Hospital-acquired infections (HAIs), synonymously called nosocomial infections, are detrimental to patient safety and recovery. HAIs are a threat in all hospitals, but the intensive care unit (ICU) documents the highest HAI rates1. Moreover, HAIs are most common with the central line bloodstream and ventilator usage costing an extra 9.5 and 9.1 days of hospital stay, respectively2. Annually, over 12 million deaths occur due to HAIs globally, of which 95% occur in low- and middle-income countries (LMICs)3. In the United States alone, over 2 million infections are caused by HAIs, particularly associated with antibiotic resistance (AR) and/or multidrug-resistant organisms (MDROs), of which about 23,000 fatalities occur, annually2.
Gram-negative bacteria are considered more resistant to current antibiotics compared to gram-positive bacteria mainly due to the presence of cell wall that limits the uptake of antibiotics in the former and is lacking in the latter4. Escherichia coli is among the gram-negative bacteria commonly causing HAIs5, including food-borne outbreaks6, and ultimately deaths5. In virulent strains such as E. coli O1:K1:H7 (CN1018), expression of adherent fimbriae has been associated with increased bacterial survival and inflammatogenic response in the urinary tract infection mouse model7. Moreover, bacterial isolates lacking expression of fimbriae have demonstrated reduced growth, adhesion, biofilm formation and virulence, thereby resulting in low infections and improved survival in animal models and humans when infected with the virulent strains8. Hence, a need for better antimicrobial agents that can specifically target the cell wall and/or adherent fimbriae expression cannot be overstated, and engineered nanomaterials (ENMs) may have the potential to serve as the next generation antimicrobial agent9.
ENMs are being used in various commercial applications (e.g., fabric, mask, medical device, paint, cement, etc.), but the potential factors and mechanisms underpinning their antimicrobial activities have remained unclear10-12 as results are largely compounded by the heterogeneity of the ENMs used, endpoints measured, and differential sensitivity of the bioassays used. As the studies on the actual mode of ENM toxicity is ongoing, recent literature suggest that the capping ligand, core particle size, and surface charge likely play a role13-17. In the recent rule published by the USEPA under TSCA (Toxic Substances Control Act), the roles of inherent particle attributes, including the particle size, surface charge (measured as zeta potential), aggregation state, or surface reactivity on ENM toxicity, were emphasized as factors commonly linked to nanotoxicity18. Necessary for future endeavors is an understanding of desired ‘nano’ properties, so certain characteristics may be wishfully tailored to render ENM safer (e.g., for environmental remediation), or more toxic (e.g., for neutralizing AR/MDROs).
Molecular or atomic scale interactions occurring at the nanoscale was previously difficult to ascertain optically owing to the Abbe diffraction19. Nonetheless, routine use of electron microscopy in tandem with surface analytical spectroscopic techniques have now enabled researchers to track individual NPs on the cell surface and its interior (cytoplasm and nucleus). Recently, Werner and colleagues19 asserted that “it would be advantageous to require the nanomaterial to be as minimally specific as necessary in order to act as delivery vector, nanosensor, or imaging agent.” This is crucial given that a given NP has several physico-chemical properties, among which only a few characteristics are routinely measured, including particle size, zeta potential, surface linker, aggregation, etc. Adhering to the ‘Occam’s razor’, also called ‘philosophy of parsimony’20, a minimal set of defined physicochemical properties of ENMs that are/can be routinely measured may enable designing ENMs for desired applications such as treating the HAIs. Herein, employing two different sizes AgNPs (5 nm versus 45 nm) with high contrasting surface charges (high positive versus high negative), including dissolved ionic Ag+ as positive control, this study aimed at: (1) elucidating potential factors and mechanisms influencing nano-bio interactions and toxicity in E. coli, a common gram-negative bacillus with high AR/MDR potential; (2) using this information, determining the most potent antibacterial AgNPs; and (3) documenting potential phenotypic plasticity (i.e., shape-shifting) in E. coli in response to different silver treatments.
Purified AgNPs were characterized using the TEM and images were analyzed using an ImageJ. Average TEM diameter of the NH2–AgNPs was 5.8 ± 2.8 nm, and were spherical in shape, crystalline, and had a high positive average zeta potential of +41.6 mV due to the cationic amino groups on the NP surface (Fig. 1; Table 1). The uniform NH2- coating thickness on the surface of AgNPs was measured in the range 0.5-1.5 nm, thereby enabling AgNPs’ stability via electrosteric repulsion. Average TEM diameter of Citrate–AgNPs was 44.8 ± 5.0 nm, and were pseudospherical in shape, crystalline, and had a high negative average zeta potential of -30 mV due to the anionic carboxyl groups on the NP surface (Fig. 1; Table 1). The uniform citrate-coating thickness on the surface of AgNPs was 1.8 nm, which enabled NPs stability via electrostatic stabilization. Further, TEM analysis of AgNPs morphology suggests that both the NH2- and citrate- ligands could effectively confer stability to AgNPs by preventing aggregation, and thus the nano-suspensions were observed to be highly stable in aqueous suspension over several years. Dynamic light scattering (DLS) analysis of average hydrodynamic diameter (HDD) for NH2–AgNPs was 4.3 nm and 11 nm for Citrate–AgNPs. UV-Vis spectrophotometry analysis showed localized surface plasmon resonance (λmax) at 416.5 nm for NH2–AgNPs, and 425 nm for Citrate–AgNPs, and that dilution (Supplementary Information Fig. S1), incubation time (72 h) and temperature (35 °C) had no effect on the stability of both the AgNPs (Supplementary Information Table S2).
Our results showed that E. coli population increased over the 72 h period for the negative controls (Fig. 2). Citrate–AgNPs (0.5 or 10 µg/mL), or Ag+ ions particularly at 0.5 µg/mL, also showed positive cell growth (OD600 > 0.6 a.u.) over 72 h. However, exposure to 10 µg/mL NH2–AgNPs showed, after initial acclimation, a steady decline in E. coli growth with population completely crashing by 72 h, confirming a bactericidal effect of NH2–AgNPs at 10 µg/mL (Fig. 2). Exposure to 10 µg/mL Ag+ ion demonstrated strong bacteriostatic effect (Fig. 2). Further, at 0.5 µg/mL NH2–AgNPs inhibited bacteria growth more effectively than the same concentration of Citrate–AgNPs or Ag+ ions (Fig. 2). Taken together, these results indicate that NH2–AgNPs possess strong bactericidal activity, while Ag+ ions demonstrated strong bacteriostatic effect, at 10 µg Ag/mL level.
Potential interaction with the bacterial cell wall upon physical contact with NPs is a plausible explanation for antibacterial effects of AgNPs12. Small size (5 nm diameter) AgNPs could competitively interact with bacterial cell wall due to electrostatic attraction (Fig. 4A). Further, high positively charged cationic AgNPs (with amino [-NH2+] as surface ligand) demonstrated higher affinity to bacterial cell surface binding compared to the negatively charged AgNPs (with citrate [-COO-] moiety) (Figs. 3 and 4).
We further assessed if E. coli cell diameter is a predictor of cell length using the General Linear Model. Results showed that cell diameter is a strong predictor of cell length (Supplementary Information Table S3). Plotting cell diameter versus cell length (Fig. 2B, C) revealed distinct cell clusters reflecting differing cell morphologies. Cells treated with 10 µg/mL Ag+ ions demonstrated change in cell morphology from individual rod-shape to hexagonal honeycomb-like clusters of cells stacking together (Figs. 3F, 5). Such hexagonally stacked honeycomb phenotype is known to minimize surface area to volume ration, thereby significantly reducing available exposure surfaces per bacterial cell to the stressor, and in this case, Ag+ ions at 10 µg/mL21. Moreover, it has been suggested that hexagonal honeycomb structure, as observed in beehives, is likely the most economically feasible shape in terms of material and energy expended to form or build the honeycomb22. To our knowledge, this is the first study to report a honeycomb-like, potentially resistant cell phenotype in gram-negative E. coli upon Ag+ ions exposure. At 10 µg/mL NH2–AgNPs treatment, E. coli survival severely diminished before the population completely crashed by 72-h as verified by FE-SEM analysis (i.e., no cell could be located in the samples under the FE-SEM). This result confirms potent bactericidal effects of NH2–AgNPs at 10 µg Ag/mL, thereby avoiding the evolution of potentially resistant phenotype as observed for Ag+ ions.
It was recently documented that bacteria could form lenticular shaped chains upon metal stress. Chakravarty & Banerjee21 studied morphological changes in Acidiphilium symbioticum H8 upon exposure to Cu (as 500 mM CuSO4) and Cd (as 12.5 M CdSO4) and found that loosely packed coccobacillus-type cells (short rod-shaped) transformed into round shape and were highly packed together upon exposure to the metals. Likewise, Khusro et al.23 noted similar stress response induced in Bacillus subtilis KPA upon exposure to mercuric chloride, wherein treated cells presented round phenotype from the native rod-shape.
Because bacteria exposed to 10 µg/mL NH2–AgNPs for 72 h did not survive and thus the cells could not be located under the SEM, we then briefly exposed the bacteria for 10 min to the same concentration (10 µg/mL) of NH2–AgNPs and mapped AgNPs-cell interaction, potential cell wall damage, and elemental Ag validation across the selected areas as shown in Fig. 4. SEM analysis of the bacterial cells did not reveal morphological damage, but a larger particulate of Ag (size range 155-157 nm in diameter) was clearly found to be located at the end of each cell (see Energy Dispersive Spectroscopy [EDS] spectra #10 and #11 in Fig. 4). It was previously shown that smaller size and positively charged AgNPs are relatively more toxic that the larger size and negatively charged AgNPs12, but the NH2–AgNPs’ localization at the end of E. coli cells has been for the first time clearly imaged in this study at high resolution using the STEM mode. It can thus be inferred that 5 nm diameter, highly positively charged NH2–AgNPs precipitated upon electrostatic interactions with the bacterial cell surfaces and/or biomolecules within the cell exterior, thereby forming larger particulates of Ag ranging in diameter 155-157 nm (Fig. 4A). Furthermore, high positively charged NH2–AgNPs attachment to the bacterial cell surface may lead to “hyperpolarization” of the cell wall, more specifically, peptidoglycan that makes up the cell wall, as described recently for gold nanoparticle interactions with E. coli K1224, subsequently leading to adherent fimbriae inhibition and cell wall damage as observed in our study (Fig. 4, Table 2).
EDS surface elemental mapping of E. coli exposed to 10 µg/mL Ag+ ions for 72 h showed a greater amount of silver on the cell surface and were found associated with common biological elements such as C, O, Na, K, Cl, (Fig. 5). In comparison, E. coli exposed to 10 µg/mL Citrate–AgNPs for 72 h confirmed the above toxicity results showing lesser amount of silver on the cell surface and were found associated with such biological elements: C, O, Na, K, and Cl but not with P and S (Supplementary Information Fig. S2).
In E. coli dh5a, mannose resistant Proteus-like [MR/P]) fimbriae are expressed7. FE-SEM image analysis revealed the presence of filamentous adherent fimbriae in the negative control group, including in both concentrations of Citrate–AgNPs and low concentration (0.5 µg/mL) of Ag+ ions (Fig. 3; Table 2). However, adherent fimbriae were completely absent in E. coli when treated with NH2–AgNPs at both concentrations (0.5 and 10 µg/mL), changing them into non-motile phenotype (Fig. 4; Table 2). Fimbriae were also absent with Ag+ ions treatment, particularly, at 10 µg/mL, when bacteria changed their morphology to hexagonal honeycomb-like phenotype (Fig. 3F; Table 2). Fimbriae expression in pathogenic bacterial strains such as E. coli is known to promote adhesion to substrates, biofilm formation, and virulence, causing higher infections and reduced survival in animals and humans7,8.
Particle size vis-à-vis material-specific properties could govern nano-bio interactions12,13,19. Previous studies that modeled artificial phospholipid bilayer interacting with rod-like biomolecules suggest that the energy on the surface may influence surface molecule-cell membrane association, potentially leading to cell entry of biomolecules26,27 and toxicity. Energy barrier that exists between NPs and the biologic surfaces ought to be overcome28 before the NPs could interact with the receptor surfaces (cell wall or cell membrane). In our study, a remarkable difference in the magnitude of surface charge between the 5 nm NH2–AgNPs (z = +41.6 mV) and 45 nm Citrate–AgNPs (z = -30 mV) can be attributed to higher attraction forces between the E. coli surface and the NH2–AgNPs surface (Table 1), enabling direct physical contact between the two surfaces (as evidenced by EDS spectra showing particulates of silver on the E. coli surfaces; see Fig. 4), leading to increased cell death after 4 h followed by complete population collapse by 72 h. On the other hand, a significantly lower charge difference observed for Citrate–AgNPs suggests that the dominant repulsive forces (lower attraction; Table 1) could have played a key role in keeping most of the Citrate–AgNPs away from the biologic surfaces (as confirmed by EDS elemental mapping showing lower amount of silver on E. coli surfaces) and this might explain the lowest toxicity observed for Citrate–AgNPs (Figs. 2, 3C,D; Supplementary Information Fig. S2). Evidence from previous studies also corroborate a potential for direct physical interactions between the NPs (e.g., graphene and derivatives) and the biologic receptors29,30, suggesting direct physical interaction may serve as a primary mechanism explaining AgNP toxicity.
Shape- and size-shifting in E. coli have been known to occur in response to varied environmental stressors. It is hypothesized that bacteria can shrink or assume spherical (from originally rod-like) shape to minimize cell surface area when conditions become more stressful. Liu et al.31 investigated potential effects of ZnONPs (diameter 70±15 nm) in E. coli and observed deformed cell membranes that led to cell leakage but did not observe any morphological changes. In another study, AgNPs could potentially penetrated into E. coli cells, resulting in membrane degradation32. At 24 h post-exposure, a significant growth inhibition was recorded as a function of AgNPs concentrations32. Formation of irregular-shaped pits in the outer membrane was attributed to inhibition in cell growth and reproduction. Further, increased AgNPs concentration resulted in smaller bacterial size32. He et al.33 treated E. coli with MgONPs with an average size of 20 nm. Using SEM, the authors revealed disrupted membranes in E. coli exposed to 20 nm MgONPs, that led to the rough and leaky E. coli cells33. Measuring the average roughness of bacteria using atomic force microscopy before and after Catechin-CuNPs (5 nm diameter) treatments, Li et al.34 found that the untreated E. coli cell roughness was 6.32 nm, 4.88 nm (Rq, Ra), whereas for the treated cells the roughness increased significantly to 71.6 nm, 57.3 nm (Rq, Ra); and for Streptococcus aureus the roughness increased from 4.51 nm, 3.13 nm (Rq, Ra) to 76.1 nm, 59.9 nm (Rq, Ra). Such evidence strongly support the premise that NPs (e.g., Catechin-CuNPs) can potentially disrupt the basic cell membrane structure (i.e., form) and consequently the membrane functions34.
Understanding NP-induced oxidative stress in human cells is key in developing potent antibacterial that is also biocompatible for human use. Reactive oxygen species (ROS) may be generated upon NP exposure. Lipid peroxidation occurs when excess ROS affects cell membranes and can also lead to oxidation and denaturation of proteins and DNA damage, further inducing inflammatory immune responses and cell death35. Our results show no lipid peroxidation (MDA) in human lung epithelial (H-6053; Cell Biologics) and dermal fibroblast (106-05A-1526; Millipore Sigma) cells upon NH2–AgNPs treatments as the MDA levels were all below the background concentrations (negative control, dilution buffer). Rather, data revealed a decrease in MDA with up to 10 µg/ml of NH2–AgNPs treatments in lung cells compared to treatments with the negative control dilution buffer or the positive control (hydrogen peroxide) (Fig. 6). Such an ability to quench oxidative stress makes NH2–AgNPs a biocompatible antibacterial candidate.
AgNPs have been reported to alter homeostasis and membrane permeability in E. coli and Vibrio cholera36. In a separate study, Kim et al.37 documented AgNPs to form pits within the cell membrane, thereby increasing cell permeability. Research also indicates Ag+ ions could inhibit bacteria growth by disrupting DNA replication and inactivating thiol groups in many enzymes36. Bruins et al.38 suggested that metals could interact with cellular components through covalent and ionic bonding. At higher exposure levels, essential metals including Co, Cu, Ni, and Zn can also inhibit microbial growth and function39. Detail characterizations of surface functionalized AgNPs before and after interactions with the E. coli cells are paramount to better elucidate how these NPs could elicit toxicity to bacteria and other pathogenic microbes. Employing electron microscopy (Figs. 4-6) coupled with surface spectroscopic techniques (Figs. 3-5, Supplementary Information Fig. S2) allowed us to confirm our in vitro E. coli toxicity results (Fig. 2).
Designing near-atomic size 5 nm diameter NH2–AgNPs with high positive surface charge is found to be significantly inhibitory to E. coli compared to the larger size 45 nm Citrate–AgNPs with high negative surface charge. The bactericidal effects of NH2–AgNPs and the non-toxic effects of Citrate–AgNPs lend credence to the hypothesis that smaller size and highly positive surface charged AgNPs could serve as a next generation antibacterial agent to addressing the growing HAIs and patient health and safety.