An AIE‐active probe for efficient detection and high‐throughput identification of outer membrane vesicles

Rapid detection and quantification of outer membrane vesicle (OMV) are of both scientific value and clinical implications. However, limited tools are available for investigations of OMVs. Herein, we report a novel fluorescent probe with aggregation‐induced emission (AIE) characteristics, namely, OEO‐TPE‐MEM (OTM), for OMV detection. OTM emits faintly in an aqueous medium, but its fluorescence could be effectively turned on upon interacting with bacteria bodies and OMVs produced by Gram‐negative bacteria. Notably, OTM could provide quantitative information on bacterial membrane remodeling and OMV secretion and be applied to high‐throughput screening of OMV‐inducing agents. This study presents a powerful AIE probe for imaging and quantitative analysis of bacteria envelop and derived OMVs, which might be applied for evaluating research and clinical antimicrobial materials in future studies.


INTRODUCTION
Outer membrane vesicles (OMVs) are spherical buds produced mostly by Gram-negative bacteria, which carry a series of bacterial molecules, such as lipopolysaccharide (LPS), flagellin, peptidoglycan, enzyme, DNA, and RNA. These bacteria components enable OMVs to participate in varied critical biological processes, such as pathogenesis, communication, quorum sensing, horizontal gene transfer, and stress response. Besides, OMVs serve as a crucial regulatory mediator on host immunity, [1][2][3] which enables OMVs to be engineered to induce humoral and/or cellular immunity and serve as adjuvants in vaccines, such as the vaccine against Neisseria meningitidis (MenB). More importantly, OMVs could be developed into cancer vaccines or cancer immunotherapy agents for preventing and treating cancers. [4][5][6] In spite of their promising implication in advanced biotechnology, our understanding of OMVs is still immature. The biogenesis of OMVs is not yet clear, and some scientists remain skeptical about the physiological roles of OMVs. The lagging understanding of OMVs may be partially ascribed to the limited tools for investigating OMVs. The small diameters of OMVs ranging from 20 to 250 nm [7,8] make OMVs very difficult to be observed and purified. Although proteomic analysis [9] and nanoparticle tracking analysis [10] are commonly used to study OMV production, tedious differential centrifugation or ultrafiltration is involved in OMV isolation. [11][12][13] Fluorescence-based techniques with high sensitivity have been widely used in biological studies, including those related to bacteria, [14][15][16][17] but it has seldom been applied to the detection of OMVs owing to lack of selectivity and high background fluorescence. Development of sensitive and wash-free fluorescent probes for OMVs would greatly accelerate research progress in OMVs. In 2001, a new class of fluorophores, known as the aggregation-induced emission luminogen (AIEgen), [18][19][20][21][22][23][24][25] was discovered, which demonstrated better performance in fluorescence imaging than traditional fluorescent probes with the aggregationcaused quenching effect. [26][27][28][29][30][31][32] Through specific targeting of AIEgens to desired bacterial strains by electrostatic interactions, [15,16,33] metabolic labeling following clickreaction, [34][35][36] or phage-guiding, [37] selective imaging of bacteria [38] has been achieved, but AIE-active probes for OMVs have not yet been reported.
Herein, we report an AIE-active probe (OEO-TPE-MEM, OTM) targeting to bacterial cell envelopes with superior properties in fluorescence imaging and quantification of OMVs. Under stress induced by antibiotics, membrane remodeling in Gram-negative bacteria was associated with a fluorescence increase of OTM in bacterial cell bodies and OMVs. Through applying a well-documented OMV inducer, ampicillin, we established a fluorescent method for fast and wash-free quantitative analysis of OMVs. The method was then applied to evaluating OMVinducing effects of several antimicrobial materials. In addition to bacterial imaging and quantification that can be achieved by using other reported fluorescent probes, this work provides a new imaging and quantitative probe for the evaluation of bacterial stress responses and OMV secretion.

RESULTS AND DISCUSSION
For rapid detection and quantitative analysis of OMVs, we decorated tetraphenylethylene (TPE) with a highly amphiphilic oligomeric ethylene glycol group and a quaternary pyridinium salt, which formed an electron donoracceptor system (OTM, Figure 1A). The positively charged quaternary pyridinium salt and the ammonium salt were expected to bind to the negatively charged LPSs and phospholipid membranes through electrostatic and lipophilic interactions. [15,39] OTM was characterized by 1 H NMR, 13 C NMR, and high-resolution mass spectrometry (Scheme S1, Figures S1-S9), from which satisfactory results corresponding to its molecular structure were obtained. OTM showed an absorption peak at 334 nm in pure water and an emission maximum at 605 nm in the solid state, resulting in a large Stokes shift of 271 nm ( Figure 1B). Notably, emission from OTM in the solution state was extremely low due to its AIE characteristics ( Figure 1B), which was favorable for imaging applications.
We first tested the envelope-targeting ability of OTM on Gram-positive bacteria (Staphylococcus epidermidis, S. epidermidis) and Gram-negative bacteria (Escherichia coli strain Nissle 1917, EcN). OTM stained the envelope of live bacteria in pure water ( Figure 1C). Nonetheless, its staining efficiency for Gram-positive and Gram-negative bacteria was distinguishably different: S. epidermidis was rapidly stained within 3 min but EcN was stained in 30 min (Figures 2A and  S10). This difference may be attributed to the shielding effect of the outer membrane of EcN on the cationic probe. [40] Interestingly, when these bacteria were stained in MOPS minimal medium, only S. epidermidis was strongly stained by OTM, whereas EcN was very weakly stained ( Figures 2B and S10). Quantitative analysis using PL spectrometry also showed that both S. epidermidis and EcN exhibited strong fluorescence in pure water ( Figure 2C), whereas only S. epidermidis emitted strongly in MOPS minimal medium ( Figure 2D). Similarly, S. epidermidis showed higher PL intensity than EcN in phosphate-buffered saline (PBS; Figure S11A) and Hanks' balanced salt solution (HBSS; Figure S11B), which suggested that ions reduced the binding affinity of OTM to the envelope of Gram-negative bacteria. These results were consistent with a previous report that an increase in ion strength could decrease the binding of cationic probes to Gram-negative bacteria. [41] Due to the low background from unbound OTM, it has great advantages over conventional fluorescent dye for quantitative analysis of bacteria. The fluorescence spectra of EcN at different concentrations in water were collected, which exhibited emission peaks at 580 nm ( Figure S12A). There was a linear relationship between the PL intensity of OTM and the EcN concentrations ranging from 5 × 10 6 to 5 × 10 8 CFU/mL and a 23-fold increase at 5 × 10 8 CFU/mL of EcN compared to the blank control ( Figure S12B). Similar linear relationships were obtained for S. epidermidis in pure water ( Figure S12C) and MOPS minimal medium ( Figure S12D), which enabled the high-throughput quantitative analysis of bacteria.
Although OTM-stained EcN fluoresced weakly in MOPS minimal medium, its fluorescence intensity was positively correlated with bacterial concentrations ranging from 2 × 10 7 to 5 × 10 8 CFU/mL ( Figure 3A). Growth curve of EcN could be determined by OTM staining in MOPS minimal medium with the wash-free protocol ( Figure S13). By culturing EcN in MOPS minimal medium supplemented with antibiotics for 5 h and quantifying bacteria amount by OTM, bacterial responses to antibiotics could be evaluated. In the presence of 1-100 µg/mL kanamycin, the fluorescence intensity of OTMstained EcN dropped to <10% ( Figure 3B). Interestingly, low concentrations (1-5 µg/mL) of ampicillin increased the fluorescence intensity of OTM-stained EcN by twofold, whereas high concentrations (>30 µg/mL) of ampicillin decreased the fluorescence to ∼20% ( Figure 3B). These observations suggested that different antibiotics exerted distinguishable concentration-dependent effects on OTM fluorescence.
To investigate the origin of the increased OTM fluorescence at a low concentration (1 µg/mL) of ampicillin, we performed fractionation of EcN cultures. Optical density at 600 nm (OD 600 ) was used to estimate the amount of bacterial cell bodies. Pellet fractions accounted for the presence of bacterial cells and 1 µg/mL of ampicillin reduced cell bodies of EcN to one fifth of the untreated control ( Figure 3C). Both bacterial pellets and culture supernatants demonstrated a threefold increase of OTM fluorescence in the ampicillin-treated group ( Figure 3D). β-Lactams targeting the cell wall of Gram-negative bacteria were shown to increase the production of OMVs through the induction of envelope stress responses. [8,42,43] Thus, we postulated that OTM could effectively detect the ampicillin-induced changes in bacterial membrane and OMV secretion in a fluorescence turn-on manner. To evaluate the sensitivity of OTM, a commercial membrane probe Vybrant DiO, which stained bacterial membranes and OMVs, was employed for comparison. Under the same staining conditions, ampicillin-treated culture products and derived fractions demonstrated remarkable increase of OTM emission, whereas those incubated with DiO did not show any emission enhancement ( Figures 3E and S14A). This could be ascribed to the relatively low sensitivity of DiO in detecting the subtle remodeling of the envelope and vesicle secretion in EcN. Afterward, we collected the PL spectra of unstained, OTM-stained, and DiO-stained samples, which confirmed that ampicillin-treated culture products and derived fractions enhanced the fluorescence of OTM but not DiO ( Figure S14B-D). In confocal imaging, untreated EcN exhibited rod-shaped morphology, whereas ampicillin led to a dramatic increase in OTM fluorescence with enlarged and elongated bacterial bodies ( Figure 3F). Meanwhile, a number of circular structures were observed in the supernatant of EcN cultures supplemented with ampicillin ( Figure 3F), which were postulated to be OMVs or OMV aggregates. In contrast, DiO staining failed to recognize the induction of bacterial bodies and vesicles in confocal imaging ( Figure  S15). To further examine whether OTM was applicable to the quantitative analysis of OMVs, we cultured different densities of EcN ranging from 1 × 10 3 to 5 × 10 6 CFU/mL for 5 h in the presence of 1 µg/mL ampicillin and stained the culture product with OTM. Increased fluorescence intensity of OTM was observed at starting EcN concentrations of above 1 × 10 5 CFU/mL ( Figure S16A). Besides, there was a linear relationship between the fluorescence intensity of OTMstained supernatants and OMV concentration in the range of 10-10 3 -fold dilution of supernatants ( Figure S16B), implying that OTM was a sensitive probe for the rapid detection of OMVs in a single step. Together, our data demonstrated that OTM could effectively detect and quantify ampicillininduced bacterial membrane remodeling and OMV secretion ( Figure 3G). Encouraged by simple operations and the high sensitivity of OTM in OMV studies, we employed it for high-throughput identification of the OMV-induction ability of different antimicrobial materials and determination of their optimal concentration ranges. Untreated EcN cultures did not show a net increase of the OTM fluorescence intensity in the supernatant ( Figure 4A), revealing that EcN did not massively produce OMVs under normal conditions. Bacterial numbers inferred from the OD 600 reading showed that EcN was sensitive to ampicillin at concentrations of above 0.5 µg/mL, and its growth was effectively inhibited by >10 µg/mL of ampicillin ( Figure 4B). Interestingly, the fluorescence of OTM-stained supernatants showed a bell curve in ampicillin concentrations ranging from 0.5 to 10 µg/mL ( Figure 4B). To verify our hypothesis, we isolated OMVs with the traditional filtration method following ultracentrifugation. We analyzed the protein contents of the total cultures and derived OMVs with SDS-PAGE following silver staining or Coomassie Blue staining. As shown in Figure S17A, silver staining revealed OMV induction in 0.75 and 5 but not in 0 or 0.1 µg/mL ampicillin-treated EcN cultures. Moreover, Coomassie Blue staining failed to recognize these inductions ( Figure S17B). These results clearly showed that our fluorescent system possessed great advantages on high sensitivity and high-throughput screening. To demonstrate the advantage of our system over other fluorescent probes, we compared OTM with DiO and a previously published AIE probe (TPE-Bac) [44] for high-throughput OMV screening. As shown in Figure S18, DiO failed to recognize the induction of OMVs, whereas TPE-Bac showed less sensitivity to the induction of OMVs and failed to figure out the most effective concentration of ampicillin for OMV induction. Meanwhile, the presence of 10 µM OTM greatly inhibited the growth of EcN in MOPS minimal medium ( Figure S19), which did not affect the staining of isolated OMVs in the supernatant and minimized the variations during the staining procedure. We then tested kanamycin with concentrations from 0.05 to 1 µg/mL, and no discernable peak was observed from its supernatants ( Figure 4C). Four other antibiotics (aztreonam, ciprofloxacin, chloramphenicol, and tobramycin) and one antimicrobial metal ion (Cu 2+ ) were also tested for determining their ability to induce OMVs. Aztreonam was a time-dependent monobactam antibiotic, and it inhibited the division of Gram-negative bacteria. [45] However, its ability of inducing OMV has not been investigated. Our data revealed that aztreonam in a relatively narrow concentration range of 1-20 ng/mL induced a strong peak of OTM fluorescence in the culture supernatant ( Figure 4D), which corresponded to OMV production. Ciprofloxacin, which was controversial for the clinical application and was reported to induce Shiga toxin-containing OMVs in enterohemorrhagic E. coli O104:H4 and O157:H7 strains, induced a 40% increase of OTM fluorescence in a narrow concen-tration range of 1-3 ng/mL ( Figure 4E), which could be ascribed to OMV production. Meanwhile, chloramphenicol and tobramycin did not induce any significant changes of OTM fluorescence in their subinhibitory concentration ranges ( Figure 4F,G). Beside antibiotics, the antimicrobial Cu 2+ caused about the twofold increase of OTM fluorescence in a relatively wide concentration range of 2.5-30 µM ( Figure 4H), which was consistent with the previous report that Cu 2+ of growth challenging concentration induced bacterial vesicles. [46] Total culture products, which contained the bacterial bodies and vesicles, were also stained with OTM, and showed similar influence on OTM fluorescence ( Figure S20).
The induction of OMVs was confirmed by dynamic light scattering (DLS) and transmission electron microscopy (TEM). For each antimicrobial material, the most effective OMV-inducing concentrations (0.75 µg/mL for ampicillin, 0.4 µg/mL for kanamycin, 10 ng/mL for aztreonam, 2 ng/mL for ciprofloxacin, 0.3 µg/mL for chloramphenicol, 2.5 ng/mL for tobramycin, and 5 µM for Cu 2+ ) were chosen for further experiments. After 5-h incubation, culture products were concentrated, and the supernatants were subjected to DLS measurement and negative staining followed by TEM characterization. In the supernatant of untreated group, the polydispersity index (PDI) was especially high (0.745), and no particles of specific diameter were detected ( Figure 5A). In the supernatant of ampicillin-treated group , particles with an effective diameter of 147.5 nm were detected with an acceptable PDI of 0.254, and round vesicles were observed by TEM ( Figure 5B). In the supernatants of samples treated by kanamycin, tobramycin, and chloramphenicol, the DLS and TEM results were similar to the untreated group that no particles of specific diameter were detected ( Figure 5C-E). However, particles with effective diameter in the 50-300 nm range were detected with acceptable PDI in the supernatants of samples treated with aztreonam (effective diameter of 154.4, PDI of 0.172), ciprofloxacin (effective diameter of 150.6, PDI of 0.187), and Cu 2+ (effective diameter of 134.3, PDI of 0.204) ( Figure 5F-H). Together, our data demonstrated that OTM staining can be a powerful tool for the screening of stimulatory antimicrobial agents that induce OMV secretion.

CONCLUSION
In this study, a novel bacterial AIE probe named OTM was designed and synthesized. It could monitor the subtle remodeling of the bacterial cell envelope and OMV production induced by antibiotics. Thanks to the low background fluorescence and high sensitivity toward OMVs, OTM could be employed to high-throughput screening of OMV-inducing antimicrobial agents in a wide range of concentration. Compared with the reported ultracentrifugation or ultrafiltration for OMV characterization, the current OTM staining enables rapid and high-throughput analysis of OMVs. We envisioned that OTM can be applied to future studies to examine OMV production in pathogenesis or other cellular events. This study may add a novel dimension to evaluate the side effects of research and clinical antimicrobial materials and pave the way for the development of personalized antimicrobial therapeutics in the future.  All chemical reactions were performed under an inert nitrogen atmosphere with the use of a Schlenk line. Glassware was dried in an oven prior to use. Commercially available reagents were used without purification. All the reagents for chemical synthesis were purchased from Tokyo Chemical Industry Co., Ltd (TCI), Sigma-Aldrich, Acros Organics, or J&K Scientific. Dry solvents were stored in the presence of activated 3 Å molecular sieves. All the reactions were monitored by thin-layer chromatography (TLC) with pre-coated aluminum plates (Merck). Intermediates were purified by silica-gel column chromatography (230-400 mesh from Merck or Davisil LC60Å 40-63 µm from W. R. Grace & Co.-Conn). The target compound of OTM was purified by column chromatography performed on aluminum oxide gel (neutral, 200-300 mesh; Macklin).

Instrumentation for chemical characterizations and photophysical measurements
1 H NMR and 13 C NMR spectra were measured in CDCl 3 or CD 3 OD on a Bruker Avance III 400 MHz FT-NMR spectroscopy by employing tetramethylsilane (TMS) as an internal standard for calibrating the chemical shift. Electrospray ionization quadrupole time-of-flight (ESI-Q-TOF) mass spectrometry was performed on the Agilent 6540 Q-TOF LC/MS system. UV-vis absorption spectra were collected on the SpectraMax M2e microplate reader (Molecular Devices) in different solutions at 293 K. The emission spectra of OTM in solutions were measured on the PerkinElmer LS 55 fluorescence spectrometer.

Bacterial cultures
A single-bacterial colony of EcN or S. epidermidis was inoculated into 5 mL LB broth medium and cultured for 12-16 h in a 37 • C incubator at 220 rpm. MOPS minimal medium was used to replace LB broth for the antibiotic screening and the OMV detection assay. For determining the influence of 10 μM OTM on the growth of EcN, 2 × 10 7 CFU/mL of EcN was first inoculated into MOPS minimal medium supplied without (Control) or with 10 μM OTM (OTM) and then incubated at 37 • C, 250 rpm. Three replicates of each group were cultured. After incubated for desired time, 100 μL of each tube were transferred to a well of a standard 96-well plate, of which the OD 600 were measured by a SpectraMax M2e microplate reader.

Bacterial staining
Bacteria grown in LB broth were harvested by centrifugation at 5000 rpm for 2 min at room temperature. The cell pellet was washed twice and resuspended in selected buffer solution (pure water, PBS, HBSS, or MOPS minimal medium  ) with 400-nm (for OTM and TPE-Bac) or 450-nm (for DiO) excitation in three replicates. Autofluorescence was measured from unstained samples and was deducted before data plotting. Generally, higher bacteria concentration will result in large autofluorescence intensity. Thus, for bacteria without antibiotics treatment, the autofluorescence will be larger than those treated with antibiotics, due to the larger quantity of bacteria in the former group. Typically, autofluorescence was about 50% of the total signal in bacteria without antibiotics treatment due to the large autofluorescence and low fluorescence signal. For bacteria treated with antibiotics, bacteria amount was low, and the autofluorescence was <20% of the total signal, as the autofluorescence was low, and the fluorescence signal was large. Spectra of unstained, OTM-stained, or DiO-stained OMVs were determined by a PerkinElmer LS 55 fluorescence spectrometer with 365 and 450-nm excitations. For isolation and purification of OMVs by ultracentrifugation, 50 mL of MOPS minimal medium added with (0.1, 0.75 or 5 μg/mL) or without ampicillin was inoculated with EcN at the density of 5 × 10 6 CFU/mL and were incubated at 37 • C, 250 rpm for 5 h. After that, OD 600 values of the total culture products were determined by a SpectraMax M2e microplate reader (Molecular Devices) with a 1 cm cuvette.
Then, the total culture products were centrifuged at 2800 × g, 4 • C for 10 min. After centrifugation, the supernatant of each group was collected and filtered with a 0.22-μm-pore-size filter, respectively. About 30 mL of each filtered supernatant was subjected to ultracentrifuge at 150,000 × g, 4 • C for 2 h. The resulting pellets were resuspended by 1 mL of MOPS minimal medium as OMVs and stored at −80 • C refrigerator. Protein contents of the total cultures and derived OMVs were extracted with NuPAGE LDS Sample Buffer (4X) and then analyzed with standard SDS-PAGE following Coomassie Blue staining or silver staining.

Measurement of the size of OMV solutions
OMV samples were prepared and stored in 4 • C fridge overnight. Sizes of OMV samples were measured with a Malvern Zetasizer.

4.9
Transmission electronic microscopy OMV samples were prepared and stored in a 4 • C fridge overnight. The next day, 5 μL OMV samples were absorbed by a carbon-coated copper grid and negatively stained with 1% uranyl acetate. All samples were observed by a CM100 Transmission Electron Microscope (Philips).

Statistical analysis
Quantitative analysis data are presented as mean ± standard deviation. All data are plotted with Origin lab software (USA). For the quantification of EcN with OTM in H 2 O, PL spectra of three replicates were measured for each sample. For the quantification of other bacterial samples with OTM, samples were aliquoted into a 96-well black-walled glassbottom plate in four replicates, which were measured by a SpectraMax M2e microplate reader. For the quantification of bacteria with OD 600 in the antibiotics screening, samples were aliquoted into a 96-well plate in four replicates, which were measured by a SpectraMax M2e microplate reader. For the quantification of bacteria and derived OMV with OTM in the antibiotics screening, samples were aliquoted into a 96-well black-walled glass-bottom plate in three replicates, which were measured by a SpectraMax M2e microplate reader. In addition, at the same time, one unstained aliquot of each sample was measured for autofluorescence. DLS data are representative data of four independent experiments.

C O N F L I C T S O F I N T E R E S T
The authors declare no conflict of interest.

D ATA AVA I L A B I L I T Y S TAT E M E N T
The data that supports the findings of this study are available in the supplementary material of this article.