Tuesday, June 20, 2017

Spooky action at a (short) distance

Entanglement in physics is about action that seems to transfer some sort of 'information' across distances at speeds faster than that of light.  Roughly speaking (I'm not a physicist!), it is about objects with states that are not fixed in advance, and could take various forms but must differ between them, and that are separated from each other.  When measurement is made on one of them, whatever the result, the corresponding object takes on its opposite state.  That means the states are not entirely due to local factors, and somehow the second object 'knows' what state the first was observed in and takes on a different state.

You can read about this in many places and understand it better than I do or than I've explained it here.  Albert Einstein was skeptical that this could occur, if the speed of light were the fastest possible speed.  So he famously called the findings as they stood at that time "Spooky action at a distance." But the findings have stood many specific tests, and seem to be real, however it happens.

Does life, too, have spooky action? 
I think the answer is: maybe so.  But it is at a very short distance, that within the nuclei of individual cells.  Organisms have multiple chromosomes and many species, like humans, have 2 instances of each (are 'diploid'), one inherited from each parent.  I say 'instances' rather than 'copies', because they are not identical to each other nor to those of the parent that transmitted each of them.  They are perhaps near copies, but mutation always occurs, even among the cells within each of us, so each cell differs from their contemporary somatic fellows and from what we inherited in our single-cell beginnings as a fertilized egg.

Many clever studies over many years have been documenting the 3-dimensional, context-specific conformation, or detailed physical arrangement of chromosomes within cells.  The work is variously known, but one catch-term is chromosome conformation capture, or 3C, and I'll use that here.  Unless or until this approach is shown to be too laden with laboratory artifact (it's quite sophisticated), we'll assume it's more or less right.

The gist of the phenomenon is that (1) a given cell type, under a given set of conditions, is using only a subset of its genes (for my  purposes here this generally means protein-coding genes proper); (2) these active genes are scattered along and between the chromosomes, with intervening inactive regions (genes not being used at the moment); (3) the cell's gene expression pattern can change quickly when its circumstances change, as it responds to environmental conditions, during cell division, etc.; (4) at least to some extent the active regions seem to be clustered physically together in expression-centers in the nucleus; (5) this all implies that there is extensive trans communication, coordinating, and physically juxtaposing, parts within and among each chromosome--there is action at a very short distance.

Even more remarkably, I think, this phenomenon seems somehow robust to speciation because related species have similar functions and similar sets of genes, but often their chromosomes have been extensively rearranged during their evolutionary separation. More than this: each person has different DNA sequences due to mutation, and different numbers of genes due to copy number changes (duplications, deletions); yet the complex local juxtapositions seem to work anyway.  At present this is so complicated, so regular, and so changeable and thus so poorly understood, that I think we can reasonably parrot Einstein and call it 'spooky'.

What this means is that chromosomes are not just randomly floating around like a bowl of spaghetti.   Gene expression (including transcribed non-coding RNAs) is thought to be based on the sequence-specific binding of tens of transcription factors in an expression complex that is (usually) just upstream of the transcribed part.  Since a given cell under given conditions is expressing thousands of condition-specific genes, there must be very extensive interaction or 'communication' in trans, that is, across all the chromosomes. That's because the cell can change its expression set very quickly.

The 3C results show that in a given type of cell under given conditions, the chromosomes are physically very non-randomly arranged, with active centers physically very near or perhaps touching each other.  How this massive plate of apparent-spaghetti even physically rearranges to get these areas together, without getting totally tangled up, yet to be quickly rearrangeable is, to me, spooky if anything in Nature is.  The entanglement, disentanglement, and re-entanglement happens genome wide, which is implicitly what the classical term  'polygenic' essentially recognized related to genetic causation, but is now being documented.

The usual approach of genetics these days is to sequence and enumerate various short functional bits as being coding, regulatory, enhancing, inhibiting, transcribing etc. other parts nearby.  We have long been able to analyze cDNA and decide which parts are being used for protein coding, at least. Locally, we can see why or how this happens, in the sense that we can identify the transcription factors and their binding sites, called promoters, enhancers and the like, and the actual protein or functional RNA codes.  We can find expression correlates by extracting them from cells and enumerating them.  3C analysis appears to show that these coding elements are, at least to some extent, found juxtaposed in various transcription hot-spots.

Is gene expression 'entangled'?
What if the molecular aspects of the 3C research were shown to be technical artifacts, relative to what is really going on?  I have read some skepticism about that, concerning what is found in single cells vs aggregates of 'identical' cells.  If 3C stumbles, will our idea of polygenic condition-specific gene usage change?   I think not.  We needn't have 3C data to show the functional results since they are already there to see (e.g., in cell-specific expression studies--cDNA and what ENCODE has found). If 3C has been misleading for technical or other reasons, it would just mean that something else just as spooky but different from the 3D arrangement that 3C detects, is responsible for correlating the genomewide trans gene usage.  And it's of course 4-dimensional since it's time-dependent, too.  So what I've said here still will apply, even if for some other, unknown or even unsuspected reason.

The existing observations on context-specific gene expression show that something 'entangles' different parts of the genome for coordinated use, and that can change very rapidly.  The same genome, among the different types of cells of an individual, can behave very differently in this sense. Somehow, its various chromosomal regions 'know' how to be, or, better put, are coordinated.  This seems at least plausibly to be more than just that a specific context-specific set of transcription factors (TFs) binds selectively near regions to be transcribed and changes in its thousands of details almost instantly.  What TFs?  and how does a given TF know which binding sites to grab or to release, moment by moment, since they typically bind enhancers or promoters of many different genes, not all of them expression-related.  And if you want to dismiss that, by saying for example that this has to do with which TFs are themselves being produced, or which parts of DNA are unwrapped at each particular time, then you're just bumping the same question about trans control up, or over, to a different level of what's involved.  That's no answer!

And there is even another, seemingly simpler example to show that we really don't understand what's going on: the alignment of homologues in the first stage of meiosis.  We've been taught that empirical and necessary fact about meiosis for many decades. But how do the two homologues find each other to align?  This is essentially just not mentioned, if anyone even was asking, in textbooks.  I've seen some speculative ideas, again involving what I'll call 'electromagnetic' properties of each chromosome but even their authors didn't really claim it was sufficient or definitive.  Just for examples, homologous chromosomes in a diploid individual have different rearrangements, deletions, duplications, and all sorts of heterozygous sequence details, yet by and large they still seem to find each other in meiosis.  Something's going on!

How might this be tested?
I don't have any answers, but I wonder if, on the hypothesis that these thoughts are on target, how we might set up some critical experiments to test this.  I don't know if we can push the analogy with tests for quantum entanglement or not, but probably not.

One might hope that 'all' we have to do is enumerate sequence bits to account for this action-at-a-distance, this very detailed trans phenomenon.  But I wonder......I wonder if there may be something entirely unanticipated or even unknown that could be responsible.  Maybe there are 'electromagnetic' properties or something akin to that, that are involved in such detailed 4D contextually relativistic phenomena.

Suppose that what happens at one chromosomal location (let's just call it the binding of a TF), directly affects whether that or a different TF binds somewhere else at the same time.  Whatever causes the first event, if that's how it works, the distance effect would be a very non-local phenomenon, one so central to organized life in complex organisms that, causally, is not just a set of local gene expressions.  Somehow, some sort of 'information' is at work very fast and over very short distances. It is the worst sort of arrogance to assume it is all just encoded in DNA as a code we can read off along the strand and that will succumb to enumerative local informatic sequence analysis.

The current kind of purely local hypothetical sequence enumeration-based account seems too ordinary--it's not spooky enough!

3 comments:

Ken Weiss said...

I want to stress, if anybody's actually reading this post, that it is not intended to dismiss decades of work on gene-regulation! Cis-acting effects (enhancers, promotors, spacers, introns, exons, poly-A tail regions, locus control regions, and others like it) are so well-documented, so broadly, and under so much very close experimental control that they can hardly be questioned. What can be questioned is how these regions are identified and acted on in 4-dimensional trans-genomic ways. This can't too easy to explain because there are no fixed rules for the location, number, type, and so on of these regulatory signaling and binding regions.

My claim, or suggestion, is that there is something _else_ going on that is not just local cis-action, but in trans. What that is, or how things work if it isn't, is what I think deserves attention, that goes beyond the current essentially cis focus.

geoff said...

Hi there, I just came across your blog via the omnigenic discussion on twitter (thanks for sharing your perspective-- interesting!), and saw that the latest post is a little closer to things I've worked on.

Briefly, there is growing body of evidence, much of it from 3C-based technologies, supporting the idea that mammalian chromosomes are organized throughout the cell cycle by molecular motors that translocate along the chromatin fiber & progressively extrude chromatin loops: in metaphase the loop extruders are present in high numbers to effectively compact chromosomes, and in interphase they are present at lower numbers and additionally halted at certain boundary elements (e.g. cohesins in interphase http://www.cell.com/cell-reports/abstract/S2211-1247(16)30530-7, condensins in metaphase https://elifesciences.org/articles/14864)

From this perspective, the very recent report of directional translocation by the SMC condensin on DNA curtains, while perhaps making things less molecularly spooky, is very exciting! http://biorxiv.org/content/early/2017/05/13/137711

Thus far we have seen no evidence for strong, specific, or direct trans interactions between mammalian interphase chromosomes: our best guess as to what is happening is a competition between loop extruders (which effectively stir together different regions on the chromosomes) and a phase-separation type phenomena between active & inactive chromatin (http://biorxiv.org/content/early/2016/12/15/094185). This can nevertheless lead to interesting impacts on the global arrangement of chromosomes, and clustering of various classes of elements in trans.

In terms of re-arrangements, there have been some very interesting papers showing how disruption of particular chromatin boundaries can lead to developmental disease & cancer (e.g. http://www.cell.com/cell/abstract/S0092-8674(15)00377-3).

The field is moving very quickly, but some more references are in these slides I posted a few months ago: https://figshare.com/articles/03-23-17_les_houches_forWeb_gfudenberg_pdf/4871948

Enjoy!
Best,
Geoff

Ken Weiss said...

Good examples, one of many, of complex cis interactions that seem well-understood by normal methods based on sequence-binding are RedGreen color vision, homeobox gene cluster usage, globin locus control regions, and various aspects of MHC usage.
Somewhat less obvious are trans coordination of immune system rearrangements such that one homologue is silenced.

But perhaps illustrating the idea that whatever is going on it isn't totally obvious is the single gene expression of olfactory receptors (ORs), in mammals and presumably other taxa. There, many clusters exist on different chromosomes. In any given olfactory receptor cell, one gene, from one cluster, from only one of the homologues is expressed. No cis sequence properties have as yet been found, to my knowledge, to account for this (though some were hypothesized). Further, there has been at least some evidence that in developing 'stripes' of nasal tissue, different OR clusters are activated, in a serial way.

To me, these facts cannot be explained by cis-regulation alone. Whether some chromosomal master centers somehow send out context-specific on/off instructions, or some other mechanism yet unsuspected is involved, I obviously have no idea. But I do believe we need to take seriously the possibility that our current types of explanation are missing something important, or even fundamental, that has not yet been suspected.