Scalable Isolation and Purification of Extracellular Vesicles from Escherichia coli and Other Bacteria

Summary

Bacteria secrete nanometer-sized extracellular vesicles (EVs) carrying bioactive biological molecules. EV research focuses on understanding their biogenesis, role in microbe-microbe and host-microbe interactions and disease, as well as their potential therapeutic applications. A workflow for scalable isolation of EVs from various bacteria is presented to facilitate standardization of EV research.

Abstract

Diverse bacterial species secrete ~20-300 nm extracellular vesicles (EVs), comprised of lipids, proteins, nucleic acids, glycans, and other molecules derived from the parental cells. EVs function as intra- and inter-species communication vectors while also contributing to the interaction between bacteria and host organisms in the context of infection and colonization. Given the multitude of functions attributed to EVs in health and disease, there is a growing interest in isolating EVs for in vitro and in vivo studies. It was hypothesized that the separation of EVs based on physical properties, namely size, would facilitate the isolation of vesicles from diverse bacterial cultures.

The isolation workflow consists of centrifugation, filtration, ultrafiltration, and size-exclusion chromatography (SEC) for the isolation of EVs from bacterial cultures. A pump-driven tangential flow filtration (TFF) step was incorporated to enhance scalability, enabling the isolation of material from liters of starting cell culture. Escherichia coli was used as a model system expressing EV-associated nanoluciferase and non-EV-associated mCherry as reporter proteins. The nanoluciferase was targeted to the EVs by fusing its N-terminus with cytolysin A. Early chromatography fractions containing 20-100 nm EVs with associated cytolysin A - nanoLuc were distinct from the later fractions containing the free proteins. The presence of EV-associated nanoluciferase was confirmed by immunogold labeling and transmission electron microscopy. This EV isolation workflow is applicable to other human gut-associated gram-negative and gram-positive bacterial species. In conclusion, combining centrifugation, filtration, ultrafiltration/TFF, and SEC enables scalable isolation of EVs from diverse bacterial species. Employing a standardized isolation workflow will facilitate comparative studies of microbial EVs across species.

Introduction

Extracellular vesicles (EVs) are nanometer-sized, liposome-like structures comprised of lipids, proteins, glycans, and nucleic acids, secreted by both prokaryotic and eukaryotic cells¹. Since the early studies visualizing the release of EVs from gram-negative bacteria², the number of biological functions attributed to bacterial EVs (20-300 nm in diameter) has constantly been growing in the past decades. Their functions include transferring antibiotic resistance³, biofilm formation⁴, quorum sensing⁵, and toxin delivery⁶. There is also growing interest in the use of bacterial EVs as therapeutics, especially in vaccinology⁷ and cancer therapy⁸.

Despite the growing interest in EV research, there are still technical challenges regarding methods of isolation. Specifically, there is a need for isolation methods that are reproducible, scalable, and compatible with diverse EV-producing organisms. To create a unified set of principles for planning and reporting EV isolation and research methods, the International Society for Extracellular Vesicles publishes and updates the MISEV position paper⁹. Moreover, the EV-TRACK consortium provides an open platform for reporting detailed methodologies for EV isolation used in published manuscripts to enhance transparency¹⁰.

In this protocol, previous methodologies used for the isolation of EVs from mammalian cell culture were adapted^11,12 to enable the isolation of EVs from bacterial cell culture. We sought to employ methods that enable EV isolation from a variety of microbes, which can be scalable, and balance EV purity and yield (as discussed in the MISEV position paper⁹). After removing bacterial cells and debris by centrifugation and filtration, the culture medium is concentrated either by centrifugal device ultrafiltration (for a volume of up to ~100 mL) or pump-driven TFF (for larger volumes). EVs are then isolated by SEC using columns optimized for the purification of small EVs.

Figure 1: Bacterial EV isolation workflow schematic overview. Abbreviations: EV = extracellular vesicle; TFF = tangential flow filtration; SEC = size exclusion chromatography; MWCO = molecular weight cut-off. Please click here to view a larger version of this figure.

A mouse-commensal strain of Escherichia coli (i.e., E. coli MP1¹³) was used as a model organism and modified to express EV-associated nanoluciferase by fusion to cytolysin A, as previously reported¹⁴. The methods used here can process at least up to several liters of bacterial cultures and effectively separate EV-associated from non-EV-associated proteins. Finally, this method can also be used for other gram-positive and gram-negative bacterial species. All relevant data of the reported experiments were submitted to the EV-TRACK knowledgebase (EV-TRACK ID: EV210211)¹⁰.

Protocol

NOTE: Ensure that all work involving bacteria and recombinant DNA follows best practices for biosafety containment appropriate for the biosafety hazard level of each strain. Work should be done in accordance with local, national, and international biosafety regulations.

1. Bacterial strains and culturing conditions

NOTE: Bacterial strains used in this study were Escherichia coli MP1¹³, Akkermansia mucinophila, Bacteroides thetaiotaomicron, Bifidobacterium breve, and Bifidobacterium dentium.

1. For E. coli, use a sterile loop to inoculate single colonies into 250 to 1,000 mL of Luria-Bertani (LB) broth and incubate aerobically in a shaking incubator at 300 rpm and 37 °C for 48 h before processing the culture. For recombinant E. coli MP1 strain harboring p114-mCherry-Clyluc (Supplemental Method and Supplemental Figure S1), add chloramphenicol to the LB agar and broth at a final concentration of 17 µg/mL.
2. For A. mucinophila, B. thetaiotaomicron, B. breve, and B. dentium, streak on Brain heart infusion (BHI) agar plates and incubate anaerobically inside a vinyl anaerobic chamber. Inoculate single colonies into 100 mL of pre-reduced BHI broth and incubate for 48 h anaerobically.

2. EV isolation

1. Clarifying bacterial culture medium by centrifugation and filtration
   1. Transfer the bacterial cell cultures inoculated in step 1 to clean 250 mL or 500 mL polypropylene centrifuge bottles by pouring. Centrifuge the bottles in a large-capacity, fixed-angle rotor at 4 °C and 5,000 × g for 15 min. Transfer the supernatant to clean centrifuge bottles by careful pouring, and centrifuge again at 10,000 × g for 15 min.
      ​NOTE: Reuse the bottles after biosafety-appropriate cleaning and decontamination.
      1. If large pellets of bacterial cells are present after the second centrifugation, repeat the centrifugation in a clean bottle to further remove cells.
   2. Transfer the supernatant to a 0.22 µm polyethersulfone vacuum-driven filter device of appropriate size by pouring. Filter by connecting the filtration device to a vacuum wall supply. If the filtration rate drops significantly, simply move any unfiltered material to a new device. Store the filtered medium at 4 °C overnight, and continue the protocol the following day if desired.
      NOTE: The centrifugations above typically allow processing of ~2x the indicated volume of cell culture through each device. For example, a single 500 mL filter device could filter ~1,000 mL of pre-centrifuged culture. These devices are not typically reused. Using syringe filters at this step is not recommended without optimization, as significant losses were noted with the tested models. This is a potential stopping point.
   3. Check for the complete removal of the viable cells at this point by spreading an aliquot of the filtered supernatant on suitable agar plates and ensure the absence of any colonies after incubation at optimum conditions for the bacterial strain. If bacteria are detected, further optimize the procedure above by performing additional centrifugations and/or filtrations.
2. Concentration of the filtered medium
   1. If working with volumes significantly >100 mL, proceed to step 2.2.2. If working with volumes of ~100 mL, load 90 mL of filtered culture medium onto the reservoir of a respective capacity 100 kDa molecular weight cutoff (MWCO) centrifugal ultrafiltration device using serological pipettes. Always balance with a matching ultrafiltration device, and centrifuge in swinging bucket rotor at 4 °C and 2,000 × g for 15-30 min intervals, until the volume of the medium in the top reservoir has been concentrated to <0.5 mL.
      1. Top up the reservoir with any remaining filtered culture medium. If "topping up," remove the flow-through in the bottom of the device and re-balance any devices.
         NOTE: It was observed that the maximum volume of filtered culture medium that can be concentrated using these devices is <2-fold the recommended volume.
      2. If the viscosity of the concentrated medium in the reservoir is visibly increased (dark, viscous material), dilute with phosphate-buffered saline (PBS) and re-concentrate by centrifugation to dilute any non-EV proteins smaller than the MWCO of 100 kDa.
         NOTE: This is a potential stopping point.
      3. Transfer the concentrated medium to a low-protein-binding tube, store at 4 °C overnight, and continue the protocol the following day if desired.
   2. If working with volumes significantly >100 mL, select an appropriately sized TFF device (100 kDa MWCO) to accommodate the volume to be processed.
      NOTE: Filtration devices for processing 100 mL to >1,000 mL are commercially available. Local availability, cost, and compatibility with the pump and tubing/connections will dictate which particular models will be most useful. Up to 2 L of culture medium were processed with the device indicated in the Table of Materials before needing to clean the filter (see step 2.3 below for the cleaning protocol).
      1. Assemble a filtration circuit with #16 low-binding/low-leaching tubing, 1/8 inch hose-barb to Luer adapters, the TFF device, and a peristaltic pump, as indicated in Supplemental Figure S2.
         NOTE: Perform TFF within a biosafety cabinet to minimize the risk of contaminating the EV preparation with environmental bacteria.
      2. At room temperature, begin circulating the filtered, conditioned medium at approximately 200 mL/min (minimum 100 mL/min). Determine the appropriate RPM corresponding to the desired flow rate by pumping 200 mL of PBS into a graduated vessel. When circulating filtered, conditioned medium, collect the molecules <100 kDa crossing the ultrafiltration membrane as waste in a separate vessel.
         NOTE: The example below will be assuming a starting volume of 2 L of culture.
      3. Continue to circulate the conditioned medium until its volume has been reduced to ~ 100-200 mL. Move to smaller vessels as needed. Dilute 2-fold with PBS, and continue to circulate with the pump, concentrating down to 75-100 mL. Dilute 2-fold with PBS, and continue to circulate to a final volume of 25 mL. Dilute 2-fold with PBS and continue to circulate until <10 mL.
      4. Lift the feed tubing out of the sample reservoir, and continue to pump to purge the filter and recover the maximum amount of sample.
         NOTE: This is a potential stopping point.
      5. Transfer the concentrated sample to a conical tube and store overnight at 4 °C if desired. Alternatively, continue with the protocol.
      6. Move the concentrated sample to a 15 mL capacity 100 kDa MWCO centrifugal ultrafiltration device. Centrifuge in a swinging bucket rotor at 4 °C and 2,000 × g for 15-30 min intervals until the volume of the medium in the top reservoir has been concentrated to <2 mL.
         NOTE: This is a potential stopping point.
      7. Transfer the concentrated medium to a low-protein-binding tube, and store at 4 °C overnight, continuing the protocol the following day if desired.
3. Cleaning the TFF device (optional)
   NOTE: The filtration rate decreases as the TFF device begins to "clog" during the process (fouling). If necessary, the filter device can be cleaned to facilitate filtration of additional samples in the same purification run. Though theoretically possible, a cleaned TFF filter has not been used for a different purification run to avoid cross-contamination.
   1. To clean, remove all tubing and caps from the TFF device and drain any residual liquid.
   2. Use the peristaltic pump and tubing to flood both the inner and outer compartments of the TFF device (i.e., via the parallel and perpendicular ports in the model listed in the Table of Materials) with ~100 mL of distilled water. Remove all tubing/caps and drain the TFF device.
   3. Cap the outer (perpendicular, filtrate) ports and circulate 250 mL of 20% ethanol in distilled water at >200 mL/min for 10 min through the inner compartment. Drain, flood with distilled water, and drain again as above.
   4. Circulate 250 mL of 0.5 N fresh NaOH solution for 30 min through the inner compartment and drain again.
   5. Reconnect all tubing and caps to the inlet, outlet, and filtrate ports, as in Supplemental Figure S2, and circulate 0.5 N NaOH solution again until a volume of NaOH > 1 mL/cm² filter surface area permeates through the filter membrane and is collected as filtrate/waste.
   6. Rinse the TFF device with distilled water as above. Use the TFF device immediately or flood the device with ~100 mL of 20% ethanol and store overnight at 4 °C.
      ​NOTE: If stored in ethanol, be sure to drain, rinse with water, drain, and circulate 250 mL of PBS through the device until a volume of >1 mL/cm² filter surface area permeates through the filter membrane and is collected as filtrate/waste to remove residual ethanol prior to sample processing.
4. Size exclusion chromatography (SEC)
   NOTE: SEC is used to increase the purity of EVs and remove non-vesicular protein.
   1. Use a small SEC column (10 mL bed volume) for the isolation of EVs from <100 mL of starting material and a larger column (47 mL bed volume) for the isolation of EVs from >100 mL of starting material.
      NOTE: The example below will list volumes for the larger column, with volumes for the smaller column in parentheses.
   2. Bring the SEC column and PBS to room temperature over several hours. Stabilize the SEC column in a vertical position using a standard laboratory stand and holder. Alternatively, use commercial chromatography column stands.
   3. Before connecting to the SEC column, hydrate the sample reservoir by allowing 5 mL of PBS to flow through the frit and into a waste container. Unscrew the inlet cap of the SEC column, add 2 mL of PBS to the sample reservoir, and carefully connect the reservoir to the column as the PBS is dripping out through the frit (not applicable for small SEC columns).
      NOTE: This previous step prevents any air bubbles from getting trapped at the top of the SEC column. If air is trapped, remove the reservoir, tap the column to get the air bubble out, and repeat the connection procedure. For the smaller column, simply uncap the top of the SEC column, and attach the sample hopper.
   4. Add 47 mL (10 mL) of PBS to the sample reservoir and uncap the bottom of the SEC column. Allow all the loaded sample buffer to flow through the column for equilibration. Discard the flow-through.
   5. Load a maximum of 2 mL (0.5 mL) of sample onto the sample reservoir, discard the flow-through, and allow the sample to enter the column completely.
   6. Immediately add PBS to the sample reservoir or hopper at a volume of 14.25 mL minus the sample volume (3 mL minus the sample volume, for the small column). Allow the solution to flow through the column and discard this amount equal to the column void volume.
      NOTE: For a typical 2 mL sample, the amount of PBS to be added to the sample reservoir or hopper will be 12.25 mL.
   7. Position a 2 mL low-binding microtube directly below the SEC column. Immediately add 2 mL (0.5 mL) of PBS to the sample reservoir and allow it to enter the column. Label this first 2 mL (0.5 mL) of flow-through as Fraction 1. Continue to add 2 mL (0.5 mL) at a time to the sample reservoir to collect each subsequent fraction.
      NOTE: Most bacterial EVs elute in the first 5 fractions. During optimization, the first 12 fractions were collected.
   8. Store the fractions at 4 °C for short-term storage (days) or -80 °C for long-term storage.
   9. Cleaning and storage of the reusable SEC columns
      NOTE: The SEC columns described in this protocol can be reused up to 5 times according to the manufacturer. If the flow rate of the SEC columns decreases after <5 uses, the manufacturer recommends centrifuging the concentrated samples at 10,000 x g for 10 min to clear any aggregates before SEC. Then load the supernatant of this centrifugation  onto the SEC column for EV isolation. 
      1. To clean and store SEC column after each use, add 2 mL (0.5 mL) of 0.5 M NaOH and allow it to enter the column completely. Run 100 mL (20 mL) of 20% ethanol through the column and store it at 4 °C until the next use. Before the next use, equilibrate the ethanol to room temperature as above, and exchange it with PBS buffer by running another 150 mL (30 mL) of PBS through the column.
      2. To clean and immediately re-use SEC column after each use, add 2 mL (0.5 mL) of 0.5 M NaOH and allow it to enter the column completely. Run approximately 150 mL (30 mL) of PBS buffer to wash away NaOH. When pH of eluate is equal to PBS (~7), a new sample may be loaded.

3. EV preparation quality control

1. Sterility testing
   NOTE: As these EVs come from bacterial cultures, it is critical to ensure sterility prior to downstream use.
   1. Obtain 100 µL (20 µL) of the fractions to be used in assays and inoculate 3 mL of the medium used to grow the source bacteria. Culture under the respective optimal conditions for at least 3 days and observe for turbidity. Alternatively, apply the fraction samples to agar plates containing the medium used to grow the producing bacteria and look for colony formation.
      NOTE: If bacterial contamination is detected, it is not recommended to use the EV preparation for experimentation. Instead, repeat the isolation, taking care to minimize the risk of bacterial contamination by (a) performing sufficient centrifugation/filtration of conditioned bacterial cell culture medium, (b) using clean bottles, tubing, filters, and chromatography columns, and (c) employing appropriate aseptic techniques.
2. Protein quantification
   NOTE: A high-sensitivity, fluorescence-based protein quantification kit was used (see the Table of Materials). The kit works with a matching proprietary fluorimeter at excitation/emission wavelengths of 485/590 nm.
   1. Bring all reagents, standards, and samples to room temperature.
   2. Prepare a master mix of protein reagent and buffer by adding 1 µL of the reagent to 199 µL of buffer for each sample and standard to be assayed. Using thin-walled 0.5 mL PCR tubes, add 10 µL standard + 190 µL of master mix to each standard tube.
      NOTE: To be within the range of the assay, the amount of each fraction to be added to each sample tube depends on the expected protein yield of the purification. Typically, 5 µL of each fraction + 195 µL of master mix were used. The final volume of sample + master mix must be 200 µL.
   3. Vortex the assay tubes, and incubate for at least 15 min at room temperature in the dark.
   4. Measure the standards on the appropriate proprietary fluorimeter (see the Table of Materials) by selecting the Protein assay option using the arrow buttons and pressing the GO button to confirm. Follow the on-screen instructions, inserting each standard tube and pressing GO.
   5. Insert the experimental sample tube; press GO to read; and note the result displayed, which is the actual protein concentration in the assay buffer/sample mixture. To obtain the protein concentration in the sample, use the arrow keys to select the Calculate sample concentration option, press GO, and use the arrow keys to select the sample volume added to the assay buffer for the given sample. Press GO and record the sample protein concentration. Repeat this step for each sample to be analyzed.
3. Particle counting and size distribution
   NOTE: Microfluidics resistive pulse sensing (MRPS) was used to quantify EV concentration and size distribution.
   1. Dilute the samples in PBS supplemented with 1% Tween-20 that has been filtered through a 0.02 µm syringe filter to a protein concentration of approximately 0.1 µg/mL.
      NOTE: The goal of dilution is to reach an expected particle concentration in the range of 10¹⁰ particles/mL in EV-containing fractions. The optimal dilution may need to be determined empirically. Few EVs are expected for later fractions (beyond Fraction 6). Thus, the particle concentration will likely be <10¹⁰ particles/mL despite analyzing at low dilutions.
   2. Load 3 µL of each sample into the disposable microfluidics cartridge with a micropipette, insert the cartridge into the MRPS instrument, and push the metal button with a blue illuminated rim.
   3. Click Go! on the acquisition software and wait for the sample to be analyzed by the instrument. Acquire 1,000 to 10,000 particle events to minimize the technical statistical error of analysis. At this point, click Stop and End Run to complete the sample acquisition.
      NOTE: Together with the raw data files, the instrument outputs a summary spreadsheet listing the particle concentration in the sample. Correct this value according to the sample dilution made.
   4. Using analysis software, load the raw data and generate customized graphs of size distribution.

4. EV storage

1. Aliquot individual or pooled fractions to 25-50% of the individual fraction size (depending on the size of column used) in low-protein-binding tubes and store at -80 °C to avoid freeze-thaw cycles.
   ​NOTE: Different applications may require smaller or larger aliquots depending on the expected amount utilized in each experiment. This will need to be determined empirically. The non-EV-containing fractions can be discarded if not applicable to the research objectives.

5. Transmission electron microscopy

1. Negative staining
   1. Add 5 µL of the EV sample to the carbon-coated copper 400 mesh grid and incubate at room temperature for 10 min. Wash the specimen side with 5 drops of 5 mM Tris buffer (pH 7.1) and then with 5 drops of distilled water.
   2. Stain specimen side with 5 drops of 2% uranyl acetate. Blot away any extra amount of stain with filter paper, and allow the grid to dry completely for several hours or overnight. Visualize the specimens with an electron microscope operated at 80 kV.
2. Immunogold labeling
   1. Apply 10 µL of the EV suspension to a formvar/carbon 400 mesh grid and incubate at room temperature for 1 h. Wash the grid in PBS three times, and then apply 4% paraformaldehyde for 10 min to fix the sample. Wash the grids five times with PBS.
   2. Block the grid with three washes of PBS containing 0.1% bovine serum albumin (BSA). Then, apply 10 µL of a primary antibody for 40 min at room temperature (here, 1 µg/mL of nluc antibody). Wash three times again with PBS containing 0.1% BSA.
   3. Add 10 µL of secondary gold-labeled antibody to the grid and incubate for 40 min at room temperature. Wash the grids three times with PBS.
      NOTE: Here, a goat anti-mouse antibody conjugated with 10 nm gold nanoparticles was used after diluting 1:10 in blocking buffer. If gold labeling obscures EV visualization, secondary antibodies with smaller gold nanoparticles (e.g. 5 nm) can be used instead.
   4. Post-fix the grid with 10 µL of 2.5% glutaraldehyde for 10 min at room temperature. Wash three times in PBS. Perform negative staining with 2% uranyl acetate (10 µL) for 15 min. Embed the samples in 10 µL of 0.5% uranyl acetate and 0.13% methyl cellulose solution for 10 min.
   5. Allow the sample grids to dry overnight at room temperature before imaging on the electron microscope.
   6. On the microscope acquisition software, determine the exposure empirically to obtain the optimal quality of the image (e.g., 0.80851 s in this particular setup) and adjust it by typing this value into the exposure time option box. Select the 80 kV option, and click Start Acquisition to capture the image.

Results

To assess which SEC chromatography fractions were enriched for EVs, the SEC column was loaded with 2 mL of E. coli MP1-conditioned culture medium that had been concentrated 1,000-fold by TFF, and sequential fractions were collected. Using MRPS, it was found that Fractions 1-6 contained the most EVs (Figure 2A). Subsequent fractions contained very few EVs, comprising instead of EV-free proteins (Figure 2B). EVs were primarily <100 nm in diameter (Figure 2C). TEM confirmed EV-enrichment and size, particularly in Fractions 2-6 (Figure 2D).

To further ensure that the methods were able to separate EVs from non-EV-associated proteins, a recombinant strain of E. coli MP1 expressing a cytolysin A-nanoluciferase fusion protein and free (non-fused) mCherry was generated (schematized in Figure 3A). Cytolysin A fusion proteins were previously shown to associate with E. coli EVs¹⁴. To monitor mCherry fluorescence, 100 µL aliquots from each chromatography fraction were transferred to the wells of a 96-well plate. Their fluorescence was measured in a microplate reader using 580 nm and 620 nm as absorption and emission wavelengths, respectively.

Similarly, for luminescence measurement, a 20 µL aliquot of each fraction was mixed with an equal volume of the Luciferase assay solution in a 384-well plate, incubated for 15 min, and visible light luminescence was measured. It was observed that EV-enriched fractions (Fractions 2-7) had high nanoluciferase activity comparable to that of later fractions but only a very low fluorescent signal from non-EV-associated mCherry (Figure 3B). Background signal from an equal amount of negative control EVs (isolated from a matched bacterial strain lacking nanoluciferase expression) was >1,000-fold lower in EV fractions. The signal from the positive control (a nanoluciferase-expressing bacterial cell pellet) was approximately 1,000-fold higher than that from the EV fractions (not shown). The latter was normalized to the initial volume of cell culture material yielding the analyzed cells or EVs, respectively. The EV-association of nanoluciferase was confirmed in Fractions 2-5 by immunogold labeling (Figure 3C), observing the labeling within the small ~20 nm EVs isolated.

Finally, to assess the applicability of this protocol to other bacterial species, EVs were isolated from ~100 mL cultures of the following diverse anaerobic bacteria: A. mucinophila, B. thetaiotaomicron, B. breve, and B. dentium prepared in BHI culture medium. As a control, EVs were isolated from the fresh BHI culture medium. While EV yield varied by species, it was again observed that early chromatography Fractions 1-4 were enriched for EVs (Figure 4A). The complex BHI medium also contained EV-sized particles, albeit at <25% of the total EV yield in these preparations (Figure 4A, black bars). EVs of these bacteria were also found to be primarily <100 nm in size (Figure 4B).

Figure 2: Representative E. coli MP1 EV elution in early chromatography fractions. (A) Particle count by MRPS in sequential 2 mL chromatography fractions. SEC input was 2 L of bacterial culture supernatant concentrated to 2 mL. Solid line represents mean; shaded area denotes SEM. (B) Protein concentration in each fraction. (C) Size distribution of Fraction 3 EVs measured by MRPS. Solid line represents mean; shaded area represents 95% CI. Note that the instrument cannot quantify particles <50 nm. (D) Transmission electron micrographs of sequential, pooled chromatography fractions and fresh culture medium (control). All images were taken with the same scale (100 nm). Wide-field TEM images are shown in Supplemental Figure S3. Abbreviations: EV = extracellular vesicle; MRPS = microfluidics resistive pulse sensing; SEC = size exclusion chromatography; SEM = standard error of the mean; CI = confidence interval. Please click here to view a larger version of this figure.

Figure 3: Separation of E. coli MP1 EVs from EV-free proteins by SEC. (A) Schematic of E. coli MP1 expressing recombinant mCherry (red) in the cytoplasm and periplasm-trafficking cytolysin A-nanoLuciferase fusion protein (yellow diamonds). (B) nanoLuciferase bioluminescence activity and mCherry fluorescence were monitored in sequentially eluted chromatography fractions. Vertical dotted line represents the limit of EV-enriched fractions based on analyses in Figure 2. (C) EV fractions (F2-5) were immunogold labeled following staining with anti-nanoLuciferase antibody. Cyan arrowheads point to gold-conjugated secondary antibody colocalizing with small EVs. Control EVs from wildtype E. coli MP1 had negligible non-specific staining. Both images were taken with the same scale (100 nm). Wide-field TEM images are shown in Supplemental Figure S4. Abbreviations: EV = extracellular vesicle; SEC = size exclusion chromatography; F = fraction; TEM = transmission electron microscopy; ClyA-nLuc = cytolysin A-nanoLuciferase fusion protein. Please click here to view a larger version of this figure.

Figure 4: Isolation of EVs from diverse bacterial species. Indicated species were cultured for 48 h in BHI medium under anaerobic conditions. EVs were isolated by ultrafiltration + SEC. (A) EV concentration in the first 4 SEC fractions, measured by MRPS. Mean ± SEM. Black bars represent particles detected in a batch of fresh BHI medium (control). (B) Size distribution of EVs in Fraction 2. Abbreviations: EV = extracellular vesicle; MRPS = microfluidics resistive pulse sensing; SEC = size exclusion chromatography; BHI = brain heart infusion. Please click here to view a larger version of this figure.

Supplemental Figure S1: p114-mCherry-Clyluc plasmid. (A) Map of p114-mCherry-Clyluc plasmid. (B) Sequence of the J23114-mCherry-clyA-nluc region. Violet, J23114 Promoter; Pink, mCherry gene; Grey, clyA gene; Green, Linker sequence; Orange, nLuc gene. Please click here to download this File.

Supplemental Figure S2: Tangential flow filtration (TFF) setup schematic. The tubing with barb-hoses was connected to the TFF device, as shown. Feed flow tubing begins submerged in the filtered, conditioned culture medium vessel, continues through the peristaltic pump, and connects to the TFF device inlet port. The return flow begins at the TFF device outlet port and ends above the surface of the filtered, conditioned culture medium. Optionally, a backpressure clamp (e.g., a screw nut + bolt or simple paper clamp) can be used to increase the rate of filtration. As the pump circulates the conditioned medium, developed pressure within the TFF device leads to ultrafiltration and removal of components <100 kDa through the filtrate/waste flow tubing, which can be collected in a separate vessel for disposal (magenta). Please click here to download this File.

Supplemental Figure S3: Widefield transmission electron micrographs. TEM images of sequential, pooled chromatography fractions and fresh culture medium (control) shown in Figure 2D. The images show that the E. coli MP1 extracellular vesicles elute in early chromatography fractions. Please click here to download this File.

Supplemental Figure S4: Widefield TEM images for Figure 3C. The EV fractions (F2-5) were immunogold-labeled following staining with anti-nano-Luciferase antibody. Cyan arrowheads point to gold-conjugated secondary antibody colocalizing with small EVs. Please click here to download this File.

Supplemental Method: For recombinant E. coli MP1 strain harboring p114-mCherryClyluc. Please click here to download this File.

Discussion

In the protocol above, a method is described that is scalable and reliably isolates EVs from various gram-negative/positive and aerobic/anaerobic bacteria. It has several potential stopping points throughout the procedure, although it is better to avoid taking longer than 48 h to isolate EVs from conditioned bacterial culture media.

First, it consists of culturing bacteria to generate conditioned bacterial culture medium. It was found that increasing the culture time to at least 48 h and using the optimal growth medium helps to maximize the EV yield. It is likely that each bacterial species will need to be optimized with regard to these two parameters. The volume of the bacterial culture is also important to ensure sufficient EVs are isolated for the desired application. For in vitro studies, the EVs are typically isolated from a minimum of 100 mL, while for in vivo studies, EVs are typically isolated from >1 L of culture medium. Again, the EV production characteristics of each bacterial strain and the required EV amount for downstream assays will dictate the minimum starting culture volume.

Once conditioned culture medium is available, cells and large non-EV debris must be removed. It was found that centrifugation is a critical step in this process. As noted in the protocol above, two increasing g-force centrifugations were performed. Occasionally, an additional 10,000 × g centrifugation is performed if it was noted that the pellet of the second spin is not compact. Subsequently, sterile filtration of this supernatant is performed through a 0.22 µm filter. Insufficient centrifugation leads to clogging and poor performance of this filtration step. It was noted that continuing to filter the supernatant after the filtration rate has significantly slowed can lead to filter malfunction and contamination of the EV preparation with the parental bacteria. The solution to persistent clogging of the filter is to re-centrifuge and/or re-filter the supernatant, ensuring sterility. The centrifugation and vacuum-driven filtration steps described were tested for up to a total of 4 L of bacterial culture. Further scale-up of EV isolation may require modifications. For example both of these steps could be potentially be substituted with sequential pump-driven filtration using compatible filter devices with decreasing pore size, down to 0.22 μm. However, this remains to be tested.

In the current protocol, two variations are described, depending on the starting volume of the bacterial culture. For volumes <100 mL, use centrifugal ultrafiltration devices to concentrate the culture media. The MWCO is critical in these steps. For mammalian EVs, >300 kDa MWCO were previously used^11,12. However, this resulted in very poor EV yields from bacteria, presumably because of the smaller size distribution. Thus, it is recommended to use 100 kDa MWCO. A smaller MWCO can also be used but is associated with longer centrifugation times and less removal of small molecular weight contaminants, increasing sample viscosity. It is also helpful to have various sizes of ultrafiltration devices at 100 kDa MWCO to help concentrate different starting volumes of sample throughout the protocol.

Alternatively, for sample sizes significantly >100 mL, use pump-driven TFF to concentrate the sample; again, using a 100 kDa MWCO is critical. This method allows for processing large volumes of culture medium in a semi-automated fashion. It is important to obtain an appropriately-sized TFF device for the starting culture volume. The device used is rated at processing up to 200 mL of material by the manufacturer. It was possible to process up to about 2 L. However, a severe drop in the filtration rate was observed when trying to process larger volumes, requiring the process to be stopped and the device cleaned before additional processing. Thus, the characteristics of each bacterial culture and the amount of starting material will dictate the required size of the TFF device. Furthermore, the attainable pump speed is another important parameter for TFF. At low rates of ~100 mL/min, it was necessary to increase the backpressure in the TFF device using a clamp, as indicated in Supplemental Figure S2, to facilitate filtration, which increases the fouling rate of the filter. The tubing was reused up to 2 times after appropriate decontamination and autoclaving.

Once the sample is concentrated, it can then be loaded onto an SEC column to isolate the EVs. Commercial columns optimized for small EV isolation were used. For small starting samples, use columns with 0.5 mL loading volume, and use the columns with 2 mL loading volume for larger starting samples up to 2 L. It is likely that the processing of starting cultures >2 L will require larger columns. Manufacturers of EV-optimized columns currently offer SEC columns capable of accepting >100 mL of concentrated material.

Various methods are used to characterize the isolated EVs, most of which are widely available. Normalization was based on the protein concentration for most assays because this is not affected by the inability of other quantification methods (namely, particle quantification by technologies such as MRPS) to detect very small EVs <50 nm. MRPS and other nanoparticle quantification technologies remain useful in the relative quantification of EVs among the different fractions.

One critical aspect of MRPS quantification is the level of dilution. When diluted appropriately, the frequency of detected EVs should continue to increase to the limit of detection in most cases, as the instrument cannot quantify particles <50 nm. Insufficient dilution will lead to high instrument noise, which will generate an artifactual bell-shaped curve with a peak >65 nm (when using the recommended C-300 microfluidics cartridge). During size frequency distribution data analysis, an artefactual peak between 50 nm (the absolute limit of detection of the instrument) and 60 nm is still sometimes observed, despite adequate dilution. This is likely due to the presence of significant numbers of very small bacterial EVs (as visualized in TEM, Figure 2D) that are below the limit of accurate detection by MRPS and again lead to instrument noise. In this case, exclude data points smaller than the observed "peak," which becomes the de facto lower limit of quantitation of the given sample.

As described in this protocol, the quantification of EV abundance, total protein concentration, and abundance of non-EV proteins in the eluted chromatography fractions can help users decide which fractions to use for downstream assays. For example, small EVs were detected in pooled Fractions 7-8 (Figure 2D); however, their abundance was lower than that in the immediately preceding fractions, while the total protein concentration (Figure 2B) was higher. This may suggest that Fractions 7-8 contain higher amounts of non-EV-associated proteins and may thus not be desirable for certain downstream applications.

In summary, a versatile EV isolation protocol that relies on commercially available materials is described here. The significance of this methodology compared to widely used ultracentrifugation-based methods is that it comprises steps that can be easily reproduced by different users and is highly scalable. This is especially important to facilitate the generation of sufficient material for in vivo studies. It was used to isolate EVs from cultures of 100 mL to 2 L. Given the wide range of available TFF devices, it is possible that this protocol could be adapted to larger-scale purifications with some modification. The isolation protocol described is primarily based on the physical properties of EVs, namely their size, and is likely applicable to bacterial species beyond those described in this study.

One limitation of the protocol described is that it favors the isolation of small EVs, particularly <100 nm, as seen in the representative results. Prior reports also describe the presence of larger bacterial EVs^15,16. Isolation of larger bacterial EVs may require modifications of the protocol above, for example, by using SEC columns optimized for larger EVs. Such SEC columns are also commercially available. Moreover, other protocols can likely attain higher EV purity (for example, density gradient ultracentrifugation or immuno-isolation). However, these methods lack the throughput and scalability of the methods described in this study. Modification of this protocol with additional purification steps in the future may further increase the yield and purity of preparations, which could be important for experimental and therapeutic applications.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
0.5 mL flat cap, thin-walled PCR tubes	Thermo Scientific	3430	it is important to use thin-walled PCR tubes to obtain accurate readings with Qubit
16% Paraformaldehyde (formaldehyde) aqueous solution	Electron microscopy sciences	15700
250 mL Fiberlite polypropylene centrifuge bottles	ThermoFisher	010-1495
500 mL Fiberlite polypropylene centrifuge bottles	ThermoFisher	010-1493
65 mm Polypropylene Round-Bottom/Conical Bottle Adapter	Beckman Coulter	392077	Allows Vivacell to fit in rotor
Akkermansia mucinophila	ATCC	BAA-835
Amicon-15 (100 kDa MWCO)	MilliporeSigma	UFC910024
Avanti J-20 XPI centrifuge	Beckman Coulter		No longer sold by Beckman. Avanti J-26XP is closest contemporary model.
Bacteroides thetaiotaomicron VPI 5482	ATCC	29148
Bifidobacterium breve	NCIMB	B8807
Bifidobacterium dentium	ATCC	27678
Brain Heart infusion (BHI) broth	Himedia	M2101	After autoclaving, Both BHI broth and agar were introduced into the anaerobic chamber, supplemented with Menadione (1 µg/L), hematin (1.2 µg/L), and L-Cysteine Hydrochloride (0.05%). They were then incubated for at least 24 h under anaerobic conditions before inoculation with the anaerobic bacterial strains.
C-300 microfluidics cartridge	Spectradyne
Chloramphenicol	MP Biomedicals	ICN19032105
Electron microscope	FEI company	Tecnai G2 SpiritBT
Escherichia coli HST08 (Steller competent cells)	Takara	636763
Escherichia coli MP1	Dr. Mark Goulian (gift)		commensal bacteria derived from mouse gut
Fiberlite 500 mL to 250 mL adapter	ThermoFisher	010-0151-05	used with Fiberlite rotor to enable 250 mL bottles to be used for smaller size of starting bacterial culture
Fiberlite fixed-angle centrifuge rotor	ThermoFisher	F12-6x500-LEX	fits 6 x 500 mL bottles
Formvar Carbon Film 400 Mesh, Copper	Electron microscopy sciences	FCF-400-CU
Glutaraldehyde (EM-grade, 10% aqeous solution)	Electron microscopy sciences	16100
Hematin	ChemCruz	207729B	Stock solution was made in 0.2 M L-histidine solution as  1.2 mg/mL
Infinite M Nano+ Microplate reader	Tecan		This equibment was used to measure the mCherry fluorescence
In-Fusion  HD Cloning Plus	Takara	638909	For cloning of the PCR fragements into the PCR-lineraized vectors
JS-5.3 AllSpin Swinging-Bucket Rotor	Beckman Coulter	368690
Lauria Bertani (LB) broth, Miller	Difco	244620
L-Cysteine Hydrochloride	J.T. Baker	2071-05	It should be weighed and added directly to the autoclaved BHI media inside the anaerobic chamber
Masterflex Fitting, Polypropylene, Straight, Female Luer to Hose Barb Adapter, 1/8" ID; 25/PK	cole-parmer - special	HV-30800-08	connection adapters for filtration tubing circuit
Masterflex Fitting, Polypropylene, Straight, Male Luer to Hose Barb Adapter, 1/8" ID; 25/PK	cole-parmer - special	HV-30800-24	connection adapters for filtration tubing circuit
Masterflex L/S Analog Variable-Speed Console Drive, 20 to 600 rpm	Masterflex	HV-07555-00
Masterflex L/S Easy-Load Head for Precision Tubing, 4-Roller, PARA Housing, SS Rotor	Masterflex	EW-07514-10
Masterflex L/S Precision Pump Tubing, PharmaPure, L/S 16; 25 ft	Cole Palmer	EW-06435-16	low-binding/low-leaching tubing
Menadione (Vitamin K3)	MP	102259	Stock solution was made in ethanol as 1 mg/mL
MIDIKROS 41.5CM 100K MPES 0.5MM FLL X FLL 1/PK	Repligen	D04-E100-05-N	TFF device we have used to filter up to 2 L of E. coli culture supernatant
Nano-Glo Luciferase Assay System	Promega	N1110	This assay kit was used to measure the luminescence of the nluc reporter protein
NanoLuc (Nluc) Luciferase Antibody, clone 965808	R&D Systems	MAB10026
nCS1 microfluidics resistive pulse sensing instrument	Spectradyne
nCS1 Viewer	Spectradyne		Analysis software for particle size distribution
OneTaq 2x Master Mix with Standard Buffer	NEB	M0482	DNA polymerase master mix used to perform the routine PCR reactions for colony checking
Protein LoBind, 2.0 mL, PCR clean tubes	Eppendorf	30108450
Q5 High-Fidelity 2x Master Mix	NEB	M0492	DNA polymerase master mix used to perform the PCR reactions needed for cloning
qEV original, 35 nm	Izon		maximal loading volume of 0.5 mL
qEV rack	Izon		for use with the qEV-original SEC columns
qEV-2, 35 nm	Izon		maximal loading volume of 2 mL
Qubit fluorometer	ThermoFisher		Item no longer available. Closest available product is Qubit 4.0 Fluorometer (cat. No. Q33238)
Qubit protein assay kit	ThermoFisher	Q33211	Store kit at room temperature. Standards are stored at 4 °C.
Sorvall Lynx 4000 centrifuge	ThermoFisher	75006580
SpectraMax i3x Microplate reader	Molecular Devices		This equipment was used to measure the nanoluciferase bioluminescence
Stericup Quick-release-GP Sterile Vacuum Filtration system (150, 250, or 500 mL)	MilliporeSigma	S2GPU01RE S2GPU02RE S2GPU05RE	One or multiple filters can be used to accommodate working volumes. In our experience, you can filter twice the volume listed on the product size.
Uranyl acetate	Electron microscopy sciences	22400
Vinyl anaerobic chamber	Coy Lab
Vivacell 100, 100,000 MWCO PES	Sartorius	VC1042
Whatman Anotop 10 Plus syringe filters (0.02 micron)	MilliporeSigma	WHA68093002	to filter MRPS diluent