#microtwjc Do Electric Guts Dream of Android Burgers ? part 2

In part 1 of this post, I briefly described the key piece of equipment used in this study- the TIM2 machine, designed to model the large intestine.

So let's see what this machine can tell us about antibiotic associated diarrhoea. This disease happens after patients have been exposed to a long course of antibiotics, and the body's natural microbiota is wiped out. The usual treatments for this are the application of probiotics. But when do we apply them so that they have the most beneficial effect ? Do we give probiotics simultaneously with antibiotics and hope they flourish as well as they would without ? How will this affect the way the bacteria metabolise nutrients ?


Figure 2.

Quite helpfully, they put in a diagram of their experimental design. Essentially they took stool samples from 10 healthy adults of both genders *. From these samples they prepared a standardised set of microbiota **. They then put these microbes through the system for three treatments. The Control simply had no added clindamycin or VSL#3 and was monitored for seven days. The second group had both clindamycin and VSL#3 added, and was also monitored for seven days. The third group was essentially the same  as the second group, except that it was followed for seven days longer with VSL#3 alone.***

For each day that these experiments were run, samples from the dialysis fluid (essentially representing the nutrients absorbed from the stomach) and samples from within the fluid of the machine. I chip analysis was performed every 7 days- this could be used to assess the bacteria present within this electric gut.

Figure 3

So the first question that was asked was what effect these different treatments have on the production of beneficial metabolites.  So in this figure, they looked at short chain fatty acids, such as acetate, propionate and butyrate, measured daily using gas chromatography. The line graphs show cumulative frequency, i.e. the total accumulation of these metabolites throughout the experiment.

B and C  shows what happens to these metabolites with and without the clindamycin + VSL#3 treatment, and found these there wasn't much of a difference, save for butyrate production.

A  shows the clindamycin+VSL#3 followed by VSL#3 alone. Whilst it looks the same as the other groups for the first 7 days, after that there seems to be a sharp rise in the amounts of metabolites present. Considering that clindamycin is a bacteriostatic, it isn't such a great surprise that metabolites increase when it is removed.

At least until you realise the controls haven't received any antibiotics. So we could speculate that the presence of VSL#3 increases the presence of short chain fatty acids.

D- Ignore the title, as it's really misleading. What this shows is the total amount of short chain fatty acids that have accumulated over the first seven days for the first three groups, and then the accumulation of fatty acids that have accumulated over the last seven days for the long term group. This graph shows that the final seven days of the long term group (with VSL#3 alone) produced more propionate than the other groups.

Figure 4

So what about other compounds, such as lactate ? They look at that in their next figure.

These are another set of graphs showing the cumulative production of a compound, in this case- Lactate. This is structured the same way as the previous figure, and this time the differences are far more clear cut. L-Lactate and D-lactate (molecules of lactate that are structured as mirror images) are produced in roughly equal amounts for all groups. There is a drastic difference in cumulative production between the controls and the groups receiving the treatment. And as before, once clindamycin is removed, bacteria (most probably from the VSL#3 formulation) dramatically increase their production of lactate.

D demonstrates that when VSL#3 is acting alone, there is an increase in D-Lactate which accounts for much of the increase in lactate production.

Figure. 5

So we've looked at some of the beneficial compounds that can be made by probiotics, but can these probiotics prevent the production of potentially harmful products ?

This figure looks at the production of branched chain fatty acids (BCFA's), which are a signifier of protein fermentation which can lead to production of toxic metabolites.

This figure is set up in the same way as the previous ones, only this time showing the production of branched chain fatty acids. Keep in mind that for the first 7 days, the lines on A and B should be very similar. And in this figure, they are quite similar (just put a ruler against your screen to check, and try that out for all of the graphs)

Although the total levels in A are higher than in B, but I cannot tell whether it's outside the natural variation because I don't know whether the error bars they aren't showing are standard deviation or standard error.

Anyway, what this graph shows is that the clindamycin + VSL#3 treatment appears to reduce the presence of these branched chain fatty acids. But in group A , after clindamycin is no longer applied there is a dramatic increase in these products, and this is mirrored in D.

What they write about this in the results section is rather baffling though. What they set down in their study design was that for group A, VSL#3 was given throughout. What they imply in their results is that Clindamycin was the only treatment given for the first three days in this group, then follwed by VSL#3, and thus the reduction of  BCFA's must be due to the probiotics, but only when given with clindamycin. But then if that was the case the whole time, then we could attribute the higher rate of lactate production in the previous experiment to clindamycin alone, which is insane.

The data seems to show that the probiotics do produce more branched chain fatty acids when left unchecked by clindamycin. There is a small decrease when both clindamycin and VSL#3 are administered with respect to the control.

Figure 6

So what about ammonia production. As you are probably aware, ammonia is a fairly hazardous compound, which is produced as part of protein digestion. Reducing the presence of detectable ammonia would be beneficial, so they analysed the production of ammonia over time.

This graph is simpler to understand than the others. A shows that the control group produces more ammonia than the other two groups over the last seven days. The change in the gradient of the longer term group indicates that ammonia production is increased when clindamycin is removed.  Looking at the absolute amounts comparatively, it looks like the reduction of ammonia is due to the presence of clindamycin, as when it is removed ammonia production returns to the level seen in the controls.

Figure 7

So now we'll take a look at the bacteria that have been causing all of the chemical changes seen in the previous figures. So they used a microarray designed to detect the presence of genes expressed by specific species of bacteria. It can only detect a certain species of bacteria if there are more then ten million present per gram of fluid, so this analysis will miss out some of the less abundant bacteria.

The data shown is fold change compared to the controls, and this time they tested clindamycin alone compared with clindamycin + VSL#3, and VSL#3 after a weeks application of clindamycin. Red indicates that certain species are reduced with respect to the controls, and green indicates where some are increased, and blank spaces indicate no change. 

The interesting thing is looking at the blank spaces, and seeing that clindamycin and VSL#3 together induce the greatest changes in bacterial flora. and when clindamycin is removed, and VSL#3 is added, a lot of bacterial species had a reduction in their presence.

Lactobacilli are increased when both clindamycin and VSL#3 are given together, but when clindamycin is given alone, there is no change, and when it is given and then VSL#3 is given, then the numbers of lactobacilli decrease.

Summary

The Good

• Probiotics - They show that probiotics may still be active even in the presence of antibiotics, and their data suggest that it would be better for them to be administered with antibiotics rather than straight after.
• TIM-2- They make the TIM-2 system seem like an interesting machine, although that they resorted to cumulative data indicates that there must be a great deal of variance on a day to day basis which could hide certain trends.
• I-Chip- I think that what they did here was quite good, not just making a list of bacteria, but actually trying to understand how the various treatments are affecting these species.

The Bad

• Graphs- For figures 3-5, the data was presented in a way to confound. If you want to show the effect of a treatment on a specific output (like BCFA, or lactate production) then it would be great to see all those different treatments on the same graph. Like how they presented figure 6. I really don't like pressing a ruler against my computer screen to check if there are massive differences between groups.

The Ugly

• Poorly communicated experimental design :Figure 2  throws open some of the interesting problems of this paper. I generally quite like it when people sketch out their experimental design. But the problem with figure 2 is that the design sketched there isn't the design they keep to. In fact, for half the paper I was baffled as to what they were supposed to be doing. In figure 2 they suggest that for the long term treatment group they were giving VSL#3 throughout, but towards the end of the paper they straight up contradict this and say they only gave VSL#3 after 7 days to this group. Which is it ?
• The Clindamycin +VSL#3 followed by just VSL#3 The long term group itself is problematic. In the original paper where the TIM-2 machine was devised, it was noted that for bacteria derived from fecal samples, the compositions can change over time. So why didn't they keep all the groups running for 14 days? They could have generated directly comparable data. As it stands , it is very difficult to say that the late stage changes in bacterial and metabolite composition are not due to them being kept in the machine longer than the other groups.
• Error Bars- Where are the error bars? I see that some of the graphs have horizontal lines dangling off them, but what are those? are they standard deviation? standard error of the mean ?. The bar charts are handily labelled with numerical values which purport to show the averages. But without showing SEM, we are still in the dark as to how certain we can be of those averages.
• Replicates- How many replicates were performed for each of these experiments ? 
• Statistical analysis-We got some statistics for looking at the microarrays, which is fine. However, where are the statistics for all of the other experiments ? How do we know whether any of these findings are not simply due to the natural variation of bacteria and metabolites within the TIM2 system? We don't know.

So did the VSL#3 decrease toxic metabolites? Don't know, the graphs say one thing and the authors say another. 

Does it increase beneficial metabolites? Don't know, it may, but I'm not sure enough about what was done in each experiment to tell whether it was VSL#3 or Clindamycin doing the hard work.

What about when clindamycin is given followed by VSL#3 ? Still not sure, as I have no proper control to compare that data with. And to be honest,the confusion in the methods left me baffled as to what they actually did. 

I am certain that if these issues were cleared up, then the rest of the paper would fall into place. There is enough here to convince me that there could be a great paper in here somewhere, but not enough for me to find it.

Rehman, A., Heinsen, F.A., Koenen, M.E., Venema, K., Knecht, H., Hellmig, S., Schreiber, S. & Ott, S.J. (2012). Effects of probiotics and antibiotics on the intestinal homeostasis in a computer controlled model of the large intestine, BMC Microbiology, 12 (1) DOI: 10.1186/1471-2180-12-47

*Yes, I saw the spelling error, and as someone who constantly makes spelling errors, I can say that it does not bother me. 

**I have no idea how, as the paper with this technique was impossible for me to access- venema et al 2000 in Ernährung/Nutrition