Does the IUPAC Collaborative trial study meet the criteria for a qualitative study?

A good method would screen for the full range of possible events within a commodity using a single PCR protocol.

Screens based on 35S detection are required within my company to be followed up with variety-specific screens. I had at first thought that that was overkill, operating on the assumption that the CaMV would be unlikely to infect non-Brassica crops. However, Wolf et. al. in Eur Food Res Technol (2000) 210:367-372 found evidence for CaMV infection in 7 food samples (vegetable and soy products) which tested 35S positive and variety negative. They warn that the sequences of the various strains for CaMV are not well enough known to be able to say for certain that their PCR test will detect all strains. We have seen 35S positive-variety negative results here on my lab. Of course, we have specific tests for only a few varieties, but even so, without a test for CaMV, I'm inclined to view the 35S screen with some suspicion.

Given the fact that 35S and NOS have been the common promoter/terminator sequences used in GMO's they provide an excellent target to assay (IUPAC Collobarative) irrespective of modification type. But as Beth has indicated, 35S has been derived from CaMV and therefore can give false positive results. With that in mind the 35S/NOS approach to detect GMO's becomes an initial sceening test. Depending on the outcome of the 35S/NOS results verification of positives with another round of PCR targeting CaMV and the specific transgeneic gene appears to be necessary for confirmation. Quantitaive PCR would then follow after confirmation. I think it might be worth while to develop a flow diagram of a testing strategy to help build a framework for the specific test(s) needed to get at PCR GMO detection.

When using 35s or nos (or any other system that is not event specific) the quantitative result obtained can be confusing, as the different GMO may contain different copy numbers of the target sequence. Thus when using an assay calibrated with a material having a single inserted copy may lead to a two fold overestimation when detecting a GMO having two inserted copies - but still, given the number of GMO that need to be deteceted, such quantitative screening tests are useful - as long as the result is below the desired threshold. A comment to Beth Curran: The paper we published together with Christian Wolf focused on specific samples where a suspicion on false positive existed - thus we can not make any assumption on the frequency of such false positives - in fact there are few matrices where we observe such false positives with significant frequency whereas in other (most) matrices we never observed false positives. The decision tree that is proposed clearly must be matrix specific as well.

David, can you help us get started on a flow diagram for a qualitative assay? Does anyone know of other promotor/terminator sequences that we really need to be concerned about? Is there anything likely to appear in the next year or so that we should consider or include?

Early in this discussion, it is important for us to identify which crops we are going to discuss. For example, if canola is included in this discussion, the 35S/NOS screen is of little value for detecting the GMO/non-GMO status of a bulk shipment. In Canada (a major canola producer),some of the approved events do not possess these elements as part of the transgenes. So lets define the crop types we wish to include, to limit or perhaps broaden the perspective of this discussion group.

I would like to echo Andreas caution of quantitative PCR determinations based on transgene elements such as 35S, Cry1a, and NOS. It is documented that some corn events contain multiple insertions (MON 802) and/or partial insertion (MON 809) events. These multiple insertions can create serious problems with quantitative PCR estimations especially in bulk shipments since they are more than likely a mixture of varieties containing both single insertion and multiple insertion events. One simple method to overcome multiple insertional events is to design primers which span the transgene/genomic borders. These amplicons are single copy and event-specific.

About quantitative PCR: We have proposed in our company to perform a RR/soya specific quantification and 35S-promoter/maize quantification so far. I think it is clear that there are nearly no problems with determination and quantification of RRS? For maize we are aware of the problem that there are different copy numbers in the GM-maize events but an accurate quantification of ALL specific GM-maize events is not only awfully time consuming but also not possible if there is a mixture of at least two different GM-maize events. Therefore we think it is much better to have a GOOD estimation with only ONE run. Yes, for some events we might overestimate the amount of GM-maize - but for our needs we think it is better to overestimate some samples and to analyze it in an adequate time. The copy number problem can be "solved" when you use standards of the same events you want to quantify (at the moment only Bt11 and Bt176 available). For sure we always determine the GM-crop by another specific PCR method (CryIA, nos, etc)after screening for 35S.

Concerning promoter sequences: the FMV promoter is also widely used.

Question for Andreas: What was it about those specific 35S positive/variety negative samples that raised your suspicions? Were there other samples having similar results that were not considered suspicious? The reason I am asking is that we have seen this type of result twice in our lab, but we've analyzed only a few dozen samples so far. That's a far greater proportion than is cited in your paper. Perhaps it's merely random ill fortune! But it will be difficult to distinguish between false (trace) positives due to CaMV and trace positives due to contamination of one commodity by another (e.g. contamination of maize by soy), since both will fail the variety-specific tests.

How will new corn varieties that carry an EPSP synthase gene (cDNA) from corn at a higher copy number than wild-type corn be detected by PCR?

The comments on using 35S promotor to detect can result in false positives due to CaMV were interesting to me. It is a concern since the USDA is considering validation of labs on this method and may be unaware of this significant drawback.

One answer to Johns question: Since the new transgenes will be introduced into a new genomic location, a set of primers which span the trangene genomic border, would be specific to the new insertional event. With this strategy, the specific test could have two primer sets (one for each side of the insertion) and interrogating for the presence of both amplicons would increase the specificity and accuracy of the analysis. Since each amplicon is single copy, they should be useful for quantitative PCR as well.

One concern that I have with the 35S and nos approach is that the second-generation biotech corn and soybean products (both agronomic and value output) are in general not going to have either the 35S or nos sequence. We expect to start seeing some of these products on the market next planting season although the planting will be low acreage.

Hybrid corn seed containing a B.t. or another trait is typically developed by introduction of the trait from one parent. As a result, segregation for the trait will occur in the subsequent grain (i.e., 25% homozygous positive, 50% heterozygous, 25% homozygous negative). Thus, 25% of the grain from a GM field will be non-GM and will not contant the transgenic DNA. How is this factored into the quantitative analysis by PCR?

Dirk is exactly correct in his concern that a testing routine based on 35S and NOS as a genral screen will not be able to detect coming events which lack these regulatory sites. In that case, the number of PCR tests, and the costs, will have to escalate in order to include primers for those new regulatory constructs.

As Chuck has indicated with corn and a unstacked event one will get a 1 : 2 : 1 genotypic distribution in the grain in a field. With a stacked event such as the Mon 810 and T25 one can get 9 different genotypes and approved and unapproved EU kernels in the same field. We expect more of the stacked events in the future and need to be prepared to handle those, as they become available.

It was stated at the USDA/GIPSA meeting in Kansas City (February 25, 2000), that obtaining a grain sample with the 1 : 2 : 1 genotypic ratio then the ratio of transgene to house keeping gene averages out. Which brings up several questions: Does anyone have the experimental data (not assumption) that the homozygous and heterozygous grains will behave in a PCR reaction to give the expected averaged result? Should the reference materials used for the validation of the PCR reaction be the grains in the genotypic ratio expected in a farmer’s field? Although sampling is probably not a part of the AACC plans, if the sample is not sufficient to contain the 1 : 2 :1 ratio of the grain in a field can the analytical method provide quantitative data?

Heinz Schimmel at the USDA/GIPSA meeting indicated some problems with the housekeeping gene in some of the European studies. Are there plans for selecting a housekeeping gene for the AACC studies or will that be left open for the participating laboratories?

I just logged in for the first time and I hope this works. My name is Gert van Duijn. Within the TNO Food and Nutrition Research Institute in the Netherlands I am active as Product Manager GMO Foods and therefore heavely involved in GMO detection. We develop methods based on Protein as well as DNA detection. For the DNA detection using PCR we developed specific methods for various transgenic crops authorized in Europe. In routineous control we always run these types of specific PCR methods in parallel with 35S screening methods. Further of course , we discriminate between qualitative and quantitative assays. I agree with Andreas that we should be very carefull with quantitative TAQMAN results because of the problem of copy numbers. Next time more.

Gert

Dirk, could you please give a bit more detail on what the problems with the house-keeping genes are? When you use the term "housekeeping genes", I'm assuming you are talking about the sequence information used as internal controls for transgene detection and quantitation.

I believe that we would be better served to get Schimmel involved in the discussion or perhaps either vanDuijn or Wurz could address the problems that Schimmel discussed in Kansas City.

As for the housekeeping gene: all of the commercial labs that are doing quantitative analysis today in the US, Europe and Japan that have shared their quantitative calculations with me are doing quantitative GMO analysis based on a relative ratio. A ratio between the number of copies of transgenic sequences (usually a 35S) and the number of copies of a housekeeping gene (native gene such as invertase in corn). There have been questions as to different number of copies of a housekeeping gene in for instance French Maize and US maize. As well as the questions raised by Schimmel and I quote "Investigations indicate differences in degradation behavior of target sequences (transgenic 35S and/or EPSPS and natural lectine gene) commonly used for the PCR detection of RR soybean”. This is more of a sample preparation problem but needs to be considered to develop a solid quantitative method.

Comment to Beth: It is correct that in case of trace contamination distinction between a CaMV contamination and a real GM-contamination is very difficult - there is no proper solution except doing many replicates of the analyses or scaling up extraction.

Comment on the coming generation GMO lacking 35s or nos sequences: We and others are on the way to develop DNA-array based techniques ("Chips") to have more multiplexed screening abilities to match this issue - this will help to reduce the analytical effort.

Comment of Dirk concerning the housekeeping gene: Indeed I think that either (performance-) criteria for the use of the "housekeeping" gene or instead a fixed "housekeeping" gene should be envisaged. By the way the reference gene does not need to be a houskeeping gene in strict sense - it should be specific, and as invariable as possible with respect to copy number and allelic variation.

Question to Dirk: What experiment would you suggest to find out if the hypothesis on "averaging out" of the 1:2:1 ratio in bulk samples is true?

I fully agree with comments made so far concerning the implications of multiple copy numbers on the quantification of GMO's. As stated a set of screening methods could be used to overestimate in the worst case the GMO content which would be beneficial for consumer protection. However it would be important for international trade to finally come up with a figure which gives the true content, e.g. by using trait or event specific sequences. The DNA separation from seeds to grain certainly takes place and there needs to be international consensus whether and how to correct for that. I am not aware of any detailed study trying to find out what really happens. A study could for instance be performed in such a way that one simply uses a batch of well defined seeds (as we use it in the meantime for CRM production) and to grow it on a field. Afterwards single grain DNA analyses need to be carried out. During my presentation in Kansas I showed our findings with concerning the influence of the production parameters of the RR soybean reference materials on the quantitative results. There have been two series of RR soybean CRM's consisting of 0, 0.1, 0.5, 2 % and 1 %, 5 % materials respectively which were produced under slightly different conditions. The production conditions were such that they can occur in industrial food processing (addition of water to make a slurry and mixing). Both series are based on the same batch of raw materials and contain the amount GMO specified in the certificate. Users of our reference materials made us aware of the fact that the two series are not consistent and that the 1 % material almost behaved like the 2 % material in quantitaive PCR using e.g. the 35 S promoter for screening and the lectine gene as the housekeeping gene. Consequently we launched a study and investigated the problem in more detail. Using the two official methods for RR-soybeans in Switzerland, one targeting to the 35 S promoter and one being RR specific, i.e. in the EPSPS region, and using lectine gene for the correction we got for the materials of the two series discrepancies of a factor 2 to 3 in the quantiative result. The amount of lectine was more or less constant. This means that the two different target sequences decompose to a different degree. On the other hand the two series were very consistent in the evaluation of the SDI ELISA for EPSPS protein. We then started to make a third series of all concentration levels still working in an aquous slurry but we cooled the mixture. The base materials were the same as for the first two series. Again the two Swiss official methods were used. All three series were analysed in the same laboratory in a short time. The third series has been fully consistent and the discrepancies between the two target sequences disappeared. This means that the DNA degradation processes leading to the effects with the first two series have been stopped by cooling. However if industry is going to process soybeans they are not going to cool the slurry after water addition. Therefore these DNA degradation processes will influence the quantitative result significantly if target and housekeeping gene sequences are not chosen in such a way that the degaradation behaviour is the same. In maize we have not observed these problems and the series are consistent although we also prepared two series under the same conditions as the soybeans. Therefore the above mentioned effects seem to be species perhaps also trait or even event specific. As a consequence of the above such effects have to be taken into consideration when methods are validated. Getting consistent and agreeing results in a ring trial using only one type of material does not proof that the method always gives the correct result in an absolut sense especially if the calibrants and the blind samples originate from the same series. According to our oppinion various materials with known GMO content (unprocessed and processed) need to be investigated. A g

This part of my previous message was obviously lost

According to our oppinion various materials with known GMO content (unprocessed and processed) need to be investigated. A good method would find back the GMO amounts added. In the future we will therefore try to provide two types of materials: a series of materials with as much as possible intact DNA and a material of the same series but treated e.g. with water or heat. If a method finds back the same result in both types of materials, which were originally exactly identical, it should work, but if it does not the method should not be used. Since up to now such method validations have not been done we are currently pushing a uniform system of methods and reference materials which may give results deviating somewhat from the truth but which will give the same result everywhere.

Hello everyone. I find the discussion interesting, but also feel the need to clarify some information. GIPSA is in the process of developing a program to accredit laboratories that use PCR for the analysis of biotech grains. The accreditation program will involve review of the facility, review of SOP's, analysis of challenge samples, on-site audit, and participation in a monitoring program. Laboratories can request accreditation for qualitative non-event specific testing and/or qualitative event-specific testing. I have some concerns with quantitative analysis by PCR, related to both the technology and the genetics, so for now we do not plan to accredit laboratories for quantitative testing.

The accreditation program is performance based; GIPSA will not specify either the methodology or the technology to be used by the laboratory. Some people have questioned this approach, but given the various approaches people are using, and lack of uniform agreement in this area, we concluded that assessing performance, along with the other criteria, will help address the short-term concerns. From discussions with many laboratories, I know that many are using the promoters and/or terminators to screen for the presence of biotech grain. That is up to the laboratory, and if they can meet our performance criteria, along with everything else, they will be accredited.

I don't want to go on too long here, if you have more questions let me know. I do have a question that I don't believe has been addressed in the discussion:

Given comingled corn of unknown origin that potentially could contain event T-25 (Aventis Liberty Link), MON810 (Monsanto Bt) and the MON810/T25 hybrid; can PCR definitively identify which particular events are present?

Don

Don

I am having trouble understanding the utility of qualitative PCR. Is this simply to determine whether there is any GMO present in support of a "non-GMO" claim, or do you envision this type of testing being done in conjunction with Identity Preservation systems? Have you decided which events will be part of the challenge sample set? Will you also prepare mixtures of events? At different proportions? Have you decided at what concentration you will mix GMO with non-GMO?

Jim

Jim, apparently qualitative testing does meet some customer needs because laboratories are being asked to provide qualitative testing. I can see that qualitative for the presence of biotech grain will help in deciding on the appropriate market channels. In addition, qualitative testing for the presence of unapproved events, depending on the market, could also be valuable. In the past testing has been the final "answer." In this arena, testing (PCR or ELISA) should be part of an overall system, whether it is a formal IP system, or something else in place to provide some protection for both the producer and buyer.

With respect to accreditation of laboratories, all commercially available events approved in the U.S. will be in the challange samples. The challenge samples will generally contain a mixture of events at various concentrations. Unfortunately, obtaining appropriate reference materials to prepare samples has proven to be more difficult than anticipated, so everything is on hold until we get the proper materials.

Don