Chemicals and reagents

Commercial chemicals were obtained from Sigma-Aldrich (Sigma-Aldrich Co., St Louis, USA), Sangon (Sangon Biotech Co., Ltd., Shanghai, China), Weikeqi (Weikeqi Biological Technology Co., Ltd., Sichuan, China), and Sinopharm (Sinopharm Chemical Reagent Co., Ltd., Shanghai, China). BBR chloride was dissolved in DMSO to obtain stock solution (60 mM). THF was dissolved in 2-mercaptoethanol (1.0 M) to generate stock solution (10 mM, neutralized with 1 M KOH). ATP was dissolved in Tris–HCl (100 mM, pH = 7.9) to obtain stock solution (10 mM). PBS was prepared as follows: 29.22 g (500 mM) NaCl, 0.20 g (2.7 mM) KCl, 1.44 g (10.1 mM) Na2HPO4, and 0.24 g (1.8 mM) KH2PO4 were dissolved in 1 L of double distilled water (pH = 7.0).

Bacterial strains and culture conditions

E. coli DH5α and BL21 (DE3) as well as their derived strains were grown in LB (lysogeny broth) medium [37], with the addition of kanamycin (100 μg/mL) when needed. The E. coli DH5α and E. coli BL21 (DE3) strains were used for gene cloning and protein expression, respectively.

Peptostreptococcus anaerobius ATCC27337, Fusobacterium nucleatum ATCC10953, and Clostridium symbiosum ATCC14940 were grown in the brain heart infusion (BHI) broth supplemented with hemin, K2HPO4, vitamin K1, and l-cysteine [38]. Streptococcus faecalis ATCC19433, Lactococcus lactis DSM20481, Lactobacillus acidophilus ATCC4356, Lactobacillus reuteri DSM17938, Lactobacillus plantarum DSM13171, Lactobacillus fermentum CGMCC1.1880, and Lactobacillus bulgaricus BD15 were grown in the MRS medium [39]. Anaerobic bacteria are cultured in an anaerobic chamber (Whitley A35 Anaerobic Workstation, Don Whitley Scientific Limited, Bingley, West Yorkshire, UK).

Synthesis of the BBR-biotinylated probe

The BBR-biotinylated probe (BBP, 4), consisting of BBR, biotin, and C-9 hydrophilic linker, was synthesized as described previously (Fig. S2) [24]. In brief, the compound 1 (1.006 mmol BBR) dissolved in DMF (10 mL) was mixed with propargyl bromide (2.415 mmol). The mixture was stirred and then recrystallized from diethyl ether to generate compound 2 as a brown solid. Next, in the presence of CuSO4 and sodium ascorbate, BBP (4) was obtained from compound 2 and azide biotin (3) through click chemical reaction. Finally, the solvent was concentrated, and the residues were purified to obtain BBP (C40H50N7O9S+).

Identification of BBR-binding proteins

P. anaerobius ATCC27337 was grown in the BHIS medium under anaerobic conditions. The cells were collected by centrifugation (8000 × g; 5 min; 4 °C) at an OD600 of 1.0. Cells were resuspended in lysis buffer (PBS + 1 mM PMSF) and then lysed using a cell disruptor (French Press, Constant Systems Limited, Northants, UK). The lysate was centrifugated at 15,000 × g for 60 min at 4 °C. The supernatant (containing 200 μg of total protein) was incubated with BBP (100 μM) for 2 h at room temperature. The mixture was added to streptavidin agarose resin and then incubated for 0.5 h at room temperature on a rotating apparatus. After five times of washing with lysis buffer (PBS + 0.05% Tween 20 + 1% PEG, pH 7.0) and centrifugation (1000 × g at 25 °C for 10 s) to remove unbound components, the BBP-bound proteins in streptavidin agarose resin were loaded on a 12% SDS-PAGE gel and then stained by Coomassie Blue. The specific bands of interest were cut out and trypsinized overnight. The extracted peptides were identified by LC–MS (Orbitrap™ mass analyzer, Thermo Fisher Scientific, Vilnius, Lithuania). Here, the following controls were set: (i) the aforementioned supernatant (containing 200 μg of total protein) was mixed with BBP (100 μM) and free BBR (1 mM); (ii) the supernatant (containing 200 μg of total protein) was mixed with biotin (100 μM).

To validate BBR-binding protein, the recombinant target proteins (3 μM) were prepared and subjected to the same manipulations as described above. The following controls were set: (i) 3 μM recombinant target protein mixed with BBP (100 μM) and free BBR (1 mM); (ii) 3 μM recombinant target protein mixed with biotin (100 μM).

Gel digestion and liquid chromatography-mass spectrometry analysis

The gel bands of interest were cut out from SDS-PAGE, destained with 30% ACN/100 mM NH4HCO3, and dried in a vacuum centrifuge. The in-gel proteins were reduced with dithiothreitol (10 mM DTT/100 mM NH4HCO3) for 30 min at 56 °C and then alkylated with iodoacetamide (200 mM IAA/100 mM NH4HCO3) in dark at room temperature for 30 min. Next, gel pieces were rinsed with 100 mM NH4HCO3 and ACN and then digested overnight with trypsin (12.5 ng/μl) in 25 mM NH4HCO3. The peptides were extracted with 60% ACN (containing 0.1% TFA), and pooled and dried completely by a vacuum centrifuge.

The LC–MS analysis was carried out on a Q Exactive mass spectrometer combined with an Easy HPLC system (Thermo Scientific, Vilnius, Lithuania). Each sample was loaded onto a reverse phase trap column linked to a C18-reversed phase analytical column and eluted with a flow rate of 0.3 μl/min for 30 min. The mobile phase A and B for HPLC separation were 0.1% formic acid in deionized water and 84% acetonitrile, respectively. The chromatography gradient was set up with the following linear increase: 5% to 35% of B within 22 min; 35% to 100% of B within 5 min; 100% of B for the last 3 min.

The mass spectrometer was operated in a positive ion mode with the following parameters: MS spectra in the range of 300–1800 m/z. Survey scans were acquired at a resolution of 70,000 at m/z 200, isolation width was 2 m/z, microscans to 1, and maximum inject time to 50 ms. Normalized collision energy was 27 eV, and the underfill ratio was defined as 0.1%. The instrument was run with peptide recognition mode enabled.

The Mascot 2.2 search engine (Matrix Science Ltd., London, UK) was used for protein identification. Searches of the MS data were performed based on the P. anaerobius UniProt database.

Plasmid construction and site-directed mutagenesis

All the primers and plasmids used in this study were listed in Tables S1 and S2 respectively. The vector of pET28a-PaFtfL for expressing the ftfl gene from P. anaerobius ATCC27337 was constructed as follows. In brief, the DNA fragment of Paftfl was obtained by PCR amplification using the primers Paftfl-F/R and the genomic DNA of P. anaerobius ATCC27337 as the template. And the DNA fragment was then inserted into the pET28a plasmid (digested with BamHI and NdeI) using a ClonExpress MultiS One Step cloning kit (Vazyme, Nanjing, China). The plasmids for expressing the other ftfl genes were constructed via the same steps.

The vectors pET28a-PaFtfLY229A for expressing the mutated ftfl gene of P. anaerobius ATCC27337 was constructed as follows. In brief, the DNA fragment of the mutated ftfl gene was obtained by PCR amplification using the vector pET28a-PaFtfL as the template and the primers PaftflY229A-F/R. The PCR product was treated with DPN1 to remove the original methylated plasmids and then transformed into E. coli DH5α. The vectors expressing the other mutated ftfl genes were constructed via the same steps.

Gene expression and protein purification

The proteins used in this study were expressed in the E. coli BL21 (DE3) strain. Protein expression was induced with 0.3 mM IPTG at 16 °C when the cell density reached OD600 of 0.6 ~ 0.8. After 16 h of induction, cells were collected by centrifugation (12,000 × g for 5 min at 4 °C) and then resuspended in a lysis buffer (100 mM Tris–HCl, pH 7.9, 100 mM NaCl) with 1 mM PMSF. Cells were broken by using a cell disruptor (French Press, Constant Systems Limited, UK), and the lysate was centrifuged (15,000 × g for 60 min at 4 °C). The supernatant was loaded onto a Ni2+ Sepharose™ 6 fast flow agarose resin (GE Healthcare, Waukesha, WI, USA). The resin was then washed with lysis buffer (containing 15 mM, 30 mM imidazole) for the removal of non-target proteins, and eluted with lysis buffer (containing 300 mM imidazole). The eluted fractions were freed from imidazole by an Amicon Ultra 15 Centrifugal Filter (Millipore Billerica MA) with lysis buffer. Finally, the purified proteins were identified by SDS-PAGE and then concentrated (Amicon Ultra-4, Millipore, USA) and stored at − 80 °C.

Bio-layer interferometry (BLI) assay

The interactions between BBR and the PaFtfL (including its variants), hsMTHFD1, and hsMTHFD1L proteins were determined by using the ForteBio Octet RED 96 platform (Forte Bio, San Francisco, USA). A streptavidin matrix-coated sensor chip (SA chip) was firstly equilibrated with buffer A (lysis buffer with 0.05% Tween 20) followed with the immobilization BBP (200 μM) on the SA chip. Next, proteins with increasing concentrations (0.23, 0.91, 3.63, 14.50, and 58.00 μM) were passed on to the chip for the measurement of changes in response unit (nm). The program comprises the stabilization of the baseline with the buffer A for 1 min, 6 min incubation of the BBP with the SA chip for immobilization, stabilizing the baseline again for 3 min, association enabling interaction between proteins and compounds for 4 min, and dissociation for 4 min followed by a regeneration step. The interaction between BBR and the PaEF4 protein (58 μM) was determined by using the same method. Raw data were pre-processed, analyzed, and fitted using the 1:1 binding model in the manufacturer’s software (9.0, Pall ForteBio Corp, Menlo Park, CA, USA) to generate kinetic parameters.

Determination of inhibitory activity assay of BBR to FtfL

The activities of the FtfL enzymes were measured according to the protocol described previously [26].

The assay of the BBR’s inhibition on FtfLs was carried out as follows. In brief, 20 nM FtfLs (100 nM for LpFtfL because of its low activity) was preincubated with BBR (5, 10, 50, 100, and 200 μM) at 30 °C for 5 min. Then, the reaction was initiated by adding substrates (0.1 mM tetrahydrofolate, 2 mM MgCl2, 0.05 mM ATP, and 8 mM sodium formate). After 2 min of reaction, the reaction was terminated by adding HCl (0.36 M, 2 × volume of the reaction mixture).

The assay of the BBR’s inhibition on hsMTHFD1/hsMTHFD1L was performed as follows. Briefly, 400 nM hsMTHFD1 or hsMTHFD1L was mixed with BBR (200 or 400 μM) and preincubated at 30 °C for 5 min. Then, the reaction was initiated by adding substrates (2 mM tetrahydrofolate, 10 mM MgCl2, 40 mM sodium formate, and 5 mM ATP). After 2 min of incubation, the reaction was terminated by adding HCl (0.36 M, 2 × volume of the reaction mixture).

The product was detected with a maximum absorbance at 350 nm (ε = 24.9 mM−1 cm−1) using FLUOstar OPTIMA (BMG LABTECH, Offenburgh, Germany). The reaction was performed at 30 °C. The control reaction was performed by replacing BBR with DMSO. The half maximal inhibitory concentration (IC50) of BBR was adopted to represent the inhibitory efficiency of BBR on different FtfLs. Data analysis was performed in GraphPad Prism 7.0.

The enzyme kinetic constants of PaFtfL were determined using two kinds of reaction mixtures (200 μl): (i) 8 mM sodium formate, 0.05 mM ATP, 2 mM MgCl2, 100 nM PaFtfL, and different amounts of THF (32.5, 65, 130, 325, 650, and 1300 μM); (ii) 8 mM sodium formate, 0.04 mM THF, 2 mM MgCl2, 100 nM PaFtfL, and ATP (10, 20, 50, 100, and 200 μM). The reaction mixture was incubated at 30 °C for 2 min. Then, the reaction was terminated by adding HCl (0.36 M, 2 × volume of the reaction mixture). The inhibition constants (Ki) of BBR on PaFtfL based on THF or ATP were determined by measuring the apparent Km with the addition of 40 μM BBR into the reaction mixture.

Crystallization and structure determination

The crystals of PaFtfL apo were obtained by using a sitting drop vapor diffusion method at 22 °C with 1 μl drops containing a 1:1 mixture of crystallization buffer (15% w/v PEG 3350, 0.15 M cesium chloride) and 10 mg/ml protein. The crystals were grown for 7 days and then applied for X-ray diffraction data collection. Before frozen in liquid nitrogen, crystals were stabilized in cryoprotectant (0.15 M cesium chloride, 17.5% v/v ethylene glycol, and 15% w/v PEG 3350).

PaFtfL-ATP crystals were obtained by soaking PaFtfL apo crystals in crystallization buffer (200 mM potassium formate, 20% w/v PEG 3350) containing 1 mM ATP. Before frozen in liquid nitrogen, crystals were stabilized in cryoprotectant (200 mM potassium formate, 20% w/v PEG 3350, and 12.5% v/v 1,2-butanediol).

For PaFtfL-BBR crystals preparation, the PaFtfL protein in 10 mg/ml was mixed with 1 mM BBR (stock in 100 mM, 100% DMSO) and then incubated at 25 °C for 1 h. The crystals were obtained by a sitting drop vapor diffusion method at 22 °C with 1 μl drops containing a 1:1 mixture of 10 mg/ml PaFtfL-BBR mixture and crystallization buffer (2.0 M ammonium sulfate, 0.1 M HEPES pH 7.5), The crystals were grown for 7 days before X-ray diffraction data collection. Before frozen in liquid nitrogen, crystals were stabilized in cryoprotectant (2.0 M ammonium sulfate, 0.1 M HEPES pH 7.5, and 3 M l-proline). Data were collected at Shanghai Synchrotron Radiation Facility (SSRF) beamlines 02U1 and then were processed using XDS [40] for PaFtfL apo and PaFtfL-BBR datasets and xia2-3dii for PaFtfL-ATP dataset. The molecular replacement was performed by MorDa [41] at CCP4 online server (https://ccp4online.ccp4.ac.uk/). The model building and refinement were performed in Coot [42] and Phenix [43].

Phylogenetic analysis of FtfL homologs

The amino acid sequences of FtfL homologs were obtained by using the TBLASTN search. The obtained protein was aligned with Clustal W software. The alignment was visualized with MEGA 7 program [44]. The maximum likelihood phylogenetic tree was generated from this alignment in MEGA7. Finally, annotation was made manually.

Determination of FtfL abundance using healthy human metagenomic data

All the amino acid sequences from homologous enzymes in the phylogenetic tree were used to generate the SSNs [45], using the EFI-EST webtool (http://efi.igb.illinois.edu/efi-est/). Then, the network was generated with initial edge values of 215 as previously reported [46]. The abundance of FtfLs in human metagenomes was obtained by using ShortBRED.