This streamlined protocol details a workflow to detect and image bacteria in complex tissue samples, from fixing the tissue to staining microbes with fluorescent in situ hybridization.
Measuring the localization of microbes within their in vivo context is an essential step in revealing the functional relationships between the microbiota and the vertebrate gut. The spatial landscape of the gut microbiota is tightly controlled by physical features – intestinal mucus, crypts, and folds – and is affected by host-controlled properties such as pH, oxygen availability, and immune factors. These properties limit the ability of commensal microbes and pathogens alike to colonize the gut stably. At the micron-scale, microbial organization determines the close-range interactions between different microbes as well as the interactions between microbes and their host. These interactions then affect large-scale organ function and host health.
This protocol enables the visualization of the gut microbiota spatial organization from distances between cells to organ-wide scales. The method is based on fixing gut tissues while preserving intestinal structure and mucus properties. The fixed samples are then embedded, sectioned, and stained to highlight specific bacterial species through fluorescence in situ hybridization (FISH). Host features, such as mucus and host cell components, are labeled with fluorescently labeled lectins. Finally, the stained sections are imaged using a confocal microscope utilizing tile-scan imaging at high magnification to bridge the micron to centimeter length scales. This type of imaging can be applied to intestinal sections from animal models and biopsies from human tissues to determine the biogeography of the microbiota in the gut in health and disease.
Microbial visualization techniques have origins that are as old as microbiology itself, when Antonie Van Leeuwenhoek used his microscope to observe bacteria (which he called "animalcules") from tooth plaque and stool in the 17th century. Since then, numerous techniques have been developed to visualize the spatial organization of the consortium of bacteria, fungi, and viruses that live associated with a host-the microbiota1. Elucidating the localization of these microbes is essential to determining their function within their animal host. The biogeography of bacteria is important in commensal and pathogenic taxa alike, as proximity to specific locations (e.g., the epithelium), nutritional substrates, and specific microbes may dictate bacterial production of metabolites underlying interspecies and inter-kingdom interactions.
A key structure separating bacteria and the host tissues in several disparate body sites, such as the oral cavity, intestine, or lung, is the mucus – a host-produced layer that both prevents microbial translocation onto the host epithelial cells and serves as a nutritional resource for the microbiota2,3,4. Characterizing breaches and changes in this barrier is of key importance, leading to mechanistic insight into host-microbiota interactions that would not be obtained by sequencing alone5,6,7. For example, imaging enabled the discovery that antibiotic exposure can disrupt the mucus layer and microbiota organization7,8, and that laxatives may deplete the mucus, correlating with large changes in immune parameters5.
This protocol outlines a general framework for fixing, staining, and imaging the microbiota and the host tissue (Figure 1), built upon the work of Johansson and Hansson9. While this protocol is modeled in the context of intestinal sections, it can be easily adapted to other tissue types. This protocol enables the processing of either experimental animal or human clinical samples; notes for processing both types of samples have been included. In the example presented here, the host epithelium and luminal bacteria have been simultaneously labeled with 4′,6-diamidino-2-phenylindole (DAPI) staining, mucus with the fluorescein (FITC)-conjugated lectin, Ulex europaeus agglutinin I (UEA-1), and a specific bacterial taxon using FISH. FISH probes are usually designed against a taxon's 16S rRNA genes to ensure the high signal from binding a high-copy transcript.
In this example, the probe targets the 16S rRNA of the Muribaculaceae bacterial family (Figure 2). However, the stains are readily substituted with different FISH probes and/or lectins to accommodate the appropriate biological question. Previously validated FISH probes can be found on ProbeBase10, an online resource for rRNA-targeted oligonucleotide probes, or on SILVA11, a ribosomal RNA database. For the design of new probes, the reader may refer to new pipelines such as HiPR-FISH8 or Oligominer12. Using this protocol, it is possible to observe the close packing of bacteria in the intestinal lumen and the different features of intestinal mucus throughout the digestive tract. The workflow described here enables the quantitative analysis of the microbiota in the spatial context of its host environment.
All animal experiments and tissue collection described in this protocol were performed in compliance with the Canadian Council on Animal Care (CCAC) guidelines and were approved by the Animal Care Committee at the University of British Columbia. Germ-free Swiss Webster mice were used. All animals were 8-10 weeks of age, both sexes were used, and all mice were co-housed with at least two mice per cage. Animals were euthanized using carbon dioxide with secondary cervical dislocation or cardiac puncture.
1. Designing an imaging experiment: considerations and sample collection
NOTE: When staining multiple community members, probes can be used combinatorially (e.g., the combination of a family-specific probe and a genus-specific probe) if the fluorophores used do not overlap in excitation and emission spectra (unless imaging on a system with the ability to perform linear unmixing). Additionally, specificity may be demonstrated by including specificity controls/scrambled probes or by competition with non-fluorescent probes.
2. Tissue sample fixation and infiltration
3. Staining bacteria and host features
4. Imaging and image analysis
To investigate the localization of specific gut commensal bacteria in the mouse intestine, germ-free Swiss Webster mice that had been colonized with individual bacterial isolates were used in this study. For this experiment, the bacterial species labeled were i) a human isolate of the Muribaculaceae family5, Muribaculum intestinale, an abundant and prevalent family of bacteria in the murine microbiota15, as well as ii) Bacteroides thetaiotaomicron, a genetically tractable and common gut bacterium. Following two weeks of equilibration, the mice were euthanized, and a segment of the colon containing a fecal pellet was sectioned and fixed in freshly prepared methacarn. After two days of fixation, the sections were dehydrated by processing through methanol, ethanol, and xylenes, infiltrated with paraffin, embedded, and then sectioned to luminal 4 µm slices. These samples were then stained with a FISH probe (3' Cy3-tagged) specific for the Muribaculum isolate, designed using Oligominer software12, and checked for secondary and tertiary structure and minimal non-specific binding using NuPACK and mathFISH16,17. The samples were also counterstained with the Rhodamine-bound lectin UEA-1 (which stains fucosylated glycans in mucus) and DAPI. The stained sections were imaged using a 40x oil objective and super-resolution module.
Figure 1: Workflow of visualization of the microbiota-host interface in the intestine. (A) An illustrated workflow for the pipeline. (B) The two sectioning planes will yield either transverse sections (preferred for this pipeline) or cross-sectional "doughnuts." (C) Embedding tissues of very different thicknesses may result in slices that are only optimal for one tissue or another. Abbreviations: FISH = fluorescence in situ hybridization; DAPI = 4′,6-diamidino-2-phenylindole. Please click here to view a larger version of this figure.
Figure 2: Imaging the colon shows the localization of Muribaculum intestinale relative to mucus in a gnotobiotic mouse model. Sections were stained with DAPI (blue), Muribaculaceae FISH probe (red), and UEA-1 (green). (A) Distal colon of mouse mono-colonized with Muribaculum intestinale and (B) distal colon of mouse bi-colonized with Muribaculum intestinale and Bacteroides thetaiotaomicron. Scale bar = 20 µm. (C and D) Cy3-FISH (left), DAPI (middle), and combined Cy3-FISH + DAPI channels for luminal inset portions of (A) and (B), respectively. Scale bar = 5 µm. In (A) and (C), all bacteria-shaped and bacteria-sized DAPI signals are labeled with Cy3 (single arrowheads), and in (B) and (D), in addition to these Cy3- and DAPI-double-positive cells (single arrowheads), there are DAPI-stained bacterial cells that are Cy3-negative (double arrowheads), as expected for mono- and bi-colonization states. With the exception of longer filamentous bacteria, larger (>4 µm in length) DAPI-positive structures (blunt arrows) are plant material or nuclei from host cells. Abbreviations: FISH = fluorescence in situ hybridization; DAPI = 4′,6-diamidino-2-phenylindole. Please click here to view a larger version of this figure.
Figure 3: Common problems. (A) Shallow sectioning will not provide luminal slices of the intestine. (Left) An intestinal segment that has been sectioned at a depth that provides a view of the lumen as well as longitudinal views of the epithelium. (Right) An intestinal segment that has been sectioned too shallowly, revealing only cross-sections of crypts and no constant mucus layer or bacteria. (B) Uneven tissue/coverslip coverage may result in blurry and unevenly illuminated tile-scans. A tile-scan was set up with positions that were out of focus (top right corner). (C) Example of normal background (left) and high background (right). In this example, for the DAPI signal-note the signal coming from the epithelium, outside the nuclei. All scale bars = 20 µm. Abbreviation: DAPI = 4′,6-diamidino-2-phenylindole. Please click here to view a larger version of this figure.
The protocol described above provides a reproducible method to visualize the host-microbiota interface. These assays have considerably benefited from protocol development, starting from the optimization of FISH labeling18 to mucus preservation and imaging9. Fixation and embedding also provide useful storage of samples; furthermore, paraffin-infiltrated cassettes can be mailed without any restrictions, as the samples are completely fixed and inert.
Significance and alternative methods
The combination of FISH and mucus staining enables the analysis of microbiota composition at specific tissue locations and the interaction between individual bacteria and the host. Alternative techniques that involve the analysis of specific sites within the microbiota are usually unable to explore the single-cell nature of these interactions, such as in the case of laser capture microdissection coupled with sequencing19. However, sequencing-based techniques have the advantage of being able to capture the genetic make-up of the microbiota at a finer level and more broadly than 16S rRNA probes.
A pitfall of the protocol presented is that it provides a single-time snapshot of the microbiota in relation to the host. For real-time imaging in live animals, special imaging techniques are required to facilitate the acquisition of a fluorescence signal in deep tissue (e.g., intravital two-photon microscopy and light-sheet fluorescence microscopy20,21). In these methods, the model organism is either colonized with bacteria that are stained with molecules that bind to their envelope20 or genetically modified bacteria that harbor fluorescent proteins21. In the latter case, the maturation of fluorescence proteins is a key issue as most standard fluorescent proteins require oxygen to emit light. Therefore, model organisms that have at least nanaerobic environments (nanomolar concentration of oxygen) are required, such as the aerobic zebrafish gut21.
In this protocol, methanol-Carnoy is used as a fixative instead of paraformaldehyde or formalin because it does not contain water. This prevents hydration, dehydration, and collapse of the mucus layer during processing and facilitates accurate measurements of mucus thickness. While methacarn fixation and paraffin embedding have become a standard in the field, different fixatives and embedding techniques have been investigated to optimize mucus preservation9,22, with some studies indicating that resins may be superior to paraffin embedding22. Another important limitation to paraffin embedding and methacarn fixation is the loss of fluorescent protein fluorescence, which can be avoided by using alternative fixatives (e.g., formalin or paraformaldehyde (PFA)) or modified embedding techniques23. Each fixative has benefits and drawbacks, and the choice of the fixative depends on the imaging priorities. Imaging modalities can be combined if biological replicates are fixed in either PFA and methacarn and images are obtained from both samples.
FISH has been combined with immunofluorescence in isolated instances with specific anti-Muc2C3 rabbit polyclonal antibodies9,24. These results have not been widely reproduced with other Muc2 antibodies, indicating that the choice of antibodies may be critical in the success of the FISH-immunofluorescence combination. The conditions for successful antibody staining for a given antibody may not permit retention of FISH staining.
Troubleshooting
In this protocol, sample collection, sectioning, and staining represent critical steps to prevent the creation of imaging artifacts. For example, pressing on the tissue upon collection and before embedding can lead to mislocalization of the microbiota and affect mucus quality. Another important consideration is to avoid combining segments of very different thicknesses in a single cassette, such as a relatively empty piece of the small intestine and a pellet within the distal colon. The different thicknesses lead to difficulties in imaging: after embedding, transverse sectioning at a specific depth for one segment may be at the wrong depth for imaging for the other one (e.g., the lumen will be at different depths for each tissue) (Figure 1B,C and Figure 3A). Furthermore, for imaging animal model stool pellets, they may be wrapped in peritoneal membrane24. In this technique, a small section of freshly isolated peritoneum is gently folded around the stool and preserves the structure of the pellet during the wash steps outlined in the protocol.
Unlike processing animal tissue with contents, imaging human intestinal samples presents some challenges. Specifically, luminal contents and mucosal lining will likely be disrupted by the colonoscopy bowel preparation usually performed before intestinal biopsies or resection procedures. Additionally, when biopsies are collected, it is helpful to note the tissue orientation to aid in the identification of specific features (e.g., lumen vs. submucosa) in the absence of intestinal contents. Pre-embedding with 3% agar and/or agar:gelatin mixtures may be useful in maintaining tissue polarity25; however, there are issues with the dehydration and sectioning of agar, as reported in the literature, and processing times may need to be altered.
A key aspect to achieving good FISH staining comes from adjusting the formamide concentration for specific FISH probes. Formamide will control the hybridization stringency of FISH probes to their targets. It is important to ensure that a) the proper formamide concentration is chosen for staining a given community, and b) that a suitable formamide concentration is even possible for the probe combination that is being utilized-this may affect the optimal probe choice when utilizing multiple probes. Websites such as mathFISH (http://mathfish.cee.wisc.edu/) may help with calculating the proper formamide concentrations for a given probe. In the absence of information about appropriate formamide concentrations, this protocol can be tested with 0% and 10% formamide.
Sectioning the embedded samples requires cutting the block to a sufficient depth to obtain luminal slices, as sectioning too shallowly into a block will result in a cross-sectional view of the villi/crypts with no luminal contents (Figure 3A). Stain the samples soon after sectioning for optimal FISH signal, and use fresh DAPI (or DAPI stored at -20 °C) to resolve background issues (Figure 3C). FISH staining of sections that are more than a month-old may result in poorer/inconsistent staining.
If the tissue or coverslip is not perfectly flat with respect to the slide, the sample may not be in focus at every position within a tile scan (Figure 3B), leading to wasted effort and potential photobleaching. This can be solved by taking smaller tile scans in which all positions have been verified to be in focus. Sectioning with dull blades can also lead to the detachment of the mucus and luminal contents, which appear as large dark areas while imaging. Finally, the evaporation of the solution or bubbles during the hybridization step can lead to uneven staining of the FISH probes in the tissue.
Another critical parameter that affects the efficiency of the FISH probe binding is the competition between different probes. A useful check to verify that the FISH probe is labeling bacteria is to examine the colocalization of the signal from the FISH probe with DAPI (Figure 2C,D). In samples that require the use of multiple rRNA probes, adjusting parameters such as hybridization temperature, probe concentration, and formamide becomes critical to ensure that the probes are binding to the correct species efficiently18.
Notes about imaging
FISH-stained bacteria are best visualized using a confocal microscope and an objective of at least 40x magnification. While 63x magnification is preferred to image bacteria at the single-cell level, 40x magnification can also be used by maximizing the digital zoom or employing a super-resolution system. Systems equipped with super-resolution capabilities may also provide sub-cellular resolution of individual bacterial cells. Lower magnification objectives may be utilized to get a general sense of bacterial localization and mucus thickness.
Acquiring 16-bit images instead of 8-bit images is also highly recommended, as the higher dynamic range will help to capture the wide range of the DAPI signal from the extremely bright nuclei of the epithelium and the relatively weak signal of luminal bacteria. Aside from using the imaging setup described above, an important imaging consideration is the choice of fluorophores. For best separation between bacterial types, ensure that the fluorophores used do not overlap in excitation and emission spectra (unless imaging on a system with the ability to perform linear unmixing). Additionally, probes can be used in combination (e.g., the combination of a family-specific probe and a genus-specific probe) to identify additional community members. If using linear unmixing or unvalidated probes, it is essential to test the accuracy and level of FISH staining on pure cultures8,14.
Once the images are collected, multiple tools are available for the quantitative analysis of the host-microbiota interface, such as BacSpace26, HiPR-FISH8, and other segmentation tools27. BacSpace is a MATLAB software that allows the segmentation of the tissue and luminal components and the removal of plant material from images26. The program also provides analysis of mucus thickness in 2D and quantification of spatial distribution based on pixel distances in different channels26. Conversely, the HiPR-FISH Image Analysis software allows the segmentation of FISH-stained cells8. Finally, other bacterial segmentation tools that have been optimized for in vitro experiments27 could also be adapted to this application.
Conclusion
In situ imaging enables the quantitative analysis of individual microbial taxa in the context of other community members and the various microenvironments created by diet, mucus, and host epithelial crypts. The interaction between organisms dictates the biology of host-associated microbial systems and provides an essential analysis of change in these structures during perturbation that has implications for health. Notably, the ability to image the microbiota in situ through the protocol described above provides an incredible window into the biogeography of the microbiota in relation to the animal host.
The authors have nothing to disclose.
The authors would like to thank Dr. Kristen Earle for invaluable technique development, Amber Hann for troubleshooting help, Dr. Jen Nguyen and Deanna Pepin for critical review of the manuscript, and Dr. Hesham Soliman and the Rossi lab at the University of British Columbia for providing microscope access. The authors acknowledge support from CIHR Team Grant: Canadian Microbiome Initiative 2 (C.T.), Crohn's and Colitis Canada (to C.T. and K.M.N.), CIFAR (to C.T. and K.M.N.), Michael Smith Foundation for Health Research Scholar Award (to C.T.), the Weston Foundation: Weston Family Microbiome Initiative (to C.T.), Paul Allen Distinguished Investigator Award (to C.T.).
Aluminum foil | |||
Chloroform | Fisher | C298-500 | Toxic |
Coplin jar | Fisher | 08-813E | |
Cover glass, 22 × 22 mm, no #1 | VWR | CA48366-227-1 | |
DAPI | Sigma-Aldrich | D9542 | Resuspended to 5 mg/mL |
Empty pipette tip box(es) | To melt paraffin | ||
Ethanol, anhydrous, molecular grade | — | — | |
FISH probes | |||
FISH hybridization solution | — | — | 20 mM Tris–HCl pH 7.4, 0.9 M NaCl, 0.01% (w/v) SDS in nuclease-free water. Optional addition of 5–50% (v/v) formamide |
FISH washing buffer | — | — | 20 mM Tris–HCl pH 7.4, 0.9 M NaCl |
Fluorescently-labeled Ulex Europaeus Agglutinin (UEA-1) lectin | Vector Laboratories | FL-1061 (Fluorescein-labeled) or RL-1062-2 (Rhodamine-labeled) | UEA-1 labels fucosylated glycans, so it is not appropriate for germ-free mice or for samples from fucosyltransferase 2 (FUT2)-deficient hosts. |
Forceps | |||
Formamide | Sigma-Aldrich | F9037 | |
Fumehood | |||
Glacial Acetic Acid | Fisher | A38-212 | Corrosive and flammable |
HybriSlip flexible plastic cover slips, 22 x 22 mm × 0.25 mm | Sigma Aldrich | GBL722222 | for FISH hybridization steps, can substitute glass coverslips |
ImmEdge Hydrophobic Barrier PAP Pen | Vector Laboratories | H-4000 | |
Kimwipes | |||
Methanol | Fisher | A412 | Toxic and flammable |
Microtome | for sectioning | ||
Multipurpose Specimen Storage Containers, 473 mL | Fisher | 14-955-117A | Can substitute with a glass or polyethylene container of your choice. Methacarn is not compatible with polystyrene or metal. |
Nail Polish | Electron Microscopy Sciences | 72180 | Any nail polish should work |
Oven | Necessary range: 45-60 °C | ||
Paraffin: Leica Paraplast | Leica | 39601006 | |
PBS, Phosphate-buffered Saline (10x solution) | Fisher | BP3991 | Dilute to 1x for protocol |
Sodium chloride | Fisher | S271 | |
Sodium Dodecyl Sulfate (SDS), 20% Solution, RNase-free | Invitrogen | AM9820 | |
Tissue-Loc HistoScreen Cassettes | Thermo Scientific | CA83009-999 | |
Trizma hydrochloride solution, 1M Tris-HCl, pH 7.4 | Sigma-Aldrich | T2663-1L | |
Vectashield | Vector Laboratories | H-1000 | |
Water, nuclease-free | Fisher | BP2484100 | |
Xylenes | Fisher | X3P-1GAL | Toxic and flammable |