I'd check it with luciferase reporter assay, yes, but you'd have to put a minimal promoter for the RNA Polymerase. If the binding site is near an exon, I'd clone this exon too in order to add the sequence context. It's also possible to clone the genomic region with a minimal promoter. Nevertheless, if you have identified this site by ChIP (that it's a in vivo experiment) maybe it'd be enough if you checked your data by EMSA and the expression of the target gene when the TF it's overexpressed by transient expression, for example.

The same case has been reported for FLC in Arabidopsis thaliana, which it's capable to bind both the promoter region and the first intron sequence of FT gene.

Dear Maria,

Yes, doing the so called enhancer trap experiment using luciferase based reporter can provide you an additional level of proof. The next level would be so called 3C (Chromosome Conformation Capture) experiments and show that the implicated enhancer gets to the proximity of the promoter. Absolute proof requires genome engineering.

Regards,

Laszlo

Cloning the putative regulatory sequence into a reporter construct would test that the DNA element can regulate gene expression (enhancer function should be independent of upstream/downstream). Genetic deletion of the binding site would give you a stronger argument that your TF binding site is important for regulation of it's target gene. If it's an enhancer you could additionally use chromosome conformation capture to show that it is interacting with the promoter of your gene of interest.

I'd check it with luciferase reporter assay, yes, but you'd have to put a minimal promoter for the RNA Polymerase. If the binding site is near an exon, I'd clone this exon too in order to add the sequence context. It's also possible to clone the genomic region with a minimal promoter. Nevertheless, if you have identified this site by ChIP (that it's a in vivo experiment) maybe it'd be enough if you checked your data by EMSA and the expression of the target gene when the TF it's overexpressed by transient expression, for example.

The same case has been reported for FLC in Arabidopsis thaliana, which it's capable to bind both the promoter region and the first intron sequence of FT gene.

Hi Maria,

Yes it make sense but to really prove it, I think you will have to do one more thing.

You can clone :

- your predicted regulatory region in the reporter vector of your choice and hopefully you we see that it will drive the expression where you expect it --> gives a good clue

- clone a second time the same region but introduce a/some mutation(s) in the binding site (but for that you have first predict where is locate your binding site within the intron) and hopefully you will loose the expression --> in combination with the previous point it validates your binding site

You can also use a mutant (of your TF):

- put your construct in the mutant context and see if you are loosing expression.

- see if the gene of interest is down regulated (or upregulate if your TF is a repressor) in this context (qRT-PCR, or RNA-seq if you want to test a lot of genes).

Best,

Delphine

Dear Maria, Beside the suggested techniques by the people who answered already, it would be also great to see the typical markers of an enhancer on the binding site, at a minimum change in H3K27 acetylation upon activating the gene or by silencing the identified TF factor. Showing change in the enhancer transcripts on the binding site by QPCR upon activation or silencing the TF would give another layer to the details of the regulation.

Hi Maria,

I would compare the WT system to a mutant one. The mutant system should contain maybe a 1 bp mutation that removes binding to that binding site(s). You can then measure the expression level with qPCR in the WT vs the mutant. Changes in expression levels should indicate that binding in the intron is functional.

Radu

Hi Maria,

the classical experiment would be to clone the respective region into a yeast one-hybrid compatible vector system, once with a native binding consensus sequence and a second time with a mutated binding site. Then co-expression of a GALa-AD-fusion of your TF-of-interest and look for reporter gene activation (both auxotrophy complementation/growth [qualitative] and b-GAL assay [quantitative]).

Anyway, for publication, it does not really make sense to verify an 'in vivo' technique, such as ChIP, with an 'in vitro' assay - Therefore, whatever you do, you better go for a in vivo method of validation. Thus, qPCR of x-ChIP - as was suggested above - will only qualitatively validate your ChIP experiment, but not validate your binding in an intron sequence.

An 'intronic' sequence upstraem of a reporter gene does not make any sense, as you lack the UTR and possibly a core-promoter. Hence, your ribsomal attachment site is missing and your protein reporter will probabbly end up in the NMD or ERAD pathways.

You can use such an assay in combination with a minimal core-promoter that does not elicit any background activity on it's own. There are suitable vectors for plants and worm, but I don't know of any that are applicable in mammals (human or mouse).

Dear Maria,

I would suggest accumulating other ChIP-seq data to help determine whether the site might be an enhancer. So, H3K4me1 or DNAse! HS data from ENCODE might help if your cell type is in their pipeline. Also see if it is conserved. Then you would need to test whether it is an enhancer by luciferase assays....I would clone the region of conservation into a luciferase vector with a basal promoter and test its function in some appropriate cell line. This is what we did to the intronic enhancers fo the E2f2 and E2f4 genes - see Tallack et al, . Journal of Biological Chemistry (2009) Jul 31;284 (31):20966-74

Hi folks,

I see your points - but, if Maria really is after a TF-binding motif, I am not really sure how the ENCODE data or Histone-modifications will be supportive for the binding of a specific TF under investigation?

As she did not disclose, which organism she is interested in, we can not really conclude specificity of TF-binindg from luciferase-assays if she does not have k.o.-mutants for her TF-gene under investigation. And making TF-mutants in humans is a ethical issue, isn't it.

Therefore, she needs to use a reporter-assay, where she can assay binding of her TF in a native environment, but on mutatetd binding sites or with mutated TF.

Otherwise the data will be distractive.

Cloning of the intronic site into a reporter vector will for sure reveal some activity, but it will not relate to the specific binding of a TF-of-interest.

The fastest assay to reveal in vivo binding specificity is the yeast one-hybrid. In addition, it can most likely be adopted also to your organism of choice. e.g. we used a one-hybrid assay vector on a LUC-reporter in plants (Brand et al, 2013, PloS ONE).

Two experiments. First, just put the binding site from intron in the basal promoter region of promoter-reporter construct, and see its effect. Second, perform luciferase or other reporter assay placed downstream of genic fragment without the intron in question.

Above suggestions are good. Alternatively, you can do 3C or 4C to show the functionality of the regulatory elements with T binding.

reporter assays are good. Alternatively, you can do 3C or 4C to show the functionality of the regulatory elements with T binding.

Dear Colleagues,

thank you all very much for your helpful input and inspiration. Summarizing your comments, I think my next steps would be:

1) Cloning the regulatory intronic sequence into a reporter construct upstream of a core minimal promoter. The ChIP-Seq peak ranged smth around 300bp and lays in the 4 intron, with around 1500bp distance to the neighboring exons. I think I will generate two constructs, one with the whole inton 4 and parts of exons for context (as suggested by Jose) and one with only putative sequence?

2) I will also clone the proximal promoter of GOI (though no ChIP-Seq peak there) into the reporter plasmid to test in parallel, just curious.

3) I was also thinking of EMSA but unfortunately within the ChIP-Seq peak no canonical binding site for my TF was predicted by Transfac and NURSA (my TF is a nuclear receptor). Or, if I use less stringent parameters, too many sites are predicted with weak scores, so that I could not develop a particular strategy to go for that..

4) Activation of TF leads to the strong upregulation of mRNA of GOI, confirming again the hypothesis that this TF could be a direct regulator. The same with knock-down of TF and concomitant downregulation – so this part is already done. But, as Dierk mentioned, these studies confirm the fact of regulation but do not give me the clue about the direct binding.

5) Great suggestion from Delphine to use mutant cDNA (dominant negative) of my TF – but it again does not help much in the identification of a direct binding site.

6) Genetic deletion of the binding site in the natural environment would be really a great thing. How can I do it?

Hi Maria,

You can do genome editing with CRISPR/cas9 but I don't think it's the easiest way to prove it. Just introducing the mutation in the TFBS in the region and clone it again seems easier.

But for any of these solutions you first need to know where is located the binding site(s) (unless you want to remove the entire peak?).

May be I can help for trying to detect a good TFBS if you can explain a bit more what you've been using to find it (PWM, tool, background etc...). Here are some things you can think about if it's not already done:

- Did you performed a motif discovery across you ChIP peak, and where you able to find back something similar to a NR motif?

- Then to find you TFBS did you use a PWM from the literature or a motif generated from your peaks?

- Do you take into account conservation of the predicted BS across species (more chance to be functional if it is conserved)?

Hi Maria,

Maybe, you want to try screening for binding sites.

We recently published the DPI-ELISA screen method, which is allows screening for any binding site of any given TF, which is detectable by antbodies (Brand et al, 2013, PloS ONE).

In principle, it follows the technique of Protein Binding Microarrays (PBM) described earlier, but it is much cheaper. Our DPI-ELISA screen will give you at least a hint, or readily disclose the binding consensus if it's not longer than 6mers - for the library we regularly use. Given, you still work in Stutgart, maybe you want to make a trip to Tuebingen...

Anyway: For conclusive EMSAs you will need to have purified protein - or you need to make supershifts, which sometimes does not work properly. In addition, 300 bp might be already cause problems du to its size.

You should also keep in mind that the public database releases contain only about 10% of eukaryote TFs with a known binding motif (except for Arabidopsis, 19%) - they are very much biased by bactrial TFs. Thus, it is not really surprising that you did not find anything of significance! In addition, they work good on bZIP and zinc-fingers. But they lack TFs-info for other families, e.g. scaffold-binding or minor-groove-binding proteins, that usually make contact with only 4 bp!

Genetic deletions in human cell lines might work with TALENs. However, if your TF-DNA interaction is compulsory for proper cell growth and development, you will never get your mutant line... Thus, you will not know WHY and IF your binding protein is required and sufficient for this 'phenotype'.

Hi Delphine/Maria,

I just wanted to point out that reliable motif finding is one of the hardest problems in computational biology. You should read Tompa et al. (2005, Nature Biotechnology 23, 137 - 144; doi:10.1038/nbt1053). They showed that virtually all motif finders were insufficient.

Radu

Hi Radu,

I was just asking questions about what they did. Even if I agree motif discovery is not straightforward, I think it is still very useful (also new tools have been published since 2005 and the Tompa paper is not using ChIP-seq data but smaller data sets).

If I talked about motif discovery, it's because :

-I always trust more ChIP data when I am able to find back a similar motif to the known one in the peaks.

-just running a motif discovery analysis can help to understand why she a not finding a very good candidate in the 300bp peak. For example it can help to choose the best PWM to use (e.g.: in drosophila we are having multiple PWMs for NRs).

maybe someone's already mentioned this but to validate enhancer activity in a reporter assay you'd want to clone the target sequence in forward and reverse orientation. The chromatin features mentioned of course say nothing about functionality. I would follow up with EMSA if you can.

Dear Delphine,

I agree that motif discovery can be informative. I was just trying to draw the attention that it is not a straightforward method, but rather a complex approach. The in vivo binding (that is observed in the ChIP-seq) is controlled by the binding energy between the TF and the DNA words (represented through the PWM), but also by other factors, such as DNA accessibility or TF concentration in the nucleus. This is why I think that this method should be complemented by other approaches (e.g. conservation of the sites, like you suggested, or/and in vitro binding methods, such as SELEX or PBM).

Radu

What about in vivo experiment? If you work on human of course this is very limiting...!

1) show in vivo that your 2 genes are coexpressed in the same cells and same tissue: mouse, human other?

2) clone your enhancer and deleted enahncer into reporting construct with minimal promoter to detect the region(s) necessary for the activity. Do trangenics (zebrafish, mouse) for all of them. Find this way what was the minimum region leading to specific expression

3) Mutate your binding in the context of the enhancer. Check in transgenic if the expression persist or not.

4) concatenate your motif, do transgenic and check if this minimal contrsuct is sufficient to drive expression.

Functional analysis of enhancer is tricky and time consuming. Then it depend what level of confidence you want to achieve

TRANSFAC and similar predictive methods are notoriously known to not be very reliable. First, if you have a recombinant TF of your interest, the binding region is confined to ~300 bp through the "peak" analysis, you can conduct the old fashioned in vitro DNase I footprinting assay to precisely locate its binding site(s). Alternatively, you test a series of 40-50bp partially overlapping oligonucleotides by EMSAs. To locate the binding site, you will need to test a series of mutations as soon as you identify the bound probe(s).

Concerning the reporter gene assay, using luciferase implicates that you want to test the enhancer activity in cell culture systems. While there is many great cell lines, the issue is the type of your gene of interest, if it is a tightly developmentally controlled gene, analysis in transgenic mouse using GFP reporter would be preferred. Alternatively, zebrafish is a good system. To proof that the site(s) is functional, its mutagenesis within the reporter construct is highly desirable.

Many thanks once again to all of you for the helpful comments and feedback!!

Of course the tests in vivo using mutant animals would be fantastic, but frankly it extends a bit the initial boundaries of this particular project, which emerged anyway by accident. So, I will try to use the tools we have more or less established in the lab.

By th way, I have checked the ENCODE ChIP-Seq data for the enhancer specific histone modification mark, H3K4me1, for this region in the cell line of tissue of interest - and it was positive! Although there were several peaks for H3K4me1 within this intron and one of them covered the TF peak of my data. Of course, it´s just a hint, but at least in a right direction.

Another thing, whcih a colleague of mine recommended me to do - to generate luciferase reporter constructs containing the regulatory element either upstream or downstream of luciferase gene. I guess it´s also worth trying..

And yes, Ales, that´s really good suggestion - to conduct DNase I footprinting assay, which I unfortunately know only in theory. Thank you!

Hi!

I did not read all the answers to your question but i ll try to have a relevant input. ChIP-seq experiments do not actually tell you that the protein is binding DNA directly. There is a very good chance that if you can not find a canonical binding sequence for your TF it is because it is recruited to the Region of interest by interaction with another factor. I would suggest that you look for binding sites of known interactor proteins.

Also, most canonical sequences can vary greatly. I would suggest that you look at your sequence manually and try to find a sequence that looks similar to the consensus one. Then test binding with EMSA.

From what you are describing, I would say it is fairly reasonable to think it is an intronic enhancer. These can function in many ways and often increase splicing or elongation. I would for binding of factors involved in either process.

Hi Maria, you probably got the answer you are looking for already, but just in case you are still looking, I just designed some similar in vitro functional assays to validate chip seq data, and used the pgl4 vector with the minimal promoter (pgl4.23). You can clone any putative regulatory sequence into this vector, even it's not naturally near the TSS, as this contains the required elements for basic expression. It worked well for us.

Thanks Agnes, have you generated/tested both constructs - with the regulatory sequence upstream and downstream of luciferase?

The place where you clone is always upstream of the luciferase gene (take a look at the vector map, it makes it a bit more clear). It is the natural position of the regulatory element in the genome that doesn't matter because you don't need the minimal promoter at the TSS, this is provided by the vector. I've cloned distal enhancers, proximal promoters and introns into this vector and co-transfected with my TFs. Hope this helps clarify, but let me know if I can provide any more info.

Hi, I haven't read all answers but I'll give you my opinion before trying wet stuffs: you may want to try to find Gene Expression profiling data for your TF KD in this or a similar cell line or tissue (ENCODE or GEO datasets) and check if your target gene's expression is affected. That would be a good indication that if your TF binding is limited by lack of proteins the functional output is deregulated. Good luck!