Detection of Gram-negative bacterial outer membrane vesicles using DNA aptamers

Infection of various pathogenic bacteria causes severe illness to human beings. Despite the research advances, current identification tools still exhibit limitations in detecting Gram-negative bacteria with high accuracy. In this study, we isolated single-stranded DNA aptamers against multiple Gram-negative bacterial species using Toggle-cell-SELEX (systemic evolution of ligands by exponential enrichment) and constructed an aptamer-based detection tool towards bacterial secretory cargo released from outer membranes of Gram-negative bacteria. Three Gram-negative bacteria, Escherichia coli DH5α, E. coli K12, and Serratia marcescens, were sequentially incubated with the pool of random DNA sequences at each SELEX loop. Two aptamers selected, GN6 and GN12, were 4.2-times and 3.6-times higher binding to 108 cells of Gram-negative bacteria than to Gram-positive bacteria tested, respectively. Using GN6 aptamer, we constructed an Enzyme-linked aptamer assay (ELAA) to detect bacterial outer membrane vesicles (OMVs) of Gram-negative bacteria, which contain several outer membrane proteins with potent immunostimulatory effects. The GN6-ELAA showed high sensitivity to detect as low as 25 ng/mL bacterial OMVs. Aptamers developed in this study show a great potential to facilitate medical diagnosis and early detection of bacterial terrorism, based on the ability to detect bacterial OMVs of multiple Gram-negative bacteria.

Binding affinity of isolated aptamers. Next, we quantified binding affinities of GN6 and GN12 aptamers against three Gram-negative bacteria used in SELEX. Various concentrations of aptamers labelled with FAM (carboxyfluorescein) at the 3′-end were added to 10 8 cells of bacteria. After measuring the fluorescence signal, the binding saturation curves were fitted using non-linear regression model, based on the following equation: S = B max × C/(K d + C) (where S represents the fluorescence (FAM) intensity, B max , the maximum binding intensity, K d , the dissociation constant, and C, concentrations of aptamer) (Fig. 2a-c). The dissociation constants (K d ) were observed in the range of 20.36 to 59.70 nM against three bacterial strains tested (Table 2). These values were in similar range with previous experimental results in other studies 26 . The binding affinities of two aptamers were compared with that of random ssDNA (N40) at 250 nM concentration in 10 8   bacteria, respectively (Fig. 2d,e). Both GN6 and GN12 aptamers showed average 2.55-and 3.19-times higher binding towards 10 8 cells of bacteria than N40 random sequence. Likewise, it showed average 2.89-and 3.03-times higher binding to 10 5 cells of bacteria than the same control.
Aptamer cross-reactivity towards multiple Gram-negative bacteria. Next, the broad cross-reactivity of selected aptamers was analysed by performing binding assay against multiple Gram-negative bacteria, based on the assumption that they share similar structural components as common cell surface epitopes. Both GN6 and GN12 (250 nM) showed high binding to 10 8 cells of various Gram-negative bacteria tested, including the well-known pathogenic species, K. pneumoniae, E. cloacae and S. sonnei (Fig. 3a). It should be noted that aptamers showed lower binding to one Gram-negative bacteria tested, S. trueperi, which is a Sphingomonas species that belongs to α-proteobacteria 35 . In contrast, significantly reduced binding efficiency was observed against all Gram-positive bacterial strains tested. GN6 aptamer could detect Gram-negative bacteria 4.2-times higher than Gram-positive bacteria (p < 0.0001). GN12 aptamer also showed 3.6-times higher binding to Gram-negative bacteria than Gram-positive bacteria (p < 0.0005) (Fig. 3b). These broad cross-reactivity and specificity to Gram-negative bacteria were also observed when 10 5 cells of bacteria were incubated (Fig. S1a). Both GN6 and GN12 aptamer at 250 nM concentration were able to detect Gram-negative bacteria 3.9-times and 3.4-times higher than Gram-positive bacteria, respectively (p < 0.0001) (Fig. S1b). Without negative selections against Gram-positive bacteria during Toggle-cell SELEX, selected aptamers showed no binding preference to them, suggesting that the unknown targets of aptamers on cell surface exclusively expressed in Gram-negative bacteria.  Table 1. ssDNA sequences of two isolated aptamers after Gram-negative bacterial Toggle-cell SELEX.  www.nature.com/scientificreports www.nature.com/scientificreports/ Future studies are required to identify the specific targets of GN6 aptamers on OMVs surface. Furthermore, we also noticed inconsistency between dissociation constants and binding profiles of both aptamers regarding their binding abilities to Gram-negative bacteria. Higher binding affinity of aptamers with low K d values generally represents the higher intermolecular interaction between a single target on bacteria and its bound aptamer. In contrast, Fig. 3 showed the maximum binding capacity (B max ) when all targets are fully saturated by excessive amounts of aptamers (250 nM). These two binding parameters, affinity and capacity, can be separately interpreted because capacity is affected by several factors such as multivalent interactions when an aptamer shares multiple antigens of bacteria or the density of targets on bacterial outer membrane 29,36 . Isolation and characterizations of bacterial OMVs. It has been known that OMVs budding out from outer membranes of Gram-negative bacteria carry several virulence biomolecules and endotoxins 7 . We isolated bacterial OMVs using ultracentrifugation. DLS analysis exhibited the size distribution of OMVs ranging from 84.29 to 176.4 nm (Table 3), within the range of general agreement 4,5 . There are no general bacterial OMV markers, but OMV proteins such as OmpA (35.2 kDa) in E. coli 37 and Serralysin (~52-55 kDa) in S. marcescens 38 could be used for characterization (Fig. S2). Next, magnetic bead-pull down assay using streptavidin-coated beads and 3′-biotinylated GN6 aptamer was performed to capture OMVs 39 (Fig. 4a). The binding between E. coli DH5α OMVs and GN6 was visualized by scanning electron microscope (SEM) (Fig. 4b). Without GN6 aptamer, spherical nanoparticles of E. coli DH5α OMVs were not shown. It suggests that the targets of GN6 on E. coli DH5α are   eLAA platform for detecting Gram-negative bacterial oMVs. Using the selected aptamer, GN6, we have developed an Enzyme-linked aptamer assay (ELAA) to detect OMVs originated from various Gram-negative bacteria. Instead of conventional ELISA using the detection antibody conjugated with HRP (Horse radish peroxidase), GN6 aptamer was used as a bait (Fig. 5a). GN6 ELAA showed highly recognizable signals against three Gram-negative bacterial OMVs. The dissociation constants of GN6 to OMVs derived from E. coli DH5α, E. coli K12 and S. marcescens were 0.13 ± 0.01 μg/ml, 3.70 ± 0.98 μg/ml and 0.23 ± 0.16 μg/ml, respectively (R 2 = 0.99) (Fig. 5b). It especially showed the highest binding affinity and capacity to E. coli DH5α OMVs. Meanwhile, this assay also showed high sensitivity, which could detect as low as 25 ng/mL of Gram-negative bacterial OMVs. Consistent with GN6 binding affinity to bacterial cells, it showed the similar pattern to binding affinity to bacterial OMVs, strongly suggesting that target of GN6 aptamers is present both on Gram-negative bacterial cells and on the surface of Gram-negative bacterial OMVs. Consistent with the previous results, OMVs from Gram-positive bacteria, L. grayi and B. megaterium, showed no recognizable binding to GN6 (Fig. 5c,d). These results indicate that GN6 ELAA could detect multiple Gram-negative bacteria derived OMVs. It opens the new possibility of developing cell-free bacterial sensor using bacterial OMVs as substrates instead of living bacterial cells.

conclusion
While various systems are being developed to detect pathogenic bacteria, few studies have reported the isolation of broadly cross-reactive aptamers against various species of bacteria. Here, we developed highly specific DNA aptamers, GN6 and GN12, against many Gram-negative bacteria, including pathogenic strains. Selected aptamers after Toggle-cell SELEX showed broad cross-reactivity towards many Gram-negative bacteria tested. Using GN6 aptamer, we developed an GN6-ELAA to detect Gram-negative bacterial OMVs from cell-free supernatant. This is because unknown targets of GN6 on the original bacterial outer membrane could also be expressed in the surface of bacterial OMVs. Moreover, the GN6-ELAA had high sensitivity to low concentration of Gram-negative bacterial OMVs and high specificity exclusively bound to them. Further studies will require identification of GN6 aptamer targets and increasing the yield and purity of OMVs. If we increase the final yield and purity of OMVs, it will be possible that the detection of bacterial OMVs in cell-free medium leads to the accurate identification of the originated bacteria. We believe that the aptamer-based Gram-negative bacterial OMV detection has a great potential to facilitate medical diagnosis and early detection of bacterial terrorism.  www.nature.com/scientificreports www.nature.com/scientificreports/ collected at 6,000 rpm for 5 min at 4 °C. The bound ssDNA was separated from bacterial cells by elution in Tris-EDTA buffer (50 mM Tris, pH 8.0, 10 mM EDTA, 1 mM MgCl 2 , 5 mM KCl, 100 mM NaCl) at 95 °C for 5 min and cooled on ice for 5 min. The ssDNA after elution was collected at 13,000 rpm, 4 °C for 5 min, purified using PCI solution (Sigma Aldrich, USA), and precipitated in ethanol including 5 v/v% ammonium acetate and 1.5 v/v% glycogen. The PCR mixtures were made by ssDNA template (~20 ng), 0.4 μM of forward and poly-A tailed reverse primer, MyTaq Reaction buffer, and 0.25 U My Taq DNA polymerase (Bioline, UK). After PCR, the reaction products were separated by 10% PAGE gel in 1X TBE (Tris-borate-EDTA). To separate the interested ssDNA, gel purifications in UREA-PAGE gel were performed using asymmetric poly-A tailed reverse primer. The dsDNA PCR product separated by forward primer in UREA gel was separated, heated at 65 °C for 30 min in elution buffer (10 mM Tris, pH 7.4, 1 mM EDTA, 750 mM NH 4 OAc, 0.1% SDS) and collected in 0.3 M sodium acetate and 0.25 μg/μL glycogen (pH 5.2). Subsequently, 2.5 volumes of 100% ethanol added and incubated at −80 °C for 12 h, followed by centrifugation at 13,000 rpm, 4 °C for 30 min. The DNA pellets were dissolved in water and quantified using BioSpecNano Spectrophotometer (SHIMADZU, Japan). For the next round of selection, 100 pmol of ssDNA from the first round was mixed with 10 8 cells of E. coli K12. The following procedure was the same as above. Subsequently, ssDNA was isolated against S. marcescens and the same procedures were repeated four times (total 12 rounds). After the final round, the isolated ssDNA pools were amplified by RBC T&A cloning vector kit (Real Biotech, Taiwan). ssDNA aptamer candidates were transformed into competent E. coli DH5α and then plasmid DNA were purified using QIAprep Spin Miniprep Kit (Qiagen Inc., Germany). The secondary structures were computationally predicted using M-fold algorithm (http://mfold.rit.albany.edu) at RT, [Na + ] = 100 mM, and [Mg 2+ ] = 5 mM.

Methods
Binding enrichment test using quantitative real-time PCR. The standard controls were made by serial dilution of samples. Each ssDNA sample was mixed with 0.2 μM of forward, reverse 18-nt primers and SYBR ® Premix Ex Taq TM (TliRNaseH Plus, Takara Bio Inc, Japan), followed by relative quantification analysis using StepOne TM Real-Time PCR System (Applied Biosystems ® , USA) according to the manufacturer's protocol.
fluorescence-based binding assays for aptamers. The binding affinity and capacity to bacteria were quantified by binding 3′-FAM labelled aptamers to the bacterial cells. For measuring the dissociation constants, 10 8 cells of bacterial species were bound to different aptamer concentrations at RT for 15 min. To determine whether the aptamers showed non-specific binding to bacteria, the same concentrations of the negative controls as 3′-FAM labeled N40 library were incubated in the same bacteria used above. After incubation, the samples were washed three times in Tris buffer to remove unbound ssDNA, at 10,000 rpm, 4 °C for 10 min. Samples were resuspended in water, and their fluorescence intensity was measured using VICTOR X2 Multilabel Plate Reader (PerkinElmer, USA). The dissociation constant was measured by a non-linear regression fit model of SigmaPlot12.0. The binding efficiency or capacity of the two selected aptamers at 250 nM concentration was measured by incubating 10 8 and 10 5 cells of bacteria. It was also compared with that of 3′-FAM-labeled N40 random ssDNA upon incubation with 10 8 and 10 5 cells of bacteria. Isolation and characterizations of OMVs. Bacterial cultures grown overnight in media were pelleted at 10,000 rpm for 30 min. The supernatant fraction was filtered through a 0.45 μm syringe (Merck Millipore, USA) to remove any remaining cell debris, and concentrated 50-fold by ultrafiltration using 100 kDa Amicon ® Ultra-0.5 device (Merck Millipore). One more filtration was performed using 0.22 μm syringe filter (Merck Millipore). Next, OMVs were isolated by ultracentrifugation (Optima MAX-XP, Beckman Coulter, Inc., USA) at 150,000 rpm for 3 h at 4 °C, resuspended in PBS and stored at −80 °C. The protein concentrations were measured using micro BCA assay (Thermo Scientific, USA). To estimate the size distributions of the isolated OMVs, dynamic light scattering (DLS) was carried out using Zetasizer Nano ZS90 (Malvern, UK). Samples were diluted 1:1000 in PBS and processed at 25 °C under standard settings (Dispersant Refractive Index = 1.331, viscosity (cP) = 0.89). To visualize the binding between GN6 and E. coli DH5α OMVs, 200 μg of streptavidin-coated magnetic beads (Dynabeads ™ M-280 Streptavidin, Thermo Scientific, USA) and 50 pmol of GN6 aptamer were mixed for 30 min in mild shaking. After washing once, 10 μg/mL of OMVs were added and incubated for 15 min, followed by washing several times to remove unbound OMVs. These samples were fixed in a 2% paraformaldehyde solution for 2 h and diluted in distilled water, followed by immobilization on the clean silicon chips under drying conditions. To make surface conductive, Au-Pd alloy was applied by sputtering before imaging. SEM using JSM-7100F was performed in 2 or 5 kV of beam energy.
Aptamer-based direct oMVs detection. For GN6 ELAA using bacterial OMVs, Nunc-Immuno 96 MicroWell solid plates (Thermo Scientific, USA) were used to immobilize bacterial OMVs in Tris buffer. After incubating OMVs at various concentrations in the plate overnight at 4 °C, the plate was washed twice and blocked using 2% BSA-Tris buffer for 2 h. After blocking, 20 pmol of GN6 aptamer and N40 control were separately added and incubated for 1 h. After washing 4 times, streptavidin-Poly HRP conjugate (Pierce, USA) was added and incubated for 30 min. After thoroughly washing 5 times in Tris buffer with 0.05% Tween-20, Ultra TMB-ELISA reagent (Thermo Scientific, USA) was added. After 15 min, 1 M sulfuric acid as stop solution was added. Absorbance at 450 nm was measured using Multiskan microplate photometer (Thermo Scientific, USA). The measured values were analyzed using non-linear regression fit model of SigmaPlot 12.0.