Aminofutalosine Deaminase in the Menaquinone Pathway of Helicobacter pylori

Abstract

Helicobacter pylori is a Gram-negative bacterium that is responsible for gastric and duodenal ulcers. H. pylori uses the unusual mqn pathway with aminofutalosine (AFL) as an intermediate for menaquinone biosynthesis. Previous reports indicate that hydrolysis of AFL by 5′-methylthioadenosine nucleosidase (HpMTAN) is the direct path for producing downstream metabolites in the mqn pathway. However, genomic analysis indicates jhp0252 is a candidate for encoding AFL deaminase (AFLDA), an activity for deaminating aminofutolasine. The product, futalosine, is not a known substrate for bacterial MTANs. Recombinant jhp0252 was expressed and characterized as an AFL deaminase (HpAFLDA). Its catalytic specificity includes AFL, 5′-methylthioadenosine, 5′-deoxyadenosine, adenosine, and S-adenosylhomocysteine. The k_cat/K_m value for AFL is 6.8 × 10⁴ M⁻¹ s⁻¹, 26-fold greater than that for adenosine. 5′-Methylthiocoformycin (MTCF) is a slow-onset inhibitor for HpAFLDA and demonstrated inhibitory effects on H. pylori growth. Supplementation with futalosine partially restored H. pylori growth under MTCF treatment, suggesting AFL deamination is significant for cell growth. The crystal structures of apo-HpAFLDA and with MTCF at the catalytic sites show a catalytic site Zn²⁺ or Fe²⁺ as the water-activating group. With bound MTCF, the metal ion is 2.0 Å from the sp³ hydroxyl group of the transition state analogue. Metabolomics analysis revealed that HpAFLDA has intracellular activity and is inhibited by MTCF. The mqn pathway in H. pylori bifurcates at aminofutalosine with HpMTAN producing adenine and depurinated futalosine and HpAFLDA producing futalosine. Inhibition of cellular HpMTAN or HpAFLDA decreased the cellular content of menaquinone-6, supporting roles for both enzymes in the pathway.

Graphical Abstract

Individuals infected with Helicobacter pylori have an increased risk of developing gastric malignancies, which are the third most deadly cancers according to GLOBOCAN 2018 data.¹ Menaquinone (MK or vitamin K₂) is a membranous quinone and a bacterial redox cofactor that transfers electrons in the electron transport chain of prokaryotes.² It is an essential component of anaerobic respiration in Gram-negative bacteria.^2–5 The classical MK biosynthetic pathway in Escherichia coli consists of nine men genes and is well conserved in prokaryotes.⁶ More recently, an alternative pathway, the futalosine pathway, was discovered and is found in Streptomyces coelicolor, Chlamydia trachomatis, Thermus thermophilus, Acidothermus cellulolyticus, Wolinella succinogenes, Helicobacter pylori, Campylobacter jejuni, and others [mqn genes (Figure 1)].^5–9 Analyses of the taxonomic distribution of mqn genes suggest that the futalosine pathway may have appeared earlier than the classical men pathway in evolution.⁶

Figure 1.

Futalosine pathway route that diverges at aminofutalosine to form futalosine (red, deamination) and nucleoside hydrolysis by MTAN (colored blue to form dehypoxanthinefutalosine). Other organisms are reported to express an MqnB, but this activity was not observed in H. pylori.

H. pylori is a Gram-negative pathogenic bacterium that colonizes the human gastric epithelium.¹⁰ More than half of the global population is infected by H. pylori, which can lead to duodenal and gastric ulcers.^11,12 H. pylori-induced gastritis is responsible for >60% of the world’s gastric carcinogenesis patients, making it the only bacterium identified as a class I carcinogen.^13–15 Frequent antibiotic use is causing drug resistance in H. pylori to compromise the effectiveness of conventional antibiotic-based therapy.16 Clarithromycin-resistant H. pylori strains are listed in the high-priority group requiring new antibiotics by the World Health Organization (WHO).¹⁷ H. pylori solely relies on the futalosine pathway for MK production, as most of the men genes are absent.^5,7,18 The futalosine pathway is rarely utilized in human gut microbiota. Thus, the futalosine pathway provides potential for developing H. pylori-targeted therapeutic agents that can treat ulcers and reduce gastric cancer incidence with minimal impact on healthy gut flora.

One of the unique steps in the mqn pathway is the formation of dehypoxanthine futalosine (DHFL). Previous studies have reported that aminofutalosine (AFL) is converted into futalosine (FL) by AFL deaminase followed by MqnB-catalyzed hydrolysis to form DHFL in T. thermophilus and S. coelicolor (Figure 1).^5,7 In contrast, the presence of an active 5′-methylthioadenosine nucleosidase (MTAN) in H. pylori directly hydrolyzes AFL to form DHFL, while HpMTAN cannot utilize FL as a substrate.^7,19 Moreover, it was originally reported that H. pylori does not possess a typical ortholog of Acel_2064, the gene encoding AFL deaminase in A. cellulolyticus.⁷

MTAN is a multifunctional enzyme that can hydrolyze the N-ribosidic bond of 5′-methylthioadenosine (MTA), S-adenosylhomocysteine (SAH), 5′-deoxyadenosine (5′dAdo), and AFL.^7,19,20 The broad substrate specificity of MTAN allows it to participate in several biological processes, including polyamine biosynthesis (MTA degradation), S-adenosyl-L-methionine (SAM) metabolism, the methionine salvage pathway, bacterial quorum sensing, and MK biosynthesis. Organisms lacking MTANs utilize a 5′-methylthioadenosine → 5′-methylthioinosine → hypoxanthine + ribose pathway for MTA degradation.²¹ Previously, we reported an MTA deaminase in Pseudomonas aeruginosa that catalyzes the deamination of both adenosine and MTA.²¹ Interestingly, a previous study reported an adenosine deaminase capable of catalyzing SAH deamination in H. pylori strain 26695,²² leading to the question of whether deamination of MTA/AFL also takes place in H. pylori.

Here, we report the protein encoded by jhp0252 to be an AFL/MTA deaminase of H. pylori strain J99. In vitro kinetic analysis characterized substrate specificity, and inhibitor screening indicates 5′-methylthiocoformycin (MTCF) as a slow-onset inhibitor for AFL deaminase. MTCF was shown to have inhibitory effects on pylori growth. The crystal structure of MTCF-bound jhp0252 revealed structural interactions at the active site. Additionally, metabolite analyses in intact H. pylori cells established the intracellular activity and inhibition of jhp0252 activity by MTCF.

MATERIALS AND METHODS

Chemicals.

Coformycin (CF), 5′-methylthiocoformycin (MTCF), 5′-methylthio-2′-deoxycoformycin (MTDCF), 5′-propylthio-2′-deoxycoformycin (PrTDCF), and 5′-phenylthio-2′-deoxycoformycin (PhTDCF) were synthesized by methods reported previously.²³ Inhibitors were dissolved in distilled water, and the concentrations determined from the extinction coefficient (8.25 mM⁻¹ cm⁻¹ at 282 nm). 6-Amino-6-deoxyfutalosine (AFL) and futalosine (FL) were synthesized by methods reported previously.^6,19,23,24 All other chemicals and reagents were purchased from Sigma or Fisher Scientific and were of reagent grade.

Plasmid Construction and Enzyme Purification.

The jhp0252 gene from the H. pylori strain J99 genome database²⁵ was predicted to encode an AFL deaminase (HpAFLDA) and was used for plasmid design. The E. coli codon-optimized jhp0252 gene was cloned into the N-terminal TEV cleavable N-terminal His₆ tag containing vector pJ express414 from ATUM DNA2.0 Gene Design. E. coli BL21 (DE3) was transformed with plasmid pJ411-HpAFLDA for recombinant protein expression. LB medium (50 mL) with 100 μg/mL ampicillin was inoculated with one single colony, and the culture was incubated at 37 °C overnight before being transferred into 1 L of LB medium. When the OD₆₀₀ reached 0.6–0.7, 1 mM IPTG was added to induce protein expression at 23 °C for 10 h. The cells were harvested by centrifugation at 4500g for 30 min. The cell pellet was resuspended in 20 mL of lysis buffer containing 1× BugBuster (Millipore Sigma), 100 μg/mL lysozyme (Sigma), 25 μg/mL DNaseI (Sigma), and 1 tablet of EDTA-free protease inhibitor (Roche). After being stirred for 20 min, the lysed cells were centrifuged at 16000g for 30 min at 4 °C to remove cell debris. The collected supernatant was incubated with Ni-NTA agarose [1.7 mL of slurry/g of cell pellet (Qiagen)] for 1 h with rocking. The mixture was then transferred into a column, and the resin was washed with 200 mL of wash buffer [50 mM HEPES (pH 7.4), 50 mM KCl, and 1 mM DTT]. The target protein was eluted with 50 mL of elution buffer [50 mM HEPES (pH 7.4), 50 mM KCl, 1 mM DTT, and 50 mM imidazole]. Sodium dodecyl sulfate–polyacrylamide gel electrophoresis analysis was used to determine the presence and purity of the target protein. Size-exclusion chromatography (Hi-Load Superdex 200 column) was used as the final purification step. The eluted protein was stored in 50 mM HEPES (pH 8.0), 50 mM KCl, and 0.5 mM DTT.

Enzymatic Assays for HpAFLDA.

The deaminase activity on AFL, MTA, 5′dAdo, adenosine, and SAH was measured by the absorbance change (Δε = 8012 M⁻¹ cm⁻¹) at 263 nm.²⁶ Reactions were performed in 1 cm cuvettes at 25 °C. Assay mixtures (1 mL) contained 50 mM HEPES (pH 7.0), variable concentrations of the substrate, and appropriate amounts of purified HpAFLDA. Reactions were started by addition of HpAFLDA, and the initial rates were monitored on a CARY 300 ultraviolet–visible spectrophotometer for steady state kinetic calculation. The kinetic constants were obtained by fitting initial rates to the Michaelis–Menten equation in Graphpad Prism 8.

Inhibition Assays.

Inhibition assay reactions were initiated with 77 nM HpAFLDA in an assay mixture containing 400 μM MTA and variable concentrations of the inhibitor in 50 mM HEPES (pH 7.0) at 25 °C. Controls without enzymes were included in all of the inhibition assays. Inhibition constants were obtained by fitting initial rates with variable inhibitor concentrations to eq 1 using Graphpad Prism 8.

$$\frac{v_{\text{i}}}{v_0}=\frac{[\text{S}]}{K_{\text{m}}+[\text{S}]+\frac{K_{\text{m}}[\text{I}]}{K_{\text{i}}}}$$ (1)

where v_i and v₀ are the initial rates with and without an inhibitor, respectively, [S] and [I] are the substrate and inhibitor concentrations, respectively, K_m is the Michaelis constant, and K_i is the inhibition constant. The inhibitor concentration was corrected by eq 2 when it was <10 times the enzyme concentration.²⁷

$$[\text{I}{]}^{\prime }=[\text{I}]-\left(1-\frac{v_{\text{i}}}{v_0}\right)E_{\text{t}}$$ (2)

where [I]′ is the effective inhibitor concentration, [I] is the inhibitor concentration in the assay mixture, and E_t is the total enzyme concentration.

Inhibition of H. pylori Growth.

H. pylori J99 was cultured as described previously.²⁸ Briefly, it was grown under microaerophilic conditions (5% O₂, 10% CO₂, and 85% N₂) at 37 °C in brain heart infusion medium (Oxoid) with 10% fetal bovine serum (Sigma). The half-maximal inhibitory concentration (IC₅₀) for MTCF was determined in an H. pylori culture in the exponential growth phase diluted to an OD₆₀₀ of 0.010. Samples of 100 μL were transferred to 96-well plates. Increasing final concentrations of MTCF (1–250 μM) were added to wells followed by incubation for 72 h under microaerophilic conditions at 37 °C.

A SpectraMax M5 plate reader (Molecular Devices) was used to determine the OD₆₀₀ value of each well, and the IC₅₀ value was obtained using a nonlinear regression curve fit in GraphPad Prism 8. Rescue of H. pylori growth from MTCF treatment experiments used 6, 13, 25, 50, and 100 μM MTCF supplemented with 50 μM FL. Actively growing H. pylori was inoculated in each treatment with the initial OD₆₀₀ of 0.005. Cultures were maintained under microaerophilic conditions at 37 °C for 96 h followed by OD₆₀₀ analysis.

Intracellular Metabolite Analysis.

The metabolite extraction method was described previously.^29,30 Briefly, ice-cold extraction buffer (500 μL; 40:40:20 acetonitrile/methanol/water mixture and 0.1 M formic acid) with an internal standard at known concentrations [(RS)-S-adenosyl-L-methionine-d₃ (CDN Isotopes)] was added to pelleted H. pylori cells. Samples were vortexed and incubated on ice for 10 min followed by centrifugation at 20200g for 10 min at 4 °C. The supernatant was transferred to a fresh tube and dried under vacuum overnight. Prior to injection, samples were dissolved in 50 μL of distilled H₂O. The lipid extraction for menaquinone-6 analysis was carried out using a chloroform/methanol-based method described by Bligh and Dyer.³¹ H. pylori cells (~2.5 million per sample) were washed in PBS buffer after being harvested; 750 μL of a 1:2 chloroform/methanol mixture with added menaquinone-4 at known concentrations (internal standard) was added to cell pellets followed by a 2 min vortex. Then, 250 μL of chloroform was added to the mixture followed by a 30 s vortex. After that, 250 μL of 1.5 M NaCl was added to the extraction mixture followed by a 30 s vortex. After centrifugation at 20200g for 20 min, the bottom layer (chloroform) was collected and lyophilized in separate tubes, and 100% methanol was used to dissolve extracted lipids before LC-MS analysis. Biological triplicates were prepared for each treatment group in this study.

High-performance liquid chromatography (HPLC) with detection by mass spectrometry (LC-MS) was conducted on an Agilent Technologies (Santa Clara, CA) 1200 system coupled with an Agilent Technologies 6410 triple quadrupole mass spectrometer. The system was operated with the associated MassHunter software package, which was also used for data collection and analysis. Assay mixtures were separated on a Zorbax Rapid Resolution SB-C18 column (2.4 mm × 35 mm, 3.5 μm particle size) equilibrated in 99% buffer A [5 mM perfluoroheptanoic acid and 6 mM ammonium formate (pH 3)] and 1% buffer B (100% acetonitrile). A gradient of 1% to 12% buffer B was applied from 0 to 2 min, followed by a gradient of 12% to 30% buffer B from 2 to 3 min. Then, a gradient of 30% to 50% buffer B was applied from 3 to 5 min before returning to 1% buffer A from 5 to 6 min. The column was allowed to re-equilibrate for 1.5 min under initial conditions before subsequent sample injections. Detection of products was performed by two separate injections using electrospray ionization in negative mode (ESI⁻) with the following MRM methods (Table S1). For the analysis of menaquinone-6, the extracted lipids were separated on the SB-C18 column equilibrated in 12% buffer A (0.1% formic acid) and 88% buffer B (100% acetonitrile). A gradient of 88% to 99% buffer B was applied from 0 to 0.5 min followed by 22 min with 1% buffer A and 99% buffer B. The column was equilibrated back to initial conditions within 1.5 min before the next injection.

AFL and FL Hydrolysis by HpMTAN and H. pylori Cell Mixture.

AFL and FL hydrolysis by HpMTAN was measured in a discontinuous assay in 1 mL mixtures of 50 mM HEPES (pH 7.0), variable concentrations of the substrate, and purified HpMTAN. Reactions were started by addition of HpMTAN and quenched by addition of sulfuric acid to a final concentration of 0.01 M. Reactions were analyzed by LC-MS using the same method as for metabolite analysis. Initial rates were used for steady state kinetic calculations in GraphPad Prism 8. Live H. pylori cell mixtures were prepared from overnight cultures and washed twice with PBS prior to incubations. AFL or FL (25 μM) was added to ~10⁸ cells in PBS (pH 7.0, 700 μL) to initiate incubations. Supernatants were collected at various time points and analyzed by LC-MS to determine metabolites.

Crystallization.

HpAFLDA (5 mg/mL) was screened for crystallization with commercially available Microlytic (MCSG1–4) and Hampton (crystal screenHT) crystallization conditions. The crystallization of the apoenzyme was performed by sitting drop vapor diffusion at 22 °C. Crystallization used the CRYSTAL-GRYPHON crystallization robot (Art Robbins) in 96-well INTELLI plates (Art Robbins). Crystallization drops contained 0.5 μL of apo HpAFLDA and 0.5 μL of a well solution. The well solution volume was 70 μL. Diffraction-quality crystals were obtained in 2 weeks from 100 mM sodium malonate (pH 4.0) and 12% (w/v) polyethylene glycol 3350.

Co-crystallization of HpAFLDA with MTCF used 5 mg/mL enzyme mixed with MTCF in a 1:2 molar ratio, incubated on ice for 2 h. Crystallization was performed as described above for apo HpAFLDA. Crystals appeared in 2 weeks from a crystallization condition of 100 mM sodium acetate (pH 4.6) and 8% (w/v) polyethylene glycol 4000. Crystals were cryoprotected with 20% ethylene glycol and frozen in liquid nitrogen before X-ray diffraction data collection.

Data Collection and Data Processing.

Diffraction data of apo and MTCF-bound HpAFLDA crystals were collected at the LRL-CAT beamline (Argonne National Laboratory, Lemont, IL) irradiated at a wavelength of 0.97931 Å to resolutions of 2.79 and 1.89 Å, respectively (Table 4). The data were processed using iMOSFLM and scaled by the AIMLESS program of the CCP4 suite in the P12₁1 space group (Table 4).^32,33 The quality of the crystal diffraction data was analyzed by SFCHECK and XTRIAGE.^33,34 The Matthews coefficient (V_m) calculations were performed to calculate monomers present in the asymmetric unit. The data collection and processing statistics are summarized in Table 4.

Table 4.

Data Collection and Refinement Statistics of HpAFLDA Structures^a

	apo-HpAFLDA	complex of HpAFLDA with MTCF
Unit Cell Data
space group	P1211	P1211
cell parameters	a = 73.29 Å, b = 73.29 Å, c = 157.49 Å	a = 72.90 Å, b = 73.07 Å, c = 157.44 Å
	α = 90.0°, β = 98.6°, γ = 90.0°	α = 90.0°, β = 98.3°, γ = 90.0°
Vm (Å3/Da)	2.2	2.2
no. of subunits in the asymmetric unit	4.0	4.0
Data Collection
beamline	LRL-CAT	LRL-CAT
wavelength (Å)	0.97931	0.97931
temperature (K)	100	100
resolution range (Å)	155.73–2.79 (2.90–2.79)	155.79–1.89 (1.92–1.89)
total no. of observed reflections	140666 (14721)	494028 (24293)
no. of unique reflections	39040 (4432)	129546 (6388)
Rmerge (%)b	22.7 (105.7)	12.1 (68.3)
CC1/2 (%)	97.7 (54.7)	99.2 (66.4)
〈I/σ(I)〉c	5.0 (1.3)	6.5 (1.9)
completeness (%)	94.2 (95.7)	98.8 (98.1)
multiplicity	3.6 (3.3)	3.8 (3.8)
Wilson B-factor (Å2)	34.7	21.0
Refinement
Rwork (%)d	25.0	23.1
Rfree (%)e	28.7	26.9
no. of atoms	12744	14029
protein	12570	12904
ligand	–	84
solvent	174	1041
model quality
root-mean-square deviation from ideal values
bond lengths (Å)	0.006	0.003
bond angles (deg)	0.98	0.73
average B-factor
protein atoms (Å2)	41.3	26.2
ligand atoms (Å2)	–	28.8
waters (Å2)	30.5	31.4
Ramachandran plotf
most favored regions (%)	94.1	96.6
allowed regions (%)	5.4	3.2
outlier regions (%)	0.5	0.2
PDB entry	7LKJ	7LKK

^aValues in parentheses refer to the highest-resolution shell.

^bR_merge = [Σ_hklΣ_i|I_i(hkl) − 〈I(hkl)〉|]/Σ_hklΣ_i〈I_i(hkl)〉, where I_i(hkl) is the intensity of the ith measurement of reflection (hkl) and 〈I(hkl)〉 is its mean intensity.

^cI is the integrated intensity, and σ(I) is its estimated standard deviation.

^dR_work = (Σ_hkl|F_o − F_c|)/Σ_hklF_o, where F_o and F_c are the observed and calculated structure factors, respectively.

^eR_free is calculated as described for R_work but from a randomly selected subset of the data (5%), which were excluded from the refinement calculation.

^fCalculated by MOLPROBITY.

Structure Determination and Refinement.

Crystal structures of apo and MTCF-bound HpAFLDA were determined by molecular replacement using PHASER.³⁵ Chain A of wild-type Nitratiruptor amidohydrolase [aminodeoxyfutalosine deaminase, Protein Data Bank (PDB) entry 3V7P, UniProt entry A6Q234] was used as the initial phasing model. The model obtained from PHASER was autobuilt using BUCCANEER and then manually adjusted and completed using COOT.^36,37 Refinement of the structures was performed by PHENIX-REFINE.³⁴ The final refinement statistics of the structures are summarized in Table 4.

Structure Analysis.

The crystal structure of apo-amidohydrolase (aminodeoxyfutalosine deaminase, PDB entry 3V7P, chain A, UniProt entry A6Q234) from Nitratiruptor was used for structural comparisons. The MTCF complex with P. aeruginosa adenosine deaminase (PDB entry 4GBD, chain A) was also used in the structural comparisons. Structural superimpositions used the SSM protocol of COOT. The geometry analyses of the final model used MolProbity.³⁸ The B-factors of the structures were calculated with the BAVERAGE program of the CCP4 suite. Structural figures were produced by PyMOL. For HpAFLDA structures, subunit A was used for all structural analyses and comparisons.

RESULTS AND DISCUSSION

jhp0252 Encodes an Aminofutalosine Deaminase.

It was previously reported that H. pylori lacks AFL deaminase orthologs, and its DHFL production depends solely on AFL hydrolysis by HpMTAN.⁷ However, the KEGG database for the H. pylori J99 genome contained a gene entry, jhp0252, annotated as a potential aminofutalosine deaminase (AFL deaminase). We expressed and purified the recombinant jhp0252 protein and investigated its substrate specificity (Table 1). The recombinant enzyme displayed catalytic activity for deamination of AFL, MTA, 5′dAdo, adenosine, and SAH, suggesting a high degree of substrate promiscuity. Among the tested substrates, AFL was the most favorable substrate with a k_cat of 0.62 s⁻¹ and a K_m of 9 μM (k_cat/K_m of 6.8 × 10⁴ M⁻¹ s⁻¹). The jhp0252 deaminase showed comparable activity for deamination of MTA and 5′dAdo with k_cat/K_m values of 1.3 × 10⁴ and 1.2 × 10⁴ M⁻¹ s⁻¹, respectively. The catalytic efficiency for adenosine was 2.6 × 10³ M⁻¹ s⁻¹, which was 26-fold lower than that for AFL. Recombinant jhp0252 showed lower activity in catalyzing SAH deamination (k_cat/K_m of 5.5 × 10² M⁻¹ s⁻¹). On the basis of the substrate profile, we concluded that jhp0252 encodes an active AFL deaminase (HpAFLDA) in H. pylori J99.

Table 1.

Substrate Specificities of HpAFLDA and Steady State Kinetic Constants

substrate	Km (μM)	kcat (s−1)	kcat/Km (M−1 s−1)
AFL	9 ± 1	0.62 ± 0.03	6.8 × 104
MTA	100 ± 10	1.24 ± 0.07	1.3 × 104
5′dAdo	130 ± 50	1.6 ± 0.3	1.2 × 104
adenosine	170 ± 40	0.44 ± 0.06	2.6 × 103
SAH	220 ± 50	0.12 ± 0.02	5.5 × 102

Deamination activity on AFL has been reported in Sav2595 from Steptomyces avermitilis as well as its distantly related orthologs (Acel0264, Dr0824, and Nis0429) from three thermophilic bacteria.³⁹ The characterized catalytic efficiencies for AFL deamination range from 4.8 × 10⁶ to 7.9 × 10⁵ M⁻¹ s⁻¹, which are approximately 1–2 orders of magnitude higher than that for jhp0252. However, jhp0252 has greater k_cat/K_m values for MTA and 5′Ado than other AFL deaminases except Dr0824 from Deinococcus radiodurans. It is worth noting that the kinetic characterization of these four enzymes was carried out at 30 °C instead of 25 °C, which was used to assay jhp0252. The catalytic activity for jhp0252 on AFL (6.8 × 10⁴ M⁻¹ s⁻¹) is similar to that for the activity of Plasmodium falciparum adenosine/MTA deaminase on MTA (9 × 10⁴ M⁻¹ s⁻¹).²³ Despite the well-defined catalytic potential for the substrates listed in Table 1, other substrates remain an alternative possibility for this enzyme.

Inhibitors of HpAFLDA.

Coformycin (CF) and 2′-deoxycoformycin (2′d-CF), originally isolated from Streptomyces, are natural product transition state analogues of human adenosine deaminase (HsADA) with submicromolar K_i values.^23,40 Coformycin analogues, methylthio-coformycin (MTCF) and methylthio-2′-deoxy-coformycin (MT-2′d-CF), were designed to be specific inhibitors of MTA deaminase from P. falciparum, as the 5′-methylthio group provided additional target selectivity. We tested (8R)-CF, (8R)-MTCF, (8R)-MT-2′d-CF, (8R)-PrT-2′d-CF, and (8R)-PhT-2′d-CF as inhibitors of HpAFLDA (Figure 2A). Initial rates were determined and used to calculate initial inhibition constants (K_i). The MTCF-inhibited reactions showed a second phase of inhibition after reaction for 30 min (Figure 2B). A similar inhibition pattern for MTCF has been reported for PaMTADA,²¹ establishing MTCF as a slow-onset inhibitor of HpAFLDA. Reaction rates following slow-onset inhibition were used to obtain the second-phase inhibition constant (K_i*). MTCF was the most potent HpAFLDA inhibitor with a K_i* of 63 nM (Table 2). Surprisingly, MT-2′d-CF did not inhibit HpAFLDA activity, even at 10 μM, establishing that the 2′-hydroxyl group is essential for binding of the CF analogue to HpAFLDA. Additional 2′d-CF analogues with modifications at the 5′-subsitutent, including PrT-2′d-CF and PhT-2′d-CF, showed some inhibition, indicating that 5′-substituents also play a critical role in inhibitor affinity.

Figure 2.

(A) Coformycin analogues tested for inhibition of HpAFLDA. (B) Inhibition of HpAFLDA by MTCF. The initial inhibition constant (K_i) was calculated from data collected for the initial rates of the reactions. The slow-onset inhibition constant (K_i*) was calculated from data collected following completion of the slow-onset inhibition.

Table 2.

Summary of K_i Values for HpAFLDA^a

inhibitor	Ki (nM)	Ki* (nM)
(8R)-CF	N/O	N/O
(8R)-MTCF	560 ± 170	63 ± 4
(8R)-MT-2′d-CF	N/O	N/O
(8R)-PrT-2′d-CF	420 ± 140	N/O
(8R)-PhT-2′d-CF	850 ± 200	N/O

^aN/O means inhibition not observed.

Inhibition of H. pylori Growth.

Is FL essential for bacterial growth? Addition of MTCF to H. pylori cultures demonstrated an inhibitory effect on H. pylori growth in liquid medium with an IC₅₀ of 14 ± 1 μM (Figure 3A). Moreover, an addition of 50 μM FL partially reversed the growth inhibition (Figure 3B). Thus, FL is an important metabolite in H. pylori growth, and impairment of HpAFLDA intracellular activity inhibits H. pylori growth.

Figure 3.

(A) Inhibition of H. pylori growth by MTCF. (B) The inhibitory effect of MTCF on H. pylori growth is reduced by supplementation with 50 μM futalosine (FL).

Inhibition of Intracellular HpAFLDA Activity by MTCF.

Inhibition of cell growth, assumed to result from inhibition of HpAFLDA, requires experimental validation. Cells were cultured with and without MTCF at 14 μM (IC₅₀ value), and cell extracts were analyzed for MTA, MTI, AFL, FL, and DHFL. In the MTCF-treated groups, MTI and FL levels were significantly reduced [p < 0.0001 and p < 0.005, respectively (Figure 4)]. Depletion of the HpAFLDA products in MTCF treatment cells establishes that HpAFLDA activity was reduced.

Figure 4.

Metabolic analysis of cellular inhibition of HpAFLDA. (A) Inhibitory effect on the conversion of MTA to MTI. (B) Inhibitory effect on the conversion of AFL to FL. (C) Reduced HpAFLDA or HpMTAN activity led to decreased DHFL production. (D) Menaquinone-6 biosynthesis was negatively modulated by HpAFLDA or HpMTAN inhibition.

MTA and AFL both serve as substrates for HpAFLDA, and both are expected to accumulate upon inhibition of HpAFLDA if HpAFLDA catalyzes their major metabolic path in vivo. MTCF treatment caused an increase in AFL abundance [p < 0.0001 (Figure 4B)] caused by decreased HpAFLDA activity. However, the level of cellular MTA did not increase following MTCF treatment. MTA and AFL are also both substrates of HpMTAN, a nucleoside hydrolase with broad substrate specificity.²⁰ The k_cat/K_m values for MTA and AFL hydrolysis by HpMTAN were 8.3 × 10⁶ and 6.9 × 10⁴ M⁻¹ s⁻¹, respectively (Supporting Information). As the catalytic efficiency of MTA hydrolysis is >100-fold higher than that of AFL hydrolysis, this metabolic analysis indicates that HpMTAN is efficient at the hydrolysis of MTA but not AFL when intracellular HpAFLDA activity is inhibited.

DHFL is the reaction product of HpMTAN-catalyzed AFL hydrolysis^7,19 and is an essential menaquinone precursor in the futalosine pathway (Figure 1).^5–9 The role of HpAFLDA in menaquinone biosynthesis was examined by treatment with MTCF at the IC₅₀ concentration. The cellular DHFL levels decreased significantly [p < 0.005 (Figure 4C)] in the MTCF-treated groups relative to the untreated controls. Significant decreases in cellular DHFL concentrations were also observed in cells cultured with BuT-DADMe-ImmA (BTDIA), an HpMTAN transition state analogue,¹⁹ at its half-maximal inhibitory concentration (14 nM). Additionally, we measured the levels of menaquinone-6, the dominant isoprenoid quinone in Helicobacter,⁴¹ in lipid extracts. The menaquinone-6 concentrations decreased significantly in both MTCF- and BTDIA-treated groups [p < 0.005 and p < 0.005, respectively (Figure 4D)]. It is known that HpMTAN activity is critical to MK production in H. pylori.^7,19 The inhibition of HpAFLDA by MTCF and the inhibition of HpMTAN by BTDIA both led to decreased levels of DHFL, a downstream menaquinone precursor, and menaquinone-6, the final product of the pathway. Overall, the metabolic analysis suggests that HpAFLDA plays a significant, but as yet undefined, role in modulating menaquinone biosynthesis in H. pylori.

Exploring Futalosine N-Ribosyl Hydrolase in H. pylori.

In microorganisms utilizing the AFL → FL → DHFL pathway for menaquinone biosynthesis, the MqnB hydrolase converts FL to DHFL and hypoxanthine (Figure 1).^5,7 However, no MqnB activity or genome coding region has been reported in H. pylori. In the past, HpMTAN has been misannotated as HpMqnB and has been reported to be active for hydrolysis.⁷ We tested HpMTAN for FL hydrolase activity and confirmed that HpMTAN was capable of catalyzing only AFL hydrolysis but not FL hydrolysis, in agreement with published studies (Table 3).⁷ Metabolism of AFL and FL in H. pylori was investigated by adding AFL or FL to live H. pylori cells, followed by metabolite analysis. H. pylori cells readily converted AFL to FL and adenine (Figure 5A). Thus, AFL hydrolysis and deamination occur simultaneously with accumulation of products. Cells incubated with FL caused no significant changes in the FL concentration, and no hypoxanthine was formed (Figure 5B). Under these conditions, H. pylori cells contain no functionally significant FL hydrolase.

Table 3.

Kinetic Analysis of AFL and FL Hydrolysis by HpMTAN

substrate	Km (μM)	kcat (s−1)	kcat/Km (M−1 s−1)
AFL	1.6 ± 0.3	0.022 ± 0.02	1.4 × 104
FL	n/a	<1 × 10−5	n/a

Figure 5.

Metabolism of AFL and FL by H. pylori cell culture. (A) Conversion of AFL to adenine by hydrolysis and to FL by deamination in H. pylori cells. (B) Incubation of FL with H. pylori cells.

Investigating Potential Feedback Mechanisms in the Futalosine Pathway.

The involvement of HpAFLDA in MK biosynthesis is supported by the inhibition of HpAFLDA causing metabolic perturbations in the futalosine pathway (Figure 4). We examined the possibility of potential feedback mechanisms involving HpAFLDA activity, namely, the production of FL as a feedback control when the pathway was overproducing AFL. Supplementation of culture growth medium with 200 μM AFL did not cause a significant impact on cell growth when BTDIA was used to stress MK biosynthesis (Figure 6A). Supplementation with FL at concentrations ranging from 1 to 200 μM did not affect cell growth (Figure 6B). A simple feedback regulation involving HpAFLDA activity within the futalosine pathway was not observed.

Figure 6.

Investigation of the potential effect of metabolite supplementation on cell growth. (A) Supplementation with AFL in the presence of BTDIA treatment. (B) Supplementation of FL as a potential feedback regulator of the pathway.

HpAFLDA Structure and Comparisons with Homology.

The crystal structures of apo-HpAFLDA and MTCF-bound structures were determined in space group P12₁1 with acceptable R and R_free values (Table 4). The PISA analysis supports a dimer with two catalytic sites as the crystallographic state. Except for a few residues at the N-terminus and surface side chain residues, the entire polypeptide chain of apo and HpAFLDA structures was well-defined in the electron density.

The crystal structure of apo-HpAFLDA (PDB entry 7LKJ) was determined to 2.79 Å resolution with four monomers (chains A–D) in the asymmetric unit. Polypeptide chains A and C were well-defined in the electron density map, whereas chains B and D were disordered in Phe70–Lys74 and Gly94–Asn99. The solvent accessible surface area of the dimer interface is 1196 Ų. HpAFLDA contains two domains, including a small N-terminal β-barrel domain positioned between two α-helices (Met1–Leu58 and Glu354–Ile409) and a C-terminally distorted TIM barrel domain (Pro59–Leu360) (Figure 7). The density for metal binding was observed in the active site of HpAFLDA. Both Fe²⁺ and Zn²⁺ were modeled and refined on the basis of the coordination geometry and refinement parameters. The precedent of the Fe²⁺ and Zn²⁺ atoms was established in the original description of the closely related enzyme (Hp0267, 96% identical) from H. pylori 26695 and reported to contain both Fe²⁺ and Zn²⁺ by direct metal analysis and by metal substitution experiments.^22,39,42

Figure 7.

Crystal structure of HpAFLDA. (A) Quaternary homodimeric structure of HpAFLDA. The active site of the enzyme is deeply buried in the structure and highlighted with black arrows. The two domains including the TIM barrel domain (blue) and β-barrel domain (magenta) are highlighted. (B) The active site residues involved in inhibitor binding are shown with bound MTCF. The metal ion (brown) and water (red) molecules are also indicated. The hydrogen bond and metal coordination bonds are shown with the black dotted lines. (C) Stereoview structural comparison of apo-HpAFLDA (PDB entry 7LKJ) with the MTCF-bound structure (PDB entry 7LKK). The gray structure is the unliganded monomer A of HpAFLDA, and the blue structure is the MTCF-bound monomer. There are no major structural changes observed between apo and MTCF-bound structures.

Most adenosine deaminases (both eukaryotes and prokaryotes) use Zn²⁺ as the metal cofactor.^43–45 The metal content of HpAFLDA was quantified for both Fe²⁺ and Zn²⁺ content (Supporting Information). The results showed a molar ratio of 0.38 for Fe²⁺ and enzyme active sites and a molar ratio of 0.60 for Zn²⁺ and enzyme active sites. This heterogeneous Fe²⁺ and Zn²⁺ content for HpAFLDA was similar to that found for the adenosine deaminase from Hp0267.²² Fe²⁺- or Zn²⁺-bound only enzymes were prepared for kinetic comparison in metal removal and replacement experiments (Supporting Information). The k_cat for AFL with fully Fe²⁺-substituted HpAFLDA is roughly 40% faster than that for fully Zn²⁺-substituted HpAFLDA. However, the overall catalytic efficiency for AFL for fully Fe²⁺-bound HpAFLDA is lower than that for the Zn²⁺-bound enzyme (k_cat/K_m values of 2.0 and 5.0 × 10⁴ M⁻¹ s⁻¹, respectively) (Table S2).

The structure of HpAFLDA bound to MTCF (PDB entry 7LKK) was also determined with four monomers (chains A–D) in the asymmetric unit in space group P12₁1 at 1.89 Å resolution with the entire polypeptide chain well resolved in the electron density. MTCF is clearly resolved in the electron density map (Figure 7 and SI-1). Inhibitor binding does not cause major conformational changes or loop movements, as superposition of apo-HpAFLDA and MTCF-bound (chain-A) structures gives a root-mean-square deviation (RMSD) of 0.256 Å. The solvent accessible surface area of the dimer interface with MTCF bound was 1239 Ų, greater than that of apo-HpAFLDA.

The structure of HpAFLDA was compared to those of AFLDA from Nitratiruptor and MTADA from P. aeruginosa (PaMTADA).^21,39 Human adenosine deaminase (PDB entry 3IAR) shares only 11.9% sequence identity with HpAFLDA and was not included in the comparison. HpAFLDA shares 37.2% and 25.0% sequence identity with the AFLDA from Nitratiruptor and MTADA from P. aeruginosa, respectively. Upon comparison of HpAFLDA and Nitratiruptor AFLDA (PDB entry 3V7P), the RMSD for Cα was 0.854 Å, suggesting nearly identical secondary structures. The Leu12–Glu18 and Lys222–Thr239 loops of HpAFLDA differ slightly from those of Nitratiruptor AFLDA (Figure 8.). Comparison of HpAFLDA with PaMTADA (PDB entry 4GBD) gives a RMSD value of 1.924 Å, and significant differences between HpAFLDA and PaMTADA structural folds. The substrate binding loop (Leu146–Ala160 of HpAFLDA) is open but a closed conformation in PaMTADA consistent with structural specificity for the smaller substrate (MTA). The helix covering the catalytic site of HpAFLDA (Lys222–Thr239) is missing in PaMTADA (Figure 8). The amino acid residues involved in metal coordination are conserved in all three structures.

Figure 8.

Structural comparisons of HpAFLDA with orthologues. (A) Stereoview of the structural comparison of HpAFLDA (blue, PDB entry 7LKK) with Nitratiruptor AFLDA (cyan, PDB entry 3V7P). The minor structural differences between Leu12–Glu18 and Lys222–Thr239 loops are denoted with the black and red stars, respectively. (B) Structural comparison (stereoview) of HpAFLDA (blue, PDB entry 7LKK) with PaMTADA (gray, PDB entry 4GBD). The open conformation of the HpAFLDA substrate binding loop from residue Leu146 to Ala160 is highlighted with a black star. The catalytic sitecovering helix of HpAFLDA from Lys222 to Thr239 is highlighted with a red star.

Catalytic Site and MTCF Binding to HpAFLDA.

The catalytic site of HpAFLDA is located deep in the monomer (Figure 7). In apo-HpAFLDA, only a few ordered water molecules are observed (Figure SI-2). The Me²⁺ binding site is coordinated with NE2(His65), NE2(His67), NE2(His207), OD2(Asp314), and O8 of the MTCF inhibitor (Figures 9 and 10). The residues interacting with the MTCF inhibitor with 4 Å are His65, His67, Phe70, Trp85, Asn124, Glu144, His180, His207, Glu210, His261, Leu292, Asp314, and Ser317 (Figure 9). With respect to the 5′-alkyl group, a hydrophobic packet is formed by Leu86, Val89, Leu90, Leu146, Ser148, Tyr183, Ser184, Phe228, Tyr229, and Leu233 (Figure 9). This hydrophobic pocket provides adequate space for binding the 5′-group of the AFL substrate (Figure 1). The adenine mimic of the inhibitor is sandwiched between His207 and Trp85 with Phe70 providing a hydrophobic interaction with the β-ribosyl face of the inhibitor. Phe70 is disordered in the apo-HpAFLDA structure. N6 of MTCF is hydrogen bonded with OD1 and OD2 of Glu201. The OG(Ser317) and a water molecule have hydrogen bond interactions with N1 of MTCF. The ribosyl O2 and O3 atoms of MTCF are hydrogen bonded with OE1(Glu144) and OD1(Asn124). A solvent ethylene glycol is found in the hydrophobic packet beyond the 5′-alkyl thio group of the inhibitor (Figures 9 and 10).

Figure 9.

Binding of MTCF to HpAFLDA (stereoviews). (A) Active site of HpAFLDA in complex with MTCF (yellow). The amino acid residues interacting with MTCF are indicated. The binding of metal (brown), water (red), and ethylene glycol (EG, light blue) is also highlighted. Selected hydrogen bond interactions and metal coordinations are shown as black dotted lines (see also Figure 8). (B) Amino acids forming the hydrophobic pocket extending beyond the 5′-methylthio group of MTCF. This pocket provides adequate space for the substrate (AFL) to bind to the enzyme. (C) Surface representation of the HpAFLDA binding site bound with the MTCF. The hydrophobic pocket extends from the 5′-methylthio group of MTCF toward the viewer and the surface of the protein.

Figure 10.

Two-dimensional active site distance map comparing (A) PaMTADA and (B) HpAFLDA. The hydrogen bonds (dotted black lines) and metal coordination (dotted red lines) are indicated. The distances are in angstroms.

A comparison of the catalytic site contacts to MTCF in PaMTADA and HpAFLDA shows conservation for metal binding. Adenine interactions differ with residues Phe70, Phe82, Val89, Leu90, and Ser317 in HpAFLDA being replaced by Met77, Leu93, Leu89, Trp98, and Ala312, respectively, in PaMTADA. The Ser317 hydrogen bond interaction with N1 of MTCF in HpAFLDA is replaced with a water molecule in PaMTADA.²¹ The substrate binding pocket of PaMTADA is smaller than that in HpAFLDA to allow 5′-substituent binding with hydrophobic interactions. The binding conformation of ribose and the 5′-substituent for MTCF also differ in PaMTADA such that MTCF is a more potent inhibitor for PaMTADA.

Structure–Inhibition Relationship of MTCF.

MTCF is a slow-onset tight-binding inhibitor of HpAFLDA with a K_i* of 63 ± 4 nM. However, it binds 13000-fold tighter to PaMTADA with a K_i* of 4.8 ± 0.5 pM.²¹ The active site of PaMTADA is designed to accommodate a 5′-methylthio group from its 5′-methylthioadenosine substrate (Figure 8). The catalytic pocket of HpAFLDA is much larger to accommodate the 5′-substituent of AFL, leaving an unfavorable water–hydrophobic site interaction with this inhibitor (Figure 9). Larger hydrophobic groups attached to the 5′-alkyl of MTCF have the potential to increase the binding affinity and provide an ongoing synthetic chemistry challenge.

CONCLUSIONS

Chronic H. pylori infections cause gastric and duodenal ulcers and are considered a strong risk factor for developing gastric adenocarcinoma.⁴⁶ The current standard-of-care treatment for H. pylori infection fails at a rate of 30–40% due to resistance against commonly used antibiotics.¹⁶ Thus, there is growing interest in developing narrow-spectrum therapeutics for H. pylori infection. Menaquinone is an essential component in the electron transport chain in bacteria that adopt anaerobic respiration.⁴⁷ MTAN-catalyzed AFL hydrolysis has been reported to be a drug target for anti-H. pylori therapeutic development.^19,48,49 Here, we have characterized jhp0252 to be an AFL deaminase in H. pylori J99. HpAFLDA catalyzes AFL deamination to form FL, and the AFL → FL process is a biologically significant, possibly regulatory, step for growth of H. pylori. MTCF is a nanomolar inhibitor of HpAFLDA and is functional in growing cells. MTCF demonstrated an anti-H. pylori growth activity in bacterial culture, partially reversed by FL, highlighting the importance of FL production to bacterial survival. The crystal structure of apo-HpAFLDA and in complex with MTCF revealed favorable inhibitor binding interactions at the catalytic site. A large hydrophobic pocket in the crystal structure is present extending from the 5′-methylthio group of the inhibitor. The binding mode of MTCF in the HpAFLDA binding site provides the basis for designing the next-generation inhibitors.

The catalytic specificity for AFL implies that the primary biological function of HpAFLDA resides in the futalosine pathway. However, an active AFL → FL → DHFL pathway has not been established in H. pylori. No whole cell activity for the FL → DHFL step could be detected. The current literature favors HpMTAN in the primary metabolic path for DHFL production in H. pylori. Transition state analogues of HpMTAN inhibit H. pylori growth with IC₉₀ values as low as 6 ng/mL, supporting the essential role of HpMTAN in AFL hydrolysis.^7,19,48 Kinetic and metabolic analysis of HpAFLDA demonstrates its capability of deaminating AFL in vitro and in vivo. The formation of intracellular FL from AFL is readily detected in viable cell assays, establishing the entry of AFL into cells and release of FL following AFL deamination. Intracellular AFL and FL levels changed substantially when HpAFLDA was inhibited by MTCF, further establishing AFL deamination as a physiological reaction in H. pylori. H. pylori cell growth inhibition by MTCF and cell growth rescue by supplementation with FL support a role for HpAFLDA in menaquinone synthesis. A significant decrease in DHFL and menaquinone-6 concentrations was observed when cells were cultured under HpAFLDA inhibition. The missing link in establishing a role for HpAFLDA in the pathway is the lack of a measurable catalytic activity for FL conversion to DHFL in H. pylori, possibly suggesting an activity needed for other environmental conditions. A potential feedback regulation involving HpAFLDA formation of FL did not identify growth modulation via MK biosynthesis. The possibility that formation of FL by HpAFLDA plays a regulatory role in other pathways essential for growth of H. pylori remains interesting.

HpAFLDA catalyzes the deamination of MTA and 5′dAdo in addition to AFL. These metabolites are products of the SAM-dependent polyamine pathway and of radical SAM reactions, respectively,⁵⁰ allowing the reaction products to enter salvage pathways.⁵¹ MTI, converted from MTA, can undergo N-ribosyl phosphorolysis to yield hypoxanthine and 5-methylthioribose 1-phosphate, a methionine salvage precursor in P. aeruginosa.²¹ Deamination of 5′dAdo produces 5′-deoxyinosine, a precursor for purine recycling in P. falciparum.^23,52 The MTA metabolite is reported to cause potent product inhibition on polyamine biosynthesis,⁵³ and accumulation of 5′dAdo inhibits radical SAM reactions, such as biotin synthesis.⁵⁴ Being capable of metabolizing MTA and 5′dAdo efficiently suggests that HpAFLDA has a function in recycling end products in several pathways in addition to a potential regulatory role in menaquinone biosynthesis.