Changsoo Lee, Jaai Kim, Seung Gu Shin, Seokhwan Hwang ∗

Absolute and relative QPCR quantification of plasmid copy number in Escherichia coli

Real-time QPCR based methods for determination of plasmid copy number in recombinant Escherichia coli cultures are presented. Two compatible methods based on absolute and relative analyses were tested with recombinant E. coli DH5
harboring pBR322, which is a common bacterial cloning vector. The separate detection of the plasmid and the host chromosomal DNA was achieved using two separate primer sets, specific for the plasmid -lactamase gene (bla) and for the chromosomal d-1- deoxyxylulose 5-phosphate synthase gene (dxs), respectively. Since both bla and dxs are single-copy genes of pBR322 and E. coli chromosomal DNA, respectively, the plasmid copy number can be determined as the copy ratio of bla to dxs. These methods were successfully applied to determine the plasmid copy number of pBR322 of E. coli host cells. The results of the absolute and relative analyses were identical and highly reproducible with coefficient of variation (CV) values of 2.8–3.9% and 4.7–5.4%, respectively. The results corresponded to the previously reported values of pBR322 copy number within E. coli host cells, 15–20. The methods introduced in this study are convenient to perform and cost-effective compared to the traditionally used Southern blot method. The primer sets designed in this study can be used to determine plasmid copy number of any recombinant E. coli with a plasmid vector having bla gene. © 2005 Elsevier B.V. All rights reserved

Changsoo Lee, Jaai Kim, Seung Gu Shin, Seokhwan Hwang ∗

Absolute and relative QPCR quantification of plasmid copy number in Escherichia coli

Real-time QPCR based methods for determination of plasmid copy number in recombinant Escherichia coli cultures are presented. Two compatible methods based on absolute and relative analyses were tested with recombinant E. coli DH5
harboring pBR322, which is a common bacterial cloning vector. The separate detection of the plasmid and the host chromosomal DNA was achieved using two separate primer sets, specific for the plasmid -lactamase gene (bla) and for the chromosomal d-1- deoxyxylulose 5-phosphate synthase gene (dxs), respectively. Since both bla and dxs are single-copy genes of pBR322 and E. coli chromosomal DNA, respectively, the plasmid copy number can be determined as the copy ratio of bla to dxs. These methods were successfully applied to determine the plasmid copy number of pBR322 of E. coli host cells. The results of the absolute and relative analyses were identical and highly reproducible with coefficient of variation (CV) values of 2.8–3.9% and 4.7–5.4%, respectively. The results corresponded to the previously reported values of pBR322 copy number within E. coli host cells, 15–20. The methods introduced in this study are convenient to perform and cost-effective compared to the traditionally used Southern blot method. The primer sets designed in this study can be used to determine plasmid copy number of any recombinant E. coli with a plasmid vector having bla gene. © 2005 Elsevier B.V. All rights reserved

How did you generate the standards?

Have you used the appropriate standard curves for each primer pair?

Have you thoroughly quantified both standards (if these are two different standards)?

How do the amplification curves look like (make sure there is no artefact leading to grossly wrong ct values)?

Assuming that:

1) no one made a math or pipetting error in making the standard curves

and

2) your primers can't be amplifying other targets in the genome or in any contaminating bacteria/fungi/whatever

and

3) you aren't accidentally amplifying genomic DNA with one primer pair

then you have something weird.

My money is on either genomic DNA amplification or math error.

What do your controls look like?

First, than you for your answers!

Jochen Wilhelm :

I purified each PCR product and made 10-fold dilutions with it. I checked the PCR products with gel and found only one product. I didn't check the products of qPCR but the melting curves showed 1 peak only.

Katie A Burnette

I checked the maths with another personn, they seem ok. Controls are ok.

I have a third primer pairs for the same gene which I didn't try yet: I'm going to check what kind of result I get with it and compare.

I'll let you you know if I figure out what was the origin of my issue!

Thanks again

How have you quantified the standards?

How do the amplification curve look?

Jochen Wilhelm

I used nanodrop to quantify standard concentration. I first diluted at 10 ng/uL (checked again with nanodrop) and up to 10e-8 by 10-fold dilutions (90+10).

The amplification curves seem to look good (I attach screen copies).

Yesterday I tried a third primer pair on the gene and got a third different quantification of the copy number. So I still don't understand the problem... maybe it's not good to use directly the PCR products for the standard curves and it would be better inside plasmids?

Ok, thank you. I still have more questions, hoping to find the reason for your problem.

Do you use standard conc in "ng/µl" or have you determined "copies/µl" ?

What are the ct values of your samples (are they >28)? If so, did you check the linearity of the standard curve at cycles >28?

What about the intergrity of the RNA? Low integrity might have an impact on the observed quantity from differentprimer-pairs (depending on what RT-primers are used and on where on the mRNA and how long the PCR products are).

Hi,

If you want to quantify expression levels, you will have to normalize your expression.

You need to control for:

- input

- reverse transcriptase: which is very variable

So, either:

- you normalize for multiple stable reference genes (by preference)

- you could normalize on spike ins (less preferable, as you do not account for differences in input)

Even when you normalize, you can still take standard curves to assess efficiency and linear range.

If you determine a standard curve from a plasmid:

- linearize your plasmid

- use by preference a matrix comparable to your samples (f.e. spike in your plasmid in a similar buffer and treat it the same way as your samples)

- keep in mind that determining copy numbers from nanodrop is not a precise technique, only an estimate. (But this is not that important when you normalize)

What also could have an effect, is RNA degradation, depending on on which side of the transcript your amplicon is. This could (partly) be responsible for the difference you observe.

Comparing several genes with qPCR is hard, as the sequence, amplicon size, primers, RNA degradation, RT efficiency and so on will have an effect.

I can recommend you the MIQE guidelines: https://academic.oup.com/clinchem/article/55/4/611/5631762

Your amplification curves look beautiful!

Ideally, you should have ONE standard that all of your primers could amplify from, ideally a linearized plasmid. Using PCR product as a standard isn't ideal for several reasons, but mostly because the risk of cross-contaminating your workspace is really high.

You should base your standard on number of copies/microliter.

If you know the exact basepairs of your standards, you can correct your math on your standard curves to account for exact copy numbers.

Sounds like it's a math issue and not a technique issue.

If you're calculating [template] in ng.ul-1, but your templates have different lengths, this will obviously contribute some error: an 80bp amplicon at 10ng.ul-1 will contain three times as many target molecules as a 240bp amplicon at 10ng.ul-1.

I am also confused about your '24x factor': do you literally mean one gives a value 24-fold greater than the other? Because that's like, almost five cycles of difference, and your dilution series amp curves look remarkable similar. At best I might accept 2.4-fold difference, based on those curves, and that brings us into the sort of ranges affected by amplicon length as noted above. How are you calculating your Cq values?

Also (unrelated, but generally useful) remember that cDNA is usually single stranded, while PCR/plasmid templates are double stranded, so it takes ~1 cycle for cDNA qPCR to catch up with standards. Adjust your estimates of copy numbers accordingly.

Jochen Wilhelm John Hildyard Katie A Burnette

Thank you for your answers!

I used concentration (ng/uL) to establish the standard curves. So the software calculated the concentrations of the samples from the Cq with the corresponding curves, which I finally turned into copies/uL using the formulae. Do you think converting ng/uL into copies/uL at first or at the end changes something in the result ?

In the example I gave you, the sample gives Cq=26.41 (calculated concentration 5.289E-7 ng/uL) with the first primer pair, and Cq=28.22 (calculated concentration 4.074E-8 ng/uL) with the 2nd primer pair. The duplicates are ok and I thought the Cq are not too high to be reliable. After calculations, I found 9176 copies and 376 copies, respectively...

I didn't specify first but in my example, I actually used the same forward primer, and two different reverse. So the targeted region is somewhat similar, one amplicon being twice longer. Molar mass are 34582,8 and 65070,8 g/mol, respectively. Also I have now checked the qPCR product on gel : there is one product only for each, at the expected size. So I don't think there is a contamination.

But I still don't find a satisfying explanation, though !

Strange.

ng/µl refers to what, precisely? To the amplicon sequence? I hope you purified the PCR products before measuring their concentration (you still did not provide the details how the standard were generated!).

Degradation of your RNA might be a possible cause, because the longer product gives the lower quantity.