RIP-Seq

[SeeElle]

Hello all!

I'm Cell Biology grad student. My training thus far has been in the wet-lab format so I know very little knowledge in bioinformatics.

I'm about ready to generate samples for NGS and I realize that my lack of experience with analyzing such data has created a hole in my experimental setup. I need advice on a controls that I think I will need to incorporate before I submit my samples and I would like any advice! sooner the better

I am doing a RNA-IP-Seq to identify target RNAs bound to my protein of interest. I would like to be able to identify the target RNAs by the enrichment of x genes from the deep seq data and the enrichment would be relative to my negative controls: a) a sample in which I expect the protein I'm fishing with not to bind target RNAs or RNAs in general b) a sample where my protein of interest should not be present.

In reference to neg. control b: do people have experience seeing reads in this type of sample just from background from non-specific binding by beads? And are these non-specific binding consistent or variable?

In a small-scale experiment with the read-out being qRT-PCR, I could run a Western and normalize between samples based on the amount of protein from the IP. Two Issues with NGS after the IP: a) I'm not sure if I can normalize the coverage/reads to relative protein levels; b) I have to pull RNA from multiple IPs in which there is not a big enough gel to compare all my samples.

I planned out my experiment to be consistent between samples, but I think there should be some kind of readout demonstrating that the enriched RNAs isn't just from more IP beads, more protein, etc. How do you incorporate a control to demonstrate consistency between sample to allows for relative "levels" comparison?

[SeeElle]

I see that folks have viewed my post, but there hasn't been a response. Has my question already been addressed elsewhere? Could someone direct me to the thread(s)? Thanks!

[bioinfosm]

My guess is that not many people are doing RIP-Seq yet!

[GKM]

People aren't doing it, because of the very high background associated with the RIP procedure. If your protein is an RNA-biding protein, i.e. you know it binds RNA specifically and tightly, CLIP is a better alternative, it's much harder to do though

[bioinfosm]

CLIP! Guess will have to search that one up..

[JGU]

CLIP it, not RIP it

As mentioned, the RIP assay can be pretty high background assay and it is also littered with misinterpretations. Anybody thinking about RIP should read this paper and then decide whether they want to proceed:

http://rnajournal.cshlp.org/content/10/11/1692.full.pdf?ck=nck

The question of affinity is really where things get complicated and are you pulling down RNAs that are relevant to biology in the cell or what happens in the lysate conditions? All skepticism aside, RIP has been used in a number of nice pubs by Pat Brown & Dan Hershlag's labs for pulling down RNA binding proteins (yeast) and then doing arrays, but I'm not sure if they have published RNA seq from those pulldowns yet. I know they are doing these kinds of expts now.

As mentioned by others, CLIP is better because it offers the high stringency washes, crosslinking of the RNA to the protein, nitrocellulose selection of protein-RNA adducts, etc. BUT, the problem with CLIP is the need for a high quality antibody. If the organism of interest is amenable to tagging at the endogenous locus (e.g. S. cerevisiae), then there are great commercial antibodies (e.g. Myc, FLAG) to overcome the need for a good antibody. This has also been done some in mammalian cells by overexpressing a tagged version of a protein (see Jeremy Sanford's work on ASF/SF2) One has to be careful with overexpression stuff though...

[SeeElle]

I considered doing the CLIP, but does anyone understand the chemistry to support that ONLY RNA-protein binding occurs (and not ptn-ptn; RNA-RNA)?

[JGU]

RIP seq xlinking

The first published method for CLIP (PMID: 16314267) involved shortwave UV light as the inducer of crosslinks. These are essentially "zero length" crosslinks, meaning that there is no linker chemical so it has to be direct crosslinking of an RNA base to a protein. There is definitely DNA-DNA crosslinking happening under these conditions (T-dimers) but I'm not sure about RNA-RNA...probably some, but those guys probably aren't converted to cDNAs especially well.

In any case, this type of UV crosslinking is quite inefficient and only a percent or two of the RNA gets crosslinked to a protein. My guess is that it is mostly adducts formed by stacking of an RNA base with an aromatic amino acid. Because of this, some proteins that actually bind very well to RNA do not crosslink efficiently. That's obviously an issue and yet another reason why a really good antibody is necessary. You need to capture anything that is actually crosslinked. I don't think a lot of protein protein crosslinks are formed under UV light, and in any case you couldn't see those by RNA seq anyway.

To increase the efficiency, Tom Tuschl's lab (PMID: 20371350) began feeding the cells 4-thio-uridine containing nucleotides...this is an old trick used to investigate the spliceosome (PubMed search for Sontheimer and Steitz, 1992-1994). ThioU makes the cells pretty sick, but it also makes it so that you can use longer wavelength UV (365nm) and then you only get the 4-S-U crosslinking to nearby RNA or protein. It's more efficient, so this way you get more protein-RNA adducts as starting material for your pulldown. Obviously this doping with ThioU is impossible to do in an organism/organ context like the first CLIP papers did.

[ETHANol]

CLIP-seq looks like a promising progression but it is limited to proteins that bind directly to the RNA. Does anybody know of a method that will identify RNA binding sites of proteins that are in a complex bound to RNA but not directly bound to the RNA, i.e. they are bound to the RNA via a protein-protein interaction?

I see people are doing 'RNA-ChIP' using formaldehyde x-linking, is there a reason that this won't work for 'RNA-ChIP-seq'?

One more question, in CLIP-seq the proteins are X-linked with UV. These X-links are not reversed. Doesn't this interfere with reverse transcription? Does the method possible select for a subset of RNA's that that still form templates for reverse transcription and a subset that doesn't, there by skewing the results?

[patateplate]

Does anybody have a good detailed protocol for RIP-seq or CLIP-seq which includes the library preparation. I'm mostly interested in an in-house protocol for library preparation instead of the illumina or epicentre kits. Thanks,

[NextGenSeq]

Once you get the RIP or CLIP cDNA you can use any protocol. You usually get tiny amounts of nucleic acid from any IP though and you'll have to increase the amplification cycle number.

[yjhua2110]

there is a public platform, starBase, for decoding protein-RNA and microRNA-target interaction maps from CLIP-Seq (HITS-CLIP, PAR-CLIP) and degradome sequencing (Degradome-Seq, PARE) data.

[tboothby]

SeeElle, did you ever try RIP-seq? What was the outcome?

Anyone else know anything more about RIP-seq? Pit falls, controls, etc?

[mudshark]

One more question, in CLIP-seq the proteins are X-linked with UV. These X-links are not reversed. Doesn't this interfere with reverse transcription? Does the method possible select for a subset of RNA's that that still form templates for reverse transcription and a subset that doesn't, there by skewing the results?

as far as i know the UV crosslinks do interfere with reverse transcription. I think that is the reason why you can exploit CLIP for identification of the binding site on the RNA as well - that's where RT stalls.

[[email protected]]

I've got the RIP-seq analysis outputs. We used an isotype control antibody VS antibody against an RBP to pull down corresponding RNAs!

I did the following pipeline: Hisat2 (mapping the reads after QC) -> htseq-count (getting read counts) -> DESeq2 (getting enriched RNAs in test VS isotype control) -> volcano plot shows up-regulated and down-regulated genes in the test sample.

Logically, I should have seen just enriched RNAs (up-regulated genes), but surprisingly there are some RNAs down-regulated in the test sample (or up-regulated in the isotype control sample)!!!!!
How can we explain and interpret this???? it seems some genes are pull down in isotype control sample as well. Are these background? Is my computational pipeline right?

BTW, it's worth to mention that all enriched RNA (called up-regulated genes in DESeq2 pipeline) are significantly relevant and meaningful to my experiment. However, I am not sure how much confident I can be on this output? And how should I explain down-regulated RNAs? because normally we just expect enriched RNAs in RIP-seq experiment.