Monday, November 28, 2016

Nostalgic for Fly's Eyes

Kumar Thangdu's mission of prodding me has contributed to a bout of nostalgia for one of my graduate student rotation projects.  He asked in a tweet how I'd allocate funds if I was given stewardship of a billions in grant money.  If you want to get some big results but are willing to be very patient, then a great way way to invest is in model organisms.  As a rotation student at Harvard, I spent several months pushing Drosophila melanogaster, the not-so-humble fruit fly.

I had planned to leave my experimental career, and experimental frustrations, behind in Newark.  However, the graduate program I was in at Harvard had minted only one prior computational-only Ph.D. (Jim Ostell, the force behind NCBI for many years).  Having someone slide through without ever doing a single wet experiment was still controversial, so the edict was laid down that I needed to go through one experimental rotation.

Luckily, I had been introduced to Bill Gelbart, one of a cohort of Drosophila geneticists in the department (his premature death last year was a terrible surprise).  Bill was part of a multi-site group trying to land a grant to digitize all Drosophila genetic knowledge into FlyBase, with the relational database itself falling into Harvard's sector.  Bill knew my crazy informatics leaning, and asked me to review the schema.  I had to confess I had never looked at such a document nor did I understand relational technology, but he said not to worry about that and gave me the document anyways.  I can't remember if I found anything intelligent to say about the schema, but it opened a line of communication.

In undergraduate genetics we had some fly labs, and to be honest I hated them.  The problem was my miserable eyesight.  If I looked through the compound scopes with my glasses on, then I had almost no field-of-view to work with.  I could take my glasses off and still focus the scope, but any notes I took would be utterly unreadable.  My attempt at a solution was to sort the flies into piles with my glasses off, then count the piles with my glasses on.  Would have worked, except the Fly-Nap would wear off during the counting phase and my piles would start merging under their own power.

But Harvard employed a system in which carbon dioxide was continually bled into the pad under the compound scope.  With that system, you can leave flies knocked out for literally hours, though some geneticists worry about ill effects to the flies from extreme exposure.  I didn't ever take that long, but my sort-then-count scheme worked well.

I found a lot of appeal in the fly room.  Because Drosophila had been worked on for nearly a century at that point, there were all sorts of nifty genetic tricks known. A lot of fly work requires less than your full brain, so you can spend the rest thinking over experiment designs or discussing science with your colleagues who are also sorting flies.  It also helped that all of the scientists and techs in the Fly Room were very welcoming.  So I really did seriously consider staying in Bill's lab.

One person who was very happy I didn't was my wife.  She visited me exactly once in the Fly Room; the stench of fermenting bananas is something I can block out but it utterly revolted her.  She also hated when I was collecting virgin flies, as I was going in three times a day for about two weeks straight on a very strict schedule; she complained that flies were running our lives.

Bill gave me an interesting rotation project.  He had three or four position mutants that he proposed having me try to find modifiers.  These mutations each display a different stable, reproducible pattern of coloration in their eyes; Bill was challenging me to find mutations which change those patterns.

Drosophila eyes are normally a red color.  The first mutation found by Thomas Hunt Morgan was the white mutation, which erases this color.  As an illustration of the fun twists which genetics can take, if you think that the white gene is responsible for making the responsible pigment, you'd unfortunately be wrong.  The white gene is a transporter of pigment, a member of the ginormous ABC transporter family which includes the cystic fibrosis gene and exporters of drugs (both in patients and in producers).  There are many families of transmembrane exporters, but the ABC group is probably the largest and most diverse.  White is also a sex-linked gene in Drosophila, carried on the X.  Alas, my nitpicking gene was activated today reading Siddhartha Mukherjee's excellent The Gene, which puts white on the Y.

Anyways, just because the white gene is normally on the X chromosome doesn't mean geneticists can't put it somewhere else.  A wild-type white gene missing a full complement of transcriptional signals will not be transcribed.  However, if installed on a transposon (aka "jumping gene") which hops around the genome, some of the resulting insertions will result in the white gene being successfully transcribed in the eye.  By the correct set of crosses or other tricks, the transposon and its ability to hop (the transposase) are separable, so that the insertions are fixed in place.  In most cases, the entire eye will turn red.  But in position mutants, a reproducible pattern is formed in the eye.

If I remember correctly, I was given four different such position mutants, but to my embarrassment I can remember only three.  I'll sketch them below, but here is the description.  One was a P-element insertion which had been generated and  published by Tulle Hazelrigg which resulted in only the bottom half of the eye being dark red; in young flies the top half would be white, ripening to a dull orange as the flies aged.  Another created a tiny crescent of color in the upper part of the eye.  But the one I found most striking was one which created a wedge of color pointing to a single line of colored facets crossing the eye.

Those facets, or ommatidia, are what make a compound eye compound.  Each is a subassembly of photoreceptors.  If you've seen insect vision at a zoo or museum (or on The Simpson's) depicted as each facet with a different image, that's not remotely like the truth.  Each ommatidium is closer to one pixel, with different photoreceptors within the subassembly recording different wavelengths of light.

The most famous photoreceptor type is R7, which senses ultraviolet light.  The great Seymour Benzer had previously developed a genetic mapping system for phage so precise he could map at nucleotide resolution.  Looking for a new challenge, he decided to explore the genetic basis for fruit flies positive phototaxis for ultraviolet light.  So he mutagenized flies and then had them pass through a series of Y-shaped tubes.  Only one side of each Y was illuminated with UV light, so wild-type flies would always go that way, but any flies deficient for R7 would pick randomly.  Using this approach, Benzer isolated sevenless (sev).

Sevenless became a paradigm for understanding cell fate decisions.  One approach to further elucidating the genetics of this system was to isolate various additional mutants which interacted with sevenless, with such whimsical names as bride of sevenless (boss) and son of sevenless (sos). Because these circuits turned out to be biochemically related to key circuits involved in human cancer (sev is a receptor tyrosine kinase; sos is a nucleotide exchange factor for RAS), the sevenless genetic circuit has proven valuable for contributing to our understanding of a wide range of human cancers.

So my task was a similar idea, but applied to these position mutants.  Obtain a large number of unmated ("virgin") females homozygous for the mutant in question.  Females won't respond to male's mating approaches for a time after emerging from their pupal state, so during that window one can collect them.  That was the procedure that my wife disliked due to insects controlling our schedule.  Collect a similarly large number of homozygous white males, which is easier because you really don't care if they have mated.  Mutagenize the males with gamma rays to create double-strand breaks, using a huge metal contraption that insured that nothing larger than a small bottle could ever be exposed to the gamma ray source.  Now mate the flies and start looking through the offspring for alterations in the eye patterns.

Now, what I've described is a simple screen which will pick up dominant mutations. A quarter century of not thinking about the screen has erased my memory of whether I tried to cross the progeny with each other to look for recessive mutations (I think I did).  

I do remember finding a great example of a Drosophila oddity: a gyandromorph.  In humans, anatomical males are created by a gene on the Y-chromosome activating in testes tissue and then triggering the secretion of hormones into the bloodstream.  Cells around the body see this signal, which synchronizes them to be male.  A lack of this signal, and all cells progress down a female genetic program.  

But in flies, each cell decides on its own.  Rather than the presence of a Y, the determinant is the ratio of X chromosomes to autosomes.  One X is male, two Xs are female.  In most flies, every cell will have the same chromosome complement, so all of the fly's cells express the same gender.  But, if a damaged X-chromosome (such as one zapped by a gamma ray) is present, then sometime after early mitoses in the embryo that second X can be lost.  So now a subset of cells have one chromosome (male) and others two (female).  In the case of the fly I found, one X  probably dropped in the first mitosis, as it appeared be almost precisely half-and-half.  The front of the fly showed several female-specific anatomical features, whereas the abdomen showed male-specific features.  Other planes of symmetry (or asymmetry) are possible and seen.

I looked through a lot of offspring; I think the goal was either 5,000 or 10,000 offspring per cross.  Or did I isolate 10,000 virgin females?  Again, memory is failing me.  I didn't find any stunning variants, but I do remember pulling a small number of candidates for modifiers. Then my rotation ended, and even if I had stayed it would have only been a side project at most.

Still, those position mutations are very interesting.  Assuming that each faithfully represents the actual expression pattern of at least one eye-specific gene, why these patterns?  Why does some fly gene express only in the lower (ventral) portion of the eye?  Why would some gene shows that crazy wedge plus line pattern?

But perhaps more interesting would be if I had found modifiers, as these would potentially be the genes enforcing these rules of expression.  While there was clearly leakiness over time, in the young flies the pattern was essentially binary; ommatidia were either red or white.  Obtaining such sharpness of gene control is a marvel, so dissecting it would be fascinating.

It is worth comparing these against possible genomic approaches.  Given current methodologies, it is not unreasonable to think that one could use various single cell transcriptomics approach to map out all the different expression patterns in a fly eye.  The position mutants capture these in a striking visual way, but we have no means of knowing all the possible position mutants that have never been seen.  

But finding the enforcers?  That would likely be difficult or impossible from transcriptome data.  Such tight patterns are most likely a combination of effects from multiple genes, creating a  complicated circuit ANDs, ORs and NOTs.  These interactions could be at the level of assembling transcription complexes, or activating them, or numerous other levers which eukaryotic cells possess for shaping gene expression.  Genetics approaches are a proven route for identifying such interactions, particularly valuable since it is agnostic to the biochemical mechanism of control.

I haven't kept up with Drosophila genetics other than at a very cursory level.  Perhaps these days one would screen a large RNAi or CRISPR collection to look for mutants, rather than gamma rays.  Perhaps today I would have used a chemical mutagen, particularly now that it is so easy to determine the precise mutation using inexpensive whole genome sequencing.  The technology can only get better, but the core strengths of Drosophila will always be there: fast generation time and the ability to run huge screens, so long as you can stand the smell of the Fly Room.


Mike the Mad Biologist said...

Just one minor nitpick/quibble. Jim Ostell is still at NCBI and still a force there.

Keith Robison said...

Thanks - somewhere I picked up the misimpression he had retired.