Web of Ecological Interactions in an Experimental Gut Microbiota

Web of Ecological Interactions in an Experimental Gut Microbiota : 

Web of Ecological Interactions in an Experimental Gut Microbiota By Birendra Tripathi, Sara Vitorino, Ekaterina Yarovitsyna and Kalyani Raut 5th November 2010 Authors: Pål Trosvik, Knut Rudi, Knut Olav Strætkvern, Kjetill S. Jakobsen, Tormod Næs and Nils Chr. Stenseth

Introduction : 

Introduction

Human Gut Microbiota : 

Human Gut Microbiota The gut flora (microbiota) is the human flora of microorganism that normally live in the gastrointestinal and can perform a number of useful functions for their hosts . Examples: Immunity (the response to pathogens and commensals)

Human Gut Microbiota : 

Human Gut Microbiota Nutrition (the degrading dietary substances that are indigestible by the host) Development (millions of years of coexistence a human and his resident bacteria contributed to human’s development and physiology ) Homeostasis (regulate its inner enviroment to ensure its stability in response to fluctuations in the outside environment and the weather ) Number of pathologies

Gastrotestinal Tract as an Ecosystem : 

Gastrotestinal Tract as an Ecosystem The mammalian gastrointestinal (GI) tract is the most densely inhabited of all known ecosystems, with microbial counts reaching 1011 - 1012 cells per ml of luminal content in the colon Current estimates of species richness generally range from 500 to 1000 representing several thousand strains In spite of this biological abundance, the microbiota of the human GI tract shows a low degree of diversity at the phylum level, with the majority (> 90%) belonging to just two out of 70 known bacterial phyla (microbiota core): Bacteroidetes and Firmicutes Others like Preoteobacteria and Actinobacteria represent the less abundant phyla

Well-Established Theory in Animal and Plant Ecology : 

Well-Established Theory in Animal and Plant Ecology Dynamics behaviour as an ecosystem Intrinsic density-depended effects Non-linear interaction Linear interaction Environmental forcing

Ecological Theory in Bacteria Community : 

Ecological Theory in Bacteria Community The traditional application of the ecological theory to bacterial community has been only in terms of individual growth rates and nutrient availability, in the original given environment In low-complexity scenarios this approach can be extremely fruitful, but it may not be valid in more complex community settings It is a widely held view that microbe ↔ microbe interactions are important in shaping gut communities and that gut microbial communities are, to a large extent, structured through environmental forcing

Objectives : 

Objectives The present work aims to provide a functional description of the dynamic structure of a simplified gut microbiota, without environmental forcing by host selection, helping to describe and quantify interactions between typical members of the gut microbiota This work’s findings highlight the potential importance of intrinsic, density-dependent ecological interactions in the establishment and maintenance of a stable and structured microbial community

Why is this Important? : 

Why is this Important? An especially intriguing issue is the initial colonization of infants: Since infants do not have trained immune systems, it is likely that interactions between bacteria are an important factor in the initial colonization of the GI tract. Understanding the ecological mechanisms underlying the bacterial colonization of the infant gut is therefore of great importance.

Experimental MethodsvsResults and Discussion : 

Experimental MethodsvsResults and Discussion

Bacterial Strains : 

Bacterial Strains 4 bacterial species representing the 4 main phyla of GI tract colonizers : Bacteroides thetaiotaomicron (Bt) → Bacteroidetes phylum 6% of the human gut micro biota Symbiotic Clostridium perfringens (Cp) → Firmicutes phylum normal gut micro biota of humans and animals major pathogen Escherichia coli (Ec) → Proteobacteria phylum Commensal member of the normal mammalian gut microbiota Several pathogen strains Bifidobacterium longum (Bl) → Actinobacteria phylum Relatively low quantities in the normal mammalian GI microbiota Encoutered frequently as ingrediants in health foods

Growth in a Chemostat : 

Anaerobial growth in a suitable medium Continuously logging of RedOx potential and pH results Growth in a Chemostat

Mean Time Series of RedOx Potential and pH : 

Mean Time Series of RedOx Potential and pH With standard errors represented with the vertical bars pH RedOx time time

Mean Time Series of RedOx Potential and pH : 

Mean Time Series of RedOx Potential and pH The bulk of variation in both pH and RedOx was observed during roughly the first 10 h of the two chemostats Due to the discrepancy between the replicate results, and the relative stability of RedOx potential and pH during the larger part of the experiments, they were not included as parameters in further analysis It is also interesting that differing time trajectories of pH and RedOx potential did not have an appreciable effect on the population trajectories

Growth in a Chemostat : 

Automatically withdrawing of cell samples every hour with subsequently freezing and optical density measuring of samples at 600 nm Growth in a Chemostat Chemostat 37oC ≈33h

Overall Growth Curve from the Optical Density Measurements at 600 nm : 

Overall Growth Curve from the Optical Density Measurements at 600 nm The chemostats reached a steady overall density level after approximately 10 h of growth

DNA extraction, PCR amplification and DNA sequencing : 

DNA extraction, PCR amplification and DNA sequencing To quantify individual species in mixed population from each samples from the chemostat All DNAs were extracted A stretch of 35 nucleotides in the 16S rRNA gene were amplified by PCR DNAs were sequenced and presented in electropherogram spectral data The data were pre-processed and entered into the Partial Least Squares Regression (PLSR) All PLSR computations were carried out using the R-package ‘pls’ statistics programme

The Mean Time Population Series : 

The Mean Time Population Series Standard errors were heavily inflated during the first 10–12 h of the experiments, reflecting the fact that either Cp or Ec showed the fastest initial growth Cp is the dominant species in the community throughout the series Bt shows slow initial growth, but its density increases steadily, reaching levels comparable to Cp after ≈60 h Bacteroides thetaiotaomicron Bifidobacterium longum Clostridium perfringens Escherichia coli

The Mean Time Population Series : 

The Mean Time Population Series Ec abundances are generally around 10%, while Bl occurs only at low levels All four subpopulations were regulated by intra-specific density-dependence (involved in regulating abundances of the other species) ↓ Competitive Interactions Bacteroides thetaiotaomicron Bifidobacterium longum Clostridium perfringens Escherichia coli

The Mean Time Population Series : 

The Mean Time Population Series Cp is the dominant species throughout most of the observed period (bacteria of this species are extremely fast growing). But it is never able to establish complete dominance in this system. This appears, at least in part, to be due to the presence of Bl, which restricts the growth of Cp Bacteroides thetaiotaomicron Bifidobacterium longum Clostridium perfringens Escherichia coli

Pre-Processing of the Time Series Data : 

Pre-Processing of the Time Series Data In order to reduce the effect of noise in the data it was filtered In further analysis only experiments with a stable total bacterial abundance (from 11 h and onward) were processed giving data for a total of 81 time points (hours 12–92) Smoothing To provide a means of decomposing a periodic signal on a local timescale, which makes it possible to find the dominant periodic phenomena in a time series, as well as providing time resolution for these the wavelet analysis was used

Time Trends in The Filtered Time Series : 

Time Trends in The Filtered Time Series Highly significant time trends were detected for all series except for Bl

Smoothed and De-Trended Population Series : 

Smoothed and De-Trended Population Series De-trending results in the centering of the series so that the ordinate axes indicate fluctuations in relative abundance around zero

Autocorrelation Correlograms of the 4 Population Time Series : 

Autocorrelation Correlograms of the 4 Population Time Series Clear patterns of cycling with periods in excess of 20 h in the case of Bt, Bl and Cp

Wavelet Power Spectra of the 4 Population Time Series : 

Wavelet Power Spectra of the 4 Population Time Series The abscissae indicate the time from 12 to 92 hours of the series The ordinates indicate the period of detected oscillations

Wavelet Power Spectra of the 4 Population Time Series : 

Wavelet Power Spectra of the 4 Population Time Series It suggested that the cycles were strongest in the populations of Bt and Cp, whereas the signal for Bl was more diffuse For Ec the wavelet analysis showed signs of an oscillation of a shorter period (≈8 h) at the beginning of the series, but this oscillation gave way to a period around 20h towards the end

Regression Modelling of Time Series : 

Regression Modelling of Time Series In order to identify the biotic interactions driving the dynamics in the system, the filtered and de-trended time series were analysed by fitting generalized additive models (GAMs) to each population time series From the set of full models the following were selected according to the generalized cross-validation (GCV) criterion Four selected GAMs:

Regression Modelling of Time Series : 

Regression Modelling of Time Series Where Btt+1, Blt+1, Cpt+1, Ect+1 – relative bacterial abundance in (t+1) time measurement Btt, Blt, Cpt, Ect – relative bacterial abundance in t time measurement bi – intercepts, (since the times series were centred the intercept terms, bi, are all zero); fi, hi, gi, ki – nonparametric smooth functions specifying the effects B. thetaiotaomicron, C. perfringens, B. longum and E. coli respectively on the target response term; ɛit – the noise terms

Checking of the GAMs : 

Checking of the GAMs Except for a few outliers, the model residuals were found to be approximately normally distributed, homoscedastic , and without appreciable remaining autocorrelation structure

Plots of Smooth Terms from the four Selected GAMs : 

Plots of Smooth Terms from the four Selected GAMs

Plots of Smooth Terms from the four Selected GAMs : 

Plots of Smooth Terms from the four Selected GAMs The fact that no non-linear model terms were indentified in Equation indicates that the GAMs are essentially identical to their corresponding linear parametric models

The Predicted Series Overlaid Over the Observed series (filtered and de-trended) : 

The Predicted Series Overlaid Over the Observed series (filtered and de-trended) There is generally good accordance between prediction and observation Although some time points are poorly predicted: In particular the final observation (at 92 h) is not well predicted for Bt and Cp

Conclusions : 

Conclusions

The Whole Analysis: Network of Interactions Governing the Dynamics of the System : 

The Whole Analysis: Network of Interactions Governing the Dynamics of the System Arrows represent interactions inferred from statistical modelling The strength and nature of a given interaction is indicated by the number next to the arrow For example, the arrow labelled -0.35 from Cp to Bt signifies that a 10% increase in the abundance of the former tends to cause a 3.5% decrease in abundance of the latter The curved arrows indicate the effects of the target species on itself (i.e. intra-specific interactions)

Final Considerations : 

It is thought that community structure in the GI tract is primarily a result of top-down selection by the host organism But even in the gross simplified system of the GI tract described in this article it may be seen that the temporal system structure and dynamics is a result of microbe ↔ microbe interactions All the interactions identified were of a competitive nature. However, it was not able to identify any symmetrical interactions where two species competed in a reciprocal fashion, and this may be an indication that scramble competition for nutrients was not the only underlying phenomenon Final Considerations

Final Considerations : 

It was verified that such a simple model system spontaneously adopts a structure similar to that of the human GI microbiota Importantly, GAMs were able to explain large proportions of the observed dynamics, demonstrating that such an indirect means of inferring biotic interactions can be a worthwhile exercise Final Considerations

Final Considerations : 

Final Considerations This approach allows to isolate and describe the sort of interactions that take place among typical gut bacteria. Since the health status of a human is greatly influenced by its GI microbiota, knowledge of the basic microbial ecology of the gut is of great importance. The ways in which gut microbes interact with one another, as well as their response to environmental changes, are likely to play significant roles in determining the dynamic behaviour of the GI ecosystem. Identification and description of the factors that regulate dynamics in microbial communities provide a means of predicting their behaviour, and manipulation of key dynamic parameters represents a potential way of controlling such systems

Final Considerations : 

Final Considerations Identifying biotic interactions between B. Longum and C. perfringens (potentially harmful bacteria) should be of particular importance in newborns and infants, where the GI microbiota is characterized by strong fluctuations in community composition during the first year and host control of the gut microbiota appears to be greatly reduced due to the fact that they do not have fully developed immune systems. Thus, ecological interactions between bacteria may be expected to play a significant role as determinants of early microbiota structure. In this way, the approach taken in this article could be useful for devising new prevention strategies and therapies for early microbiota-associated diseases

Thank You for the Attention : 

Thank You for the Attention