ChIP seq - Different DNA size on agarosgel and Bioanalyzer

[Leeenaaa]

Dear all,

I'm performing chromatin imnmunoprecipitation and are shearing my crosslinked cells with a Bioruptor. After reverse crosslinking 65C, 5 h followed by DNA purification with E.Z.N.A cycle pure kit (column based) I analyze my Input chromatin (no IP) on a 1 % agarosgel and with Agilent DNA high sensitivity kit. The problem is that these two assays give totally different results. Agarosgel shows DNA library size of avarage 300 bp while bioanalyzer shows an avarage of 2500 bp?!?!

Which assay should I trust? What could be the reason for this difference?

See attached document.

[dariober]

Hi,
Unfortunately I don't have an answer to your question, but I would also like to hear about it. I've seen before similar inconsistencies between gel and Agilent. If you ask me, I believe the gel more than the Agilent. I suspect the Agilent is more sensitive to contamination, overloading, or fragments out of the recommended chip range (and you have a long smear). Did you try to quantify your DNA by Nanodrop? Just to see if the Agilent is getting the concentration right (although the spectrofotometer has also its issues with contamination...).

If you find something... please post!

Good luck
Dario

[pmiguel]

Difficult to say what is going on for a number of reasons. But we can probably draw a few conclusions.

First, the amount of DNA in your 2.5 kb bioanalyzer peak is very small -- 0.1 ng/ul. You don't say if the sample on the bioanalyzer is at a different dilution than the sample on the gel.

As you wrote above your agarose gel shows a very intense peak at 300 bp that is not present at all in your bioanalyzer High Sensitivity chip chromatogram--except possibly a tiny blip of a peak at 266 bp.

Here are a couple of possibilites:

(1)The intense "300 bp" peak is single-stranded DNA and/or RNA. But this would require that whatever fluor the DNA High Sensitivity kit uses, does not detect ssDNA and/or RNA.

[Note added 3/2/2012 -- 13 months later: I now know this is not the case. DNA chips and RNA chips detect both single and double stranded polynucleotides, although the intensity is somewhat lower for single stranded.]

If you have enough of your sample, you could run a Nano chip. This will detect both single and double stranded polynucleotides. Alternative you could test the concentration of your sample using double-strand-specific and non-double-strand-specific fluors. (For example, picogreen and ribogreen, respectively link ) If your ribogreen assay detects vastly more nucleic acids than your picogreen assay, this would suggest that your sample is largely single stranded.

(2)That little peak at 266 is really your 35 bp marker. But this would not explain what the big peak at 35 bp is. Still, if you can take a look at the .XAD file containing your chromat using the bioanalyzer software, you could select that peak and set it to be the lower size marker band. Then just see if your peak now agrees with the size on your agarose gel. This possibility seems unlikely to me.

Please let us know what your final result is.

--
Phillip

[Neuromancer]

similar problem here

hey guys,

is there any update about this issue? I'm having more or less the same problem, or at least I think I do.
Untill now I could not get the DNA highsensivity chip to work well for ChIP samples, input or sheared chromatin - but I think that is what the ChIP is made for!?

thanks for any info on this issue!

[pmiguel]

Untill now I could not get the DNA highsensivity chip to work well for ChIP samples, input or sheared chromatin

What exactly do you mean? Sizing difference between agarose and lab chip? Or something else?

--
Phillip

[Neuromancer]

Yes, we have exactly the same problem. On agarose gel ist looks like avg. size of 200-400bp. On the Bioanalyzer DNA HS kit it gives a slight smear in that range plus a peak at ca. 2000 as has been seen in the post before (see attachment).
Could it be that the DNA HS kit has problems with sheared DNA?

[pmiguel]

Yes, we have exactly the same problem. On agarose gel ist looks like avg. size of 200-400bp. On the Bioanalyzer DNA HS kit it gives a slight smear in that range plus a peak at ca. 2000 as has been seen in the post before (see attachment).
Could it be that the DNA HS kit has problems with sheared DNA?

Maybe it is ssDNA? We ran the RNA ladder standard on a high sensitivity chip (see here

http://seqanswers.com/forums/showthread.php?t=12852

with results you may find surprising.

ssDNA would not stain very brightly with EtBr (if that is what you used in your gel...) but it shows up fine on a High Sensitivity Chip. Only problem then is that the ssDNA molecules would need to be ~1.5 thousand nt in length to run as a band of that apparent molecular weight on the High Sensitivity Chip.

So, maybe the temperature of your shearing (sonication?) was not sufficient to prevent some strand denaturation? If single stranded molecules are more recalcitrant to shearing than double stranded molecules that would explain the second peak...

Speculation on my part, obviously...

--
Phillip

[ETHANol]

Your agarose gels looks massively overloaded which makes the chromatin appear to run as a small band like you see. Can you rerun some serial dilutions of your input on an agarose gel.

Don't believe me, check out the figure in this protocol and you'll see that when the gel is overloaded the chromatin looks small and well sheared.

http://myers.hudsonalpha.org/documents/Myers%20Lab%20ChIP-seq%20Protocol%20v042211.pdf

However, their assessment of the gel is 100% wrong.

I've done some work looking at X-linking conditions and shearing conditions with the Covaris so if you give more details about your x-linking, chromatin preparation and shearing conditions I could probably give a better guess.

This might interest you:

http://ethanomics.wordpress.com/2011/11/30/177/

As well as this:

http://ethanomics.files.wordpress.com/2011/10/chip_covaris.pdf

[pmiguel]

Hi Ethan,
My interpretation of the gel in the HudsonAlpha pdf is that the lower molecular weight DNA is pulling EtBr out of the gel as it migrates -- effectively destaining the gel as it passes. In the case of the highest loadings no EtBr remains to stain the larger (~5 kb) MW "smear".

That assessment may coincide with the description "the gel is overloaded" in the lanes where the high molecular weight smear is not visible.

Then there appears to be another issue where the lower molecular weight smear trends downward in apparent size as the gel is more heavily loaded. That is more the canonical "overloaded gel" effect. Like the gel matrix (agarose?) is failing to retard the progress of the larger numbers of molecules passing through it as effectively as it does a smaller number.

Note that when examining a gel of this sort we are essentially being asked to pre-suppose a number of salient factors:

(1) What "staining" dyes are being used. Eg, EtBr. What concentration. To what components (eg, loading mixture and/or gel, and/or running buffer, or post-run stain only.)
(2) What was the composition of the running buffer, the gel buffer and the loading buffer.
(3) What were the conditions of the run (V, time, length of gel, distance between electrodes.)

Some of these details may be given, but rarely are they all provided. I have my own sort of default presumptions:

Most gels are agarose gels with EtBr in the gel itself, but not in the running buffer or the loading dye mix. The loading dye mix is some glycerol concentration that has been sitting on a bench for over 1 year and has too much Bromphenol Blue in it -- possibly also Xylene Cyanol (also too much). The gel (it will be a minigel most of the time) will have been run just below the voltage that would melt the agarose that composes it. Illumination will be a medium wave length UV transilluminator.

But the chance of all of my presumptions being correct is miniscule.

One plus to use of bioanalyzer chips is that, although secret, many of the parameters are controlled and thus will be the same chip to chip, lab to lab. Also, the protocols offered by Agilent will militate (in some cases successfully) against overloading.

The down-side to using bioanalyzer chips is: it is largely a black box. We may all be using the same (or a similar) black box, but there is much we do not know.

--
Phillip

[pmiguel]

Your agarose gels looks massively overloaded which makes the chromatin appear to run as a small band like you see. Can you rerun some serial dilutions of your input on an agarose gel.

Don't believe me, check out the figure in this protocol and you'll see that when the gel is overloaded the chromatin looks small and well sheared.

http://myers.hudsonalpha.org/documents/Myers%20Lab%20ChIP-seq%20Protocol%20v042211.pdf

However, their assessment of the gel is 100% wrong.

What assessment is that?

By the way, I don't see anything about the protein being purified away prior to running the agarose gel. Does that mean it would still be there, possibly binding the DNA?

I am not familiar with protein binding experiments. Are the proteins that bind the DNA and end up cross-linked to it going to have a positive charge that would tend to drag them back towards the anode? Or do you add SDS to the agarose gels or loading buffer?

--
Phillip

[Neuromancer]

Maybe it is ssDNA? We ran the RNA ladder standard on a high sensitivity chip (see here

http://seqanswers.com/forums/showthread.php?t=12852

with results you may find surprising.

ssDNA would not stain very brightly with EtBr (if that is what you used in your gel...) but it shows up fine on a High Sensitivity Chip. Only problem then is that the ssDNA molecules would need to be ~1.5 thousand nt in length to run as a band of that apparent molecular weight on the High Sensitivity Chip.

So, maybe the temperature of your shearing (sonication?) was not sufficient to prevent some strand denaturation? If single stranded molecules are more recalcitrant to shearing than double stranded molecules that would explain the second peak...

Speculation on my part, obviously...

--
Phillip

Ok, we also ran the RNA ladder on the DNA HS chip and have similar results to yours (5 peaks with completely weird sizes). But that might also be due to different conditions of the gel, dye and migration program on the DNA vs RNA chips...? like you said, it's a black box...
However, I don't see why we should have ssDNA in our soultion.

Maybe a few key features of our protocol will help:
We are using Diagenode's low cell# ChIP kit and X-linking is done with 1% Formaldehyde. Shearing for 25 minutes in the Bioruptor with changes of icy water every 5 minutes. For reverse x-linking we use several ours with prot k @ 55°C and then purify DNA with phenol-chloroform.

I don't see where in this protocol you can denature such a huge amount of DNA...

Furthermore, I really do not see that our gel is overloaded (but I will try rerunning lower amounts!). And there are several arguments against the EtBr-pulling theory. First, EtBr runs to opposite direction compared to the DNA, second there should be a lot more EtBr than could actually be taken up by the DNA, third high MW DNA stains a lot better with EtBr due to its lenghth (more space for binding of EtBr).

Still not happy with this Bioanalyzer thing...

So how do other labs check the correct size and concentration of samples before ChIP-seq library prep??
As we don't have a Qbit we need to use the Bioanalyzer to check concentrations after ChIP, I really thought that the DNA HS chips were desinged for exactly that purpose...

[pmiguel]

I don't see where in this protocol you can denature such a huge amount of DNA...

Okay, I have no reason to doubt you. Except that in combination with the shadowing effect I mention below, it would easily explain what you see. Also, if the DNA was strand-denaturing it might be harder to shear... Still pure speculation on my part.

Furthermore, I really do not see that our gel is overloaded (but I will try rerunning lower amounts!). And there are several arguments against the EtBr-pulling theory. First, EtBr runs to opposite direction compared to the DNA,

Yes it runs in the opposite direction, and when there is sufficient DNA, it casts a "shadow", binding it up before it reaches the higher molecular weight fragments.

Do you add EtBr to the running buffer as well? Or just to the gel?

second there should be a lot more EtBr than could actually be taken up by the DNA,

I don't see you have a basis for this claim. How do you know?

third high MW DNA stains a lot better with EtBr due to its lenghth (more space for binding of EtBr).

But it will see a lower concentration of EtBr because the lower molecular weight smear will have bound much of it.

This is going to be the case unless the gel is stained after running. There are lots of factors in play, but in cases where there is the amount of DNA I see in the smear in your gel I do expect little of the dye to be available to the higher molecular weight (~5kb).

Still not happy with this Bioanalyzer thing...

So how do other labs check the correct size and concentration of samples before ChIP-seq library prep??
As we don't have a Qbit we need to use the Bioanalyzer to check concentrations after ChIP, I really thought that the DNA HS chips were desinged for exactly that purpose...

I see no reason in this case not to trust your Bioanalyzer result. I think the agarose gel is the one misleading you.

--
Phillip

[Neuromancer]

Hey Philip,

thanks for the detailed analysis. You might have a point there...

I will try to figure it out and keep you posted about it!

[angelawu]

Hi!

Not sure if this issue of gel vs bioanalyzer has been resolved yet, but I have some thoughts about this.

1) In your second uploaded pdf, the intensity on the bioanalyzer chip looks very high. If the bioanalyzer chip is overloaded, I have observed the recorded size of the DNA to appear much larger than it actually is (the quantity of DNA causes its migration in the microfluidic chamber of the chip to change such that the chip detects the DNA at a later time than it should, which displays as a larger size). And in addition, not only will this happen if your one sample is overloaded, but if OTHER samples on the chip are overloaded, it will run-over into the subsequent wells, showing up as a high mol weight DNA smear. So, you may want to look back and see if the other samples on the same chip were overloaded.

2) In your second pdf, for the agarose gel image, I actually do see a faint band at the top of your gel. This band could show up as being much "brighter" in the agilent chip because of the way the intensity is measured, maybe? If you look at your quantitation for the bioanalyzer, the software can calculate for you the number of actual molecules you have in a particular size range (use "regions" instead of "peaks"). When you compare the number of molecules in the high mol. region vs low mol. regions, you might find that you actually have way more molecules in the low mol region anyways.

3) When you purify your DNA, do you use any sort of carrier RNA or protein such as glycogen? Is it possible that those carried forward into your bioanalyzer reaction? Those may not have an effect on the agarose gel, but might show up as big sized stuff in the chip.

[ETHANol]

Furthermore, I really do not see that our gel is overloaded (but I will try rerunning lower amounts!).

Care to bet on that. If I win you'll send me a couple good beers. If you win I'll send you some sun.