A preparation of bacterial outer membrane with osmium tetroxide and uranyl acetate co-stain enables improved structural determination by transmission electron microscopy

Abstract

Biological nanoparticles, such as bacterial outer membrane vesicles (OMVs), are routinely characterized through transmission electron microscopy (TEM). In this study, we report a novel method to prepare OMVs for TEM imaging. To preserve vesicular shape and structure, we developed a dual fixation protocol involving osmium tetroxide incubation prior to negative staining with uranyl acetate. Combining osmium tetroxide with uranyl acetate resulted in preservation of sub-50 nm vesicles and improved morphological stability, enhancing characterization of lipid-based nanoparticles by TEM.

The rapid advancement of nanotechnology in recent decades has contributed to rising interest in the characterization and application of nanoparticles [1,2]. The study of biological nanoparticles has garnered interest due to their involvement in human physiology, homeostasis and pathology [3,4]. A subset of biological nanoparticles are extracellular vesicles (EVs) that are ubiquitous to both prokaryotic and eukaryotic cells [5–9]. Gram-negative bacteria secrete EVs that have similar biomolecular components as their cell of origin [10–12]. Gram-negative bacteria, in particular, release outer membrane vesicles (OMVs) that can elicit potent immune responses in the host cells and are of significant interest in drug and vaccine development [12–15].

Accurate imaging of OMVs aids in the characterization of their physical properties (i.e. shape, lipid and protein profiles) and mechanisms of vesicle biogenesis and interactions [16–18]. These vesicles range widely in size from 20 to 200 nm in diameter and arise from the bacterial outer membrane [5]. The mechanisms of biogenesis are poorly understood with the current models including activation of stress-response pathways, reduced cross-linking of peptidoglycan to the outer membrane and accumulation of biomolecules in the periplasmic space [11,19,20]. While OMVs cannot be visualized in detail through conventional light microscopy techniques, transmission electron microscopy (TEM) enables high-resolution imaging of OMVs, most notably through negative staining [21,22].

Uranyl acetate (UA), a commonly utilized negative stain for TEM for >60 years [23,24], is an acetate salt of uranium oxide that binds to a wide range of biomolecules [25,26]. Although UA can stain lipids with sialic acid groups like glycoproteins and gangliosides [26], OMVs are composed primarily of lipopolysaccharides (LPSs), which lack the moieties necessary for interactions with the uranium oxide ions [27], leading to issues with imaging the OMVs post-UA staining such as poor contrast and image quality. Another problem that arises is that drying of the UA-stained vesicles can lead to modification of the vesicle form and structure, resulting in issues with downstream analysis of the OMVs [28,29]. Osmium tetroxide is another oft-used electron microscopy stain where the osmium ions integrate into the phospholipid membrane and rapidly oxidize to osmium dioxide, giving the cellular substructures greater electron density, allowing them to appear darker in the microscope’s electron beam [30]. This mechanism of action also allows the molecule to act as a fixative as it preserves the structure of cellular samples and membranes through its reaction with the unsaturated double bonds of adjacent lipids, enabling the formation of a stable polymerized ester structure in the sample that can resist physical and chemical adjustments [30–32]. Thus, we hypothesized that the reaction of osmium tetroxide with unsaturated lipids is a superior alternative for the negative staining of EVs.

To test this hypothesis, we examined the addition of osmium tetroxide treatment to OMVs in conjunction with UA staining to enhance the quality of TEM images. In this report, we compared the negative staining of OMVs from enterotoxigenic Bacteroides fragilis (ETBF) and non-toxic B. fragilis (NTBF) under two conditions: UA staining alone (Fig. 1) versus co-staining osmium tetroxide with UA (Fig. 2).

Fig. 1.

OMV staining with UA alone. TEM micrographs of ETBF (a and b) and NTBF OMVs (c and d) stained with UA. Arrows denote vesicles that appear collapsed or that experienced a loss of structure. Scale bars, 100 nm. ×50 000 magnification.

ETBF, enterotoxigenic Bacteroides fragilis; NTBF, non-toxic Bacteroides fragilis; OMV, outer membrane vesicle; TEM, transmission electron microscopy; UA, uranyl acetate.

Fig. 2.

OMV staining with OsO₄ and UA. TEM micrographs of ETBF (a and b) and NTBF (c and d) OMVs stained with osmium tetroxide and UA. Vesicles are expected to be spherical in shape. Scale bars, 100 nm. ×50 000 magnification.

ETBF, enterotoxigenic Bacteroides fragilis; NTBF, non-toxic Bacteroides fragilis; OMV, outer membrane vesicle; TEM, transmission electron microscopy; UA, uranyl acetate.

Using only UA staining (Fig. 1), we observed collapsed vesicular forms (a and b) and the formation of irregularly shaped vesicles and lipid fragments (c and d), indicative of OMV degradation. Additionally, the low image contrast, combined with irregular vesicular morphologies, prevents accurate quantification of critical parameters. However, co-staining with UA and osmium tetroxide improves contrast and morphological characterization of vesicles (Fig. 2) as demonstrated by enumeration and by calculating the roundness of the vesicles observed (Fig. 3).

Fig. 3.

Quantitative analyses of OMV size (a and b) and roundness (c and d). Comparisons between different staining methods are shown as a box blot (a and c) and density plot (b and d). Osmium tetroxide and UA co-stain was able to visualize smaller vesicles and rounder vesicles in greater quantity. Statistics were calculated using Welch’s t-test.

OMV, outer membrane vesicle; UA, uranyl acetate.

Enumeration, size and morphology analyses of vesicles from obtained images show that the co-stain is able to reveal significantly smaller, more round vesicles compared to UA alone (Fig. 3e). With respect to the co-stain, the fixative properties of osmium on lipids allow it to stain vesicles to preserve the structure to a much greater degree than UA alone. We note that using cryo-EM would be advantageous in revealing accurate dimensions and shape and that using phosphotungstate, silicotungstate or ammonium molybdate could also improve negative staining.

In summary, the added contrast of UA in the co-stain facilitates visualization of smaller vesicles, enabling a greater enumeration of OMVs for representative quantification. The low contrast of UA alone, in combination with irregular vesicle structure, yields lower enumeration sample size and prevents accurate measurement of the OMVs. Our results demonstrate that the co-staining of OMVs with osmium tetroxide and UA can better preserve the structure of vesicles, resulting in higher-quality images.

OMVs from ETBF and NTBF were isolated from stationary bacterial cultures using the ExoBacteria OMV Isolation Kit (System Biosciences, EXOBAC100A-1) as per the manufacturer’s instructions and were concentrated using centrifugal concentration tubes (Thermo Scientific, 88532).

Concentrated vesicles were stained with UA or co-stained with osmium trioxide and UA. For UA, formvar-coated copper grids (Electron Microscopy Sciences, Hatfield, PA) were incubated on 10 µL drops of OMV-containing solution for 5 min, followed by two washes on deionized water drops for 2.5 min each and finally staining with 2% UA for 1 min. For the osmium tetroxide and UA co-stain, 10 µL of OMVs were mixed with an equivalent volume of 4% osmium tetroxide for 10 min on formvar-coated copper grids followed by staining with 2% UA for 1 min. All grids had excess liquid blotted off with filter paper and were allowed to dry overnight. The grids were then imaged using a TEM (JEM-1010, JEOL Inc., Tokyo, Japan) at 60 kV with a spot size of 2 and a magnification of ×50 000; a minimum of seven images per test condition were obtained.

The diameter and size of OMVs were measured using the Olympus CellSens Dimensions Software (Olympus America Inc., Version 2.2). Prior to analysis, a number generator was used to pick five random images per experiment to sample for counting. To determine the ‘roundness’ of the OMVs, we used Eq. 1 as previously described [33]. Briefly, the diameter of the OMVs in the x-direction was divided by the diameter from the y-direction and subtracted from 1, and the absolute value of the resulting number was then recorded. Values closer to 0 indicate a rounder shape, while values closer to 1 indicate a more linear form.

(1)

Funding

Funding for this project was provide by a grant to K.L.G. from the NIH (1R15AI156742). Funding support for A.S. was provided by Baylor University Biology Department.