A broad-spectrum lasso peptide antibiotic targeting the bacterial ribosome

Abstract

Lasso peptides (biologically active molecules with a distinct structurally constrained knotted fold) are natural products that belong to the class of ribosomally synthesized and post-translationally modified peptides^1-3. Lasso peptides act on several bacterial targets^4,5, but none have been reported to inhibit the ribosome, one of the main targets of antibiotics in the bacterial cell^6,7. Here we report the identification and characterization of the lasso peptide antibiotic lariocidin and its internally cyclized derivative lariocidin B, produced by Paenibacillus sp. M2, which has broad-spectrum activity against a range of bacterial pathogens. We show that lariocidins inhibit bacterial growth by binding to the ribosome and interfering with protein synthesis. Structural, genetic and biochemical data show that lariocidins bind at a unique site in the small ribosomal subunit, where they interact with the 16S ribosomal RNA and aminoacyl-tRNA, inhibiting translocation and inducing miscoding. Lariocidin is unaffected by common resistance mechanisms, has a low propensity for generating spontaneous resistance, shows no toxicity to human cells, and has potent in vivo activity in a mouse model of Acinetobacter baumannii infection. Our identification of ribosome-targeting lasso peptides uncovers new routes towards the discovery of alternative protein-synthesis inhibitors and offers a novel chemical scaffold for the development of much-needed antibacterial drugs.

Extended Data Fig. 1 ∣. Paenibacillus sp. M2 strain produces colistin and LAR.

(a) The antibacterial activity of partially fractionated extract from Paenibacillus M2 against A. baumannii C0286 (‘Ab’) and the following E. coli BW25113 strains: wild type (WT), colistin-resistant expressing mcr-1, and the antibiotic hypersusceptible ΔtolC/ΔbamB mutant (ΔT/ΔB). (b) Bioactivity assay of RP-HPLC fractions of the pre-fractionated Paenibacillus sp. M2 extract against A. baumannii C0286 strain, showing the presence of two distinct antibiotics. (c) Liquid chromatography-mass-spectrometry analysis of fractions 7 and 21 (see panel b) shows the presence of LAR and colistin, respectively. The upper panel shows the extracted ion chromatogram, and the bottom panels represent mass spectra of corresponding fractions. LAR, lariocidin; LAR-B, lariocidin B, and LAR-C, lariocidin C.

Extended Data Fig. 2 ∣. LAR does not affect the bacterial cell envelope.

(a) Inner membrane permeabilization assay in E. coli TOP10 cells with pUC19 plasmid (containing the lacZ gene) using membrane-impermeable dye Ortho-nitrophenyl-β-D-galactopyranoside (ONPG). LAR (40 μg/ml) did not facilitate the uptake of ONPG, unlike colistin (5 μg/ml), which creates membrane pores. Data are representative of two independent experiments. (b) Membrane depolarization assay using DiOC₂(3) dye. The fluorescence of the dye quenches when it enters depolarized cells (due to membrane potential disruption). LAR was used at 10xMIC (40 μg/ml). Protonophore, CCCP (20 μM) served as a positive control. Data represent three biological experiments, with error bars indicating SD of three replicates (c) Scanning electron microscopy images of E. coli treated with 10xMIC of LAR, showing no obvious changes in morphology or defects in the cell envelope. The images are representative of two independent samples.

Extended Data Fig. 3 ∣. Synthesis of lariocidin-fluorophore conjugates and confocal microscopy.

(a) Synthesis of lariocidin-fluorophore conjugates via click-chemistry. TEA, triethylamine; PyBOP benzotriazol-1-yloxytripyrrolidinophosphonium hexafluorophosphate; TFA, trifluoroacetic acid; DMSO, dimethylsulfoxide; DCM, dichloromethane. (b) MIC of LAR-fluorophores against E. coli BW25113 strain in MOPS minimal medium. (c, d) Confocal microscopy images of E. coli BW25113 cells treated with LAR-BODIPY (20 μg/ml) (c) or LAR-rhodamine (20 μg/ml) (d) indicating intracellular accumulation of probes. Fm-4-64 was used to stain the membrane, and Hoechst 33342 for DNA visualization. The images are representative of three biological replicates. Scale bar is 5μM.

Extended Data Fig. 4 ∣. Lariocidin utilizes membrane potential to enter the bacterial cytoplasm.

E. coli BW25113 was used in all the experiments. (a) Anaerobic conditions result in an increase of LAR MIC determined in different media. MOPS MM is MOPS minimal medium and MHB is cation-adjusted Mueller-Hinton Broth. (b) Addition of bicarbonate, known to potentiate certain antibiotics like macrolides and aminoglycosides by enhancing the active membrane potential, reduces LAR MIC in the MOPS and MHB media. MIC is also reduced in RPMI medium, which mimics physiological conditions better than MHB. (c) Effect of lower pH (known to decrease the membrane potential) on MIC. The experiment was conducted in MOPS MM. (d) The protonophore, CCCP, which eliminates the membrane potential, protects the cells from the killing action of LAR. cfu were enumerated 1 h after treatment with LAR (40 μg/ml) and/or CCCP (20 μM) in MOPS MM. However, no significant change in MIC of LAR was observed at this concentration of CCCP, suggesting that cells may overcome the effect of CCCP during prolonged growth. Data are plotted and mean±SD of three biological replicates (e) LAR-BODIPY uptake under various treatment conditions using confocal microscopy showing that lowing pH or pretreatment of cells with the protonophore CCCP significantly reduces the accumulation of the LAR-BODIPY fluorescence inside the cells. Fm-4-64 was used to stain the membrane, and Hoechst 33342 for DNA visualization. The images are exemplary of two biological experiments. Scale bar is 5μM.

Extended Data Fig. 5 ∣. LAR-resistance mutations in the 16S rRNA

(a) Characteristics of LAR-resistant mutants selected in E. coli SQ110 ΔtolC. (b) Location of the mutations conferring increased resistance to LAR in the 16S rRNA helices h31 and h34. (c) Odilorhabdins (NOSO-95719) and LAR have different resistance profiles, as evidenced by the lack of LAR MIC changes in most NOSO-95719-resistant strains with point mutations in the 16S rRNA gene. (d) Spatial arrangement of the LAR-resistance mutations in the 16S rRNA shown in the structure of the E. coli small ribosomal subunit (PDB ID: 4YBB).

Extended Data Fig. 6 ∣. Intramolecular H-bonds of LAR and comparison of LAR and LAR-B structures.

(a, b) Structure of ribosome-bound LAR is shown from two opposite sides, highlighting intramolecular H-bonds that stabilize its fold. Carbon atoms are colored yellow, nitrogens are blue, oxygens are red. (c, d) Superposition of the structures of ribosome-bound LAR and LAR-B. Residues whose positions differ between the two isoforms are labeled. Nitrogen and oxygen atoms in the isopeptide bonds are colored blue and red, respectively.

Extended Data Fig. 7 ∣. LAR effectively kills A. baumannii C0286 in vitro and ex vivo.

(a) In vitro time-kill assay in MHB medium showing the bactericidal effect of LAR as denoted by reduction in viable cfu counts. Data are plotted as the mean of three biological replicates with the error bars indicating SD. (b) Cidality of LAR in the ex vivo model. A. baumannii C0286 was inoculated in human blood, and bacterial cfu were enumerated after 4 h treatment with LAR. Data are plotted as mean ±SD of three biological experiments. Significance was determined using one-way ordinary ANOVA with Dunnett's multiple comparisons test. (*P=0.0118; **P=0.0017). (c) MIC of LAR against A. baumannii C0286 in MHB with or without the addition of serum. FBS=fetal bovine serum; HS=human serum (heat inactivated).

Extended Data Fig. 8 ∣. lrc-like biosynthetic gene clusters in various bacterial genomes.

(a) Gene composition of the representative set of lrc-like BGCs from different bacterial phyla. Each gene is represented by an arrow, and the proposed functions of the encoded proteins are listed on the right. The lrc BGC from Paenibacillus sp. M2 is labeled in bold. Note that BGCs from Actinomycetota (including the BGC of triculamin from S. griseocarneus ATCC 29818 (ref. 35)) lack the homologs of lrcB1B2 genes as the encoded precursors contain the unusual C-terminal follower peptide instead of the N-terminal leader peptide typical for other LPs. (b) Phylogenetic tree of lrc-like BGCs (n = 29) built based on the amino acid sequence similarity between LrcC (lasso cyclase) homologs. The lasso cyclase from the biosynthetic gene cluster of the lasso peptide paeninodin served as the outgroup. The alignment of the amino acid sequences of the precursor peptide(s) core parts is shown on the right. The amino acids are colored according to their physico-chemical properties. The consensus sequence and the extent of sequence conservation for each position are shown below. Note that for three triculamin-like peptides from Actinomycetota containing a follower rather than the leader peptide, the actual cleavage site for the follower-peptide (marked with black asterisks) could not be identified unambiguously. RRE – RiPP recognition element, GNAT – Gcn5-Related N-Acetyltransferase, OM – outer membrane.

Extended Data Fig. 9 ∣. Common mechanisms conferring resistance to clinically relevant ribosome-targeting antibiotics do not impact LAR antibacterial activity.

The graph shows the MIC increase of LAR and corresponding control antibiotics (Ab) upon overexpression of the designated resistance determinants in E. coli BW25113 ΔtolCΔbamB. The strain design and details of plasmids are described by Cox et al. The color of the gene name reflects the mechanism of resistance: antibiotic modification/inactivation – black, rRNA modification – orange, ribosome protection – blue. Functions of specific genes are- aad(3′′)(9)= spectinomycin adenyltransferase; apmA, aac(2′)-Ia, aac(6′)-Ib, and aac(2′)-IIa= aminoglycoside N-acetyltransferases; aph(4)-Ia, aph(3′)-IIIa, aph(3′)-IVa, and aph(6)-Ia= aminoglycoside phosphotransferases; sat= streptothricin acetyltransferase; tetX= tetracycline inactivation; vph= viomycin phosphotransferase; cat= chloramphenicol acetyltransferase; linB= lincosamide nucleotidyltransferase; vatD= streptogramins acetyltransferase; vgb= streptogramin B lyase; kamB, npmA, armA and rmtB= 16S rRNA methyltransferases; cfrA and ermC= 23S rRNA methyltransferases. Control antibiotics (Ab)- apramycin (apmA, kamB, npmA), gentamicin (aac(2′)-IIa, armA, rmtB), hygromycin B (aph(4)-Ia), kanamycin (aac(6′)-Ib, aph(3′)-IIIa), kasugamycin (aac(2′)-IIa), ribostamycin (aph(3′)-IVa), streptomycin (aph(6)-Ia), clindamycin (cfrA, linB, ermC), flopristin (vatD), quinupristin (vgb), nourseothricin (sat), oxytetracycline (tetM, tetX), viomycin (vph). For the controls marked with asterisks (*), the actual MIC change value is higher than that of the one presented; this reflects the growth of bacteria in the presence of the antibiotic in the highest concentration tested. Antibiotics targeting the small ribosomal subunit are in the light blue background, large ribosomal subunit – is in the light orange.

Figure 1 ∣. Lariocidin and its biosynthetic gene cluster.

(a) Top, gene composition of the lrc BGC. Bottom, posttranslational modification of the LrcA precursor peptide leads to the production of LAR and its variant(s). The proteins B1 and B2 recognize and cleave the leader peptide, respectively, followed by the cyclization by LrcC. LrcF cleaves the terminal glycine leading to the formation of LAR-B and LAR-C variants. It remains unknown how the second isopeptide bond of LAR-B is formed. (b) Heterologous expression of lrc BGC in S. lividans and analysis of LAR in cell-free supernatant. The panel shows cation-exchange chromatographic analysis of LAR produced in the heterologous host. The curves are adjusted to show all three chromatograms in 3D plane. The major peak around 12.5 min corresponds to LAR as confirmed by mass spectrometry. SL-Lrc: S. lividans transformed with pIJ10257-lrc; M2 control: LAR purified from the native producer Paenibacillus M2; SL-control: S. lividans transformed with the empty vector. (c) LC-MS analysis of LAR purified from the heterologous host shows all three isoforms produced; LAR – m/z of 936.039; LAR-B – m/z of 898.523, and LAR-C – m/z of 907.528. All masses shown are ions corresponding to [M+2H]²⁺. (d) LC-MS analysis of LAR produced by heterologous host from the lrc operon with the lrcF gene deleted shows exclusive production of LAR, but not LAR-B and LAR-C variants, suggesting the role of LrcF in the biosynthesis of these variants.

Figure 2 ∣. LAR exhibits bactericidal activity and targets bacterial protein synthesis.

(a) Reduction in colony forming units (cfu) of E. coli cultures treated with LAR at 10xMIC (40 μg/ml) or cell membrane-targeting lytic antibiotic colistin at 10xMIC (5 μg/ml). (b) Effect of LAR or colistin on the density of exponential E. coli culture. In a and b, data are presented as mean ±SD (standard deviation) of three technical replicates and are representatives of two biological replicates with similar results. (c) A propidium iodide accumulation assay to assess the effect of LAR on cell permeabilization. The y-axis represents relative fluorescence units (RFU) normalized by the initial fluorescence at time 0. Cell-permealizing colistin served as positive control. (d) LAR-BODIPY is found in the cytoplasm of the E. coli cells. Green; LAR-BODIPY fluorescence; red: membrane stain Fm-4-64; blue: DNA stain Hoechst 33342. The image is representative of three experiments. (e, f) Effect of LAR on E. coli protein synthesis in cell-free transcription-translation system programmed with firefly luciferase-encoding plasmid (e) or GFP mRNA (f). (g) of the effect of LAR on protein synthesis in rabbit reticulocyte lysate programmed with luc mRNA. In c, e–g, data points represent the mean±SD of three experiments. (h) Translocation inhibition assay. Adding LAR or a control antibiotic negamycin (NEG), a known inhibitor of translocation, interferes with the movement of mRNA/tRNA complex through the ribosome. (i) Toeprinting analysis of ribosomes stalled by LAR on a model mRNA. The inhibitor of translation initiation retapamulin (RET) served as a control. ‘None’ – no antibiotic control. Black arrowhead indicates ribosomes stalled at the start codon, and open arrowheads those stalled at internal codons. (j) LAR induces miscoding as evidenced by the ability of E. coli cells harboring the lacZ gene with a premature stop codon to produce functional β-galactosidase (visualized as a blue halo bordering the zone of growth inhibition). Miscoding-inducing gentamicin (GEN) and streptomycin (STR) were used as positive controls and chloramphenicol (CHL) as a negative control. h–j, data is representative of two independent experiments with similar results.

Figure 3 ∣. Structures and electron density maps of ribosome-bound LAR and LAR-B.

(a, c) Schematic diagrams of the lasso peptides LAR (a) and LAR-B (c) highlighting their N-terminal residues 1-7 (yellow), branching point at residue 8 (red), and C-terminal residues 9-18 (blue). Lys2 residue of LAR-B forming the second isopeptide bond is colored orange. (b, d) 2Fo-Fc Fourier electron density maps of LAR (b) and LAR-B (d) in complex with the T. thermophilus 70S ribosome (blue mesh). The refined models of lariocidins are displayed in their respective electron density maps after the refinements contoured at 1.0σ. Color scheme as in panels (a) and (c), respectively.

Figure 4 ∣. Structure of LAR in complex with the T. thermophilus 70S ribosome.

(a, b) Overview of the LAR binding site (yellow) in the bacterial ribosome, viewed from two different perspectives. The 30S subunit is shown in light grey, the 50S subunit is dark grey, the mRNA is cyan, and the A-, P-, and E-site tRNAs are colored teal, blue, and orange, respectively. In (a), the 30S subunit is viewed from the inter-subunit side. The view in panel b is a cross-cut section through the nascent peptide exit tunnel. (c, d) Close-up views of the LAR’s interactions with the small ribosomal subunit. The E. coli numbering of the 16S rRNA nucleotides is used. H-bonds between LAR, rRNA, and A-site tRNA are indicated with dashed lines. In (d), the mRNA nucleotides are numbered relative to the first nucleotide of the P-site codon. (e, f) Superposition of structures of antibiotics binding in the vicinity of the decoding center on the small ribosomal subunit. Overview (e) and close-up view (f) of the ribosome-bound LAR (yellow) relative to the binding sites of other antibiotics targeting the A site of the small ribosomal subunit: odilorhabdin (ODL, teal), tetracycline (TET, blue), negamycin (NEG, green), streptomycin (STR, light red), paromomycin (PAR, dark blue), viomycin (VIO, orange).

Figure 5 ∣. Therapeutic efficacy of LAR in a mouse neutropenic thigh infection model.

(a-c) Reduction in bacterial burden (A. baumannii C0286) in mice after 24 or 48 h postinfection with (treated) or without treatment (control) with LAR as measured by colony forming units (cfu) per gram of tissue or per mL of blood. LAR was administered intraperitoneally at 50 mg/kg at 1 h, 4 h, 8 h, and 20 h postinfection. (a) Bacterial burden in the spleen in control (n = 21) and treated (n = 15) groups. (b) Thigh bacterial burden in control (n = 32) and treated (n = 30) groups. (c) Blood bacterial burden in control (n = 10) and treated (n = 8) groups. Data points are from individual animals (in panel ‘b’ one point represents one thigh) and horizontal lines represent the group means. Significance was determined with a two-tailed Mann-Whitney test (***P=0.003;****P<0.0001). (d) Kaplan-Meier test showing group survival across select time points throughout A. baumannii thigh infection in vehicle control (n=10) and LAR-treated mice (n=8). The mice were administered with LAR as mentioned above. ****P =0.0003, Log-rank (Mantel-Cox) test.