Chiral peptidoglycan mimics target bacterial wall biosynthesis for pathogen intervention

Abstract

The global burden of microbial infections and antimicrobial resistance, coupled with the absence of precise bacterial recognition modalities, demands innovative breakthroughs in antibacterial treatment. Here, we report a chirality-specific biomimic—D-alanine-conjugated peptidoglycan mimics (D-PM)—designed for bacterial recognition. D-PM exhibits broad-spectrum, effective recognition across ESKAPE pathogens, antibiotic-resistant strains, and clinical isolates, while displaying minimal interaction with eukaryotic cells. We elucidate the bacterial recognition mechanism, wherein D-PM—act as a biosynthetic substrates—become incorporated into peptidoglycan biosynthesis. This reveals a mechanism by which macromolecular mimetics are assimilated into bacterial biosynthesis, providing insights into bacterial recognition. Beyond recognition, D-PM enables the construction of pathogen-specific imaging agents and antibiotic-targeted delivery systems. In localized and systemic infection models, D-PM achieves efficient pathogen localization, tissue penetration, and enhanced therapeutic outcomes. This work presents a molecularly engineered strategy for bacterial recognition and intervention, offering a translational approach to address the escalating threat of infectious diseases.

Improved delivery of antibiotics is needed. Here, the authors report on chirality-specific peptidoglycan mimics that hijack bacterial cell wall synthesis, enabling pathogen recognition, bioimaging, and targeted antibiotic delivery while minimizing off-target effects on host cells.

Introduction

Microbial infections continue to pose a formidable threat to human health, accounting for more than 7 million deaths annually^1,2. This escalating crisis highlights an urgent need for effective approaches that enable the precise diagnosis and targeted intervention of pathogenic infections^3,4. Central to this challenge is the accurate recognition of bacterial pathogens—an indispensable prerequisite for effective treatment and resistance mitigation^5,6. Despite significant advances in bacterial recognition, differentiating pathogenic bacteria from host components within complex biological environments remains profoundly difficult, posing a major obstacle to clinical antibacterial therapy^7,8. As a result, the development of robust strategies for bacterial recognition has emerged as a critical scientific imperative in the global battle against infectious diseases.

The bacterial cell wall, built upon a peptidoglycan scaffold, constitutes one of the most fundamental distinctions between prokaryotic pathogens and eukaryotic hosts⁹. Peptidoglycan—universally composed of sugar moieties and chirality-specific amino acids—is highly conserved across bacterial taxa (Supplementary Fig. 1)^10–13, which presents unique molecular signatures for precise bacterial recognition^14–18. Although peptidoglycan-targeting antibiotics—such as vancomycin—have served as pivotal agents for bacterial eradication and detection^19–23, their efficacy is increasingly compromised by target mutations and emerging resistance^24–26. Concurrently, increasing efforts have been directed toward biosynthetic labeling for bacterial recognition, drawing inspiration from the bacterial-specific sugars (e.g., N-acetylglucosamine, N-acetylmuramic acid) and amino acids (e.g., D-alanine) essential for peptidoglycan biosynthesis^27–32. However, these small-molecule probes are intrinsically constrained by limited structural tunability, suboptimal recognition efficiency, and restricted biological permeability^33,34. These persistent limitations necessitate bacterial recognition platforms that break through conventional intervention paradigms and molecular designs.

Emerging evidence suggests that macromolecular analogs of bacterial peptidoglycan, endowed with superior biomimetic properties, represent promising candidates for targeted bacterial intervention³⁵. For instance, chitin—a natural polysaccharide sharing partial structural features with peptidoglycan—exhibits comparable recognition by host pattern recognition receptors^36,37, while synthetic glycoconjugates have been shown to elicit peptidoglycan-mimetic immune responses^38,39. Inspired by these findings, we rationally designed macromolecular mimics of peptidoglycan substrates that recapitulate the chirality-specific amino acid motifs and glycan backbone architectures (Supplementary Fig. 2). In this work, we delineate the design principles of peptidoglycan mimics, elucidate their structure–function relationships and biological mechanisms of bacterial biosynthetic incorporation (Fig. 1a and Supplementary Fig. 3), and demonstrate their capability to selectively distinguish pathogenic bacteria from host cells. Furthermore, we engineered these peptidoglycan mimics into nanoparticulate probes and antibiotic delivery platforms, enabling targeted intervention of bacterial infections. This multidisciplinary study establishes a transformative solution for bacterial diagnostics and therapeutics, providing effective strategies to tackle the global burden of resistant microbial infections.

Fig. 1. Chiral peptidoglycan mimics for bacterial recognition.

a Schematic illustration of D-peptidoglycan mimics incorporated into cell wall biosynthesis and biomedical applications. b Relative positive rate of different pathogenic bacteria, including S. aureus ATCC29213, S. aureus ATCC25904, S. aureus RN4220, E. faecalis ATCC29212, E. faecalis ATCC19433, S. aureus ATCC43300 (MRSA), and a clinical isolate of S. aureus, incubated with FITC-labeled CS, L-PM, and D-PM at a concentration of 0.5 mg ml⁻¹ for 1 h (n = 6 biological replicates per group). Absolute values are shown in Supplementary Fig. 12. c Representative CLSM images of FITC-labeled CS, L-PM, and D-PM incubated with S. aureus ATCC29213 at a concentration of 0.5 mg ml⁻¹ for 1 h. Scale bar represents 2 μm. Data in (c) are representative of three biologically independent experiments.

Results

Chiral peptidoglycan mimics label bacteria

To enable the controlled synthesis of chiral peptidoglycan mimics (PM), protected chiral alanine (Ala) was conjugated to glucosamine (GlcNH₂) in chitosans (CS, (GlcN)_n analogous to the polysaccharide chain of peptidoglycan) via amide bond formation under racemization-free conditions^40,41. A series of chiral amino acid-conjugated (GlcN)_n polymers was synthesized with varying degrees of L-Ala (7.5 ± 0.3%, 11.3 ± 0.4%, 17.7 ± 0.3%, and 23.5 ± 0.4%) or D-Ala (7.4 ± 0.6%, 11.6 ± 0.4%, 17.2 ± 0.3%, and 22.9 ± 0.8%) modification, respectively (Supplementary Fig. 4). Following deprotection, L-PM of (GlcN–L-Ala)_n and D-PM of (GlcN–D-Ala)_n were obtained with distinct Ala modification but identical fluorescein isothiocyanate (FITC) labeling. The bacterial interactions of the chiral PM were quantified using Staphylococcus aureus (S. aureus ATCC29213) as a model strain. L-PM exhibited consistently low binding efficiency with minimal dependence on the modification ratio, yielding FITC-positive bacterial populations ranging from 2.9 ± 0.2% to 14.1 ± 0.4% (Supplementary Fig. 5). In contrast, D-PM displayed a pronounced modification-ratio-dependent recognition behavior, with the proportion of FITC-positive bacteria increasing from 27.3 ± 2.7% to 94.1 ± 0.3%. Notably, maximal bacterial labeling was achieved at a D-Ala modification degree of 17.2%, whereas further increasing the substitution to 22.9% resulted in diminished labeling efficiency, likely due to steric congestion that impairs enzymatic incorporation during cell wall biosynthesis. On this basis, the 17.2% modification rate of D-PM was selected for all subsequent studies. ¹H nuclear magnetic resonance (NMR) spectra confirmed the successful conjugation of chiral amino acids onto (GlcN)_n, as evidenced by the concurrent appearance of characteristic signals corresponding to sugar rings and amino acids (Supplementary Fig. 6). Both L-PM and D-PM exhibited zeta potentials comparable to that of native CS (approximately +23 mV) in aqueous solution (Supplementary Fig. 7a). Dynamic light scattering (DLS) analysis indicated that CS could form aggregates around 350 nm in an aqueous solution, whereas both L-PM and D-PM assembled into smaller nanostructures of 160 nm (Supplementary Fig. 7b–d). Transmission electron microscopy (TEM) revealed well-defined nanostructures of both L-PM and D-PM, highlighting their potential as functional nanomaterials. The circular dichroism spectra revealed distinct trends, with L-PM showing a more pronounced attenuation of the negative band at 212 nm, but D-PM exhibiting a greater increase in the positive band at 193 nm (Supplementary Fig. 8). These spectral variations are likely attributed to the modifications induced by the chiral amino acids.

Chiral peptidoglycan mimics with abundant free amino groups facilitated effortless conjugation with diverse fluorescent molecules, including FITC, 6-carboxytetramethylrhodamine (TAMRA), and sulfo-cyanine5 (Cy5), facilitating sophisticated biological imaging (Supplementary Fig. 9). Moreover, both L-PM and D-PM enhanced the stability of the fluorescent probes, with no detectable fluorescence decay over 10 days (Supplementary Fig. 10). To assess bacterial labeling performance, FITC-labeled CS, L-PM, and D-PM were applied at identical concentrations and normalized to equivalent fluorescence intensities (Supplementary Fig. 11). Strikingly, flow cytometric analysis revealed that D-PM displayed high-affinity interactions across a broad panel of Gram-positive pathogenic bacteria⁴², including multiple Staphylococcus aureus strains (S. aureus ATCC29213, S. aureus ATCC25904, and S. aureus RN4220), two Enterococcus faecalis strains (E. faecalis ATCC29212 and E. faecalis ATCC19433), methicillin-resistant Staphylococcus aureus (MRSA ATCC43300), as well as clinically isolated Staphylococcus aureus (Fig. 1b and Supplementary Fig. 12). Meanwhile, D-PM also demonstrated significantly stronger recognition and binding toward Gram-negative ESKAPE pathogens, including Klebsiella pneumoniae (K. pneumoniae ATCC13883), Acinetobacter baumannii (A. baumannii ATCC19606), Pseudomonas aeruginosa (P. aeruginosa ATCC27853), and Escherichia coli (E. coli ATCC25922), compared with CS and L-PM (Supplementary Fig. 13). In addition, the relatively lower labeling efficiency observed for certain Gram-negative species in the D-PM-treated group, including K. pneumoniae and E. coli, is likely attributable to the presence of an outer membrane that partially restricts D-PM access to the peptidoglycan⁴³. By contrast, both CS and L-PM exhibited uniformly low binding efficacy toward both Gram-positive and Gram-negative bacterial strains. More importantly, D-PM exhibited robust bacterial binding and fluorescent labeling as visualized by confocal laser scanning microscopy (CLSM), clearly delineating the coccoid morphology of S. aureus ATCC29213 (Fig. 1c and Supplementary Fig. 14). D-PM exhibited pronounced dose-dependent binding to bacteria, indicative of its potential accumulation on the bacteria (Supplementary Fig. 15). Taken together, D-PM exhibits chirality-dependent biological effects that enable broad-spectrum and highly efficient recognition of pathogenic bacteria, including antibiotic-resistant strains.

To investigate the bacterial recognition advantage conferred by chiral peptidoglycan mimic design, we compared FITC-labeled D-PM with FITC-labeled small-molecule D-Ala probes at matched D-Ala molar equivalents. Two control conditions were employed: (1) a D-Ala/FITC-labeled D-Ala mixture calibrated to match the fluorescence intensity of FITC-labeled D-PM, and (2) fully FITC-labeled D-Ala with substantially higher overall fluorescence than FITC-labeled D-PM (Supplementary Fig. 16). Under fluorescence-matched conditions of probe, only 2.6 ± 0.2% of bacteria were FITC-positive following treatment with the D-Ala/FITC-labeled D-Ala mixture, which was dramatically lower than the 85.7 ± 0.2% FITC-positive bacteria observed in the FITC-labeled D-PM group. The mean fluorescence intensity (MFI) in the FITC-labeled D-PM-treated bacteria was about 8-fold higher than that of the D-Ala/FITC-labeled D-Ala mixture. Notably, even compared with fully FITC-labeled D-Ala with approximately 4-fold higher fluorescence than FITC-labeled D-PM, both the proportion of FITC-positive bacteria and MFI remained higher in the FITC-labeled D-PM-treated group. CLSM images clearly demonstrate the overwhelming performance of D-PM in bacterial fluorescent labeling (Supplementary Fig. 17). These results underscore the superior bacterial recognition efficiency afforded by D-chiral peptidoglycan mimics, highlighting the intrinsic advantages of chirality-guided macromolecular design over small-molecule D-Ala–based probes for pathogenic bacteria.

D-PM incorporates biosynthetic downstream

To elucidate the recognition mechanisms, microscale thermophoresis (MST) was employed to generate binding curves that characterized the interactions between chiral peptidoglycan mimics and S. aureus⁴⁴. The results revealed a significant interaction between D-PM and bacteria, with a dissociation constant (K_d) of 5.4 ± 0.3 μM—an order of magnitude lower than that of both CS and L-PM (Fig. 2a). This finding indicated that the D-chiral strategy enhanced the affinity between D-PM and pathogenic bacteria (S. aureus ATCC29213). In parallel, flow cytometry analysis demonstrated that FITC-labeled D-PM rapidly bound to bacteria over time, achieving near-complete binding efficiency (Fig. 2b and Supplementary Fig. 18). The maximal binding rates of fluorescent CS and L-PM with bacteria remained below 20%. These striking differences suggested that D-PM may engage distinct mechanisms in binding to bacteria, potentially involving biosynthetic processes.

Fig. 2. D-PM incorporate bacterial metabolism downstream.

a MST analysis of the binding affinities of CS, L-PM, and D-PM to S. aureus ATCC29213 (n = 3 biological replicates per group; data presented as mean ± s.d.). b Flow cytometry analysis of binding kinetics of 0.5 mg ml^-1 FITC-labeled CS, L-PM, and D-PM to S. aureus ATCC29213 at different time points (n = 3 biological replicates per group; data presented as mean ± s.d.). c UPLC-MS analysis of UDP-MurNAc-pentapeptide accumulation following treatment with CS, L-PM, D-PM, and Van in S. aureus cultures. Extracted ion chromatograms (d, EIC, theoretical m/z 322.1609) and MS (e, observed m/z 322.1613) of GlcN-D-Ala-D-Ala after CS, L-PM, and D-PM incubation with S. aureus cultures. EIC (f, theoretical m/z 925.4360) and MS (g, observed m/z 925.4353) of GlcN-D-Ala-D-Ala-L-Lys-D-iGlu-L-Ala-MurNAc after CS, L-PM, and D-PM incubation with S. aureus cultures. Molecular docking of GlcN-D-Ala with Ddl (h) and GlcN-D-Ala-D-Ala with MurF (i) visualized by PyMOL, with specific residues shown in blue. j Effect of different bacterial biosynthesis inhibitors on the incorporation of D-PM into S. aureus ATCC29213. All inhibitors were used at a concentration of 100 μg ml⁻¹ (n = 6 biological replicates per group; data presented as mean ± s.d.). The tested inhibitors included D-cycloserine (Dcy), amoxicillin (Amo), meropenem (Mer), tetracycline (Tet), and ciprofloxacin (Cip). Statistical analysis was performed using a two-tailed unpaired t-test (a, j); ***P < 0.001, ****P < 0.0001.

Uridine 5’-diphosphate-N-acetylmuramic acid pentapeptide (UDP-MurNAc pentapeptide) is a well-established biomarker of cell wall synthesis intervention, whose accumulation is notably induced by vancomycin (Van)^23,45. An ultra-high-performance liquid chromatography system coupled with a high-resolution quadrupole–orbitrap mass spectrometer (UPLC-MS) revealed the presence of a characteristic peak for UDP-MurNAc pentapeptide at 1,150.3583 in the bacterial wall of S. aureus following exposure to D-PM, consistent with the signal observed after vancomycin treatment (Fig. 2c and Supplementary Fig. 19). In contrast, UDP-MurNAc pentapeptide was undetectable in both the CS and L-PM groups. Moreover, flow cytometry results indicated that pretreatment with vancomycin significantly reduced D-PM binding to S. aureus, likely due to competitive interference in cell wall synthesis (Supplementary Fig. 20).

Inspection of UPLC-MS results revealed that D-peptidoglycan mimics interfered with the biosynthesis of the bacterial cell wall. The observed peak at m/z 322.1613 was assigned to the protonated molecular ion [GlcN–D-Ala–D-Ala + H]⁺ (theoretical m/z 322.1609), which likely arises from the ligation of an additional D-Ala to the D-PM substrate (GlcN–D-Ala) catalyzed by D-Ala–D-Ala ligase (Ddl, Fig. 2d, e)²⁷. This intermediate, GlcN-D-Ala-D-Ala, could subsequently be incorporated into the cell wall biosynthesis pathway to form GlcN-D-Ala-D-Ala-L-Lys-D-iGlu-L-Ala-MurNAc through the biocatalyst of UDP-N-acetylmuramoyl-tripeptide–D-alanyl-D-Ala ligase (MurF), with a characteristic ion peak at m/z 925.4353 (Fig. 2f, g). To further elucidate the structural basis of D-PM interference in cell wall synthesis, we used the catalytic domain of Ddl bound to GlcN-D-Ala and determined the structure of the complex. Docking analysis revealed that D-PM, as a substrate bind to Ddl in a pocket highly analogous to that of D-Ala within the overall complex structure (Fig. 2h, i and Supplementary Fig. 21a, b). Furthermore, the strong binding affinity of D-PM to Ddl was further supported by a lower docking score of −6.775 kcal mol⁻¹, compared to the D-Ala/Ddl complex, which had a docking score of −4.077 kcal mol⁻¹. Similarly, the intermediate of GlcN-D-Ala-D-Ala exhibited a better docking score of −8.088 kcal mol⁻¹ as compared to natural D-Ala-D-Ala (docking score of −6.042 kcal mol⁻¹) when bound to the catalytic pocket of MurF (Supplementary Fig. 21c, d). Subsequently, molecular dynamics simulations revealed that the GlcN–D-Ala/Ddl and GlcN–D-Ala–D-Ala/MurF complexes exhibited sustained structural stability, with markedly stronger interactions than those observed between conventional small-molecule substrates (D-Ala or D-Ala–D-Ala) and their corresponding ligases (Supplementary Figs. 22 and 23). UPLC-MS evidence and molecular simulations jointly confirmed that chirality-specific peptidoglycan mimics could efficiently integrate into the metabolism downstream of bacterial cell wall biosynthesis, thereby facilitating the robust recognition and biosynthetic labeling of pathogens.

In addition, key cell wall synthesis inhibitors, including D-cycloserine (D-Ala-D-Ala ligase inhibitor), amoxicillin, and meropenem (penicillin-binding protein transpeptidase inhibitor)⁴³, were able to block the incorporation of D-PM into downstream metabolic pathways (Fig. 2j). In contrast, a protein synthesis inhibitor of tetracycline and a DNA synthesis inhibitor of ciprofloxacin showed no effect on the incorporation of D-PM into S. aureus. In addition, none of the inhibitors significantly affected the binding of CS and L-PM to the bacteria (Supplementary Fig. 24). The chiral-specific design of D-peptidoglycan mimics enables their integration into bacterial cell wall synthesis metabolism, thereby facilitating efficient bacterial recognition and targeting.

Encouraged by the molecular-level confirmation of downstream metabolism, we proceeded to investigate the impact of D-PM on the morphology of pathogenic bacteria. FITC-labeled peptidoglycan mimics were incubated with S. aureus over three passages in Luria–Bertani broth (Fig. 3a). As shown in Fig. 3b, only a small proportion of fluorescent L-PM bound to S. aureus ATCC29213, always with around 10% of bacteria exhibiting fluorescence. This minimal binding of L-PM is attributable to weak interactions governed by a dynamic adsorption-desorption equilibrium, as evidenced by the reduced bacterial surface zeta potential following supramolecular electrostatic interactions between L-PM and the bacterial envelope (Supplementary Fig. 25). Under the same conditions, continuous co-incubation with D-PM from Passage 1 to Passage 3 led to a progressive increase in the percentage of fluorescent bacteria, rising from 16.1% to 44.3%. This outcome was due to the incorporation of D-PM into bacterial metabolism, enabling the accumulation over successive generations during bacterial proliferation. To visualize this generational integration, super-resolution three-dimensional structured illumination microscopy (3D-SIM) provided direct evidence of D-PM incorporation into bacterial cell wall synthesis (Fig. 3c, d). In Passage 1, Cy5-labeled D-PM well-delineated the bacterial cell wall (red fluorescence). In Passage 2, TAMRA-labeled D-PM was incorporated into the biosynthesis of bacterial envelopes (purple fluorescence), while Cy5-labeled D-PM remained prominently visible on the bacteria. The approximately half reduction in Cy5 fluorescence intensity was attributed to fluorescence dilution during bacterial division in Passage 2. In Passage 3, co-incubation with FITC-labeled D-PM (green fluorescence) demonstrated clear co-localization of red, purple, and green fluorescence along the bacterial contours, indicating stable incorporation of D-PM into the downstream metabolism of cell wall synthesis. In contrast, L-PM showed minimal co-localization with bacteria, consistent with flow cytometry results, and failed to accumulate within the bacterial cell wall.

Fig. 3. D-PM accumulation into bacterial cell wall.

a Schematic illustration of FITC-labeled D-PM multiple-round incubations with S. aureus ATCC29213. b Flow cytometry analysis of S. aureus ATCC29213 at Passage 1 to Passage 3 (P1, P2, and P3) following multiple-round incubations with FITC-labeled L-PM and D-PM at 0.5 mg ml^-1 every 1 h. c Schematic illustration of Cy5, TAMRA, and FITC-labeled D-PM multiple-round incubations with S. aureus. d Representative 3D-SIM images of S. aureus ATCC29213 after incubation with Cy5-labeled L-PM or D-PM (red) for P1, TAMRA-labeled L-PM or D-PM (purple) for P2, and FITC-labeled L-PM or D-PM (green) for P3. Scale bar represents 2 μm. e SEM characterization of P3 bacterial morphology after treatment with phosphate-buffered saline (PBS), CS, L-PM, or D-PM. Scale bar represents 2 μm. f Flow cytometry analysis of PI-positive rate in P3 bacteria following treatment with PBS, CS, L-PM, or D-PM (n = 6 biological replicates per group; data presented as mean ± s.d.). Statistical analysis was performed using a two-tailed unpaired t-test (f); ****P < 0.0001.

Scanning electron microscopy (SEM) images of bacteria over three passages in the presence of L-PM or D-PM revealed distinct morphological differences (Fig. 3e). In the L-PM-treated groups, bacteria retained well-defined surfaces, appearing consistent with both the control and CS-treated groups. However, D-PM induced notable structural abnormalities, characterized by visible protrusions on the bacterial surface. These deformities are likely attributed to the incorporation of D-PM into cell wall synthesis, disrupting the formation of a cohesive peptidoglycan network. Additionally, propidium iodide (PI) probe assays confirmed the compromised bacterial envelope integrity after three passages in the presence of D-PM (Fig. 3f). The PI influx in the D-PM-treated group was significantly higher than that observed in the control and L-PM-treated groups, indicating abnormal envelopes. In addition, colony-forming unit (CFU) assays demonstrated that incorporation of D-PM into bacterial cell wall biosynthesis elicited a modest bactericidal effect at Passage 1 (Supplementary Fig. 26). With successive passages up to Passage 3, progressive accumulation of D-PM within the peptidoglycan increasingly compromised cell wall integrity, leading to a stepwise enhancement of bactericidal activity. These findings confirmed that D-PM could be efficiently incorporated into bacterial cell wall biosynthesis and progressively accumulates within the peptidoglycan.

D-PM recognizes bacteria over mammalian cells

D-PM achieved efficient integration into bacterial biosynthesis, exhibiting satisfactory bacterial recognition. By leveraging the intrinsic differences in chiral amino acid metabolism between prokaryotic pathogens and mammalian cells, D-PM demonstrated significant potential for selectively distinguishing these entities, paving the way for advancements in the diagnosis and treatment of bacterial infections. As expected, CS, L-PM, and D-PM exhibited minimal interactions with animal cell lines at concentrations corresponding to those used for bacterial incubation, including mouse NIH 3T3 cells, human umbilical vein endothelial cells, smooth muscle cells, and monkey Vero cells (Supplementary Fig. 27). Additionally, MST measurements revealed no detectable specific binding of CS, L-PM, or D-PM to NIH 3T3 cells (Supplementary Fig. 28), and UPLC-MS analysis targeting characteristic peptidoglycan-derived motifs detected no corresponding signals in NIH 3T3 cells following incubation with CS, L-PM, or D-PM (Supplementary Fig. 29). Co-culture with prokaryotic bacteria and eukaryotic cells in the presence of fluorescent D-PM revealed a clear preference for bacterial recognition by flow cytometry analysis (Fig. 4a). As shown in Fig. 4b, D-PM selectively associated with S. aureus, with approximately 30% of bacteria exhibiting FITC-positive fluorescence, whereas binding to mammalian NIH 3T3 cells was much lower. Additionally, D-PM exhibited robust specificity for identifying S. aureus over a range of the mammalian cell lines (Supplementary Fig. 30). CLSM images directly showed that demonstrated that D-PM (green) exhibited strong co-localization with S. aureus but poor co-localization with NIH 3T3 cells, confirming the remarkable selectivity of D-PM in distinguishing bacteria from mammalian cells (Fig. 4c). In addition, flow cytometry and CLSM images revealed that CS and L-PM exhibited minimal binding to S. aureus and NIH 3T3 cells under co-culture conditions (Supplementary Figs. 31 and 32). Similarly, D-PM effectively distinguished multiple Gram-negative bacterial species from NIH 3T3 cells under co-culture conditions (Supplementary Fig. 33). These findings highlight the potential of D-PM as a broad-spectrum recognition tool for pathogenic bacteria.

Fig. 4. D-PM for bacteria recognition and antibiotic targeted delivery.

a Schematic representation for D-PM recognition of bacteria and mammalian cells. b Flow cytometry analysis for S. aureus ATCC29213 and NIH 3T3 cells co-cultures after incubation with FITC-labeled D-PM at 0.5 mg ml^-1 for 1 h (n = 6 biological replicates per group; data presented as mean ± s.d.). c Representative CLSM images for S. aureus ATCC29213 and NIH 3T3 cells co-cultures after incubation with FITC-labeled D-PM at 0.5 mg ml⁻¹ for 1 h. S. aureus and NIH 3T3 cells were stained with Hoechst 33342 (blue). Outline the NIH 3T3 cells with white dashed lines based on the bright-field images. Scale bar represents 20 μm. d Schematic illustration for the fabrication process of T/D-PM with the size distribution and TEM image. Scale bar represents 100 nm. e CFU count after treatment with 2 μg ml⁻¹ Tet for 4 h in PBS with different treatment groups. S. aureus ATCC29213 treated with PBS served as the control group (n = 6 biological replicates per group; data presented as mean ± s.d.). f Representative CLSM images of Tet, T/CS, T/L-PM, and T/D-PM incubated with S. aureus ATCC29213 at 2 μg ml⁻¹ Tet (green) for 1 h. Hoechst 33342 was used to stain nuclei (blue). Scale bar represents 2 μm. g Representative CLSM images of Tet, T/CS, T/L-PM, and T/D-PM incubated with S. aureus ATCC29213 at 2 μg ml⁻¹ Tet for 1 h. Bacteria were stained with Hoechst 33342 (blue) to visualize nuclei, and PI was used to identify dead bacteria (red). Scale bar represents 10 μm. Statistical analysis was performed using a two-tailed unpaired t-test (b, e); ****P < 0.0001. Data in (d) and (f, g) are representative of three independent experiments.

Based on its strong bacterial recognition and self-aggregation into nanoparticles, D-PM holds great potential as a vehicle for antibiotic loading and targeted delivery. Pyrene probes, serving as guest molecules, were readily encapsulated into the internal pockets of nanoparticulate CS, L-PM, and D-PM, demonstrating their potential as antibiotic carriers (Supplementary Fig. 34). Using hydrophobic tetracycline (Tet) as a model drug, Tet was successfully encapsulated into the pockets of D-PM to form Tet-loaded D-PM (T/D-PM)⁴⁶, which exhibited a uniform size distribution of approximately 180 nm and well-defined nanostructures (Fig. 4d and Supplementary Fig. 35). The drug-loaded T/D-PM delivery system exhibited a loading capacity of ~6.6% and an encapsulation efficiency of ~32.9% (Supplementary Fig. 36). T/D-PM significantly enhanced the antibacterial efficacy of antibiotics, reducing the minimum inhibitory concentration (MIC) of tetracycline by 4-fold against S. aureus ATCC29213 (Supplementary Table 1). Moreover, CFU assays further validated that, under the same Tet concentration conditions, T/D-PM completely eradicated S. aureus, while free Tet, T/CS, and T/L-PM failed to eliminate the bacteria (Fig. 4e). CLSM images revealed that D-PM improved the internalization of tetracycline through targeted delivery, showing significantly stronger green fluorescence signals compared to free Tet, T/CS, and T/L-PM (Fig. 4f and Supplementary Fig. 37a). Targeted antibiotic delivery by D-PM substantially promoted antibacterial efficacy of antibiotics, as demonstrated by the highest proportion of PI-positive dead bacteria (red, Fig. 4g and Supplementary Fig. 37b). Notably, T/D-PM potentiated the antibacterial efficacy of tetracycline against MRSA, halving its MIC and markedly suppressing bacterial colony formation on agar plates (Supplementary Table 2 and Supplementary Fig. 38).

As expected, tetracycline-containing formulations exhibited no detectable cytotoxicity toward multiple mammalian cell lines even at concentrations exceeding 40 times the effective antimicrobial dose (Supplementary Figs. 39 and 40). Additionally, treatment of erythrocytes collected from BALB/c mice with these formulations at high concentration revealed no hemolysis, demonstrating excellent hemocompatibility (Supplementary Fig. 41). The combination of high-efficiency targeted delivery and favorable biocompatibility underscores the potential of chiral-specific D-peptidoglycan mimics as promising carriers for biomedical applications.

D-PM intervene wound bacterial infections

The robust incorporation into bacterial metabolism, high-efficiency bacterial recognition, and potential for bacteria diagnosis and therapy of D-PM prompted further investigation of their efficacy in vivo. FITC-labeled D-PM effectively marked nearly all S. aureus, which were used to infect wounds (~80 mm²) on the dorsal skin of BALB/c mice with a bacterial load of 1.0 × 10⁶ CFU ml⁻¹ per mouse (Fig. 5a). Following bacterial infection, the wounds were treated with Cy5-labeled CS, L-PM, or D-PM, respectively. After an additional 4 h, wound tissues were excised, cryosectioned, and visualized using CLSM. As illustrated in Fig. 5b, red fluorescence from D-PM exhibited prominent co-localization with the green fluorescence of S. aureus, verifying their targeted association. In contrast, no significant red fluorescence was observed in the CS- or L-PM-treated groups, and uninfected wounds also exhibited no fluorescence signal from D-PM. Semi-quantitative analysis of co-localization rates demonstrated that D-PM selectively traced a substantial number of pathogens, a capability absent in CS- and L-PM-treated groups (Supplementary Fig. 42). Moreover, their facile conjugation enables the incorporation of diverse imaging probes into D-PM, potentially allowing for multimodal diagnostics.

Fig. 5. D-PM for bacterial recognition and antibacterial treatment in wound infection.

a Schematic diagram of bacterial recognition in the wound infection model. b Representative CLSM images of Cy5-labeled CS, L-PM, and D-PM (red) recognizing fluorescent S. aureus ATCC29213 (green) in the wound infection area. Cells were stained with Hoechst 33342 (blue). Scale bar represents 100 μm. Pearson’s correlation coefficients (r) quantifying colocalization are provided in Supplementary Fig. 42 (n = 3 biologically independent animals). c Schematic diagram of antibacterial administrations in the wound infection model. Photographs with serial dilution (d) and quantification (e) of bacterial colonies remaining in wounds after different administrations of PBS, Tet, T/CS, T/L-PM, and T/D-PM (n = 6 biologically independent animals; data presented as mean ± s.d.). f Representative Gram-stained images of infected wounds after different treatments on day 1. Scale bar represents 20 μm. g Changes in wound area in S. aureus ATCC29213 infected mice after different treatments (n = 6 biologically independent animals; data presented as mean ± s.d.). h Representative Masson-stained images of wound regeneration after different treatments on day 8. Scale bar represents 100 μm. Statistical analysis was performed using a two-tailed unpaired t-test (e, g); *P < 0.05, ****P < 0.0001. Data in (b) and (h) are representative of three and six biologically independent experiments, respectively.

To assess the targeted delivery capabilities of D-PM, a wound infection model was established by inoculating BALB/c mice with a suspension of S. aureus (1.0 × 10⁶ CFU) for 4 h (Fig. 5c). Infected mice were treated with PBS (control), free Tet, T/CS, T/L-PM, or T/D-PM, at a dosage of 10 mg Tet per kg of body weight. Plate counting analysis demonstrated that T/D-PM exhibited the most effective antibacterial performance in infected wounds, attributable to its targeted antibiotic delivery. As illustrated in Fig. 5d, T/D-PM markedly enhanced the antibacterial efficacy of tetracycline, nearly eradicating all bacteria within the wound area. In contrast, T/CS and T/L-PM treatments failed to improve the antibacterial activity of tetracycline (Fig. 5e). Gram staining of wound tissue sections further corroborated these findings, showing that T/D-PM eliminated bacteria from the wound tissues, whereas abundant purple bacteria persisted in the wounds of other treatment groups (Fig. 5f).

At the end of the observation period, T/D-PM exhibited the greatest efficacy in promoting infected wound healing, as evidenced by the smallest wound area (Fig. 5g and Supplementary Fig. 43). Hematoxylin and eosin (H&E) staining revealed that T/D-PM treatment promoted the formation of a thicker dermal layer in the infected wounds, underscoring its superior antibacterial efficacy compared to free Tet, T/CS, and T/L-PM (Supplementary Figs. 44 and 45). Masson staining further indicated enhanced collagen deposition in the T/D-PM group, suggesting accelerated repair of the infected wounds (Fig. 5h, Supplementary Figs. 46 and 47). Picrosirius red staining revealed that the proportion of type I collagen relative to the total collagen was increased in the wound tissue of T/D-PM-treated wounds, suggesting enhanced replacement of type III collagen by type I collagen to promote more effective wound healing (Supplementary Figs. 48 and 49). Moreover, the infected wounds treated with T/D-PM exhibited elevated CD31 expression relative to other groups, indicative of increased capillary formation with mature vascular structures (Supplementary Figs. 50 and 51). The enhanced healing observed in infected wounds treated with T/D-PM is primarily attributable to its potent antibacterial activity, whereas evaluation in non-infected wound models further revealed that both D-PM and T/D-PM possessed intrinsic bioactivity that promotes wound repair (Supplementary Fig. 52). Collectively, these findings highlight that T/D-PM not only enhances the in vivo antibacterial potency of tetracycline but also facilitates wound healing through promoting dermal regeneration, collagen deposition, and neovascularization.

D-PM manages systemic bacterial infection

Systemic bacterial infections, such as peritonitis, present a rigorous model to evaluate the targeted bacterial recognition capability of D-PM under disseminated infection conditions. This model leverages the highly specific integration of D-PM into bacterial metabolic pathways. To explore this, a bacterial peritonitis model was established through the intraperitoneal injection of S. aureus ATCC29213 (2.5 × 10⁷ CFU) into the right lower abdomen of BALB/c mice (Fig. 6a). One hour post-infection, fluorescence CS, L-PM, and D-PM were intraperitoneally administered into the left lower abdomen of the infected mice, respectively. An in vivo imaging system provided a visual representation of the biomacromolecule distribution within the body. As the observation time extended, fluorescence signals from CS and L-PM gradually diminished, becoming almost undetectable by 4 h (Fig. 6b). In contrast, fluorescence signals from D-PM remained stable even at 4 h post-administration and demonstrated site-specific accumulation in the peritonitis mice. On the other hand, fluorescence signals of D-PM also decreased in uninfected mice within 4 h, attributable to the absence of bacterial infections. Semi-quantitative analysis of whole-body fluorescence, based on average radiant efficiency, demonstrated the progressive clearance of both CS and L-PM from peritonitis mice, as reflected by the time-dependent attenuation of fluorescence signals (Fig. 6c). In contrast, D-PM demonstrated prolonged retention in peritonitis mice, with fluorescence intensities approximately 3-fold higher than those of CS and L-PM. However, the signal decay of D-PM in uninfected mice followed a similar pattern to that of CS and L-PM in bacterial peritonitis models. This infection-dependent retention of D-PM is primarily due to the efficient incorporation into bacterial metabolic pathways linked to cell wall biosynthesis, whereas no appreciable D-PM retention is observed in uninfected mice.

Fig. 6. D-PM for bacterial recognition and antibacterial treatment in peritonitis.

a Schematic diagram of bacterial recognition in the peritonitis model. S. aureus ATCC29213 were inoculated on the right abdomen, and formulations were administered contralaterally. In vivo imaging (b) and fluorescence intensity (c) of Cy5 signals at different time points following different administrations (n = 3 biologically independent animals; data presented as mean ± s.d.). Abbreviation: average radiant efficiency, ARE. Ex vivo imaging (d) and fluorescence intensity quantification (e, f) of spleens and lungs following different treatments over 6 h (n = 3 biologically independent animals; data presented as mean ± s.d.). g Representative images of frozen tissue sections from spleens and lungs showing drug distribution. Scale bar represents 20 μm. h Schematic diagram of antibacterial administrations in the peritonitis model. Quantitative results of bacterial colonies for ascitic fluid (i), spleens (j), and lungs (k) following different treatments (n = 6 biologically independent animals; data presented as mean ± s.d.). Statistical analysis was performed using a two-tailed unpaired t-test (c, e, f, i–k); *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Data in (g) are representative of three biologically independent experiments.

Next, we sought to investigate the bacterial targeting ability of D-PM at the organ level, given that peritonitis often leads to multiorgan infections. As shown in Fig. 6d, ex vivo imaging revealed pronounced accumulations of D-PM in non-metabolic organs, specifically the spleens and lungs, distinguishing from low fluorescence signals in the CS- and L-PM-treated groups. Plate counting revealed comparable bacterial loads in the spleens and lungs of all peritonitis mice (Supplementary Fig. 53). However, D-PM exhibited selective enrichment in these infected organs compared to the control, CS-, and L-PM-treated groups. Notably, uninfected organs showed no obvious D-PM accumulation, highlighting the specificity of bacterial tracing (Fig. 6e, f). Frozen sections of spleen and lung tissues were analyzed to investigate the tissue distribution of D-PM within infected organs (Fig. 6g and Supplementary Figs. 54 and 55). A significantly higher density of fluorescence signal spots was observed in the D-PM-treated group compared to the L-PM group. The semi-quantitative analysis confirmed that the fluorescence intensity in the D-PM-treated sections was substantially greater than that of the L-PM-treated group. These findings, spanning from the systemic level to the organ level, consistently highlight the robust bacterial targeting capabilities of D-PM, characterized by enhanced retention in infected mice and selective accumulation within infected organs.

Beyond the ability to target and track bacteria, D-PM holds potential as a therapeutic carrier for systemic bacterial infections. Following establishment of the S. aureus ATCC29213 peritonitis model, antibacterial treatment was administered using the different antibiotic delivery systems (Fig. 6h). As shown in Fig. 6i, the bacterial colony count in ascitic fluid was undetectable in the T/D-PM-treated group, which was markedly lower than that observed in the control group (4.5 × 10³ CFU ml⁻¹), the free Tet-treated group (1.0 × 10³ CFU ml⁻¹), the T/CS-treated group (2.2 × 10² CFU ml⁻¹), and the T/L-PM-treated group (2.8 × 10² CFU ml⁻¹). Meanwhile, T/D-PM exhibited the highest antibacterial efficacy, effectively eradicating bacteria in organs, including the spleens and lungs (Fig. 6j,k). Furthermore, T/D-PM exhibited broad-spectrum and potent antibacterial activity in vivo, demonstrating robust therapeutic efficacy in murine peritonitis models challenged with antibiotic-resistant S. aureus ATCC43300 (MRSA, Supplementary Fig. 56) or the Gram-negative pathogen E. coli ATCC25922 (Supplementary Fig. 57). These findings highlight the potential of D-PM as a drug delivery platform to enhance antibiotic efficacy and mitigate bacterial resistance.

Last but most importantly, we conducted a comprehensive in vivo safety evaluation of D-PM, which is critical for its future clinical translation. D-PM was administered intraperitoneally to female BALB/c mice at doses of 200 or 500 mg per kg body weight twice on day 1, followed by a 7-day observation period (Supplementary Fig. 58a). No significant changes in body weight were observed throughout the study (Supplementary Fig. 58b). On day 7, complete blood counts in D-PM-treated mice remained within normal ranges compared with control groups (Supplementary Fig. 59). Consistently, total plasma immunoglobulin G (IgG) levels and plasma biochemistry indicators, including alanine aminotransferase (ALT), aspartate aminotransferase (AST), and blood urea nitrogen (BUN), showed no abnormal alterations (Supplementary Figs. 60 and 61). Moreover, H&E histological examination of major organs, including the heart, liver, spleen, lungs, and kidneys, revealed no apparent pathological lesions (Supplementary Fig. 62). Collectively, these results demonstrate that D-PM exhibits excellent in vivo biocompatibility, even at doses exceeding those required for bacterial recognition and antibiotic-targeted delivery.

Discussion

Precise bacterial recognition remains a critical bottleneck in addressing the escalating global burden of pathogenic infections. Despite intensive efforts, existing detection and intervention strategies are constrained by limited selectivity and suboptimal efficacy, necessitating innovative concepts and tools to overcome these intrinsic challenges. Here, we introduce a pioneering biomimetic strategy based on the rational design of chirality-specific peptidoglycan mimetics constructed from D-Ala-decorated chitosans. By mimicking the structural and stereospecific features of bacterial peptidoglycan, D-PM introduces a framework in which macromolecular substrates actively integrate into bacterial biosynthetic pathways—a mechanism fundamentally distinct from traditional bacterial targeting approaches. Systematic investigations demonstrated that D-PM achieved high-affinity targeting and robust labeling across a broad spectrum of pathogenic bacteria, including the clinically formidable ESKAPE pathogens, antibiotic-resistant MRSA strains, and clinical isolates. In contrast, natural CS and their L-enantiomeric derivatives of L-PM exhibited negligible bacterial recognition, while conventional small-molecule D-Ala probes displayed inferior labeling efficiency, underscoring the unique specificity and efficacy endowed by the D-PM architecture. Through a combination of biological assays and molecular simulations, we elucidated that D-PM actively integrates into bacterial cell wall biosynthesis, giving rise to distinct biosynthetic intermediates (GlcN-D-Ala-D-Ala and GlcN-D-Ala-D-Ala-L-Lys-D-iGlu-L-Ala-MurNAc), thereby providing a molecular rationale for their high-affinity and durable bacterial labeling. This unveils a previously unrecognized mechanism of pathogen-specific macromolecular recognition, reshaping our understanding of bacterial metabolic targeting. Beyond recognition, D-PM presents a versatile platform for pathogen-specific applications—macromolecular scaffolds of D-PM can be readily conjugated with diverse imaging probes, enabling efficient bacterial imaging and discrimination between prokaryotic and eukaryotic cells; moreover, D-PM undergoes spontaneous self-assembly in aqueous environments to form nanocavities capable of encapsulating antibiotics, thereby facilitating targeted delivery and enhancing antimicrobial efficacy, including antibiotic-resistant strains. In vivo evaluations in both localized and systemic infection mouse models further validated the translational potential of D-PM. In local infection settings, D-PM exhibited deep tissue penetration, precise bacterial imaging, and accelerated antibacterial therapy and wound healing. In systemic infections, D-PM enabled targeted localization of infected organs and markedly improved therapeutic outcomes. Furthermore, D-PM demonstrated favorable safety, underscoring its potential for translation in antimicrobial diagnosis and therapy. Looking forward, the broad-spectrum recognition of D-PM across diverse pathogenic bacteria, including methicillin-resistant Staphylococcus aureus, supports future investigations of the applicability to a wider range of multidrug-resistant pathogens and to bacteria in distinct physiological states, such as dormant and stationary-phase populations. Collectively, this study transcends traditional molecular frameworks for bacterial intervention by introducing macromolecular mimetics that directly incorporate into bacterial biosynthetic machinery. With interdisciplinary innovation spanning biology, chemistry, materials science, and medicine, our findings provide strategies for precise microbial targeting and intervention, offering urgently needed solutions to combat pathogenic infections and address the mounting challenge of antimicrobial resistance.

Methods

Bacterial strains and culture conditions

The bacterial strains utilized in this study are detailed in Supplementary Table 3. The strains were revived from glycerol stocks and stored at −80 °C until further use. For liquid culture, bacterial strains were cultured in tryptic soy broth (TSB) or Luria–Bertani broth (LB) at 37 °C with continuous shaking at 220 rpm. For solid culture, bacteria were plated on LB agar or mannitol salt agar plates and incubated at 37 °C.

Synthesis and characterization of peptidoglycan mimics

Synthesis of chiral peptidoglycan mimics

Chiral peptidoglycan mimics (D-PM and L-PM) were synthesized as follows: 200.0 mg of chitosans (CS, Mw ≈ 5000 Da) and 94.6 mg of Boc-D-Ala, combined with 191.2 mg of 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC.HCl), 115.6 mg of N-hydroxysuccinimide (NHS), and 384.0 μl of N, N-diisopropylethylamine (DIPEA), were dissolved in a 3:7 (v/v) mixture of dimethylformamide (DMF) and H₂O solution. The reaction was stirred for 48 h at room temperature. Afterward, the solvent was removed, and the residue was purified by washing 3 times with methanol and dichloromethane. This step yielded Boc-protected D-PM. To deprotect the Boc group of D-Ala, trifluoroacetic acid (TFA, 1.0 ml) was added to the Boc-protected D-PM, and the mixture was stirred for 6 h. The TFA was evaporated under reduced pressure, and the resultant residue was precipitated in anhydrous diethyl ether. The D-PM product was obtained as a yellow solid (yield: 61.2%). The synthesis of L-PM was performed under identical conditions, yielding a yellow solid (yield: 58.9%).

Characterizations of peptidoglycan mimics

To confirm the modification of chiral Ala, ¹H NMR spectra of Boc-protected L-PM or Boc-protected D-PM were recorded on a Bruker Avance III 400 MHz spectrometer in D₂O, with a sample concentration of 5 mg ml⁻¹. The modification rate of chiral Ala was calculated based on the integral areas of the proton signals corresponding to the sugar rings (δ 3.94–3.10, 5H) and the methyl group of Ala (-CH₃, δ 1.32–1.26, 3H). The modification rate of chiral Ala was also determined using the amine titration method. Fluorescamine—a reagent for detecting primary amines—was employed to assess the amine content in Boc-protected L-PM and Boc-protected D-PM. The fluorescamine solution was prepared in DMSO at a concentration of 1 mg ml⁻¹, containing 2% DIPEA. CS were used to construct a standard curve with concentrations ranging from 0 to 200 μg ml⁻¹ on a 96-well microplate. Fluorescence measurements were taken using a plate reader (SpectraMax M2, Molecular Devices), with excitation at 402 nm and emission at 470 nm. The modification rate of Boc-protected L-PM and Boc-protected D-PM was then calculated based on the standard curve.

Zeta potentials and nanostructures

The zeta potential and size distribution of CS, L-PM, and D-PM were measured using a Zetasizer Nano (ZSE, Malvern Panalytical) at a concentration of 400 μg ml⁻¹. The nanostructures of CS, L-PM, and D-PM were observed using transmission electron microscopy (TEM, JEM-2100Plus, JEOL) after staining with phosphotungstic acid.

Circular dichroism (CD) measurements

CD spectra were conducted for CS, L-PM, and D-PM at a concentration of 0.5 mg ml⁻¹ in aqueous solutions using a CD spectrometer (J-1500, Jasco). The measurement parameters included a data pitch of 0.1 nm, a scanning rate of 100 nm min⁻¹, a bandwidth of 1.00 nm, and a wavelength range from 190 to 260 nm.

Fluorescent PM constructions

Peptidoglycan mimics possess abundant amino groups, making them suitable for bioconjugation with fluorescent molecules such as FITC, TAMRA, and Cy5, which facilitates the development of imaging probes. To construct fluorescent D-PM, 20.0 mg of D-PM was dissolved in 1 ml of water, and the pH was adjusted to 8.0 using 1 M NaOH. The fluorescent molecule solution was added dropwise to the D-PM solution, and the reaction was stirred in the dark for 12 h. The resulting fluorescent D-PM was purified by dialysis (MWCO 2000) in water, followed by freeze-drying. The preparation of fluorescent CS and fluorescent L-PM was carried out using the same procedure. To ensure uniform fluorescence intensity across the experimental groups, unlabeled PM solutions of equivalent concentration were used to adjust and standardize the fluorescence intensities. Fluorescent PM labeled with FITC, TAMRA, and Cy5 were excited at wavelengths of 488 nm, 544 nm, and 637 nm, respectively. Emission spectra were collected between 460 and 760 nm. The fluorescent stability of FITC-labeled CS, FITC-labeled L-PM, and FITC-labeled D-PM was assessed at a concentration of 0.02 mg ml^-1 in an aqueous solution. The samples were incubated in the dark at room temperature for different time points. Fluorescence intensity was measured using a plate reader, with an excitation wavelength of 488 nm and an emission wavelength of 520 nm.

Bacterial recognition ability and mechanisms

Flow cytometry assays

Exponential-phase S. aureus ATCC29213 was harvested and resuspended in 10 mM PBS to achieve an optical density at 595 nm (OD₅₉₅) of 0.3. The bacteria were incubated with FITC-labeled CS, L-PM, or D-PM at specific concentrations for designated periods. Following incubation, the bacteria were collected by centrifugation at ~9000 g for 5 min at 4 °C, washed, and resuspended in 10 mM PBS. The treated bacteria were subsequently analyzed using flow cytometry (CytoFLEX, Beckman Coulter). The bacterial recognition of chiral PM was assessed using the same treatment protocols for all pathogenic Gram-positive and Gram-negative bacteria. All representative gating strategies are shown in Supplementary Fig. 63.

CLSM observation

Exponential-phase S. aureus ATCC29213 was harvested and resuspended in 10 mM PBS to achieve an OD₅₉₅ of 0.3. The S. aureus was incubated with FITC-labeled CS, L-PM, or D-PM at a concentration of 0.5 mg ml⁻¹ for 1 h. After washing off unbound fluorescent PM, the bacterial nuclei were stained with a 1 μg ml⁻¹ Hoechst 33342 solution for 15 min. The fluorescence signals of the fluorescent PM located with the bacteria were observed using CLSM (Stellaris 5, Leica).

MST assays

The binding affinity of CS, L-PM, and D-PM to S. aureus ATCC29213 was determined using MST (Monolith NT.115, NanoTemper). S. aureus (OD₅₉₅ = 1.0) was stained with 10 μg ml⁻¹ FM4-64 dye in 10 mM PBS. Serial dilutions of CS, L-PM, and D-PM were prepared to create a concentration gradient and incubated with S. aureus for 10 min at room temperature. The mixtures were loaded into capillaries, and MST measurements were conducted and analyzed using Nanotemper Analysis software.

UPLC-MS assays

To investigate the effect of D-PM on cell wall biosynthesis, vancomycin was used as a positive control based on the accumulation of UDP-MurNAc-pentapeptide in cultures. Exponential-phase S. aureus ATCC29213 (OD₅₉₅ = 0.3) grown in TSB was pretreated with 130 μg ml⁻¹ chloramphenicol for 15 min at 37 °C. Subsequently, 1 mg ml⁻¹ of CS, L-PM, and D-PM, along with 10 μg ml⁻¹ vancomycin, were added to the S. aureus cultures. After 1-h incubation, the S. aureus was harvested, and the pellets were resuspended in 500 μl of aqueous solution. The cell suspension was boiled for 30 min and then centrifuged at ~14,000 g for 15 min. The supernatant was analyzed using UPLC-MS (Orbitrap Exploris 120, Thermo), and the data were analyzed by FreeStyle (Thermo Fisher Scientific, Xcalibur Version 4.2). By inputting the chemical formula of UDP-MurNAc-pentapeptide into FreeStyle, the targeted molecular weight of 1,149.3516 was used to capture the molecular signal in UPLC-MS results.

To reveal the mechanism of D-PM intervention in synthesis, biosynthetic intermediates of GlcN-D-Ala-D-Ala and GlcN-D-Ala-D-Ala-L-Lys-D-iGlu-L-Ala-MurNAc were tracked and studied. S. aureus (OD₅₉₅ = 0.3) was treated with CS, L-PM, or D-PM. The supernatant was then analyzed by UPLC-MS to detect and analyze the molecular signals of GlcN-D-Ala-D-Ala (Mw = 321.1536) and GlcN-D-Ala-D-Ala-L-Lys-D-iGlu-L-Ala-MurNAc (Mw = 924.4288).

Molecular docking and dynamics simulation

Molecular docking studies were conducted to predict the binding interactions between peptidoglycan mimics and key proteins involved in peptidoglycan biosynthesis, specifically Ddl and MurF. The three-dimensional structures of these proteins were obtained using AlphaFold Protein Structure Database, and docking simulations were performed using AutoDock Vina 1.2.3. The docking scores were computed using the built-in scoring functions of AutoDock Vina, which evaluate binding affinities by considering intermolecular interactions. Following the docking procedure, molecular dynamics (MD) simulations were carried out with GROMACS 2023.2, employing the CHARMM36m force field and the TIP3P water model. The force field parameters for D-Ala, D-Ala-D-Ala, GlcN-D-Ala, and GlcN-D-Ala-D-Ala were generated using CGenFF. Hydrogen bonds were constrained using the LINCS algorithm, with an integration step of 2 fs, and nonbonded interactions were truncated at 12 Å. The resulting data were visualized using PyMOL 3.0.4.

Biosynthetic inhibition

Exponential-phase S. aureus ATCC29213 (OD₅₉₅ = 0.3) was resuspended in TSB and pretreated with bacterial biosynthetic inhibitors, including Dcy (a D-Ala-D-Ala ligase inhibitor), Amo (penicillin-binding protein transpeptidase inhibitor), Mer (penicillin-binding protein transpeptidase inhibitor), Tet (a protein synthesis inhibitor), and Cip (a DNA synthesis inhibitor), all at a concentration of 100 μg ml⁻¹ for 2.5 h. After the removal of the inhibitors, the pretreated S. aureus (OD₅₉₅ = 0.3) was incubated with 0.5 mg ml⁻¹ FITC-labeled D-PM in 10 mM PBS for 1 h to analyze bacterial recognition using flow cytometry. The un-pretreated group served as a positive control, allowing for a comparison of bacterial recognition in the presence of each inhibitor.

D-PM accumulation in bacterial biosynthesis

Chirality-dependent bacterial accumulation

The accumulation of fluorescent peptidoglycan mimics in bacteria was detected after multiple rounds of incubation with peptidoglycan mimics. Exponential-phase S. aureus ATCC29213 (OD₅₉₅ = 0.003) was collected and resuspended in an LB medium. After incubation with FITC-labeled L-PM or FITC-labeled D-PM at a concentration of 0.5 mg ml⁻¹ for 1 h, unbound L-PM and D-PM were removed by washing, yielding Passage 1 (P1) S. aureus. P1 S. aureus was then subjected to a second incubation with FITC-labeled L-PM or FITC-labeled D-PM for 1 h, followed by washing to remove unbound PM, resulting in P2 S. aureus. This process was repeated to obtain P3 S. aureus. The accumulation of chiral PM into each passage of S. aureus biosynthesis was quantified by flow cytometry. Exponential-phase S. aureus ATCC29213 was harvested and diluted in PBS to an OD₅₉₅ of 0.3. The bacterial suspensions were then incubated with CS, L-PM, or D-PM at a concentration of 0.5 mg ml⁻¹ for 1 h. After incubation, the zeta potential of each sample was measured using a Zetasizer Nano instrument.

3D-SIM for PM hierarchical accumulation

Super-resolution 3D-SIM (N-SIM S, Nikon) was utilized to visualize the hierarchical incorporation of D-PM into the bacterial cell wall biosynthesis. The bacterial incubation and treatment protocols were consistent with those described in the flow cytometry assay for studying chirality-dependent bacterial accumulation. To observe the hierarchical accumulation of PM into the bacterial cell wall, P1 S. aureus ATCC29213 was derived from incubation with Cy5-labeled L-PM or Cy5-labeled D-PM, and P2 S. aureus was observed following incubation with TAMRA-labeled L-PM or TAMRA-labeled D-PM, and P3 S. aureus was examined after incubation with FITC-labeled L-PM or FITC-labeled D-PM. Each passage of S. aureus (P1, P2, and P3) was observed using 3D-SIM to assess the hierarchical accumulation of PM into the bacterial cell wall.

SEM for morphology observation

Exponential-phase S. aureus ATCC29213 was incubated with CS, L-PM, or D-PM for three consecutive passages to assess morphological alterations. Following treatment, the bacteria were thoroughly rinsed with PBS and sequentially dehydrated in a graded series of ethanol solutions. After dehydration, the samples were air-dried and sputter-coated with gold to enhance conductivity. The bacterial surface morphology was then observed using SEM (MIRA4 LMH, Tescan).

Propidium iodide (PI) influx assay

Exponential-phase S. aureus ATCC29213 (OD₅₉₅ = 0.003) was harvested and resuspended in LB medium. The bacterial suspension was incubated with CS, L-PM, or D-PM at 0.5 mg ml⁻¹ over three consecutive passages. Following treatment, bacteria were stained with 10 μg ml⁻¹ PI for 15 min in the dark, and fluorescence signals were quantified by flow cytometry.

CFU count assays

Exponential-phase S. aureus ATCC29213 was diluted in PBS to a final concentration of 5 × 10⁵ CFU ml⁻¹. The bacterial suspension was incubated with CS, L-PM, or D-PM at a concentration of 1 mg ml⁻¹ for 1 h to generate Passage 1 (P1). A 10 μl aliquot from each suspension was plated onto mannitol salt agar. P1 bacteria were then subjected to a second round of incubation with the same materials under identical conditions for 1 h to obtain Passage 2 (P2), followed by plating of a 10 μl aliquot. This incubation–plating cycle was repeated once more to generate Passage 3 (P3). All agar plates were incubated at 37 °C overnight, after which colony-forming units (CFUs) were enumerated.

Targeted delivery systems based on D-PM

Differentiation of bacteria and mammalian cells

NIH 3T3 fibroblasts and S. aureus ATCC29213 were co-cultured in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS). FITC-labeled CS, L-PM, or D-PM (0.5 mg ml⁻¹, equal fluorescence intensity) were added to the co-culture system and incubated at 37 °C for 1 h. For flow cytometry, unbound fluorescent macromolecules were removed by washing with PBS. S. aureus was collected by centrifugation, and NIH 3T3 cells were enzymatically detached using 0.25% trypsin. For CLSM imaging, NIH 3T3 cells and S. aureus were stained with Hoechst 33342 (1 μg ml⁻¹) for 10 min after removing unbound macromolecules.

Preparation of Tet-loaded D-PM

Tet (20 mg ml⁻¹ in 50 μl DMSO) was mixed with D-PM (20 mg ml⁻¹ in 250 μl water) and sonicated in a water bath. The mixture was then added dropwise into 700 μl water under ultrasonic conditions, obtaining nanoparticulate Tet-loaded D-PM (T/D-PM). Unencapsulated Tet and residual organic solvents were removed by dialysis (MWCO: 2000 Da) against distilled water for 24 h. And the solution was freeze-dried to obtain the final T/D-PM formulation. Other formulations, including T/CS and T/L-PM, were prepared using the same procedure. Encapsulation efficiency and drug loading content were quantified by UV–Vis spectrophotometry (UV-1280, Shimadzu) at 360 nm, based on a standard curve generated from known Tet concentrations. Pyrene was used as a fluorescent probe to evaluate the loading capacity of D-PM for guest molecules. Particle size distribution and zeta potential were measured by Zetasizer Nano. The nanostructure of T/D-PM was visualized using TEM.

Antibacterial activity of T/D-PM

The minimum inhibitory concentrations (MICs) of free Tet, T/CS, T/L-PM, and T/D-PM were determined using the standard broth microdilution method following Clinical and Laboratory Standards Institute guidelines. The exponential-phase bacteria were diluted in fresh TSB to a final concentration of 5 × 10⁵ CFU ml⁻¹ to prepare the working suspension. A 100 μl aliquot was added to each well of a 96-well plate. Free Tet, T/CS, T/L-PM, or T/D-PM were then added to the first row at designated initial concentrations (final volume: 200 μl), followed by a series of 2-fold serial dilutions across the plate. After incubation at 37 °C for 24 h, the MIC was defined as the lowest antibiotic concentration that completely inhibited visible bacterial growth. To further evaluate bactericidal activity, PI staining and CFU assays were conducted. Exponential-phase S. aureus cultures were diluted to OD₅₉₅ = 0.3 in PBS and treated with each formulation containing Tet at a final concentration of 2 μg ml⁻¹ for 1 h at 37 °C. Following treatment, bacterial suspensions were incubated with PI (10 μg ml⁻¹) for 15 min in the dark, followed by CLSM to visualize PI-positive (dead) bacteria. For CFU assays, exponential-phase S. aureus ATCC29213 and ATCC43300 were diluted in PBS to 5 × 10⁵ CFU ml⁻¹. S. aureus ATCC29213 was incubated with free Tet, T/CS, T/L-PM, or T/D-PM (equivalent to 2 μg ml⁻¹ Tet) for 1 h, while S. aureus ATCC43300 was treated with the same formulations at 0.25 μg ml⁻¹ Tet for 2 h. Each suspension (10 μl aliquot) was plated onto mannitol salt agar, and CFUs were enumerated after overnight incubation at 37 °C.

Bacterial internalization of T/D-PM

Exponential-phase S. aureus ATCC29213 (OD₅₉₅ = 0.3) was treated with different formulations (equivalent to 2 μg ml⁻¹ Tet) for 1 h. After treatment, S. aureus was stained with Hoechst 33342 (1 μg ml⁻¹, 15 min) to visualize nucleic acids. The intracellular localization of Tet was examined by CLSM using excitation at 488 nm to detect the intrinsic fluorescence of Tet.

Animal study

All animal experiments were approved by the Laboratory Animal Ethics Committee of Hunan University (SYXK(XIANG)2018-0006). Ethical guidelines related to animal experiments were strictly followed. Animals were housed in a controlled environment with a temperature of 20–26 °C, humidity of 40–70%, and a 12 h light/dark cycle.

D-PM for local infection imaging and treatment

Following anesthesia, circular full-thickness wounds (~80 mm²) were created on the dorsal skin of female BALB/c mice (6–8 weeks old), followed by inoculation with FITC-D-PM-labeled S. aureus ATCC29213 (1 × 10⁶ CFU). The infected mice were subsequently treated topically at the wound site with PBS or Cy5-labeled CS, L-PM, or D-PM (1 mg per kg body weight). After 4 h, wound tissues were collected, embedded, cryosectioned (~10 μm), and stained with Hoechst 33342 for CLSM analysis. Uninfected mice also received Cy5-labeled D-PM as a control. The Pearson’s correlation coefficient (r) for the colocalization between bacteria (green) and peptidoglycan mimics (red) was quantified using Fiji/ImageJ (version 1.54f).

For antibacterial efficacy evaluation, the same wound infection model was employed. After 4 h of infection, the mice were treated with PBS, free Tet, T/CS, T/L-PM, and T/D-PM at a dosage of 10 mg Tet per kg of body weight. Treatments were administered topically every 12 h for a total of 2 doses. Throughout the treatment period, the wound area and the body weight of infected mice were closely monitored. At designated time points, the infection wounds were swabbed under sterile conditions, and the bacteria were suspended in PBS and plated on mannitol salt agar for CFU enumeration. Following euthanasia, wound tissues were fixed in 4% polyformaldehyde, paraffin-embedded, and sectioned for histological analysis. Gram staining assessed bacterial burden, while the stainings by H&E, Masson, picrosirius red, and CD31 immunofluorescence were performed to evaluate tissue regeneration and neovascularization.

D-PM for systemic infection imaging and treatment

A mouse peritonitis model (female BALB/c) was established by intraperitoneal injection of S. aureus ATCC29213 (2.5 × 10⁷ CFU in 100 μl PBS) into the right side of the abdomen. After 1 h of infection, Cy5-labeled CS, L-PM, or D-PM (5 mg Tet per kg body weight) were administered intraperitoneally into the left side of the abdomen. Control animals received the treatment with Cy5-labeled D-PM in the absence of infection. In vivo imaging was conducted at 10 min, 1 h, 2 h, and 4 h post-administration using an in vivo imaging system (IVIS Spectrum CT, PerkinElmer). Major organs were harvested and subjected to ex vivo fluorescence imaging. Fluorescence intensity was quantified using Living Image software. Bacterial burdens in tissues were determined by plating tissue homogenates on mannitol salt agar. These tissues were embedded, cryosectioned, and mounted as described for wound samples, followed by CLSM imaging to assess the tissue distribution of Cy5-labeled CS, L-PM, or D-PM.

For antibacterial efficacy assessment, female BALB/c mice were intraperitoneally inoculated with 100 μl of bacterial suspension containing S. aureus ATCC29213 (2.5 × 10⁷ CFU), S. aureus ATCC43300 (MRSA, 2.5 × 10⁷ CFU), or E. coli ATCC25922 (1.0 × 10⁷ CFU) in PBS. After 1 h, mice were treated with PBS, free Tet, T/CS, T/L-PM, or T/D-PM. The administered dose was 10 mg Tet per kg body weight for S. aureus ATCC29213 and MRSA ATCC43300 infections, and 20 mg Tet per kg body weight for E. coli ATCC25922 infection, delivered intraperitoneally every 12 h for a total of two doses. After 24 h of the first treatment, mice were euthanized, and ascitic fluid, along with major organs, were aseptically collected. Samples were homogenized, serially diluted, and plated on agar for CFU enumeration.

Safety evaluation of D-PM

D-PM suspended in PBS was administered intraperitoneally to female BALB/c mice at 200 or 500 mg D-PM per kg body weight, with PBS alone serving as the control. Body weights were recorded daily throughout the study. On day 7 post-administration, whole blood and plasma were collected for routine hematological analysis and plasma biochemical profiling, including immunoglobulinG (IgG), alanine transaminase (ALT), aspartate transaminase (AST), and blood urea nitrogen (BUN). Mice were then euthanized, and major organs (heart, liver, spleen, lungs, and kidneys) were excised, fixed overnight in 4% paraformaldehyde, paraffin-embedded, and sectioned for hematoxylin and eosin (H&E) staining and histological examination.

Statistical information

Statistical significance between two groups was assessed using a two-tailed unpaired t-test. All experiments were performed with at least three independent biological replicates. Data are presented as mean ± s.d. Differences were considered statistically significant at P < 0.05 (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001); P > 0.05 was considered not significant (NS).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

X.X. and K.D. conceived and designed the study. K.D., D.Z., and Y. L. synthesized the chiral peptidoglycan mimics. K.D., D.Z., and Z.Z. investigated bacterial recognition and the incorporation mechanisms of peptidoglycan mimics into bacterial biosynthesis. K.D., W.G., and D.Z. conducted molecular dynamics simulations. K.D., D.Z., Z.Z., B.G., and Y.X. contributed to the in vivo animal studies. X.X. and K.D. wrote and revised the manuscript. K.D., D.Z., Z.Z., B.G., Y. L., and X.X. provided critical feedback on the manuscript. X.X. supervised the project.

Peer review

Peer review information

Nature Communications thanks Dicky Pranantyo, Wei Chen, and the other anonymous reviewer for their contribution to the peer review of this work. A peer review file is available.

Data availability

All data supporting the findings of this study are available within the Article and its Supplementary Information. Source data are provided for Figs. 1–6, and for Supplementary Figs. 4, 5, 7–13, 16, 18–21, 23–29, 31, 33–39, 41–43, 45, 47, 49, and 51–61 in the associated Source Data file. Protein structures were retrieved from the AlphaFold Protein Structure Database: Ddl (P96612, https://alphafold.ebi.ac.uk/entry/P96612) and MurF (P96613, https://alphafold.ebi.ac.uk/entry/P96613). Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-69967-z.