Botulism or not?

By Siouxsie Wiles 01/09/2013

MPI says Fonterra whey powder not contaminated with botulism-toxin producing bacterium.

Over the past few days I’ve been contacted by a number of journalists asking how this case of mistaken identity could have been made. “What did they do wrong?” has been the main question. To answer that we need to know all the details of the experiments and see the actual data. This critiquing process is what scientist call peer review. When scientists have finished their experiments and want to tell other scientists about their findings, they will write a paper explaining how they did their experiments (the Materials & Methods) followed by showing their data (the Results) and then they usually speculate wildly about what those results mean (the Discussion). The paper is submitted to a journal where it is then sent to a number of experts for critique, an often brutal process, with reviewers pulling no punches!

One of the most important questions a reviewer asks is this: “did the experiments actually work as they should?”. To answer this we need to understand the most important part of an experiment, the controls. Without proper controls, an experiment is meaningless. So lets go through the tests that could be performed to see if a bacterium contains the botulinum toxin, and I’ll explain what the controls would be and what could go wrong.

The first test involves looking for the presence of the botulinum toxin in the bacterium’s genetic material or DNA. To do this we use an exquisitely sensitive technique called the polymerase chain reaction or PCR for short. As I’ve explained before, PCR is a clever technique that amplifies very small amounts of genetic material to generate over a billion copies, taking something that is almost invisible and making it detectable. One of the crucial ingredients in a PCR reaction is a set of very specific ‘primers’ which recognise the region of genetic material that you want to amplify. You need primers to each end of the region of interest and PCR amplifies the bit between the primers.

One of the downsides of the sensitivity of PCR is that it is very easy to contaminate, so proper controls are really important. This means that when carrying out a PCR reaction, alongside the samples being tested, it is crucial to include the following:

1. A negative control which has water in place of any target genetic material. This will tell you whether you have had a contamination problem or not.

2. A negative control which has control genetic material that does not contain any of the target sequence. This will tell you if your primers are specific enough.

3. A positive control which has genetic material that does contain the target sequence. This will tell you if your reaction has worked.

When the PCR machine has finished you are left with a little tube filled with either billions of copies of the target sequence if the sample was positive, or none if the sample was negative. This can then be visualised using a technique called gel electrophoresis and you are left with something like the picture below.

Example of PCR samples visualised by gel electrophoresis

So to recap:

1. The primers need to be specific so that they only amplify what you are targeting and nothing else. If the primers are not specific, you will get a false positive result.

2. You have to be very very very careful not to contaminate the reaction. If you do inadvertently contaminate your sample, you will get a false positive result.

Once it has been confirmed that the bacterium has the botulinum toxin gene, the next assay that can be done is called the mouse bioassay. Suspected toxin samples are injected into the abdominal cavity of mice at different dilutions, and the animals monitored for signs of illness over a 48 hour period. If the toxin is present, mice injected with the sample solution will show signs of botulism. The time it takes for them to become unwell will depend on the concentration of toxin present in the sample injected. To find out which toxin is present, groups of mice are also injected with antitoxins to each of toxins A, B, E and F. If the correct antitoxin is injected alongside the sample, the toxin will be neutralised and those animals will not show any signs of botulism.

In the case of the mouse bioassay, a group of animals would be injected with an inert substance to show that the injection alone doesn’t cause the animals to become sick. These would be the ‘negative’ control animals. Another group would be injected with a sample known to contain botulinum toxin, and would be expected to become sick. These are the ‘positive’ control animals. If the ‘negative’ control animals become sick, this suggests the animals may have an underlying illness, been incorrectly dosed, or that the reagents have been inadvertently contaminated.

The fact that PCR reactions can be inadvertently contaminated, or animals incorrectly dosed, brings us to another important aspect of doing science: replication. Experiments should be performed several times, on different days (and preferably by different people) using different batches of reagents, to find out if the result is reproducible. That is, was the result a fluke or does it happen reliably?

I hope this post has helped explain how it could be that different labs have presented different results for whether botulinum toxin was present in Fonterra’s contaminated whey powder. But until we have been told how all the botulism testing was done, how many times experiments were repeated and how the controls performed, we can’t say anything about what did go wrong.

UPDATE: MPI release their methodology and results for botulism toxin testing – available here. Alas Appendix 11 and appendix 12 which outline what testing Fonterra had done have been “withheld from public release”. .

As an aside, it is amusing to note the disclaimer at the beginning of the MPI report:

any and all responsibility for any inaccuracy, error, omission, lateness or any other kind of inadequacy, deficiency, or flaw in, or in relation to, the information.

0 Responses to “Botulism or not?”

  • There is little done by way of PCR testing in the dairy industry – something we’re helping some labs with. Nice bit Siouxsie; just on the subtyping though- more commonly done by PCR as well. We have developed real-time PCR assays for a clostridium toxin-producing screen (not just botulinum) and also the sub-types (as per article A.B, E and F). Real-time PCR = faster, cheaper and much lower chance of PCR contamination.

  • I completely agree with the fact that there is need for peer-review on the actual data and experimental design. But the fact of the matter is (as you would find from MPI released reports) that most, if not all, all information are omitted due to commercial sensitivity of the results and/or for the protection of “NZ inc”.

    In addition to the experimental designs, there is also a question of the sample integrity. According to the latest MPI report, samples were sent overseas for testing. Therefore, sample reparation and integrity during transit is also very important and can influence the final results.

    It is a frustrating time for scientists when there is a lack of transparency. This lack of transparency undermines the integrity of NZ science and raises doubts about the accuracy of the statements issued on MPI website (see latest news on MPI website) since no evidence was presented, nor was any reference made to validate the use of techniques to determine whether botulinum is present.

    Bacteria are constantly evolving. In light of the O104 outbreak in Germany a couple of years back. It is entirely possible the new Clostridia species have emerged, which may not be detected by existing PCR techniques. I am worried by the latest MPI which categorically stated the organism in question was Cl. sporogenes with little or no evidence.

    Yes, the ban has been lifted and perhaps trade activity will recover to normal. But, with the lack of evidence, I question the drivers behind these bold statements and worry for the future of NZ science.

  • PCR (“conventional PCR”) is as you described eloquently Siouxsie – amplifying the target of interest and analysing it on a gel. Opening the reaction tube generates aerosols and the chance for contamination of following reactions – one of the reasons PCR has to be performed with separate areas (and ideally rooms).
    Real-time PCR has a fluorescent chemistry – as the target is amplified, fluorescence is generated by one of a number of chemistries available. That fluorescence is analysed each cycle as the reaction progresses – or in ‘real-time’, Thus once the PCR is finished, so is the detection. The reactions are analysed through clear films and never need to be opened – thus little chance of amplified DNA contaminating future mastermix reagents. It also has quantitative aspects used in research for gene expression
    Real-time PCR is currently the major molecular diagnostic technology (human, animal and plant molecular diagnostics).

  • Having been many years since molecular bio, and not keeping up with the technology I had never heard of real-time PCR. Then I was asked to audit a test in our company that used it.

    One question I had around the operation of the instrument was; Is it necessary to check the temperature of the reaction with a reference thermocouple? At least periodically.
    As the entire reaction is controlled very precisely with temperature a fault here could invalidate results (ideally this would be picked up via the controls Siouxsie describes).

    Do you have info on this I might feed back?

  • The use of thermocouples came about from standard PCR machines – because there was no way to tell how well a particular fragment may or may not be amplifying (a neat and a million lower-fold concentration template look the same on a gel after amplification).
    With real-time PCR we see a lot more information – namely the cycle where exponential amplification starts to occur (the basis of the ‘q’ in quantitative PCR – another name for real-time PCR).
    This Cq is extremely reproducible.

    Therefore, with the use of well-maintained, appropriate concentration and well-stored controls (which is a whole other topic in molecular diagnostics) and the recording of Cq’s, faults in qPCR instrument heating can be identified very quickly. Therefore the relatively crude thermocouples are not required IMO