Mathematical models, with some initial assumptions (based on known properties of protein DNA interaction), diffusion and other properties are used to create these notions. At present many of these are theoretical explanations of how transcription factors bind to their targets. So, it is rather a feet of imagination to understand these models. Some of the features (known currently) that might help you imagine the process better are

1) For any given gene in a eukaryote, there are two copies (diploid), ergo limited number of transcription factor binding sites. These few sites are sought by at least several hundred transcription factors randomly scanning the genome. So the probability of binding goes up.

2) In a non-mitotic cell the chromatin is thought to be a mesh with several parts of actively transcribed regions brought close together and heterokaryon spaced out. This demarkation might act to funnel the transcription factors to the binding site.

3) It is also known that tertiary interactions mediated by secondary factors (non-coding RNA) that associate with the RNA polymerases cause formation of "transcritosomes" or clusters that drive the expression of several genes regulated by a single transcription factor simultaneously. (Such transcriptosomes have been observed for yeast Gal4 systems)

To address your second concern:

Transcription factor not-binding will not be detected (No transcription). But, because genes do express, it is safe to assume that one of the several hundred transcription factor molecules did eventually bind to DNA. So the better question is : What is the delay between activation of the signal and the binding of the transcription factor? This depends on the percentage of active transcription factors and diffusion. Now two factors that hasten the diffusion process are the dense packing of the DNA in nucleus (or the proximal positioning of the transcription factors to DNA in prokaryotes) and low affinity of the factors for non-canonical sites. So they take less than a few nanoseconds to come off of an non-canonical site and have to scan for sites in a small closed space. Diffusion at the molecular level is like driving with a constant rocket booster. These factors together make transcription factor binding a routine and not a rare event.

The bigger problem is transcription factor accidental binding. Again one could only "speculate about fail safe mechanisms" to avoid transcription factor "accidental binding"

1) Most transcription factor binding initiates recruitment of other transcription factors and then the transcription machinery. Some of the other factors will also need appropriately placed DNA binding sequences. This could serve as a check list for initiating transcription. (For example the MAD-Madea complex in Drosophila TGF-beta signaling pathway)

2) Transcription inhibitors or suppressors could compete for the same DNA binding site and prevent accidental initiation. (In TGF-beta signaling this role is played by inhibitory SMADs)

3) Leaky transcription is known to occur in several genes. These events could be left to evolution (natural selection) to prune out.

This is a very important question and we are very far from giving a good answer.

One need to keep in min that there are very different DNA-binding proteins that display a broad range of binding capabilities.

Some proteins read the sequence, other measure the size of the minor groove and again some others might solely recognize possible binding sites due to DNA-shape.

One also need to keep in mind that our transcription factor (TFs) knowledge is biased for prokaryote TFs. Eukaryote TFs are rarely understood - for about 10 % of the TFs in a model species a possibke binding consensus is described. But these might not reflect the in vivo binding sites. Many of them recognize a range of binding sites, both low and high affinity sites.

From our data, we see that the specificity of a TF is context dependent, which implies that several different factors communicate with each other in form of a module. Finally, leaving only the high-affinity binding sites that are localized in the appropriate genomic environment that generate the specificity of the output.

In vitro techniques exist that allow the estimation of binding strength. From these data, it is very unlikely that TFs dissociate from DNA at 'room temperature' in the living cell. Hence, permanent removal of TFs by the proteasome is required in eukaryotes.

Hence, they will probably not 'slide'.

Recomended reads are:

Badis et al. (2009). Diversity and complexity in DNA recognition by transcription factors. Science 324, 1720-1723.

Rohset al. (2009). The role of DNA shape in protein-DNA recognition. Nature 461, 1248-1253.

Sathiya and Deirk, thanks for your outstanding and detailed replies. I accept most of the points you make but I have some lingering questions:

1. I seriously doubt that "random scanning" of the genome is an efficient enough process. Is there any evidence that TFs search by "brute force"?

2. How do TFs avoid accidental collisions within the nucleus?

3. If TFs fail to bind correctly with their recognition site, will some other TF try to do so? Is there, in a sense, competition like between sperm cells for the ovum?

4. How does the cell send TFs to the proteasome for recycling after they have done their job and prevent them from continuing to activate genes?

1. personally, I am more and more convinced that most TFs bind randomly, where the chromatin is accessable. Maybe, they are thrown off the DNA at low affinity sites by competition. Maybe, they scan until they find a binding consensus that fits, stay there until they are competed or degraded. I am a plant scientist, and we found that several TFs are always bound to DNA, despite the fact that they transduce a specific signal. Thus, I think that much of a TF function we notice as a distant observer is the concerted action of the genomic context, where more than only the TF of our interest has bound.

2. I am convinced that they probably will collide with almost everything insite the nucleoplasm. Most probably they will also hit DNA from the wrong angle and not with its DNA-binding domain in the first place.

3. We try to nail that by our DPI-ELISA, but it is much more tricky than I expected, because of several reasons. EMSA are only qualitative and are not sufficient to address the question. At present there is convincing in vitro or vivo data that supports a competition of let's of one TF at a low affinity site by a high affinity binding protein. I don't know of any publication describing such an approach in a sound manner.

4. So far, it seems that there is constant degradation, possibly at random. Maybe without specific SCF or RING involvement. The nice thing about this model is that it readily explains how the signal is mantained or switched off. Either the TFs are replaced from a pool inside the nucleus, or not.

Sorry, there is a typo:

@3: We have NO convincing evidence for competition!!

without an evidence it is difficult conclude and therefore there are hypothesis. As Dierk mentioned, TFs have broad range of binding capabilities but most of them has specific type of mode of actions.

Because of their role in the cell i am of the view that DNA binding should not be random. first of all it does not appear to be energy efficient and several TF act within no time. if it is random scanning it has be be extremely fast and energy intensive.

Manoj, I know this may appear silly, but could it be that TFs somehow "know" where to look for their binding sites and don't just search for them exhaustively by random scanning. Could there be some guiding force or principle at work? As you say, a blind search would be energy inefficient, time consuming, and I suspect would result in a lot of accidents and collisions.

Joseph, well this is a open question. there are theoretical models that try to explain such processes (http://www.yumpu.com/en/document/view/17964505/a-model-of-transcription-factor-dynamics-on-dna-mit-press). Walking, sliding, hopping or jumping have been suggested to play role in the process of finding its target site.

I personally believe there is a guiding principle that we are not completely clear about. for example, suppose a TF-X target 10 genes i.e. these 10 genes has a binding site in their promotes where TF-X can bind. Now all these proteins are not required at a single time but they have specific roles in various environmental conditions. how TF-X can bind to required gene promoter whose product is required at that point of time and to some other gene if you change the environmental condition. i dont think that is random. it is a very specific process.

there is another article which is talking about role of DNA sequence correlations in this process (http://www.cell.com/biophysj/abstract/S0006-3495(11)00513-3).

I concur with most of the points that Dierk has raised. I also want to add the role of non-coding RNAs and chromatin modifiers in eukaryotic transcription. It is possible that some non-coding RNAs could act as liaisons, linking the transcription factor to its DNA binding pocket (like an anchor to prevent it from drifting off). In this scenario, the protein could be favorably placed for activation. The chromatin modifiers could make different genes accessible or otherwise according to the cell's needs. Again these are tentative and not experimentally verified hypothesis.

Regarding evidence for brute force scanning: There is no direct evidence but some extrapolation from bacterial enzymes that work on DNA. Restriction enzymes are thought to have a weak binding to DNA molecules in general. Once they bind to the DNA, they could scan a short stretch like this animation shows and come off. This model provides a different dynamic interaction between the DNA and the protein, than brute force bouncing model. It is not a random dart game of hitting the right sequence (TFs are not randomly bouncing off the DNA), but rather a chance game of spinning a wheel and stopping when you get there (Bind and scan by traveling a short distance- If you find a target activate target. If no target then unbind.

http://www.youtube.com/watch?v=aA5fyWJh5S0

Like Dierk pointed out, proteins do collide with their environmental molecules, presumably in all possible ways. In contrast to a macroscopic collision these collisions are rarely detrimental to the protein. (At the atomic level no two molecules come too close unless they form a bond, because of electrostatic repulsion). So even if the proteins collide- So what?. They are going to survive.

Yes, competition exists for the limited sites for TFs to bind. But the transcription factor binding affinities are not the same. Neither are their concentration. So it is safe to say that the TF with the most affinity for a given site, and the one that has the highest affinity gets the most "binding time" (if you will). It is analogous to the game musical chair- Imagine being played by people of different size (large person = high concentration, quick person =high affinity). So a large and quick person wins by sitting on the chair more times than the others. But they all have to get up and run when the music is turned on. So do TFs due to the so called breathing effect or Off rate.

Degradation is not necessarily the only way to prevent transcription.There are several ways in which a transcription factor could be inactivated. Most famous transcription factors in Eukaryotes have to be phosphorylated to be active. Phosphorylation domain is different from the DNA binding domain and the phosphorylation is thought to activate DNA binding by changing the conformation of the protein. So one could again imagine dephosphorylating it will act as an inactivation signal. Ex. SMADs in TGF-beta signaling. Small molecule binding may be necessary for activation. Inactivation means simply withdrawal of the small molecule (by diffusion/ competition). Ex. Progesterone Binding receptors. It is also important to take into account the transport of transcription factors in and out of the nucleoplasm as it could alter the dynamics of transcription factor binding.

very interesting question! exactly the very question in my mind. good luck. I hope you can figure something out.