Jump to content

  • Log in with Facebook Log in with Twitter Log in with Windows Live Log In with Google      Sign In   
  • Create Account

Submit your paper to J Biol Methods today!
Photo
- - - - -

qPCR for telomere length measurement - efficiency issues


  • Please log in to reply
40 replies to this topic

#1 dtae

dtae

    Enthusiast

  • Active Members
  • PipPipPipPipPip
  • 59 posts
1
Neutral

Posted 17 August 2010 - 08:49 AM

Hello,

I am trying to get a qPCR assay to measure human telomere lengths up and running (http://www.ncbi.nlm....les/PMC2647324/). It uses SYBR green chemistry on an iCycler iQ (BioRad) machine and is multiplex (telomere and control single-copy gene amplified in same tube) by using some fancy primer design. My efficiency values are tending to be too low (often <80%). I think the first advice most people would give me is to redesign my primers, but given some complex primer design and that I am trying to replicate the published protocol which others have used successfully, I am hoping to stick with these primers. What other things would you recommend first to try to fine tune to increase my efficiency values? I have tried increasing and decreasing my annealing temperature a few degrees without much luck.

There are some other complexities to the assay which I am trying to spare you, but please either check the above link, or let me know if there is any more information I can provide to help you help me.

Thank you very much for your time.

Dan

#2 dtae

dtae

    Enthusiast

  • Active Members
  • PipPipPipPipPip
  • 59 posts
1
Neutral

Posted 17 August 2010 - 04:23 PM

As a follow up, I have been told by a few others in my building that worrying about efficiencies (as calculated from standard curves) is not needed as long as my R2 is good and the traces and melt curves look ok. I've been told that the other labs who claim consistently good (90-110%) efficiencies across runs are probably exaggerating.
How much should I be worrying about efficiency values?

#3 Nathalie Allard

Nathalie Allard

    member

  • Members
  • Pip
  • 3 posts
0
Neutral

Posted 23 August 2010 - 05:00 AM

Hello!

I'm also trying to get Cawthon's protocol running in our lab. I have the problem that my efficiencies are too high. Did you recognized primer dimer? At the telomere Tm my NTC are coming, but far enough from the rest to worry about.

The beta Globulin Primer are not so good. Those for Albumin as SCG are better and R. Cawthon published some new Primer in a blog in this Forum (somewhere). If you want I can give those primers tp you.

Which concentration of SCG and telomere primers do you use?

First of all the efficiencies of telg/telc an SCG should be the same so you are able to calculate your T/S! Normally I find 80-90% is a good efficiency. 100% does not reflect reality. efficiency over 100% stands for contamination or primer dimer in the reaction. I would use result until 105% efficiency. After all a non-varying efficiency is the best.

Good luck, Nathalie

#4 dtae

dtae

    Enthusiast

  • Active Members
  • PipPipPipPipPip
  • 59 posts
1
Neutral

Posted 23 August 2010 - 04:10 PM

Hi Nathalie (and others),

I am very pleased to have someone else who understands the plight I am in :)

I was previously fooling around with using the Fermentas SYBR green with Fluorescein MasterMix (on Bio-Rad iCycler iQ). With that I tended to get high efficiency values as well (~110-140, but also rather inconsistent and sometimes dropping down into the 80s). Are you using Cawthon's original homewbrew mastermix recipe or an off the shelf mastermix? I assume you are trying with the 2009 protocol right?

I've avoided using the beta-globin primers too--mainly because there are several unconfirmed SNPs overlapping with those primers. Glad to hear I am at least using the better primer set on more empirical grounds. I'll look for thew new Cawthon primers posted on the blog and post them here if I find them (if not I would of course appreciate if you can post).

Using the originally recommended 900nm for all primers, I found that the telomere peak on melting curve tends to be almost non-existant on the highest concentrations on my standard curve (88ng genomic DNA). I have switched to 500 for telomere primers and 300 for albumin and that has corrected the melt-curve problem.

What do you make of the strategy employed in Ehrlenbach et al which seems to yield much lower CV values? If I am understanding correctly, they seem to embrace (rather than battle) the low efficiency values and use them to help calculate the T and S values with LinRegPCR. I am experimenting with LinRegPCR myself now. So far it looks like mean efficiency of 89% for telomere and 79% for albumin (SCG). This interesting because efficiencies calculated with the standard curve method almost always yield higher values for the albumin amplicon.

The more I read and talk to people, the more I am beginning to believe that the 85 or 90 to 105 or 110 rule of thumb for standard curve calculated efficiencies is not particularly based in empirical reality. My current, perhaps incorrect, synthesis of the literature is that anything under ~105% with a clean exponential phase and melt curve is ok as long as efficiency values are similar across samples in the same amplicon (and if LinRegPCR is employed, perhaps not even needing similar values across samples).

Thanks for the feedback and looking forward to more.

Dan


Hello!

I'm also trying to get Cawthon's protocol running in our lab. I have the problem that my efficiencies are too high. Did you recognized primer dimer? At the telomere Tm my NTC are coming, but far enough from the rest to worry about.

The beta Globulin Primer are not so good. Those for Albumin as SCG are better and R. Cawthon published some new Primer in a blog in this Forum (somewhere). If you want I can give those primers tp you.

Which concentration of SCG and telomere primers do you use?

First of all the efficiencies of telg/telc an SCG should be the same so you are able to calculate your T/S! Normally I find 80-90% is a good efficiency. 100% does not reflect reality. efficiency over 100% stands for contamination or primer dimer in the reaction. I would use result until 105% efficiency. After all a non-varying efficiency is the best.

Good luck, Nathalie



#5 dtae

dtae

    Enthusiast

  • Active Members
  • PipPipPipPipPip
  • 59 posts
1
Neutral

Posted 23 August 2010 - 04:19 PM

Found the other primers:
http://www.protocol-...osts/14233.html

Have you had better luck with them?

I also wonder if the thermocycling profile Cawthon recommends is better for the iCycler as well...

On a related note, I have been extending my number of cycles out to 37 to better catch some of the low C amplicons. Because of the LinRegPCR suggesting I drop samples that haven't reached plateau phase, I am thinking of extending out to 40 or or 45.

D

#6 Nathalie Allard

Nathalie Allard

    member

  • Members
  • Pip
  • 3 posts
0
Neutral

Posted 24 August 2010 - 05:36 AM

Hello!

First I tried SybrGreen Supermix from BioRad, but the results were not reproducable (with three different SCG).

Today I gave EvaGreen a try (just both Albumin Primer as SCG, Hbgu is too bad). There I had quit good results with the Albugcr MM. EvaGreen shows much better results in the MC and the NTC don't show up. Albu MM was not good, the efficieny was too high (150%). But at all the T/S Ratio was nearly the same with both SCG. Tomorrow the same experiment again to verify the results.

Do you have the possibility to chek your results with southern blot or something else?

I also reduced the concentration of the Primer to 500nm for the telomere, 200nm for Albugcr and 300nM for Albu. For the SCG 32 cycles is ok. But you need more cycles for LinRegPCR, or?

What do you use as standard?

I didn't try the mentioned protocol for the LightCycler. I have a BioRad IQ but if I have still problems I will try it!

With LinRegPCR I have no experience. I downloaded it today to test it.

I hope the best to get it run!

Nathalie

#7 dtae

dtae

    Enthusiast

  • Active Members
  • PipPipPipPipPip
  • 59 posts
1
Neutral

Posted 24 August 2010 - 10:25 AM

I think I had problems with the Bio-Rad supermixes as well--although I would have to go back into my notes to see more. I think I didn't have much luck with EvaGreen either--but again would have to go dig up why (would be happy to for either if it would be helpful).

No plans to use a southern validation--although if I find time it of course would be a good idea. My impression of the literature is that it is beginning to be acceptable to publish without it.

I think the need for more cycles are in part because the low efficiencies result in higher Cqs. Also, LinRegPCR alogirthm seems to like it more. However, if I end up using the LinRegPCR method exclusively, it is probably not necessary as I won't need a standard curve with low c samples.

I've been using DNA extracted from myself as a standard. I considered using the O'Callaghan method of using an oligo as a standard--but I become worried about primer degradation, and lack of comparability between DNA extracts and oligo samples.

Best,
Dan

Hello!

First I tried SybrGreen Supermix from BioRad, but the results were not reproducable (with three different SCG).

Today I gave EvaGreen a try (just both Albumin Primer as SCG, Hbgu is too bad). There I had quit good results with the Albugcr MM. EvaGreen shows much better results in the MC and the NTC don't show up. Albu MM was not good, the efficieny was too high (150%). But at all the T/S Ratio was nearly the same with both SCG. Tomorrow the same experiment again to verify the results.

Do you have the possibility to chek your results with southern blot or something else?

I also reduced the concentration of the Primer to 500nm for the telomere, 200nm for Albugcr and 300nM for Albu. For the SCG 32 cycles is ok. But you need more cycles for LinRegPCR, or?

What do you use as standard?

I didn't try the mentioned protocol for the LightCycler. I have a BioRad IQ but if I have still problems I will try it!

With LinRegPCR I have no experience. I downloaded it today to test it.

I hope the best to get it run!

Nathalie



#8 dtae

dtae

    Enthusiast

  • Active Members
  • PipPipPipPipPip
  • 59 posts
1
Neutral

Posted 26 August 2010 - 11:04 AM

An update, using LinRegPCR, I can get T/S CVs on replicates down in the ~10% range. However, T/S values vary substantially between different DNA concentrations and across runs. See the attached for more details.Attached File  10-08-25-1 qpcr set-up - summary.xls   54.5KB   350 downloads

#9 dtae

dtae

    Enthusiast

  • Active Members
  • PipPipPipPipPip
  • 59 posts
1
Neutral

Posted 26 August 2010 - 02:25 PM

You don't happen to know how the albugcr primers are supposed to be better do you? What kind of difference do you notice between it and the original albumin primers? I figure I'll ask you before bugging Cawthon more.

Thanks,
Dan

#10 Nathalie Allard

Nathalie Allard

    member

  • Members
  • Pip
  • 3 posts
0
Neutral

Posted 01 September 2010 - 02:03 AM

Hello!

I run the PCR product on a 3% gel. ALbu has a product in the NTC which has the same size like the samples. Albugcr NTC is empty. I suppose that the Albugcr primer are nearly on the same place on the sequence because they have the same size on the gel.
To set this method up is not so easy. I'm not getting repducable results! I don't know why the variation between the runs is so high.

Did you try the cycling program for the LightCycler also on the BioRad? If yes, how does it work? This will be my next try for establishing this method. I will also try the selfmade mastermix in the paper. I'm sure that DTT and betaine could be very helpful in the PCR because of the GC clamp.

Attached is a gel I made!

Attached Files



#11 dtae

dtae

    Enthusiast

  • Active Members
  • PipPipPipPipPip
  • 59 posts
1
Neutral

Posted 03 September 2010 - 11:42 AM

Hi Natalie,

A few notes on reproducibility of results (hopefully I am not repeating things you have already told me or we have already talked about in the above discussion):
1. Be very careful that DNA concentration is the same across all samples. For example, I think that using 10ul of template at 2ng/ul will not work as well as 2ul of 10ng/ul template. I think this is because there tends to be more error introduced in diluting down to lower concentrations.
2. Be careful about DNA quality. Aliquot DNA you intend to analyze multiple times and freeze it so you don't have it sitting in the fridge degrading or multiple freeze thaws of the same sample. Some of my preliminary evidence suggests that freeze/thaw or sitting around in the fridge causes problems.
3. Making mini-master mixes for replicates might also help. That is, I combine my DNA and mastermix for ~3.2x reactions worth in one separate tube and then aliquot out into the qPCR plate using a multichannel into three seperate wells (with the .2x excess to compensate for mix getting stuck on sides of tubes/tips).
4. DTT degrades readily in light and with freeze thaws - if you decide to use it be careful. Also, I have found it helpful to just make my own mastermix using Cawthon's recommendations and make a whole bunch of single-use aliquots.

I didn't try the LightCycler conditions--in retrospect, I think cycling conditions are different on LightCyclers because they go through the cycles faster than a normal thermocycler.

Thanks for the gel images. I wouldn't worry about positives in your negative controls as long as the initial concentrations (i.e. Cq values) are very low.



I had the following correspondence with Richard Cawthon about some of these issues which he said would be ok for me to post here and might be of interest:

"The posted primers eliminate a 4-bp overlap and perfect match at the 3’ ends between one telomere primer and one albumin primer from the MMQPCR paper, which may sometimes cause a problem with a telalb primer dimer appearing.

I’ve changed the albumin primers again however, and now recommend:

Albugcr2
cggcggcgggcggcgcgggctgggcgg CCATGCTTTTCAGCTCTGCAAGTC

and

Albdgcr2
gcccggcccgccgcgcccgtcccgccg AGCATTAAGCTCTTTGGCAACGTAGGTTTC

which contain a few internal mismatches from native sequence, in order to minimize primer dimers. These work great, under the following conditions:

Final concentrations of primers in the PCR:

tel g: 200 nM
tel c: 700 nM
albugcr2: 700 nM
albdgcr2: 500 nM

Thermal profile:

95 x 15 min
-----------------
94 x 15 sec
49 x 60 sec
Repeat for a total of 2 cycles
-----------------
85 x 20 sec
59 x 30 sec
Repeat for a total of 4 cycles (this stage is amplifying the telomere product, without amplifying the scg product)
-----------------
94 x 15 sec
59 x 30 sec with signal acquisition
84 x 30 sec
85 x 20 sec with signal acquisition
Repeat for a total of 30 cycles
------------------
Melting curve (Optional), from 59 to 95 degrees, 0.5 degrees per step, 5 sec per step

Regarding telomere amplicon melting peaks disappearing at high input DNA concentrations: this is due to the GC-rich scg product “stealing” the SYBR Green I. SYBR Green I is not saturating in the reaction (it would inhibit the PCR if it were saturating), and more dye binds as the GC-content of the ds-DNA rises. So if you stop the cycling after you’ve plateaued the telomere amplicon, but when you’ve made only half the plateau amount of the scg amplicon, then you’ll see the telomere amplicon’s melting peak just fine, even in the high DNA input reactions.

I haven’t tried LinRegPCR yet.

My understanding is that having separate standard curves for each distinct amplicon prevents even large differences in amplification efficiency between those distinct amplicons from being any problem at all. However, if one finds that amplification efficiencies for a single amplicon are varying significantly from sample to sample in the same run, then one definitely needs software to correct for those sample-to-sample efficiency differences (or, alternatively, one needs to re-optimize one’s DNA extraction and PCR conditions).

Sincerely,

Richard"

Hello!

I run the PCR product on a 3% gel. ALbu has a product in the NTC which has the same size like the samples. Albugcr NTC is empty. I suppose that the Albugcr primer are nearly on the same place on the sequence because they have the same size on the gel.
To set this method up is not so easy. I'm not getting repducable results! I don't know why the variation between the runs is so high.

Did you try the cycling program for the LightCycler also on the BioRad? If yes, how does it work? This will be my next try for establishing this method. I will also try the selfmade mastermix in the paper. I'm sure that DTT and betaine could be very helpful in the PCR because of the GC clamp.

Attached is a gel I made!



#12 dtae

dtae

    Enthusiast

  • Active Members
  • PipPipPipPipPip
  • 59 posts
1
Neutral

Posted 08 October 2010 - 03:48 AM

I have been having some problems with edge effects - my melt curves for edge (and especially corner) wells look different than those from inner wells. Cq values also tend to be higher in edge and corner wells, and correspondingly, excluding sample replicates in the edge wells gives me a lower CV. This is using the original albumin/telomere primers in Cawthon 2009. I am considering using the modified albumin primers above which Cawthon gives lower recommended acquisition temps for and/or lower cycling temperatures a bit more.

Has anyone else had problems with edge effects with this assay? I have had the edge effects both using the iCycler iQ and iCycler MyiQ.

Thanks,
Dan

#13 dtae

dtae

    Enthusiast

  • Active Members
  • PipPipPipPipPip
  • 59 posts
1
Neutral

Posted 08 October 2010 - 10:15 AM

Also of note to those considering using the new scg primers recommended by Richard Cawthon - there is one uncomfirmed SNP (rs75523493) that overlaps with those primer sequences. If this SNP is real, it might cause some noise and perhaps systematic bias.

#14 Tom-S

Tom-S

    member

  • Members
  • Pip
  • 4 posts
0
Neutral

Posted 25 November 2010 - 01:35 AM

Dear Nathalie and Dan,

I'm also intending to implement this technique in my new lab. I carefully read everything you have posted above but i'm still trying to spare out some troubles.
My first tries were made with the initial 2004-protocol from R.Cawthon (home-made master mix) and failed. I then go on with Biorad SYBR-mastermix using the following conditions:
_Telomere f: 270nM
_Telomere r: 900nM,
95°, 3min
95°, 15s
54°, 2min, 30cycles (indeed, i increased it)

_AT1 (as SCG) f: 500nM
_AT1 r: 500nM
94°, 3min
95°, 15s
60°, 1min, 40cycles
I used separated plate for tel and SCG
Standard DNA concentration were from 10ng to 0.31ng
I'm using low DNA concentration of 1ng/µL (directly from my stock sample DNA)(I know that it can increase sample to sample variations due to pipetting error!)

After two separate experiments, I found the following results: Tel efficiency at 67 and 65% respectively with R˛ around 0.995, slope around -4.492. AT1 efficiency 88%, 85% respectively with R˛ around 0,994, slope around -3,652. The melting curves were good, indicated no primer dimer or non-specific amplicon. I found out very low sample to sample variations in term of Ct.
Nevertheless, I'm still wondering why I can't reach better tel efficiency?

Thanks again for the already published useful informations and thanks by advance for your answer
Tom

#15 dtae

dtae

    Enthusiast

  • Active Members
  • PipPipPipPipPip
  • 59 posts
1
Neutral

Posted 25 November 2010 - 11:06 AM

I'm not sure what's up, but hear are a few thoughts:
1. Some people are publishing with low efficiencies and using LinRegPCR to correct for this. The greater than 90% efficiency rule of thumb seems to be just that, a rule of thumb with limited empirical basis. The fact that the E between T and S amplicons are similar is reassuring.
2. The sample you are using to make your standard curve might have inhbitors/enhancers in it. Try to exclude low or high C samples and see what it does to efficiencies. You might also try using different samples to construct your standard curve with.
3. See what kind of well-specific efficiencies you are getting. You might find that those are substantially different than using the standard curve method.
4. Beware of well-position effects--they can cause unpredictable and unstable efficiencies.
5. Ct CVs really don't tell you a great deal. T/S ratio CV is what really matters here.


Dear Nathalie and Dan,

I'm also intending to implement this technique in my new lab. I carefully read everything you have posted above but i'm still trying to spare out some troubles.
My first tries were made with the initial 2004-protocol from R.Cawthon (home-made master mix) and failed. I then go on with Biorad SYBR-mastermix using the following conditions:
_Telomere f: 270nM
_Telomere r: 900nM,
95°, 3min
95°, 15s
54°, 2min, 30cycles (indeed, i increased it)

_AT1 (as SCG) f: 500nM
_AT1 r: 500nM
94°, 3min
95°, 15s
60°, 1min, 40cycles
I used separated plate for tel and SCG
Standard DNA concentration were from 10ng to 0.31ng
I'm using low DNA concentration of 1ng/µL (directly from my stock sample DNA)(I know that it can increase sample to sample variations due to pipetting error!)

After two separate experiments, I found the following results: Tel efficiency at 67 and 65% respectively with R˛ around 0.995, slope around -4.492. AT1 efficiency 88%, 85% respectively with R˛ around 0,994, slope around -3,652. The melting curves were good, indicated no primer dimer or non-specific amplicon. I found out very low sample to sample variations in term of Ct.
Nevertheless, I'm still wondering why I can't reach better tel efficiency?

Thanks again for the already published useful informations and thanks by advance for your answer
Tom






Home - About - Terms of Service - Privacy - Contact Us

©1999-2013 Protocol Online, All rights reserved.