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.
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.
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 adapted11,12 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.
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.
2. EV isolation
3. EV preparation quality control
4. EV storage
5. Transmission electron microscopy
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.
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 used11,12. 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 EVs15,16. 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.
The authors have nothing to disclose.
The research described above was supported by NIH TL1 TR002549-03 training grant. We thank Drs. John C. Tilton and Zachary Troyer (Case Western Reserve University) for facilitating access to the particle size analyzer instrument; Lew Brown (Spectradyne) for technical assistance with analysis of the particle size distribution data; Dr. David Putnam at Cornell University for providing pClyA-GFP plasmid14; and Dr. Mark Goulian at the University of Pennsylvania for providing us with the E. coli MP113.
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 | |
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 | F12-6×500-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 |