Enhancing biocatalytical N-N bond formation with the actinobacterial piperazate synthase

Natural compounds with nitrogen-nitrogen bonds are diverse and have applications in medicine and agriculture. L -Piperazic acid (Piz), an α -hydrazino acid, is one of few naturally occurring compounds of its kind. Yet, Piz and its derivatives are valuable building blocks for bioactive compounds. Few NNzymes, enzymes capable of forming N – N bonds, have been identified thus far. The hemoenzyme KtzT from Kutzneria sp. 744 catalyzes the intra-molecular N – N bond formation of N 5 ‑ hydroxy-L -ornithine (OH-Orn) to form Piz, a natural building block of kutznerides. The latter has antifungal and antibiotic properties. In our study, we established an improved expression method, with significantly improved yields ( ca . 35-fold) of heme-loaded enzyme, making the enzyme much more accessible for laboratory studies. In vitro biochemical characterization under conditions for N – N bond formation indicated a considerable thermo-and pH-flexibility, with optimal reaction conditions at 30 ◦ C and 10 mM Tris buffer at pH 9 together with low salinity, paving the way for more complex applications involving KtzT. We have also identified two homologous enzymes from extremophilic organisms to exhibit piperazate-forming activity. In silico structural studies, combined with phylogenetic analysis, resulted in a heme-and substrate-binding model, suggesting target enzyme residues that we propose are critical for the structural integrity and catalytic activity of KtzT. Following this approach, we investigated the potential role of a cysteine residue in a dimer-stabilizing disulfide bridge. The interplay of in vitro and in silico data therefore provides crucial functional information on this enzyme class.


Introduction
The group of natural compounds containing a nitrogen-nitrogen (N-N) bond still has relatively few members and is only occasionally reported on, despite their structural and functional diversity and their various applications in medicine, agriculture, and other fields.With about 200 known N -N compounds until 2013 [1] and 300 compounds as of 2022 [2], interest in these compounds and applied methods for their discovery notably rose within the last decade.This includes even synthetic biology approaches towards new bioactive compounds [3].However, many of these rare products are likely to remain undiscovered, and details of their biosynthetic mechanisms of their bioactive functions remain to be elucidated.Among the discovered compounds, over 30 families were found to contain an α-hydrazino acid, L-piperazic acid (systematic name: (S)-hexahydropyridazine-3-carboxylic acid; abbreviated as Piz).As examples, sanglifehrins, matlystatins and luzopeptins shall be named, with the first ones to be discovered being the family of monamycins in 1959 [4].Today, Piz still remains the only known naturally occurring α-hydrazino amino acid [5].The antibiotic functions of molecules containing N-N bonds make Piz and its known natural derivatives (5-chloro-Piz, 1,6-dehydro-Piz, 1-6-dehydro-5-hydroxy-Piz) desired building blocks for the development of novel bioactive compounds.Traditional formation of the hydrazinic N -N bond by chemical means is achieved only via inefficient and hazardous procedures [6,7] such as asymmetric α-hydrazination [8,9], and only a handful of enzymes forming N -N bonds (NNzymes) have been identified thus far and remain severely underexplored [1,2].
Among the Piz-containing compounds are also kutznerides, antifungal and antibiotic compounds produced by a non-ribosomal peptide synthetase complex in the actinobacteria Kutzneria [10].In 2017, the hemoenzyme KtzT (EC 4.8.1.1)from Kutzneria sp.744, until then only annotated as an FMN-dependent regulatory factor in databases, was proven to catalyze the formation of an N-N bond in N 5 -hydroxy-L-ornithine (OH-Orn) in a cyclocondensation reaction forming L-Piz [11].Respectively, we can designate it as a piperazate synthase (PizS).These studies have also provided evidence that KtzT forms a homodimer and coordinates a molecule of heme b, essential for its catalytic function, via a histidine residue (His65) that is highly conserved among KtzT homologues.Further, it was proposed that the ring-closing condensation reaction may occur via intramolecular nucleophilic attack of the α-nitrogen at the N 5 atom following polarization of the N 5 -OH bond by ferric or ferrous heme-iron acting as a Lewis acid catalyst, mimicking what is believed to drive hydrazine formation in bacterial annamox processes [12].Until now, however, no detailed biochemical characterization of KtzT has been reported.We therefore describe here an improved heterologous gene expression method towards enzyme production and define optimized in vitro conditions for the KtzT-catalyzed N-N bond formation starting from OH -Orn.Together with a heme-binding structural model that explores how the KtzT homodimer is stabilized and how the active site may coordinate OH -Orn, our data provide crucial information for designing biotechnological applications employing this NNzyme and related homologs.

Production of recombinant KtzT and homologous
Different methods were investigated to optimize the recombinant expression of ktzT in Escherichia coli, and we here describe the method producing the best results.The ktzT gene and homologous genes from Micromonospora tarapacensis (WP_230415462) and Streptacidophilus pinicola (WP_111501290) were codon-optimized (Table S3) for recombinant expression in E. coli NiCo21(DE3) and ordered from Twist Bioscience in the pET28a vector with a C-terminal His 6 -tag (accession numbers: OR145751, OR145752 and OR145753, respectively).Heatshock competent E. coli NiCo21(DE3) cells were transformed with the expression vector pET28a containing the respective genes and vector pKJE7 carrying chaperone genes dnaK, dnaJ and grpE [16].Expression cultures were grown in 1 L LB (lysogeny broth; 10 g/L tryptone, 10 g/L NaCl, 5 g/L yeast extract) medium supplemented with kanamycin sulfate (30 mg/L) and chloramphenicol (50 mg/L) for selection and L-arabinose (500 mg/L) for induction of pKJE7-based gene expression at 37 • C and 160 rpm in an unbaffled flask.Upon reaching an optical density of 0.7 at 600 nm, 0.1 mM IPTG, 1 mM FeSO 4 and 1 mM δ-aminolaevulinic acid were added and the temperature was lowered to 20 • C. Cells were harvested after 16-20 more hours of incubation, washed once in Tris buffer (50 mM Tris-HCl, 500 mM NaCl, pH 8) and then resuspended in Tris buffer supplemented with DNase I and lysozyme.Cell suspensions were homogenized via sonication (Sonoplus MS 72, 10 min, 30% amplitude) and cell debris was removed by centrifugation for 30 min at 4 • C and 12,420 × g.The supernatant was filtered through a 0.22 µm membrane and His 6 -tagged KtzT was purified over a Ni-NTA column using an imidazole gradient (50 mM Tris-HCl, 500 mM NaCl, 0-500 mM imidazole, pH 8).Combined elution fractions were dialyzed overnight at 8 • C against buffer (50 mM Tris-HCl, 10 vol% glycerol, pH 8) in a 12-14 kDa MWCO membrane and afterwards stored at − 80 • C. Protein concentrations were determined from a Bradford assay [17] and a bovine serum albumin (BSA) standard curve.As purified KtzT is usually not saturated with cofactor, the heme concentration in purified samples was determined using the pyridine hemochromogen assay according to previously reported protocols [18,19].

Site-directed mutagenesis
For the generation of mutant KtzT C197S , a pair of overlapping oligonucleotide primers (Table S4) was used to introduce the desired point mutation in the pET28a-KtzT plasmid via a 3-step PCR using Pri-meSTAR Max DNA Polymerase (Takara Bio).Amplification was performed for 30 cycles with 15 s at 95 • C, 30 s at 61 • C and then 30 s at 72 • C. The template DNA was digested by adding 10 U of DpnI and incubating the reaction overnight at 37 • C. Heat shock-competent E. coli DH5α cells were transformed with 1 µL of the reaction and grown overnight at 37 • C on an LB agar plate.Individual colonies were cultivated, and plasmids were isolated from the culture using a NucleoSpin Plasmid kit by Macherey-Nagel.Purified plasmids were sequenced by Microsynth Seqlab to verify the respective mutation.

Enzyme assays
General procedure.For assessment of KtzT activity under different conditions, 300 nM of heme-loaded KtzT (based on the pyridine hemochromogen assay results performed on purified KtzT) in 50 mM Tris-HCl pH 8 were added to reaction vessels (50 µL total volume per reaction) and incubated for 60 s at 30 • C. Afterwards, 1.25 mM of synthetic OH -Orn were added, and reactions incubated for 60 s at a target temperature.Reactions were quenched by addition of one volume equivalent of cold 90 vol% acetonitrile and 0.2 vol% formic acid.Samples were centrifuged for 5 min at 17,000 × g to remove precipitated protein.Reactions were performed in triplicates and analyzed using the liquid chromatography-tandem mass spectrometry (LC-MS/MS) method described in 2.5.
Temperature dependence.Prior to substrate addition, reaction mixtures were incubated for 60 s at the target temperature.Reactions were then treated as described at the target temperature.
Thermal stability.Prior to substrate addition, reaction mixtures were incubated for 30 min at the target temperature.Reactions were then treated as described at 30 • C. pH optimum.Instead of Tris buffer, Britton-Robinson buffer [20] at different pH values was used for the reactions.Reactions were then treated as described with 1 µL of 2 M CaCl 2 solution being added before centrifugation to remove phosphates from solution prior to LC-MS/MS analysis.
pH stability.Concentrated KtzT solution was incubated for 30 min in Britton-Robinson buffer of different pH values.Enzyme was then added to 75 mM potassium phosphate buffer at pH 8, and reactions were treated as described with 2 µL of 2 M CaCl 2 solution being added before centrifugation to remove phosphates from solution prior to LC-MS/MS analysis.
Influence of different buffers and salts.Reaction mixtures were supplemented with respective additives prior to enzyme addition, and then treated as described before.

LC-MS/MS method
Analytes were chromatographically resolved over an ammonium sulfonate hydrophilic interaction chromatography (HILIC) column (EC NUCLEOSHELL HILIC, 150 × 2 mm, 2.7 µm, 90 Å, Macherey-Nagel) and measured on a Shimadzu LCMS-8030 device in positive electrospray ionization mode with a triple quadrupole detector.Methods were run isocratically with 78 vol% acetonitrile, 22 vol% water and 20 mM ammonium formate buffered to pH 6 with a flow rate of 0.3 mL/min at 40 • C. Piz as authentic standard was purchased from BLDpharm.OH-Orn was synthesized according to the method described in the supplementary material.Chemical standards were used to develop S. Schröder et al. optimized multiple reaction monitoring (MRM) methods for each analyte, as described in detail in the supplementary material.Standards were measured in different concentrations to produce linear standard curves for quantification.5 µL of sample were applied per measurement.

Structure model development by means of an iterative refinement approach
The amino acid sequence of KtzT was entered into AlphaFold2 [21,22] using the ColabFold [23] interface.Energy minimization was performed on the best model using YASARA [24].A heme b molecule was inserted into one of the proposed active sites of the KtzT dimer (binding groove surrounding the His65 residue) using YASARA and AutoDock Vina [25].This resulted in 25 different conformations, varying in the angle of the porphyrin plane and positioning of the carboxylic acid residues.To identify the conformation that fits best to our model, a custom approach of evaluating structures based on characteristic distances was developed.First, the smallest distance between the His65 Nπ and Nτ atoms and the heme iron (His65-Fe) was determined, as this coordination is known to be essential for heme binding and catalytic activity.Secondly, the smallest distance between any of the carboxylic acid oxygen atoms of heme b and the heme iron (COOH-Fe) was calculated because the oxygen atoms tend to coordinate heme iron as a second axial ligand, thereby blocking the site for the virtual docking of substrates.Small His65-Fe distances and large COOH-Fe distances stable during energy minimization and simulation should thus indicate heme conformations as being suited for further modeling.To further develop an enzyme-cofactor complex structure that is suited for the employed simulation model [26,27], the narrowed-down selection of heme conformations was analyzed further using a bias for binding the native substrate OH -Orn.To each KtzT-heme conformation, OH -Orn was docked 20 times.Each substrate configuration was energy minimized and then simulated for 5000 fs three times with set random seeds.Enzyme-cofactor-substrate complexes were then scored based on: 1) a large COOH-Fe distance, 2) a small His65-Fe distance, 3) a small distance between the hydroxy oxygen of OH-Orn and the heme iron (O-Fe) to allow polarization of the N -O bond, and 4) a small distance between both nitrogen atoms of the substrate (N-N) to favor cyclization.For the latter two scoring distances, however, one must consider that the used YASARA2 force field [26] does not consider bond polarization.Average post-simulation scoring distances were calculated for all configurations, median values were determined, and the best heme conformation was selected based on its performance in the described criteria.The process was repeated for the second active site.Python scripts were used for automatization of the process.

Expression optimization
Initial attempts to overproduce KtzT in E. coli under standard conditions resulted in unsatisfactory yields.In addition, it was found that only a small fraction of the total KtzT yield was loaded with the heme cofactor required for the N -N bond forming activity of the enzyme.This means that under typical standard recombinant expression conditions (pET28a-KtzT, -S, 0 h) only about 50 µg of heme-loaded KtzT per liter of culture was isolated.Thus, the production conditions were optimized in an iterative manner by stepwise changes in the expression conditions: with and without addition of iron and the heme maturation agent δ-aminolevulinic acid (+/-S), with different induction delay times and by using different co-expression strategies described in literature, either by co-expressing a ferrochelatase gene [28] or three chaperone genes [16] (Fig. 1).
Generally, and for all strategies, the yield of active enzyme increased with prolonging induction delay times, i.e. incubation times between reaching an OD 600 of 0.7 and induction of ktzT expression.Furthermore, the obtained amount of heme-loaded KtzT was generally higher in cultures supplemented with FeSO 4 and δ-aminolevulinic acid compared to the respective non-supplemented cultures.As co-expression strategies can aid in increasing yields of active proteins, different approaches were Fig. 1.Yield of heme-loaded and total KtzT obtained from different expression strategies.pET28a-KtzT: cells transformed only with the ktzT gene-carrying vector; expression was induced by adding 0.1 mM IPTG.+pASK-HemH: also transformed with the pASK-IBA37plus-HemH vector carrying the hemH ferrochelatase gene; expression was induced by adding 200 µg/L anhydrotetracycline-+pKJE7: also transformed with the pKJE7 vector carrying the chaperone genes dnaK, dnaJ and grpE; expression was induced by supplementing the medium with 500 mg/L L-arabinose.pETDuet: cells transformed only with the pETDuet-KtzT-HemH vector carrying both the ktzT and hemH genes; expression was induced by adding 0.1 mM IPTG.+S / -S: culture was / was not supplemented with 1 mM FeSO 4 and 1 mM δ-aminolevulinic acid prior to induction.0/1/2/3 h: time between the culture reaching an OD 600 of 0. taken.Co-expression of the ferrochelatase gene hemH increased the total protein yield, but not to the extent observed in cases of co-expressing other chaperone genes (from pKJE7).
The most efficient protein production strategy was found to be based on gene expression with supplements and chaperone gene co-expression (+pKJE7, +S, 3 h).Under these conditions, yield of active enzyme was increased around 35-fold to a total of (1.74 ± 0.24) mg/L culture of hemeloaded KtzT (24% saturation).Albeit the overall amount of total enzyme was surpassed by the non-supplemented hemH-co-expression strategies (+ pASK-HemH, -S), very low heme saturation levels were achieved in these cases (between 2.7 and 4.7 %).The highest relative heme loadings were observed in strategies involving addition of supplements without co-expression (pET28a-KtzT, +S) and from the pETDuet cultures, irrespective of supplementation.Here, saturations from 33 up to 44 % were achieved, yet at the cost of very low overall protein yields.Lowering the pH of the medium to 6.0 to increase iron uptake resulted in significantly reduced biomass and protein yields (data not shown).Interestingly, coexpression of all three, ktzT, hemH and the chaperone genes from pKJE7, resulted in lower yields than co-expression of ktzT and pKJE7 chaperone genes without hemH (data not shown), potentially indicating levels of stress on the cells under these conditions too high for normal growth.

Assay development
Enzymatic activity can only be determined if a suitable assay is available.The natural reaction of KtzT, the cyclocondensation of OH -Orn to Piz (Fig. 2), cannot easily be monitored by high-throughput methods.With the aim to circumvent tedious and potentially errorprone derivatization methods that would enable e.g.reversed-phase-HPLC-based separation with spectrophotometric detection [29], we established a direct measurement method based on liquid chromatography coupled to mass spectrometry (LC-MS/MS).Using hydrophilic interaction liquid chromatography (HILIC) coupled with MS/MS detection in multi reaction monitoring (MRM) mode, it was possible to develop a direct and reliable quantification method of substrate consumption and product formation (supplementary information).By this, we were able to directly monitor enzymatic activity and determine product amounts in our biophysical characterization studies.

Temperature dependence and thermal stability
Having a suitable protein production strategy and activity assay at hand, in-depth experiments on the proteins' biophysical properties were made possible.First, the temperature optimum and thermal stability of KtzT were investigated to uncover catalytic performance during the assays.Fig. 3 shows the observed activity rates at different temperatures and after 30 min of pre-incubation referring to protein stability under temperatures applied.The optimal reaction temperature was found to be 30 • C with a decrease to 86 % and 71 % of the optimal reaction rate when changing the temperature to 25 and 35 • C, respectively.While the observed residual activity drops below 20 % at 45 • C, notably, the enzyme still shows around 30% residual activity between 5 and 10 • C. KtzT seems to be thermally stable during 30 min pre-incubation up to 30 • C, before rapidly losing more than 25% activity at 35 • C relative to the reaction pre-incubated at 10 • C.After pre-incubation at 50 • C, KtzT is almost completely inactivated with only 7 % of observed residual activity.
Though Kutzneria sp.744 has not been characterized extensively yet, these findings agree with observations made on other strains of Kutzneria, for which an optimal growth temperature between 25 and 30 • C is reported, though also described as thermotolerant [30,31].A PKS gene cluster in K. albida encodes ω-3-polyunsaturated fatty acid synthases, producing lipids which are known to mediate membrane stability at lower temperatures [32].Kutzneria sp.744 was first isolated from mycorrhizal root tips in a Lithuanian forestry nursery [10], and thus, a moderately cold region.This isolation was part of a larger study investigating microorganisms acting as antagonists against root-rot fungi in north-temperate and boreal forests [10], and would explain why kutzneride and piperazate biosynthesis are designed to function even in colder climates.From a biotechnological view, this relatively high activity at such low temperatures could prove beneficial, as lower temperatures can increase KtzT stability and possibly reduce the rate of basal hydroxylamine degradation in solution, while also decreasing the necessity of external heating, improving the energy efficiency of this process.

Determining optimal buffer conditions
The buffer composition was optimized regarding pH, type and concentration of buffer and effect of added salts.To this end, the pHdependent activity of KtzT was determined in Britton-Robinson buffer at different pH values and after pre-incubation in different pH ranges (Fig. 4).
The highest activities were observed at slightly basic pH values between 8.5 and 9.0.The observed activity always remains above 60 % between pH 7.5 and 10.0, before falling under 10 % below pH 6.0 and under 2 % above pH 11.5.The stability of KtzT is not significantly impacted in the pH range of 5.5 -9.5 but decreases rapidly at higher pH values.
Some strains of Kutzneria are known to grow in pH ranges between 5 and 9, with a growth optimum around 7.0.There are reports of root-rot fungi alkalinizing their host medium [33], which might induce production of specific secondary metabolites.It is therefore not inconceivable that kutzneride biosynthesis in Kutzneria sp.744 (isolated from mycorrhizal root tips) may have evolved towards increased activity at slightly alkaline values, as KtzT seems to operate best at pH values  Next, the activity of KtzT was determined in different buffering systems with ideal buffer capacities close to the optimal pH, i.e., Tris buffer, potassium phosphate buffer (PPB), PIPES buffer, carbonatebicarbonate buffer (CBCB) and HEPES buffer, to determine if the chemical composition of the buffer influences the enzymes' activity (Fig. 5).
The highest activity of KtzT was observed in Tris buffer.The reduced activity observed with PIPES and HEPES buffer systems may be related to their piperazine moieties potentially binding the active site.Reduced activity in phosphate and carbonate buffers may be related to the low halotolerance of KtzT, as determined later, and in the latter case also to CO 2 coordinating to the heme iron and inhibiting catalysis.
As the best results were obtained with Tris buffer, different concentrations of this system were investigated as well (Fig. 5).Maximum KtzT activity was observed with only 10 mM of Tris buffer, decreasing at both higher and lower concentrations of buffering agent.This might indicate that the systems' ionic strength affects the catalytic behavior of KtzT, possibly also depending on the nature of the ions present.For example, increased ionic strength could potentially lead to increased dissociation of the KtzT homodimer, thus leading to a decrease in observed activity [34].
To verify this assumption, the influence of different salts was further investigated.Fig. 6 shows the observed rates of KtzT in Tris buffer supplemented with 100 mM of different selected salts.
All tested salts reduce the observed activity of KtzT under the chosen conditions, with sodium chloride impacting the reaction the least and magnesium chloride the most.To investigate the salt tolerance of KtzT further, different NaCl concentrations were investigated as shown in Fig. 6.
The observed activity of KtzT steadily decreases with increasing amounts of NaCl present during the reaction, though still maintaining 64 % of the maximum activity when supplemented with 300 mM of sodium chloride.Hence, increased ionic strength lowers the observed activity which might be caused by potential disruption of the dimerization interface and preventing the correct association of the active site.

Piz productivity of KtzT and structural homologs
As KtzT was initially annotated as an FMN-binding regulatory protein and is so far the only confirmed piperazate synthase (PizS), we became interested in homologous sequences with potential PizS-like activity.Among the identified sequence homologs (Figure S1) which are also annotated as FMN-binding regulatory proteins, two enzymes from the extremophilic organisms Micromonospora tarapacensis (MtPizS) und Streptacidiphilus pinicola (SpPizS) were chosen to confirm PizS activity.MtPizS and SpPizS share on amino acid level a sequence similarity with KtzT of 52 % and 59 %, respectively.Both putative PizS's were heterologously produced and purified following the optimized KtzT protocol with final yields of 0.47 ± 0.07 and 0.36 ± 0.03 mg/L culture of heme-loaded protein.The newly obtained putative PizS's were compared to KtzT in terms of Piz formation activity (Fig. 7).
KtzT showed the highest initially observed rate constant of (3.72 ± 0.16) s − 1 compared to SpPizS with (2.87 ± 0.35) s − 1 and MtPizS with (2.42 ± 0.15) s − 1 .SpPizS reached a comparable amount of produced Piz after 5 min under reaction conditions optimized for KtzT.As both homologues originate from extremophiles, their ideal reaction conditions may differ from that of PizS from Kutzneria and have yet to be elucidated.Nevertheless, it was shown that the optimized protocol for recombinant production of KtzT can directly be used to access two hitherto uncharacterized sequence homologs from different origins, SpPizS and MtPizS.Both KtzT homologs convert OH -Orn into Piz, confirming their activity as PizS's.

Structure model and phylogenetic analysis
A structural model of the heme-free KtzT homodimer was generated using AlphaFold2 [21][22][23][35][36][37][38][39].The predicted structure as well as secondary structure elements and regions predicted to form two active sites are visualized in Fig. 8 and Fig. 9. Functional KtzT forms a homodimer including one heme per monomer which relates it to the structurally most similar protein PaiB (PDB ID: 2OL5) [11].The structural alignment reveals a RMSD of 2.167 between KtzT and PaiB.From this, His65 was identified as potential heme binding or coordinating amino acid of KtzT.According to the model, two similar heme binding sites are located at the monomer interfaces, so that each heme and substrate binding pocket is composed of amino acids from both monomers.This peculiar setup was worthy of more in-depth investigation by an in silico modeling approach.Two molecules of heme b were docked and refined with a bias towards optimal binding of OH -Orn as its currently accepted native substrate.The found optimal heme conformation is shown in Fig. 9.
In the generated model, heme b binds to each of the H65 residues of the homodimer with one carboxylate group close to residue N67, facing the solvent, and one inside a small pocket lined by residues V131 and L159.The active sites are formed by regions β1, β2, β3, α4 and α5 of the heme-binding H65 monomer (Fig. 8, purple) and regions NT', β4', α3', β5' and β6' of the opposing, substrate-binding monomer (Fig. 8, orange).
To elucidate if this binding site is conserved among homologues, we determined the level of conservation at each KtzT position among 300 homologous sequences.As shown in Fig. 10, these regions display indeed a high level of conservation, making their necessity for functional PizS's more likely.The N-terminal region stands out as being less conserved.This is because about 25 % of the aligned homologues contain a truncated N-terminus missing the first 14 residues found in the KtzT sequence.The involvement of the N-terminus on piperazate forming activity remains to be elucidated, and it is not yet known whether PizS's with this truncation are active at all.
Based on the enzyme-cofactor-substrate model, we predict several interactions to be essential for binding and activation of OH-Orn besides the coordination of the N -OH group to the heme iron, as shown in Fig. 11.Conserved lysine and threonine residues (K177, T106) may coordinate the carboxylate group of OH -Orn, and a highly conserved glutamate residue (E181) could be involved in coordinating the α-amine group via hydrogen bonds.It is currently hypothesized that the α-amine will attack the N -OH group as intramolecular nucleophile for Piz formation [11].For this, the α-amine, likely protonated under physiological conditions, must first be deprotonated to free an electron pair for nucleophilic attack.This may likely be facilitated both by an increased pH value (vide supra, 3.4) as well as by E181 and/or a neighboring, conserved tyrosine residue (Y7).We thus predict at least one of these residues to be essential for efficient catalysis.The two active sites in this model are located on the dimer interface (Fig. 9), and for each site, one monomer provides the heme-binding H65 residue, and the other monomer provides the residues we propose to be involved in catalysis.The other active site is analogous, but the roles of the subunits are swapped there.While monomeric KtzT may thus be able to weakly bind the substrate, we propose that Piz-formation can exclusively occur in the dimer.
Regarding interactions stabilizing the dimer (Fig. 12), besides nonpolar van der Waals interactions, we found a potential salt bridge between residues E198 and R201 of opposing monomers located in the region between α6 and α7 helices.This specific interaction could, however, be exclusive to KtzT, as phylogenetic analysis among homologues shows high conservation only of R201 but not of E198, a position where most homologues have an arginine or a tyrosine instead.Dimerstabilizing hydrogen bonds are sparsely found throughout the model, with E139 in α4 contacting Y92 and S94 in β4', H41 in β2 contacting    S111 in β5', I39 in β2 contacting Q113 in β5'.Of these residues, H41, Y92, S94 and E139 are highly conserved among KtzT homologues.Interestingly, most dimer-crossing hydrogen bonds are located in the loop segments between β1 and β2 of both monomers, forming a contact region involving residues N30, G31, G34, A35, A36, P37 of the loop region as well as N81 and R115 from β4 and β5, respectively.Of these residues, N30, G31, P37 are highly conserved among KtzT homologues.
In addition, we found the only cysteine residues of both subunits (C197) in the contacting α6-α7 loops of both monomers, suggesting a disulfide bond may also be involved in the dimerization of KtzT.

Dimerization of wildtype KtzT and C197S
In order to confirm the hypothesis that C197 indeed forms a disulfide bond for dimer stabilization, we generated the point mutant KtzT C197S (see 2.3 and 3.6) that was compared with the wildtype via SDS-PAGE under both reducing and non-reducing conditions (Fig. 13).
Under non-reducing conditions, a second KtzT WT band between 40 and 55 kDa appears, likely corresponding to covalently linked, dimeric KtzT.This band does not appear after cysteine substitution, proving that a disulfide bond is at least partially involved in KtzT dimerization.Under those non-reducing conditions, the additional signal is composed of a stronger and a weaker band.Whilst this is in agreement with the earlier report on KtzT [11], a potential reason, such as a discrepancy in heme loading of the dimer, is speculation at current stage.
Interestingly though, the C197S variant exhibits similar, if not increased levels of Piz-forming activity (k obs in phosphate buffer of (2.82 ± 0.56) s − 1 and (3.85 ± 0.94) s − 1 for WT and C197S, respectively).Thus, while KtzT may be stabilized by a disulfide bond, it is not required for a catalytically active enzyme.In the C197S mutant, the previously discussed salt bridge E198-R201 may be able to preserve the homodimer by itself, especially considering the proximity of these residues to the proposed disulfide bridge.

Conclusion
Our here presented studies provide valuable insights into the biochemical properties and behavior of KtzT, the first identified PizS of the highly potent enzyme class of NNzymes.To study the catalytic properties of KtzT in vitro, we first developed a suitable activity assay based on the conversion of OH -Orn to Piz using a LC-MS/MS based method.This direct measurement now allows the reliable and quantitative determination of Piz formation and ultimately the determination of the kinetic properties of KtzT.This activity assay was used to study the optimal reaction conditions for KtzT-catalyzed Piz formation.In vitro, KtzT appears to be thermo-and pH-flexible, with a temperature optimum at 30 • C and stability up to that temperature, a pH optimum at pH 9 and stability between pH 5.5 and 9.We have also shown that among commonly used buffers, the best performance of KtzT is achieved in low concentrations of buffer and other additives, with highest activities obtained in 10 mM of Tris-HCl buffer.Moreover, we have demonstrated that increased salinity of the reaction medium decreases Piz-forming activity.These insights are highly valuable for planning other reactions involving KtzT, enabling confident assessments on how well, for example, cascade reactions involving KtzT are expected to work.The still significant activity at low temperatures makes the energy efficiency of potential future biotechnological applications involving KtzT very   promising.
By improving conditions for heterologous expression, involving medium supplementation and co-expression of chaperone genes, we significantly increased the yield of heme-loaded enzyme by a factor of approximately 35 compared to the original conditions used in previous reports [11].It has yet to be elucidated whether protein folding and heme incorporation in the cell can still be optimized to give even higher yields.Possibilities to further optimize gene expression to yield soluble protein and higher heme incorporation would include using a codon-harmonized ktzT gene, a fusion construct with increased solubility, a heme-incorporating E. coli strain such as Nissle 1917 [43] or different lysis conditions.Under the herein identified optimized conditions, homologous enzymes MtPizS and SpPizS were expressed and purified in sufficient amount and shown to produce Piz, however, to a lower extent compared to KtzT.As such, two additional representatives of the PizS family could be successfully identified, recombinantly expressed and Piz-forming activity confirmed.We thus made KtzT and homologs more accessible for laboratory studies and potential biotechnological applications.
Our structural models of the enzyme-cofactor and enzyme-cofactorsubstrate complexes provided valuable insights into potential structureactivity relationships.The active site is shown to be located directly at the dimer interface, making it highly unlikely for KtzT to be catalytically active as a monomer.By identifying target positions for investigating KtzT dimerization, substrate coordination and the catalytic mechanism, we opened possibilities to understand and improve the activity of this enzyme.The fact that the C197S mutant appears slightly more active than the wild-type, and that this amino acid is not a conserved residue among homologs, could suggest a more flexible dimer to be favorable for increased catalytic activity.Closer inspection of the dimer interface also revealed salt bridges and hydrogen bonds that could be more relevant for forming the functional dimer, as increased salt concentrations that may disrupt these interactions were shown to cause a decrease in Pizforming activity.In addition, we could uncover the conserved residues Y7 and E181 that might serve as important catalytic residues to enable the nucleophilic attack on the terminal N-OH.
Recently, Kong and co-workers demonstrated that Piz can be produced effectively by means of an engineered fungal strain which allowed to reach 1.1 g L − 1 from 120 h fed-batch process [44].This is hardly comparable to the herein presented cell-free in vitro approach at small scale.However, if one considers full conversion by means of the in vitro approach, a theoretical yield of about 10 % compared to the whole-cell approach might be achievable.This still leaves room for improvement, for example by increasing substrate loading and space-time-yield.Most promising might be to engineer a respective KtzT enzyme to higher productivity and employ it in the format of the whole-cell system to simplify production.
In conclusion, we have demonstrated the biochemical, catalytic and structural properties of the first PizS, KtzT, and identified additional homologs for which Piz-forming activity could be confirmed.These results are expected to pave the way not only for further elucidation of the reaction mechanism of this peculiar class of enzymes, but also for the development of KtzT and related enzymes as biocatalysts for intramolecular N-N bond formation, thus providing access to a variety of bioactive compounds containing N -N bonds.
Fig.1.Yield of heme-loaded and total KtzT obtained from different expression strategies.pET28a-KtzT: cells transformed only with the ktzT gene-carrying vector; expression was induced by adding 0.1 mM IPTG.+pASK-HemH: also transformed with the pASK-IBA37plus-HemH vector carrying the hemH ferrochelatase gene; expression was induced by adding 200 µg/L anhydrotetracycline-+pKJE7: also transformed with the pKJE7 vector carrying the chaperone genes dnaK, dnaJ and grpE; expression was induced by supplementing the medium with 500 mg/L L-arabinose.pETDuet: cells transformed only with the pETDuet-KtzT-HemH vector carrying both the ktzT and hemH genes; expression was induced by adding 0.1 mM IPTG.+S / -S: culture was / was not supplemented with 1 mM FeSO 4 and 1 mM δ-aminolevulinic acid prior to induction.0/1/2/3 h: time between the culture reaching an OD 600 of 0.7 and induction.All cultures were grown in 1 L of LB medium in a Fernbach flask under shaking at 37 • C and cooled down to 20 • C after reaching an OD 600 of 0.7.(A): Overview of all performed experiments.(B): Selection of experiments showing the increase in heme-loaded KtzT yield with different strategies.

Fig. 3 .
Fig. 3. Temperature-dependent reaction rates (A) and thermal stability (B) of KtzT.(A): Reaction mixtures of 2 mM OH -Orn and 300 nM heme-loaded KtzT were incubated at a target temperature for 60 s after substrate addition.(B): For thermal stability experiments, the buffered enzyme was pre-treated for 30 min at a target temperature before performing the reaction at 30 • C. The reaction rate after the 10 • C pre-incubation is set to 100 % (corresponds to k obs = 5.35 s − 1 ).Product quantification was performed via LC-MS/MS and an optimized MRM method.Error bars indicate the standard deviation of at least three biological replicates.

Fig. 4 .
Fig. 4. pH-Dependent reaction rates (A) and pH stability (B) of KtzT in Britton-Robinson buffer.Reactions were incubated for 60 s after substrate addition (1.28 mM OH -Orn).For pH stability experiments, 200 nM of the buffered, heme-loaded enzyme were pre-treated for 30 min on ice at a target pH before performing the reaction at pH 8.The reaction rate after pre-incubation in pH 8.5 is set to 100 % (corresponds to k = 4.39 s − 1 ).Product quantification was performed via LC-MS/MS and an optimized MRM method.Error bars indicate the standard deviation of at least three biological replicates.

Fig. 5 .
Fig. 5. Influence of different buffering systems and buffer concentrations on the KtzT reaction.(A) Influence of different buffers (50 mM each).Reactions were incubated at 30 • C for 60 s after substrate addition.(B) Influence of different concentrations of Tris-HCl.Reactions were incubated at 30 • C for 60 s after substrate addition.All reactions were performed with 200 nM of hemeloaded KtzT and 1.28 mM of OH -Orn.Product quantification was performed via LC-MS/MS and an optimized MRM method.Error bars indicate the standard deviation of at least three biological replicates.

Fig. 6 .
Fig. 6.Influence of different salts and salt concentrations on the KtzT reaction.(A) Influence of different salts (each 100 mM).(B) Influence of different concentrations of NaCl.Reactions were incubated at 30 • C for 60 s after substrate addition.Product quantification was performed via LC-MS/MS and an optimized MRM method.Error bars indicate the standard deviation.

Fig. 7 .
Fig. 7. Comparison of Piz-forming activity of KtzT and selected homologues.Reactions were performed with 150 nM active enzyme at 30 • C. Product quantification was performed via LC-MS/MS and an optimized MRM method.Error bars indicate the standard deviation.

Fig. 8 .
Fig. 8. AlphaFold2 predicted structural model of KtzT homodimer with secondary structures and presumed active sites.(A): Structural model with predicted secondary structures, termini and His65 residue.The structure is shown twice to focus on both subunits separately.(B): Schematic sequence of secondary structure motifs.Rectangles below indicate residues making up the two similar active sites, with purple residues from one monomer and orange residues from the other, see also Fig. 9. (C): Amino acid sequence of KtzT with the described rectangle coloring scheme and important residues underlined and in bold.

Fig. 9 .
Fig. 9. Optimized configuration of heme b in the two active sites of homodimeric KtzT.

Fig. 10 .
Fig. 10.Conservation of residues among KtzT homologues.(A): Bar graph showing percentages of the most common residue at each aligned position among 300 homologous sequences.Residues of the active site are colored orange and purple as defined in Fig. 8C.(B/C): Cartoon/surface structure model of KtzT (B: monomer, C: dimer) colored by the conservation of each residue (white: relative frequency < 50 %; red: frequency = 100 %).An arrow indicates the active site.(D): Dimer interface of the KtzT homodimer colored by the same scheme.

Fig. 12 .
Fig. 12. Potential dimer-stabilizing interactions in the KtzT homodimer model.Cartoon representation of the homodimer.Atoms potentially involved in dimerstabilizing interactions are shown as spheres, with sulfur, oxygen, nitrogen, and hydrogen colored in yellow, red, blue, and gray, respectively.

Fig. 13 .
Fig. 13.SDS-PAGE result of KtzTWT and KtzT C197S under reducing and nonreducing conditions. 1 µg of protein in sample buffer with or without β-mercaptoethanol and dithiothreitol was loaded per lane.Electrophoresis was performed for 2 h at 100 V and room temperature.Gels were stained overnight in a colloidal Coomassie staining solution and afterwards washed with water.Pag-eRuler™ Prestained Protein Ladder, 10-180 kDa by Thermo Scientific, was used as a size standard marker (M).