SDS, Cell numbers, dilution - (Jul/28/2009 )

MunkySpunk on Jul 31 2009, 10:56 AM said:

One of the benefits of doing the plate based ChIP with the Flanagin et al. method is that you don't have to use beads. We just bind the protein A directly to the wells. I didn't think there would be enough surface area but Steve found that high-bind polystyrene plates from Corning work great. They do give more background than magnetic beads (similar to using agarose beads) but I'll take that background if it means less sample transfers (i.e. keeping it in the plate is much simpler and easier). Typically we use 50,000-100,000 cells per well though the method still works down to 10,000 cells. Using the method as it is in the paper we elute in 100µl so that should give you an idea how many PCR reactions you can do.

Let me know if you're interested and I can give you a detailed protocol and some tips to get you started. It is a little more involved for getting started than Fast ChIP (since Fast ChIP is essentially the same method everyone uses just with a small tweak at the end) but once you get it working you quickly see the advantage. A few times I harvested cells in the morning, sonicated just before noon, ran 150 ChIP samples (2 plates) in the afternoon, and loaded up a PCR plate before going home, and was much more labor than doing 24 ChIP samples with Fast ChIP starting with already sonicated chromatin (this is assuming you have a repeat pipettor or multi-channel pipettor for all the washes and blocking step).

Also, I'm curious what your sonication profiles will look like. Definitely post them. Good luck.

Joel

Tried the new ChIP buffer. The results didn't look very good. I performed 4 lysis/sonication reactions. Each in a volume of 500ul with 10e7 cells/ml (I split up 2X10e7 cells into 4 equal parts for sonication). The only sonication that gave me anything whatsoever as far as DNA in concerned was the sample I pulsed 12X10s at 75% power. The 12X10s pulses, coincidentally, gave me a smear where I wanted one, but none of the other lanes showed ANY DNA, much less a smear at the wrong size.

The fact that none of the other lanes from that particular 500ul sonication/lysis reaction showed ANY DNA leads me to conclude my problem was downstream of the sonication step. But I also did not get DNA in my no sonication negative control, which would lead one to clnclude that it's the steps where I use the new lysis buffer.

Here's what I do post-sonication: I spin down the sample as per step 8 of Nelson et al. 2006. I take that supernatant, add 1.5ul of RNAse A, 37 for 30 minutes, add 1.5ul of Prot K, 62 for 2 hours. After the Prot K step, I had lots and lots of precipitate, so I spun that down, added loading dye, and loaded 1/2 the sample without disturbing the pellet.

Is it possible that my DNA is in the pellet from the precipitate post-sonication and I just disturbed that one pellet a bit when loading the gel?

Am I utilizing the new lysis buffer properly? I add it, pipette, promptly spin, remove supernatant, add more, spin again, and call it lysed (as per steps 4-6 of Nelson et al.

I'm about to run the rest of my samples on another gel, pellet and all, to test my wild guess.

Any answers prior to me running this ge would be greatly appreciated.

MunkySpunk on Aug 3 2009, 10:06 AM said:

The precipitate is interesting. Probably that's where your material is. I would consider skipping the EtOH precipitation step altogether. Just add your chromatin to 100ul chelex (if the volume is too large just spin down the chelex, remove the supe, and add your chromatin to the pellet of chelex resin), add proteinase K, digest and boil and run the supe on the gel.

I don't do EtOH precipitation. This is just a shearing test and my results are quite mind-boggling.

The DNA is not in the pellet portion of the sample. I just ran it on a gel and got the same one smear in a single lane with no DNA (smear or otherwise) anywhere else.

Again, what do you guys do, exactly, to lyse the cells with the IP buffer? I'm losing my DNA somewhere between lysis and running a smear on a gel. How many times do you pipette the first time (step 4 Nelson et al.)? How vigorously do you pipette? How long does it stay in this buffer before you spin it down? How vigorously do you resuspend the pellet in the second wash (step 6)?

I've got a picture of my mystery gel below. Each lane indicated the number of pulses each sample received and the duration of each pulse.

(-) is negative control. Just lysis buffer, no sonication.

The group of samples on the left were washed together in the same tube and sonicated together in the same tube (I sonicate twice, remove 50ul, and so on...) Same for the group on the right.

As you can see, my 12X10s pulse lane has DNA fragments of decent size, albeit a bit small. Given that all of the 10s pulse samples came from the same tube, were lysed together, and sonicated together, my gel is telling me that either the cells spontaneously grew DNA somewhere between 10 and 12 pulses, or the DNA is being lost after sonication somewhere.

Can anyone suggest what may be causing this? Where could my DNA possibly be hiding out.

Thank you all.

MunkySpunk on Aug 3 2009, 01:19 PM said:

As far as lysis goes, I just pipet up and down enough to break up the pellet. I'm not to vigorous about it.

So, after sonication and clearing by centrifugation, what percentage of the original cell pellet would you say the resulting post sonication pellet is? For me it is typically 5% or less.

Hi guys,

I finally gave up on trying to get my nuclei to lyse in non-SDS buffer. They're just too stubborn. I have made a cell lysis buffer with NP-40 and a nuclear lysis buffer at 1% SDS. I figure if I start with enough cells, I can dilute it down to just .1% or lower for the IP reaction itself.

I'm running into another issue: A very bright LMW band in my gels. (Picture below)

Here is what I do after sonication, the SDS concentration is at about .5% at this point.
.5) Spin down sample, discard pellet.
1) Add 1.5ul RNAse A at 10mg/ml to 50ul of lysate
2) 37 1 hour
3) Ass 1.5ul Prot K at 20mg/ml to the 50ul of lysate
4) 67 degrees 2 hours
5) Spin down, load half of supernatant on a gel

It seems my home-made buffers work MUCH better than the millipore supplied ones, as I'm getting gobs and gobs more DNA per lane now.

My question is this: How do I get rid of that LMW band at about 150bp? An 18S rRNA band would be expected at abou 200bp. But this CAN'T be rRNA as there is no 28S band, and the 28S band would be expected to be twice as intense as the 18S band. Could this band be DNA that was wrapped around a histone (and thus semi-protected) at the time of sonication (146bp) only to come loose in the Prot K step? How would I go about avoiding the formation of this band? I sonicated once before and didn't get the band, but I couldn't replicate the results. If this was DNA wrapped and protected by a histone at the time of sonication, it could explain the buildup of ~150bp DNA.

Thanks in advance.

MunkySpunk on Aug 10 2009, 10:32 AM said:

Well, if you treated your samples with RNase you wouldn't expect the 28S and 18S bands to show the normal pattern (the 18S would be dominant with partially digested RNA). But I wouldn't expect that you would be able to see any RNA after RNase treatment. On the other hand, I doubt that the LMW band is DNA since I've never seen anyone be able to get a band smaller than 200bp. I don't think the ends of the DNA protruding from the nucleosome are long enough for the shear forces of sonication to fragment. My guess is that the LMW band is incompletely digested RNA. Maybe your RNase just isn't very efficient or isn't getting enough time to completely digest the RNA (hard to believe but I can't think of anything else).

Hey guys,

Stuff has started working for me, so I thought I'd give an update for other folks who may be experiencing the same issues:

Sonication did NOT happen in NP-40/Triton for me. I'm using LNCaP cells. I am forced to use SDS for my sonication, but it works quite well.

I am using the buffer recipes found at: http://www.jamequist.umn.edu/Protocols/SOP-ChIP.htm

As far as I can determine, these are the same buffers included in the upstate/millipore kits. So if you are looking to save money, the first thing you can do is cut the 'kit' out of the equation.

At any rate, I harvest 10e7 cells in the media and crosslink as per usual.

I resuspend 10e7 cell nuclei in 500ul of nuclear lysis buffer (with SDS). I then split this into two 250ul portions and add 250ul of dilution buffer to each for a total of 5e7 cells/500ul in two tubes at an SDS concentration of ~.5%.

I sonicate on a Fisher dismembranator 60. I use 8X10s pulses at a setting of '8' which gives me a nice smear. My earlier problems with my diagnostic gels for smearing were fixed by de-crosslinking prior to ProtK treatment. After I included the de-crosslinking step, my genomic DNA smears became textbook quality.

After sonication I pool the 2X500ul portions back into 1, spin and save the supernatant as my lysate. I use 100ul + 400ul dilution buffer for each IP.

After IP and the washes (as per typical), I use the Chelex method exactly as described in the FastChIP protocol. I can't see any difference between Chelex versus Phenol/Ethanol as far as PCR results go, except the DNA concentration of the Chelex method is lower, but this makes sense as it is in a larger volume of water than I typically resuspend an ethanol'd pellet.

I am using 3.5ul of the elution in a 25ul PCR with homemade standard Taq, but the bands are somewhat faint. I am going to bump it up to 5ul and see what happens.

I am getting bands on my Normal IgG negative control, but there is a clear enrichment for my product of choice when using specific antibodies. I am going to wash 2X in my high salt buffer for a longer period of time per wash to try and reduce this. I'd rather not have to pre-clear the lysate with beads, as that can get kind of expensive.

I am using NEB Magnetic Protein A beads and ChIP qualified AB's, typically from Millipore.

MunkySpunk on Aug 27 2009, 08:42 AM said:

Glad the shearing is working out for you now. Did you ever figure out what the low MW stuff was?

If you really want to get rid of the background you can always try using blocking agents (5% BSA and 100µg/ml sheared salmon sperm DNA; tRNA works just as well if you don't want to have any contaminating salmon DNA in your preps). I would suggest that you do all of the IP steps (pre-incubation of beads in buffer, incubation of antibody and chromatin, and incubation of antibody/chromatin complexes with beads) in the presence of these reagents.

I probably sound like a broken record but I've also found that reducing the amount of chromatin used is the best way to reduce background.

On the other hand I've actually found a small amount of background signal to be helpful in controlling for a common but rarely recognized problem with ChIP. That is the tendency for ChIP to have a bias towards non-specific binding of heavily transcribed regions. For instance I have found, on multiple occasions, that I get a higher pull down for several transcription factors at the TSS of actively transcribed genes than at the sites where the factors are supposed to bind when I express the results as a % of input. However, when I look at the ratio of transcription factor IP to mock IP the results show the greatest binding at the TF binding sites and not at the TSS, as I would expect. In addition, when I look at factors that shouldn't bind to the TSS, I still find that, what I am assuming to be background binding, is higher at the TSS than at intergenic sites, using the % of input calculation. The IP/mock ratio, however, is the same for all of the negative control regions. If you ever see a pattern of binding where you get higher levels at transcribed regions it's always good to look at the IP/mock ratio just to make sure the pattern is real.

Joel