradation. Nat Metab 5:1–14
Brown JB, Conner C, Nichols GA (2010) Secondary failure of metformin monotherapy in clinical practice. Diabetes Care 33:501–506
Liu H, Wang SJ, Zhou NY (2005) A new isolate of Pseudomonas stutzeri that degrades 2-chloronitrobenzene. Biotechnol Lett 27:275–278
Kabsch WXDS (2010) Acta Crystallogr D Biol Crystallogr 66:125–132
McCoy A et al (2007) PHASER crystallographic software. J appl Crystallogr 40:658–674
Jumper J et al (2021) Highly accurate protein structure prediction with AlphaFold. Nature 596:583–589
Adams P et al (2011) The Phenix software for automated determination of macromolecular structures. Methods (San Diego Calif) 55:94–106
Emsley P, Lohkamp B, Scott WG, Cowtan K (2010) Features and development of COOT. Acta Crystallogr D 66:486–501
Williams CJ et al (2018) More and better reference data for improved all-atom structure validation. Protein Sci 27:293–315MolProbity
Trott O, Olson AJ (2010) AutoDock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. J Comput Chem 31:455–461
Madeira F et al (2022) Search and sequence analysis tools services from EMBL-EBI in 2022. Nucleic acids res 50:W276–W279

There is NO Competing Interest.

Download PDF

Journal Publication

published 20 Jul, 2024

Read the published version in Nature Communications →

Version 1

posted

You are reading this latest preprint version

Discovery of a unique Ni²⁺-dependent heterohexameric metformin hydrolase

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Results

A Ni²⁺-dependent metformin hydrolase

MetCa and MetCb form a heterocomplex

Characterization of the metformin hydrolase

Architecture of the MetCaCb complex

Active-site architecture of MetCaCb

MetCaCb and paired arginase family proteins

Discussion

Materials and methods

Cloning, heterologous expression, and purification of recombinant proteins

Molecular weight determination

Metformin hydrolase activity assays

NMR spectroscopy

Genetic manipulation

Metal analysis of metformin hydrolase by ICP-MS

Crystallization and structure determination

Sequence comparison and phylogenetic analysis

CryoEM data acquisition and image processing

Declarations

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1

Discovery of a unique Ni2+-dependent heterohexameric metformin hydrolase

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Results

A Ni2+-dependent metformin hydrolase

MetCa and MetCb form a heterocomplex

Characterization of the metformin hydrolase

Architecture of the MetCaCb complex

Active-site architecture of MetCaCb

MetCaCb and paired arginase family proteins

Discussion

Materials and methods

Cloning, heterologous expression, and purification of recombinant proteins

Molecular weight determination

Metformin hydrolase activity assays

NMR spectroscopy

Genetic manipulation

Metal analysis of metformin hydrolase by ICP-MS

Crystallization and structure determination

Sequence comparison and phylogenetic analysis

CryoEM data acquisition and image processing

Declarations

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1

Discovery of a unique Ni²⁺-dependent heterohexameric metformin hydrolase

A Ni²⁺-dependent metformin hydrolase