Click-correlative light and electron microscopy (click-AT-CLEM) for imaging and tracking azido-functionalized sphingolipids in bacteria

Sphingolipids, including ceramides, are a diverse group of structurally related lipids composed of a sphingoid base backbone coupled to a fatty acid side chain and modified terminal hydroxyl group. Recently, it has been shown that sphingolipids show antimicrobial activity against a broad range of pathogenic microorganisms. The antimicrobial mechanism, however, remains so far elusive. Here, we introduce ‘click-AT-CLEM’, a labeling technique for correlated light and electron microscopy (CLEM) based on the super-resolution array tomography (srAT) approach and bio-orthogonal click chemistry for imaging of azido-tagged sphingolipids to directly visualize their interaction with the model Gram-negative bacterium Neisseria meningitidis at subcellular level. We observed ultrastructural damage of bacteria and disruption of the bacterial outer membrane induced by two azido-modified sphingolipids by scanning electron microscopy and transmission electron microscopy. Click-AT-CLEM imaging and mass spectrometry clearly revealed efficient incorporation of azido-tagged sphingolipids into the outer membrane of Gram-negative bacteria as underlying cause of their antimicrobial activity.

www.nature.com/scientificreports/ on the interaction of the modified sphingolipid with the bacterial membrane, or accumulation in the cytoplasm where it might interfere with the microbial metabolism. Various imaging methods can be used to analyze the mode of action of antimicrobial agents and complement biophysical studies. The combination of high-resolution imaging with fluorescence imaging and refined labeling techniques paves the way for tracking and quantification of an antimicrobial compound throughout the cell.
In this study, we established a protocol based on a combination of correlative light and electron microscopy (CLEM), i.e. super-resolution array tomography (srAT) 17,18 and click chemistry 19 , which we call in short 'click-AT-CLEM' . This combination of fluorescence and electron imaging applied here to the Gram-negative bacterium N. meningitidis allowed to determine morphological and ultrastructural changes after treatment with azido-modified sphingolipid analogs as well as to visualize their subcellular localization. A dibenzocyclooctyne (DBCO)-containing fluorescent dye (Alexa Fluor 488 DIBO analog (AFDye 488 DBCO AF)) was used for fluorescent labeling of azido-tagged sphingolipid analogs 13,16,[20][21][22] via Cu(I)-free strain-promoted alkyne-azide cycloaddition click chemistry reaction (SPAAC) 23,24 . Our data demonstrate that srAT technology in combination with click chemistry-based labeling reaction is ideally suited to visualize, track and quantify azido-modified antimicrobial sphingolipids in bacteria.

Results
Inhibitory and bactericidal activity of a novel sphingolipid analog, ω-N 3 -sphingosine, against N. meningitidis. We have recently documented minimal inhibitory and bactericidal concentration (MIC and MBC) values for the antimicrobial sphingolipid sphingosine as well as for short-chain and long-chain ceramides and several ceramide analogs, including α-N 3 -C 6 -ceramide, ω-N 3 -C 6 -ceramide, α-N 3 -C 16 -ceramide and ω-N 3 -C 16 -ceramide, against the Gram-negative microorganism N. meningitidis by broth microdilution assays and time killing studies 13 . The data demonstrated a potent bactericidal activity of sphingosine and the synthetic ceramide analog ω-N 3 -C 6 -ceramide against N. meningitidis 13 . Here, we now determined the MIC and MBC values for a newly synthesized sphingosine analog, ω-N 3 -sphingosine 22 , against N. meningitidis. ω-N 3 -sphingosine displayed a MIC value of 4 µg/ml and a MBC value of 8 µg/ml comparable to MIC/MBC values observed for the unmodified sphingosine for N. meningitidis (Table S1) 13 . In line with our previous study, we included Escherichia coli and Staphylococcus aureus as control organisms and MIC/MBC values of 16 µg/ml (MIC)/16 µg/ml (MBC) (for E. coli) and 8 µg/ml (MIC)/16 µg/ml (MBC) (for S. aureus) were determined (Table S1).

Functionalized sphingolipids induce concentration dependent membrane alteration and disruption in N. meningitidis.
To investigate the effects of two azido-modified sphingolipids, ω-N 3sphingosine and ω-N 3 -C 6 -ceramide, on N. meningitidis, bacteria were treated with different amounts of the two compounds and analyzed by scanning electron microscopy (SEM) and transmission electron microscopy (TEM). Azido-modified sphingolipids allowed imaging of sphingolipids in prokaryotes 13 via conventional fluorescence microscopy 20 or high-resolution microscopy 13 . As control, N. meningitidis were treated with ethanol (solvent control). Ethanol treated control N. meningitidis exhibited coccus morphology with typical 'kidney or coffee-bean shape' and SEM showed the characteristic pairs of N. meningitidis cells forming a diplococcus (Fig. 1C). Control bacteria showed a slight increase in their electron density by TEM, but no morphological changes compared to non-solvent treated bacteria ( Fig. 1A-C, Fig. S1). In contrast, N. meningitidis treated with a functionalized ω-N 3 -sphingosine or a ω-N 3 -C 6 -ceramide were distorted to various degrees and their surfaces were wrinkled when treated with a concentration corresponding to 1 × MBC ( Fig. 1G-I, N-P). Bacteria showed an increased number of external blebs in SEM sections. In addition, bacteria appeared in various stages of lysis with compromised cell wall and plasma membrane. Bacteria treated with a lower concentration of the functionalized sphingolipids corresponding to 0.1 × MBC showed only a slight alteration of their surfaces and shape compared to control N. meningitidis.
In thin sections, control N. meningitidis showed the typical coccus morphology, the outer and inner membrane and the release of outer membrane vesicles were visible ( Fig. 1A-B). In N. meningitidis treated with 0.1 × MBC of either ω-N 3 -sphingosine or ω-N 3 -C 6 -ceramide, some cells were still intact with an outer and inner membrane ( Fig. 1D-F, K-M). In some bacteria, there was intact intracellular content, however with strong elongation of the outer membrane (Fig. 1E). N. meningitidis treated with 1 × MBC of both compounds showed different stages of disintegration and lysis (Fig. 1G,H,N,O). The cytoplasm was not uniform with flocculation and aggregation of intracellular contents. Bacteria showed intracellular inclusion bodies appearing in high electron density in electron micrographs and additional vesicles that filled the intracellular content (Fig. 1O).
We also investigated the effects of both azido-modified sphingolipids on E. coli und S. aureus by TEM analysis. Control S. aureus had a typical 'coccus' morphology and formed grape-like clusters (Fig. S2). S. aureus treated with ω-N 3 -C 6 -ceramide were unaffected and showed intact cell walls and the cytoplasmatic membranes were clearly visible (Fig. S2G,H). In line with this finding, when E. coli were treated with ω-N 3 -C 6 -ceramide bacteria were also unaffected, showed intact cell wall and membranes and the cytoplasm had a uniform granularity (Fig. S2E,F). In contrast, both S. aureus and E. coli treated with 1 × MBC of ω-N 3 -sphingosine were distorted to various degrees. ω-N 3 -sphingosine-treated E. coli showed electron dense intracellular inclusion bodies and additional vesicles that filled the intracellular content (Fig. S2A,B). Like that seen in E. coli, S. aureus treated with ω-N 3 -sphingosine also contained electron dense intracellular content and, in some bacteria, the remaining cell wall and cytoplasm were not intact and an aggregation of flocculation of intracellular content appeared (Fig. S2C,D).
Click-AT-CLEM imaging reveals localization of azido-functionalized sphingolipids in N. meningitidis. While   www.nature.com/scientificreports/ the membrane and cell wall, we aimed to use a labeling protocol to track and quantify the sphingolipid distribution at subcellular level. As this is very challenging with established immunolabeling approaches for electron microscopy, we developed a novel pre-embedding correlative light and electron microscopy (CLEM) protocol to achieve a more precise localization of the uptake of the two azido-modified sphingolipids, ω-N 3 -sphingosine and ω-N 3 -C 6 -ceramide, by N. meningitidis. Recently, we successfully demonstrated that the toolkit of azido-modified sphingolipids can be used for imaging sphingolipids in mammalian cells 20 and prokaryotes 13 via conventional fluorescence microscopy 20 or high-resolution microscopy 13 . Super-resolution array tomography (srAT) 17,18,25,26 is an antibody-based staining CLEM workflow that combines preparation of the samples for super-resolution microscopy (e.g. structural illumination microscopy (SIM)) followed by electron microscopy, LR-White embedding and preparation of 100 nm ultrathin sections. Staining with sphingolipid specific antibodies, to detect only the incorporated sphingolipids, then takes place on the array of serial sections. Due to the embedding step prior to labeling, only a proportion of the epitopes are accessible to the staining protocol with antibodies, whereas a larger number of epitopes can be hidden in the resin making them inaccessible for the antibody 27 .
In contrast, azido-functionalized sphingolipid analogs can be visualized and tracked directly by fluorescence microscopy. To overcome this limitation, we used the functional azido group coupled to the ω-position of sphingosine or the amine-bound fatty acid side chain of ceramides, respectively, utilizing a copper-free click chemistry reaction. A dibenzocyclooctyne (DBCO)-containing fluorescent dye [Alexa Fluor 488 DIBO analog (AFDye 488 DBCO AF)] was used for the fluorescent labeling, that has been tested to preserve its fluorescence embedded in the LR-White resin, thus allowing CLEM for super-resolution fluorescence analysis followed by EM preparation protocols. Due to the pre-embedding labeling, the entire cell surface was accessible for direct labeling with the dye and not only the parts of the cell membrane facing the section surface as it would be the case in post-embedding on-section staining. In addition, click chemistry allowed us to specifically stain only the sphingolipids, which becomes especially important due to the lack of lipid specific antibodies (Fig. 2).
To test the applicability of copper-free click chemistry for CLEM, bacteria were treated with different concentrations of ω-N 3 -C 6 -ceramide or ω-N 3 -sphingosine and labeled with AFDye 488 DBCO AF. Control cells that were not treated with functionalized sphingolipid analogs exhibited only minimal levels of unspecific background fluorescence after labeling with the dye (Fig. S3). Treatment of N. meningitidis with 0.1 × the MBC of ω-N 3 -C 6 -ceramide and subsequent labeling with AFDye 488 DBCO AF showed a similar morphological pattern by CLEM and only a subset of bacteria was affected ( Fig. 3a-ii). Whereas the majority of the bacteria still displayed an unaltered 'kidney or coffee-bean shape' , some bacteria showed elongations of their outer membrane ( Fig. 3aii). In addition, all bacteria showed an unaffected cytosolic compartment with even dense content, which could also be seen on DNA staining images ( Fig. 3a-iii). Interestingly, for some bacteria CLEM imaging revealed that dye-labeled ω-N 3 -C 6 -ceramide clearly integrated into the bacterial membrane and was found especially in the elongated outer membrane ( Fig. 3a-i,iv). In contrast, treatment of N. meningitidis with 1 × the MBC of ω-N 3 -C 6ceramide and subsequent dye labeling resulted in distortion of the bacteria to various degrees and condensation of electron dense material in the cytosol of bacteria ( Fig. 3b-ii). Moreover, dye-labeled ω-N 3 -C 6 -ceramide could not be detected in the bacterial membrane (Fig. 3b). CLEM imaging revealed that the electron dense material clearly correlated with accumulation of the labeled sphingolipid analog.
Next, bacteria were treated with a low concentration of ω-N 3 -sphingosine (0.1 × the MBC) and subsequently labeled with AFDye 488 DBCO AF and imaged by CLEM. Treatment of N. meningitidis with 0.1 × the MBC of CLEM imaging showed that the functionalized sphingolipid integrated into the membrane extensions and could barely be found in the cytosol of the bacteria ( Fig. 4a-i,iv). When N. meningitidis were treated with 1 × the MBC of ω-N 3 -sphingosine and subsequently labeled with the dye, bacteria displayed a similar morphology compared to bacteria treated with the functionalized ω-N 3 -C 6 -ceramide without dye labeling and analyzed by SEM (Fig. 4b). Bacteria appeared in various stages of lysis with compromise of the cell wall, outer and inner membrane and demonstrated an altered shape compared to control bacteria ( Fig. 4b-ii,iii). By SEM we observed a strong accumulation in the electron dense regions in the cytosol of the bacteria. By CLEM imaging we could clarify that the electron dense region correlated to staining with and accumulation of the dye labeled ω-N 3sphingosine ( Fig. 4b-i,iv).
Functionalized sphingolipids primarily affect the outer membrane of N. meningitidis as confirmed by mass spectrometry. Due to the lower resolution of the fluorescence imaging, compared to the electron microscopy, we aimed to confirm the indicated localization in the outer membrane, after low concentration treatment, by HPLC-MS/MS. For that, we separated the inner and outer membrane of the bacteria after incubation with 0.1 × the MBC of either ω-N 3 -sphingosine or ω-N 3 -C 6 -ceramide and quantified the amount of functionalized lipids in the different membrane fractions (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12). Both treatments show a similar picture in which the lactate dehydrogenase (LDH) activity, a marker for the inner membrane, decreases with the increasing fractions number (from 1 to 12), whereas the amount of the outer membrane protein OpcA increased (Fig. 5a,b). The amount of ω-N 3 -sphingosine (Fig. 5a) or ω-N 3 -C 6 -ceramide (Fig. 5b), measured in the same fractions that have been used for the fraction characterization, show a negative correlation with the LDH activity and a positive correlation with the OpcA amount. Classification of the fractions into inner membrane, intermediate and outer membrane was done with the ethanol (EtOH) treated control group, based on the LDH activity, and transferred to the treated samples (Fig. S4).
For ω-N 3 -sphingosine and ω-N 3 -C 6 -ceramide, fraction number 9 seemed to be the cleanest outer membrane fraction with the lowest LDH activity and the strongest OpcA signal. In this specific fraction, the amount of functionalized lipids was strongly increased, especially compared to the inner membrane fractions indicating the primary effect on the outer membrane. Of note, functionalized sphingolipids were quantified in relation to www.nature.com/scientificreports/ the canonical sphingolipid C16 sphingomyelin (C16 SM), a typical membrane lipid whose occurrence in membrane fractions is most likely due to the medium used for cultivation of N. meningitidis strain MC58 (PPM+). HPLC-MS/MS experiments reveal that the medium contained 1.79 ± 0.01 pmol/ml C16 SM. Therefore, bacteria of the membrane separation experiment were exposed to 179 pmol C16 SM via the cultivation medium. It therefore seems plausible that the bacteria have absorbed the C16 SM detected in the membrane fractions (0.38-4.80 pmol/ml) from the medium and integrated it into their membranes. For an endogenous synthesis of sphingomyelins in N. meningitidis there is no evidence in the literature.

Discussion
In this study we observed that beside ω-N 3 -C 6 -ceramide 16 an azido-modified sphingosine, ω-N 3 -sphingosine 22 , exhibited significant bactericidal activity. We show that an observable effect on bacteria of both modified sphingolipid analogs is distortion of the cell membrane. The introduced srAT technology in combination with click chemistry-based labeling reaction (Click-AT-CLEM) clearly revealed that the azido-modified sphingolipids localize to the outer membrane resulting in strong elongation of the outer membrane. Pre-embedding labeling as used in this study added another level of versatility to the srAT workflow. It enabled super-resolved localization of incorporated fluorescent tags in the full ultrastructural background without on-section labelling that is normally necessary in AT workflows 25,26 . This adds a level of simplification to the protocol as on section labelling steps are typically laborious and can be error prone [28][29][30] . To verify our findings, we separated the outer and inner membrane of bacteria after treatment with ω-N 3 -C 6 -ceramide or ω-N 3 -sphingosine for qualitative observations by quantitative mass spectrometric measurements. Indeed, HPLC-MS/MS results confirmed the primarily effect of the azido-modified sphingolipids on the outer membrane of N. meningitidis.
Human lipids with antimicrobial properties include free fatty acids, monoglycerides and sphingolipids, and their activity against Gram-positive and Gram-negative bacteria have been demonstrated in various studies 3,4,[8][9][10][11][12][13] . Sphingolipids, including ceramides, form a diverse group of structurally related lipids and are composed of a backbone of sphingoid bases coupled to a fatty acid side chain. A broad range of different head motives and complex glycosylation pattern results in further variability 1 . Sphingoid bases and certain fatty acids are present in the oral mucosa and saliva. They are produced by either the oral epithelium or sebaceous glands. Due to their antimicrobial activity, recent work suggested that these lipids are also likely involved in innate immune defense against mucosal microorganisms.
The nasopharynx is the site of colonization by meningococci and the primary site of invasion prior to the development of systemic infection, such as sepsis and meningitis. Meningococci adhere to the nasopharyngeal mucosa via interactions between the human epithelial cells and a variety of adhesins and invasins [31][32][33][34][35][36] . It is likely that a number of factors that contribute to the integrity of the mucosal barrier and prevent both colonization and invasion and antimicrobial lipids may probably play a role in preventing colonization and invasion of these bacteria. We have recently analyzed the antimicrobial properties of sphingolipids against N. meningitidis 13 , including sphingosine, short-chain C 6 and long-chain C 16 -ceramides as well as azido-functionalized ceramide analogs. We found that treatment with sphingosine, short-chain C 6 ceramide and ω-N 3 -C 6 -ceramide lead to efficient killing of N. meningitidis 13 . Of note, short-chain C 6 ceramide and ω-N 3 -C 6 -ceramide were inactive against E. coli and S. aureus 13 . Here, we now extended the findings of our previous study and included an azidofunctionalized sphingosine to determine its antimicrobial activity against N. meningitidis. ω-N 3 -Sphingosine displayed significant antimicrobial activity against N. meningitidis, and MIC/MBC values were comparable to MIC/MBC values displayed for the unmodified sphingosine.
However, while the inhibitory activity of those antimicrobial sphingolipids has been investigated for a longer time, using biological approaches by us and other researchers 3,4,13 , the exact mechanism of the antibacterial activity on the bacterial cell was only recently investigated for the natural sphingosine 14 . The authors showed a massive increase in membrane permeability in Pseudomonas aeruginosa and S. aureus after treatment with sphingosine and linked this to the interaction between the protonated form of the sphingosine NH 2 group and the highly negatively charged bacterial membrane lipid cardiolipin. However, because ceramides including ω-N 3 -C 6 -ceramide lack this specific NH 2 group, due to the fatty acid side chain, the mode of action of ω-N 3 -C 6ceramide remained elusive.
Gram-positive bacteria are characterized by having a cytoplasmatic membrane and a thick peptidoglycan cell wall, which confers the characteristic cell shape and provides the cell with mechanical protection 37 . Gramnegative bacteria such as the β-proteobacterium N. meningitidis are characterized by the presence of two distinct membranes, called inner and outer membrane and a thin peptidoglycan cell wall between them 37 . The bacterial cell membranes are mainly formed by polar lipid bilayers (e.g. phospholipids) and it is likely that sphingolipids may insert into the outer membrane or the cytoplasmatic membrane of Gram-negative bacteria. Insertion into bacterial membranes may directly change the physical properties of the bacterial membrane and render the membrane non-functional. Alternatively, sphingolipids may penetrate and accumulate in the cytoplasm and may interfere with the cell metabolism. To address the question of the mode of action, electron microscopy (EM) has been widely used to visualize the antibacterial activity of antimicrobial lipids by determining morphological changes after treatment of bacteria with the respective antimicrobial lipid. Most studies utilize transmission electron microscopy (TEM) for imaging the effects of lipids [38][39][40][41][42][43] , since this technique allows the characterization of surface morphology along with the density of inner cytoplasmatic constituents, presence of fibers or cell vacuolization. Here we first visualized the effect of two azido-modified sphingolipids on N. meningitidis by scanning electron microscopy (SEM) and TEM. Treatment of N. meningitidis with 0.1 × MBC of both compounds showed strong elongation of the outer membrane, however the cell wall and intracellular content were intact. When N. meningitidis were treated with 1 × MBC of both compounds, bacteria showed different stages of disintegration and lysis. The cytoplasm was not uniform with flocculation and aggregation of intracellular contents. Bacteria showed electron dense intracellular inclusion bodies and additional vesicles that filled the intracellular content. Interestingly, our results are in line with recent published data observed for E. coli and S. aureus after treatment with different sphingoid bases 12 . In a study by Fischer and colleagues the authors examined the effects of different sphingoid bases, including sphingosine, dihydrosphingosine and phytosphingosine, on E. coli and S. aureus in detail by TEM and SEM 3,12 . Sphingosine treated cells of both E. coli and S. aureus contained electron dense intracellular bodies similar to treatment with the azido-modified sphingosine as reported in our study. Moreover, treatment of E. coli with unmodified sphingosine resulted in surface bleb formation, while the cell wall appeared to be intact 12  www.nature.com/scientificreports/ to those obtained when cells are treated with the unmodified sphingosine: bacteria are also in various stages of disintegration and lysis and cellular debris are clearly visible near damaged cells 12 .
It is interesting to note that azido-modified ceramides had no growth inhibitory effect on E. coli or S. aureus 13 . In case of S. aureus, this finding might be explained due to the fact that azido-modified ceramides initially integrate into the outer membrane-which is absent in Gram-positive bacteria-to exhibit toxic effects or that azido-modified ceramides cannot pass the thick peptidoglycan cell wall to interact with the cytoplasmatic layer. In case of E. coli, we tested the hypothesis that due to the presence of the complex glycolipid, the lipopolysaccharide (LPS), in the outer leaflet of the lipid bilayer, azido-modified ceramides may be hindered to intercalate into the outer membrane. In contrast to E. coli, N. meningitidis expresses a lipooligosaccharide (LOS), lacking the O-antigen of the classical LPS 44 . We therefore tested an E. coli K12 strain MG1655, lacking the O-antigen 45,46 for its susceptibility to treatment with azido-modified ceramides and determined MIC and MBC values. For E. coli K12 strain MG1655, lacking the O-antigen, MIC/MBC values ˃ 64 µg/ml were determined and demonstrated that bacteria are still resistant (Table S1).
Antibody based CLEM approaches for precise localization of epitopes of interest in the ultrastructural context combine the advantages of light and electron microscopy. Said that, often the limit is the antibody itself due to multitude of possible technical obstacles, such as weakness of interaction in the EM-sample preparation steps, pure size of the antibody that hinders precise localization or, in the worst case, complete lack of a suitable antibody for a given epitope. The click-chemistry approach paves the way to circumvent several of these obstacles of antibody based CLEM approaches as successfully shown for click-labelling on cryosections 47 . Although a big breakthrough in CLEM, labelling of sections, be it cryo or resin-sections, is often laborious and technically challenging. To circumvent these limits of on-section labelling, we established a novel pre-embedding labelling protocol based on super-resolution array tomography (srAT) 17,18,25,26 and click chemistry reaction 19 for fluorescence and electron imaging of azido-modified sphingolipid analogs on N. meningitidis. The improved srAT-CLEM approach (Click-AT-CLEM), developed for this study, combines the strength of classical AT-CLEM protocols with the advantages of click chemistry 18,48 . The possibility of epitope localization within the EM resolution range is strongly improved by the unmatched specificity of click chemistry reactions. In the classical, antibody-based AT approach, post fixation and embedding staining leads to a reduction of accessible epitopes 27 . Due to the prefixation and embedding staining in the novel Click-srAT protocol, we achieved an accessibility of all surface epitopes thus leading to an improved signal with less hand on time.
Our CLEM studies confirmed the results of the morphological studies und clearly showed the incorporation of azido-modified sphingolipids into the outer membrane of N. meningitidis. This novel pre-embedding Click-srAT protocol has the advantage that it can be combined with post-embedding antibody staining as in classical AT and srAT. This adds another level of versatility to our novel approach. Notably, as the clicked dye is integrated within the sections and not only on the surface the strength of the signal can be increased by turning to thicker sections if the detection of the signal is limiting and thereby adding up signal throughout the thickness of the section. Furthermore, also the fluorescent channels could be potentially switched or even combined in a multi-channel pre-embedding labelling approach, as we could show that also rhodamine stays fluorescent in the LR-White resin throughout the embedding steps 49 . An important point in qualitive experiments, like the established Click-AT-CLEM approach, is the proof of observation by quantitative experiments. For that purpose, we successfully separated the inner (fractions 1-4) and outer membrane (fractions 8-12) of N. meningitidis after treatment with the functionalized lipids, as shown by the comparison of LDH activity and OpcA abundance, and used these samples for mass spectrometric analyses (Fig. 5). The results clearly showed a pronounced increase in concentration with fraction 9 being the one with the highest observed amount of both functionalized sphingolipids. Remarkably, this fraction appeared to be the cleanest outer membrane fraction with a strong OpcA signal and the lowest LDH activity. These results confirm our observations during the Click-AT-CLEM observations and emphasize the primary effect of the outer membrane during low concentration treatment. This novel approach complements biological approaches, such as growth inhibitory assay and SEM or TEM, and enables to decipher the mechanism of the antibacterial activity of sphingolipids.
Neisserial membrane separation. Membrane separation of the meningococcal outer and inner membrane (OM/IM) was performed with minor changes, as previously described 53 . Briefly, either ω-N 3 -C 6 -ceramide, ω-N 3 -sphingosine or a corresponding amount of EtOH were added to a final concentration of 0.1 × the MBC to 100 ml cultures of N. meningitidis strain MC58 with an OD600 of 0.1. After 3 h of incubation at 37 °C in an Scientific Reports | (2021) 11:4300 | https://doi.org/10.1038/s41598-021-83813-w www.nature.com/scientificreports/ orbital shaker, bacteria were harvested and washed three times with PBS and in the end resuspended in 50 mM Tris-HCl (pH 8). Afterwards, lysozyme (100 µg/ml) and EDTA (pH 8, 5 mM) were added and the bacteria were incubated for 1 h at RT while shacking. This step was followed by one freeze thaw cycle (− 80 °C to 37 °C) and sonication (10 s pulses with 5 s breaks, at 100% magnitude for a total of 5 min). Unbroken bacteria were removed by centrifugation at 1200g and 4 °C for 10 min. Then, the supernatant was loaded on top of a sucrose gradient, consisting of a 3 ml 15% top layer over a 2 ml 55% cushion (both with 5 mM EDTA pH 8), and centrifuged at 280,000g and 4 °C for 2 h. The crude membrane fraction was then collected, the sucrose concentration estimated with a refractometer and lowered to 30% by dilution with dH 2 O (with 5 mM EDTA, pH 8). Afterwards, the sample was loaded onto a sucrose gradient consisting of 1.3 ml 50% cushion and 45, 40, 35% layer on top (2.4 ml each). The gradient was then centrifuged for 41 h at 268,000g and 4 °C and afterwards 800 µl samples were collected from top to bottom. Then, the purity of the isolated OM and IM fraction was analyzed. Lactate dehydrogenase (LDH) was chosen as inner membrane marker 54 and the presence of the meningococcal OpcA protein as outer membrane marker. LDH was determined with a commercial LDH activity assay and used according to the manufacturer's instructions (Sigma Aldrich). Activity was calculated as LDH activity/µg protein (determined by Bradford assay). To visualize the relative amount of OpcA in the samples, the protein amounts were adjusted and utilized in a dot blot assay with a monoclonal mouse anti-OpcA antibody (clone B306, kindly provided by M. Achtman). 4 µl of each sample was spotted onto a nitrocellulose membrane. After drying, nonspecific binding sites were blocked with 5% skim milk in PBS supplemented with 0.05% Tween 20 (PBS-T) for 1 h at RT on an orbital shaker. Afterwards, primary antibody incubation was carried out with a 1:10,000 dilution of the anti-OpcA antibody for 30 min followed by three washing steps with PBS-T (5 min each). For secondary antibody incubation, an anti-mouse IgG antibody, conjugated to horseradish peroxidase (HRP), was used with a 1:1000 dilution in 5% skim milk in PBS-T. After 30 min incubation at RT, the membrane was washed again three times with PBS-T and once with PBS for 5 min each. Finally, ECL substrate (BioRad) was added for 1 min and the protein was visualized using the ChemiDoc MP Gel and Blot Imaging System (BioRad).