How does diet affect our gut microbe health? - Microbiome composition of Boozy Flies by measured by next-gen deep sequencing.

With further awareness of the intricacies of our bodies as the product of not only human but also microbial cell populations, the interplay of microbial contributions to health has been pushed to the forefront of popular science. While our microbiome is tailored to each individual and may help maintain health, many factors can alter microbial composition which may lead to disbiosis: the unbalance of microbial populations that may lead to downstream disease or pathology. With many University student’s diet being supplemented with regular drinking, we are interested in the effect of an ethanol supplemented diet on gut microbiome composition. To elucidate this question, will use a fruit fly (Drosophila melanogaster) model fed with either conventional or alcohol supplemented (0.5% Ethanol) diets for 1 week. Microbial populations will be determined using deep sequencing of bacterial 16S DNA and compared between the two diets.

Previous studies in mice have found significant shifts in abundance of different microbial populations such as Bacteriodetes and Firmicutes in addition to Proteobacteria, Actinobacteria and Corynebacterium in response to alcohol supplemented diets, although these animals were fed a high alcohol diet (5% ethanol) to simulate chronic alcohol pathology. This equates roughly to a bender of standard 5% ABV beer for 7 weeks, and while sounds fun, hopefully is not representative of the standard drinking experience. We expect to see shifts in similar microbial populations in our boozy flies, though likely to a much lesser extent than seen in previous pathology based models.

Methods

Flies and modified diets
Vials were seeded with 10 Drosophila melanogaster flies. Only virgin female flies were chosen to prevent gender biases as well as population expansion. Vials were raised on either a chemically defined agar based holidic diet, or a holidic diet containing 0.5% ethanol at 25˚C for 14 days. Flies were then sedated and euthanized at -20˚C.

Isolation amplification and sequencing of bacterial 16S DNA
Homogenization and DNA purification using an Ultraclean Microbial DNA Isolation Kit following slightly modified manufacturer’s protocol. Bacterial 16S DNA was then amplified via conventional PCR using primers specific for a broad range of bacterial, but not fly 16S DNA. PCR products were quantified using Quant-iT high-sensitivity DNA assay which measures dsDNA concentration based on fluorescence of intercalating dye. Amplified DNA was tagmented (fragmented and tagged) with 2 sets of indexes specific for each group’s library. Fragments were sequenced using an Illumina sequence by synthesis multiplexed cluster technique. Unfortunately, a mechanical failure occurred during the sequencing run, so data analyzed was provided by Dr. Edan Foley from a previous experimental trial run on male flies. Sequence data was given to Dr. Bart Hazes, and analyzed for microbial diversity using Unix based bioinformatics programs and databases. Illumina sequence by synthesis is explained in detail in the video below from Illumina.
                                          

Results and Discussion
It is commonly thought that the composition of the gut microbiome is incredibly important for host health. With the huge amount of high sugar or fat content foods readily available alongside the myriad diets coming in and out of fashion, it is important to determine the effect that nutrition has on our gut flora. Specifically, the effect of alcohol on microbial populations interested us greatly, being among one of the most common nutritional additives to diet relative to the student lifestyle.

Upon sequencing gut microbes from flies fed an alcohol supplemented diet, we found that overall, the diversity of gut microbiome decreased (Figure 1). This drop in diversity was calculated quantitatively using Shannon diversity scores which take into account not only number of species, but relative abundance of each (Table 1). Control flies fed a ‘holidic’ diet with a known composition yielded a score of 5.92, while flies fed a holidic diet with 0.5% ethanol decreased the Shannon score from 4.613.
 

Figure 1. Gut floral composition of ethanol diet Drosophila by 16S DNA sequence. Male Drosophila melanogaster flies fed either holidic control or alcohol supplemented diets for 7 days and then euthanized. Bacterial 16S ribosomal DNA was isolated from whole fly homogenates are amplified using conventional PCR. Bacterial DNA was then tagmented with known adapters prior to Ilumina sequence by synthesis. Identity and proportions of bacteria were calculated based on abundance and sequence identity. a) Proportion of bacterial population on family levels. b) Proportion of population on genus level. Teal indicates population consisting of 39 various Acetobacterea genus'. Other colors correspond to numerous other small microbial populations.

.

Table 1. Diversity indexes of gut flora in flies fed an alcoholic diet. Shannon diversity scores were calculated using operational taxonomic units generated by analysis of deep sequencing data. Higher scores are achieved by a combination of variety and abundance of individual species. A score of 0 reflects a purely homogeneous population. Experimental data derived from given data, paper data derived from an upcoming publication from the Foley lab.

When looking at the effects of alcohol on specific populations of gut bacteria, we found a ~20% increase in population of members within the Acetobacteraceae family. This population consisted largely of Acetobacter species (52% and 63% of control and alcohol groups respectively) with the remaining populations consisting of 39 various other genus’. These results are not surprising, as acetic acid bacteria are able to utilize oxidation of ethanol into acetic acid. In addition, the vast majority of other bacterial populations decreased in size. This result was again unsurprising, as ethanol is toxic to most cells that are unable to metabolize it. While low concentrations of 0.5% may seem insignificant, long term exposure may have had the combined effect of increasing the growth of ethanol oxidizing bacteria (ie. Acetobacter) while slowing growth of other species. 

Although it’s hard to point out the biological significance of these results with looking solely at microbial composition, there are several important implications that could be made. The decrease in microbial diversity may have long term effects on nutrient access/ absorption, as many vitamins and other nutrients required for optimal growth are microbial metabolites. Reduction or depletion of these populations may also deplete the host of their nutritional benefits.

Significantly, Shin et all found that Acetobacter species modify Drosophila insulin and insulin growth factor signalling which regulate a huge range of physiological parameters such as body size, metabolism and rate of development (2). Thus, the effect of alcohol on microbial populations likely plays a role in the development of embryos and young children that require precisely coordinated signals while growing. It seems possible that the effect of fetal alcohol syndrome may be, at least in part, due to altered host-microbiome cross-talk. Additionally, depletion of microbial diversity may be affecting myriad possible interactions between resident gut microbes and the host that have yet to be described which could have varying impacts on an equally vast array of physiological parameters.

This experiment highlights the drastic changes in gut microbe composition that can be caused by even small changes in diet. Although this experiment provides important information of the impact of alcohol on the microbial populations, it does not provide any insight into what impact these population shifts have on the host. It is also important to realize that individuals within heterogeneous populations can have a very different gut microflora, and that each microbiome may act differently to identical changes in diet, especially given the complexity of interactions between the host and it’s commensals.


Technical discussion

Although the calculated Shannon scores for our experiment are 5-20 fold higher than those seen in the Foley paper, the trends seen are the same, strengthening the robustness of our results. The difference in magnitude is likely due to the way the data was analyzed. Each bacterial 16S DNA sequence is included into an Operational taxonomic unit (OUT), since sequence doesn’t necessarily describe a species. Stringency of OTUs can be adjusted to be more precise, giving more clusters of association, or less stringent, giving clusters of more broadly related microbes. We obtained 954 individual OTUs across all samples using a stringency of 97% sequence resemblance, which is ~30 fold higher than the 30 species known to be Drosophila gut commensals. This huge number is likely what increased the magnitude of our diversity scores and was likely caused by using whole fly homogenates where all external bacteria, commensal or otherwise, would be included. Homogenized isolated fly guts used in the Foley paper would include only gut microbes and likely a much smaller number of OTUs, though total OTUs obtained in their results was not included in the paper. Another reason is possible contamination from processing samples outside of a biosafety cabinet.  In addition, since only bacterial 16S ribosomal DNA was amplified, the effect of alcohol on eukaryotic commensals such as yeasts remain unknown.



Now with the final post of my MMI 590 blog, its time to embrace the student lifestyle and transition this experiment to ‘human trials’ with a pint or two.

Cheers!

Mike Wong
University of Alberta


1. Bull-Otterson et al. Metagenomic Analyses of Alcohol Induced Pathogenic Alterations in the Intestinal Microbiome and the Effect of Lactobacillus rhamnosus GG Treatment (2013). PLoS ONE. DOI: 10.1371/journal.pone.0053028

2. Shin et al. Drosophila microbiome modulates host developmental and metabolic homeostasis via insulin signaling. Science. Vol.334 no. 6056 pp. 670-674 DOI: 10.1126/science.1212782

Fluorescence microscopy and live cell imaging

Fluorescence microscopy has long been employed to observe biological processes within fixed cells or tissue sections. We are now more than ever able to expand our field of knowledge using the plethora of dyes and fluorophores to stain these samples and gain higher resolution of multiple processes in multiparametric fashion. Although using these dyes on fixed samples from different intervals (ie. Fixed at different times post treatment) can reveal kinetic trends, it does not allow the resolution to observe biological processes in real time. With the advent of live cell imaging, we are now capable of viewing cellular kinetics in real time within individual cells. Here in we demonstrate the use of live cell imaging to track the movement of mitochondria within live cells, and the effect of photo-bleaching when doing live cell experiments. We also demonstrate the use of live cell imaging to follow kinetics of nuclear NFκB translocation to the nucleus in response to recombinant TNF-α.

Methods

Mitochondria movement
Protocols are outlined in the MMI 490 Fall Semester 2015 lab manual with minor changes. Briefly, HeLa cells were seeded onto glass coverslips at 20% confluence in cDMEM (10% FCS) and incubated at 37°C with 5% CO₂. Slips were transferred to 6 well plates and stained with 1:333 Pico green, 0.5µM Mitotracker red CMX ros and 0.8µM Hoescht 33258 for 1 hour. Cover slips were individually transferred to live cell imaging chambers containing 0.5mL media before sampling on a Leica SP5 scanning confocal fluorescence microscope. Images were acquired once a minute for 40 minutes and analyzed using FIJI image analysis software to obtain mean fluorescence for whole cells in each channel.

Transfection of GFP-NFκB and NFκB translocation
HeLa cells were seeded onto glass cover slips at 20% confluence overnight. 100µL 1:33 Fugene 6 HD in Opti-MEM was mixed with 1 volume 200µM GFP-NFκB encoding plasmid in Opti-MEM. This mixture was incubated at 37°C for 20 minutes before drop-wise addition to cells in 1mL Opti-MEM (3% FCS). Cells were incubated at 37C 5% CO₂ before transfer of coverslips to a live cell imaging chamber. GFP distribution was then analyzed using the Olympus Wave FX 1 spinning disk fluorescence microscope. Data obtained from Dr. Steve Ogg contains images that were captured every 12 seconds for 1 hour with TNF addition at 3 minutes into the run. Fluorescence was measured using FIJI analysis software by comparing mean pixel fluorescence of the nucleus to that of the entire cell.

Results and Discussion

Mitochondrial movement in live HeLa cells
Mitochondria are often thought of as stationary blobs within the cell, but in fact their movement can be quite pronounced. Although this is the case, the health and state of cells can impact the mitochondrial behaviour, where our time lapse videos failed to show much significant movement at all. This could be due to many factors including density, activation states, temperature or cell ‘happiness’ outside of 5% CO₂ since the SP5 is not equipped with a live cell incubation unit. Another issue we had was the predominant staining with Pico green, an intercalating dye that binds to the small groove of dsDNA (2). We believe that the addition of Pico green at high concentration before the addition of Hoescht bound all the nuclear DNA and excluded the Hoechst dye from being displayed (Figure 1) since green signal can be seen dispersed throughout the cell, but also very brightly over the nucleus. The mitrotracker red displayed great affinity for mitochondria, which was expected since it localizes to mitochondrial membrane potential.

Figure 1. Photo-bleaching of vital dyes during live cell imaging. Hela cells were stained with pico green and mitotracker red previous to image capture on an SP5 spinning disk fluorescence microscope. Images were captured once a minute for 40 minutes. Images displayed from the beginning of the run and at the end, in both green and red channels.

Figure 2. Quantification of mean fluorescence of vital dyes during live cell imaging. Mean whole cell fluorescence was measured for each channel respectively in each acquired image and compared to the starting frame as calculated by (frame (x) mean fluorescence/ start frame mean fluorescence). 

Photo-bleaching is the degradation of fluorophores by light sources and is concern when doing any type of fluorescent assay. It is especially important for time-lapse imaging, since bleaching can prevent capture of further time points. Hence it is important to consider longer image capture times on longer assays to minimize exposure to high energy light and thus prevent extensive photo-bleaching. In this assay, one image was taken per minute in 3 channels for 40 minutes. Even at this setting, both detectable fluorophores (Mitotracker Red and Pico green) displayed significant levels of photo-bleaching with visually detectable differences from beginning to end (Figure 1). When quantified using FIJI (is just ImageJ), end mean fluorescence for individual cells was seen to drop by ~30% and 50% from start mean fluorescence for Pico green and Mitotracker red respectively (Figure 2). If more images had been taken during the same time frame, the images may have been dimmer or non-existent due to higher light exposures. This is especially problematic since bleaching is known to be exponential, and a lack of signal prevents any further data from being extracted from the assay.

NFkB translocation in TNFα stimulated HeLa cells
NFκB is a critical transcription factor involved in stimulating inflammatory reactions. Many studies observe the percent of cells that have translocated after specific treatments, etc., but that percentage is just a snap shot of what is happening in a cell. Using live cell imaging, we are able to show the kinetics of how fast HeLa cells translocate NFκB to the nucleus in response to TNFα stimulation. In data obtained from Dr. Steve Ogg, we observed translocation in all 3 cells found in one frame (Figure 3a), but not those found in another. When total cell fluorescence was compared to that of the nucleus, we found 3 cells experienced a mean nuclear intensity that increased steadily from 70-80% to 80-100% of that of the total cell (Figure 4). While each cell had slightly different kinetics, indicated by varying slopes of each line, each cell experiences filling in of the nucleus with the GFP tagged NFkB that is evident in the image. One the other hand, another set of cells did not display any translocation, with nuclear intensity actually dropping after treatment with TNF (Figure 3b). This could be due to the state of growth of the cells, the density on the cover slip, relative health, or crowding as outlined by the Pelkmans group in their 2009 Nature publication¹. They found that heterogeneity of cells within a single dish was not due to random variation, but could be predicted based on morphology and relative niche of each grouping of cells due to differential growth kinetics, activity, etc. Further, they could predict lower infectivity of certain cell subsets based on their positioning based on growth phase dependent expression of certain phospholipids that simian virus 40 (SV40) uses for entry.

                                                                                     

Figure 3. Translocation of NFkB-GFP in TNF stimulated HeLa cells. Transfected HeLa cells were imaged for GFP translocation to the nucleus for 30 minutes after stimulation with recombinant TNFa. a) Cells that exhibited translocation, cells 1-3. b) Cells that showed no translocation (cell 4). Yellow lines show representative regions of interest used to quantify fluorescence in Figure 4. c) Response to cell 4. Time lapse video of d) NFkB translocating cells e) non-translocating cells. (Apologies for the sloppy figure, Blogger doesn't seem to like embedding videos very much!)



d-e

                                                                                                                                                        e.

Figure 4. Quantification of NFkB translocation via nuclear fluorescence intensity. Individual cells indicated in Figure 3 were quantified for mean fluorescence in the nucleus compared to that of the whole cell for one frame every 2 minutes. A reading of 1 indicates a mean fluorescence of the nucleus identical to the whole cell. the red arrow indicates TNF addition at time = 3min.



Likely, this may contribute to why our personal experiments didn’t work, as we may have chosen to image cells that either fell into a growth phase that doesn’t express much TNF receptor, or that is somehow inhibited in the NFkB pathway to prevent proliferation of activated cells. Antibody staining for TNFR could be used to rule out the first possibility. Alternatively, it could have been due to the TNF becoming non-functional from too many freeze thaws, as we performed 3 separate technical replicates to try to obtain our own data. Problems with a speckle reducer, a module in the scope which smooths out light coming through the 50 fiber optic threads within the cable, prevented quality image acquisition on our first run. Additionally, a faulty temperature probe allowed the cell incubator to reach over 44 °C which likely had a detrimental effect on our cells. Though, as with all science one cannot expect results from the first run, and patience must be observed while trying to troubleshoot possible sources of error to fully optimize an assay.

1. Snijder, B. et al. Population context determines cell-to-cell variability in endocytosis and virus infection. Nature 461, 520–523 (2009).

2. Dragan, A. et al. Characterization of PicoGreen interaction with dsDNA and the origin of its fluorescence enhancement upon binding. Journal of Biophysics (2010).
doi: 10.1016/j.bpj.2010.09.012.