Note: this is a search engine friendly version of my lab notebook, please see the pdf version of this document for a more human friendly, printer friendly version.

Chapter 4
chromatin immunoprecipitation based determination of transcription factor binding sites: SAPE returns

THIS CHAPTER/PROJECT IS IN PROGRESS
In Chapter 3, I sketched out some ideas and ran some preliminary experiments to try and achieve high-throughput transcription factor target discovery by adapting ideas from SAGE (serial analysis of gene expression) and DNA footprinting to ChIP (chromatin immunoprecipitation). However, my ChIP protocol at the time was quite tedious and I found a similar project had already been completed by another, so it was no longer necessary to publish yet-another-ChIP-pilot study.
With my recent factorial and response surface optimizations of the ChIP protocol (Chapter 2), the determination of all of the binding sites for all of the transcription factors in E. coli looks like an attainable goal. In addition, highly parallel sequencing has really taken off in the last year, so finding a place to outsource the sequencing should be easier.
The general strategy for determining the transcription factor binding sites will be   to run a standard ChIP protocol (using one of my optimized protocols) but with an additional digestion step to shorten the length of the transcription factor-bound fragments. Finally, I'll purify the DNA fragments, barcode them, sequence them, and map them back to the genome (Figure ).
Please see the pdf version for figures
Figure 4.1: A schema for highly-parallel discovery of transcription factor binding sites.

4.1  Will restriction enzymes digest crosslinked DNA

Tue Nov 27 19:26:01 EST 2007
A useful component to generating an in vivo footprint is the ability to digest or cut DNA. The factorial ChIP protocol optimizations I've done, yielded one protocol that uses only 0.1% formaldehyde (see Chapter 2 for details and http://blog-di-j.blogspot.com/2007/12/optimized-chip-protocols.html for the optimized protocols. I've been told by a number of folks that it's difficult or not possible to generate a footprint with ChIP because crosslinked DNA is not readily digestable by restriction enzymes.
For this experiment, I'm going to cut pUC19 (my favorite piece of DNA) with a bunch of restriction enzymes (mostly 4-mer cutters) to see how well they cut crosslinked DNA.
I crosslinked the DNA using 0%, 0.1%, or 1% formaldehyde for 10 minutes at RT followed by glycine quenching just like I do with my standard ChIP protocol. I used 7 mg of pUC19 in a total volume of 200 ml TE for each formaldehyde concentration. I added 0.74 ml formaldehyde and 7.4 ml formaldehyde for the 0.1% and 1% solutions respectively. All 3 samples were quenched with 10 ml of 2.5M glycine.
I EtOH precipitated all 3 samples with 1 ml of glycoblue and washed them all 2x with 750 ml of 70% EtOH. I resuspended each sample into 40 ml TE.
I split all 3 samples into 7 for the digestions. Each sample used: 0.5 ml enzyme, 2 ml BSA, 2 ml NEBuffer, 5 ml pUC19 (from the 40 ml total), and 10.5 ml H2O . I used either buffer 2 or buffer 4 (whichever was better for the particular enzyme). The seven tested enzymes were: HinPI, MseI, MspI, SphI, BamHI, NlaIII, BfuCI. All were incubated at 37C for 15 minutes and run directly on a 1.5% agarose gel (Figure ).
Please see the pdf version for figures
Figure 4.2: gel of cut pUC19 with different enzymes and different concentrations of formaldehyde used for crosslinking
Brief Conclusions:   1% clearly sucks and is hardly cuttable as I've heard from other folks. However, 0.1% formaldehyde cuts almost as well as the 0% (no crosslinking) Figure 4.2, so given that the 1.5 day ChIP protocol works with this amount, we definitely have some potential to trim our ChIP DNA (probably in between the TE washes) to obtain something footprint-like. It will hopefully also make it easier to clone or circularize (for RCA) the ChIP DNA. If circularization is easier, could be very useful for ChIP-Chip without needing LM-PCR (hopefully cleaner and easier). Though, I'm more interested in using this for ChIP-Seq.
random thoughts:
might want to digest with a large cocktail (or multiple different cocktails) of enzymes
digesting for different times might allow something more of a distribution of sites that are centered around the binding site instead of just multiple clones of the binding site (which could be too small to map back?)

4.2  Cloning ChIP DNA

Given the result above that shows the 0.1% formaldehyde used in the 1.5 day protocol is cuttable with restriction enzymes, I've decided to try and clone the DNA into pUC19 so I can sequence it.

4.2.1  preparing the enriched/cut Chromatin

Dec 5, 2007
I ran the 1.5 day ChIP protocol Version 1.0 using samples 1 (lrp) and 4 (pdhR) from section 2.19 on page pageref. As in that section, I used 16 ml sheared chromatin with 64 ml of dilution buffer. I used 2.7 ml antibody (3.3 mg ) and 100 ml beads. The only alteration to the protocol was a cutter-wash step in 100 ml of cutter mix after the two TE washes.
Each 100 ml of cutter mix contained: 0.5 ml BfuCI, 0.5 ml NlaIII, 10 ml NEBuffer4, 10 ml BSA, 79 ml H2O . The beads + cutter mix was incubated at 65C for 15 minutes prior to eluting via the standard dynal elution procedure.
Dec 6, 2007
The two samples each with +/- antibody (for tubes total) were cleaned up with a Qiagen PCR purification kit. Unlike the typical ChIP procedure where I elute into 100 ml for downstream PCR, this time I wanted the DNA more concentrated for cloning. I eluted into the minimum volume of 30 ml .
DNA yield
Dec 6, 2007
Since the DNA was a little more concentrated than usual, I decided to see if I could concentrate it for the first time. I used 10 ml of the 30 ml sample with the Qubit hsDNA dye [Invitrogen]. The yields were:
sample yield
lrp 1 (with antibody) 1.356 ng
lrp 1 (without antibody) too low to measure
pdhR 4 (with antibody) 0.612 ng
pdhR 4 (without antibody) too low to measure
Brief Conclusions:   It's cool to see my ChIP DNA yield for the first time! I'm also quite happy to see that when I don't use antibody, I don't pull down a measurable amount of DNA (though the PCR reactions strongly suggest I do have some DNA in my no antibody control).

4.2.2  cloning the cut chromatin

Dec 6, 2007
I cut 1 mg of pUC19 cloning vector with each of the follow combinations of restriction enzymes (three samples total; one for each combination): BamHI (buffer 3 + BSA); SphI (buffer 2); BamHI + SphI (buffer 3 + BSA). These three vector cuts should allow me to capture any of the possible cut combinations on my ChIP DNA. I used 10 ml of each ChIP sample (I only used the two samples that contained the antibody not the negative controls), 2 ml ligase buffer, 5 ml H2O , 2 ml vector mix (all three cuts mixed together; 3 ng total), and 1 ml T4 DNA ligase.
I ran standard heat shock tranformation and plated 75 ml of the 300 ml of cells on an ampicillin plate.
Dec 7, 2007
No colonies...
Brief Conclusions:   No luck this time.

4.3  Cloning ChIP DNA: try 2

Wed Dec 12, 2007
I'm going to try both cutting and not cutting the chromatin this time; I'm also going to try both the 1.5 day samples and the 2.5 day samples. For the 1.5 day, I used samples 2 (lrp) and 9 (pdhR) from section 2.19 on page pageref. For the 2.5 day, I used samples 4 (lrp) and 6 (pdhR). As in those sections, I used 16 ml sheared chromatin with 64 ml of dilution buffer for the 1.5 protocol and 20.5 ml sheared chromatin with 59.5 ml dilution buffer for the 2.5 day protocol. I used 2.7 ml antibody (3.3 mg ) and 100 ml beads. The only alteration to the protocol was a cutter-wash step in 100 ml of cutter mix after the two TE washes.
I ran the digestions as in the previous section. The digestions were for 15 min at 37C. In addition, for each sample (i.e. 2, 9, 4, 6) I also included a replicate where I did not digest the chromatin (I still incubated at 37C for 15 minutes, but without the enzyme).
Thur Dec 13, 2007
I screwed up and forgot to reverse the crosslinks overnite, so I reversed them during the day today for 10 hr and 50 min at 65C followed bh 1 hr at 55C after the addition of proteinase K. I cleaned up the 16 samples with a Qiagen PCR purification kit and eluted into 30 ml of EB buffer.
Fri Dec 14, 2007
I used the Qubit hsDNA kit and 10 ml of each sample to quantify the DNA yield of each ChIP reaction:
sample yield (ng)
1.5 day, lrp 2, antibody, cut 0.588
1.5 day, lrp 2, no antibody, cut < 0.3
1.5 day, lrp 2, antibody, uncut 1.2
1.5 day, lrp 2, no antibody, uncut < 0.3
1.5 day, pdhR 9, antibody, cut 3.216
1.5 day, pdhR 9, no antibody, cut < 0.3
1.5 day, pdhR 9, antibody, uncut 2.832
1.5 day, pdhR 9, no antibody, uncut < 0.3
2.5 day, lrp 4, antibody, cut 2.166
2.5 day, lrp 4, no antibody, cut < 0.3
2.5 day, lrp 4, antibody, uncut 3.99
2.5 day, lrp 4, no antibody, uncut 0.498
2.5 day, pdhR 6, antibody, cut 0.894
2.5 day, pdhR 6, no antibody, cut 0.306
2.5 day, pdhR 6, antibody, uncut 0.336
2.5 day, pdhR 6, no antibody, uncut < 0.3
Brief Conclusions:   With no exceptions, the no antibody sample always has less than the antibody sample. The lrp sample always has more DNA when the chromatin is not cut, which makes sense. THe pdhR samples are always the opposite, which doesn't make sense. The yields seem to vary pretty drastically across all of these samples. It's hard to say if this is real, or if I'm just at the edge of the sensitivity of the Qubit dye, so the measurements are noisy. If I used 20 ml of the ChIP sample, I should be able to get a slightly more robust concentration estimate.
I ran out of time to clone this DNA, because my thesis was on Dec 17. I want to try this again with freshly prepared chromatin. I want to quantify them with 10 ml , try qPCR with another 10 ml , and try to clone with the remaining 10 ml . I think I'll switch to using only one restriction enzyme for now to simplify things.