Can you measure your PCR efficiency? I do not believe it is poor quality primers that is your problem. HPLC purification will only remove the shorter, incomplete sequences in the primers, and improves specificity, but not necessarily efficiency, in my experience.

If your standard curve is off, it suggests that your PCR isn't completely optimized/efficient, as with each cycle, it is not doubling the product. Designing new primers, and/or altering cycling conditions could resolve your problem

I don't believe you have a problem here at all actually. I think what this is showing you is just simply that your assay is below the linear range of quantitation at concentrations at or below standard #5. You should add another standard with a concentration above that of standard 1 and use those 5 standards to define the linear range of your assay. Honestly, given the concentrations you are working with the qPCR data you show is completely what I would expect. It's really too bad that you have invested so much time in troubleshooting - often we are trained to think that when unexpected results occur that it is something we have done wrong but much of the time the data is simply telling us that the assay, like all assays, have limits of detection both on the high and low end and that concentrations at or below #5 just can't be quantified accurately in this system. Here is why I think this is what's going on:

1. The curves for standards 6, 7, and 8 are clustered, clearly indicating that these concentrations are below the linear range. It's not unusual to find a standard like #5 which is intermediate - it doesn't cluster with the other "too low" standards but it still deviates significantly from the expected linear relation.

2. You've absolutely done every possible imaginable control and troubleshooting step. I can't possibly see how this has anything to do with your procedure, you've even investigated whether vortexing the sample differently has an effect! Probably time to just accept that this assays lower detection limit is at standard #4.

3. If I have understood your description of the procedure correctly then standard #5 contains a concentration of 0.65 pg of DNA per uL. The mass of the E. coli genome is ~0.0056 pg. So your 1 uL of DNA solution only contains about 110 genomes. It is critically important to remember that genomes and genes are quantized - they only come in multiples of 1. Consider:

If you have a DNA concentration of 0.0027 pg/uL you have "1/2" a genome (i.e. 1/2 a copy of the 23s RNA gene) per uL, but you can't have "1/2" a copy of a gene, so if you ran this on qPCR you would find that 1/2 your samples would show the gene detected and 1/2 wouldn't, because statistically there would be only 1 copy of the target in every 2 uL of solution.

Now if the concentration were 0.0056 pg/uL you would think every sample would come up positive, but that's not the case. Because the molecules are moving around in solution due to thermal motion every time you pipette out a uL of sample you are actually taking a statistical sample of the average composition of the mixture, so after many trials the average concentration would be 1 genome/uL, but sample by sample you would find that about 33% have either 0 or >1 copy and about 66% have exactly 1 copy. The exact stats depend on many factors including the temperature and viscosity of the sample.

4. PCR is strongly non-linear at lower template concentrations. This is because of the chemistry of what is going on in the tube, without getting into the mathematics and excess detail of it, consider that there are a number of steps to the PCR reaction and that each of these steps has a certain rate constant that varies depending on a number of factors. Remember, everything in the tube is just bumping around randomly due to thermal motion, so first the primer has to randomly bump into the target, then the polymerase has to randomly bump into the hybridized primer, etc... When there's lots of template around then all of these steps happen quickly relative to the hybridization time of the PCR reaction and things are nice and linear because if there's twice as much template then the primer is twice as likely to bump into it etc.... However, at very low template concentrations the reaction efficiency becomes highly dependent on temperature and time, i.e. does the 30 s of hybridization time give the primer enough time to actually find a template molecule? After the required time does every template molecule have a primer bound? Does every primer-template hybrid have a polymerase? Primers are also dissociating as well as associating during hybridization so if the Koff so high that primers tend to dissociate faster then more primers can associate? and so on.... So another long story short, PCR is highly non-linear at low concentration, which is another reason that your data looks to me very much like what I would expect from the experiment described.

In conclusion, quantitative PCR is really only useful for quantification where the sample contains about 1000 copies of the target (gene/genome) or more. That said, below this concentration the assay may be useful for detecting the presence or absence of a target sequence, but not its concentration.

To briefly address the comments of our colleagues above:

Florian C Priller

Florian C Priller

l will accept on your word that PCR reactions can be designed to achieve linear results at very low template concentrations, but I think you are being a bit hard on these folks by implying their assay is "badly designed". Regardless of whether it is possible in theory or with sufficient experience to push the limit of qPCR down to linear detection of low template concentrations in practice for most users the type of problems described by the original poster are quite typical - regardless of the biochemical reasons qPCR reactions just simply aren't observed to be linear at extremely low template concentrations. Therefore, without the specialized skill and effort to produce such highly accurate assays most labs will have to be content to simply define the linear and therefore useful range of the assay based on the data obtained. Unless the original poster has a need for quantification below the 0.65 pg limit is it worth it to them to redesign the assay?

Again, I do agree that I would like to see no signal in the NTC. Can you explain though why if there were bacterial contamination the Ct of the fifth standard is higher than expected instead of lower? In my experience contamination is indicated by a lower Ct value. Besides, although some contamination may be present it is giving a Ct value in the NTC of around 33 versus 28.5 in standard 5 so whatever contamination may be present is there at about 4% relative to the total DNA in sample 5 and should therefore account for no more than a variance of 0.06 or so in the Ct values of the standard, yet their deviation is clearly much greater than that.

agree with @Zachery R Belak

regards

I cannot thank Florian C Priller

I would like to place the product on an agarose gel and will do so at the beginning of next year when possible. Typically our genomic host cell DNA concentrations for our samples result between 0.001 to 0.1% with sample plasmid DNA concentrations anywhere between 1 to 10mg/mL. Is our standard curve range off scale from our target samples? My frustration is that this assay has been working in the past and has stopped. This has led me to believe the probe or master mix formulation has changed etc. because we are too consistently replicating this new phenomenon.

Again, I can't thank you all enough for your extremely valuable input, literally my job depends on it!

Florian C Priller

Florian C Priller

Heather Davidson A couple things to think about though: Has the method used for quantifying the standard changed or is the standard something you purchase commercially? It could be that if the standard were quantified differently or prepared differently this could be a cause of standard 5 no longer falling on the linear range of the assay. You are in a cGMP facility right? It should be possible to go back and check the lot numbers of every reagent used at any point in this procedure and see if there was one reagent in particular where you started using a new lot number at the same time these problems appeared. In critical applications like this I always make sure to buy a new batch of reagent before the old one runs out and always run them side by side to verify batch-to-batch performance for myself. It is entirely possible that the manufacturer made some slight (and they probably thought inconsequential) change in the product that is changing your results. It might not be possible to ever get the assay to work exactly as before. If you can, buy a commercial bacterial genomic DNA standard, and if you are quantifying/preparing the standard yourself please kindly give some details about how that is done as there may be something there that I can identify as the cause of the problem.

Am I to understand that the idea here is that you need to show that in your plasmid preparations which have 1 - 10 mg/mL total DNA that no more than 0.1 - 0.001% is genomic DNA? So if that is what you are going for here then consider that 0.001% of 1 mg/mL is 10 ng/mL or 10 pg/uL whereas standard #5 is 0.65 pg so in my opinion, and as you mentioned in your last post, I do think your assay is missing the target range by about an order of magnitude. I would start at the lowest concentration of genomic DNA that you want to test for in your plasmid preparation, say 0.001%, or 10 ng in a 1 mg/mL plasmid solution, then make dilutions going up from there. That way any sample with any Ct higher than the Ct of a reaction containing 10 ng can be said to have "less than 0.001% genomic DNA". In fact, standard #4 is already low enough in concentration that you can use that as your highest dilution standard. All you need to do is add one standard with higher concentration than standard #1 to get the 5 data points you need.

We have encountered issues in our labs from time to time where assays don't work because of some subtle change in reagents or lot numbers, and when it happens we just have to redesign the assay from scratch sometimes. Sometimes the linear range of an assay is just what it is and you have to report your detection range based on what you observe. So, bottom line, your assay in it's current form IS detecting bacterial genomic DNA at a concentration of 10 ng/mL (10 pg/uL) or 0.001% of total DNA (if total DNA is 1 mg/mL). So if that is lower limit of the range you are trying to detect, call it a success, add another standard above standard 1 to get your 5 data points, and have a merry christmas! (P.S. I realize it may be harder to convince management of this, but every assay has it's limit of reliable detection and in this case standard #5 is below that limit. Note I said reliable detection, because one instance of what can happen is that you might be able to detect that level of DNA in the past, but part of what you are likely seeing here is that as time goes on, that level of DNA is not reliably detectable from lot to lot of reagent etc.)

Good Luck.