NagS is required for GlcNAc toxicity and amino sugar sensing

Mutants of S. coelicolor that lack nagB fail to grow on minimal media (MM) supplemented with GlcNAc. In order to obtain mutants that are affected in GlcNAc sensing and nutrient control of development and antibiotic production, mutants were analyzed that we previously obtained in a suppressor screen, selecting for spontaneous mutants of S. coelicolor nagB mutants that survive specifically on GlcNAc [30]. Sequencing of suppressor mutant SMA11 revealed a single nucleotide polymorphism (SNP) at nucleotide position 535 of SCO4393 (G to A substitution), leading to a nonsilent change from aspartate to asparagine (D179N) in the predicted gene product. SCO4393 contains a so-called SIS domain [32], which spans the entire protein (see below). As SCO4393 was shown to be involved in N-acetylglucosamine sensitivity, this gene was renamed nagS.

To study the role of NagS in GlcNAc metabolism and nutrient sensing, we created a nagB-nagS double mutant. While the parental strain S. coelicolor M145 grew normally on MM with 1% (w/v) mannitol and 10 mM GlcNAc, the nagB mutant was unable to grow under these conditions. Conversely, both the suppressor mutant SMA11 and the ∆nagB∆nagS mutant grew well in the presence of GlcNAc. To ascertain that the deletion of nagS was the sole cause of the observed phenotypes described, we genetically complemented the ∆nagB∆nagS mutant with a native copy of nagS. Expression of nagS restored sensitivity of the mutants to GlcNAc (Fig 1a). These data strongly suggest that the deletion of nagS was the sole cause of the restored growth on GlcNAc in nagB mutants.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. NagS is a novel GlcNAc-6P dehydratase.

(a) Sensitivity of S. coelicolor mutants to GlcNAc. Spores (5 × 10⁵ CFU) of S. coelicolor M145 and its mutant derivatives ∆nagB, SMA11, ∆nagB∆nagS, ∆nagB∆nagS^C (∆nagB∆nagS expressing nagS) and ∆nagB∆nagS^E (∆nagB∆nagS with empty plasmid pSET152) were streaked on MM agar plates with 1% mannitol (Mann) and 1% mannitol plus 10 mM GlcNAc (GlcNAc). Strains were cultured for 72 h at 30 °C. Note that nagB mutants can grow in the presence of 10 mM GlcNAc only when SCO4393 (nagS) has been mutated (suppressor SMA11) or deleted. (b) NagS and its role in GlcNAc sensing. Spores of M145 and ∆nagS were plated on MM and R5 with 0, 0.001, 0.01, 0.1, 1, 5, 10, 20, 50, 100, 150, and 200 mM GlcNAc. Note that nagS mutants hardly respond to GlcNAc. (c) Phylogenetic tree of several different types of sugar isomerase (SIS) domain enzymes in bacteria, including NagS, Glucose-6P isomerase (Pgi), Phosphoheptose isomerase (GmhA), N-acetylmuramic acid-6-phosphate etherase (MurQ), Glutamine-fructose-6-phosphate aminotransferase (GlmS), and putative D-galactosamine-6-phosphate deaminase (Agas). The phylogenetic tree was made by MEGA11 (Neighbour-joining method) and built based on the alignment of the amino acid sequences. (d) ITC analysis of NagS with N-acetyl-6-phosphate amino sugar metabolites, namely GlcNAc-6P, ManNAc-6P, and GalNAc-6P. Both GlcNAc-6P and ManNAc-6P are bound well by NagS, while the enzyme did not bind to GalNAc-6P. (e and f) confirmation of NagS products by LC-MS. Extracted ion chromatograms for GlcNAc-6-P (black trace) and its dehydrated product (pink trace) in the enzymatic reaction mixture of GlcNAc-6-P with either active NagS (e) or heat-inactivated NagS (f). Peaks relating to the product (m/z = 282.0387) are indicated by red arrows. (g) Kinetics of NagS in the presence of 6-PG (0, 0.33, 1 mM). The V₀ data used in d were plotted against the substrate concentration, and each assay was performed in triplicate and expressed as a mean ± standard error. The data underlying this Figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3003514.g001

We then wondered if NagS might play a role in GlcNAc-dependent control of development and specialized metabolism. To investigate this, S. coelicolor was cultured on MM and nutrient-rich R5 agar plates with increasing concentrations of GlcNAc. Expectedly, GlcNAc enhanced development and antibiotic production of wild-type S. coelicolor on MM and inhibited these processes on R5. In contrast, the nagS mutant showed normal sporulation on R5 agar even at GlcNAc concentrations up to 200 mM, and the mutant failed to show the typical enhanced antibiotic production when MM was supplemented with GlcNAc (Fig 1b). In other words, GlcNAc sensing had been lost in nagS mutants. To further investigate this, we also looked at siderophore production, as GlcNAc also efficiently blocks siderophore biosynthesis in S. coelicolor M145 on R5 agar plates [18]. We therefore wondered if the biosynthesis of siderophores would still be subjected to catabolite repression by GlcNAc. Interestingly, nagS null mutants had lost the ability to suppress siderophore production on R5 with added GlcNAc (S2 Fig). Taken together, these observations indicate that NagS plays a pivotal role in amino sugar sensing in streptomycetes, highlighting its essential function in the nutrient-dependent regulation of differentiation depending on the extracellular concentration of GlcNAc.

To assess, suppressor mutants can still be obtained in the S. coelicolor nagB null mutant when NagS or NagA levels are enhanced, we expressed either nagA or nagS using the strong and constitutive ermE promoter (S3a Fig). Constitutive expression of nagS in ∆nagB reduced the appearance of suppressor mutants by an order of magnitude when the strains were grown on GlcNAc, while over-expression of nagA prevented the emergence of suppressor mutants even at 10⁸ colony-forming units (CFU). A control experiment showed that introducing P_ermE-nagA in the nagBS double mutant or P_ermE-nagS in the nagAB double mutant did not prevent growth on GlcNAc (S3b Fig). These experiments underline that both NagA and NagS mediate GlcNAc toxicity. Since GlcN-6P, the product of NagA, is not a catalytic substrate of NagS, these data suggest that NagA might have a second function by acting on the product of NagS, leading to the formation of toxic compounds derived from GlcNAc (this is worked out further below).

Phylogenetic analysis of NagS

To see how NagS is distributed across different phyla, we performed phylogenetic analysis. NagS contains a conserved SIS domain [32], predicted to span almost the entire length of the protein. SIS domains are typically found in phosphosugar-binding proteins, including N-acetylmuramic acid-6-phosphate etherase (MurQ) and glutamine-fructose-6-phosphate aminotransferase (GlmS) [33–35]. Phylogenetic analysis of NagS homologs together with other SIS domain proteins revealed that NagS is related to MurQ but forms a well-defined and separate clade in the phylogenetic tree (Fig 1c). Phylogenomic analysis of all NagS proteins in StringDB revealed that NagS is conserved in all streptomycetes except for two marine-derived species (CNH-099 and CNQ-509 [36,37]) in the target database (StringDB v12.0) [38]. Orthologues of NagS were identified exclusively in the Streptomycetaceae, namely in Streptomyces, Kitasatospora, Streptacidiphilus, Actinacidiphila, and a few rare genera with a single representative (Figs 2 and S4a; S4 Table). Their overall amino acid identity was 58.7% or more. Outside the Streptomycetaceae more distant homologs were identified, with an amino acid identity of around 30%. Enzyme assays showed that these enzymes do not have NagS activity and have therefore not been included in the phylogenetic tree (see below). Gene synteny analysis showed that the genomic region around nagS is also conserved, whereby nagS is invariably located adjacent to and divergently expressed from the iron master regulatory gene dmdR1 (SCO4394) [39,40]. Importantly, this linkage between nagS-dmdR1 exists in all Streptomycetaceae (S4b Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Phylogenomic analysis of NagS.

Phylogenetic tree of NagS proteins in the StringDB database. The branches are annotated with the strain names in which the orthologues occur. Streptomyces and Kitasatospora strains are denoted by blue and green markers, respectively. NagS was exclusively found in members of the Streptomycetaceae. The NagS proteins from S. coelicolor and Streptacidiphilus jiangxiensis, whose GlcNAc-6P dehydratase activity was verified by enzyme assays in vitro, are highlighted with a red marker.

https://doi.org/10.1371/journal.pbio.3003514.g002

GlcNAc-6P and ManNAc-6P are substrates of NagS

In order to identify the substrate of NagS, we first expressed S. coelicolor NagS-His₆ in Escherichia coli and purified the protein to homogeneity using Ni-affinity chromatography. Size-exclusion chromatography revealed it to be a dimer (S7 Table, S5 Fig). Isothermal titration calorimetry (ITC) was performed with Glc-6P, Fru-6P, GlcN-6P, GlcNAc, GlcNAc-1P, and GlcNAc-6P. Of these, only GlcNAc-6P showed significant binding to NagS (S6a Fig), suggesting that both the N-acetyl moiety and the C6 phosphate group are important for NagS binding. We then tested binding of GlcNAc-6P and its epimers N-acetylmannosamine-6-phosphate (ManNAc-6P) and N-acetylgalactosamine-6-phosphate (GalNAc-6P) as ligands. This revealed that both GlcNAc-6P and ManNAc-6P were bound by NagS, while GalNAc-6P was not (Fig 1d). Notably, a change in absorbance at 230 nm was detected when either 1 mM GlcNAc-6P or ManNAc-6P was incubated with purified NagS at 30 °C (S6b–S6d Fig), suggesting that NagS converted both compounds in vitro into hitherto unknown products. This is consistent with the ITC result that did not display a simple binding isotherm, as both GlcNAc-6P and ManNAc-6P are substrates of NagS, and substrate turnover is therefore expected to occur during the measurement.

NagS is an N-acetylglucosamine-6-phosphate dehydratase

The product formed from the conversion of GlcNAc-6P by NagS was identified based on NMR and Liquid Chromatography–Mass Spectrometry (LC–MS). Following in vitro reactions using GlcNAc-6P as the substrate, new peaks were detected in the ¹H NMR spectrum specifically for the reaction with active NagS, which were not seen following incubation with heat-inactivated enzyme (S7a Fig). The chemical shifts of the reaction product pointed at the formation of phosphorylated Chromogen I, the 2,3-dehydro derivative of GlcNAc (S5 Table, S8 Fig) [41]. 6P-Chromogen I (2-acetamido-2,3-dideoxy-6-phosphate-D-erythro-hex-2-enofuranose; designated 1) exists as a mixture of its α and β anomers. This was further confirmed by LC–MS analysis, which showed the appearance of two peaks in the reaction mixture containing the active NagS (Fig 1e, 1f). The HRESIMS spectrum of the peaks (S7b Fig) established a molecular formula of C₈H₁₄NO₈P (m/z 282.0387), which is consistent with the structure identified by NMR for 1 (S6 Table). NagS also acts on the other substrate, ManNAc-6P, resulting in its dehydration, although the NMR signals corresponding to 1 were relatively lower in intensity as compared to the reaction with GlcNAc-6P (S7c Fig). Kinetics demonstrated a markedly higher catalytic efficiency of NagS for GlcNAc-6P, with k_cat and K_m values of 24.67 s⁻¹ and 0.45 mM, respectively, which is approximately 40 times greater than that for ManNAc-6P, measured at 0.90 s⁻¹ and 0.68 mM, respectively (Table 1 and S7e, S7f Fig). Taken together, our data show that NagS dehydrates both GlcNAc-6P and ManNAc-6P to produce 1 (S7d Fig). NagS and its homologs constitute a completely novel family of GlcNAc-6P dehydratases within the SIS domain superfamily.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Kinetic parameters for S. coelicolor NagS and NagA.

https://doi.org/10.1371/journal.pbio.3003514.t001

N-acetylglucosamine-6-phosphate dehydratase activity of NagS orthologs

Next, we tested SIS domain proteins with lower homology to NagS, namely proteins from Streptacidiphilus jiangxiensis (TrEMBL A0A1H7F721), Clostridium amylolyticum (TrEMBL A0A1M6IM34), Paenibacillus selenitireducens (TrEMBL A0A1T2XKX6), and Acidothermus cellulolyticus (TrEMBL A0LSD9). The amino acid identities with S. coelicolor NagS are 65.5%, 33.2%, 25.7%, and 35.7%, respectively (S9a Fig). The enzyme assays showed that only the protein from S. jiangxiensis exhibited significant activity, and thus qualifies as a bona fide NagS (designated NagS^Scjia). The ortholog from C. amylolyticum showed negligible activity, while the other enzymes failed to catalyze the dehydration of GlcNAc-6P (S9b–S9e Fig). To verify these data, we investigated whether expression of the different genes could genetically complement S. coelicolor ∆nagB∆nagS by restoring sensitivity to GlcNAc on MM agar plates. Expectedly, introduction of a clone expressing nagS^Scjia in S. coelicolor ∆nagB∆nagS restored sensitivity of the strain to 10 mM GlcNAc (S9f Fig), which confirms that NagS from S. jiangxiensis indeed acts as a real NagS enzyme in vivo. Conversely, introduction of constructs expressing any of the other genes in the nagB-nagS double mutant failed to restore GlcNAc sensitivity (S9f Fig). Therefore, it is likely that the more distant homologs from species outside the family of Streptomycetacea cannot use GlcNAc-6P as a substrate.

6-phosphogluconate is a competitive inhibitor of NagS

Next, we sought to identify potential other interaction partners of NagS to better understand its role in S. coelicolor carbon metabolism. Since 6-phosphogluconate (6-PG) structurally resembles linear GlcNAc-6P (S10a Fig), we assessed the interaction between 6-PG and NagS using thermal denaturation assays. Interestingly, the addition of 2 mM 6-PG shifted the T_m of NagS from 46.45 °C ± 0.4 °C and 58.0 °C ± 0.5 °C, to 48.47 °C ± 0.28 °C and 58.33 °C ± 0.29 °C; while the addition of 5 and 10 mM 6-PG shifted the T_m from two separate peaks to a single merged peak with of T_m at 53.08 °C ± 0.03 °C at 10 mM, when compared against the NagS control (S10b Fig). This suggests that 6-PG may bind to NagS. 6-PG was not metabolized when incubated with NagS (S10c Fig), but instead it effectively inhibited the conversion of GlcNAc-6P (S10d Fig). Further kinetic analysis revealed that while the V_max remained constant, the K_m values for GlcNAc-6P increased with rising concentrations of 6-PG, observed at 0.33 and 1 mM (Figs 1g and S10e). The inhibition constant (K_i) for 6-PG was determined to be 0.28 ± 0.03 mM, supporting its role as a competitive inhibitor. Additionally, the sensitivity of ∆nagB to GlcNAc was mitigated by supplementation with exogenous D-gluconate (S10f Fig).

The crystal structure of NagS

The atomic structures of apo-NagS and its complexes with GlcNAc-6P and 6-PG were determined by X-ray crystallography to resolutions of 2.3, 2.6, and 1.7 Å, respectively (S3 Table). The structures revealed that the biologically active dimer of NagS interacts along a crystallographic 2-fold axis with C2 symmetry (Fig 3a). Two identical substrate binding cavities are located at the dimeric interface (S11a Fig). This dimer interface, which sandwiched the substrate, is stabilized by evolutionarily conserved salt bridges and hydrogen bonds. Structurally, each monomer adopts an α/β configuration with a three-layered αβα sandwich architecture, featuring a standard parallel β-sheet surrounded by nine α-helices. The central β-sheet consists of five strands (β1–β5), contributing to the stability and functionality of the enzyme (Fig 3b).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Crystal structure of NagS.

(a) Dimeric structure of NagS determined by X-ray crystallography. The individual monomers (mon-A and mon-B) of the dimer are shown in green and yellow, with the N- and C- terminus indicated. Sulphates are shown in spheres. (b) Secondary structure of monomeric NagS. α-helices are shown in red, while β-sheets are shown in yellow and loops are shown in green. Monomer of NagS contains nine α-helices, H1-H9, and five β-sheets, β1-β5, as indicated. (c) Alignment of the secondary structure of GlcNAc-6P-bound NagS monomer (cyan) with that of monomer NagS in complex with 6-PG (magenta), with an RMS deviation of 0.17 for all atoms. (d) Alignment of the secondary structure of NagS apo monomer (white) with that of monomer NagS in complex with GlcNAc-6P (cyan), with an RMS deviation of 0.22 (200 to 200 atoms). The significant change of the loop (residues 226-232) is shown in the red box and zoomed in. The loop of NagS in complex with GlcNAc-6P shows a 5.8 Å shift as indicated by the yellow dotted line.

https://doi.org/10.1371/journal.pbio.3003514.g003

The protein structures of NagS bound to either GlcNAc-6P or 6-PG were virtually indistinguishable from each other with an RMS deviation of only 0.17 for all atoms (Fig 3c). However, the apo-form deviated from the substrate-bound structures. In the substrate-bound conformations, the loops connecting H8 and H9 (residues 226–232) had closed in on the substrate (Figs 3d and S11b), resulting in a maximum main chain shift of 5.8 Å at the Cα of Val229. These loops are located at opposite sides of the dimer, and no other structural shifts were observed, indicating that substrate binding is unlikely to be collaborative. The function of the movable loop is therefore most likely to ensure a compact complex between NagS and GlcNAc-6P, and to trigger catalysis by bringing some of its catalytic side chain residues in contact with the substrate.

Analysis of the GlcNAc-6P binding site

Clear electron density at the NagS catalytic site demonstrates that residues from both monomers collaborate to form a tight substrate binding pocket (Fig 4a, 4b). Adjacent to this site, a well-ordered water molecule, likely involved in the initial ring-opening step of catalysis, is observed (Fig 4c). Additionally, several other well-ordered water molecules are integral to the active site structure (see S11c Fig). The phosphate groups of GlcNAc-6P and 6-PG engage Ser54, Ser119, and Ser121 from one monomer, resembling the phosphate-binding mechanism seen in MurQ, where three serine residues play a similar crucial role in phosphate-binding [42,43]. The GlcNAc moiety is stabilized through interactions with Ser54, Ser91, and Glu94, as well as the main chain amide of His53 from the same monomer. A putative transition state after ring-opening, prior to rotations about the C5–C6 and C1–C2 bonds of GlcNAC-6P (Fig 4c), may presumably precede subsequent ring closing. Furthermore, GlcNAc-6P interacts with the side chains of ArgB64 and AsnB228, and the main chain amides of AlaB65 and GlyB227 from the adjacent monomer. Similar interactions are found in the 6-PG-bound state (Fig 4d, 4e). We propose that 6-PG is a competitive inhibitor that stabilizes the enzyme in a catalytically relevant state. Its high affinity for NagS is explained by its nine hydrogen bonds with NagS, whereas GlcNAC-6P forms eight direct hydrogen bonds with NagS, and one indirect hydrogen bond through an ordered water molecule.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Analysis of the binding site of NagS and activity of NagS mutants.

(a) NagS active site with bound GlcNAc-6P (gray carbons), protein residues are coloured with pale blue and yellow carbons to indicate the 2 monomers forming the active site. 2|Fo|-|Fc| electron density contoured at 1.2 σ is shown as cyan mesh. Hydrogen bonds are indicated by dashed yellow lines. (b) GlcNAc-6P binding site of NagS, with hydrogen bonding distances, or distances between hydrogens and hydrogen bond acceptors indicated. The other molecule of the dimer contributes amino acids ArgB64, AlaB65, GlyB227, and AsnB228. The inset indicates the likely electron rearrangements required for ring-opening, the first step of catalysis. (c) Putative transition state after ring-opening, prior to rotations about the C5–C6 and C1–C2 bonds of GlcNAC-6P (indicated in red), that presumably precede subsequent ring closing. These rotations are associated with the rearrangement of hydrogen bonds. This likely requires conformational changes that in the crystal are inhibited by crystal contacts, explaining why the crystals are not enzymatically active. (d) NagS active site with bound 6-phosphogluconate (gray carbons), protein residues are coloured with pale blue and yellow carbons to indicate the 2 monomers forming the active site. 2|Fo|-|Fc| electron density contoured at 1.2 σ is shown as cyan mesh. (e) Hydrogen bonding distances observed in the 6-phosphogluconic-inhibited state of NagS. The inhibited state suggests how GlcNAc-6P rearranges upon ring-opening, and likely reflects the transition state prior to ring closing, which probably involves Ser91, Glu94, and AsnB228. (f) NagS enzyme activity in vivo. GlcNAc sensitivity of ∆nagB∆nagS harboring clones expressing NagS mutants H53A, S54A, R64A, S91A, E94A, S119A, S121A, D179A, or N228A were grown on MM agar supplemented with 1% mannitol (Mann) and 1% mannitol plus 10 mM GlcNAc (GlcNAc). Single colonies are most likely spontaneous suppressors. (g) In vitro activity (V_max) for wild-type NagS (WT) and NagS variants, with the substrate of GlcNAc-6P. The data underlying this Figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3003514.g004

To investigate the roles of specific residues in binding and catalysis, alanine mutants of all residues that interact with GlcNAc-6P through their side chains together with Asp179, which was determined to be essential for NagS activity from the suppressor mutant, were generated and expressed in the S. coelicolor double mutant ∆nagB∆nagS. The phenotypes of these mutants were analyzed on MM agar plates, with and without GlcNAc. Expression of wild-type nagS or its mutant variants, S54A, S119A, or S121A, fully restored GlcNAc sensitivity in the double mutant, indicating that these residues are not crucial for the catalytic activity of NagS. Conversely, mutants S91A and N228A only partially restored GlcNAc sensitivity, underscoring their relevance in the enzyme’s catalytic function. However, constructs expressing the H53A, R64A, E94A, or D179A mutants of NagS showed no restoration of GlcNAc sensitivity, indicating that these residues are essential for NagS activity (Fig 4f).

To corroborate these findings, we determined the kinetic parameters of all of the NagS variants in vitro, which aligned closely with our in vivo observations (Table 1 and S12 Fig). Variants S54A, S119A, and S121A displayed relatively small changes in K_m and k_cat values, compared to the other mutants, confirming these residues are not crucial for catalysis. In contrast, variants S91A and N228A exhibited significantly reduced activity, with very low V_max values (Fig 4g). Specifically, S91A demonstrated a 40-fold decrease in k_cat/K_m, while N228A showed a 208-fold decrease in k_cat/K_m and a 3.9-fold reduction in K_m, indicating that they are important for catalysis. Furthermore, the variants H53A, R64A, E94A, and D179A were completely inactive (Table 1 and Fig 4g), confirming that these residues are critical for catalysis.

A novel promiscuous function for NagA as 6P-chromogen I deacetylase

The observation that the deletion of nagA relieves the toxicity of GlcNAc to nagB null mutants created a mystery in terms of the enzyme’s role in GlcNAc metabolism. NagS is the first step towards a toxicity pathway, and in the absence of NagA, the NagS substrate GlcNAc-6P is expected to accumulate in large amounts when cells grow on GlcNAc, which should therefore be more, rather than less, toxic to the cells. However, nagA nagB double mutants grow very well on GlcNAc. The most likely explanation was that NagA has an additional role besides deacetylating GlcNAc-6P. We therefore wondered if NagA might also deacetylate 6P-chromogen I (1), the reaction product of NagS. To test this, S. coelicolor NagA was expressed in E. coli and purified to homogeneity, and its activity with different substrates was tested. NagA deacetylated GlcNAc-6P, ManNAc-6P and GalNAc-6P in vitro, with a clear preference for GlcNAc-6P (k_cat/K_m of 1.01 × 10⁵ M^-1·s⁻¹) as compared to 6.06 × 10³ M⁻¹·s⁻¹ for ManNAc-6P and 1.03 × 10⁴ M⁻¹·s⁻¹ for GalNAc-6P (Table 1 and S13a–S13c Fig). These kinetic parameters are comparable to those for NagA isozymes reported in E. coli and Mycobacterium tuberculosis [44,45], which shows that the purified NagA was fully active in vitro. Next, we tested if NagA could act on 1. GlcNAc-6P was first incubated with NagS at 30 °C; next, either active NagA or heat-inactivated NagA was added, and the reaction mixture was analyzed by LC–MS. No difference in peak intensity was seen when heat-inactivated NagA was added, showing that 1 was stable under the given reaction conditions. Excitingly, 1 disappeared after adding active NagA, and new products with an exact mass of 240.0280 (S6 Table) were detected in the reaction mixture (Fig 5a, 5b). This shows that NagA is a promiscuous enzyme, whereby besides its textbook function it can also deacetylate 1, the product of NagS. The HRESIMS spectrum of the peaks established a molecular formula of C₆H₁₂NO₇P (m/z 240.0280), which is the deacetylated product of 1 (S13d Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. New function of S. coelicolor NagA and an updated amino sugar metabolic pathway.

Extracted ion chromatograms for compound 1 (m/z 282.0387, shown in black lines) and compound 2/3 (m/z 240.0280, shown in red lines) in the enzymatic reaction mixture of GlcNAc-6P with NagS and NagA (a) or deactivated NagA (b). (c) Updated metabolic pathway of amino sugar in Streptomyces. Based on the metabolic pathway in S1 Fig, we propose a new metabolic route in GlcNAc metabolism. Apart from the canonical reaction whereby GlcNAc-6P is metabolized by NagA and NagB to fructose-6P, GlcNAc-6P can also be dehydrated by NagS and subsequently deacetylated by NagA to form compound 3 (shown in light red), whose is a likely toxic compound whose chemical structure is similar to ribose-5P. (d) Ribose by-passes GlcNAc toxicity. Spores (5 × 10⁵ CFU) of the S. coelicolor M145 nagB mutant were spotted on MM supplemented with 1% mannitol plus 10 mM GlcNAc and different concentrations (0–20 mM) of either glucose or D-ribose, followed by incubation for 72 h at 30 °C. Note that 1 mM ribose or more alleviates GlcNAc toxicity and allows the cells to grow, while even at 20 mM glucose the colonies still are sensitive to GlcNAc. Single colonies are most likely suppressors.

https://doi.org/10.1371/journal.pbio.3003514.g005

Thus, our work reveals a novel metabolic route in central amino sugar metabolism (Fig 5c). Herein, GlcNAc-6P is dehydrated by NagS to 1. Subsequently, 1 is deacetylated by NagA to form enamine 2 (2-amino-2,3-dideoxy-6-phosphate-D-erythro-hex-2-enofuranose), which is unstable and spontaneously converts to the corresponding imine 3 (2-imino-2,3-dideoxy-6-phosphate-D-erythro-hexofuranose).

Supplementation of ribose relieves GlcNAc toxicity to nagB mutants

We then wondered what the basis might be for the toxicity of compound 3. We noticed that imine 3 shows significant structural similarities to ribose-5-phosphate, an intermediate in pentose metabolism and the sugar moiety of nucleic acids. This suggests that 3 may interfere with the synthesis of nucleotides when accumulating at high concentrations. If this is the case, high concentrations of ribose may relieve the toxicity. Therefore, either ribose or glucose was added to cultures of the S. coelicolor nagB mutant grown in the presence of GlcNAc, to determine if these sugars could alleviate the toxicity of GlcNAc. Importantly, nagB mutants were significantly less sensitive to GlcNAc when grown in the presence of higher concentrations of ribose, while glucose did not alter the sensitivity (Fig 5d). This provided an important clue as to the toxicity of 3, which may act by interfering with ribose metabolism, and hence with the synthesis of nucleotides.

Metabolic toxicity and nutrient signaling in Streptomyces

Lysis of the vegetative or substrate mycelia is a key event during colony growth and development, through a process of PCD. Cytotoxic compounds with antiproliferative activity such as prodiginines [17] or anthracyclines [47] play a major role in the onset of PCD in streptomycetes, by killing biomass in the center of the colony [17]. The accumulation of GlcNAc around the colonies is perceived by Streptomyces as the start of lytic degradation of the cell-wall, a major signal for the onset of development. Our data show that amino sugar toxicity mediated by NagS and NagA is required for GlcNAc sensing during the control of early development. Indeed, in the absence of nagS or nagA, GlcNAc sensing is lost. Deletion of nagS in the wild-type strain did not affect the timing of aerial development, but led to reduced formation of aerial biomass. This will negatively affect sporulation, which is a major competitive disadvantage in the natural habitat. The fact that compound 3 structurally resembles ribose-5P provided a clue as to the nature of the toxicity. Indeed, the addition of ribose relieved the toxicity of GlcNAc to nagB mutants, which suggests that the toxicity pathway governed by NagS and NagA might interfere with nucleic acid synthesis. This raises the interesting possibility that compound 3 could modulate enzymes utilizing ribose-5P, which warrants further investigation in future studies.

Based on our data, we propose a model for the role of NagS in GlcNAc signaling (Fig 6). As vegetative growth progresses, degradation of the substrate mycelium is initiated to provide the required for the buildup of the new aerial biomass. Cell-wall lysis results in the release of GlcNAc, which serves as a substrate for the toxicity pathway governed by NagS and NagA. The accumulation of the toxic compound will aid in promoting cell death. In turn, the activity of NagS is metabolically inhibited by 6-PG. 6-PG is an intermediate of the pentose phosphate pathway (PPP) and we therefore propose a salvage pathway (Fig 6). 6-PG is derived in just a few metabolic steps from GlcN-6P, by its conversion to fructose-6P by NagB, then to glucose-6P by phosphoglucose isomerase (Pgi), and to 6-PG via the PPP. This probably also explains why nagB mutants are hyper-sensitive to GlcN-6P or GlcNAc-6P; after all, nagB mutants cannot convert GlcN-6P into Fru-6P and then to 6-PG, thus blocking the salvage pathway, so that NagS stays active. GlcNAc-6P is a key nutrient for streptomycetes, and once levels are low again, aerial growth is initiated. In laboratory experiments using agar plates the pool of GlcNAc (some 20 ml of agar containing 10 mM GlcNAc) can never be depleted by the colonies, so that development is permanently blocked. Therefore, the discovery of NagS appears to be the final piece of the jigsaw in terms of GlcNAc signaling and the control of development.

In conclusion, we report on the function and structure of the amino sugar dehydratase NagS, a novel enzyme that forms a gateway for GlcNAc sensing in Streptomyces. We thereby provide new insights into the molecular basis for the link between amino sugar metabolism and the control of development and antibiotic production in streptomycetes.

Bacterial strains, culture conditions, plasmids, and oligonucleotides

All strains described in this work are listed in S1 Table. Escherichia coli was grown and transformed according to standard procedures [48], with E. coli JM109 serving as the host for routine cloning, and E. coli ET12567 [49] for the isolation of nonmethylated DNA for transformation into Streptomyces [50]. For protein heterologous expression, E. coli Rosetta (DE3)pLysS was used for NagS and E. coli BL21 (DE3) was used for NagA. E. coli was grown in Luria-Bertani (LB) media in the presence of selective antibiotics as required.

Streptomyces coelicolor A (3)2 M145 [51] served as the parent of all mutants. S. coelicolor nagB mutants, ∆nagB, and nagAB mutants, ∆nagAB, and nagB suppressor mutants SMA11, have been described previously [30]. All Streptomyces media and routine techniques are described in the Streptomyces manual [50]. SFM (soy flour mannitol) agar was used for conjugation and cultivation of the spores. Phenotypic characterization was done on R5 medium and MM with 1% (w/v) mannitol supplemented with sugars as stated and, where appropriate, with the antibiotics apramycin (20 μg/ml) and/or thiostrepton (20 μg/ml) as selective markers.

All plasmids and oligonucleotides are summarized in S1 and S2 Tables of the supplemental material, respectively. The shuttle vector pSET152 [52] was used for genetic complementation and over-expression experiments, while the unstable multi-copy shuttle vector pWHM3 [53] was exploited for gene replacement strategies [54]. Cre recombinase expressing plasmid, pUWLcre [55] was used for the creation of deletion mutants via genetic excision via loxP marked sites (see below for details). Expression vector pET-15b and pET-28a (+) (Novagen) were used for proteins heterologous expression. All DNA sequencing was performed by BaseClear BV (Leiden, the Netherlands).

Gene knock-out, complementation and over-expression

The detailed procedure for the creation of S. coelicolor gene replacement and deletion mutants is described previously [56]. Gene replacement mutants were generated via homologous recombination, with the gene of interest replaced by the apramycin resistance cassette aac(C)IV. For this, the upstream and downstream flanking regions of nagS were PCR-amplified from S. coelicolor M145 genomic DNA using primer pairs SCO4393-LF/LR and SCO4393-RF/RR and cloned into pWHM3 using engineered EcoRI/XbaI and XbaI/HindIII restriction sites, respectively. The apramycin resistance cassette flanked by loxP sites was cloned in-between as an XbaI fragment. The resulting knock-out plasmid, designated pKO-4393, was introduced into S. coelicolor and ∆nagB via protoplast transformation. Correct recombination events were verified by appropriate antibiotics resistance and confirmed by PCR. To obtain deletion mutants, the apramycin resistance cassette was excised by introduction of the Cre recombinase expressing plasmid, pUWLcre [55], which allows for efficient removal of the cassette via the loxP recognition sites [57]. Deletion mutants were checked for the appropriate antibiotic sensitivity (loss of apramycin resistance) and confirmed by PCR.

For the complementation of nagS, DNA fragment of nagS coding region was amplified from the S. coelicolor genome DNA and cloned into pSET152, giving the vector pCOM-4393. NagS site-directed mutated sequences were amplified using two pairs of primers: 4393-compF/4393-compR and the corresponding mutated primers. The two amplified fragments were cloned into pSET152 by Gibson assembly [58], producing mutated nagS complementary vectors. The complementary constructs were introduced into nagB-nagS double mutant to obtain the corresponding complemented strain. For nagS and nagA over-expression, a 796-bp DNA fragment containing nagS coding region and a 1185-bp DNA fragment containing nagA coding region were amplified with the primers SCO4393-OE-F/R and SCO4284-OE-F/SCO4284-OE-R, and ligated simultaneously with the 310-bp highly efficient promoter ermE [59] into pSET152 to produce nagS-overexpressing vector pOE-4393 and nagA-overexpressing vector pOE-4284. The vectors were transformed into ΔnagB to generate nagS and nagA-overexpressing strains, respectively.

Heterologous expression and purification of S. coelicolor NagA

For heterologous expression of S. coelicolor NagA in E. coli, the 1146-bp nagA coding regions was amplified from genome DNA using primer pair NagA-exp-F/NagA-exp-R. The PCR fragment was ligated into pET-28a (+) from XhoI and NcoI, generating expression vector pEX-4284. The NagA expression vector was transformed into E. coli BL21(DE3), and the expression of C-terminal His₆-tagged NagA recombinant protein was induced by addition of Isopropyl β-D-1-thiogalactopyranoside at a final concentration of 1.0 mM when the cell density reached around an optical density at 600 nm of 0.6. In addition, 1 mM ZnCl₂ was also supplemented into the cells as a range of NagA isozymes require divalent metal cations for their function [60,61], followed by incubation at 26 °C overnight. Cells were harvested, washed, and disrupted in lysis buffer (50 mM sodium phosphate, 300 mM NaCl, 10 mM imidazole, pH 7.4) by sonication on ice. The preparations were then centrifuged, and the soluble His₆-tagged NagA from supernatant was purified using HisPur Cobalt Resin (Thermo Fisher Scientific; USA). After properly washing with wash buffer (50 mM sodium phosphate, 300 mM NaCl, 10 mM imidazole, pH 7.4), recombinant NagA was eluted from the resin with elution buffer (50 mM sodium phosphate, 300 mM NaCl, 300 mM imidazole, pH 7.4), and then desalted into phosphate buffer (25 mM phosphate pH 7.4, 100 mM NaCl, 1 mM DTT) and stored at −80 °C.

Phylogenetic analysis of NagS

Protein homology searches were performed using BLASTp [62]. The phylogenetic tree of different SIS containing proteins was made by MEGA11 [63]. SyntTax was used for gene synteny [64]. Protein alignments were analyzed by Clustal Omega (www.ebi.ac.uk/Tools/msa/clustalo/) and visualized using Jalview (Version 2.11.2.7).

For comprehensive NagS phylogenetic analysis, S. coelicolor NagS is used as a query to search StringDB (v12.0 10.1093/nar/gkac1000) proteins using diamond (version 2.1.8, 10.1038/s41592-021–01101-x) with “--sensitive” switch and a default e-value threshold of 0.001 (S4 Table). Proteins found are extracted from the database and aligned using MAFFT (version 7.520, 10.1093/molbev/mst010) with a maximum of 1,000 iterative refinement. The alignment is then piped into FastTree (version 2.1.11, [65]) to make a tree with bootstrap values of 1,000 repeats. The phylogenetic tree produced by FastTree is then uploaded to iTOL (version 7.0, 10.1093/nar/gkae268) to produce the figure. The corresponding tree file and annotation dataset can be found in Supplementary material.

Heterologous expression and purification of NagS

For in vitro enzyme experiments and structure elucidation via X-ray crystallography, N-terminally His₆-tagged NagS, its variants with site-directed mutations (H53A, S54A, R64A, S91A, E94A, S119A, S121A, D179A, and N228A) were expressed in E. coli Rosetta (DE3)pLysS (Novagen). Briefly, the nagS coding sequence was PCR-amplified from S. coelicolor genomic DNA or synthesized DNA fragments, while its site-directed mutated sequences were amplified from the vectors for mutated NagS complementation (pCOM-4393-H53A to pCOM-4393-N228A). The amplified fragments were cloned into pET-15b to obtain the expressing plasmids, which were transformed into E. coli Rosetta (DE3)pLysS competent cells. Expression host cells were grown at 37 °C with shaking at 200 rpm in LB media supplemented with chloramphenicol (25 μg/ml) and ampicillin (100 μg/ml) until an OD₆₀₀ of 0.6–0.8 was reached. Protein expression was induced with isopropyl β-D-1-thiogalactopyranoside to a final concentration of 0.5 mM, followed by further incubation at 37 °C with shaking at 200 rpm for 4 h. His₆-tagged NagS and its variants were purified using a Ni-NTA column (GE Healthcare) with Isolation Buffer (500 mM NaCl, 5% glycerol, 50 mM HEPES, 10 mM β-mercaptoethanol, pH 8.0) containing 250 mM imidazole, as described [66]. Fractions containing the target proteins were pooled and concentrated before further purification by size-exclusion chromatography (Superdex 200) on an AKTA Pure FPLC system (Cytiva) with Isolation Buffer. Proteins were desalted with HEPES buffer (20 mM HEPES, 300 mM NaCl, 5% glycerol, 1 mM DTT, pH 7.4) prior to use in crystallization trials and enzyme assays.

Enzyme assays

All enzyme experiments were performed at 30 °C using a Cary 60 UV-Vis spectrophotometer (Agilent). Each kinetic assay was carried out in triplicate and in quartz cuvettes (Hellma) with a path length of 10 mm. The initial velocity (V₀) measurements were reproducible within 10% error. All stocks of enzymes and substrates were kept on ice during the entire experiments. The chemicals used were purchased from Sigma–Aldrich unless stated otherwise.

For NagS kinetic analysis, the rate of reaction was measured by following the increase in UV absorbance at 230 nm as a function of time. The reaction (300 μl in total) was started by the addition of the substrate (final concentration 0–5 mM) to a mixture containing 150 μl phosphate buffer (100 mM phosphate, 100 mM NaCl, pH 7.4) and 220 nM NagS. The extinction coefficient of the product of NagS was experimentally determined to be 2.17 × 10³ M⁻¹·cm⁻¹ (S14 Fig).

Kinetic analysis of S. coelicolor NagA was performed using a previously described direct continuous spectrophotometric assay [44]. Here, the rate of reaction was obtained from measurement of the decrease of the amide absorbance of GlcNAc-6P/ManNAc-6P/GalNAc-6P at 215 nm. The reaction mixture (300 μl) contained 150 μl phosphate buffer, S. coelicolor NagA (150 nM for GlcNAc-6P and 400 nM for ManNAc-6P and GalNAc-6P) and the substrates, and the reaction was initiated by the addition of substrate (concentration from 0.33 to 6.67 mM). The extinction coefficients were 500 M⁻¹·cm⁻¹ for GlcNAc-6P, 408 M⁻¹·cm⁻¹ for ManNAc-6P and 612 M⁻¹·cm⁻¹ for GalNAc-6P, all determined spectrophotometrically. For all kinetics, initial rate data were analyzed and fitted with the Michaelis-Menten model using GraphPad Prism (version 8.3.0).

Identification of the products of the enzymatic reactions

NagS product was identified by a combination of Nuclear Magnetic Resonance (NMR) and LC–MS analysis. The reaction samples for NMR were prepared in phosphate buffer, in which 6.67 mM GlcNAc-6P/ManNAc-6P was incubated with 220 nM NagS at 30 °C, and a sample with boiled NagS was used as a control. After reaching equilibrium at 10 min, the reaction mixture was freeze-dried and then dissolved in 170 μl D₂O, and transferred to NMR tubes. The NMR spectra were acquired on a Bruker AVIII-600 spectrometer (Bruker BioSpin GmbH) at a field strength of 600 MHz.

Structural data collection and structure determination

Purified NagS at a concentration of 15–20 mg/ml was screened for crystallization by sitting-drop vapor-diffusion using the PGA Screen (Molecular Dimensions), Clear Strategy Screens CSS-I and CSS-II (Molecular Dimensions), JCSG+ (Qiagen/Molecular Dimensions), and the PACT screen (Molecular Dimensions) as well as optimization screens at 20 °C. The 75 μl reservoir of 96-well Innovaplate SD-2 plates was pipetted by a Genesis RS200 robot (Tecan) and drops were made by an Oryx6 robot (Douglas Instruments). Hexagonal NagS crystals (space group P6₅22) were obtained from JCSG number 83 (96-well G11), which consisted of 2.0 M Ammonium sulphate, 0.1 M BIS-Tris, pH 5.5. Crystals were soaked in mother liquor with 10%–20% glycerol as cryoprotectant, that included no other additives, or either 100 mM of the substrate GlcNAc-6P, or 200 mM of the inhibitor 6-phosphogluconate. After loop mounting, they were flash-frozen in liquid nitrogen.

X-ray data were collected at the European Synchrotron Radiation Facility (Grenoble, France) on beamline ID-23 for the apo-enzyme: 1,410 frames were collected on a Pilatus 6M detector at an X-ray wavelength of 0.9724 Angstroms, an exposure time of 0.037 s, transmission of 10% and an oscillation range of 0.2°. NagS in complex with GlcNAc-6P data were collected on beamline ID-29 with a Pilatus 6M detector. For the native crystal, 1,020 images were collected at 1.2727 Å wavelength with an exposure time of 0.02 s, transmission of 100% and an oscillation range of 0.05°. We collected 680 frames at 0.976251 Angstroms wavelength with an exposure time of 0.02 s, transmission of 47.34% and an oscillation range of 0.1°. XDS [67] was used to process all the data collected. Scaling and merging were done using the CCP4 program ‘aimless’ [68]. Subsequent structural analysis of the crystals soaked in substrate, revealed that the crystals did not significantly catalyze its conversion, see Results section. The diffraction data of the 6-PG inhibited crystals were collected at the Diamond synchrotron radiation facility on beamline I04. 3,600 images were collected at a wavelength of 0.9 Å with an exposure of 0.05 s, transmission of 100% and oscillation of 0.05°. Data processing was performed via Xia2 [69]. The resolution of the inhibited crystal form was 1.7 Å, which was significantly higher than the apo- and substrate-bound crystals (2.3 Å and 2.6 Å, respectively).

The structure of apo-NagS was solved by molecular replacement using the structure of a putative phosphoheptose isomerase from B. halodurans C-125 (PDB code 3CVJ) as the model. Two subunits were present in the asymmetric unit, and in the initial stages of structure refinement, noncrystallographic symmetry restraints were imposed. Clear densities corresponding to well-ordered water molecules and either 6-PG, or GlcNAc-6P for the substrate-bound crystal forms, were observed for all three crystal forms. Models that included the inhibitor, or the substrate where appropriate, were inspected and interactively rebuilt using Coot [70,71].

Water molecules were introduced automatically in refinement using Phenix [72]. Then, noncrystallographic symmetry restraints were removed and six consistent TLS fragments per monomer were automatically generated for the inhibited crystal form and refined by Phenix. Refinement of the apo- and substrate-bound structures benefitted from reference model torsion angle restraints provided by the inhibited structure, as judged by a significant improvement in R-free. TLS refinement using the same fragments as in the inhibited structure, improved the substrate-bound, but not the apo-structure. S3 Table shows the data collection and refinement statistics for the data sets obtained.

ITC assays

To identify the possible substrates of NagS, ITC tests were performed using a MicroCal PEAQ-ITC Automated microcalorimeter (Malvern Panalytical Ltd, Malvern, UK). A 700 µM solution of substrates (GlcNAc-6P, ManNAc-6P, or GalNAc-6P) in 20 mM Tris pH 7.4, 100 mM NaCl, 50 µM DTT, was titrated into a 50 µM enzyme preparation in the same buffer. Control titrations included the titration of buffer into enzyme and substrate into buffer. The samples were equilibrated to 25 °C prior to measurement. The titrations were conducted at 25 °C under constant stirring at 750 rpm. Each experiment consisted of an initial injection of 0.4 µL followed by 18 separate injections of 2.0 µL into the sample cell of 200 µL. The time delay between each injection was 180 s, the measurements were performed with the reference power set at 5 μcal∙s⁻¹ and the feedback mode set on “high”. The calorimetric data obtained were analyzed using MicroCal PEAQ-ITC Analysis Software Version 1.20 (Malvern Panalytical Ltd, Malvern, UK). ITC data fitting was made based on the software’s “one set of sites” fitting model. The best fit is defined by chi-squared minimization. Thermodynamic parameters are reported as the average of three experiments with the standard deviation, unless stated otherwise.

Analytical size-exclusion chromatography

The oligomeric state of purified NagS was characterized by analytical size-exclusion chromatography. Protein at various concentrations (35 μM and 70 μM) was injected onto a Superdex 200 10/300 GL gel-filtration chromatography column (Cytiva) equilibrated in a buffer containing 20 mM HEPES pH 7.5, 300 mM NaCl, 5% glycerol, and 1 mM DTT. A calibration curve (S5 Fig) relating elution volume (S7 Table) to molecular weight was generated by injecting a Protein Standard Mix (15–600 kDa; Sigma, product number: 69385) onto the same column with the same buffer.

Extinction coefficient measurement of the product of NagS

To measure and calculate the extinction coefficient of NagS, an indirect method was used instead of spectrophotometry. As only a portion of the substrate is converted by NagS after the reaction, the concentration of the product is obtained by subtracting the final concentration of GlcNAc-6P, from the initial concentration of GlcNAc-6P (3.33 mM). First, the standard curve of the linear relationship between the concentrations of GlcNAc-6P (1–5 mM) and the peak areas of GlcNAc-6P measured by LC-MS was constructed (S14 Fig). The peak areas of GlcNAc-6P after the reaction were also measured (mean = 31467.63), thus the concentration of the remaining GlcNAc-6P was calculated as 2.15 mM using the standard equation. Thus, the concentration of NagS product was calculated as 1.18 mM (initial substrate concentration 3.33 mM minus final substrate concentration 2.15 mM). The final absorbance caused by the product was measured spectrophotometrically, which was 2.564. Therefore, the extinction coefficient of the product of NagS was calculated as 2.17 × 10³ M⁻¹ cm·¹ by dividing the absorbance brought by the product by its concentration. All measurements involved were done in triplicate.

NagS-NagA coupled enzymatic assays

For NagS-NagA coupled assay, the reactions were carried out in the volatile buffer (50 mM N-ethylmorpholine/acetate pH 7.4), which is compatible with the subsequent LC–MS analysis. GlcNAc-6P (10 μl, 100 mM) was first incubated with NagS, followed by the addition of 150 nM NagA or heated-inactivated NagA after the NagS-catalyzed reaction reached equilibrium. The reaction mixture was monitored at 230 nm until the absorbance stabilized, and then subjected to mass spectrometry analysis.

Thermal denaturation assays

The thermal stability and melting point of the protein were identified through measuring the intensity of SYPRO orange fluorescence. Upon the unfolding of the protein when temperature rises, the intensity of SYPRO orange increases due to the increased exposure of hydrophobic regions [73]. This change is monitored using a TOptical Real-Time qPCR Thermal Cycler (Biometra). NagS was tested for the thermal ability with or without 6-PG. For this, 200 μl of NagS at 4 μM was made up using the isolation HEPES buffer and 5X SYPRO orange (Invitrogen). A set of 3 repeats with a final volume of 50 μl of the protein solution per well was dispensed into a MicroAmp Optical 96-Well Reaction Plate (Applied Biosystems) and sealed by a Microseal ‘B’ seal (BIO-RAD). The plate was then centrifuged at 2,000 rpm for 2 min at 10 °C using an Allegra X-15R centrifuge fitted with a µSX4250 rotor prior to the experiment. The level of SYPRO orange intensity was monitored between 20 and 75 °C. Additionally, 6-PG (2, 5, and 10 mM) was tested to determine its possible effect on the thermal stability of NagS.

Inhibitor kinetic assays

The inhibitory activity of 6-PG with respect to the enzymatic activity of NagS was determined. For this, the rate of reaction was measured by following the increase in UV absorbance at 230 nm as a function of time. Each assay (300 μl in total) was initiated by the addition of GlcNAc-6P (final concentration 0–8 mM) to a mixture containing 150 μl phosphate buffer (100 mM phosphate, 100 mM NaCl, pH 7.4), 6-PG (0.33 or 1 mM), and 220 nM NagS. Initial rates were analyzed and fitted with the competitive inhibition model using GraphPad Prism (version 8.3.0). The rate equation used for K_i calculation is K_m’ = K_m*(1+[I]/K_i).

Detection of compound 1–3 by LC–MS

For LC–MS analysis, the reactions were performed in a volatile buffer. In brief, 1 mM GlcNAc-6P was incubated in 50 mM N-ethylmorpholine acetate pH 7.4 at 30 °C, and the reactions were initiated by the addition of 220 nM active NagS or boiled NagS (for the control sample). After the absorbance of the reaction mixtures had stabilized at 230 nm, the enzyme in the reaction mixture was precipitated with methanol and the supernatant obtained was then used for LC–MS analysis. LC–MS/MS acquisition was performed using a Shimadzu Nexera X2 UHPLC system, with attached PDA, coupled to a Shimadzu 9030 QTOF mass spectrometer, equipped with a standard ESI source unit. A total of 2 µL of each sample was injected into a Waters Acquity HSS C₁₈ column (2.1 × 100 mm), which was run at a flow rate of 0.5 ml min⁻¹ using 0.1% formic acid in H₂O as solvent A, and 0.1% formic acid in acetonitrile as solvent B. The elution gradient used was 5% B for 1 min, 5%–17% for 1 min, 17%–20% for 8 min, then 20%–100% for 1 min. All samples were analyzed in negative polarity, using data dependent acquisition mode, in which full scan MS spectra (m/z 100–1,700, scan rate 10 Hz, ID enabled) were followed by two data dependent MS/MS spectra (m/z 100–1,700, scan rate 10 Hz, ID disabled) for the two most intense ions per scan at a collision energy of 20 eV. The parameters used for the ESI source were: interface voltage −3 kV, interface temperature 300 °C, nebulizing gas flow 3 L min⁻¹, and drying gas flow 10 L min⁻¹.