Inhibition of bacterial FMN transferase: A potential avenue for countering antimicrobial resistance

Abstract

Antibiotic resistance is a challenge for the control of bacterial infections. In an effort to explore unconventional avenues for antibacterial drug development, we focused on the FMN‐transferase activity of the enzyme Ftp from the syphilis spirochete, Treponema pallidum (Ftp_Tp). This enzyme, which is only found in prokaryotes and trypanosomatids, post‐translationally modifies proteins in the periplasm, covalently linking FMN (from FAD) to proteins that typically are important for establishing an essential electrochemical gradient across the cytoplasmic membrane. As such, Ftp inhibitors potentially represent a new class of antimicrobials. Previously, we showed that AMP is both a product of the Ftp_tp‐catalyzed reaction and an inhibitor of the enzyme. As a preliminary step in exploiting this property to develop a novel Ftp_Tp inhibitor, we have used structural and solution studies to examine the inhibitory and enzyme‐binding properties of several adenine‐based nucleosides, with particular focus on the 2‐position of the purine ring. Implications for future drug design are discussed.

Short abstract

PDB Code(s): 7MGT;

1. INTRODUCTION

We recently described a periplasmic post‐translational process (FMNylation) that uses FAD to covalently link FMN to target proteins in Treponema pallidum, the syphilis spirochete. ¹ , ² This pathway likely works in concert with a flavin‐uptake system ³ to enable a “flavin‐centric” metabolic lifestyle for this bacterial pathogen. ¹ , ² The organism's flavin‐trafficking protein (Ftp_Tp) ostensibly plays a key role in FMNylation and flavin homeostasis in the periplasm. An important feature of Ftp_Tp is its dual Mg²⁺‐dependent enzymatic activities: FAD hydrolysis (to FMN and AMP) and FMN transfer, catalyzing the post‐translational FMNylation of flavoproteins. ¹ , ⁴ , ⁵ In this latter reaction, the side‐chain hydroxyl moiety of a conserved threonine residue of the FMNylation motif in an acceptor protein serves as the nucleophile that attacks a phosphorus atom in FAD, ultimately resulting in a covalent link (FMNylation) between FMN and the target protein via a phosphoester–threonyl bond ¹ , ⁶ ; AMP is the other product of FAD scission.

FMNylation potentially is a vulnerable “Achilles heel” in pathogenic bacteria because this reaction is necessary for the proper formation of flavin‐ and quinone‐based redox‐driven Na⁺ pumps (NQR, Na⁺‐translocating NADH:quinone oxidoreductase and RNF, Rhodobacter nitrogen fixation) that often provide the only ion‐motive (chemiosmotic gradient) electron‐transport chain in these organisms. ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ Additionally, Ftp homologs are not present in higher eukaryotes. ¹² It is now appreciated that Ftp's FMN‐transferase activity is required for the periplasmic FMNylation and electron‐transfer functions of subunits (e.g., NqrC and RnfG) of these flavin‐based cytoplasmic membrane‐bound NQR and/or RNF redox complexes, providing a link between those periplasmic functions and the redox reactions that occur within the bacterial cytoplasm. ¹ , ¹³ , ¹⁴ , ¹⁵ Disruption of the chemiosmotic gradient generated by these redox complexes has the potential to arrest essential bacterial functions such as ATP synthesis, flagellar rotation, and nutrient transport. ¹¹ , ¹⁶ , ¹⁷ Indeed, interruption of the Ftp gene in Listeria monocytogenes limited the bacterium's ability to grow on a carbon source that requires extracellular electron transport for utilization. ¹⁸ Thus, inhibitors of FMNylation (i.e., of Ftp) have the potential to curtail bacterial growth and therefore could lead to new antimicrobials to combat antibiotic‐resistant bacteria, which are a growing menace. For example, in the two co‐circulating strains of syphilis‐causing T. pallidum, single‐nucleotide polymorphisms associated with macrolide resistance are highly prevalent (71% in the “SS14” strain and 35% in the “Nichols” strain). ¹⁹ The structural, biochemical, and biophysical studies described herein target the adenosine‐binding portion of the Ftp_Tp active site as a first step in exploring the feasibility of using Ftp inhibitors as antimicrobials.

2. RESULTS

2.1. The crystal structure of Ftp_Tp bound to 2‐Cl‐ADP

Previous work established that Ftp_Tp undergoes single‐turnover kinetics due to the inhibitory effect of the product, AMP. ¹ , ⁴ Also, crystal structures of FAD and adenine‐based nucleotides in the active site of Ftp_Tp indicated that the protein might accommodate exocyclic groups on the 2‐ and 3‐positions of the purine base. ⁴ The availability of ADP analogs with modifications at the 2‐position spurred us to co‐crystallize Ftp_Tp with 2‐Chloroadenosine‐5′‐O‐diphosphate (2‐Cl‐ADP), and we determined the structure at a resolution of 1.54 Å (Table 1). The overall structure of the protein is essentially the same as previously reported (Figures 1, S1, and S2, Supporting Information). ⁴ The root‐mean‐square deviation of the 319 C_α atoms that are comparable between the current structure and a structure with ADP bound was 0.3 Å.

TABLE 1.

Data collection and refinement statistics

Parameter	Value
PDB accession no.	7MGT
Data collection
Space group	C2
Unit cell dimensions (Å)
a	116.5
b	47.1
c	57.5
α (°)	90
β (°)	102.3
γ (°)	90
Resolution (Å)	35.8–1.54 (1.40–1.54) a
Completeness (%)	97.0 (97.6)
Multiplicity	6.8 (5.9)
Unique reflections	43,888 (2,200)
R merge b	0.039 (0.41)
<I>/σ I	40.8 (4.2)
Wilson B (Å2)	12.1
Refinement
Resolution (Å)	38.5–1.54
No. residues	319
No. nonsolvent atoms	34
No. solvent atoms	125
Maximum‐likelihood coordinate error (Å)	0.16
Average B‐factors
Protein (Å2)	17.7
Solvent (Å2)	23.0
R‐values
R work c	0.150
R free d	0.198
Ramachandran statistics
Outliers (%)	0.0
Most favored region (%)	98.4
RMS deviations
Bonds (Å)	0.006
Angles (°)	1.0

^a Numbers in the parentheses are reported for the highest‐resolution shell of reflections.

^b $R_{\text{merge}}={\sum }_{\mathit{hkl}}{\sum }_i\left|I_{h,i}-⟨I_h⟩\right|/{\sum }_{\mathit{hkl}}{\sum }_i{I}_{h,i}$, where the outer sum (hkl) is over the unique reflections and the inner sum (i) is over the set of independent observations of each unique reflection.

^c $R_{\text{work}}={\sum }_{\mathit{hkl}}\left|\left|F_o\right|-\left|F_c\right|\right|/{\sum }_{\mathit{hkl}}\left|F_o\right|$, where F _o and F _c are observed and calculated structure factor amplitudes, respectively.

^d R _free is calculated using the same formula as R _work, but the set hkl is a randomly selected subset (5%) of the total structure factors that are never used in refinement.

FIGURE 1.

Structural aspects of 2‐Cl‐ADP binding to Ftp_Tp. (a) The overall structure of Ftp_Tp bound to 2‐Cl‐ADP. β‐strands are colored tan, α‐helices green, and regions of irregular secondary structure light blue. The atoms of bound 2‐Cl‐ADP are shown as spheres, with carbon atoms colored gray, oxygens red, nitrogens blue, phosphoruses pink, and the chlorine green. (b) Adenine and ribose contacts. Protein carbons are colored tan. The black dashes represent putative hydrogen bonds. An omit map (mF _o  – DF _c ) is shown contoured at the 3σ level

The 2‐Cl‐ADP molecule bound almost exactly as ADP in a previous structure, ⁴ and this binding configuration overlaps with the FAD‐binding site in Ftp_Tp (Figures 1 and S2). Two Mg²⁺ ions coordinated the phosphate moieties of the bound 2‐Cl‐ADP, which were also contacted by several amino acid side‐chains in the active site (K165, S240, D284, and T288; see Figure S3). Only one direct contact between the ribose moiety and the protein was made (via D159; Figure 1). By contrast, hydrogen bonds between the protein and the adenine ring system were made by main‐chain atoms (from L258, T96, and N98). The exocyclic chlorine atom of the inhibitor was nestled in a hydrophobic pocket between F97 and P99. This chlorine atom also appears to make a hydrogen bond with a water molecule that is 3.4 Å away (Wat 115); notably, this water molecule was not modeled in any previous Ftp_Tp structure, and thus its binding appears to be correlated with the bound nucleotide having an appropriate exocyclic group at the 2‐position. By contrast, a nearby water atom proximate to N3 of the purine base (Wat 114) is observed in most Ftp_Tp structures. ⁴

2.2. Inhibition of the FMN‐transferase activity of Ftp_Tp

The co‐crystallization of 2‐Cl‐ADP and Ftp_Tp suggested that adenine‐based nucleotides might serve as competitive inhibitors for FMN‐transferase activity. We therefore studied Ftp_Tp's FMN‐transferase activity in the presence of five adenine‐based nucleosides and nucleotides: three naturally occurring ones (Adenosine [ADO], AMP, and ADP), and two modified at the 2‐position of the purine ring (2‐Methylthioadenosine‐5′‐O‐diphosphate [2‐MeS‐ADP] and 2‐Cl‐ADP) (Figure 2a). When included at 1 mM in the reaction along with the TP0171 FMNylation target/substrate, ¹ ADP and 2‐Cl‐ADP effectively halted FMN‐transferase activity (Figure 2b). By contrast, 2‐MeS‐ADP had a minimal effect. Adenosine (ADO) showed evidence of weak inhibition, and AMP, one of the products of the FMN‐transferase activity, evinced a more discernible diminution of activity. The observed inhibition pattern is consistent with our previous findings that both AMP and ADP also inhibit Ftp_Tp. ¹ , ⁴

FIGURE 2.

ADP and analog structures and Ftp_Tp inhibition patterns. (a) Structures of ADP and two analogs. (b) Inhibition of the FMN‐transferase activity of Ftp_Tp by various compounds. “Specific Activity” is defined in section 4, and all activities have been normalized to the “FAD” value

2.3. Direct binding of 2‐Cl‐ADP to Ftp_Tp

The assay employed above requires high enzyme and substrate concentrations and is thus not suitable for quantitative enzymology, for example, for the determination of K _i . However, it was possible to explore the thermodynamics of the binding of adenine‐based nucleosides to Ftp_Tp using microscale thermophoresis (MST) (Figures 3 and S4). ADP demonstrated a robust, submicromolar apparent equilibrium dissociation constant (K _D,app) for binding to Ftp_Tp (48 nM) (Figure 3b and Table S1). We qualified the K _Ds in this report as “apparent” because there was evidence that some of the Ftp_Tp used in the assay was occupied by an unknown, co‐purified competitor. The 2‐Cl‐ADP inhibitor had a similar, but demonstrably higher K _D,app (106 nM) (Figure 3 and Table S1). Notably, these values are several times lower than typical K _Ds for Ftps binding to FAD ⁴ , ¹⁴ (e.g., 670 nM for FAD binding to the Ftp homolog from Escherichia coli, Ftp_Ec; the K _D of FAD binding to Ftp_Tp is unknown because it readily hydrolyzes this substrate). ADO and AMP had similar K _D,apps (1,380 and 860 nM, respectively), and 2‐MeS‐ADP displayed the highest K _D,app (56 μM), which was 1,000‐fold higher than that of ADP binding to Ftp_Tp. Thus, the binding data comport well with the enzyme‐inhibition studies (Figure 2b).

FIGURE 3.

MST evidence for direct binding of nucleotides to Ftp_Tp. (a) In the upper graph, the respective value of the unliganded T‐Jump of the fluorescently labeled Ftp_Tp has been subtracted from the curves. Markers represent the averages of all included replicates, and error bars show the respective standard deviations. The lines are the respective fits to the data. The lower graph shows the residuals between the data and the fit lines. (b) K _Ds derived from the data in (a). The K _D values are depicted as red diamonds, with the actual values shown above, and the respective 68.3% confidence intervals are shown as gray boxes

3. DISCUSSION

The structural data in this report highlight possible avenues of Ftp_Tp inhibitor improvement (Figure 1). The positioning of the hydrophilic chlorine atom of 2‐Cl‐ADP in a largely hydrophobic environment (Figure 1) could account for this compound's slight diminution in binding compared to ADP (Figure 3 and Table S1). Also, a model of 2‐MeS‐ADP bound in this active site (not shown) suggests that the only way to accommodate the extra bulk of the methyl group would be for it to be oriented toward Wat114 and Wat115, potentially displacing them and consequently disrupting water‐mediated contacts to the protein. Given the nearly complete burial of ADP in the active site, decorating future inhibitors with moieties that could occupy the positions of these waters appears to be a promising avenue for inhibitor improvement via medicinal chemistry.

The inhibitors highlighted in this study were designed to exploit the tight binding of the product AMP to Ftp_Tp; as such, they likely will be effective only against the subset of Ftp enzymes that exhibit single turnover and strong product inhibition. ¹ Examples of pathogens containing such enzymes are T. pallidum, Treponema denticola, Listeria monocytogenes, and Enterococcus faecalis. However, the general strategy of disabling Ftp may be effective against a broad spectrum of pathogenic microbes, as Ftp homologs are found in many bacteria ⁵ , ²⁰ that are, concerningly, obtaining new resistances and are a focus of programs to combat them. ²¹ Thus, Ftp inhibitors could open up a new front in the battle against Ftp‐containing pathogens that have acquired multiple antibiotic resistances, like Neisseria gonorrhoeae, ²² Vibrio cholerae, ²³ and the aforementioned E. faecalis. ²⁴ Proteins for the Nqr apparatus often occur on the same operons as the respective Ftps, and it is therefore likely that inhibition of the Ftps would result in reduced efficiency of the respective Nqr machineries and hence abrogate growth of the bacteria. In this way, FMNylation interventions might also render the affected bacteria more susceptible to other antibiotics, constituting an effective adjunct therapy in combination with other antibiotics in the treatment of infections caused by bacteria possessing multiple drug resistances. ²⁵ , ²⁶ , ²⁷ Finally, attenuating Ftp activity may interfere with the life cycle of the eukaryotic pathogen Trypanosoma brucei, ²⁸ raising the possibility that Ftp inhibitors could prove useful in treating trypanosomiasis.

4. MATERIALS AND METHODS

4.1. Materials

ADP analogs (2‐Cl‐ADP and 2‐MeS‐ADP) and β‐octylglucoside were purchased from Biolog Life Science Institute (Germany) and Dojindo (Japan), respectively. Other reagents were from Sigma–Aldrich and all were of analytical grade.

4.2. Protein preparation and concentration determination

Recombinant TP0171 and flavin‐trafficking protein of T. pallidum (Ftp_Tp) preparations were as previously described. ¹ , ⁴ Protein concentrations were determined in Buffer A (20 mM HEPES, pH 7.5, 0.1 M NaCl, 2 mM β‐octylglucoside) from their deduced extinction coefficients using the ProtParam utility of Expasy. ²⁹ Ftp_Tp contained a mutation at position 96 (A96T) that did not affect the enzymatic activity nor the course of the backbone (Figure S1).

4.3. FMNylation assays

Purified proteins in Buffer A were incubated with the indicated concentrations of exogenous FAD and MgCl₂ in a 50‐μL reaction volume for 1 hr at 30°C. ¹ Approximately 100 μM of FMNylation target protein (TP0171) was incubated in Buffer A containing ~25 μM Ftp_Tp, 4 mM MgCl₂, and 1 mM FAD. When used, inhibitors were added to final concentrations of 1 mM prior to the addition of FAD. All reactions were performed in duplicate. The reactions were stopped by adding 8 μL of the mixtures to 100 μL of 2X SDS‐PAGE sample buffer containing 2‐β‐mercaptoethanol, and these mixtures were boiled for 5 min. Then, 20‐μL samples of boiled reaction mixtures were separated on a 4–15% SDS–PAGE gel (Bio‐Rad) and visualized by UV illumination (to detect bound FMN) with a Gel Logic 200 imaging system (Kodak) before Coomassie blue staining. Sometimes, boiled reaction mixtures were frozen until used. The results were quantified using ImageJ; “Activity” was defined in arbitrary units as the fluorescence signal on TP0171 divided by the signal for the protein from the respective lane in the Coomassie‐stained gel. To obtain “Transferase Activity,” the Activity value was divided by the signal for Ftp_Tp in the respective lane from the Coomassie‐stained gel, and “Normalized Transferase Activity” refers to the Transferase Activity values being normalized by the average value of Transferase Activity in the absence of inhibitors.

4.4. Microscale thermophoresis

Ftp_Tp in Buffer A was labeled with Cyanine‐5 maleimide ester using standard procedures. ³⁰ Labeled protein was diluted to 20 nM in MST Buffer (Buffer A+4 mM MgCl₂). Titrations were prepared as described previously. ³⁰ After 90 min of incubation in the dark at room temperature, titration mixtures were transferred to a respective Premium‐coated capillary tube (NanoTemper GmbH). The data were collected using a NanoTemper NT.115 instrument at ambient temperature using the red filter with an LED Power of 20%. The MST protocol was 5 s pre‐IR time, 20 s IR‐on time at 40% power, followed by 5 s of post‐IR time. Three replicates for all inhibitors were collected. The data were analyzed using the “T‐Jump” region of the data traces using PALMIST. ³¹ Any trace exhibiting a nonsmooth approach to the new post‐IR equilibrium fluorescence was excluded from the analysis; remaining T‐Jump values were averaged among the replicates. Binding curves were rendered in GUSSI. ³²

4.5. Crystallization and cryoprotection

Approximately 10 mg/mL of purified Ftp_Tp was reconstituted with 4 mM MgCl₂ plus 1 mM of 2‐Cl‐ADP in Buffer A and incubated at 4°C overnight. Crystals appeared after 3 days in sitting drops at 20°C in 0.1 M magnesium formate dihydrate with 15% (wt/vol) PEG 3350. They were serially transferred to buffers containing 0.1 M magnesium formate dihydrate, 15% PEG 3350, 100 mM NaCl, 20 mM HEPES, pH 7.5, 1 mM 2‐Cl‐ADP, and varying concentrations of ethylene glycol, with a final concentration of 35% (vol/vol) ethylene glycol. After about 1 min in this final solution, the crystals were flash‐cooled in liquid nitrogen and stored until data collection.

4.6. Data collection, structure determination, and refinement

X‐ray diffraction data were collected from the crystals at beamline 19‐ID at the Structural Biology Center of the Advanced Photon Source at Argonne National Laboratories. The crystals possessed the symmetry of space group C2, and the d _min was 1.54 Å. The data were integrated and scaled using HKL2000. ³³ The scaled data were put on an absolute scale and negative intensities removed ³⁴ in PHENIX. ³⁵ Phaser ³⁶ was used to perform molecular replacement; the search model was from the structure of apo Ftp_Tp (accession code 4IFU) ⁴ with all ligands and solvent molecules removed. Electron‐density maps after the successful (log‐likelihood gain = 7,066, translation‐function Z score = 68.4) molecular replacement showed clear density for the 2‐Cl‐ADP and the associated Mg²⁺ ions. These molecules and atoms were placed in the model and iterative rounds of refinement with riding hydrogen atoms in PHENIX using the simulated annealing, positional, and anisotropic B‐factor protocols were performed, with model adjustment as necessary in Coot. ³⁷ Weak density for the chlorine atom of 2‐Cl‐ADP necessitated occupancy refinement for it. Final refinement statistics are in Table 1. Structural figures were rendered in PyMol (Schrödinger, Inc.). The crystal structure was deposited in the Protein Data Bank with accession number 7MGT.

AUTHOR CONTRIBUTIONS

Ranjit Deka: Conceptualization (equal); formal analysis (equal); investigation (equal); project administration (equal). Akanksha Deka: Conceptualization (supporting); investigation (equal). Wei Liu: Investigation (supporting); project administration (supporting). Michael V. Norgard: Conceptualization (equal); funding acquisition (lead); project administration (equal); supervision (equal). Chad Brautigam: Conceptualization (equal); formal analysis (equal); investigation (equal); project administration (equal); supervision (equal).

CONFLICT OF INTEREST

The authors declare no potential conflict of interest.

Supporting information

Appendix S1: Supporting Information.

Deka RK, Deka A, Liu WZ, Norgard MV, Brautigam CA. Inhibition of bacterial FMN transferase: A potential avenue for countering antimicrobial resistance. Protein Science. 2022;31:545–551. 10.1002/pro.4241