I am posting a blog post that was written by Benedikt Geier, a Ph.D. candidate who just handed in his Ph.D. thesis at the Max Planck Institute for Marine Microbiology in Bremen, Germany. In my eyes, these last couple of months before submitting a doctoral thesis are the hardest, and that’s why I am even more impressed with Benedikt’s summary of one of his Ph.D. chapters that was recently published. When I received the first draft of this post, I literally got goosebumps and was holding my breath while reading it. So spectacular. Enjoy!
Naturalists like Amalie Dietrich, Marian Farquharson, Alfred R. Wallace, or Charles Darwin visualized the form and function of animals, plants, and environments by drawing the organisms and their behaviour. Today, we know that the natural world consists of far more organisms and processes than that we can see by eye and that must be considered to understand biological systems. A perfect example is bacteria. Every animal and plant is not only in physical contact with bacteria but also interacts with them through a hidden chemical language. The aim of our study and my PhD in general was to develop approaches to visualize and understand the molecular interactions between animals and almost every animal or plant on our planet.
A simple example for capturing a biological process could be to film the symbiosis between anemones and clownfish in an aquarium. It would show how both partners meet and then establish and maintain their symbiotic interaction. For instance, their behaviour would show that the anemone serves as protection and food source for the clownfish and in return, the clownfish ventilates and frees the anemone from parasites. However, to further explain these processes, we have to look behind the scenes at a microbial and chemical level to understand how the clownfish can avoid the stinging cells of the anemone and how the anemone recognizes the clownfish itself (Roux, et al. 2019).
Visualizing such basic processes for the interactions between animals and bacteria presents a very difficult challenge. Both partners interact on an extremely small scale and mainly through the exchange of molecules, which both cannot be captured with standard imaging techniques. To top it off, the majority of animal–microbe symbioses cannot be maintained in an aquarium, terrarium, or lab culture. The main reason is that they depend on interactions with other organisms and the environment, from tropical rain forests to the bottom of the ocean. Altogether, these interactions are nearly impossible to reproduce.
Over the past four years my colleagues and I tackled this challenge by visualizing the symbiotic interactions of a non-culturable animal–bacteria symbiosis from the deep sea. We wanted to understand the symbiotic interactions between deep-sea mussels from the genus Bathymodiolus and their intracellular symbiotic bacteria. Bathymodiolus mussels live at hydrothermal vents and cold seeps (Dubilier, et al. 2008; see Figure 1). The mussels’ survival strategy is based on chemosynthesis. Independent of photosynthetic products, they can survive in the deep sea by feeding on organic molecules that are synthesized by their chemoautotrophic bacterial symbionts and by directly digesting the bacteria intracellularly (Zheng, et al. 2017). These chemosynthetic bacteria oxidize chemicals from the vent fluids and mainly consist of two genotypes that we refer to as methane oxidizer (MOX) and sulfur oxidizer (SOX), depending on their substrate. Both MOX and SOX colonize the epithelial cells of the gill (Figures 2 and 3), called bacteriocytes, which turn the gill into a respiratory and symbiotic organ at once.
Unlike the phenotype of an animal, which can have a specific anatomy, coloration or behaviour, the phenotype of bacteria is more difficult to visualize. Although bacteria occur in many shapes and sizes, the most precise representation of their phenotype that describes their composition and behaviour are the molecules bacteria produce. One example is their so-called metabolic phenotype (Phelan, et al. 2011). Metabolites are small molecules below 1500 Da and are often involved in communication and defence or act as nutrients that are exchanged between hosts and their microbes (Brunetti, et al. 2018). The Bathymodiolus symbiosis has been extensively studied at a genomic, transcriptomic and proteomic level from bulk measurements, but how metabolites are distributed between both symbiotic partners was completely unknown before I started my PhD.
Compared to the clown fish–anemone symbiosis that we can observe by eye, we faced two substantial challenges in trying to visualize the metabolic interactions between host and bacteria of the Bathymodiolus symbiosis:
- How can we visualize the intracellular bacteria?
- Once we can see the bacteria, how can we document their “interactions” with the animal host?
In addition, we had to resolve both challenges under the prerequisite that we are currently not able culture host and symbionts together or individually, which did not make things easier.
Fortunately, we did not have to start completely from scratch and could build upon previously developed methods, such as fluorescence in situ hybridization (FISH) to address our first challenge of “how to visualize the intracellular bacteria” (Figure 2). FISH has become one of the state-of-the-art techniques for microbial ecologists to visualize bacteria based on their taxonomic identity within complex communities (Amann and Fuchs 2008). In a classical FISH approach, fluorescently labelled oligonucleotide probes target unique regions of the conserved 16S ribosomal RNA of bacteria. Using different fluorescent labels, individual bacterial genotypes can then be imaged and distinguished with fluorescence microscopy. Using FISH on tissue sections of the gill allowed us to resolve our first question and visualize the bacteria inside the bacteriocytes within the animal tissue.
To resolve our second major challenge of “how to document the interactions between animal and bacteria” was the most difficult one. Unlike animal–animal interactions, like clown fish and anemone, we could not simply observe the behaviour between mussel and bacteria. For one reason, because they were dead. We had to fix the animal with its bacteria on board of the research vessel, immediately after they were brought back on deck. Without preservation, the animals from these extreme depths would slowly lose their symbiotic bacteria and die without additional food sources. Another reason why we would not be able to observe the interactions between the animal and its microbes is that the majority of their interactions takes place through exchange of metabolites. Even if we could see the living bacteria within the host cells, for example through a phase-contrast microscope, all we might see would be the bacteria, wiggling around, divide or “die” within the bacteriocytes. Therefore, we had to not only visualize the bacteria and host cells but also the metabolites of both to study their metabolic interactions.
Fortunately, there are modern mass spectrometers, so-called molecular microscopes, which allow us to image metabolite distributions even from fixed tissues (Watrous and Dorrestein 2011). Our technique of choice was mass-spectrometry imaging (MSI). MSI can generate metabolite images by scanning across a sample surface similar to a scanning electron microscope. Instead of electrons, ionized molecules are detected by a mass spectrometer in each spot, resulting in the individual pixels of a metabolite image. Among different MSI approaches, we chose matrix-assisted laser/desorption (MALDI)-MSI, which is particularly powerful for imaging and identifying metabolites from tissue sections of complex samples, ranging from mixed microbial communities to plant and animal tissues. Over the last decade, substantial technical advances have improved the spatial resolution and sensitivity of MALDI-MSI, so that hundreds to thousands of metabolites can now be mapped at the micrometer scale (Kompauer, et al. 2017; Niehaus, et al. 2019).
Conventionally, samples for MALDI-MSI are snap-frozen. This has the advantage that all metabolic processes are stopped simultaneously and the spatial distribution of all metabolites, even within body fluids is preserved. After the frozen samples were transported back to our institute we sectioned the tissue with a cryotome, a microtome within a cooled chamber. This way we could section the frozen sample into tens of micrometer thin tissue sections and preserve the anatomy and chemistry for MALDI-MSI.
We developed a protocol to combine MALDI-MSI and FISH to reveal the metabolite distributions and community composition from a single tissue section of the host animal. This combined approach was essential to determine which metabolites were produced by bacteriocytes or host cells that did not contain symbiotic bacteria. Additionally, we optimized this approach for the application at micrometer-scales to resolve the metabolome and community composition of individual host cells. The detailed “how to” protocol can be found on protocols.io and in our publication on ‘Spatial metabolomics of in situ host–microbe interactions at the micrometre scale’ (Geier, et al. 2020). Combining MALDI-MSI and FISH on the same tissue section allowed us to create a snapshot of the symbiotic bacteria within the host tissue, the host cells and the metabolites, produced by both partners.
Returning to the clown fish and anemone example that one could observe over time, one question remained: “How could we claim to visualize an active process like an interaction or activity between organisms if they were dead?” Imaging the individual metabolites provided us with one substantial advantage over conventional microscopy techniques. The metabolite images not only showed us specific structures or cells, but also provided qualitative information of which metabolites were present in the tissues and cells at the point in time when the animal was frozen. Considering that each metabolite has a specific function, in combination with the location of the host and symbiont cells from the FISH images, we could draw conclusions how the metabolites could be used by both partners in the symbiosis. Therefore, we did not simply visualize cells and molecules, but through spatial correlations and superimposing the data of both techniques, could link the taxonomic identity and metabolic activity of both partners.
Although both techniques could be applied separately, only once we acquired the correlative MALDI-MSI and FISH data, which means from the same tissue surface, we could interpret the images. The type of imaging data that is generated with MALDI-MSI and FISH is comparable to visualizations that we can see every day on the weather forecast. For example, when temperatures are visualized as a heatmap that is superimposed onto a geographical map of a country. The geographical map represents the FISH image and the heat map showing temperature would correspond to the metabolite images. On the one hand, if we only had a geographical map, we would exactly know the position of streets, rivers, mountains, etc., but have no knowledge of the location’s condition. On the other hand, if we only have a heatmap of the temperatures, we would see changes across the map, but could not link the temperatures to the locations that we know. Only by superimposing both datasets (see Figure 3), we can immediately understand changes between specific locations across the map.
This visualization allowed us to show that symbiotic bacteria of the same genotype, produced different membrane lipids in different bacteriocytes, depending on their position in the gills’ filaments. For us this finding was interesting because it showed that the metabolic phenotype of the microbes is much more diverse than what we can resolve at a genomic level. In conclusion, the combination of MALDI-MSI and FISH did not only reveal new insights into the Bathymodiolus symbiosis, but could provide a tool to address one of the biggest challenges in molecular (microbial) ecology: to link the metabolic activity of individual cells to their taxonomic identity in complex samples taken directly from their natural habitat.
Training with one of the most complicated systems in terms of samples has paid off. Both approaches, MALDI-MSI and FISH, have the substantial advantage that they can be applied in situ to snap-frozen samples. This means that a scientist can go into the field, collect their favourite organism, freeze it in liquid nitrogen, and create a snapshot of the animal’s histology, associated microbes, and metabolites at the time of sampling. Considering the vast majority of uncultured host–microbe symbioses, our approach could be applied to most of them including bacteria, fungi, plants or animals that live in mutualistic, commensal or pathogenic relationships. All that is needed is a frozen sample.
Of course, as exciting this combined MALDI-MSI and FISH approach is, it relies on the integration with other molecular omics techniques. For instance, combining our spatial metabolomics pipeline with one of the emerging spatial transcriptomics pipelines (Rodriques, et al. 2019; Vickovic, et al. 2019) could allow us to link the relative metabolite abundances to differential expression patterns. With MALDI-FISH, our aim was to contribute a “molecular pencil and paper” to visualize and document the metabolic interactions in animal–bacteria symbioses from a new perspective that provides a link between form and function.
References:
Amann, R., & Fuchs, B. M. (2008). Single-cell identification in microbial communities by improved fluorescence in situ hybridization techniques. Nature Reviews Microbiology, 6(5), 339-348. doi:10.1038/nrmicro1888
Brunetti, A. E., Carnevale Neto, F., Vera, M. C., Taboada, C., Pavarini, D. P., Bauermeister, A., & Lopes, N. P. (2018). An integrative omics perspective for the analysis of chemical signals in ecological interactions. Chemical Society Reviews, 47(5), 1574-1591. doi:10.1039/c7cs00368d
Dubilier, N., Bergin, C., & Lott, C. (2008). Symbiotic diversity in marine animals: the art of harnessing chemosynthesis. Nature Reviews Microbiology, 6(10), 725-740. doi:10.1038/Nrmicro1992
Geier, B., Sogin, E. M., Michellod, D., Janda, M., Kompauer, M., Spengler, B., . . . Liebeke, M. (2020). Spatial metabolomics of in situ host-microbe interactions at the micrometre scale. Nature Microbiology. doi:10.1038/s41564-019-0664-6
Kompauer, M., Heiles, S., & Spengler, B. (2017). Atmospheric pressure MALDI mass spectrometry imaging of tissues and cells at 1.4-mu m lateral resolution. Nature Methods, 14(1), 90-96. doi:10.1038/Nmeth.4071
Niehaus, M., Soltwisch, J., Belov, M. E., & Dreisewerd, K. (2019). Transmission-mode MALDI-2 mass spectrometry imaging of cells and tissues at subcellular resolution. Nature Methods, 16(9), 925-931. doi:10.1038/s41592-019-0536-2
Phelan, V. V., Liu, W. T., Pogliano, K., & Dorrestein, P. C. (2011). Microbial metabolic exchange–the chemotype-to-phenotype link. Nature Chemical Biology, 8(1), 26-35. doi:10.1038/nchembio.739
Rodriques, S. G., Stickels, R. R., Goeva, A., Martin, C. A., Murray, E., Vanderburg, C. R., . . . Macosko, E. Z. (2019). Slide-seq: A scalable technology for measuring genome-wide expression at high spatial resolution. Science, 363(6434), 1463-1467. doi:10.1126/science.aaw1219
Roux, N., Lami, R., Salis, P., Magre, K., Romans, P., Masanet, P., . . . Laudet, V. (2019). Sea anemone and clownfish microbiota diversity and variation during the initial steps of symbiosis. Scientific Reports, 9(1), 19491. doi:10.1038/s41598-019-55756-w
Vickovic, S., Eraslan, G., Salmen, F., Klughammer, J., Stenbeck, L., Schapiro, D., . . . Stahl, P. L. (2019). High-definition spatial transcriptomics for in situ tissue profiling. Nature Methods, 16(10), 987-990. doi:10.1038/s41592-019-0548-y
Watrous, J. D., & Dorrestein, P. C. (2011). Imaging mass spectrometry in microbiology. Nature Reviews Microbiology, 9(9), 683-694. doi:10.1038/nrmicro2634
Zheng, P., Wang, M. X., Li, C. L., Sun, X. Q., Wang, X. C., Sun, Y., & Sun, S. (2017). Insights into deep-sea adaptations and host-symbiont interactions: A comparative transcriptome study on Bathymodiolus mussels and their coastal relatives. Molecular Ecology, 26(19), 5133-5148. doi:10.1111/mec.14160