Visualization of endogenous gut bacteria in Drosophila melanogaster using fluorescence in situ hybridization

All metazoans are colonized by a complex and diverse set of microorganisms. The microbes colonize all parts of the body and are especially abundant in the gastrointestinal tract, where they constitute the gut microbiome. The fruit fly Drosophila melanogaster turned out to be an exquisite model organism to functionally test the importance of an intact gut microbiome. Still, however, fundamental questions remain unanswered. For example, it is unknown whether a fine-tuned regionalization of the gut microbiome exists and how such a spatial organization could be established. In order to pave the way for answering this question, we generated an optimized and adapted fluorescence in situ hybridization (FISH) protocol. We focused on the detection of the two major Drosophila gut microbiome constituting bacteria genera: Acetobacter and Lactobacillus. FISH allows to detect the bacteria in situ and thus to investigate their spatial localization in respect to the host as well as to other microbiome members. We demonstrate the applicability of the protocol using a diverse set of sample types.


Introduction
Microbiome research has become an emerging and central field in biology and biomedicine. This is based on the fact that all metazoans are colonized by billions of microorganisms, which deeply impact the overall physiology of the host. The human body, for example, consists of 3 x 10 13 cells, but harbors another 4 x 10 13 microorganisms of enormous diversity [1]. The majority of bacteria colonize the gut, where they not only facilitate nutrient access and uptake, but also e.g. protect from pathogens or are important to maintain the intestinal barrier function of the gut [2]. Through a variety of enzymes and metabolic pathways the gut microorganisms are able to metabolize complex molecules and thus make them available for the host [3]. Further, certain metabolites which cannot be synthesized by the host itself can be provided by its gut microbiome [4]. A dysbiosis of the gut microbiome accordingly results in serious health problems and is also associated with the occurrence and progression of severe chronic diseases such as obesity, inflammatory bowel disease, diabetes mellitus, nonalcoholic fatty liver disease (NAFLD) and hepatocellular carcinoma [5][6][7]. Many gut microbiome members cannot be cultured in the lab. Yet, investigations targeting the functional importance of the gut microbiome require the ability to query the presence of distinct bacterial species. Thus, molecular biology techniques including (quantitative) PCR and next generation sequencing are routinely used to probe the microbiome composition under various conditions. A shortcoming of these methods is that they lack spatial information both in relation to the host as well as to other microorganisms. The fluorescence in situ hybridization (FISH) method overcomes this shortcoming and was previously used to detect e.g. the bacteria in human saliva [8]. The FISH method relies on fluorescently labeled probes, which hybridize to their specific DNA or RNA sequence targets [9,10]. For microbiome applications, those target sequences are usually regions of the 16S rRNA of the given gut bacterium.
Drosophila melanogaster is commonly used to study various aspects of the gut microbiome. So far, however, FISH was not extensively applied to probe for the spatial organization of the bacteria present in the gut. This is intriguing, as the low complexity of the fly microbiome with only 10 to 20 bacterial species should be well suited. While one study applied FISH to investigate the symbiont/pathogen Wolbachia and the gut microbiome member Acetobacter pasteurianus in gnotobiotic animals [11] and one study used the eubacteria probe Eub338 with flies [12], most microbiome studies in Drosophila melanogaster either utilized genetically modified, fluorescent versions of gut bacteria [13], or used tracers to test for e.g. the vitality of the gut bacteria [14]. Those methods have the draw-back that they either require the addition of exogenous, genetically modified bacteria, which could alter their abundance and location in the gut, or the lack of species differentiation. We thus generated an adapted and optimized protocol to add FISH as a method to probe the microbiome composition of Drosophila.

Fluorescence in situ hybridization probes
The used probes for FISH (Table 1) consisted of a generic probe, which should detect all eubacteria (Eub338; S-D-Bact-0338-a-A-18) [15,16], a probe which specifically detects bacteria of the genus Lactobacillus (Lacto722; S-G-Lacb-0722-a-A-25) [16,17], and a third probe which hybridizes with the genus Acetobacter (Aceto). The sequences for the Eub338 and the Lacto722 probes were obtained from Probebase (http://probebase.csb.univie.ac.at/node/8). For the design of the Aceto probe we used the webtool decipher (http://www2.decipher.codes/ DesignProbes.html). As the Eub338 probe generated high levels of background signal in the feces samples (S1A Fig) as well as in the Drosophila larval and adult guts, we only used this probe as a control with the type strain bacteria to check whether the strain-specific Aceto and Lacto722 probes result in distinct hybridization with the corresponding bacteria. All probes were ordered from Biomers (https://www.biomers.net/) as "DOPE" double labeled fluorophore probes (each of the double labeled oligos carries one fluorescent label at the 5'-and 3'-end, respectively).

Fluorescence in situ hybridization of bacterial cell suspensions
The basis for the herein established method was a FISH protocol described by Valm et al., (2011). To establish the protocol and to test for specificity of the probes, bacterial cells of type strains ( Table 2) grown as overnight cultures were used.

Fluorescence in situ hybridization of isolated Drosophila gut bacteria
For the isolation of the Drosophila gut bacteria about 30 six-day old male flies were kept at -20˚C for 10 minutes. Afterwards the flies were put in a reaction tube and washed with 1 ml of a 10% bleach solution followed by two additional washing steps with 1 ml 70% EtOH and 1 ml sterile PBS to remove external bacteria. Afterwards, the whole flies were homogenized in 200 μl sterile PBS using a tissue grinder (Kontes TM Pellet Pestle TM Motor, Kimble Chase). , we washed off the colonies with the corresponding liquid medium (thus, e.g. MRS liquid medium was used to wash off the colonies from the MRS agar plates and corresponding procedures were performed for the ACE and YPM samples). These resuspended bacteria colonies were then incubated over night at 28˚C. Subsequently, the cells were processed as described in the section "Fluorescence in situ hybridization of bacterial cell suspensions". Samples were imaged with an Operetta CLS high content screening microscope (Perkin Elmer) with the settings provided in Table 1 and a 40x air objective.

Fluorescence in situ hybridization of Drosophila feces
To detect and visualize bacteria in the feces of adult Drosophila, about 30 flies of mixed age were transferred to new food vials which were then covered with a microscope glass slide. The flies were kept in the vial and on the glass slide for about 24 hours, after which the glass slide was removed. The feces, which were visible on the slide, were dried at 50˚C on a thermal block for 2 hours. All further steps were performed directly on the glass slide. The feces were fixed with 500 μL 5% paraformaldehyde solution for 15 minutes and were afterwards washed with PBS and treated with 10 mg/mL lysozyme in PBS for 15 minutes at 37˚C. Then the hybridization buffer (including 40% formamide and 4 μM probe each) was layered over the feces and incubated for three hours at 46˚C. The feces were afterwards washed in wash buffer 2 for 30 minutes at 48˚C and finally washed with resuspension buffer. The buffer was removed, and the feces left to dry completely before they were covered with 30 μl Prolong Gold Antifade reagent and a cover slip. Samples were imaged with an Operetta CLS high content screening microscope (Perkin Elmer) with the settings provided in Table 1 and a 40x air objective.

Generation of axenic Drosophila animals
The protocol to generate axenic animals was adapted from [18]. In brief, adult Drosophila animals were placed in a fly cage, which was covered with an apple juice agar plate (for approximately 20 plates: 20 g agar, 8.5 g sucrose, 500 ml dH 2 O, 170 ml naturally cloudy apple juice, 10 ml Nipagin solution [700 ml 96% EtOH p.a., 300 ml dH 2 O, 100 g Nipagin]) with a dab of fresh yeast. After 24 h, the apple juice agar plate was replaced by a fresh one. After another 24 h, Drosophila embryos were washed off of the agar plate using embryowash (1 ml Triton X-100, 14 g NaCl, 200 ml dH 2 O, sterile filtration) and collected in a 50 ml centrifugation tube. Afterwards, the embryos were dechorionated using 50% bleach for 2 minutes, followed by a washing step with dH 2 O. The tube was centrifuged for 15 seconds at 300 rpm to pellet the embryos. Finally, the embryos were washed with 70% EtOH p.a., followed by another washing step with dH 2 O. Embryos were resuspended in 200 μl embryowash and 20 μl were pipetted onto axenic food vials [19].

Genomic DNA (gDNA) extraction and PCR-based validation of axenic state
Genomic DNA (gDNA) was isolated from single adult Drosophila flies which were either conventionally reared or axenic. Single flies were placed in an Eppendorf tube and homogenized in 50 μl squishing buffer (10 mM Tris-HCl [pH 8.2], 1 mM EDTA, 25 mM NaCl, 200 μg/ml proteinase K). The homogenate was incubated for 30 minutes at 37˚C prior to heat inactivation of the proteinase K at 95˚C for 2 minutes. The homogenate was centrifuged for 1 minute at 13.000 rpm and the supernatant was transferred to a new reaction tube.

Fluorescence in situ hybridization of Drosophila larval and adult guts
Drosophila white [-] flies were kept on a diet containing 8% yeast extract (to avoid yeast-based autofluorescence signal), 8% cornmeal, 0.8% agar, 0.4% propionic acid and 0.15% nipagin. For the larval samples, wandering third instar larvae were isolated and dissected. For the adult samples, virgin male flies were collected and aged for six days prior to the experiment. The flies were then starved overnight on 0.5% agarose and then refed with bacteria-containing yeast extract/cornmeal diet to motivate food uptake for about 4 hours prior to dissection. Axenic Drosophila animals were kept on the same yeast extract / cornmeal-based food, which only contained in addition antibiotics (ampicillin, erythromycin, kanamycin, and tetracyclin [50 μg/ml as final concentration/antibiotic]) and processed identically to the conventionally reared flies.
Guts of wandering third instar larvae and six days old adult flies were dissected in ice-cold PBS and fixed in Carnoy's solution (60% ethanol, 30% chloroform, 10% acetic acid) for 5 minutes. The guts were washed twice in PBS and treated with 10 mg/mL lysozyme in PBS for 15 minutes at 37˚C. As controls, we included experiments where either any probe was omitted in the subsequent hybridization (S3 Fig) or where the guts were pre-treated with 50μg / mL RNase A for 30 minutes prior to the hybridization with the probes (S4 Fig). The tissues were subsequently incubated in hybridization buffer containing 40% formamide with 4 μM of each probe. Hybridization was allowed to happen for about 16 hours at 46˚C. The guts were subsequently washed in 500 μl wash buffer 2 for 30 minutes at 48˚C. Then the wash buffer 2 was removed and 500 μl resuspension buffer were added. The guts were mounted on microscope slides in 30 μl Prolong Gold Antifade reagent. Samples were imaged with an Operetta CLS high content screening microscope (Perkin Elmer) with the settings provided in Table 1. The overview images of the guts were recorded as tile scans with a 5x air objective. The high-resolution zoom-in images were recorded with a 40x air objective.

Quantitative analysis of microscopy data
During method optimization, we followed the impact of varying parameters (e.g. probe and formamide concentration) on the staining intensity of the type strain stainings by recording microscopic images and quantifying the staining results via image segmentation. An overview of the image segmentation routine carried out with the KNIME data analysis platform and the image segmentation module is found in S5 Fig. We used the TO-PRO-3 DNA stain as general bacterial marker and then calculated the relative staining efficiencies of the given probes in response to the different parameter variations. The analysis pipeline with example images is deposited in the KNIME pipeline hub and can be accessed at the following URL: https://hub. knime.com/matbeller/spaces/Beller-Laboratory/latest/Akhtar_et_al_Bacterial_FISH_ Example_Workflow.

Results
The goal of the present study was to establish a FISH protocol for Drosophila melanogaster. Thus, we started selecting and designing probes for the most abundant bacterial genera living in Drosophila guts: Acetobacter and Lactobacillus. As a control, we used the universal bacterial probe Eub338 (S-D-Bact-0338-a-A-18) (see Table 1), which binds to the 16S rRNA sequence of all eubacteria [15,16]. The Lactobacillus specific Lacto722 (S-G-Lacb-0722-a-A-25) probe was obtained from the probe-database Probebase (http://probebase.csb.univie.ac.at/node/8) and also targets the 16S rRNA sequence as does the Acetobacter-specific probe which we designed using the webtool decipher (http://www2.decipher.codes/DesignProbes.html).

Validation of the FISH probe specificity using bacterial type strains and isolated Drosophila gut bacteria
We started with the protocol optimization by testing several critical parameters of the FISH protocol such as the probe concentration (we used 2 μM, 4 μM, 8 μM and 16 μM), the formamide concentration (10%, 20%, 30% and 40%) as well as different fixation times (15,30,60 and 90 minutes). The various tests were quantified with a plate reader analysis (S1B Fig) and an automated image analysis pipeline (S5 Fig) and resulted in the selection of a probe concentration of 4 μM, a formamide concentration of 40% and 15 minutes of fixation (for details see material and methods). Longer fixation times resulted in a prominent decrease in signal (S1B Fig). In order to increase the signal intensity for the gram-positive Lactobacillus, we also tested whether a treatment with 10 mg/ml lysozyme for 15 minutes at 37˚C could enhance the probe signal, which indeed was the case. Therefore, we decided to include this step in every experiment.
The specificity of the strain-specific probes was first tested on bacterial cultures of E. coli and of various Lactobacillus and Acetobacter type strains (see Table 2). All probes were tested with all type strains individually to test for potential cross-reactivities. Lactobacillus and Acetobacter representatives are typical members of the Drosophila gut microbiome [21,22]. Fig 1 displays exemplary results of FISH stainings with bacterial cells from liquid cultures which were hybridized with each of the three probes.
The universal Eub338 probe hybridized with all tested bacterial species, whereas the Lactobacillus and Acetobacter probes only showed a strong hybridization signal with the specific species (Fig 1). Both genus specific probes only showed minute hybridization signals with E. coli cells. In order to test for a potential bleed-through fluorescence across the channels as well as for a putative background fluorescence of a given bacterial species, we recorded in each single probe staining experiment all three channels with the identical settings (S6 Fig). No bleedthrough signals or prominent background fluorescence signals were present. Thus, our initial tests demonstrate the specificity of the utilized probes and specificity of the detection by the absence of fluorescence bleed-through. In order to further test the applicability of the probes, and to make sure that the probes work with the bacteria present in the Drosophila gut, we tested them on unpurified bacterial cultures (Fig 2) that we previously isolated from adult Drosophila flies using Lactobacillus (MRS) and Acetobacter (YPM, ACE) enriching agar plates (see material and methods). Here, the bacteria were isolated from whole flies (see material and methods for details) and the mix of bacteria which grew on the respective enrichment media agar plate was used for the staining experiment. This way, we could make sure that the probes do not only detect the type strain representatives (Fig 1), but also the bacteria present in Drosophila (Fig 2).
Using the unpurified strains obtained from YPM and ACE plates, the signal of the Acetoprobe was much stronger as compared to the Lacto722 probe (Fig 2). FISH staining with the MRS isolates resulted in the juxtaposed results, as expected. The eubacteria Eub338 probe resulted in a prominent signal in all isolates and all bacteria could be detected. Thus, our probes in combination with the protocol used by us are capable to detect Drosophila resident bacteria.

FISH with Drosophila feces
Next, we tested whether it is possible to visualize gut bacteria in feces of adult Drosophila flies. The generic Eub338 probe hybridized with many bacteria present in the feces samples. However, the Eub338 probe also resulted in high background signals, presumably based on an unspecific binding to food remnants (S1A Fig). Thus, we decided to only move on with the strain-specific Aceto and Lacto722 probes. Fig 3 shows a typical FISH staining of Drosophila feces collected from adult flies (see material and methods). Both the Aceto and the Lacto722 probe hybridized with bacteria dispersed in between remnants of the food, which resulted in some background signal. Based on the morphology and staining properties, however, specific signals for both bacterial genera could be identified (see asterisks and arrowheads in Fig 3).

FISH with larval and adult Drosophila guts
The final aim of establishing the staining protocol was to perform FISH on bacteria localized in the Drosophila gut. In pilot experiments, we realized that a standard paraformaldehyde (PFA)-based fixation, which we used for the isolated bacteria (Fig 2) and the feces (Fig 3)  We first started with FISH stainings on larval guts of wandering third instar animals. The staining procedure resulted in the detection of both Lactobacillus and Acetobacter cells (Fig 4). Especially in the green spectrum, however, a prominent background signal was present, likely originating from food present in the gut lumen as well as tissue autofluorescence. As for the feces samples, a differentiation based on the morphology and size between the bacteria and background signals was nevertheless possible. In order to test further for the specificity of the staining, we also performed control stainings without probes (S3 Fig) or with RNase A pretreatment (S4 Fig) with guts of conventionally reared larvae as well as FISH with guts of axenic larvae (S8 Fig). Across all control stainings a weak and mostly homogeneous background fluorescence was visible which again was most prominent in the green channel. When the control stainings (S3, S4 and S8 Figs) are compared to the FISH stainings performed with the conventionally reared larvae (Fig 4), however, only the latter showed prominent and distinct signals, thus demonstrating specificity of the staining.
As Drosophila larvae are constant feeders, the detection of food-borne bacteria is relatively easy. However, adult Drosophila are intermittent feeders [24] which potentially complicates the timing for the detection of bacteria in the gut. Therefore, we starved the six-day old animals overnight and afterwards put them on food vials containing bacteria for four hours, prior to dissection and analysis of the adult guts. This regimen allowed the detection of both Acetobacter and Lactobacillus in adult fly guts (Fig 5). Axenic animals did not show any clear-cut signals (S9 Fig). Our initial staining results demonstrate specific signals for both Acetobacter and Lactobacillus. Whether a clear-cut regionalization of the gut colonization is present, future experiments need to show.

Discussion
In this study, we describe a method to detect Acetobacter and Lactobacillus, two of the most abundant commensal Drosophila gut bacteria [21,22], in Drosophila larval and adult fly guts as well as in Drosophila feces and for isolated gut bacteria using fluorescence in situ hybridization (FISH). Many studies aiming at the analysis of gut microbiota apply (quantitative) polymerase chain reaction ((q)PCR) [25,26] or next generation sequencing (NGS) [27] to query the microbiome composition and abundance of certain species. Both methods, however, fail to provide spatial information. FISH in contrast enables the spatially precise and genus-specific distinction of bacteria. This has multiple advantages such as the analysis of a potential preference of the one or the other bacterial species for a certain gut compartment or a co-occurrence of certain bacteria in the form of consortia. As an exchange of metabolites between gut microbiome members of Drosophila was demonstrated [28,29] such a co-occurrence seems likely. When, where and how the bacteria interact or whether the metabolites instead are exchanged via a long distance, however, is not yet fully understood. A potential caveat of the FISH staining procedure and the dissection of the gut is that both manipulations potentially interfere with the normal bacteria distribution. Yet, the procedure is probably still closer to the physiological situation as compared to feeding genetically-modified fluorescent versions of the gut bacteria in large amounts to axenic animals in order to generate gnotobiotic variants.
Bacterial FISH has been used to investigate the microbiome in other insects such as beetles [30], the Asian citrus psyllid [31] or Drosophila suzukii [32] and is a standard method for corresponding studies in mammals [33][34][35]. Yet, in Drosophila only one study so far utilized the Eub338 probe to detect endogenous bacteria in the gut [12] and one study detected Acetobacter and Wolbachia in gnotobiotic animals [11]. Here, we present protocols suited for the detection of Drosophila gut bacteria in culture, in feces or in the gut of larvae or adult animals. During the design of our experiments, we realized that FISH protocol parameters vary widely between studies. To determine suitable staining parameters, we tuned multiple critical parameters such as the formamide concentration in the hybridization buffer, the probe concentration as well as the fixative and fixation time. For the sake of simplicity, we determined the effect of parameter variations in stainings of pure bacteria cultures and by quantifying the staining results using plate-reader as well as microscopy and image-segmentation based methods (S1 and S5 Figs). A parameter untouched by us was the number of fluorophores per probe and variations of the fluorophores themselves. Changing these parameters might allow an even more sensitive detection of bacteria as well as beneficial staining characteristics, as especially in the green channel background fluorescence of the food and the tissue in the larval and adult samples could be noted (S3, S4, S8 and S9 Figs). The biggest impact in our experiments had the fixation time and the fixative used. Previous FISH protocols performed with tissue samples mostly utilized a PFA-fixation, sometimes combined with an alcoholic dehydration prior to the actual hybridization (e.g. [12]). In our experience, the fixation with Carnoy's solution resulted in a superior signal to background behavior and better preservation of the overall morphology of the gut than the PFA-fixation (S7ii- S7iv Fig). Several previous reports state that the use of PFA for fixing guts results in the entire loss or collapse of the mucus layer [34,36,37] which is in line with our observations. Carnoy's was also used before for various staining protocols including FISH in invertebrates [31], yet often with much longer fixation times. The fixative performance likely depends on multiple parameters such as the fixation and hybridization time and the presence or absence of e.g. an additional dehydration step.
Altogether, our protocol presented herein should serve as a starting point for future experiments and provide a guideline for optimizing e.g. novel probes capable to differentiate bacteria down to the species level. Expanding the range of detected bacteria by multiplexing with such additional probes and assaying for the impact of environmental or genetic perturbations on the gut microbiome organization will pave the way to a better understanding of the gut microbiome compartmentalization and interactivity. plantarum, L. brevis, and L. fructivorans with the three different probes Eub338 (blue-detection in the DAPI channel), Aceto (green-detection in the GFP channel), and Lacto722 (reddetection in the RFP channel) (4 μM/probe) and 40% formamide for three hours at 46˚C. Lactobacillus strains were treated with 10 mg/ml lysozyme for 15 minutes at 37˚C prior to the hybridization. While each bacterial species sample was only stained with one probe (A: Eub388, B: Aceto, C: Lacto722) all detection channels were imaged with constant settings in order to test for potential fluorescence bleed-through. The figure shows representative examples from at least three separate experiments. Scalebar represents 20 μm. (TIF)

S7 Fig. Fixation of larval Drosophila guts using PFA and Carnoy's solution.
Larval Drosophila guts were fixed for 15 minutes using PFA (i) and for 5 minutes with Carnoy's solution (iiiv). DNA was stained with TO-PRO-3. (i) and (ii) were imaged using the same microscope settings and show the much higher staining intensity using the Carnoy's solution. (iii) TO-PRO-3 stained bacteria in larval Drosophila gut lumen. (iv) Zoom-in of (iii) showing gut bacteria marked by arrowheads. The asterisk highlights the mucosa layer. Scalebars represent 100 μm (i-ii), 50 μm (iii), and 20 μm (iv). (TIF) S8 Fig. Fluorescence in situ hybridization of axenic larval Drosophila guts. FISH was performed with axenic larval Drosophila guts using the genera-specific probes Aceto (green) and Lacto722 (red) (4 μM/probe) and 40% formamide for 16 hours at 46˚C. Prior to hybridization, larval guts were treated with 10 mg/ml lysozyme for 15 minutes at 37˚C. An overview of the entire gut was imaged and detailed zoom-ins of six different regions are shown. The figure  shows representative examples from at least three separate experiments. In each experiment 5 to 10 guts per condition were dissected. Scalebars represent 500 μm (overview) and 10 μm (zoom-ins). (TIF)

S9 Fig. Fluorescence in situ hybridization of axenic adult Drosophila guts.
FISH was performed with axenic adult Drosophila guts using the genera-specific probes Aceto (green) and Lacto722 (red) (4 μM/probe) and 40% formamide for 16 hours at 46˚C. Prior to hybridization, guts were treated with 10 mg/ml lysozyme for 15 minutes at 37˚C. An overview of the entire gut was imaged and detailed zoom-ins of six different regions are shown. The figure shows representative examples from at least three separate experiments. In each experiment 5 to 10 guts per condition were dissected. Scalebars represent 500 μm (overview) and 10 μm (zoomins). (TIF)