Transformable nano-antibiotics for mechanotherapy and immune activation against drug-resistant Gram-negative bacteria

Abstract

The dearth of antibiotic candidates against Gram-negative bacteria and the rise of antibiotic resistance create a global health concern. The challenge lies in the unique Gram-negative bacterial outer membrane that provides the impermeable barrier for antibiotics and sequesters antigen presentation. We designed a transformable nano-antibiotics (TNA) that can transform from nontoxic nanoparticles to bactericidal nanofibrils with reasonable rigidity (Young’s modulus, 21.6 ± 5.9 MPa) after targeting β-barrel assembly machine A (BamA) and lipid polysaccharides (LPSs) of Gram-negative bacteria. After morphological transformation, the TNA can penetrate and damage the bacterial envelope, disrupt electron transport and multiple conserved biosynthetic and metabolic pathways, burst bacterial antigen release from the outer membrane, and subsequently activate the innate and adaptive immunity. TNA kills Gram-negative bacteria in vitro and in vivo with undetectable resistance through multiple bactericidal modes of action. TNA treatment–induced vaccination results in rapid and long-lasting immune responses, protecting against lethal reinfections.

Physical disruption of bacterial envelope effectively avoids bacterial resistance and reduces mortality from reinfection.

INTRODUCTION

Many Gram-negative bacteria present acquired resistance against microbicides, such as carbapenem or third-generation cephalosporins (1–4), in treating hospital-acquired infections (5–8). In particular, resistance against colistin, an antibiotic of the last resort, is on the rise globally (9), making it urgent to develop novel antibiotics against Gram-negative microbes. The dearth of antibiotic candidates against Gram-negative bacteria is largely due to the presence of the unique outer membrane not existing in Gram-positive bacteria (10). The load bearing and stiff outer membrane provide an impermeable barrier for antibiotics and antimicrobial peptides (10), and also sequester antigen presentation to dampen host immune response by camouflaging their antigen epitopes or secreting vesicles (1, 11). Therefore, to develop a strategy that can not only break up the outer membrane but also promote antigen presentation would be the most effective way to combat against drug-resistant Gram-negative bacteria.

To realize such an effective strategy, the key is to precisely target and damage the outer membrane, and burst the envelop antigen release to promote antigen presentation for subsequent immune activation. For disrupting the outer membrane, amphiphilic cationic peptides (ACPs), which destroy the bacterial membrane by forming channels, have less clinical uses because the antibacterial effect induced by electrostatic interaction leads to poor selectivity (12, 13). Molecular rotor, another membrane disruption strategy, opens the lipid bilayer by rotating and changing their conformation with ultraviolet (UV) light irradiation (14, 15), but the application is limited by the tissue penetration depth of ultraviolet light (16, 17). Studies including our previous results have shown that the in situ transformation of peptide analogs from nontoxic nanoparticles to nanofibrils quickly destroy cell membrane of cancer cells and even subcellular organelles (18–21). For antibacterial purpose, although human defensin-6 or its analog self-assembles into nanofibrous networks to enclose bacteria, disruption of outer membrane is not involved (22, 23). On the other hand, bacterial immunotherapy based on organic nanostructures is emerging (24–27). Assembly of nanofibrils from peptide/protein has been found to regulate the phenotype of immune cells (28–32), astrocytes (33), and epithelial-like cells (34). Therefore, the outer membrane targeted in situ morphology transformation of nano-antibiotics is anticipated to be but one effective approach for outer membrane disruption and antigen release, and the newly formed peptide nanofibrils and released antigens should enable the subsequent activation of host innate and adaptive immune system.

Here, we designed an in situ shape transformable nano-antibiotics (TNA) to mechanically disrupt the outer membrane and activate innate and adaptive immune system of the host for treatment of Gram-negative bacteria without developing drug resistance (Fig. 1). TNA is designed to form nanoparticles in the solution and target two types of conserved molecules of the outer membrane, lipopolysaccharide (LPS) and β-barrel assembly machine A (BamA) (35–41). This recognition induces the transformation from nanoparticles to nanofibrils with reasonable rigidity and diffusivity to effectively impair the envelope of pan–drug-resistant and colistin-resistant (PDR) Acinetobacter baumannii (PDR A. baumannii). This nanomechanical action of TNA is found to interfere with electron transport chain (ETC) complex I and disrupt multiple conserved biosynthetic and metabolic pathways, resulting in undetectable drug resistance after ~500 generations by repeated TNA stimulation of bacteria. The TNA disruption bursts the bacterial antigen release to activate the innate and adaptive immune system, including M1-like polarization of macrophages, maturation of dendritic cells (DCs), proliferation of CD4⁺ and CD8⁺ T cells, and acquisition of memory B cells during the acute infection stage. In murine models of superficial wound and peritonitis infection, immunization with TNA treatment offers substantial protection and improves survival after reinfection.

Fig. 1. Illustration of in situ transformation of TNA from nontoxic nanoparticles to supramolecular nanofibrils for mechanotherapy as well as immune activation to fight against drug-resistant Gram-negative bacteria.

The rigid TNA nanofibrils triggered by LPS and BamA can mechanically penetrate the outer membrane, disrupt electron transport and multiple conserved biosynthetic and metabolic pathway needed for bacteria to survive, burst bacterial antigen release from the outer membrane, and subsequently trigger the innate and adaptive immunity. Mice that have been vaccinated with PDR A. baumannii infection and TNA treatment can effectively repel PDR A. baumannii reinfection through a combination of pathogen-specific antibodies and memory B cells.

RESULTS

Design rationale of TNA

The TNA molecule, a peptide analog Py-F₂K₃-Sul, is composed of four individual functional motifs (Fig. 2A): (i) Py (pyrene) is a hydrophobic core to induce the formation of micellar nanoparticles, and the aggregation-induced excimer emission is convenient for fluorescence observation; (ii) F₂ (phenylalanine dipeptide) is the sequence of the β sheet–forming peptide (42); (iii) K₃ (lysine tripeptide) is for the outer membrane LPS recognition (Fig. 2B) (43); (iv) Sul (sulfanilamide) is for the outer membrane BamA recognition (Fig. 2C). On the basis of the design rationale, TNA molecules are capable of (i) assembling into TNA nanoparticles in aqueous solution and (ii) transforming into rigid TNA nanofibrils on binding to LPS and BamA (35–41), two well-conserved molecules at the outer membrane of Gram-negative bacteria.

Fig. 2. Design of TNA molecule and its morphology transformation.

(A) Structure of TNA molecule and its analogs. (B and C) Molecular dynamics simulation of the binding of TNA molecule to (B) LPS and (C) BamA. (D) Transmission electron microscopic (TEM) images of initial TNA nanoparticles and the transformation into nanofibrils after interaction with BamA and LPS at 5, 20, and 40 min. The TEM image is representative of three independent experiments.

LPS and BamA co-induce fibrotic transformation of TNA nanoparticles

The binding of TNA to LPS and BamA is governed by K₃ and Sul motif based on the molecular dynamics (MD) simulation. The binding between TNA molecule and LPS is mainly instructed by the hydrophilic K₃ motif with the binding free energy of −9.04 kcal mol⁻¹ (Fig. 2B, fig. S1, and table S1). The ligand-receptor binding between TNA and BamA should be attributed to the Sul motif with binding free energy of −7.05 kcal mol⁻¹ (Fig. 2C, figs. S2 and S3, and table S2), and the main binding sites are located at Phe⁴, Asn⁵, Met³³⁶, Pro³³⁸, Leu³³⁹, Pro³⁴¹, and Ile³⁶⁵ in the β-barrel structure of BamA (Fig. 2C and fig. S2). Obvious deformation of the β-barrel structure of BamA can be seen after binding with the TNA molecule (Fig. 2C). Replacing Sul motif by an acetyl group in the TNA molecule does not cause observable structural change of the BamA, and the total binding free energy changes from −63.5 to −19.6 kcal mol⁻¹ (fig. S4 and table S3). Because the amphiphilic structure of TNA molecule drives the molecule to assemble into nanoparticles in aqueous solution (44), the density of hydrophilic K₃ and Sul motif is high on the TNA nanoparticle surface, thereby enhancing the Gram-negative bacteria binding ability.

TNA molecule and analogs were synthesized using standard fluorenylmethyloxycarbonyl (Fmoc) solid-phase peptide chemistry, with characterization details of the molecule provided in the Supplementary Materials (figs. S5 to S12). TNA molecule can easily self-assemble to TNA nanoparticles at concentrations above the critical aggregation concentration (CAC) (Fig. 2D and figs. S13 to S15). TNA nanoparticles (21.2 ± 3.7 nm) in the presence of only LPS or BamA retained the original morphology with the diameter of 45.6 ± 12.9 nm and 146.1 ± 45.7 nm for the presence of LPS and BamA (Fig. 2D and fig. S16), respectively; no fibrosis transformation was observed even after 24 hours (fig. S17). TNA nanoparticles transformed into beaded fibrils 5 min after incubation with LPS and BamA, followed by further assembly into longer than 200-nm straight fibrils at 40 min (Fig. 2D and figs. S16 and S18), corroborating the synergistic effect of LPS and BamA on enabling the fibrotic transformation of TNA nanoparticles.

LPS and BamA formed a copolymer with TNA in the transformation of TNA from nanoparticles to nanofibrils, and the transformed nanofibrils are characterized by pyrene-pyrene stacking and β-sheet conformation. We prepared the transformed nanofibrils by incubating fluorescein isothiocyanate (FITC)–labeled LPS (FITC-LPS) and rhodamine B isothiocyanate–labeled BamA (rhodamine B-BamA) with TNA. Emission spectra of FITC-LPS and rhodamine B-BamA appeared in the TNA nanofibril (fig. S19), indicating that both LPS and BamA are in the transformed nanofibrils. In the nanofibrils, fluorescence anisotropy of both FITC-LPS and rhodamine B-BamA (fig. S19) substantially increased, compared with that of themselves in the absence of TNA. The enhanced pyrene excimer emission at 460 to 470 nm indicates the substantially increased content of excimers due to pyrene-pyrene stacking in the TNA nanofibrils. In the Fourier transform infrared (FTIR) spectra (fig. S21), TNA nanoparticles presented the main peak at 1678 cm⁻¹, which may be assigned to the sterically constrained non–hydrogen-bonded amide C═O groups; the shoulder peak at 1631 cm⁻¹ shows that β sheets are present in the TNA nanoparticles. The transformed TNA assembly presented the main peak at 1636 cm⁻¹, which may be assigned to the amide I bands from the β sheet, indicating that more β sheets are formed with the morphology transformation after incubating TNA with LPS and BamA; the shoulder peak around 1661 cm⁻¹ may be attributed to the several turn structures (45, 46). In the circular dichroic (CD) spectra (fig. S22), a typical negative peak of β sheet at 216 nm was observed after incubating TNA with BamA and LPS.

TNA nanoparticles transform into nanofibrils and kill Gram-negative bacteria

The bacteria recognition triggers the rapid morphology transformation from TNA nanoparticles to nanofibrils. The enhanced pyrene excimer fluorescence at 460 nm due to the transformation was observed 3 s after mixing TNA (20 μM) solution with PDR A. baumannii (Fig. 3A). TNA nanoparticles covered PDR A. baumannii within 1 min (Fig. 3, B to D, and fig. S23), continuously gathered on the bacteria, transformed into short rod-like fibrils at 5 min, and formed long fibrils on bacteria at 10 min (Fig. 3, C and D and figs. S23 and S24). In the morphology transformation, the content of β-sheet conformation increased substantially from 17.0% to 86.4% according to CD measurements (fig. S25 and table S4).

Fig. 3. Morphological characterization of fibrillar-transformable TNA nanoparticles incubated with PDR A. baumannii or A. baumannii.

(A) Fluorescence image (inset, under 365-nm UV light) and fluorescence spectra of TNA, PDR A. baumannii, and TNA-treated PDR A. baumannii. (B) Schematics of in situ structure transformation of TNA on Gram-negative bacteria. (C) Scanning electron microscope (SEM) and (D) thin-section bio-TEM images of untreated A. baumannii/PDR A. baumannii (control) and A. baumannii/PDR A. baumannii treated with TNA for 1, 5, 10, and 20 min. The green and red arrows indicate holes and TNA nanofibrils inserted in the outer membrane, respectively. Experiments were repeated three times. The TNA concentration for treatment is 20 μM.

TNA presents transformability on all tested Gram-negative bacteria rather than Gram-positive bacteria. Fibril formation was observed when TNA nanoparticles were incubated with resistant or nonresistant Gram-negative bacteria, including PDR A. baumannii, A. baumannii, Pseudomonas aeruginosa, Klebsiella pneumoniae, Escherichia coli, and Salmonella typhimurium (fig. S26). Particularly, when TNA nanoparticles were incubated with the mixture of PDR A. baumannii and the typical Gram-positive bacterium Staphylococcus aureus, obvious fibrous networks were observed on the surface of PDR A. baumannii but not S. aureus (fig. S26). This selectivity is attributed to the conserved LPS and BamA molecules on the unique outer membrane of Gram-negative bacteria, while the outer membrane does not exist on Gram-positive bacteria.

TNA presents reasonable activity at ≥4 μM, minimum inhibitory concentration (MIC), against a range of Gram-negative bacteria including important drug-resistant pathogens, such as colistin-resistant, extended-spectrum β-lactamase and carbapenem-resistant, and aminoglycoside-resistant A. baumannii and P. aeruginosa (Table 1). With TNA concentration at 4 μM, nanofibrils can still form in the presence of LPS and BamA or in the presence of PDR A. baumannii (figs. S27 and S28). TNA is effective in killing PDR A. baumannii with a minimal bactericidal concentration (MBC) of 12 μM. The invading experiment of human embryonic kidney 293 cells (HEK293T) by PDR A. baumannii validated that TNA can protect the host cells from Gram-negative bacteria attack due to its rapid fibrous formation and bactericidal ability. The percentage of bacteria intrusion for HEK293T cells decreased from 38.4% to 4.3% and the cell viability of HEK293T remained as high as 98.5%, when the cell culture medium was treated with 20 μM TNA (fig. S29). TNA nanoparticles were barely observed in HEK293T cells, indicating that the TNA has little effect on human cells.

Table 1. Antimicrobial activities.

	Concentration (μM)
	TNA	Colistin
Gram-negative bacteria (MIC)
PDR A. baumannii*	4	>128
A. baumannii ATCC 19606	2	0.25
P. aeruginosa ATCC 27853	2	1
P. aeruginosa DK2	4	>128
K. pneumoniae ATCC 700603	4	0.5
E. coli ATCC 25922	2	0.5
E. coli SHBCC D24652	2	2
E. coli EIEC	2	1
S. typhimurium CGMCC1.1190	4	0.5
Gram-positive bacteria (MIC)
S. aureus ATCC 23235	>128	>128
Human cell line (IC50)
HUVEC	>128	64
HEK293T	>128	>128
L-O2	>128	>128

*Hospital isolate, Q. H. Zou laboratory collection.

Rigidity of nanofibrils influences mechanical disruption of the Gram-negative bacterial envelope

The in situ transformed rigid TNA nanofibrils is required for disrupting the outer membrane of Gram-negative bacteria. We adjusted the fibrils rigidity of TNA by changing Py-F₂, keeping the recognition motif K₃-Sul. Replacing F₂ with glycine dipeptide (G₂) (Fig. 2A), Py-G₂K₃-Sul (N-TNA) nanoparticles did not undergo fibrotic transformation in the presence of LPS and BamA (Fig. 4A and fig. S30), indicating the importance of F₂ for morphological transformation. Although the Young’s modulus, measured by atomic force microscopy (AFM), of nontransformable N-TNA nanoparticles (15.4 ± 3.8 MPa) was higher than that of the outer membrane (1.7 ± 0.1 MPa), N-TNA was not toxic to bacteria (Fig. 4, A to E), indicating the merit of in situ fibrous transformation for killing PDR A. baumannii. Replacing Py with less hydrophobic acetyl (Ac) (Fig. 2A), Ac-F₂K₃-Sul (S-TNA) nanoparticles transformed into nanofibrils with the Young’s modulus (0.7 ± 0.1 MPa) smaller than that of TNA nanofibrils (21.6 ± 5.9 MPa) and particularly the outer membrane (1.7 ± 0.1 MPa), thus was not able to disrupt the PDR A. baumannii envelope (Fig. 4, A to E, and figs. S31 and S33). Replacing Py with bis-pyrene (Bipy) (Fig. 2A), which has much strong π-π staking and hydrophobic interactions, Bipy-F₂K₃-Sul (R-TNA) transformed into highly rigid fibrils (272.3 ± 20.4 MPa), but showed substantially lower outer membrane disruption ability than that of TNA (Fig. 4, A to E, and figs. S32 and S33).

Fig. 4. Rigidity of TNA influences mechanical disruption of the Gram-negative bacteria envelope.

(A) Transformability and rigidity of TNA nanoparticles and analogs. Data are mean ± SD (n = 3). (B) SEM images and (C) thin-section bio-TEM images of untreated (control) and TNA or analog-treated PDR A. baumannii. (D) Height and stiffness (Young’s modulus) of PDR A. baumannii treated with TNA or its analogs measured by atomic force microscopy (AFM) in stroke-physiological saline solution. (E) Time-dependent killing of PDR A. baumannii by TNA or analogs. Data are mean ± SD (n = 3). (F) Confocal imaging of TNA/R-TNA (1 mM)–treated large unilamellar vesicle (LUV) preloaded with calcein (green). (G) Cantilever force versus indentation distance on PDR A. baumannii treated with TNA or its analogs, as in Fig. 3D. Force curves were registered with respect to the onset of force increase as the cantilever was lowered. The dose for TNA and its analogs is 20 μM for (B) to (E), and the treatment duration is 20 min for (B) to (E). Experiments in this figure were repeated three times.

Both TNA and R-TNA are rigid enough to penetrate the outer membrane of PDR A. baumannii, but presented different bactericidal effects (Fig. 4, B, C, and E), which might be attributed to the different ability in disrupting the inner membrane of the bacteria. The inner membrane is a fluid phospholipid bilayer, and ACPs can interact with the phospholipid bilayer via electrostatic interaction. The Shai-Matsuzaki-Huang (SMH) model, a classical model explaining the damage of the phospholipid bilayer membrane by the amphiphilic structure of cationic peptides, applies to the damage of the inner membrane by TNA and R-TNA (47, 48), but cannot interpret their different bactericidal effects. It has been reported that the weak cohesive forces within nanofibrils of supramolecular materials promote the disassociation of assembled molecules to disturb lipid membrane and induce cell death (49). It has also been reported that nondissociating rigid carbon nanotubes can penetrate cell membrane but do not cause cell death (50, 51), due to efficient self-repair of cholesterol in the phospholipid bilayer of cell membrane (52, 53). Therefore, a stronger rigidity of the assembly does not mean a higher ability in destroying the membrane structure. Compared to TNA, the high supramolecular cohesive forces of R-TNA may reduce the dissociation of assembly molecules from the nanofibrils, resulting in less disturbance of the inner phospholipid bilayer. We labeled TNA and R-TNA nanofibrils with rhodamine B via covalent reaction and measured fluorescence anisotropy. TNA nanofibrils presented lower anisotropy signal than R-TNA nanofibrils (fig. S34), which indicates more segmental motion of the fluorescent dye in the nanofibrils; more segmental motion indicates less cohesion of the nanofibrils. The difference of TNA and R-TNA in phospholipid bilayer disruption was also seen in the calcein leakage test using the large unilamellar vesicle (LUV) as the model of phospholipid bilayer. FM 4-64 dye–stained LUV encapsulating fluorescent calcein was mixed with TNA or R-TNA nanoparticles, respectively, and dye release was monitored using fluorescence imaging. It was found that less cohesive TNA caused greater calcein release in 250 s than strong cohesive R-TNA (Fig. 4F). As a control, the addition of phosphate-buffered saline (PBS) instead of TNA/R-TNA caused no notable leakage of calcein, confirming that osmotic pressure does not play a dominant role in calcein release. Consequently, all three factors, including the electrostatic interaction, amphiphilic structure, and supramolecular cohesion of TNA nanofibrils, have an impact on the toxicity of inner membrane.

To determine whether the TNA treatment affects the mechanical properties of bacterial envelope, the stiffness, permeability, and membrane potential of bacterial envelope were evaluated in the culture medium by fluorescence imaging and flow cytometry. In the cantilever force versus indentation distance curves, a substantially retarded slope was observed on the PDR A. baumannii envelope after 20-min TNA treatment (20 μM) compared to the control (PBS) group (Fig. 4G), indicating the decreased bacterial envelope stiffness after mechanical disruption of TNA. Furthermore, TNA-treated PDR A. baumannii was visualized by super-resolution structured illumination microscopy after staining with fluorescent reporters of outer membrane morphology (FM 4-64), cell wall morphology (WGA-iFluor555), and membrane integrity (SYTOX Green). After 10-min treatment with 20 μM TNA, the intact outer membrane and cell wall were damaged and SYTOX was observed in the bacterial cytoplasm (Fig. 5A), suggesting that TNA compromises the integrity of bacterial membrane. Bacterial permeability and membrane potential were studied by flow cytometry using fluorescent probes, DiOC₂(3) for membrane potential and TWO-PRO-3 for only membrane permeability. Compared to the control (PBS), the membrane-decoupler (CCCP), and the pore-forming peptide (nisin), 5-min treatment with 20 μM TNA caused notable defects of PDR A. baumannii in both membrane polarization and permeability (Fig. 5, B and C).

Fig. 5. TNA mechanotherapy damages the envelope integrity, causes multiple bacterial death pathways, and avoids drug resistance.

(A) Super-resolution structured illumination microscopic images of PDR A. baumannii treated with TNA for 10 min. (B) Flow cytometry analysis of the membrane potential and permeability of PDR A. baumannii after incubation with PBS (control), TNA (20 μM), nisin (20 μM), and CCCP (5 μM) for 5 min. (C) Schematics of flow cytometry results expected for each class of permeable, impermeable, polarized, and depolarized bacteria. (D) Volcano plots illustrate differential protein expression inside the bacteria between untreated and TNA-treated bacteria [results were deemed significant if |log₂(fold change)| ≥ 1 and P < 0.05; n = 3 biologically independent samples]. (E) Fluorescent images of ROS in PDR A. baumannii treated with TNA for 5 min. MIC change of (F) PDR A. baumannii and (G) A. baumannii treated with TNA or tigecycline after 20 days of serial passaging in each drug. Bacteria grew in 0.5× MIC of the indicated antibiotics in Mueller-Hinton (MH) broth. (H) Intrabacterial proteomic results of PDR A. baumannii (1 to 3, biologically independent samples) and A. baumannii (4 to 6, biologically independent samples) treated with TNA for 5 min. Blue and red values in the heat map indicate inhibition (≤0.5-fold) and stimulation (≥2-fold), respectively, of proteins relative to untreated (PBS) bacteria. The dose for TNA is 20 μM for (A), (B), (D), (E), and (H).

TNA causes multiple bacterial death pathways and avoids drug resistance

TNA mechanotherapy reduces NADH [reduced form of nicotinamide adenine dinucleotide (oxidized form)]–ubiquinone–oxidoreductase (Nuo) subunits of ETC complex I in the plasma membrane, which promotes generation of reactive oxygen species (ROS) to ultimately induce oxidative bacterial death (54, 55). The bacterial proteome analysis showed that the abundance of NuoE, NuoF, NuoG, and NuoL decreased to 0.2-, 0.3-, 0.4-, and 0.2-fold, respectively, in response to 5-min TNA treatment (20 μM) (Fig. 5, D and H). Instead of reacting with ubiquinone, electrons from NADH captured by crippled ETC complex I get sidetracked to dissolved oxygen to generate superoxide anion (54). Superoxide anion, though ordinarily noncytotoxic, can be rapidly converted to cytotoxic intermediates including peroxide as visualized by fluorescence microscopy imaging (Fig. 5E and fig. S35). The abundance of the major peroxide-inactivating enzymes, KatG and AhpC, increased to 1.6- and 1.3-fold (tables S5 and S6), respectively, as the result of feedback regulation of ETC complex I dysfunction-induced ROS.

TNA treatment causes outflow of intrabacterial proteins, which subsequently breaks the balance of a highly conserved set of proteins involved in vital biosynthesis and metabolic pathways. Among 1742 identified proteins in A. baumannii and PDR A. baumannii, there were 117 proteins down-regulated and 30 proteins up-regulated (P < 0.05, fold > 2 or fold < 0.5) after 5-min treatment with 20 μM TNA (Fig. 5, D and H, and fig. S36). In the outer membrane, the abundance of periplasmic chaperones DegP and the fragment of lipoprotein increased markedly, while BamA and Ompp1 decreased (Fig. 5, D and H). In the cell wall and plasma membrane, the abundance of 24 proteins substantially decreased, while only 3 proteins increased (Fig. 5H).

Among intrabacterial proteins, TNA inactivated alanine, asparagine, glutamate, and lysine-tRNA synthetases, which charge tRNA with their respective amino acids using ATP hydrolysis for energy supply (Fig. 5, D and H). TNA also directly disrupted ribosome assembly, stability, and function by markedly decreasing the abundance of proteins in 30S and 50S subunits of ribosome (Fig. 5H). As a result, cell division–associated proteins decreased, including cell division protein FtsA, ZipA, and cell division inhibitor MinD (Fig. 5H). In other metabolism pathways, e.g., energy, ester, and fatty acid metabolism, the abundance of 20 proteins markedly decreased and 5 proteins increased (Fig. 5H and fig. S36). Furthermore, as members of Clp proteases that are responsible for removing bacterial toxic misfolded and unfolded proteins (56), ClpA, ClpB, and ClpX decreased to 0.2-fold (Fig. 5, D and H).

Developing resistance should prove difficult for TNA because the TNA uses a multiple strategy to kill bacteria (Fig. 5, A to E and H, and fig. S37). It is possible that bacteria may develop resistance by cleaving the nanofibril structure of TNA or reducing the affinity of BamA and LPS for TNA; however, the time-consuming enzymatic degradation of nanofibrils cannot compete with the rapid bactericidal action of TNA (57). Furthermore, BamA and LPS are both conserved molecules on the outer membrane (35–41), and simultaneous mutation of these two molecules to resist TNA treatment not only is “costly” but also seriously affects bacterial viability. After treatment with 0.5 MIC of TNA, no resistance to TNA was developed for both A. baumannii and PDR A. baumannii for ~500 generations (Fig. 5, F and G).

TNA induces bacterial antigen release and activates immune responses in vitro

The intact outer membrane of Gram-negative bacteria prevents outer membrane from being recognized by the immune system (58). Outer membrane proteins have the most important immunogenicity (59–61); however, conventional continuous low-dose stimulation under chronic infection will produce an immunosuppressive state by derailing the acquisition of protective B cell memory (62). Therefore, mechanical disruption–induced burst release of outer membrane antigens should be expected to considerably enable antigen-presenting cell (APC) priming and evoke local antibacterial immune responses (63). Five minutes after TNA treatment (20 μM), proteins released in the protein-free culture medium of PDR A. baumannii were analyzed (Fig. 6A). The amount of released proteins estimated by the xanthoproteic test increased substantially (fig. S38). Proteomic analysis showed 122 proteins up-regulated and 81 proteins down-regulated (Fig. 6B and fig. S39). The abundance of a set of outer membrane proteins, especially outer membrane protein A (OmpA) and outer membrane porin (OprD family protein), increased markedly (Fig. 6B and table S7).

Fig. 6. TNA induces bacterial antigen release and activates immune responses in vitro.

(A) Schematics of mechanical disruption–triggered bacterial antigen release and antigen presentation. (B) Volcano plot illustrates differential proteins released between untreated and TNA-treated PDR A. baumannii [results were deemed significant if |log₂(fold change)| ≥ 1 and P < 0.05; n = 3 biologically independent samples for each treatment or control group]. (C) ELISA analysis of TNF-α and IL-12 expressed by DCs after treatment. Representative of flow cytometry plots (D) and quantification (E) of mature DCs (CD80⁺ CD86⁺) by gating on CD11c⁺ cells. (F) Schematics of TNA maintaining proinflammatory phenotypes of macrophages. (G) ELISA analysis of TNF-α, IL-12, and IL-1β expressed by PEMs. (H) Flow cytometry analysis displaying changes in CD86 (M1 marker) and CD206 (M2 marker) expression in CD11b⁺ F4/80⁺ macrophages. Representative flow cytometry profiles (I) and quantification (J) of phagocytic clearance of bacteria by macrophages. (C, E, G, and J) Data are mean ± SD (n = 5 per group), with n representing biologically independent experiments. Experiments in (D), (H), and (I) were repeated independently five times with similar results. **P < 0.01; ***P < 0.001.

To investigate the TNA treatment–induced immune responses in vitro, we incubated mouse bone marrow–derived dendritic cells (BMDCs) and mouse peritoneal macrophages (PEMs) with the excretion of bacteria. A higher population of mature BMDCs has been reported to induce a stronger adaptive immunity against bacterial reinfection (24, 25). The BMDCs cultured with the excretion of TNA-treated PDR A. baumannii showed a substantial up-regulation of proinflammatory cytokine tumor necrosis factor–α (TNF-α) and interleukin-12 (IL-12), as well as markedly increased percent of mature BMDCs (CD80⁺ CD86⁺ CD11c⁺) compared to the PBS-treated PDR A. baumannii group (Fig. 6, C to E, and fig. S40). Macrophages are also indispensable for effective innate and adaptive immune responses (Fig. 6F) (64). TNA-treated groups presented elevated release of proinflammatory cytokines including TNF-α, IL-12, and IL-1β (Fig. 6G) that are associated with enhanced antigen priming and subsequent presentation by APCs. TNA only–treated PEMs also induced a notable increase of M1 phenotype (CD86⁺ CD11b⁺ F4/80⁺) and bacterial phagocytosis rate compared to the PBS group (Fig. 6, H to J). In addition, TNA on the bacterial surface does not inhibit the uptake of PDR A. baumannii by PEMs (fig. S41).

TNA kills PDR A. baumannii in vivo and confers long-lasting protective immunity against Gram-negative bacterial reinfections

The biocompatibility of TNA was evaluated for the eligibility in vivo study. Hemolytic activity test with red blood cells showed that TNA had negligible hemolytic activity (>200 μM; fig. S42). The therapeutic index (TI) of TNA was 71, a value higher than that of colistin, which is considered as the last-resort drug for drug-resistant Gram-negative pathogens (fig. S43).

The antimicrobial efficacy of TNA in vivo was first demonstrated by the wound infection model with the test spot (five mice per test spot) of the independent mice group illustrated in Fig. 7A. Specifically, female Balb/c mice (age 6 to 8 weeks) were infected with PDR A. baumannii [2.0 × 10⁷ colony-forming units (CFU)] on the 8-mm artificial wound of the right leg muscle and treated on day 1 with TNA (5 mg kg⁻¹), colistin (5 mg kg⁻¹), or PBS. On day 2, bio-TEM (transmission electron microscopy) results showed that TNA nanoparticles transformed into nanofibrils and disrupted the envelope of PDR A. baumannii (Fig. 7B). On day 9, the absence of pus under the scab and superior wound area reduction showed a rapid healing by TNA treatment (Fig. 7C and fig. S44), whereas pus was observed under the scab in saline and colistin groups; the harvested muscle tissue showed at least six orders of magnitude reduction of bacterial colonies compared with PBS and colistin groups (fig. S45); hematoxylin and eosin (H&E) staining showed that the tissue morphology of the TNA-treated group recovered to the same as that of the uninfected group (fig. S46); in contrast, many inflammatory cells were found in PBS and colistin groups (fig. S46).

Fig. 7. Efficacy of TNA in wound infection model, peritonitis model, and establishing immunological memory.

(A) Schematics showing the in vivo antibacterial treatment and immune activation in a wound infection model. (B) Bio-TEM images of muscle tissue slices showing transformed nanofibrils and disrupted bacterial envelope in the TNA-treated group. The bio-TEM images are representative of three independent experiments. (C) Representative photographs of the infected area of the PDR A. baumannii–infected mice taken on days 1 to 9. (D and E) Quantitative analysis of different subtypes of macrophages in IDLNs by flow cytometry on days 3 and 6, respectively. (F) Quantitative analysis of mature DCs (CD80⁺ CD86⁺) in IDLNs by gating on CD11c⁺ cells on day 3. (G) Quantitative analysis of CD8⁺ T cells in spleens by gating on CD3⁺ CD45⁺ cells on day 3. (H) Schematics showing the in vivo antibacterial treatment and long-term immune memory activation using the extended peritonitis model. (I) Quantitative analysis of memory B cells (CD19⁺ CD38⁻) in spleens by gating on CD45⁺ cells on day 42. (J) Serum anti-PDR A. baumannii IgG titers on day 42. (K) Bacterial load in the peritoneal cavity of mice on day 43 after intra-abdominal PDR A. baumannii (10⁶ CFU) challenge on day 42. (L) Survival of mice intra-abdominal rechallenged with PDR A. baumannii (n = 10); no reinfection was performed in the colistin group because no mice survived on day 42. The dose of TNA used in wound infection model and peritonitis model was 5 mg kg⁻¹. (D) to (G) and (I) to (K) Data are mean ± SD (n = 5 per group), with n representing biological independent experiments. **P < 0.01; ***P < 0.001.

The antimicrobial efficacy of TNA in vivo was also demonstrated by the peritonitis model with independent mice group used for different tests (five mice per test). Specifically, mice were infected with PDR A. baumannii (2.0 × 10⁷ CFU) by intraperitoneal injection; at 0.5 and 4 hours after infection, mice were administrated by intraperitoneal injection with saline (control), antibiotic colistin (5 mg kg⁻¹), or TNA (5 mg kg⁻¹), and data were acquired on day 2. All mice in the TNA group survived without signs of animal distress, whereas only 20% mice of the saline or colistin group survived. Compared to saline and colistin groups, TNA treatment resulted in at least four orders of magnitude reduction in bacterial cell counts in the peritoneal cavity (fig. S47). H&E staining demonstrated that TNA alleviated the organ damage caused by PDR A. baumannii infection, whereas thickened respiratory membranes in the lung and thrombosis accompanied by congestion in the liver, spleen, and kidney were observed in saline and colistin groups (fig. S48). Moreover, a sharp increase in regulatory T cell (T_reg) (CD25⁺ Foxp3⁺) infiltrates was observed in the saline group compared to the TNA-treated group (fig. S49), suggesting that the high mice mortality in the saline group may be related to the immunosuppression by PDR A. baumannii.

The immune activation effect of TNA treatment was studied in the above wound infection model using the immunofluorescence method and the multicolor flow cytometry (Fig. 7A). M1-like cells (CD86⁺ CD11b⁺ F4/80⁺) serve as proinflammatory immune cells by bacterial phagocytosis and secreting inflammatory factors during acute phase of infection, whereas M2-like cells (CD206⁺ CD11b⁺ F4/80⁺) yield anti-inflammatory factors and prorepair effect during the late phase of infection. On day 3, in muscles and infection-draining lymph nodes (IDLNs), more M1-like cells and fewer M2-like cells were present in the TNA-treated group than in PBS and colistin groups (Fig. 7, D and E, and figs. S50 to S54). DCs are responsible for presenting antigens to other immune cells, including cytotoxic CD8⁺ T cells and CD4⁺ helper T cells, to stimulate adaptive immune responses during the acute inflammatory phase (65–67). Mature DC (CD80⁺ CD86⁺ CD11c⁺) population in isolated IDLNs was higher in the TNA-treated group than in PBS and colistin groups (Fig. 7F and fig. S50). As expected, enhanced T cell infiltrations, cytotoxic CD8⁺ T cells (CD8⁺ CD3⁺ CD45⁺) and CD4⁺ helper T cells (CD4⁺ CD3⁺ CD45⁺), were observed in the spleen of TNA-treated mice on day 3 (Fig. 7G and figs. S50, S55, and S56). These results collectively suggest that TNA treatment can enhance the pathogen-associated immunogenicity in vivo.

TNA treatment can activate the long-term immune memory in vivo and confer long-lasting protective immunity against Gram-negative bacterial reinfections as demonstrated by the extended peritonitis model. The time spot for PDR A. baumannii infection (2.0 × 10⁷ CFU), TNA treatment (5 mg kg⁻¹), and analysis is illustrated in Fig. 7H. Other than the first intraperitoneal infection and TNA treatment, mice were also immunized by infection and post-infection treatment on days 7 and 14 (Fig. 7H). Immune analyses for myeloid DCs and T cells were carried out on day 15. Similar with the immune response in wound infection model, substantially higher levels of mature DC (CD80⁺ CD86⁺ CD11c⁺), CD8⁺ T cell (CD8⁺ CD3⁺ CD45⁺), and CD4⁺ helper T cell (CD4⁺ CD3⁺ CD45⁺) infiltration were observed in the TNA group (figs. S57 and S58). To assess the level of pathogen-specific immunological memory stimulation, the expression of memory B cells and antibody response against PDR A. baumannii was studied on day 42. The memory B cells on day 42 (CD19⁺ CD38⁻ CD45⁺) markedly increased in the spleen of mice immunized with infection and post-infection treatment on days 0, 7, and 14 compared to naïve mice (Fig. 7I and figs. S59 and S60). On day 42, the anti-PDR A. baumannii immunoglobulin G (IgG) titer in the blood serum was substantially elevated (Fig. 7J). With reinfection by PDR A. baumannii (2.0 × 10⁷ CFU) on day 42, the immunized mice presented 80% survival rate on day 47, whereas no mice survived in the naïve group (Fig. 7L). No reinfection was performed in the colistin group because no mice survived on day 42. In a separate study, mice were challenged with 10⁶ CFU of PDR A. baumannii on day 42, and bacteria load was quantified on day 43 in the peritoneal cavity, blood, lung, liver, spleen, kidney, and heart (Fig. 7K and fig. S61). In all organs, a substantially more bacteria were detected for naïve mice than for immunized mice.

DISCUSSION

It is intriguing that the antibacterial mode of TNA resembles that of neutrophil extracellular traps (NETs). In nature, neutrophils, the most abundant innate immune effector cells, can be recruited to the inflammatory sites and release granule proteins and chromatin that together assembles into NETs to bind and degrade bacterial membrane proteins (68). Similar to the TNA nanofibrils in this work, NETs are fibrous structure composed of smooth “threads,” approximately 15 nm in diameter. The physical action on bacteria by both TNA fibrils and NETs could prevent systemic dissemination. Although the ability of NETs to address bacterial resistance has been demonstrated by biological evolutionary history, the molecular mechanisms of NETs for bacterial membrane disruption are not well understood (68, 69); furthermore, the relation of NET assembly process on bacterial surface with bactericidal property is unclear. The TNA in situ deformation process is required to generate antibacterial effects, because pretransformed nanofibrils did not affect the integrity of the outer membrane of PDR A. baumannii (fig. S62). Therefore, designing artificial NET-like structure, such as TNA, is but one promising approach to impart the nano-antibiotics with not only the NET-like function but also tunable molecular structure as well as clear mechanism for binding and bacterial membrane disruption without drug resistance.

The nanofibrous structure of TNA may act as excellent adjuvants to elicit immunity (28, 31), in addition to the in situ transformation-mediated bactericidal capability. Similar to the amphiphilic structure of TNA, host defense (antimicrobial) peptides (HDPs) can also assemble into entangled nanofibrous networks and trap bacteria. Unlike the fast bactericidal activity of TNA, HDPs have the poor direct antimicrobial activity (23, 70). On the other hand, HDPs exhibit multifaceted immunomodulatory activities, including anti-infection and selective anti-inflammation, as well as immune adjuvant activities (23). Recent studies indicate that peptide nanofibrils can act as adjuvants by promoting DC maturation and eliciting T cells into helper T cells and B cells into germinal center cells, as well as high-titer IgG (29, 31, 71). However, it is difficult to distinguish immunomodulatory contributions between molecular and nanofibrous structures because of the dynamic self-assembly and dissociation of TNA/HDP nanofibrils. This question may better be answered in the future by studying the immunomodulatory capacity of nondissociated peptide nanofibrils or other organic nanofibrils.

MATERIALS AND METHODS

Synthetic methods

TNA, S-TNA, R-TNA, and N-TNA were synthesized using standard Fmoc solid-phase peptide chemistry, and the syntheses were performed from C terminus to N terminus in the peptide synthesizer. Take synthesis of TNA as an example. The first 30-min coupling reaction was conducted at 75°C on a peptide synthesizer (Protein Technologies, SYMPHONY) by shaking the mixture of N,N′-dimethylformamide (DMF; Aladdin) solutions of Wang resin (J&K Scientific, 946450), 3 eq. of Fmoc-l-Lys(Mmt)-OH (Shanghai Apeptide), 3 eq. of O-benzotriazole-N,N,N′,N′-tetramethyluronium hexafluorophosphate (HBTU; Sigma-Aldrich), and 10 eq. of N,N-diiso-propylethylamine (DIEA; Sigma-Aldrich). Fmoc deprotection was subsequently performed at 75°C using 20% 4-methyl piperidine in DMF for 3 min. From the second to sixth coupling reaction, amino acids in the sequence were conjugated by replacing Fmoc-l-Lys(Mmt)-OH with Fmoc-l-Lys(Boc)-OH (Shanghai Apeptide) or Fmoc-l-Phe-OH (Shanghai Apeptide) with the other conditions the same as the first coupling reaction. Four washes were performed in between steps with methanol (twice, 10 ml g⁻¹) and DMF (twice, 10 ml g⁻¹). After the Fmoc deprotection in the sixth step, the resin was mixed with 3 eq. of 1-pyrenebutyric acid, 3 eq. of HBTU, and 10 eq. of DIEA in DMF and shaken for 1 hour at 75°C. After the pyrene conjugation, Lys(Mmt) deprotection was performed twice using 2% trifluoroacetic acid in dichloromethane (Aladdin), and the Wang resin was washed with methanol (two times, 10 ml g⁻¹) and dichloromethane (two times, 10 ml g⁻¹). The Sul moiety coupling reaction was conducted at 75°C for 2 hours by shaking the mixture of DMF solutions of the Wang resin, 1.5 eq. of tert-butyl (4-(chlorosulfonyl)phenyl)carbamate, 3 eq. of HBTU, 1 eq. of 1-hydroxybenzotriazole (HOBt; Aladdin), and 10 eq. of DIEA. After synthesis, Boc deprotection was performed and peptides were cleaved from Wang resin by shaking with the cleavage solution [94:2.5:2.5:1, trifluoroacetic acid:water:1,2-ethanedithiol:triisopropylsilane (both from Sigma-Aldrich)] for 2 hours. The peptides were then precipitated by adding the cleavage solution to cold diethyl ether, and the resulting crude product was purified by preparative scale reversed-phase high-performance liquid chromatography (HPLC) using a Boston Symmetrix ODS-R column (C₁₈ stationary phase, 5 μm, 4.6 × 250 mm) with a gradient of water and acetonitrile containing 0.1% trifluoroacetic acid. Pure fractions, identified using electrospray ionization mass spectrometry, were collected by rotary evaporation, lyophilized, and stored at −20°C. For the synthesis of S-TNA, 1-pyrenebutyric acid (Aladdin) was replaced with acetate (Aladdin); for the synthesis of R-TNA, 1-pyrenebutyric acid was replaced with BP-COOH (Lei Wang Laboratory at NCNST); for the synthesis of N-TNA, Fmoc-l-Phe-OH was replaced with Fmoc-Gly-OH (Shanghai Apeptide); the same experimental procedures and conditions were followed.

Structural modeling and MD simulations

For TNA modeling, the force field of nonstandard residues in TNA monomer was developed by Gaussian 09 (72) at the B3LYP/6-31G* level. For LPS modeling, the force field of LPS was generated by Gaussian 09 (72) at the B3LYP/6-31G level. For BamA modeling, the initial BamA structure was derived from Protein Data Bank (PDB) with the ID of 7NRE (37).

In MD simulations, the ff14SB force field (73), Generation Amber Force Field (74), and explicit solvent of TIP3P (75) were used to perform MD simulations in the Amber 18 software package (75). The distance between the peptide and the box edge is greater than 1.0 nm. The SHAKE algorithm was applied to constrain bonds including hydrogen bonds (76). The particle mesh Ewald summation method was used to describe long-range electrostatics (77). The cutoff distance is 1.0 nm for nonbonded interactions. The temperature of 298 K and pressure of 1 atm were controlled by the Langevin thermostat and Berendsen barostat, respectively. To avoid the bias sampling, we performed two simulations with different initial structural configurations. Greater than 200-ns trajectory was collected for each system. After equilibrium evaluation, only the final 50-ns equilibrated data were used for further analysis. In binding free energy calculations, the molecular mechanics with generalized Born surface area (MM/GBSA) (78) was performed using the modified generalized Born model with a salt concentration of 0.1 M.

Self-assembly preparation and characterization

The methods for self-assembly preparation and characterization are the same for TNA, S-TNA, R-TNA, and N-TNA. TNA (1 mg) was dissolved in 50 μl of dimethyl sulfoxide (DMSO) (Aladdin, D103273) and subsequently diluted to a concentration of 20 μM with deionized water. Fresh TNA solution (20 μM) was used for measurements as an initial state. The morphology transformation of TNA from nanoparticles to nanofibrils was performed by the addition of LPS (2 μg ml⁻¹) (ShanghaiyuanyeBio, S11060) and BamA (CUSABIO, CSB-EP364272EOD1) (2 μg ml⁻¹) to 20 μM TNA solution and cultured at 37°C. The morphology and size were measured by TEM (Hitachi-7500, Japan) and dynamic light scattering (DLS; Nano ZS90). For TEM observation, carbon-coated 200-mesh copper grids (Beijing Zhongjingkeyi, BZ110223a) were freshly treated by glow discharge using a plasma cleaner (Pluto-M, PLUTOVAC). Sample solution (10 μl) was dropped on the grid. After 1 min, the remaining liquid was blotted using the filter paper from the edge and air-dried for 30 min before TEM observation. To keep the morphology of the assembly undisturbed, no staining was performed.

CAC determination

The CAC of TNA was determined by the light scattering spectra of TNA at different concentrations. Briefly, TNA (1 mg) was dissolved in 50 μl of DMSO and subsequently diluted to concentrations as indicated with deionized water. The scattering spectra of TNA were recorded with the synchronous scanning mode of model F-7000 spectrofluorometer (Hitachi, Japan). Scan mode: synchronous, Δλ = 0 nm; wavelength: 200 to 600 nm; scan speed: 1200 nm/min; excitation slit: 5.0 nm; emission slit: 5.0 nm. A particle, assumed to be spherical, absorbs and scatters light depending on its size, shape, and refractive index relative to the surrounding medium. When the irradiation power is fixed, light scattering is proportional to the scattering cross section

C_sca = (πr²) (8/3) x⁴ [(m² − 1)/(m² + 2)]²

where r is the radius of spherical particle, m = n_sph/n_med, n_sph is the refractive index of the sphere, n_med is the refractive index of the surrounding medium, and x is the size parameter, equal to 2πrn_med/λ. Here, the scattering intensity at 500 nm is used for the calculation of CAC.

CD sample preparation and measurement procedures

For CD characterization of TNA after incubating with bacteria or the LPS + BamA mixture, overnight PDR A. baumannii cultures were diluted at 1:50 and grown in the Lysogeny Broth (LB) medium (Sangon Biotech, A507002) to an early-mid exponential phase [OD₆₀₀ (optical density at 600 nm) = 0.4 to 0.6] at 37°C. After centrifugation (6000g, 3 min) and washing with stroke-physiological saline solution (SPSS) twice, bacteria were adjusted to 1 × 10⁸ bacteria/ml (~1.0 OD₆₀₀). TNA (1 mg) was dissolved in 50 μl of DMSO and subsequently diluted to a concentration of 200 μM with SPSS. The 1.8-ml bacteria suspension was incubated with 200 μl of TNA (200 μM) at 37°C for 20 min. The CD spectra of bacteria, TNA, and bacteria + TNA were recorded with Chirascan plus CD (Applied Photophysics Ltd., Britain). Instrumental parameters are as follows: detector, PMD detector; light source, Xe lamp; time per point, 0.5 s; replication, three times; path length, 10 mm; temperature, 25°C; wavelength, 190 to 350 nm; step size, 1 nm; bandwidth, 1 nm. For CD characterization of TNA after incubating with LPS and BamA, TNA (1 mg) was dissolved in 50 μl of DMSO and subsequently diluted to a concentration of 25 μM with SPSS. The 1.6-ml TNA solution (25 μM) was incubated with 200 μl of LPS (20 μg ml⁻¹) and 200 μl of BamA (20 μg ml⁻¹) at 37°C for 40 min. The CD spectra of TNA, LPS + BamA, and TNA + LPS + BamA were recorded with Chirascan plus CD (Applied Photophysics Ltd., Britain) with the aforementioned instrumental parameters. Analysis was performed using the open access software at http://bestsel.elte.hu.

FTIR sample preparation and measurement procedures

For FTIR characterization of TNA after incubating with LPS and BamA, TNA (1 mg) was dissolved in 50 μl of DMSO and subsequently diluted to a concentration of 25 μM with SPSS. The 8-ml TNA solution (25 μM) was incubated with 1 ml of LPS (20 μg ml⁻¹) and 1 ml of BamA (20 μg ml⁻¹) at 37°C. After 40 min of incubation, the assembled TNA nanofibrils were obtained by centrifugation (1 × 10⁴ g, 5 min) and washing with deionized water twice. After centrifugation, the TNA nanofibrils are freeze-dried to remove excess water. The incubation conditions of the TNA only group and LPS + BamA only group were consistent with those of the TNA nanofibril group. The TNA only group and LPS + BamA only group were freeze-dried without centrifugation steps. The dried samples were mixed with potassium bromide, ground, and tableted. The FTIR spectra were recorded with Nicoletis 10 (Thermo Fisher Scientific, USA).

AFM measurements of the stiffness of in situ self-assembly fibrils and bacteria

Overnight PDR A. baumannii cultures were diluted at 1:50 and grown in the LB medium to an early-mid exponential phase (OD₆₀₀ = 0.4 to 0.6) at 37°C. After centrifugation (6000g, 3 min) and washing with SPSS twice, bacteria were adjusted to 1 × 10⁹ bacteria/ml (~1.0 OD₆₀₀). The 50-μl bacteria suspension was applied on the surface of the precleaned polysine microscope slide (Thermo Fisher Scientific, 6776215) at room temperature. After 20 min, excess suspension was removed to leave a layer of PDR A. baumannii on the slide surface. TNA (1 mg) was dissolved in 50 μl of DMSO and subsequently diluted to a concentration of 20 μM with SPSS. Fifty microliters of 20 μM TNA SPSS was added to the bacteria-covered slide to allow the self-assembly of TNA into nanofibrils at 37°C for 20 min. The bacteria slide was washed twice with SPSS, and bacteria were kept in SPSS throughout measurements.

The height, the cantilever force versus indentation distances of the bacteria-fiber complex, and the Young’s modulus were measured with an atomic force microscope (Bruker, Germany) in the MIROView-PeakForce QNM in Fluid-Live Cells mode, and data were analyzed using NanoScope Analysis. The probe (PFQNM-LC-A-CAL) was silicon tip on nitride lever with reflective Au on the front side, and parameters are as follows: 45 kHz (f₀), 0.1 N/m (k), 345 nm (T), 54 μm (L), and 4.5 μm (W). The Young’s modulus was extracted from the force-indentation curves based on the Derjaguin-Muller-Toporov (DMT) model (79–86), which is the Hertzian model considering adhesion force.

Electron microscopic study of bacterial disruption by TNA in vitro

Overnight PDR A. baumannii cultures were diluted at 1:50 and grown in LB medium to an early-mid exponential phase (OD₆₀₀ = 0.4 to 0.6) at 37°C. After centrifugation (6000g, 3 min) and washing with SPSS twice, bacteria were adjusted to 1 × 10⁸ bacteria/ml (~0.1 OD₆₀₀) with SPSS. TNA (10 mg) was dissolved in 200 μl of DMSO and subsequently diluted to a concentration of 200 μM with SPSS. The 180-ml bacteria suspension was incubated with 20 ml of TNA (200 μM) at 37°C with duration as indicated. At each predetermined time, 50 ml of bacteria suspension was withdrawn, centrifuged (6000g, 3 min), washed twice with SPSS, resuspended in 2 ml of SPSS, and fixed in 2% glutaraldehyde for 12 hours at 4°C.

For the observation of the surface morphology of bacteria, 1 ml of the fixed bacteria was centrifuged (6000g, 3 min), washed with deionized water twice, and dehydrated by CO₂ supercritical drying. Bacterial samples were sprayed with gold under vacuum for 80 s before scanning electron microscopic observation (Hitachi SU8082, Japan).

For the observation of changes in outer membrane, cell wall, and cell membrane, 1 ml of the fixed bacteria was centrifuged (6000g, 3 min), washed with deionized water twice, and treated with 0.5% OsO₄ for 1 hour at 4°C. After rinsing with deionized water, the bacteria were immersed in saturated uranyl acetate in 70% ethanol for 1 day at 25°C. Subsequently, the bacteria were dehydrated with 99.9% ethanol and then embedded in Epon resin. Ultrathin sections (~70 nm) were obtained with an ultramicrotome (Leica, EMuc7). The electron micrographs of bacteria envelope were obtained with a transmission electron microscope (Hitachi-7500, Japan).

Calcein leakage test using the LUV

The LUV was synthesized on the surface of agarose film according to previous reports (49, 80). Briefly, melted low-melting temperature agarose [1% (w/w); Sigma-Aldrich, A9414] was added dropwise to the surface of a tilted microscope slide to allow the slide to be covered with a thin layer of agarose. The slide was placed on a hot plate (40°C) for solidification. Afterward, 100 μl of egg phosphatidylcholine (3.75 mg ml⁻¹; Sigma-Aldrich Y0001905) in chloroform-methanol (9:1, v/v) was added dropwise to the surface of agarose-coated slide (1 inch × 3 inches), and the residual organic solvent was removed in a vacuum overnight to form a thin lipid film on the surface of agarose layer. Then, the slide was immersed for 12 hours in 15 ml of calcein (1 mg ml⁻¹) dissolved in PBS (pH 7.4). In such a way, LUV encapsulating fluorescent calcein was generated on the slide surface. After removal of the slide, LUV in the resulting solution was harvested through standard dialysis (molecular weight cutoff: 1000 kDa) for 36 hours in PBS solution. For staining of the liposome membrane, FM 4-64 (Thermo Fisher Scientific, F34653) was added to LUV solution to a final concentration of 10 μg ml⁻¹ at 4°C for 10 min. TNA (10 mg) or R-TNA (10 mg) was dissolved in 200 μl of DMSO and subsequently diluted to a concentration of 2 mM with PBS. TNA or R-TNA solutions were added to the hydrated liposomes to a final concentration of 1 mM and monitored using fluorescence imaging (LSM800, Zeiss, Germany). TNA and R-TNA nanoparticles can be adsorbed on the surface of LUV by electrostatic interaction. The disruption of LUV is affected by molecular diffusion from supramolecular materials (49). As a control, the addition of PBS instead of TNA/R-TNA caused no notable leakage of calcein, confirming that osmotic pressure does not play a dominant role in calcein release.

Sample preparation method for fluorescence imaging with three-dimensional super-resolution structured illumination microscopy

Overnight PDR A. baumannii cultures were diluted at 1:50 and grown in LB medium to an early-mid exponential phase (OD₆₀₀ = 0.4 to 0.6) at 37°C. TNA (1 mg) was dissolved in 50 μl of DMSO and subsequently diluted to a concentration of 40 μM with SPSS. One milliliter of the diluted PDR A. baumannii cells (3.0 × 10⁶ cells ml⁻¹) was incubated with 1 ml of TNA (40 μM) in an Eppendorf tube (2 ml) at 37°C for 10 min. The cell suspension was then centrifuged (6000g, 3 min) and washed with Hanks’ balanced salt solution (HBSS) twice. The cell permeability, cell wall, and outer membranes were stained with SYTOX Green (Thermo Fisher Scientific, S7020), wheat germ agglutinin–iFluor 555 (WGA-iFluor555, AAT Bioquest, 25539), and FM 4-64 (Thermo Fisher Scientific, F34653), respectively. SYTOX Green nucleic acid stain is an excellent green fluorescent nuclear and chromosome counterstain that is impermeable to live cells, making it a useful indicator of dead cells within a population. For staining of the nucleic acid of dead bacteria, SYTOX Green was added to the HBSS TNA-bacteria solution with a final concentration of 5 μM at 25°C for 20 min. For staining of the cell wall, WGA-iFluor555 was added to the bacteria medium with a final concentration of 10 μg ml⁻¹ at 37°C for 20 min. For staining of the outer membrane, FM 4-64 was added to the bacteria medium to a final concentration of 10 μg ml⁻¹ at 4°C for 10 min. The cells were washed in HBSS (6000g, 3 min) after each stain and finally fixed in 4% (w/v) paraformaldehyde in PBS for 30 min at room temperature. The fixative was removed with HBSS (6000g, 3 min), and HBSS (0.5 ml) was added to each well before imaging. The concentration of DMSO in the control group was 0.05%, and the other conditions were the same as those in the experimental group. To avoid the movement of the bacteria, 2 μl of bacteria solution was added to the bottom of the imaging petri dish, and then a newly prepared agarose gel (2.5%, low gelling temperature agarose) that matched the bottom of the petri dish was applied to the top of the bacteria before imaging. The fluorescence images of bacteria were obtained with a three-dimensional super-resolution structured illumination microscope (N-SIM S, Nikon).

Drug resistance studies

PDR A. baumannii isolated from hospital is resistant to ceftazidime, ceftriaxone, cefotaxime, imipenem, ampicillin, piperacillin, tazobactam, gentamicin, tobramycin, minocycline, levofloxacin, ciprofloxacin, polymyxin B, and colistin. PDR A. baumannii is not resistant to tigecycline. The MICs of two biological replicates of A. baumannii or PDR A. baumannii were obtained against TNA and tigecycline, respectively. TNA (1 mg) and tigecycline (1 mg) were dissolved in 50 μl of DMSO and subsequently diluted with the Mueller-Hinton (MH) broth (Coolaber, PM0530). The method used for resistance studies was adapted from that of Gullberg et al. (81). According to the method by Gullberg et al., the mean division time for A. baumannii is 20 to 30 min under suitable conditions. A. baumannii that grow in 0.5× MIC of the indicated antibiotics divides ∼25 generations every 24 hours. The bacteria were serially passaged by 400-fold dilution in 1 ml of batch culture medium every 24 hours for 20 days, which is roughly equivalent to 500 generations. After 100 generations of growth, the MIC of antibiotics was measured again. Fresh MH broth alone was used as the control, and MICs were obtained by following the same procedure. Resistance was confirmed by comparing the MICs of resistant mutants against the MICs of control groups.

ROS detection

Overnight PDR A. baumannii cultures were diluted 1:50 and grown in LB medium to an early-mid exponential phase (OD₆₀₀ = 0.4 to 0.6) at 37°C. TNA (1 mg) was dissolved in 50 μl of DMSO and subsequently diluted to a concentration of 200 μM with PBS. After 1:10 dilution with HBSS, 9 ml of bacteria was treated with 1 ml of TNA (200 μM) for 5 min at 37°C, centrifuged (6000g, 3 min), and washed with HBSS twice. ROS inside the bacteria were stained at 37°C by BBoxiProbe O13 (BestBio, BB-461113) working solution in the dark for 30 min. ROS can react with BBoxiProbe O13 and generate red fluorescence (excitation, 510 nm; emission, 610 nm). As a control, PBS was used to treat the bacteria by following the same procedure. In ETC inhibition experiments, the final concentrations of sodium azide and rotenone were 10 mM and 10 μM, respectively; the incubation time was 1 hour; the other conditions were the same as the TNA group.

Flow cytometry analysis of permeability barrier and membrane potential

Overnight PDR A. baumannii cultures were diluted 1:50 and grown in LB medium to an early-mid exponential phase (OD₆₀₀ = 0.4 to 0.6) at 37°C. TNA (1 mg), nisin (1 mg), and CCCP (1 mg) were dissolved in 50 μl of DMSO and subsequently diluted with SPSS. Each culture was diluted at 1:10 by HBSS and treated with the desired concentration of antibiotics (TNA, 20 μM; nisin, 20 μM; CCCP, 8 μM) for 5 min at 37°C. Bacteria were then stained with DiOC₂(3), the Bacterial Membrane Potential Kit (AAT Bioquest, 22038), to measure a bacterium’s membrane potential as a green-to-red ratio (excitation: 488 nm; emission: 525/550 nm for green and 610/620 nm for red). Membrane integrity was measured by staining cells with TWO-PRO-3 (AAT Bioquest, 17572), a dye that is excluded from cells with an intact membrane (excitation: 640 nm; emission: 670 nm). Fluorescence intensities of both dyes in response to antibiotic treatment were recorded with a flow cytometer (LSRFortessa, Becton Dickinson) at 20,000 events for each sample. Data were analyzed using FlowJo V10.8.1 software (FlowJo LLC, Ashland, OR).

Antigen release assay by xanthoproteic reaction

Overnight PDR A. baumannii cultures were diluted 1:50 and grown in LB medium to an early-mid exponential phase (OD₆₀₀ = 0.4 to 0.6) at 37°C. The cell suspension was centrifuged (6000g, 3 min), washed with SPSS twice, and adjusted to a concentration of 5 × 10⁸ cells/ml (~0.5 OD₆₀₀). TNA (1 mg) was dissolved in 50 μl of DMSO and subsequently diluted to a concentration of 200 μM with SPSS. Bacteria (7 ml) were incubated at 37°C with 7 ml of TNA (200 μM) in an Eppendorf tube (15 ml) for 0, 5, 10, 20, or 40 min. The concentration of bacteria and TNA used in this experiment was higher than the normal concentration (20 μM) to meet the measurement sensitivity of UV-visible spectrophotometry. At each time point, 2 ml of bacterial suspension was withdrawn and centrifuged (6000g, 3 min). For content analysis, 2 ml of the supernatant was transferred into a new tube, and 20 μl of concentrated nitric acid was added. The resulting mixture was kept at 95°C for 20 min to allow the complete replacement of the hydrogen on the tyrosine phenolic hydroxyl benzene ring of the protein by the nitro group. After cooling to room temperature, 50 μl of 20% NaOH solution was added. The resulting solution was used for photography (OnePlus GM1900) and UV-visible spectrophotometric measurements (U3010, Hitachi, Japan) using bovine serum albumin (BSA) standard solution for construction of the calibration curve. The concentration of protein was positively correlated with the absorbance in the xanthoproteic reaction test. The concentration of DMSO in the control group was 0.05%. The same procedure was followed for the S-TNA or R-TNA group. In addition, the xanthoproteic reaction did not occur in TNA solutions alone.

Antigen release assay by proteomic analysis

Overnight A. baumannii or PDR A. baumannii cultures were diluted 1:50 and grown in LB medium to an early-mid exponential phase (OD₆₀₀ = 0.4 to 0.6) at 37°C. The cell suspension was centrifuged (6000g, 3 min), washed with SPSS twice, and adjusted to 1 × 10⁸ cells/ml by protein-free M9 culture medium (Sangon Biotech, A507024-0250) with 20 mM glucose. TNA (1 mg) was dissolved in 50 μl of DMSO and subsequently diluted to a concentration of 40 μM with PBS. Bacteria (10 ml) were incubated with 10 ml of TNA (40 μM) in an Eppendorf tube at 37°C. After 5 min, bacterial suspension was withdrawn by centrifugation (6000g, 3 min) and concentrated by 20-fold with freeze-drying before SDS-PAGE (polyacrylamide gel electrophoresis) separation. SDS-PAGE sample loading buffer (3 μl; Beyotime, P0015F) was mixed with 15 μl of the supernatant, and the resulting solution was kept at 90°C for 3 min. Fifteen microliters of the resulting solution was loaded in individual electrophoresis channel for separation. The electrophoresis conditions are as follows: gel, NuPAGE (Invitrogen, NP0321BOX); running buffer, Bolt MES SDS running buffer (Invitrogen, Booo2); voltage, 160 V; current, 120 mA; time, 40 min. Coomassie brilliant blue fast staining solution (Beyotime, P0017) was used to stain the SDS-PAGE protein at room temperature for 2 hours, and excess staining was removed by rinsing with deionized water. The protein bands from 10 to 150 kDa were taken and dehydrated in acetonitrile, incubated in 10 mM dithiothreitol (DTT) in 50 mM ammonium bicarbonate at 56°C for 40 min, incubated in the dark in 55 mM iodoacetamide in 50 mM ammonium bicarbonate at ambient temperature for 1 hour, and dehydrated again. Then, overnight in-gel digestion was carried out by adding sequencing-grade trypsin (2 ng μl⁻¹) in 50 mM ammonium bicarbonate at 37°C. The resulting peptides were extracted twice with 5% formic acid/50% acetonitrile and vacuum-centrifuged to dryness. All samples were resuspended in 0.1% formic acid aqueous solution before liquid chromatography–tandem mass spectrometry (LC-MS/MS) analysis. As the control, PBS without TNA was added to the bacteria solution and the other conditions were the same as the TNA group.

For LC-MS/MS analysis, samples were reconstituted in 0.2% formic acid aqueous solution, loaded onto a 100 μm × 2 cm precolumn, and separated on a 75 μm × 15 cm capillary column with laser-pulled sprayer. Both columns were packed in-house with Luna 3 μm C₁₈(2) bulk packing material (Phenomenex, USA). An Easy nLC 1000 system (Thermo Fisher Scientific, USA) was used to deliver the following HPLC gradient: 5 to 35% B in 60 min, 35 to 75% B in 4 min, then held at 75% B for 10 min (A = 0.1% formic acid in water, B = 0.1% formic acid in acetonitrile) at a flow rate of 300 nl min⁻¹. The eluted peptides were sprayed into a Velos Pro Orbitrap Elite mass spectrometer (Thermo Fisher Scientific, USA) equipped with a nano-ESI source. The mass spectrometer was operated in a data-dependent mode with a full MS scan [375 to 1600 mass/charge ratio (m/z)] in Fourier transform (FT) mode at a resolution of 120,000 followed by collision-induced dissociation (CID) MS/MS scans on the 10 most abundant ions in the initial MS scan. Automatic gain control (AGC) targets were 10⁶ ions for Orbitrap scans and 5 × 10⁴ for MS/MS scans. For dynamic exclusion, the following parameters were used: isolation window, 2 m/z; repeat count, 1; repeat duration, 25 s; and exclusion duration, 25 s.

Data processing was carried out using Thermo Fisher Scientific Proteome Discoverer 2.4 using a database downloaded from https://www.uniprot.org (organism: A. baumannii). Carbamidomethyl (Cys) was chosen as static modification, and oxidation (Met) was chosen as variable modification. Mass tolerance was 10 ppm for precursor ions and 0.6 Da for fragment ions. The maximum missed cleavage was set to 2. Peptide spectral matches were validated using the Percolator algorithm, based on P values at a 1% false discovery rate. For label-free quantitation, normalization mode was set to “total peptide amount.” The protein abundances were calculated by summing sample abundances of the corresponding peptides after normalization. The relative content change of each protein was calculated by dividing the TNA group with the PBS group.

Proteome analysis of the proteins inside the bacteria after TNA treatment

Overnight A. baumannii or PDR A. baumannii cultures were diluted at 1:50 and grown in LB medium to an early-mid exponential phase (OD₆₀₀ = 0.4 to 0.6) at 37°C. The cell suspension was centrifuged (6000g, 3 min), washed with SPSS twice, and adjusted to 1 × 10⁸ cells/ml (~0.5 OD₆₀₀) by protein-free M9 culture medium with 20 mM glucose. TNA (1 mg) was dissolved in 50 μl of DMSO and subsequently diluted to a concentration of 40 μM with PBS. Bacteria (10 ml) were incubated with 10 ml of TNA (40 μM) in an Eppendorf tube at 37°C for 5 min. Bacteria precipitation was obtained after centrifugation (6000g, 3 min) and removal of the supernatant. The proteins inside the bacteria were collected by using the Bacterial Protein Extraction Kit (Sangon Biotech, C600596). As the control, PBS was used to treat the bacteria solution by following the same procedure. The methods for the SDS-PAGE separation, LC-MS/MS analysis, and data processing of proteins in bacterial extracts were the same as those in the previous section.

In vitro immune activation evaluation by ELISA and flow cytometry

Mouse PEMs were from peritoneal washings of female Balb/c mice [age 6 to 8 weeks, ~20 g, Chongqing ENSIWEIER Laboratory Animal Co Ltd. (China)] and cultured in RPMI 1640 medium (Thermo Fisher Scientific, 11875093) plus 10% fetal bovine serum (FBS) (Gibco, 10091148) with granulocyte macrophage colony-stimulating factor (10 ng ml⁻¹; Sigma-Aldrich, 234374) for 7 days. Mouse BMDCs were separated from leg bone marrow of female Balb/c mice (age 6 to 8 weeks) and cultured in RPMI 1640 medium plus 10% FBS with granulocyte macrophage colony-stimulating factor (100 ng ml⁻¹; Sigma-Aldrich, 234374) for 7 days. TNA (1 mg) was dissolved in 50 μl of DMSO and subsequently diluted with PBS. To study the phenotype of immune cells and cytokine secretion, mouse BMDCs and PEMs were seeded in six-well plates (10⁶ cells per well) and cultured in RPMI 1640 medium plus 10% FBS with TNA (20 μM), supernatant of PDR A. baumannii (5 × 10⁹ CFU ml⁻¹, 0.2 ml), or supernatant of PDR A. baumannii (5 × 10⁹ CFU ml⁻¹, 0.2 ml) plus TNA (20 μM) for 12 hours. Afterward, the supernatant was collected and the production of TNF-α (Beyotime, PT512), IL-12 (Beyotime, PI530), and IL-1β (Beyotime, PI301) was estimated using an enzyme-linked immunosorbent assay (ELISA) kit. To analyze phenotypes with flow cytometry, cells were then harvested and stained using fluorescent monoclonal antibodies for multicolor flow cytometry analysis. Because the antigens to be analyzed are mainly on the surface of the cells, the cells were not fixed. Specifically, cells were first collected and resuspended with PBS containing 2% FBS at 4°C for 1 hour to block unspecific protein binding. For macrophage immunophenotype analysis, PEMs were stained with anti-CD11b–phycoerythrin (PE)–Cy7 (2 μg ml⁻¹), anti-F4/80–Alexa Fluor 488 (2 μg ml⁻¹), anti-CD206–Alexa Fluor 647 (2 μg ml⁻¹), and anti-CD86–PE (2 μg ml⁻¹) antibodies at 4°C for 30 min, centrifuged (200 g, 3 min), and washed with PBS (containing 2% FBS) twice. For DC maturation analysis, BMDCs were stained with anti-CD11c–FITC (2 μg ml⁻¹), anti-CD80–allophycocyanin (2 μg ml⁻¹), and anti-CD86–PE (2 μg ml⁻¹) antibodies at 4°C for 30 min, centrifuged (200 g, 3 min), and washed with PBS (containing 2% FBS) twice. PEMs and BMDCs were finally rinsed with PBS containing 2% FBS and analyzed with a flow cytometer (ACEA Biosciences, NovoCyte 2060R). Flow cytometry data were analyzed using FlowJo V10.8.1 and GraphPad Prism software.

To evaluate the phagocytic capability of PEMs after TNA or PBS treatment, PEMs were seeded into six-well plates (10⁶ cells per well) and cultured with TNA (20 μM) or PBS for 12 hours. Afterward, PEMs were first stained with Dil (Beyotime, C1036), then infected with carboxyfluorescein diacetate succinimidyl ester (CFDA-SE) (Beyotime, C1031) labeled PDR A. baumannii (5 × 10⁶ CFU). Sixty minutes after infection, cells were washed three times with PBS to remove floating bacteria and analyzed with a flow cytometer.

To investigate the influence of TNA on bacteria for the uptake of professional phagocytes, PDR A. baumannii were labeled by CFDA-SE (Beyotime, C1031). TNA (1 mg) was dissolved in 50 μl of DMSO and subsequently diluted with PBS. CFDA-SE–labeled PDR A. baumannii (1 × 10⁸ cells/ml) were incubated with TNA (20 μM) in an Eppendorf tube at 37°C for 20 min, and then the TNA-treated PDR A. baumannii (CFDA-SE labeled) were centrifuged (6000g, 3 min) and washed with SPSS twice. Mouse PEMs were seeded into six-well plates (10⁶ cells per well), stained with Dil (Beyotime, C1036), and incubated in RPMI 1640 medium plus 10% FBS with TNA-treated PDR A. baumannii (CFDA-SE labeled) (5 × 10⁶ CFU) or untreated PDR A. baumannii (CFDA-SE labeled) (5 × 10⁶ CFU). Sixty minutes after incubation, cells were washed three times with PBS to remove floating bacteria and analyzed with a flow cytometer.

TI experiments

All animal experiments were in accordance with protocol no. SYXK 2020-0006, which was approved by the Institutional Animal Care and Use Committee (IACUC) at the Southwest University, China, and in accordance with the National Guide for Care and Use of Laboratory Animals, China. Animal experiments were conducted in accordance with ARRIVE guidelines. PDR A. baumannii was introduced to female Balb/c mice (age 6 to 8 weeks) intraperitoneally with a minimum lethal dose of 3.5 × 10⁹ CFU ml⁻¹ (0.5 ml). At this lethal dose, 100% of mice died within 48 hours after infection. Immediately after the infection, the 100 female Balb/c mice (age 6 to 8 weeks) were randomly divided into 10 groups and received intravenous injection of TNA at doses of 0.15, 0.29, 0.58, 0.87, 1.16, 11.2, 22.4, 44.8, 67.2, and 89.6 mg kg⁻¹ [TNA (0.5 g) was dissolved in 1 ml of DMSO and subsequently diluted to a concentration of 2 mg ml⁻¹ with SPSS]. The mice were monitored for survival over a 7-day period after infection and TNA treatment. The calculated ED₅₀ (effective dose at which half the mice survive) and LD₅₀ (lethal dose at which half the mice are sacrificed) of TNA were 0.4460 and 31.66 mg kg⁻¹, respectively, analyzed by the GraphPad Prism software. Hence, the TI (LD₅₀/ED₅₀) of TNA was 71. For the in vivo end-point evaluations, the investigators were blinded to group allocation during data collection and analysis. No data were excluded from the analyses.

Electron microscopic study of bacteria disruption by TNA in vivo

Fifteen female Balb/c mice (age 6 to 8 weeks) were randomly divided into three groups and infected with PDR A. baumannii (2.0 × 10⁷ CFU) on the artificial wound (diameter = 8 mm) of the right leg muscle. TNA (0.5 g) and colistin were dissolved in 1 ml of DMSO and subsequently diluted to a concentration of 1 mg ml⁻¹ with SPSS. On day 1 of infection, the three groups of mice were treated with TNA (5 mg kg⁻¹), colistin (5 mg kg⁻¹), and SPSS. Mice were sacrificed on day 2, and the wound tissues were harvested and fixed with paraformaldehyde (4%). The tissues were then washed with PBS and treated with 0.5% OsO₄ at 4°C for 1 hour. After rinsing with deionized water, the tissues were immersed in saturated uranyl acetate in 70% ethanol at 25°C for 1 day. Subsequently, the tissues were dehydrated with 99.9% ethanol and then embedded in Epon resin. Ultrathin sections (~70 nm) were obtained with an ultramicrotome (Leica, EMuc7). The electron micrographs of bacteria and tissues were obtained with a transmittance electron microscope (Hitachi-7500, Japan).

In vivo immune activation evaluation by flow cytometry

The infiltrating immune cells in spleen and IDLNs were collected and grinded at the predetermined time (five mice at each predetermined time), digested, and filtered through 70-μm filters to obtain the single-cell suspension. These groups of cell suspensions were blocked with 2% FBS and staining with corresponding antibodies for further flow cytometry analysis. Specifically, the cells were first collected and resuspended with PBS containing 2% FBS at 4°C for 1 hour to block unspecific protein binding. Because the antigens to be analyzed are mainly on the surface of the cells, the cells were not fixed. For macrophage immunophenotype analysis, PEMs were stained with anti-CD11b–PE-Cy7 (5 μg ml⁻¹), anti-F4/80–Alexa Fluor 488 (5 μg ml⁻¹), anti-CD206–Alexa Fluor 647 (5 μg ml⁻¹), and anti-CD86–PE (5 μg ml⁻¹) antibodies at 4°C for 30 min. For DC maturation analysis, BMDCs were stained with anti-CD11c–FITC (5 μg ml⁻¹), anti-CD80–allophycocyanin (5 μg ml⁻¹), and anti-CD86–PE (5 μg ml⁻¹) antibodies at 4°C for 30 min. For cytotoxic T lymphocyte and helper T lymphocyte activation analysis, the corresponding spleen cells were stained with anti-CD45–PE-Cy7 (5 μg ml⁻¹), anti-CD3–FITC (5 μg ml⁻¹), anti-CD4–PE (5 μg ml⁻¹), and anti-CD8–allophycocyanin (5 μg ml⁻¹) antibodies at 4°C for 30 min. For T_reg analysis, the spleen cells were stained with anti-CD25–FITC (5 μg ml⁻¹), anti-CD4–PE (5 μg ml⁻¹), and anti-Foxp3–Alexa Fluor 647 (5 μg ml⁻¹) antibodies at 4°C for 30 min. For B cell maturation analysis, the spleen cells were stained with anti-CD45–FITC (5 μg ml⁻¹), anti-CD19–PE (5 μg ml⁻¹), and anti-CD38–peridinin chlorophyll protein (PerCP)–Cy5.5 (5 μg ml⁻¹) antibodies at 4°C for 30 min. Afterward, the above cells were centrifuged (1000g, 3 min) and washed with PBS (containing 2% FBS) twice. After staining, the above cells were centrifuged (200g, 3 min), washed with PBS (containing 2% FBS) twice, rinsed with PBS (containing 2% FBS), and analyzed with a flow cytometer (ACEA Biosciences, NovoCyte 2060R). Flow cytometry data were analyzed using FlowJo V10.8.1 and GraphPad Prism software.

In vivo immune activation evaluation by immunofluorescence staining

The immune activation effects were analyzed by immunofluorescence staining. Simply, the spleen was collected at the predetermined time (five mice at each predetermined time), fixed with formaldehyde, embedded in paraffin, and sectioned with a tissue microtome (Leica, RM2016). The tissue sections were deparaffinized and rehydrated by consecutively immersing in xylene, ethanol, and water. Antigens were retrieved by immersing the tissue sections in EDTA antigen retrieval buffer (pH 8.0) and keeping at a sub-boiling temperature in a microwave for 8 min. After air cooling to room temperature, the tissue sections were washed three times with PBS. For the blocking of endogenous peroxidase, the tissue sections were immersed in 3% H₂O₂ for 25 min at room temperature. After washing three times with PBS, the tissue sections were blocked with 3% BSA for 30 min at room temperature. After discarding the blocking solution, the tissue sections were incubated with the primary antibody overnight at 4°C.

For T cell activation analysis, two sectioned tissues were stained with primary antibody anti-CD4 and anti-CD8 rabbit anti-mouse antibodies, respectively. After washing the primary antibodies three times with PBS, the sectioned tissues were incubated with a secondary antibody, horseradish peroxidase (HRP)–conjugated goat anti-rabbit IgG, at room temperature for 50 min in the dark. After washing the secondary antibodies three times with PBS, the sectioned tissues were incubated with Cy3-tyramide working solution and 0.003% H₂O₂ for 10 min at room temperature in the dark. After washing the sectioned tissues with PBS, spontaneous fluorescence was quenched with an autofluorescence quenching kit (Vector Laboratories, SP-8400-15) for 5 min at room temperature, and then the cellular nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI) for 10 min at room temperature. After washing with PBS, the tissue sections were sealed between the cover slide and the slide with anti-fade mounting medium. The fluorescence was observed with a fluorescence microscope (Carl Zeiss Jena, LSM800).

For M1 macrophage analysis, the primary antibody of CD86 was anti-CD86 rabbit anti-mouse antibody. The secondary antibody and conditions of tyramide signal amplification (TSA) for the analysis of CD86 were the same as those in the T cell activation analysis. After CD86 staining, the primary and secondary antibodies of CD86 were removed by immersing the tissue sections in EDTA antigen retrieval buffer (pH 8.0) and keeping at a sub-boiling temperature in a microwave for 8 min. After air cooling to room temperature, the tissue sections were washed three times with PBS and then incubated with the primary antibody of anti-F4/80 rabbit anti-mouse antibody overnight at 4°C. After washing three times with PBS, the tissue sections were incubated with Alexa Fluor 488 goat anti-rabbit IgG at room temperature for 50 min in the dark. The methods for spontaneous fluorescence quenching, cellular nuclear staining, tissue sealing, and fluorescence observation were the same as those in the T cell activation analysis.

For M2 macrophage analysis, anti-CD86 rabbit anti-mouse antibody was replaced by anti-CD206 rabbit anti-mouse antibody, and the remaining procedures and methods were the same as those in the M1 macrophage analysis.

Antibody titer analysis

Ninety-six–well assay plates (Thermo Fisher Scientific) were coated overnight with 2 μg of PDR A. baumannii bacterial lysate per well at 4°C in ELISA coating buffer (Thermo Fisher Scientific, 00-0044-59). Plates were washed three times with 200 μl of PBS supplemented with 0.05% (v/v) Tween 20 (J&K Scientific, 573169) and blocked with 200 μl of 5% (w/v) milk (Sangon Biotech, C520013-0500) dissolved in PBS supplemented with 0.05% Tween 20 for 2 hours. After washing three times with PBS, 100 μl of serially diluted serum samples was added and incubated in the plates for 2 hours. After washing with PBS, 200 μl of HRP-conjugated goat anti-mouse IgG (HRP-IgG) (Abcam, ab6789) at a 1:5000 dilution was added and incubated in the plates for another 2 hours. After washing with PBS to remove floating HRP-IgG, the plates were incubated in 100 μl of the mixture of 3,3′,5,5′-tetramethylbenzidine (1 mM) (Beyotime, ST1708) and H₂O₂ (2 mM) for 10 min, and the reaction was stopped with 100 μl of H₂SO₄ (2 M). The absorbance was recorded at 450 nm, with 570 nm as the reference using a microplate reader (Molecular Devices, SpectraMax iD3). Antibody titers were determined by fitting the data to a four-parameter logistic curve and setting the naïve mice as interpolation threshold such that the average titers of naïve mice were approximately 10.

Statistical analysis

Data are presented as mean ± SD. One-way analysis of variance (ANOVA) was used for multiple-group comparison. Intergroup comparison was analyzed by Student’s t test (two-tailed). *P < 0.05, **P < 0.01, and ***P < 0.001.

Supplementary Materials

This PDF file includes:

Figs. S1 to S62

Tables S1 to S9

Other Supplementary Material for this manuscript includes the following:

Data file S1