SEQanswers

Go Back   SEQanswers > Applications Forums > Sample Prep / Library Generation



Similar Threads
Thread Thread Starter Forum Replies Last Post
Differential histone modifications maryhagen Bioinformatics 6 02-13-2012 02:21 PM
ChIPseq for histone modifications Alex Clop Epigenetics 9 02-09-2012 11:32 PM
Chip-Seq histone modifications EdinG Introductions 0 09-21-2011 04:05 AM
ChIP-Seq: MEME-ChIP: motif analysis of large DNA datasets. Newsbot! Literature Watch 0 04-14-2011 03:50 AM
Removal of large fragments? seqAll Sample Prep / Library Generation 2 01-13-2010 10:20 PM

Reply
 
Thread Tools
Old 05-21-2012, 08:10 AM   #1
MGineste
Member
 
Location: Montpellier, France

Join Date: Feb 2011
Posts: 21
Default Large DNA fragments after ChIP on histone modifications

Dear all,

I'm encountering an already mentioned issue (http://seqanswers.com/forums/showthread.php?t=11737).

I'm creating a new thread though, as my ChIP experiments are slightly different (targeting histone modifications) and as I've already tested a number of hypotheses regarding this issue.

As mentioned in the above thread, I encounter a DNA size distribution issue after the ChIP : before ChIP, average size of the DNA fragments is around 200bp (with ≈99% of the DNA below 600bp) ; after ChIP, average size of the DNA fragments is around 1000-2000bp (with variable proportion of the DNA below 600bp from one sample to another).

Figure 1 : DNA size aberration is systematic with all antibodies.


Figure 2 : DNA size aberration is systematic with all ChIP replicates.


MG##_Sonication : Sample processed right after sonication to check the DNA fragment distribution.
MG##_ChIP-Input : Unbound fraction of a mock ChIP (no antibody).
MG##_ChIP-'antibody' : Bound fraction with a given antibody.

I reverse-crosslink overnight at 65°C with strong shaking. I perform a 3h proteinase K treatment at 55°C. All samples are then silica column purified. The DNA size aberration is not a Bioanalyzer bias as I observe exactly the same thing on an agarose gel.


I first incriminated an insufficient reverse-crosslink, but additional reverse-crosslink in optimized conditions does not change the situation. I then incriminated presence of RNAs, but additional RNase treatment does not change the situation either.

I am now very sceptical as well as desperate regarding this situation. I would be very happy if you could share your experience on this issue.

Last edited by MGineste; 05-21-2012 at 08:13 AM.
MGineste is offline   Reply With Quote
Old 05-24-2012, 06:48 AM   #2
MGineste
Member
 
Location: Montpellier, France

Join Date: Feb 2011
Posts: 21
Default

EDIT (11-08-2012) : The explanation mentioned below only partially stands to explain the size difference between the input and the chIP samples. Further explanations in the next post.

Quote:
Answer found. Newbie mistake.

I just took beads without adding any sonicated chromatin to them. Kept the supernatent on one side (MG96_Beads_Supernatent), and processed the beads as I would during a normal ChIP on the other side (MG98_Beads_BoundFraction).

Figure 3 : Do not use salmon sperm (or anything else...) DNA blocked beads


Conclusion is in the title of the figure. "Putain de merde !", ai-je envie de dire...

Last edited by MGineste; 11-09-2012 at 04:37 AM. Reason: Updated data.
MGineste is offline   Reply With Quote
Old 11-09-2012, 05:44 AM   #3
MGineste
Member
 
Location: Montpellier, France

Join Date: Feb 2011
Posts: 21
Default

Over the last few months, I performed a number of ChIP assays on different histone modifications, using non-DNA-blocked magnetic beads this time.

The pattern difference mentioned in the original post of this thread (showing a (huge) size shift between the input and the chIP samples) was still present with a reproducible trend : the less frequent the targeted epitope is in the genome, the more the size shift is observed (Figure 4).

Figure 4 : "Rare" histone modifications (H3K4me3) show a greater size shift than "frequent" histone modifications (H3, H3K27ac).


Key : Red = Input, Blue = Mock (no antibody), Green = H3, Turquoise = H3K4me1, Pink = H3K4me3, Orange = H3K27ac.

I interpreted this observation by an efficiency bias of the ChIP towards long fragments : short chromatin fragments carry few epitopes ; long chromatin fragments -even though they should be in a large minority within the sample- carry a higher number of epitopes. Longer fragments would be thus preferably bound by the antibodies, and thus immucoprecipitated with a higher efficiency than shorter fragments.

Your comments and feedback are welcome.
MGineste is offline   Reply With Quote
Old 07-11-2013, 09:53 PM   #4
wangel
Junior Member
 
Location: Madison

Join Date: Jul 2013
Posts: 1
Thumbs up

Hi, I have similar problem with the H3K27me3 ChIP. Could you please tell me how did you solve the problem? Also, do you think if the epitope has been destroyed during sonication, it is also likely to see this shift in IP samples? Thanks a lot for your help!
wangel is offline   Reply With Quote
Old 07-12-2013, 01:35 AM   #5
MGineste
Member
 
Location: Montpellier, France

Join Date: Feb 2011
Posts: 21
Default

Hi wangel,

I did not really solve the size shift issue.

Instead, I used the Bioanalyzer assay to determine for each antibody the quantity of immunoprecipitated DNA whose size is below 500bp, which is the usual upper limit of size selection during library preparation of ChIP-Seq samples. I repeated the ChIP experiment to be sure to obtain at least 10ng of immunoprecipitated DNA whose size is below 500bp.

Noteworthily, the size shift did not affect at all the quality of the ChIP-Seq sample. For example, I obtained the best signal over noise ratio for H3K4me3, which was showing the most drastic size shift.

Nevertheless, I would guess that for H3K27me3, which is a rather abundant mark, the size shift would be moderately pronounced.


If you really want to spend time on reducing the size shift issue, I would look into those directions :

- I would achieve an efficient sonication with the least sonication cycles : reducing crosslinking length (down to 8-10min), dissociating nuclei before sonication (with a syringe/needle), reducing the density of material during sonication.

- I would also increase the quantity of antibody for each immunoprecipitation, as preferential immunoprecipitation of long fragments could be a sign of the antibody not being in saturation quantity compared to the chromatin.

Good luck and keep us posted of your experience,
Mathieu
MGineste is offline   Reply With Quote
Old 12-16-2013, 03:14 AM   #6
YanMar
Junior Member
 
Location: Sweden

Join Date: May 2013
Posts: 3
Default

Hi,

I'm experiencing the same problem for H3K4me3 and H3K27me3. The size shift is much more pronounced for H3K27me3 (aka I have a large peak >1000 bp).

We performed ChIP-qPCR on these ChIP samples as a QC before proceeding with the library preparation, and we see a satisfactory enrichment of correct histone target.

Now, I'm concerned about how this size-shift will affect the library preparation since the majority of the fragments are >1000 bp. Would it be possible to size select to remove all of those large fragments before library prep or perhaps "re-shear" the ChIP material again, before library prep?
YanMar is offline   Reply With Quote
Old 08-12-2014, 07:10 AM   #7
pr393
Junior Member
 
Location: Cambridge

Join Date: Jul 2014
Posts: 3
Default

Hi MGineste. I was running a few IP samples on the Bioanalyzer prior to library preparation and I got really funny peaks compared to the ones that you've shown. I'm using the same antibodies, sanitation looked ok and the qPCR I did to control for the IP also looked great. Could you please let me know your opinion on my Bioanalyzer data. It's the first time I'm doing this and I'm not sure if this samples are worth it to prepare the library or not. I wanted to run the input in parallel with some of the samples but my Bioanalyzer decided to commit suicide.
Attached Images
File Type: jpg Bioanalyzer.jpg (70.0 KB, 109 views)
pr393 is offline   Reply With Quote
Old 09-11-2014, 02:11 AM   #8
AndreaT
Junior Member
 
Location: Germany

Join Date: Sep 2014
Posts: 1
Default Histone modification ChIP antibodies

[QUOTE=MGineste;88950]Over the last few months, I performed a number of ChIP assays on different histone modifications, using non-DNA-blocked magnetic beads this time.

The pattern difference mentioned in the original post of this thread (showing a (huge) size shift between the input and the chIP samples) was still present with a reproducible trend : the less frequent the targeted epitope is in the genome, the more the size shift is observed (Figure 4).

Figure 4 : "Rare" histone modifications (H3K4me3) show a greater size shift than "frequent" histone modifications (H3, H3K27ac).


Key : Red = Input, Blue = Mock (no antibody), Green = H3, Turquoise = H3K4me1, Pink = H3K4me3, Orange = H3K27ac.

Hi MGineste,

I am also trying to do a histone modification ChIP.
Your IPs look very fine! Can you tell me from which company you get your histone modification antibodies?

Best regards
AndreaT is offline   Reply With Quote
Old 03-02-2015, 12:33 PM   #9
Saaket
Junior Member
 
Location: Los Angeles

Join Date: Mar 2015
Posts: 2
Thumbs up Same problem

Hi,
I had the same problem with another antibody. I was wondering how to proceed with Sequencing as I am afraid that I will have data which is not clean.

I was also told that I might have performed over amplification or my fragmentation was not good, both of which were not applicable.

Cheers.
Saaket is offline   Reply With Quote
Old 03-06-2015, 01:03 PM   #10
aewills
Junior Member
 
Location: CA

Join Date: Mar 2015
Posts: 1
Default large fragment size bias after H3k27ac IP

Hi all,
I've just been battling this with a (usually very reliable) H3K27ac antibody in a new type of sample. The input is a smooth curve centered at 350bp, but following IP the peak is at 1500bp. It's now turned up in several biological replicates and with different lots of antibody. I'd really like to know the status on this question!
1. Any smoking guns on how to fix it?
2. How have others fared using these size-biased samples to make libraries? Did you use Ampure beads to size-select away the high MW fragments, or carry on with the library as-is, and just size-select after adding adapters and before PCR?
3. Has anyone else found that this only occurs in some sample types?

Thanks much!
aewills is offline   Reply With Quote
Old 03-06-2015, 01:31 PM   #11
Saaket
Junior Member
 
Location: Los Angeles

Join Date: Mar 2015
Posts: 2
Default I tried 0.4x beads

I tried 0.4X beads lost 50% of DNA but was not able to get rid higher molecular wt DNA. Will post Bioanalyzer report soon.

Ran it on sequencer and processing at the data now.
Saaket is offline   Reply With Quote
Old 03-06-2015, 04:39 PM   #12
nucacidhunter
Jafar Jabbari
 
Location: Melbourne

Join Date: Jan 2013
Posts: 1,199
Default

I do not know the exact protocol that has been followed, but possibilities to explain size shift:

1- Ligation of fragments to each other to form larger ones (least possible cause)
2- Differences in buffer composition and viscosity between input and IP DNA solution
3- Binding proteins to DNA making them less mobile in Chip
4- Denaturation and formation of single stranded DNA during protocol which will migrate differently to dsDNA
5- Preferential pull down of larger fragments
6- Combinations of above

All above possible causes can be tested one by one by proper experiments.
nucacidhunter is offline   Reply With Quote
Old 06-01-2015, 08:26 AM   #13
romain30
Junior Member
 
Location: Montréal

Join Date: Apr 2015
Posts: 1
Default

I have the same problem.
I am working on 300K FACSED cells.
I tried XChIP and NChIP with extremely different protocols. I always get a good input but my IPs contain only >1000bp fragments.
Any idea?
romain30 is offline   Reply With Quote
Old 08-05-2015, 07:31 AM   #14
Cadam
Junior Member
 
Location: belgium

Join Date: Aug 2015
Posts: 1
Default questions

We are quite surprised by the size distribution of ChIP and Input samples (bioAnalyzer plot in attachment, with 6: 9 being ChIP samples and others corresponding to input for different conditions and replicates).
It seems that most of the fragmented samples are greater than 1 kb while our customer expected a size distribution between 200 and 400 bp.
For the ChIP sample we see only a bump greater than 1 kb ("6: 9” in red).
Is this a bias of the bioAnalyzer HS DNA Chip?
Would you already perform library prep on these samples or would you try to optimize the IP protocol?

Someone with similar samples could continue with at least 10 ng of sample shorter than 500 bp to start the library prep. Can someone confirm that this works? Do you do library prep with size selection or without size selection? How do you perform the size selection, on AMpure beads or on a gel? Since the concentration of the ChIP samples is very low, we can’t repeat the library prep.
Attached Images
File Type: jpg Presentation1.jpg (44.6 KB, 61 views)
Cadam is offline   Reply With Quote
Old 11-04-2016, 03:12 AM   #15
K_Cheung
Junior Member
 
Location: Newcastle-upon-Tyne

Join Date: Nov 2016
Posts: 1
Default fragment size shift after IP

We also had this same issue with histone ChIP, DNA fragment sizes were checked with a bioanalyzer before and after IP and there appeared to be a bias towards pulling down larger fragments. The company we used for library prep and sequencing (Diagenode) simply re-sheared the post IP fragments before library prep. The data has been analyzed and it does not seem to have affected the results, peak quality is good (signal over noise ratio is good) and histone mark enrichments are where they are expected to be.
K_Cheung is offline   Reply With Quote
Old 11-26-2018, 04:19 PM   #16
KB*
Junior Member
 
Location: UK

Join Date: May 2018
Posts: 2
Default 2018 - Large fragments observed as well

Hi, guys.
The same here. I am preparing ChIP-Seq samples for a "co-factor". There is only 1 paper reporting it binds to DNA at all. In a different cell line

My IP on well sheared chromatin yields much more >1 kb fragments than the smaller ones. When DNA yield in total is about 10-40 ng/IP, the DNA yield for <=400 bp is in the range of pg! Two totally different protocols are already tried, including enzymatic chromatin digestion and sonication. No help.

Thoughts:
- might try to super scale up to get more total DNA (to increase the yield of <400 bp fragments);
- there shall be no DNA >400 kb fragments before IP. I may try to oversonicate/overdigest chromatin before IP;
- I want to try DNA beads (Select-a-Size DNA MagBeads from Zymoresearch) on my lysate to remove >400 bp fragments before IP ... not sure how it would work out as there is SDS, proteins, NP40 etc etc in the cell lysate;
- digest/sonicate DNA after IP ... If the pull down of larger fragments is specific, then why not? On the other hand, this shall bring irrelevant DNA fragments just situated in between of the "co-factor" binding sites;
- finally, they guys who published that only one paper on this co-factor, prepared the library on what they had and manually excised <400 bp fragments from from the gel. May be this is the right answer...

Any comments will be very much appreciated

Last edited by KB*; 11-26-2018 at 04:22 PM.
KB* is offline   Reply With Quote
Reply

Tags
chip, histone modifications, size

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off




All times are GMT -8. The time now is 10:59 AM.


Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2019, vBulletin Solutions, Inc.
Single Sign On provided by vBSSO