Streptomyces coelicolor M145 (parental strain) and mutants were cultured for 4–6 days at 30°C on the defined Evans-agar base (modified after Evans et al., 1970) supplemented with different nitrogen sources: 50 mM monosodium glutamate, L-glutamine, ammonium chloride, sodium nitrate, urea, cadaverine dihydrochloride, putrescine dihydrochloride, spermidine trihydrochloride and spermine tetrahydrochloride in appropriate concentrations. Growth experiments in liquid culture were performed using either complex S-medium (Okanishi et al., 1974) or YEME:TSB (1:1) (Kieser et al., 2000) and chemically defined medium Evans-medium (modified after Evans et al., 1970). If appropriate, media were supplemented with polyamines or ammonium chloride as a sole nitrogen source. Strains were cultivated at 30°C, on the rotary shaker (180 rpm) for 4–6 days. Genetic manipulation of S. coelicolor M145 was performed as described by Kieser et al. (2000) and Gust et al. (2003). For preparation of the genomic DNA, S. coelicolor M145 was grown for 4 days in S-medium and DNA was isolated with the NucleoSpin Tissue Kit (Macherey-Nagel). All strains and plasmids used in this study are listed in the Table 1.

To perform a complementation of the mutant, glnA3 gene without its native promoter was amplified by PCR using CglnA3f and CglnA3r primers (Table 2) and cloned into the multiple cloning site of pRM4 plasmid. The correct construct was confirmed by sequencing and subsequently introduced into the ΔglnA3 mutant by conjugation using the E. coli S17 strain. Clones were then selected on resistant phenotype against kanamycin and apramycin. The correct integration of the pRM4-glnA3 was confirmed by PCR and sequencing.

REDIRECT gene-replacement procedure (Gust et al., 2003) was employed to generate in-frame deletion of the glnII gene (SCO2210) in previously generated ΔglnA mutant (Fink et al., 1999). The glnII gene on the cosmid St3H12 was replaced by a hygromycin resistance cassette in E. coli BW25113. The hygromycin resistance cassette was amplified by PCR using glnII_hygr_P1 and glnII_hygr_P2 primers. The obtained cosmid St3H12ΔglnII was then transferred into E. coli ET12567/pUZ8002 by electroporation and subsequently transferred into S. coelicolor M145 ΔglnA by conjugation. Hygromycin and apramycin resistant and kanamycin sensitive trans-conjugants were isolated from MS-plates supplemented with L-glutamine. Correct marker-less deletion of glnII was confirmed by PCR followed by sequencing as well as Southern blot analysis.

Level of the intracellular polyamines was analyzed using high-performance liquid chromatography (HPLC), as described by Potter and Paton (2014). The method was modified and optimized for S. coelicolor. Cells were harvested by centrifugation (6,000 ×g, 10 min at 4°C), washed three times in phosphate-buffered saline (PBS). The 500 mg of the wet cells were resuspended in morpholinepropanesulfonic acid (MOPS) lysis buffer (100 mM MOPS, 50 mM NaCl, 20 mM MgCl2, pH 8.0) and disrupted using glass beads (150–212 μm, Sigma) and a Precellys homogenizer (6500 rpm, 20 – 30 s; Peqlab). Samples were centrifuged (13,000 ×g, 10 min at 4°C), the pellet was discarded and supernatant was transferred into new tube. Trichloroacetic acid then was added to a final concentration of 10%, and the mixture was incubated on ice for 5 min. After incubation the mixture was cleared by centrifugation (13,000 ×g, 10 min at 4°C), the pH of each sample was neutralized using HCl and the samples were stored at −20°C until analysis. Polyamines were derivatized using pre-column derivatization with ortho-phthalaldehyde (OPA)/mercaptoethanol (MCE) (Dr. Maisch GmbH, Ammerbuch). OPA-derivatized polyamines were separated on a Reprosil OPA column (150 mm × 4.0 mm, 3 μm) fitted with a precolumn (10 mm × 4 mm, same stationary phase) (Dr. Maisch GmbH, Ammerbuch) using a HP1090 Liquid Chromatograph equipped with a diode-array detector, a thermostated autosampler and a HP Kayak XM 600 ChemStation (Agilent, Waldbronn). UV detection was performed at 340 nm. The following gradient was used at a flow rate of 1.1 ml/min with solvent A (25 mM sodium phosphate buffer, pH 7.2 containing 0.75% tetrahydrofuran) and solvent B [25 mM sodium phosphate buffer, pH 7.2 (50%), methanol (35%), acetonitrile (15%) by volume]: t₀ = 60% B, t_2.5 = 70%, t₁₂ = t₂₂ = 100% B (time in minutes). Cadaverine dihydrochloride, putrescine dihydrochloride, spermidine trihydrochloride, and spermine tetrahydrochloride were purchased from Sigma and used as standards. Standard solutions of polyamines were prepared with distillated water. HPLC detection limit for different polyamines was 0.1 mM.

For the transcriptional analysis experiments, the S. coelicolor M145 wild type and the glnA3 mutant were grown in S-medium. After 4 days, cells were washed twice with defined Evans medium and cultivated for 24 h in defined Evans medium supplemented either with 25 mM ammonium chloride or 25 mM polyamine (cadaverine, spermidine, or putrescine). RNA isolation was performed with an RNeasy kit (Qiagen). RNA preparations were treated twice with DNase (Fermentas). First, an on-column digestion was carried out for 30 min at 24°C, and afterward RNA samples were treated with DNase for 1.5 h at 37°C. RNA concentrations and quality were checked using a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific). cDNA from 3 μg RNA was generated with random nonamer primers (Sigma), reverse transcriptase and cofactors (Fermentas). PCR reactions were performed with the primers listed in Table 2. The PCR conditions were 95°C for 5 min; 35 cycles of 95°C for 15 s, 55–60°C for 30 s and 72°C for 30 s; and 72°C for 10 min. As a positive control, cDNA was amplified from the major vegetative sigma factor (hrdB) transcript, which is produced constitutively. To exclude DNA contamination, negative controls were carried out by using total RNA as a template for each RT-PCR.

The ability of the S. coelicolor M145 and ΔglnA3 mutant to survive in the presence of polyamines was examined by the estimation of the dry weight biomass after 72 and 168 h of incubation in the YEME–TSB (1:1) rich complex medium. Both strains, M145 and ΔglnA3 mutant were inoculated into YEME–TSB supplemented with putrescine, cadaverine, spermidine as well as spermine (25 mM of each) and incubated for 3 days at 30°C. As a control M145 and ΔglnA3 mutant were cultivated YEME:TSB without polyamines supplementation and incubated as described above. Samples (1 ml) were taken every day and cells were harvested by centrifugation (16,200 ×g, 15 min at 4°C), the residual liquid was removed and pellets were dried for 24 h. The middle value of three technical replicated was compared with the middle value of three biological replicates and the standard error was calculated.

Morphology and viability of S. coelicolor M145 and ΔglnA3 mutant cells grown in a rich complex YEME–TSB (1:1) medium was analyzed in presence of polyamines. Samples were taken from every culture after 72 and168 h of growth, and obtained perpetrates were observed under phase-contrast microscope under 400× enlargement. To detect live and dead cells, SYTO9 and PI (propidium iodide) stains of the LIVE/DEAD BacLight Bacterial Viability Kit (Molecular Probes) were used. The SYTO 9 green fluorescent stain labels cells with intact membranes, as well as those with damaged ones. PI penetrates cells with damaged membranes, decreasing SYTO 9 stain fluorescence when both dyes are present. Thus, in the presence of both SYTO9 and PI, cells with intact cell membranes appear fluorescent green whereas cells with damaged membranes appear red. The staining solution was prepared by mixing 0.75 μl of component A and B in 500 μl of water. Stained cells were analyzed by the fluorescence microscopy using Olympus BX60 microscope with an Olympus UPlanFl 100 × oil objective and an Olympus BX-FLA reflected light fluorescence attachment. Images were taken with the F-view II camera (Olympus), using TxRed and eGFP filter sets for detection of the fluorescent markers. The ImageJ was used for image processing. Significant number of images (10) was analyzed in a minimum of three independent culture analyses.

The gene encoding GlnA3 (SCO6962) was amplified by PCR form S. coelicolor genomic DNA and ligated into the expression vector pJOE2775 (Volff et al., 1996) under the control of rhamnose inducible promoter Prha. The restriction sites NdeI and HindIII were used for cloning purposes. His-GlnA3 was overexpressed in E. coli strain BL21 (DE3) using the auto-induction method for protein production (Li et al., 2011; Studier, 2014). Cells were grown initially at 37°C until the culture density reached an optical density of ∼0.6 at 600 nm at which time the temperature of the shaking incubator was reduced to 20°C. The bacterial cells were then harvested by centrifugation after 48 h cultivation and stored at −20°C until needed. His-GlnA3 was purified by nickel ion affinity chromatography essentially as directed by the resin manufacturer (GE-Healthcare). Purified His-GlnA3 was dialyzed against 20 mM Tris, 100 mM NaCl (pH 8) and immediately used for further analysis.

HPLC/ESI-MS was used to evaluate putrescine and glutamate as GlnA3 substrates and generated gamma-glutamylated putrescine as a product. Standard reactions typically contained 20 mM HEPES (pH 7.2), 150 mM glutamate sodium monohydrate, 150 mM putrescine dihydrochloride, 20 mM MgCl₂ × 6H₂O, and 10 mM ATP were mixed with 10 μg of the purified His-GlnA3 (or without GlnA3 as a control) and incubated at 30°C for 5 min. The reaction was stopped by incubation of the reaction mixture at 100°C for 5 min. HPLC/ESI-MS analysis of the glutamylated product generated by GlnA3 was done on an Agilent 1200 HPLC series using a Reprosil 120 C18 AQ column, 5 μm, 200 mm × 2 mm fitted with a precolumn 10 mm × 2 mm (Dr. Maisch GmbH, Ammerbuch, Germany) coupled to an Agilent LC/MSD Ultra Trap System XCT 6330 (Agilent, Waldbronn, Germany). Analysis was carried out using 0.1% formic acid as solvent A and acetonitrile with 0.06% formic acid as solvent B at a flow rate of 0.4 ml min⁻¹. The gradient was as follows: t₀ = t₅ = 0% B, t₂₀ = 40% B (time in minutes). Injection volume was 2.5 μl, column temperature was 40°C. ESI ionization was done in positive mode with a capillary voltage of 3.5 kV and a drying gas temperature of 350°C.

In order to determine whether GlnA2, GlnA3, and GlnA4 were involved or not involved in glutamine biosynthesis in S. coelicolor M145, single and double mutants of genes glnA and glnII encoding proven GSs, GSI and GSII, respectively, were deleted in the parental strain M145. REDIRECT gene-replacement procedure (Gust et al., 2003) was employed to delete glnII gene (SCO2210) in a previously generated ΔglnA mutant (Fink et al., 1999) as described in Section “Materials and Methods.” Independent clones of the single ΔglnA and ΔglnII mutants, of the double ΔglnAΔglnII mutant and of the parental strain M145 were tested for their ability to grow on defined Evans-agar supplemented with ammonium chloride (50 mM) as sole nitrogen source. Glutamine synthetases, GSI and GSII catalyze condensation of L-glutamate and ammonia to form L-glutamine in the ATP dependent reaction. Consequently, growth reveals the ability of the strains to assimilate ammonium and to synthesize glutamine whereas absence of growth indicates absence of this ability. S. coelicolor M145 and the single deletion mutants, ΔglnA and ΔglnII, were able to grow on the plates, demonstrating that both glnA and glnII encode functional GSs as reported previously (Reuther and Wohlleben, 2007). In contrast, the ΔglnAΔglnII double mutant was unable to grow on this medium (Figure 2) and was auxotrophic for glutamine. Addition of L-glutamine restored growth of the ΔglnAΔglnII double mutant on this medium. These results indicated that GlnA2, GlnA3, and GlnA4 do not have a GS function that can substitute for that of GSI and GSII and avoid glutamine auxotrophy.

In an attempt to decipher the function of GS-like enzymes, the genes glnA2, glnA3, and glnA4 were deleted in the parental strain S. coelicolor M145. The ΔglnA2 (Nentwich, 2010), ΔglnA3 (Rexer et al., 2006), and the ΔglnA4 (this work) mutants were tested for their ability to utilize different nitrogen sources. Phenotypic analysis was performed on defined Evans-agar supplemented with the following sole nitrogen sources: ammonium chloride, sodium nitrate, monosodium L-glutamate, L-glutamine, putrescine dihydrochloride, cadaverine dihydrochloride, spermidine trihydrochloride, and spermine tetrahydrochloride. Growth and morphology of the parental strain M145 and the ΔglnA2, ΔglnA3, and ΔglnA4 mutants were monitored on defined Evans-agar after 3 to 12 days of incubation at 30°C. Remarkably, S. coelicolor M145 could grow on all tested nitrogen sources (Figure 3). Interestingly, growth of the S. coelicolor M145 on polyamines resulted in delayed aerial mycelium and spore formation as well as abolition of actinorhodin (blue antibiotic) and enhanced prodigiosin (red antibiotic) production (Figure 3). Growth of the glnA2, glnA3, and glnA4 mutants was similar to that of the parental strain in the presence of ammonium chloride, sodium nitrate, L-glutamine, and monosodium L-glutamate. Growth of the glnA3 mutant, but not that of the glnA2 and glnA4 mutants, was inhibited in the presence of putrescine, cadaverine, spermidine, and spermine (Figure 3). The complementation of the glnA3 mutant by the glnA3 gene under control of the constitutive promoter (PermE) restored the growth of this mutant on polyamine plates (Supplementary Figure S1). Altogether these results indicated that the disruption of glnA3, but not that of glnA2 and glnA4, completely abolished the utilization of polyamines as a nitrogen source. GlnA2 and GlnA4 have obviously a function different from that of GlnA3 since their deletion did not influence polyamine utilization. Interestingly, the glnA3 mutant showed defect in formation of aerial mycelium as well as sporulation on defined medium supplemented with glutamate (50 mM). Moreover, enhanced actinorhodin and prodigiosin production was observed on the nitrate, glutamine and ammonium plates, respectively (Figure 3), indicating an impact of glnA3 deletion on the secondary metabolites production and morphological differentiation in S. coelicolor.

Reverse transcription/PCR experiments were performed as described in Section “Materials and Methods” to determine whether glnA3 expression was influenced by N- and/or C-concentrations as reported for glnA and glnII (Amin et al., 2016) as well as to investigate the expression of glnA3 in the presence of polyamines. For this purpose, parental strain M145 was grown in complex S-medium subsequently transferred into defined Evans medium supplemented either with polyamines (25 mM) or ammonium chloride (25 mM) as a sole nitrogen source. Total RNA isolated after 24 h from S. coelicolor M145 was used to generate cDNA. Subsequently, reverse transcription/PCR analysis was performed using primers internal to glnA3. HrdB, encoding the main sigma factor of RNA polymerase, that has relatively constant levels of expression throughout growth (Buttner et al., 1990) was used as internal control. Expression of genes encoding GSs, glnA and glnII, was shown to be strongly enhanced in condition of ammonium limitation and reduced or completely abolished in condition of ammonium proficiency in S. coelicolor, irrespective of the glucose concentration (Amin et al., 2016). Interestingly, the expression of glnA3 was also strongly enhanced under ammonium limitation, but only in combination with low glucose concentration suggesting that its expression does not require polyamines. Furthermore, transcriptional analysis of glnA3 revealed strong induction of the expression of this gene in the presence of putrescine, cadaverine and spermidine (Figure 4), putative substrates of the encoded enzyme.

Polyamine uptake and utilization pathways are largely uncharacterized in streptomycetes. BlastP analysis was used to predict uptake systems and putative enzymes involved in polyamine utilization in S. coelicolor, whose amino acid sequences are closely related to the amino acid sequences of the characterized proteins from E. coli and P. aeruginosa (Kusano and Suzuki, 2015). Similarity based prediction revealed 18 strong homologs in S. coelicolor (genome accession number AL645882). Their amino acid identities ranged from 24 to 50%, while their similarities ranged from 41 to 65% (Figure 5). To demonstrate that these genes were induced by polyamines, their expression patterns were analyzed in the presence of putrescine, cadaverine, and spermidine as sole N-source as well as ammonium as control.

Analysis of the glnA3 genomic region revealed two genes SCO6960 and SCO6961 located downstream of glnA3 and likely co-transcribed with the latter. SCO6961 encodes an uncharacterized protein belonging to the amidohydrolase 2 family, whereas SCO6960 encodes a small protein of unknown function conserved among streptomycetes. However, no orthologs of these genes could be found in E. coli and P. aeruginosa, suggesting that the polyamine utilization pathway might be different in S. coelicolor. Transcriptional analysis of the SCO6960 and SCO6961 indicated that the expression of these genes was inducible by polyamine as that of glnA3 (Figure 5). In contrast, the expression of the genes SCO6963, SCO6964, and SCO6965 located upstream of this putative operon and encoding uncharacterized lipoprotein and two small proteins of unknown function were not induced in conditions of ammonium limitation nor in the presence of polyamine (data not shown).

Furthermore, BlastP analysis predicted two ABC transporters and two permeases potentially involved in the uptake of polyamines. The SCO5667-70 and SCO3453-56 predicted gene products resemble the ABC transport systems PotFGHI or SpuEFGH from E. coli and P. aeruginosa, respectively. The expression of these genes was only weakly inducible in the presence of polyamines (Figure 5). In contrast, the expression of the genes encoding the putative polyamine permeases SCO5057 and SCO5977 was strongly induced in the presence of all polyamines or only putrescine, respectively (Figure 5). The expression of SCO5658 (encoding predicted polyamine binding protein) was strongly induced in the presence of spermidine (Figure 5).

Prediction of genes encoding enzymes involved in the post-glutamylation steps in S. coelicolor revealed orthologs for gamma-glutamylpolyamine oxidoreductase (SCO5671), polyamine-pyruvate/2-oxoglutarate transaminase (SCO5655) as well as close orthologs of (gamma-glutamyl-) gamma-aminobutyraldehyde dehydrogenases PatD and PuuC (SCO5657 and SCO5666) and SCO5676 encoding predicted GabT-like gamma-aminobutyrate transaminase and finally SCO5676, encoding predicted succinate semialdehyde dehydrogenase (GabD-like). The expression of all these genes (except SCO5671) was strongly induced in the presence of polyamines (Figure 5). Altogether, putative homologs of genes encoding predicted polyamine uptake systems and proteins involved in the polyamine utilization in S. coelicolor showed enhanced expression in the presence of polyamines whereas their expression was abolished in the presence of ammonium. To summarize, the search for proteins orthologs of those involved in the catabolism of polyamine in other bacteria using BlastP and the analysis of gene expression patterns, led us to predict, so far unknown, polyamine degradation pathway in S. coelicolor (Figure 1).

In an attempt to understand the basis of polyamine toxicity in the glnA3 mutant grown in the presence of polyamine, the intracellular polyamine content of S. coelicolor M145 and glnA3 mutant was analyzed by HPLC. To do so, the parental strain M145 and the glnA3 deletion mutant were cultivated in a complex S-medium for 4 days at 30°C. Cells were harvested and washed with Evans medium without N-source to remove remaining complex N-sources. Washed cells were transferred into defined Evans medium containing polyamines as a sole N-source and incubated at 30°C for further 4 days. Samples for the biomass determination and extraction of intracellular polyamines were collected every day. Cell pellets and supernatants were used to extract intracellular and extracellular polyamines, respectively. Comparison of intracellular polyamine levels of both strains revealed three, six and twenty fold higher levels of putrescine, cadaverine, and spermidine in ΔglnA3 mutant, respectively (Figure 6) than in the parental strain after 4 days of cultivation. The intracellular concentration of polyamines decreased over time in the parental strain M145, but not in the mutant strain. After 4 days of cultivation, intracellular polyamine concentrations ranged between 0.5 and 0.7 μmol/g of biomass, whereas that of the glnA3 mutant ranged between 3.1 and 5.1 μmol/g. Furthermore, comparison of extracellular polyamine levels in the parental strain culture and glnA3 culture revealed strong differences. The residual level of polyamines in the glnA3 mutant culture supernatant remained high from the first to the fourth day of incubation (Figure 7), whereas after 4 days of incubation no peak for polyamines could be detected in the culture supernatant of the parental strain. Altogether, these data indicated that the glnA3 mutant was able to uptake polyamines, but was unable to degrade them. The absence of polyamine degradation in the glnA3 mutant obviously resulted in high concentration of intracellular polyamine, likely to be responsible for the inhibition of further uptake of external polyamines (Figure 7).

The ΔglnA3 mutant and its parental strain S. coelicolor M145 were cultivated in a rich complex liquid medium (YEME:TSB) supplemented or not (control) with polyamines (putrescine, cadaverine, spermidine, and spermine, 25 mM of each). Biomass yields as well as morphology of both strains were monitored over 7 days. The ΔglnA3 mutant yielded significantly lower biomass than the parental strain after 72 and 168 h of cultivation in the complex medium supplemented with polyamines (Figure 8). Prolonged incubation of the ΔglnA3 mutant culture did not result in any significant biomass increase. In this medium, S. coelicolor M145 showed its typical pellet growth whereas growth of the ΔglnA3 mutant was dispersed. This indicated that despite the presence of alternative nitrogen sources in this complex medium, besides polyamines, growth of ΔglnA3 mutant was impaired most likely because of intracellular accumulation of polyamines.

The mycelium of the ΔglnA3 mutant grown in the complex medium with polyamines was observed under microscope. After 3 days of cultivation, the parental strain formed a long and compact, branched mycelium constituting pellets whereas the ΔglnA3 mutant formed non-branched mycelium (Figure 9). The mycelial fragments of the ΔglnA3 mutant were significantly shorter and thicker than those of the parental strain M145 after 3 or 7 days of cultivation in the medium containing polyamines (Figure 10).

In order to assess more precisely the toxic effect of polyamines, the viability of the ΔglnA3 mutant was assessed with SYTO9/PI staining. The SYTO 9 green fluorescent stain labels all cells, whereas PI penetrates only cells with damaged membranes, decreasing SYTO 9 stain fluorescence when both dyes are present. In the presence of both, SYTO9 and PI, cells with intact cell membranes appear fluorescent green and dead cells with damaged membranes appear red. Whereas the parental strain M145 appeared green (alive) in the presence of polyamines, most filaments in the culture of the ΔglnA3 mutant were red (dead) after 7 days of incubation (Figure 9). These results demonstrated that the absence of GlnA3 led to the toxic accumulation of polyamine that impaired growth and viability and suggested that GlnA3 plays a major role in polyamine degradation.

The GlnA3 protein belongs to large class of ligases which form carbon-nitrogen bonds using energy from ATP (class 6.3.1). This class contains twenty different subclasses of enzymes including GS (class 6.3.1.2), glutamate – ethylamine ligase (class 6.3.1.6) and glutamate-putrescine ligase (6.3.1.11). GlnA3 protein sequence analysis by InterProScan¹ predicted two enzymatic domains: N-terminal GS, beta-Grasp domain (IPR008147) and the C-terminal catalytic domain of GS (IPR008146). Searching for functionally and structurally characterized GlnA3 homologs from the Protein Data Bank (PDB) revealed a gamma-glutamylmonoamine/polyamine synthetase PauA7 (PA5508) as a likely. PauA7 is involved in monoamine/polyamine gamma-glutamylation in P. aeruginosa (Ladner et al., 2012). The available crystal structure of Pau7 (Protein Data Bank entry: 4HPP) and GS GSI_St from S. typhimurium (Protein Data Bank entry: 1FPY) were used as the templates to generate a structural identical conserved binding residues for glutamate and ATP as well as two divalent metal ions (Mg⁺⁺/Mn⁺⁺). However, the binding pocket for ammonium and monoamine differ significantly. The GSI_St (as well as other eukaryotic and bacterial GS) and the PauA7 require two divalent metal ions per one subunit for activity. These ions have a structural as well as a catalytic role. Comparison of the conserved catalytic residues for the binding of divalent metal ions revealed four conserved glutamate residues E151, E153, E207, E214 and one conserved histidine residue H263 in the GlnA3 model structure (corresponding to E131, E133, E180, E187, and H236 in PauA7, as well to E129, E131, E212, E220, and H269 in GSI_St, respectively). Moreover, three conserved residues involved in coordination of the glutamate binding in GSI_St and PauA7, were found in the GlnA3 model structure. These GlnA3 residues (N260, G261, and R316) correspond to the N233, 258 G234, and R290 in PauA7 as well as to N264, G265, and R321 in GSI_St. Finally, two conserved residues (H265 and R339 both corresponding to H271 und R344 in GSI_St) important for coordination of the beta-phosphate and alpha-phosphate group of ADP were found in the GlnA3 model. This comparative in silico analysis demonstrates that GlnA3 possess conserved residues for binding of two metal ions, L-glutamate and ATP (Supplementary Figure S2). An essential element of GSI_St catalytic pocket is a loop, termed “the E327 flap” (GSI_St). The GSI active site is located between two subunits, and the E327-flap closes the catalytic site by interaction of the E335 with D50 from an adjacent subunit within the same ring, shielding the γ-glutamyl phosphate intermediate from hydrolysis. The second conserved residue D50 is involved in the deprotonation of ammonium for attack on the γ-glutamyl phosphate. Interestingly, these key acidic residues (E327 and D50) essential for the catalytic synthesis of glutamine in GSI_St are not conserved in PauA7 nor in GlnA3 (Supplementary Figure S2). In PauA7 structure and GlnA3 model these loops are much bigger and instead E327, a non-polar W327 residue occupies analogous position in the GlnA3 model structure (corresponding to W296 in PauA7). The conserved residue D50 that increases the affinity for ammonium binding in GSI_St is replaced by G40 and G63 in PauA7 and 273 GlnA3, respectively. Moreover, the conserved Y179 in GSI_St that coordinates the ammonium binding pocket is substituted by A147 in PauA7 and A169 in GlnA3, providing much more space for a bulky substrate such as polyamine (Supplementary Figure S2). The significant overall similarity of GlnA3 to the gamma-glutamylmonoamine/polyamine synthetase PauA7, the lack of the conserved residues for ammonium binding as well as our observations strongly suggest that GlnA3 may function as a gamma-glutamylpolyamine synthetase catalyzing the first step of polyamine degradation.

To determine whether GlnA3 was able to catalyze the predicted glutamylation reaction we developed an HPLC/ESI-MS based assay designed to detect a gamma-glutamyl product formed by GlnA3. For this, GlnA3 was overexpressed in E. coli BL21 (DE3) and purified by nickel ion affinity chromatography. The purified His-GlnA3 was used in an in vitro assay as described in Section “Materials and Methods.” The reaction conditions were as follows: 10 μg of His-GlnA3 were incubated at 30°C with glutamate, putrescine, MgCl₂ and ATP in HEPES buffer (pH 7.2). The reaction was stopped after 5 min. by incubation at 100°C for 5 min. HPLC analysis of generated product was performed in a positive MS mode. Result indicated that GlnA3 was able to use glutamate and putrescine as substrates in the ATP dependent reaction, generating a product with the mass to charge ratio of 218 m/z, corresponding to calculated mass of gamma-glutamylputrescine, confirming the function of GlnA3 as a gamma-glutamylpolyamine synthetase (Figure 11).