Escherichia coli EPI300-T1 (Epicentre, Madison, WI, USA) was used for the screening of the genomic libraries cloned in pCC1FOS fosmids library (Epicentre). E. coli strain Rosetta 2 (DE3) was used as the host for pET46 Ek/LIC expression vector. Nicotinamidase substrates, inhibitors, and stable iron salts were obtained from Sigma-Aldrich.

Fosmid library contained inserts of mixed genomic DNA of 194 mesophilic bacteria (MB) isolated from deep-sea hydrothermal vents sampled at various sites in Mid Atlantic Ridge. The DNA extracted from individual isolates was pooled in equal ratios and used to prepare a polygenomic library using Epicentre Fosmid Library Construction Kit (Epicentre), according to the manufacturer instructions and as described earlier, to produce a library named MB (Leis et al., 2015). The fosmids clones in E. coli EPI300-T1 were arrayed in 384-well microtiter plates and stored at -80°C in LB with chloramphenicol (12.5 μg/mL).

To detect pyrazinamidase activity in the fosmid constructions of the polygenomic libraries, we used a modification of the Wayne’s pyrazinamidase assay (Wayne, 1974). This method is based on the complex formed between AFS and POA formed in the pyrazinamidase reaction, giving rise to an intense red color. Briefly, Escherichia coli EPI300-T1 transformed with the fosmid library was grown on LB agar plates at 37°C overnight to yield single colonies, which were stored in 384-well plates with LB and chloramphenicol (12.5 μg/mL) at -80°C. Replica plating was used to transfer the library colonies to 96-well plates containing 200 μL of fresh LB with the same antibiotic and Fosmid Autoinduction Solution (Epicentre, Madison, WI, USA) for activation of the oriV origin and hence for increasing of fosmids copy number per cell. After growing at 37°C overnight, their pellets were then re-suspended in a mixture of 20 mM PZA and 1% AFS dissolved in MilliQ^® water, and incubated for 1 h at 37°C until an intense red color appeared. If desired, the revealed plates could be incubated overnight to obtain an even more intense red color with no appreciable change in control wells. Positive and negative controls were made using previously cloned Oceanobacillus iheyensis nicotinamidase (OiNic) (Sanchez-Carron et al., 2013) and empty EPI300-T1 cells, respectively.

To detect pyrazinamidase or nicotinamidase activity using SNP, culture conditions were the same as in the previous method (AFS method), but pellets were re-suspended in 1 mM PZA or NAM with 0.75% SNP. Once re-suspended, plates were incubated at 37°C for 4 h, centrifuged, and the supernatants were transferred to flat 96-well plates. The difference in absorbance between the negative control wells and those containing the fosmids clones was measured at 495 or 395 nm, to detect pyrazinamidase or nicotinamidase activity, respectively.

Positive clones were selected and their DNA inserts were sequenced using a Roche 454 GS FLX Ti sequencer (454 Life Sciences, Branford, CT, USA) at Life Sequencing SL (Valencia, Spain). Upon completion of sequencing, the reads were assembled to generate non-redundant meta-sequences using Newbler GS De Novo Assembler v.2.3 (Roche). The GeneMark software (Lukashin and Borodovsky, 1998) was employed to predict potential protein-coding regions (open reading frames (ORFs) with ≥20 amino acids) from the sequences of each assembled contig, and deduced proteins were screened via blastp and psi-blast (Altschul et al., 1997). ESPript (Robert and Gouet, 2014) was used for displaying alignments. PolyNic homology modeling was carried out with Swiss-Model¹ (Biasini et al., 2014) using Pyrococcus horikoshii nicotinamidase coordinates as template (PhNic, PDB 1IM5), and structurally aligned with Chimera (Pettersen et al., 2004).

The selected positive gene of the fosmid library (contig 00048_2428, NCBI’s BioSample accession: SAMN05386332 and NCBI’s Sequence Read Archive accession: SRP079696), named as PolyNic, was amplified by PCR with a high fidelity thermostable DNA polymerase (Pfu UltraII Fusion HS, Agilent). The respective forward and reverse oligonucleotides were 5′-ATGGAAATAAAGGAGAAATTCACCGTGTCGCCAACTCTCCATTATCAACCTAATGACGCG-3′ and 5′-CTACTGGTCGTGGACATGCAACGCGACTTCTGCGAAGGCGGAGCTCTGGCGATTGATGGC-3′. The amplified PCR product (588 bp) was inserted into pET-46 Ek/LIC vector (EMD Millipore). The vector was later transformed into E. coli Rosetta 2 (DE3). Cells were grown in 1 L of Terrific Broth (TB) supplemented with ampicillin (50 μg/mL) and chloramphenicol (12.5 μg/mL) and, when the OD₆₀₀ reached a value of 4, induced with 0.5 mM isopropyl-β-thiogalactoside (IPTG) for 16 h at 25°C with constant agitation. Cells were harvested by centrifugation, re-suspended in lysis buffer (50 mM sodium phosphate buffer pH 7.3 containing 300 mM NaCl) and disrupted using a Bead Beater (BioSpec, USA). The protein in the supernatant was then purified by Ni²⁺-chelating affinity chromatography (ÄKTA Prime Plus, GE Life Sciences) onto a HiPrep IMAC 16/10 FF 20 mL column (GE Life Sciences). Fractions containing the desired protein were desalted, concentrated and filtrated onto a Superdex 200 HiLoad 16/600 column (GE Life Sciences), obtaining an electrophoretically pure enzyme. The protein molecular mass was determined by SDS-PAGE, gel filtration and using an HPLC/ESI/ion trap system (Sanchez-Carron et al., 2011). Protein concentration was determined using Bradford reagent (Bio-Rad) and BSA as standard.

Nicotinamidase activity was determined chromatographically by HPLC following the increase in the NA area, using an HPLC (Agilent 1100 series) with a reverse-phase C₁₈ 250 mm × 4.6 mm column (Gemini C18, Phenomenex) and a mobile phase (20 mM ammonium acetate pH 6.9) running at 1 mL/min. Under these conditions, the retention times (R_T) for NA and NAM were 7 and 19.9 min, respectively. The standard reaction medium consisted of 1 mM NAM and 0.14 μM of purified PolyNic in 100 mM phosphate buffer at pH 7.3. Reactions were stopped by the addition of trifluoroacetic acid to 1% (v/v). One unit of activity was defined as the amount of enzyme required to cleave 1 μmol of NAM releasing 1 μmol of NA in 1 min. The activity toward other NAM analogs was also determined by HPLC. Finally, the inhibition constants toward nicotinaldehyde and 5-bromo-nicotinaldehyde were obtained as previously described (Sanchez-Carron et al., 2013).

In order to establish a simple whole-cell method for nicotinamidases, the first step was to test the two previously described methods for detecting pyrazinamidase activity with POA and AFS. In the first method, used for detecting PZA activity in Enterobacteriaceae, a wireloop of an overnight bacterial culture on Tryptic Soy Broth is suspended in 0.2 mL of a 0.5% PZA solution, incubated for 18 h at 37°C, and finished by the addition of two drops of a 1% freshly prepared AFS solution to the culture (Pignato and Giammanco, 1992). In the second method, E. coli cells are cultured overnight in LB containing 8 mM PZA, centrifuged for 3 min at 13800 g, washed twice with 150 mM NaCl in 100 mM Tris-HCl pH 7.5, re-suspended in 8 mM with PZA, incubated at 37°C for 120 min, centrifuged again, revealed with 0.05 volumes of 500 mM AFS and, finally, the change in OD₄₈₀ compared with a standard curve (Frothingham et al., 1996). However, when both methods were tested with a previously cloned O. iheyensis nicotinamidase (OiNic) in E. coli Rosetta 2 (Sanchez-Carron et al., 2013), several drawbacks came out. In the first method, the interferences produced with the culture medium gave faint and not very reliable colors, whereas in the second, the need of several steps, centrifugations and, especially the precipitation of POA-ferrous ion complex at pH > 6 hinders the correct spectrophotometric quantification. Thus, a more simple method suitable for HTS was developed by just resuspending the LB overnight OiNic E. coli pellets in a solution with 20 mM PZA and 1% AFS dissolved in MilliQ^® water at 37°C. The red color corresponding to POA/AFS complex was developed faster than in the above-described methods, just in 1 h (Supplementary Figure S1A), being this color even more intense after an overnight incubation, without any discernible color in control wells (Supplementary Figure S1B). Thus, to minimize the time of the assay for HTS, a clone that gave red color after 1 h was considered as a pyrazinamidase positive clone.

To test the sensitivity of the assay, the method was tested with different OiNic mutants (Sanchez-Carron et al., 2013). Clear differences between wild-type (OiNic Wt), K104A (total loss of activity), and E65H (less activity than the wild-type) mutants were observed, giving rise to a gradient of colors, depending on how such mutation affects the activity (Figure 1C). Nevertheless, when NAM is used for this screening instead of PZA, it is difficult to discriminate between more or less active clones, since a faint orange color is formed inside cells and not in the reaction medium (data not shown), making impossible to discriminate between them and control cells, even after centrifugation and subsequent detection in a microplate spectrophotometer.

Thus, only nicotinamidases with activity toward PZA could be screened with the AFS method. To circumvent this limitation, another activity assay was developed to be also used with NAM, using a different stable iron salt, SNP. This salt at neutral-basic pHs not only reacted with POA (Figure 2, curve A; 𝜀_475nm = 4374 M^-1 cm^-1) but also with PZA (Figure 2, curve B; 𝜀_495nm = 4549 M^-1 cm^-1). Surprisingly, it also gave appreciable color with both NAM (Figure 2, curve C; 𝜀_395nm = 3125 M^-1 cm^-1) and its corresponding acid (NA) (Figure 2, curve D; 𝜀_385nm = 2688 M^-1 cm^-1). To explore these different reactions (colors) of nicotinamidase substrates/products with SNP, which have never been used for whole-cell screening of this enzyme, different optimization conditions were tested with NAM and PZA as substrates. Since the change in color in the PZA/SNP method from red to orange was more evident, we started to optimize conditions with it. In order to see the decrease in the color of the wells with the naked eye, lower concentrations of PZA were needed (Figure 3A), when compared with the PZA/AFS method. At 0.3 mM PZA, the difference in color between OiNic Wt (Figure 3A, squares) and control clones (Figure 3A, circles) was more evident at 0.75% SNP. Once the SNP concentration was adjusted, different concentrations of the substrate were tested at 0.75% SNP in order to get the highest resolution. The difference in absorbance at 495 nm increased with PZA concentration up to 1 mM (about 0.2 absorbance units), tending toward a plateau (Figure 3B). On the other hand, the differences in the case of NAM were less evident by the naked eye due to the change in color from yellow to pale yellow, but still spectrophotometrically measurable. Again, 0.75% SNP (Figure 3C) and 1 mM NAM (Figure 3D) gave the best color contrast at 395 nm (also about 0.2 absorbance units). However, whereas substrate concentrations were decreased down to 1 mM in the SNP method, the revealing time increased up to 4 h to see measurable differences in color.

Collectively, these results show that this new SNP assay could be used as a re-screening method in order to find not only highly active pyrazinamidases, but also to easily distinguish among more or less active nicotinamidases in a quantitative (spectrophotometric) way.

Once the two new whole-cell screening methods were optimized with a known nicotinamidase (OiNic) (Sanchez-Carron et al., 2013), a fosmid library obtained from mesophilic MB polygenomic library was first screened with the PZA/AFS method. After induction with the autoinduction solution (Epicentre, USA) (Supplementary Figure S2), only 8 over 8832 fosmid clones screened were positive with a different degree of activity, as shown by their different color intensity (Figure 4A). These clones were then re-screened with NAM (Figure 4B, black bars) and PZA (Figure 4B, gray bars) with the SNP method, confirming not only the presence of pyrazinamidase activity but also its correlation with the nicotinamidase activity.

Clear differences between clones with respect to both activities were observed (Figure 4B), and for a proof-of-principle, one of the clones with similar high nicotinamidase and pyrazinamidase activity (Figure 4B; L24 M14) was selected to be sequenced, since it showed the most balanced decrease in absorbance with both substrates compared to the control. BLAST analysis of the amino acid sequence obtained from this polygenomic nicotinamidase (PolyNic) (Figure 5) showed a 91% identity with a nicotinamidase found in Halomonas sp. HL-48 (UniProt code: A0A0P7YXE9). In addition, PolyNic also showed variable similarity (25–46%, Supplementary Table S1) with the crystallized nicotinamidases found in the bibliography (Figure 5) (Du et al., 2001; Hu et al., 2007; Fyfe et al., 2009; French et al., 2010a; Gazanion et al., 2011; Liu et al., 2011; Petrella et al., 2011).

The alignment also revealed that PolyNic contained all the conserved catalytic triad residues of the cysteine-hydrolases family (Figure 5, filled circles): a catalytic cysteine at position 150 (C150, PolyNic numbering), an aspartate at position 25 (D25) and a lysine at position 111 (K111) (French et al., 2010b; Sanchez-Carron et al., 2013; Ion et al., 2014; Sheng and Liu, 2014). Furthermore, PolyNic showed the characteristic cis-peptide bond (Figure 5, positions L145 and A146, filled squares), which is invariably preceded by a conserved glycine (G144), and the conserved specific metal ion binding, which includes one aspartate (D67) and two histidines (H69 and H86) (Figure 5, filled triangles). Depending on the metal ion and on the structural conformation of the protein, a fourth residue may be implicated in the metal ion binding (Sanchez-Carron et al., 2013). This residue can be a glutamic acid, a histidine or a serine, being the last one the case of PolyNic (S75). Other residues also involved in the formation of the hydrophobic cavity, where NAM and the ion metal bind (Zhang et al., 2008; Sanchez-Carron et al., 2013) are also present in PolyNic, such as F30, L36, W83, A112, Y120, and V149 (Figure 5, filled stars; Supplementary Table S2).

PolyNic functional characterization was carried out after cloning its sequence into pET46 Ek/LIC vector. The recombinant clone with the highest expression was induced with 0.5 mM IPTG for 16 h at 25°C in TB medium with constant agitation and purified to homogeneity from Escherichia coli Rosetta 2 DE3 cells by a simple procedure based in a Ni²⁺-chelating affinity chromatography and a gel filtration step onto a Superdex 200, as described in the “Materials and Methods” Section. The molecular mass of the purified protein was determined by SDS-PAGE (∼23 kDa), HPLC/ESI/ion trap (24.1 kDa) and gel filtration in a Superdex 200 10/300 GL column (50.3 kDa). These experiments confirmed its dimeric nature.

The enzyme activity was both pH- and temperature-dependent. The activity increased as pH increases from pH 5.0 to pH 10.0 (Figure 6A), being the first nicotinamidase described with its optimal activity at that basic pH, since the highest optimum pH described in the bibliography was pH 8.0 for the Lactobacillus arabinosus 17-5 nicotinamidase (Hughes and Williamson, 1953). PolyNic exhibited its maximum activity at a temperature of 50°C (Figure 6B), which is 5 to 20°C higher than those described for other nicotinamidases (Pardee et al., 1971; Yan and Sloan, 1987; Zhang et al., 2008; Sanchez-Carron et al., 2013). However, and for comparison purposes, kinetic parameters were carried out at pH 7.3 and 37°C, as generally accepted in the bibliography (Zhang et al., 2008; French et al., 2010b; Sanchez-Carron et al., 2013).

The activity of PolyNic was studied toward NAM and some of its derivatives (Table 1), which include PZA, 5-methylnicotinamide and two nicotinate esters (methylnicotinate and ethylnicotinate). PolyNic showed a quite similar catalytic efficiency for PZA (k_cat/K_M 69.9 mM^-1s^-1) and for NAM (k_cat/K_M 102.0 mM^-1 s^-1). Thus, PolyNic is the first described bacterial nicotinamidase with these characteristics, since the rest of characterized nicotinamidases have a marked preference for NAM instead of PZA (more difference between their catalytic efficiencies), including that of Mycobacterium tuberculosis (Zhang et al., 2008; French et al., 2010b; Sanchez-Carron et al., 2013). Furthermore, PolyNic also showed activity toward 5-methylnicotinamide, although with a lower catalytic efficiency (k_cat/K_M 18.9 mM^-1 s^-1) compared to NAM and PZA, but it was still 2.5-fold higher than that described before for OiNic (k_cat/K_M 7.5 mM^-1 s^-1) (Sanchez-Carron et al., 2013). Interestingly, the enzyme was active toward methylnicotinate (k_cat/K_M 19.7 mM^-1 s^-1), with the highest ever-described catalytic efficiency. In fact, PolyNic is 293-fold more active toward methylnicotinate than OiNic (Sanchez-Carron et al., 2013). Finally, ethylnicotinate was a very poor substrate for PolyNic (k_cat/K_M 1.0 mM^-1s^-1) as usual in this kind of enzymes, but still ninefold higher than for OiNic (Sanchez-Carron et al., 2013).

Nicotinaldehydes have been reported to act as good competitive inhibitors for several nicotinamidases (French et al., 2010b; Sanchez-Carron et al., 2013). Thus, recombinant PolyNic was tested with nicotinaldehyde and 5-bromo-nicotinaldehyde, toward both substrates, NAM (Figure 7A) and PZA (Figure 7B), respectively. The K_i value for nicotinaldehyde with NAM was found to be 0.18 μM (Figure 7A; circles), a similar value to that found in Borrelia burgdorferi nicotinamidase (0.11 μM) (French et al., 2010b). The K_i value for 5-bromo-nicotinaldehyde with NAM was higher (0.72 μM) (Figure 7A; squares) than that of nicotinaldehyde, and similar to that found for the Plasmodium falciparum nicotinamidase (0.57 μM) (French et al., 2010b). Interestingly, the K_i values for nicotinaldehyde and 5-bromo-nicotinaldehyde obtained with PZA were 0.015 and 0.07 μM, respectively (Figure 7B), which are much lower than those obtained using NAM as substrate. These values could not be compared since no data on nicotinaldehyde inhibition with PZA have been published.

With the K_i values described above, our goal was to develop the first whole-cell screening method for testing nicotinamidase inhibitors using the PZA/AFS and NAM/SNP methods. Thus, different concentrations of nicotinaldehyde and 5-bromo-nicotinaldehyde (0–1500 μM) were added to each well inoculated with PolyNic in E. coli Rosetta 2 before adding the revealing solution (20 mM PZA/1% AFS or 1 mM NAM/0.75% SNP). Surprisingly, SNP was not suitable for inhibitor screening because the difference of absorbance between the untreated and the treated wells was not enough to rely on this method with both substrates. Furthermore, at the highest concentration of the inhibitors, the absorbance was greater than expected in comparison with the other wells, probably because of a reaction between SNP and the inhibitors themselves (data not shown).

However, in the case of the AFS method, it was possible to find a gradient of colors (from red, orange to uncolored) as the inhibitor concentration rose (Figure 8), finding by the naked eye that nicotinaldehyde (Figure 8A) is a better inhibitor than 5-bromo-nicotinaldehyde (Figure 8B), since color formation stops at lower concentrations with nicotinaldehyde (75–100 μM) than with 5-bromo-nicotinaldehyde (250–500 μM) (Figure 8), which correlates well with the above-described kinetic results. These results make this whole-cell PZA/AFS assay a simple HTS qualitative prescreening method for nicotinamidase inhibitors.