An Ultrastable Self-Assembled Antibacterial Nanospears Made of Protein

Protein-based nanomaterials have broad applications in the biomedical and bionanotechnological sectors owing to their outstanding properties such as high biocompatibility and biodegradability, structural stability, sophisticated functional versatility, and being environmentally benign. They have gained considerable attention in drug delivery, cancer therapeutics, vaccines, immunotherapies, biosensing, and biocatalysis. However, so far, in the battle against the increasing reports of antibiotic resistance and emerging drug-resistant bacteria, unique nanostructures of this kind are lacking, hindering their potential next-generation antibacterial agents. Here, the discovery of a class of supramolecular nanostructures with well-deﬁned shapes, geometries, or architectures (termed “protein nanospears”) based on engineered proteins, exhibiting exceptional broad-spectrum antibacterial activities, is reported. The protein nanospears are engineered via spontaneous cleavage-dependent or precisely tunable self-assembly routes using mild metal salt-ions (Mg 2 + , Ca 2 + , Na + ) as a molecular trigger. The nanospears’ dimensions collectively range from entire nano-to micrometer scale. The protein nanospears display exceptional thermal and chemical stability yet rapidly disassemble upon exposure to high concentrations of chaotropes ( > 1 mm sodium dodecyl sulfate (SDS)). Using a combination of biological assays and electron microscopy imaging, it is revealed that the nanospears spontaneously induce rapid and irreparable damage to bacterial morphology via a unique action mechanism provided by their nanostructure and enzymatic action, a feat inaccessible to traditional antibiotics. These protein-based nanospears show promise as a potent tool to combat the growing threats of resistant bacteria, inspiring a new way to engineer other antibacterial protein nanomaterials with diverse structural and dimensional architectures and functional properties.


Introduction
Owing to their inherent properties, including biocompatibility, bioavailability, biodegradability, ease and cost-effective bioproduction, and structural versatility, protein molecules have been exploited to design novel nanomaterials without biological functions. [1] More recently, upon tuning their properties, protein-based nanomaterials are gaining wide-ranging and promising applications in biomedicine and nanotechnology . [2] These nanomaterials integrate characteristics and functionalities afforded by the different proteins. In particular, protein nanomaterial vaccines have been developed to tackle daunting infectious diseases. [3] Yet another crisis, this time stimulated by antibiotic-resistant bacteria or "superbugs", is gradually looming. [4] Nonetheless, to date, protein-based nanomaterials to combat these resistant bacteria are lacking [5] and have received relatively less attention, despite the recent increasing reports of antibiotic resistance and their overwhelming threats to global public health. [6] The direct global deaths due to antibiotic resistance are reported to be nearly equal to global human immunodeficiency virus (HIV) and malaria deaths combined. [6] While more prolonged antibiotic exposure due to their external misuse enables bacteria to mutate incessantly to develop resistance, the scarcity of novel structural antibacterial classes and targeted approaches exacerbates the situation. [7] Numerous conventional nanomaterials (for reviews [8] ), such as metal-based nanoparticles, quantum dots, polycationic polymers, low-dimensional materials, nanotubes, nanocomposites and liposomes, have been developed and possess remarkable antibacterial activities. In spite of these, some of these nanomaterials are putatively toxic and have safety concerns. [8a] They involve extensive, complex synthesis with toxic chemicals and are less stable at desirable solution conditions such as neutral pH. [8] To this end, we proposed or envisaged that protein nanostructures with combined unique structural or morphology-induced and enzymatic actions tailored to kill bacteria rapidly upon contact without prolonged external exposure could offer a promising strategy but are currently beyond reach or unattained. Hence, tailoring the structure of protein nanoantibacterial agents may be strategically beneficial to newly effective antibacterials [9] to intimately and selectively interact or puncture bacteria upon attachment without prolonged exposure. This will enable the evasion of continuous evolutionary mechanisms favoring antibiotic resistance upon external misuse. Furthermore, the tailored geometry, shape or nanoprotrusion with high-aspect-ratios will facilitate physical deformation, piercing, lysis and rupturing bacteria on contact, creating destructive pores.
In this work, we report the discovery of unique protein nanostructures, as a novel class of antibacterial agents, made of engineered proteins consisting of genetically fused antimicrobial peptide (AMP), endolysin and self-assembly peptide, and biosynthesized via a bottom-up strategy (Figure 1a) These protein nanostructures termed "nanospears" resulted from both rational design and surprising finding and their double-pointed tips with distinct sharp and tailored nanotographies or geometries offer potent bactericidal activities against resistant Gram-negative and Gram-positive bacteria.

Design, Biosynthesis, and Characterization
As illustrated or summarized in Figure S1 and Tables S1 and S2, Supporting Information, we selected five model candidates or components using rational design choices from literature, including a strong self-assembling peptide (P 11 4; P) linkers (flexible and helical (rigid) type), an endolysin (LysAB2; AB2) and antimicrobial peptide (LL-37; L). We then designed three new peptideprotein modules, T-P-AB2, T-AB2-L, and T-P-AB2-L (Figure 1b) via genetic fusion as nanostructure-building blocks to explore this aforementioned idea and desirable nanoarchitecture. Notably, the endolysin (LysAB2) and antimicrobial peptide (LL-37) were connected with a helical linker (AEAAAKEAAAKEAAAKA) to ensure the continuous helical kink required for membrane penetration [10] and stability. [11] Insertion of helical linker has been demonstrated to relatively enhance the bioactivity of fusion proteins by 115%. [11] We also reasoned that it would contribute to the desired or predicted protein material's formation. Similar linkers have been demonstrated to induce other protein-based nanostructure formation with specific nanoarchitectures without biological functions. [12] The engineered proteins were fused to a solu-bilizing tag (Thioredoxin) to maintain soluble expression, prevent possible aggregation, and controllably induce their selfassembly at the desired time after Tobacco Etch Virus (TEV) protease cleavage. In addition, we used native endolysin, LysAB2, without a self-assembly peptide tag or antimicrobial peptide as a control to help decipher the component contributions to the protein nanostructures. These proteins were produced in Escherichia coli, purified to near homogeneity, and cleaved to release the protein nano-building blocks designated as P-AB2, AB2-L and P-AB2-L ( Figure S1a, Supporting Information). These proteins were consistent with their expected molecular weights, as confirmed by electrospray ionization mass spectrometry (ESI-MS) (Table S3 and Figure S3b, Supporting Information). As evident in the SDS-polyacrylamide gel electrophoresis (SDS-PAGE) analysis ( Figure S1a, Supporting Information), all the cleaved proteins were pure, except AB2-L. Probing this, we observed by size exclusion liquid chromatography that AB2-L had distinct biochemical characteristics, as evidenced by two fractions compared to P-AB2-L with a single fraction ( Figure S3a, Supporting Information); likewise, P-AB2 and lysAB2 proteins.
To investigate this further, dynamic light scattering (DLS) analysis was used to determine the hydrodynamic sizes of the cleaved proteins. P-AB2-L and AB2-L showed narrow and broad size distribution peaked at 8 ± 2 and 1000 ± 200 nm after cleavage (Figure 2a). Figure 2b-d presents negative-staining transmission electron microscopy (NS-TEM) images of the protein (before) and nanostructures (after cleavage). Surprisingly, for AB2-L, these showed monodisperse protein structures with well-defined dagger-like shapes and double-pointed geometries ( Figure 2d). Conversely, NS-TEM images of P-AB2-L and T-AB2-L had no similar structure (Figure 2a,b).
We hypothesized that LL-37 might be responsible for the directed spontaneous assembly of AB2-L into the nanostructures. Moreover, we reasoned that unlike AB2-L, the flexible nature of the connecting linker (GS linker) coupled with hypothetical dynamic conformations or orientations of self-assembly peptide, P 11 4 possibly interfered with or prevented the P-AB2-L spontaneous self-assembly directed by LL-37 after cleavage. We, therefore, hypothesized that it might require mild external stimuli or triggers to tune the size. [13] Toward this end, upon adding 5 mm Mg 2+ ions, P-AB2-L spontaneously switched to broader size distribution with an average peak diameter of 300 ± 50 nm, and gradually increased in size with further increased Mg 2+ ion concentration ( Figure 2e). In contrast, the cleaved lysAB2 fractions showed narrow size distribution consistent with that of the P-AB2-L monomer (0 mm MgCl 2 ), regardless of the concentration of Mg 2+ ions added. P-AB2 formed nanoparticles in the presence of MgCl 2 ( Figure S4, Supporting Information). Similar to our prior proof-of-concept work using endolysin P128, [14] AB2 did not form any nanostructure under both triggers. It is for cell wall degradation, as detailed earlier and in Table S2, Supporting Information.
Due to this, we focussed on the AB2-L and P-AB2-L for the remaining part of the work. To confirm the native protein structure (without stain), we employed scanning electron microscopy (SEM) analysis. [15] Interestingly, the protein nanospears (both AB2-L and P-AB2-L triggered with 10 mm MgCl 2 ) had identical morphological features and even showed more striking "Y or T-shaped" dagger-like structures, albeit with similar desirable Figure 1. Design, biosynthesis, and self-assembly of antibacterial nanospears. a) A schematic overview of experimental workflow. First, candidate enzymes, peptides, and linkers were rationally mined from various sources using sequence analysis (Step A). Next, the components are rationally combined via protein engineering methods to design the biomimetic amino acid sequence or building blocks (Step A), and the engineered proteins are produced in genetically modified organisms (bacteria) and protein-based nanospears are fabricated via two main self-assembly routes (Step B). Finally, the protein nanospears are characterized using different techniques and antibacterial properties are evaluated (Step C). The experiments within each process were repeated several times. b) Schematic representation of the designed gene constructs of fusion proteins used in this study.
Next, we used NS-TEM and SEM to explore or characterize the morphology of the nanostructures for P-AB2-L in response to increasing MgCl 2 . Similarly, NS-TEM images showed polydisperse protein structures with well-defined dagger-like shapes and double-pointed geometries ( Figure 2h). P-AB2-L formed longer nanospears with increasing Mg 2+ ion concentration, and these nanospears could not be wholly captured in the TEM image. However, SEM images of the P-AB2-L nanospears from the respective grids confirmed these structures with similar dimensions ( Figure S5, Supporting Information).
Using image J, we next measured the morphology of the protein nanospears based on the NS-TEM images. The morphology In further validation, we characterized protein nanospears using confocal microscopy and protein fluorimetry by staining the structures with Thioflavin T dye, an extrinsic fluorophore that emits increasing fluorescence upon binding to amyloids. [16] The protein structures had similar morphological characteristics as observed in NS-TEM and SEM ( Figure 2I, and Figure S1b, Supporting Information). Notably, P-AB2-L nanospears had a similar increasing morphological trajectory with increasing MgCl 2 as in NS-TEM. Moreover, protein nanospears showed strong rapid fluorescence with increasing MgCl 2 concentration and protein concentration for AB2-L and P-AB2-L, respectively (Figure S1c,d, Supporting Information). In contrast, P-AB2 and LysAB2 showed no significant change in fluorescence even at 20 μm, which confirms that they do not form protein nanospears.
Previously, protein biomaterials have been fabricated using diverse salt ion triggers. [13,17] Inspired by this, we next explored the ability of other salt cations (Ca 2+ and Na + ) to induce the formation of P-AB2-L nanospears. As expected, these protein nanospears had similar morphological and biophysical characteristics as those reported using MgCl 2 ( Figures S7 and S8, Supporting Information). While these protein nanospears were a surprise discovery, it might have been partly directed by the LL-37 component based on the above data. Alternatively, based on previous studies, we surmised that the protein nanospears' assemblies might also be attributed to intermolecular interactions between proteinpeptide monomers, afforded by: i) rationally matched or linked rotational symmetry [1b,18] or distinct sequence content/structure (helical and coiled-coiled segments) [19] or linker type connecting the bifunctional domains; [20] and ii) metal-ion-mediated interaction or coordination between peptide (P 11 4) regions stabilized by electrostatic interactions, [13,17] especially for P-AB2-L nanospears fabricated with divalent cations.
The former premise might be valid for the AB2-L nanospears as its protein nanobuilding block contains an endolysin with primarily alpha-helix structure (C1) fused to four bundle helix LL-37 (C4) via helix forming peptide linker, which may illuminate or expand available rational design approaches (domainswapped, symmetry, or helical coil-coil directed) [18b] to create previously unexplored self-assembling nano-architectures or geometries with desirable functions. While divalent cations have mainly been exploited as a molecular trigger to make protein nanomaterials, [21] monovalent cations have been underexploited. Therefore, we are aware that protein nanospears fabricated using NaCl require more detailed interplay or analysis, which is currently under investigation, and related findings will be separately reported. Nonetheless, there is some evidence that a similar selfassembly peptide spontaneously forms hydrogels [22] in the presence of NaCl under similar physiological conditions, albeit at very high concentrations (6.3 mm).

Stability Property and Characterization
The stability and durability of antibacterial agents under different conditions are essential for their further development and widespread application in external physiological environments and biomedical settings. [23] To this end, we evaluated the protein nanospears' stability and resistance to varying thermal, chemical, and biological conditions (Table S6, Supporting Information) using dynamic light scattering (DLS), NS-TEM, SEM and differential scanning calorimetry (DSC). P-AB2-L triggered with 20 mm MgCl 2 was used for most of the assessment. The protein nanospears showed extremely high stability and resisted disassembly upon exposure to even high temperatures (≥90°C), chelating agents (≤20 mm EDTA), different buffers, and media exchanges or dialysis ( Figure S2, Supporting Information). Most importantly, the nanospears retained intact morphology up to 95°C and maintained their structural integrity upon storage at 4°C for a month (Figure 3a-f, and Figure S9a-d, Supporting Information). SEM images from the respective grids confirmed these structures with similar morphologies and dimensions ( Figure S10, Supporting Information). Nonetheless, some disassembly or structural defects were observed for protein nanospears incubated at 37°C for 3 days followed by 2 min heat-shock at 100°C.
In comparison, protein nanomaterial made from tendon collagen [24] disintegrated upon heating to 65°C. In further analysis, we characterized the structural and folding properties of the protein nanospears using differential scanning calorimetry (DSC) ( Figure S3c, Supporting Information). P-AB2-L nanospears fabricated with 20 mm MgCl 2 manifested the highest melting temperature compared to P-AB2-L, 5 mm MgCl 2 ; P-AB2-L monomer, P-AB2, AB2-L and LysAB2, which possibly suggests that increasing divalent cation concentration increase the thermal stability of the P-AB2-L protein nanobuilding block. Congruently, other nanomaterials coordinated by other divalent cation (Zn 2+ ) [25] demonstrated similar exceptional stability traits. However, despite the extreme stability, protein nanospears rapidly disassembled in the presence of higher concentrations of sodium dodecyl sulfate (SDS) (≥1 mm) ( Figures S3f and S9d, Supporting Information), probably via disruption/cleavage of metalcoordination bond, as sodium dodecyl sulfate (SDS) negative ions(SO 4 2− ) may swamp all the positive charges in solution. [26]

Antibacterial Properties, Activity, and Investigation
Next, we evaluated the antimicrobial properties of the protein nanospears via a bacteriostatic microbial inhibitory concentration (MIC) determination and bactericidal (Agar plate spread/count, see Experimental Section for further clarification) assays against two Gram-positive (Staphylococcus aureus; MSSA 25923 and MRSA 17234) and Gram-negative (Acinetobacter baumannii, ATCC 19606S and Ab-03-149.1) bacteria (Figure 3a), recognized as leading pathogens for resistant infections. [6] The MIC values are experimentally defined as the lowest concentration of an antimicrobial agent that completely prevented visible bacterial growth, as depicted in Figure 3b. AB2-L and P-AB2-L prepared with 20 mm MgCl 2 and NaCl were used for comparison. As shown in Figure 3b, Both AB2-L and P-AB2-L Mg ions demonstrated preferential inhibitory activity toward the bacterial strains. The MICs of P-AB2-L Mg ions against Gram-positive and Gramnegative pathogens were 4 and >10 μm, respectively. We observed AB2-L showed similar MICs values, albeit vice versa (6 and >10 μm for Gram-negative and Gram-positive strains). Surprisingly, unlike the afore-tested nanospears, P-AB2-L Na ions were very active against all the bacterial strains with a MIC value of 4 μm and displayed no preferential inhibitory activity (Figure 3b), which requires further investigation. Furthermore, in further antibacterial assay via agar spread plate method [27] (Figure 3c-f), we observed that P-AB2-L nanospears at 1×MIC elicited enhanced and rapid antibacterial activity against Gram-positive pathogen with increasing MgCl 2 by completely p values were calculated using one-way ANOVA followed by comparison between the treatment groups and respective controls (*p < 0.1, **p < 0.01, ***p < 0.001, and ****p < 0.0001). The experiment was performed in at least three biological replicates on different days. The error bars represent the standard deviation (SD) of the mean.
In contrast, P-AB2-L at 1×MIC had no activity against Gramnegative bacteria (Figure 3e), albeit completely killed all the bacterial cells at 2×MIC triggered with 5 mm MgCl 2 (Figure 3f). Nonetheless, P-AB2-L at 2×MIC gradually lost its killing ability with increasing MgCl 2 . Interestingly, AB2-L at 1×MIC completely killed the tested Gram-negative bacteria and lost its activity with increasing MgCl 2 . A representative photograph of agar plates for the respective assessment is available in Figure S3, Supporting Information. Considering the above preferential inhibitory action and difference in antibacterial activ-ities, we should note that: i) the Gram-positive bacteria (S. aureus) membrane possesses a significant lipid component, cardiolipin, which binds to divalent cations and forms complexes of negative curvature required for membrane destabilization or depolarization processes; [28] and ii) the Gram-negative's LPS dense outer membrane is stabilized ionic interaction between divalent cations (Ca 2+ , Mg 2+ ) and the negative phosphate group at the LPS base. [29] Therefore, increasing divalent cation concentrations may contribute to membrane permeabilization or depolarization in S. aureus and could make the outer envelope of Gram-negative bacteria impervious to the antibacterial agent. This might account for such interplay in inhibitory or killing activity, primarily www.advancedsciencenews.com www.advmat.de caused by similar divalent salt cations used in nanospear fabrication. This divalent cation effect on other antibacterial agents has also been confirmed in other studies. [30] Nevertheless, the nanospears' distinct or usual morphology has a remarkable effect on the improved bactericidal activity, partially comparable (but not reminiscent) to that observed for star architecture by Qiao and co-workers. [31] The enzymatic action (caused by endolysin component) is also a key player. These are based on the observation that: i) the membrane active ability of the divalent cations in this study under the same nanospear conditions was discounted, as the divalent cations were found to be inactive against the S. aureus tested above; and ii) the nanospear formed at 2×MIC with MgCl 2 had at least 2-log reduction in Gram-negative bacteria colonies.
The enzymatic action (caused by endolysin component) is also a key player. These are based on the observation that: i) the membrane active ability of the divalent cations in this study under the same nanospear conditions was discounted, as the divalent cations were found to be inactive against the S. aureus tested above; and ii) the nanospear formed at 2×MIC with MgCl 2 had at least 2-log reduction in Gram-negative bacteria colonies. Therefore, further elucidation of the superior antibacterial activity and related mechanism will be essential, extend this protein nanostructure design concept and thus support the earlier observations.
Toward this end, we employed NS-TEM to directly visualize the effect of protein nanospears on bacterial cell morphology (Figure 4, and Figure S4, Supporting Information). Prior to incubation with P-AB2-L nanospears, the bacterial cells showed intact cell envelope (Figure 4a, and Figure S4a, Supporting Information). However, after incubation or treatment with P-AB2-L nanospears in a similar antibacterial assay at either 2×MIC (for Gram-negatives) and 1×MIC (for Gram-positive), intriguingly nanospears possibly interacted, distorted and disrupted the bacterial cell envelope via various modalities to elicit enhanced antibacterial activity (Figure 4b-l, and Figure S4a-l, Supporting Information).
Additional morphological changes are available in Figure S11, Supporting Information. Similar changes to bacterial morphology were observed after treatment with AB2-L and P-AB2-L (triggered with NaCl) nanospears at their respective MIC values ( Figures S12 and 13, Supporting Information). In particular, AB2-L rapidly caused more pronounced lysis to Gram-negative bacteria ( Figure S11, Supporting Information). The SEM analysis or observations ( Figure S14, Supporting Information) also agreed with the NS-TEM data, which indicated that protein nanospear interaction or association leads to bacterial envelope disruption. Recently, a handful of conventional antibacterial nanomaterials have demonstrated multimodal mechanisms, [31,32] which supports our data, indicating that a single antibacterial agent may possess more than one bacterial disruption effect. However, these antibacterial agents fail to exhibit their actions definitively. For instance, per the preliminary mechanisms explored in these studies against Gram-negative bacteria, these nanoscale materials turn to skip the highly divergent peptidoglycan layer or cell wall, and may warrant the development of unforeseeable resistance. Besides, these progressive envelope association and disruption effects are distinguished from bacterial damages caused by the previously reported antibacterial agent to the best of our knowledge.

Mechanism of Action Studies of Protein Nanospears
To further decipher the individual damage of the cell envelope components (outer membrane, cell wall and inner membrane) caused by the different protein nanospears, we performed fluorescence assays to study the effects of specific dye uptake (1-(N-phenylamino)naphthalene (NPN), propidium iodide (PI), and 3,3′-dipropylthiadicarbocyanine iodide (DiSC3(5))), on the bacterial cell membranes in the presence of the protein nanospears ( Figure S5a,b, Supporting Information). NPN is an uncharged, hydrophobic dye used for outer membrane (OM) permeabilization assay that weakly fluoresces in an aqueous environment but emits strong or bright fluorescence upon contact with a hydrophobic environment as a bacteria membrane. [33] NPN only permeate damaged bacterial OM with compromised integrity. As shown in Figure 5c, compared to the positive control, P-AB2-L monomer and AB2-L, bacterial cells (Ab 19606S) exposed to P-AB2-L at 2×MIC had a decreasing fluorescence intensity as the MgCl 2 concentration increased. In contrast, P-AB2-L at 1×MIC emitted increased fluorescence with increasing NaCl concentration, agreeing with the earlier bacterial killing data.
PI is a fluorescent dye used for bacterial membrane penetration assay to detect dead cells, penetrate bacterial cells with compromised membranes and emit bright red fluorescence upon binding to DNA. [34] Similar to the NPN assay, P-AB2-L treated Ab19606S generated decreasing PI fluorescence intensity with increasing MgCl 2 , albeit MRSA 17234 treatment quickly generated high PI fluorescence intensity with increasing Na ion trigger ( Figure S5d, Supporting Information). A similar PI fluorescence intensity profile was observed for AB2-L at 1×MIC in MgCl 2 , albeit fluorescence decreases in the presence of NaCl ( Figure S5e, Supporting Information). In contrast, both Ab19606S and MRSA 17234 subjected to similar treatment showed increased PI fluorescence intensity ( Figure S15a, Supporting Information). The PI kinetics also further confirms the bacterial killing data. DiSC 3 (5) is a potentiometric fluorescent dye used for cytoplasmic membrane (CM) depolarization that accumulates and aggregates in the phospholipid membrane and self-quench its fluorescence. DiSC 3 (5) moves to the outer environment and generates fluorescence when the membrane becomes destroyed or depolarized. [27] Surprisingly, both P-AB2-L and AB2-L nanospears depolarized the cytoplasmic membrane of Ab19606S regardless of the trigger or salt added (Figures S5f-i and S15e, Supporting Information). However, Ab19606S treated with the monomeric AB2-L at a similar concentration generated no observed DiSC 3 (5) fluorescence intensity change, explicitly indicating that protein nanospears(at 2×MIC) killed A. baumannii via a different mechanism (possibly due to morphology or structure effect), which by-pass the earlier outer envelope stabilization effects by divalent cations. [29] Also, none of the protein nanospears depolarized the S. aureus cytoplasmic membrane. Overall, these data suggest that the protein nanospears exert an inhibitory or killing effect by possibly penetrating or permeabilizing bacterial membranes, which corroborates observations in the bacterial morphological changes. Finally, we probed the enzymatic action of the protein nanospears due to the endolysin component (LysAB2) in a turbidimetric/degradation assay using a common lysozyme substrate (lyophilized Micrococcus lysodeikticus cell) via absorption measurement ( Figure S6a, Supporting Information). [35] Similar to lysozyme, endolysin (LysAB2) hydrolyzes the cell wall's (1→4) glycosidic bond between N-acetyl-d-glucosamine (NAM) and Nacetylmuramic acid (NAG) ( Figure S6b, Supporting Information). As shown in kinetics (decay) Figure S6c,d, Supporting Information, the absorbance or scattering decreases as the substrate is enzymatically degraded. Further fitting the absorbance decay ( Figure S6d, Supporting Information) with first-order Figure 5. Schematic of proposed models of spontaneous organization and self-assembly mechanism for protein nanospear formation process. a) Following cleavage of solubilizing tag (Thioredoxin, T) from T-AB2-L fusion protein, AB2-L monomers spontaneously interact and stack, leading to subsequent self-assembly into short and monodisperse protein nanospear directed by LL-37(L). The helical linker possibly maintained a rigid conformation and spatial distance between LysAB2 and LL-37. b) Following cleavage of solubilizing tag (Thioredoxin, T) from T-P-AB2-L fusion protein, self-assembly peptide P 11 4(P) possibly prevents or interferes with the spontaneous LL-37 directed self-assembly in P-AB2-L into nanospears (upper panel, 1). The flexible GS linker possibly caused dynamic and flexible P 11 4 conformational changes. However, in the presence of divalent cations/salts, the metal ions (e.g., Mg 2+ ions) bind to/coordinate with specific amino acids at the peptide-peptide interface to allow or subsequently induce nanospear formation. The addition of excess divalent cations form longer and polydisperse protein nanospears (lower panel, 2). Color representation in the cartoons: thioredoxin, T (orange); self-assembly peptide P 11 4, P (green); flexible GS linker (black); endolysin LysAB2, AB2 (pink); helical/rigid linker (blue); antimicrobial peptide LL-37, L (light green).
kinetics determined the relative activity ( Figure S6e, Supporting Information). The P-AB2-L protein nanospears displayed little to no enzyme activity with increasing MgCl 2 concentration (5 and 20 mm). Nonetheless, as expected, AB2-L nanospears retained approximately 50% of the endolysin activity. Importantly, it displayed a shorter half-life compared to the free endolysin counterparts. This reduction in protein nanospears' enzymatic action is probably a consequence of endolysin's reduced assess or availability to the substrate compared to the other free counterparts and has been previously observed in alginate-lysozyme nanoparticles. [36] Regardless, it is remarkable that enzymatic activity was mainly retained, which is plus for this nanoantibacterial agent.

Mammalian Cell Viability
To assess or investigate the biocompatibility of our protein nanospears, we performed a viability assay on mammalian cells using Alamar blue. To our surprise and contrary to the antibacterial effects, our nanospears at 1×MIC and 2×MIC were not toxic to both NIH-3T3 fibroblast and HeLa cells ( Figure S16, Supporting Information). The observed mammalian cell selectivity was hypothesized and envisioned earlier. It could be delineated by the fact that the outer surface of the bacteria cells is more negatively charged than the mammalian cells. [37] Hence, justifying the better affinity and penetration or lysis of the bacteria cells. Moreover, the cholesterol available in the mammalian cell membrane probably helped stabilize the membrane surface, [38] making them refractory to the disruption by the protein nanospears. To further corroborate, it was recently reported that adding cholesterol to cubic-phase nanoparticles drastically improved their toxicity or biocompatibility to HeLa cells. [39] Notwithstanding, it should be noted that mammalian cell culture models used in this work may not wholly emulate the in vivo toxicity profile. Regardless, our "immediate" intended application of these protein nanospears is cream or gel formulations for the topical treatment of bacterial infections. In addition, due to their stable properties, they may be formulated into sprays or wipes to disinfect external surfaces.

Molecular Model of Nanospear Formation and Mechanism of Action
Taken together, we have utilized peptide assembly and engineering routes to fabricate protein nanospears, which could arguably be due to rational design and serendipity.
Based on these findings and the collective biophysical analysis, we propose a molecular model of how the two peptides induce protein nanospears formation ( Figure 5). In the spontaneous self-assembly pathway (Figure 5a), after thioredoxin cleavage from T-AB2-L fusion protein, AB2-L monomers, regardless of the concentration, interact and stack together, leading to lateral supramolecular self-assembly to form short and monodisperse nanospears. The LL-37 possibly directed and promoted the selfassembly of AB2-L into the supramolecular arrays. Conversely, after thioredoxin cleavage from T-P-AB2-L fusion protein, P-AB2-L do not undergo spontaneous self-assembly (Figure 5b, upper panel 1), possibly because of the dynamic, flexible and bent conformation switching ability of the flexible GS linker connecting P 11 4 and LysAB2. This enables P 11 4 to block or interfere with LL-37-driven assembly and nanospear formation in P-AB2-L. However, in the presence of divalent cations (Figure 5b, lower panel 2), the metal ions bind, interact or coordinate with specific amino acids at P 11 4 peptide-peptide interface within the monomeric P-AB2-L protein, leading to properly oriented coordination motifs in a possible linear configuration. This allows subsequent LL-37directed P-AB2-L self-assembly and nanospear formation. Furthermore, the addition of excess divalent cations induces the formation of longer and polydisperse protein nanospears that grow bi-directionally or bi-dimensionally.
In addition, as evident in SEM, TEM and kinetic assays, we postulate that protein nanospears demonstrate an exceptionally multifactorial or multi-complementary mechanism of contact bacterial killing, which most probably proceeds via morphological induced bacterial membrane association/attachment, perturbation and penetration, leading to cell wall hydrolysis, membrane depolarization and eventually to bacterial cell lysis or fragmentation ( Figure S7, Supporting Information). To the best of our knowledge, these successive antibacterial events observed of the protein nanospears are very notable, unique and substantially different from any reported antibacterial agent, none we could imagine at present.

Conclusion
Our findings demonstrated that antibacterial nanostructures with notable or unusual architectures could be engineered from proteins. These protein nanospears are unique and distinguished from any previously reported protein complexes or nanomaterials. [5,12,40] Besides their remarkable antibacterial activities, they can be produced from an environmentally benign biofabrication approach, are not toxic to mammalian cells and Generally Recognized as Safe (GRAS), biocompatible and biodegradable. Before this discovery, some antibacterial nanomaterials were reported to be putatively toxic and have safety concerns. [8a] Also, they require lengthy, complex synthesis and are less stable at desirable solution conditions such as neutral pH. [8a] Our protein nanospears have come to light to solve these problems.
Moreover, the high stability of the protein nanospears promises the potential external or topical application as an effective antibacterial agent. Furthermore, the knowledge discovered in this work will guide the design of next-generation selfassembled protein nanostructures that can effectively kill bacteria. Finally, the work has established a rational approach of linking specific components (proteins or biological molecules) to form robust protein nanostructures with unique functionalities or properties, expanding the possible avenues for supramolecular assemblies to afford unexplored 'impossible geometries far beyond the unusual architecture manifested by the protein nanospears.

Experimental Section
The experimental details, materials, and methods and associated references are available in the Supporting Information.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.