The Death and Life of Great American Neuroscience Department

Unless your university's name ends in "rd," your neuroscience department will inevitably go through boom and bust cycles. Washington University recently went through one, losing Rachel Wong, Josh Sanes, and Jeff Lichtman. Similarly, Cold Spring Harbor lost Zach Mainen, Holly Cline, Roberto Malinow, and Karel Svoboda. Those institutions are now trying to rebuild with young faculty like Daniel Kerschensteiner, Dinu Florin Albeanu, and Stephen Shea

When I started grad school at Duke, it was ranked in the top 10 neuroscience departments (whatever that's worth). Seven years later, it has dropped to ~ #20, and lost five senior faculty (with rumors of more).  And the department is now starting the rebuilding process by hiring a new chair, Stephen Lisberger.

Having lived through the death spiral of a department, I figured I'd write a post-mortem of what went wrong.  This will be a from the benches report, as I don't know what was said among faculty.

1. Insufficient leadership

Around the time I started grad school, the department appointed a new chair.  Rumor has it that the choice was between Larry Katz, and Miguel Nicolelis, but for personal reasons neither of them could be appointed chair.  So a third person was picked.

I've always been baffled by the business world's obsession with leadership, what makes a leader, how to identify leaders, etc.  I thought leadership was just another important aspect of business, like management, or logistics. Now, having experienced a leadership void, I can understand. As far as I could tell, the department had no direction, no focus of research, or any way to distinguish itself from other departments. I rarely received any e-mails from the chair, or any other senior scientist.* People just did their work, in whatever direction it happened to go.  Which leads to the next point.

2. Lack of community

Duke Neurobiology, rather than being a cohesive department, was just a collection of labs in the same building.  There were few collaborations between labs. People in the lab next door would publish exciting, high impact papers, and no one would know.  There was a happy hour on Friday's with intermittent attendance. The lunch room was in the corner of the building where only a few labs went to it.

I realize that community is an intangible thing, and won't make up for deficiencies in intelligence or dedication.  And that other neuroscience departments may not have communities either.  But after reading Peopleware, and reading about teams of people working towards a common goal, I feel something was missing.  Simple things like centralized lunch rooms, or quarterly e-mails about papers from the department would have gone some way towards making people part of a group rather than data machines stuck to a desk.

3. No forward momentum

Science is a Red Queen race. In order to maintain your standing in the wider community, you need to continually improve: incorporating new techniques, trying new systems, asking new questions. For a department, this usually takes the form of hiring new faculty. Duke tried to do that, and held faculty searches four years in a row, looking for both junior and senior faculty.  Which resulted in one junior hire.

The inevitable result of this was that once the department stopped growing, it immediately started shrinking. Senior faculty were continually getting propositioned by other departments until one offer was good enough. And each faculty that left made it easier for the next to leave.

4. No money

Duke, as a "young" university, has a small endowment compared to other institutions. There are relatively few sponsored professorships for senior faculty. Other departments that go through boom/bust cycles are in similar positions, where they can hire young faculty, but can't match offers to retain them.  I have some Moneyball-style ideas for how a department can compete by hiring under-appreciated scientists, but that's for another blog post.


In the press release announcing Lisberger as the new chair, Lisberger said all the right things:

“I look forward to bringing excellent young scientists to the Department of Neurobiology. I hope that graduate training in Neurobiology can become a focus of the institution and will strive to help the neuroscience community achieve a level of interaction that makes the whole much greater than the sum of the parts."

Hopefully he can turn things around, and improve the brand of my degree.  It can only go up from here.

* Rumor also has it that the chair asked songbird and primate researchers whether they could switch model systems to mice.

A Walk Along the Paper Trail: Bitter Trail Mix

While my previous walks along the paper trail have focused mainly on olfaction, my ultimate goal is to study taste perception.  Since the last few walks have been about olfactory receptors, I figure it would be natural to transition to taste receptors.

Unlike olfaction, where it is difficult to correlate percepts like "floral" or "rotten" with individual receptors, the correlation is much easier for taste perception.  Taste receptors have been identified most of the major taste percepts: sweet, bitter, salty, and umami (there are some additional receptors for properties like carbonation and water, but those are considered secondary). The receptors for sweet, bitter, and umami are all G-protein coupled receptors (GPCRs), while sodium is detected by ENaCs.  The one percept that remains somewhat elusive is sourness, where a TRP channel PKD2L1 and a carbonic anhydrase CAR4 which are expressed in sour sensing taste cells, but the full mechanism remains unclear.

Given that background, today I'm going to (read and) review the first paper to report taste receptor activation.

Cyclohexi...blegh


The bitter receptors (T2R) were reported almost simultaneously by the Buck and Zuker labs in March and April of 2000 (I bet there's an interesting story of that race somewhere). Besides listing a bunch of receptors, these papers also found that bitter receptors are all coexpressed in the same cells, and that there is no receptor pattern on the tongue.  The Zuker lab also published an accompanying paper by Chandrashekar et. al. where they showed that individual receptors respond to different bitter compounds.

For Chandrashekar and colleagues to study the T2Rs, they needed to establish an in vitro system, but given that they just discovered the receptors, they didn't know the downstream signaling proteins (at least until the end of the paper).  To get around this they used a G-protein which binds to a broad range of GPCRs, Ga15.  They then expressed individual T2Rs with Ga15 in HEK-293T cells and measured the calcium response following tastant application.

Using this system, they screened a random set of human, rat, and mouse T2Rs against a set of 55 tastants. From this entire screen they reported three responsive T2R-tastant pairs. (They surprisingly don't report this entire response profile. And it's not like supplemental info didn't exist in 2000.)  The pair they focused on is mouse T2R-5, which responds selectively to cycloheximine (panel b, below).

mT2R-5 responds to cyclohexamine. From Chandrashekar et. al. 2000.

They "characterized" the response as having a fast rise time of <1s, which desensitizes within 10s (panel a, above).  They also noted that repeated application of cycloheximine caused smaller Ca²⁺ rises, showing there was desensitization.

Thirty years ago, researchers found that different strains of mice have different bitter sensitivities.  For example, C57 mice don't avoid cyclohexamine while DBA/2J mice do.  Chandrashekar sequenced the mT2R-5s for multiple strains of mice, categorizing the receptors as "tasters" (from mice which could taste cycloheximine) and "non-tasters." The non-taster mT2R-5s had three mutations, including a missense mutation, and when they transfected HEK cells with the non-taster mT2R-5, they found it had a lower sensitivity to cycloheximine (panel d).

In the final section of the paper they fingered gustducin as the G-protein that couples to T2R.  To do this they used a cell-free system, using membranes with mT2R-5, and measured how much gustducin was bound to the membrane.  They found gustducin selectively bound to mT2R-5 in the presence of cycloheximine (top panel), and the K_d of the system was similar to what they found in HEK cells (compare panel b below to panel d above).

Gustducin (GTPyS) binds to mT2R-5 in presence of  cycloheximine. From Chandrashekar et. al. 2000.

And that's the paper.  It's still kinda shocking to me that it took until the year 2000 for just a subset of the taste receptors to be identified. They're right there, on the tongue!  I also appreciated the cleverness of using Ga15 to allow them to record tastant responses (although this trick may be standard in molecular biology).

One problem I have with this paper is that it does not seem very... thorough.  For example, they did not report all their tastant-odorant responses, and from skimming Zuker's later papers they did not report them there either.  Or when they characterized the response, they didn't report a tau for the on or off phases. I realize these numbers may not be completely accurate, but I still want to know them.

My legs are a bit tired from this trek, so I'll stop here.  There's a lot more to know about bitter receptors, for example many people have looked at how different polymorphisms in mammals can effect perception. But that is for another time.

CHANDRASHEKAR, J., Ken L Mueller1, Mark A Hoon2, Elliot Adler2, Luxin Feng3, Wei Guo1, Charles S Zuker1, §, *, and Nicholas J.P Ryba2, § (2000). T2Rs Function as Bitter Taste Receptors Cell, 100 (6), 703-711 DOI: 10.1016/S0092-8674(00)80706-0

Blocking Exocytosis Decreases and Delays Structural Plasticity

One of my biggest beefs with "Science"* is the journal publishing culture, where incomplete stories get buried, and cool little findings have no place to be heard or read. However, as Ghandiji says, "Be the change you want to see in the world," so I'm going to present some unpublished data.


* Not the journal


Exocytosis of AMPA receptors is important in LTP


In a previous life, aka grad school, I studed AMPA receptor (AMPAR) trafficking in Ryohei Yasuda's lab. AMPAR trafficking is essential for the most common model for studying learning and memory, LTP (I won't discuss LTP here, but my favourite review was written by Derkach and Soderling).  AMPAR are ligand-gated ionotropic receptors the reside in most excitatory synapses in the brain and mediate a majority of synaptic activity.  They can move around the plasma membrane by diffusion, and are trafficked between endosomes and the plasma membrane by exocytosis and endocytosis.


Given that AMPAR are responsible for LTP, and that they move via diffusion and membrane trafficking, one might wonder which type of movement is necessary during LTP.  A series of papers from the Ehlers lab (defunct) has argued that during LTP, AMPARs are exocytsed, and that this exocytosis is necessary to quickly deliver a large number of receptors to the synapse. Last year they even identified the SNARE responsible for post-synaptic, activity dependent exocytosis: syntaxin-4 (this will be important later).

During my thesis work I used glutamate uncaging to ask where AMPAR are exocytosed during synaptic plasticity, and which signaling pathways led to exocytosis.  In a glutamate uncaging protocol, we use spine volume as a proxy for synaptic strength (there's a long literature linking the two).  If you stimulate a spine with glutamate, it gets bigger quickly, and then relaxes a bit (see Fig. 1 below).  Using this protocol we found that AMPAR are exocytosed in the stimulated spine and surrounding dendrite, and that the Ras-ERK pathway led to exocytosis.

TeTX delays and decreases structural plasticity following glutamate uncaging


You might think that in a paper about AMPAR exocytosis and structural plasticity there'd be a figure about structural plasticity when exocytosis is blocked.  And you'd be partially right: we blocked exocytosis and measured structural plasticity for ~5 min, but not long enough to look at anything like LTP.  It was only after we published that we actually did the full thirty minute experiment.

Under normal conditions, when you uncage gluatamte on a spine, it grows to +300-500% of its initial size ~ 1-2 minutes following uncaging.  Then over the next thirty minutes, the spine size plateaus to a size ~100-200% of its initial size (top left panel, black line).  To block exocytosis, we transfected cells with tetanus toxin (TeTX), which interferes with VAMP-mediated exocytosis.  When I uncaged on cells transfected with TeTX, both the initial size increase, and the late phase structural plasticity were decreased (top left panel, blue line).

Fig. 1: Blocking exocytosis decreases and delays structural plasticity. Methods can be found at Figshare.

When I was doing the experiments, I noticed that the TeTX-expressing cells grew slower than the control cells.  I tried a few ways to quantify that.  First, I normalized both Ctl and TeTX responses to their peak (top right panel), and you can see that on average the TeTX peak structural plasticity is slightly delayed.

To see how fast individual spines grew, I normalized each responses, and plotted the growth during the first five minutes following uncaging (bottom left panels).  You can see that a large majority of the control spines reach their peak size quickly, within 1-2 minutes.  In contrast, only a few TeTX spines reach their peak size within that window, and there are many slow spines that take over two minutes to peak.  The scatterplot of the individual peak times is shown on the bottom right.

How does one interpret this?  The decrease in amplitude is fairly easy: exocytosis is important for structural plasticity, and without it there is less growth.  The delayed growth is a little more complicated.  One theory from the Ehlers lab is that when an endosomes fuses with the membrane during exocytosis, it provides membrane in addition to the more traditional cargoes of receptors and membrane proteins.  Thus when we knock out exocytosis, we have decreased the supply of membrane to the spine, and it takes longer for the spine to grow.  It still can, but it takes more time to suck in membrane from the dendritic shaft.

Syntaxin-4 mediates AMPAR exocytosis during structural plasticity

As mentioned before, the Ehlers lab identified syntaxin-4 as the SNARE involved in activity-dependent exocytosis.  When they were revising their paper, they asked us to do some uncaging experiments, but they published before we could finish.  Anyway, we simply did the same experiment, and uncaged on neurons transfected with syntaxin-4 shRNA. And we found exactly what you'd expect: syntaxin-4 shRNA decreased both peak and late-phase structural plasticity, as well as delayed the peak structural plasticity (right panel).

Fig. 2: Syntaxin decreases spine density as well as structural plasticity. Brief methods at Figshare.

When I was doing the experiments, I noticed that the spine morphology was bizarre.  There was a huge decrease in spine number, but the remaining spines were quite large.  We did not quantify spine size, but we did look at spine number, and found that syntaxin shRNA indeed decreased spine density (left panel).


That's my little bit of unpublished data.  It's a pretty clean set of results, but not quite enough to publish anywhere.  Hopefully someone finds it useful.

Peopleware, Scienceware

I've been doing a lot of programming at work lately.  As I've said before, neuroscience is increasingly becoming a form of computer science, so I decided to do a little computer science reading.  To find a book, I looked to two of the best comp. sci. bloggers out there, Coding Horror and Joel on Software.  Both recommended Peopleware, which is about managing software development teams, so I read it.

While neuroscience research and software development are not directly analogous, there are some similarities.  Both disciplines involve small groups of people working on limited length projects.  They both involve applying relatively standard techniques to create something new.  They require creativity and thought.  In essence, they're both knowledge work.

While I don't manage a lab yet, I hope to someday, so I took some notes.  These are the points that stood out for me at this point in my career:

1. "The major problems of our work are not technological, but sociological in nature"

While neuroscientists try discover how the brain works, the techniques we use are fairly standard: Westerns, imaging, molecular biology, patching, etc. Certainly these techniques can be skill intensive, and require intelligence to use, but they are uncomplicated enough that hundreds of labs around the world use them.  Only a handful of labs are developing truly new techniques like STORM/PALM imaging, or FLIM imaging.  Consider the biggest technical innovation of the past ten years, Channelrhodopsin, was just smartly applied molecular biology.

Yet project failures and delays happen all the time at least in the labs I've been around.  Part of this is because, unlike software development, there are no clear goals in science (although I'm sure some software developers with waffling clients would disagree).  Yet even without clear goals, most projections still usually have a direction, some purpose.

A bigger issue is the people, and their morale.  When an experiment runs into problems, it is usually solvable: if your PCR doesn't work, you tweak your Mg²⁺ concentration, or elongation times, but eventually it works. How quickly it gets solved is determined by morale. You need to go back to work the next day, and try something different.  All too often, when an experiment fails, your boss will ask what went wrong rather than making sure you're eager to fix it. (And once a PI stops doing experiments, their technical ideas for how to fix things become quickly outdated.)

This is partly due to how we train and promote PIs: most PIs are there because of their technical acumen, not their people skills. I've only met two PIs that understand happy workers are productive workers on a deep level.  The most successful labs will realize that projects usually run into problems when their people do, and not due to technical difficulties.

2. "There ain't no such thing as overtime"

Scientists like to think that they work hard, long hours, but with a few exceptions, I haven't seen it.  Sure, people will be in the lab area 10-12 hours a day, and come in a bit on weekends, but for almost every overtime hour they work, they also work "undertime."  Sometimes this undertime comes in the natural flow of work, whether waiting for gels to run, or stacks to be acquired. But all too often, I've seen people dicking around when they're not waiting for something, then staying late to make up for it (myself included).

While I don't doubt scientists' dedication to their job, we all have lives outside of work, and we need to drop the myth of being super productive.  Vanishingly few people can work 60-80 hour weeks for months on end, and trying to hold people to that standard is only going to drive them off.  One thing the scientific community lacks is a sort of "middle class" of people who do good work, but don't have the ambition to become a PI.  And this ridiculous standard of working long hours is part of the reason.

3. "You never get anything done around here between 9 and 5"

Knowledge work, unlike say construction or retail work, requires uninterrupted thought.  It takes a while to get into the flow of working, to remember where you are in an analysis, and to solve things quickly.  And to get into the flow you need the right environment, with no interruptions.

Despite that few labs are really set up to let people think in peace.  Some labs at Duke were set up smartly, with desks next to the windows in groups of two.  Others were set up almost optimally for disruption, with all the desks crammed into an office, where one conversation could interrupt everyone.  New buildings are often laid out in an open plan to "facilitate interactions."  I can only imagine how little work gets done there, and how often the people stay home to get some work done.

...

The rest of the book covers topics like hiring, and team building.  I don't have much management experience, so I can only consider the problem from the perspective of the managed.  I'm certain the issues look different from above.  The main lesson of the book, though, is pretty simple: hire smart people, keep them happy, and remove obstacles from their progress.

A Walk Along the Paper Trail: The Carlson Lodge, Part II

As I'm new to chemosensation, I have been reading papers from prominent labs in the field.  In the previous post, I briefly reviewed two papers from John Carlson's lab, which studies olfaction in flies. Those papers established an "empty neuron" system that allows one to express arbitrary odorant receptors (Ors) and record from their olfactory receptor neurons (ORNs); and described the receptive field of a subset of larval Ors. In this post, I'm going to review two more papers that flow from that start.

The Complete Larval ORN Receptive Field

As I just mentioned, the Carlson lab used the empty neuron system to look at how some Ors responded to odorants. To follow this effort, Kreher expanded the odorant receptor-odorant search to 25 identified Ors, and a wide variety of odorants. Of the 25 putative Ors, 21 responded to one of the odorants.  At 10^-2 dosage, 20% of the Or-odor pairs were responsive, but at 10^-4 dosage, only 4% were responsive.  They plotted this in terms of "tuning curves," and showed that Ors are more narrowly tuned at lower concentrations than high concentrations.  They further looked at the representational distance between odors by composing 21-dimensional odor response matrices, and looked at the distance between odors in spikes/s.  Odors with vaguely similar structures were closer than dissmilar odors (like aliphatics versus aromatics).  They reuse this odor distance a little bit later.

To characterize the behavioural response to the odorants, they placed larva in a petri dish with the odorant on one side of the petri dish.  The 26 odorants elicited a range of attraction and repulsion, generally skewed towards attraction.  While this gets at behaviour, this does not give a sense for perceptual similarity.  To look at that, they used a masking paradigm, where they placed an odorant A on one side of the dish, and put a masking odorant B throughout the petri dish.  Here they found that certain odors were able to mask each other, indirectly showing perceptual similarity between them.

Larva are attracted to 2-Heptanone unless it is masked by e.g. 1-Hexanol; masking implies perceptual similarity. From Kreher et al, 2008.

Then, given that they had the odor representation distance, and the perceptual distance, they plotted them against each other and saw a correlation.  In other words, odors that we perceive as similar elicit similar responses even at the level of  odorant receptors (not upstream).

Odorants which mask each other are closer to each other in odor representation space.

From Kreher et al, 2008.

Finally, they noticed that an odorant, ethyl acetate (EA), elicited behaviour over a wide range of odor dilutions (panel A, below).  When they checked their Or response matrix, they found that only two receptors, Or42a, and Or42b, responded to EA.  They then tested the two Ors' concentration dependence, and found that Or42b had a lower EC50 than Or42a (panel B below). To see whether these two Ors combined to code EA for the larva, they knocked out each receptor individually, and performed the behavioural assay.  Or42b KO larva did not respond to high concentrations, while Or42a KO larva did not respond to low concentrations.  The larva's ability to detect EA was entirely determined by these two Ors.

Or42a/b combine to code for ethyl acetate. From Keher et al, 2008.

Phew! What a paper.  The first really cool finding here is that odors that are perceived as similar are also represented in the brain as similar, at least on the level of odorants.  This may seem obvious, but it is not necessarily so. For example, think of the perceptual class of "floral."  Floral odors probably share some components, but the overlap is not complete.  Before this paper, it was possible that the differences between floral odors was large enough that they might activate different sets of receptors, and that the "floral" categorization occurred as a higher order process.  Instead, ignoring what happens in higher order neurons (which are pretty obviously important given the noisiness of the correlation), there already is an overlap in represenation and perception at the level of the odorant receptor.

The other interesting thing here is how ethyl acetate's representation seems to be dictated by just two receptors.  We normally thing of compounds activating ensembles of receptors, and our brain decoding these ensembles into percepts.  Here, though, the ensemble is one neuron.  It would be interesting to see if mammals, with more odorant receptors, similarly code for some compounds on single receptors, or if the coding is tangibly different.

The coding of similar odors


And finally we get to the paper that led me down the trail, Montague et. al.'s study of pyrazine representation.  Whereas Kreher looked at a wide set of odorants, Montague focused on a small subset, the pyrazines, which are physiochemically similar (they used Noam Sobel's idea of physiochemical space, which is what attracted me here).  And they found that while pyrazines are clustered in physiochemical space, their representation is somewhat more different in ORN space (pyrazines were closer together in spikes/s than dissimilar odors, but more spread than you might expect from the physiochemical clustering).

There were a few cool points, though.  First, in Kreher 2008, they found odor responses in 21 of the 25 Ors.  In Montague, they found that one of the previously unresponsive Ors, Or33b, responded to pyrazines.  Rather than being non-responsive, it was simply highly tuned.

Second, they found that two Ors, Or33b and O59a, responded strongly to pyrazine. So they looked at the representation of two pyrazines (2-ethylpyrazine and 2-methylpyrazine) that elicited different behavioural responses, but similar ORN responses, and focused on them.  They found that over a range of odors, the behavioural response was indeed different:

2-ethyl- and -methylpyrazine elicit different behaviour responses.
From Montague et. al., 2011.

Then they reexamined the ORN response over a wide range of dilutions, and found they were indeed quite similar (see below).  They analyzed the temporal responses as well, and found that the temporal profiles were also similar.  So these two pyrazines, which elicited different behaviours, seemed to be encoded by the ORNs almost identically.  Of course, the small differences they did observe could be amplified by higher order neurons to distinguish the two odorants.

2-ethyl- and -methylpyrazine (green and yellow) elicit similar ORN responses over a range of dilutions.
From Montague et. al. 2011.

The last cool thing of the paper is that they reported "super sustained" responses from ORNs.  Most ORN responses last for a few seconds at most, but these super sustained responses could last for minutes.  They didn't have a really good explanation for this.  In lab meeting, one guy mentioned that it could be due to differences in the second messenger cascade given this is an empty neuron system.

Pyrazines can elicit "super sustained" responses in ORNs that last over minutes.
From Montague et. al. 2011.

While they could not mechanistically explain why super-sustained responses happen (that's probably the next paper), they had some cute theories of what they could be useful for.  They could act as a sort of short-term memory, so that a fly could remember encountering a pyrazine within the last minute.  Or they could also act as a consistent inhibitory input to neighboring glomeruli, and alter processing that way.

That's the end of this hike along the paper trail.  Having read this series of papers (and skimming a few others), I have a much greater appreciation and understanding of how ORNs may respond to odors, how narrowly tuned they are, and how the perceptual code starts to form at the level of the receptors.  The next step will be to see how this is processed in the glomeruli, and in mammals.

References


Kreher S a, Mathew D, Kim J, Carlson JR. Translation of sensory input into behavioral output via an olfactory system. Neuron. 2008;59(1):110-24. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2496968&tool=pmcentrez&rendertype=abstract [Accessed August 9, 2010].

Montague S a, Mathew D, Carlson JR. Similar Odorants Elicit Different Behavioral and Physiological Responses, Some Supersustained. Journal of Neuroscience. 2011;31(21):7891-7899. Available at: http://www.jneurosci.org/cgi/doi/10.1523/JNEUROSCI.6254-10.2011 [Accessed May 26, 2011].

A Walk Along the Paper Trail: The Carlson Lodge, Part I

One of the big open questions in olfaction research is how odorant receptors bind to and respond to different odorants.  In mammals, this question is complicated by the sheer number of olfactory receptors expressed (~400 in humans), but the issue is much simpler in Drosophila, which express only 60 olfactory receptors.  One of the leading labs studying the relationship between olfactory receptor genes, receptors, and behaviour is the Carlson lab at Yale (as a tyro to olfaction, I didn't know John Carlson's name, but damn that's a good publication record).  They recently published a paper looking at how a small set of odorants, pyrazines, are represented in the olfactory receptor neurons of drosophila larva. Since I don't know much about odorant receptors or fly olfaction, I decided to take a walk down the paper trail, and see what I could see.

Olfactory receptors, their neurons, and empty neurons

The first paper relevant to our trip is from 2003, and describes the expression of  two odorant receptors in adult flies, Or22a and Or22b.  Using in situ hybridization and a Gal4-UAS-GFP system, Dobritsa found that both of these receptors were expressed in a subset of basiconic sensilla (not that I know what a basiconic sensilla is), and eventually narrowed their expression down to a neuron labeled "ab3A."  They presented a number of odors, and confirmed that the Or22a/b expressing neurons had the same odor response profile as ab3A neurons.

While Or22a is expressed in ab3A neurons, it does not necessarily mean that ab3A's odorant response is solely determined by Or22a or Or22b.  To investigate that, they knocked out Or22a and/or b, and found that only Or22a was necessary for olfactory responses.  Then to see whether a single Or is sufficient to drive ORN responses, they expressed another olfactory receptor Or47a under the Or22a-Gal4 driver.  When they did this, they found that the ab3A neuron now had the response profile of Or47a neurons (identified as ab5B). 

Or22a neurons expressing Or47a (UAS-47a) have the same response profile as ab5B neurons, different from the ab3A neurons it typically shares a profile with. From Dobritsa et al, 2003.

The Carlson lab continued to use this expression system in the next three papers, dubbing it the "empty neuron" system. In the final section of the paper, they showed that the Or22a-driven UAS-47a neurons still mapped to the same glomerulus, DM2, as Or22a neurons.

Sitting here in 2011, it is easy to take many of the conclusions of this paper for granted, and I don't know the history of odorant receptor or drosophila olfaction research well enough to put this in historical context.  Of course ORN responses are determined by the Or they express.  There are some cute findings here.  The Or expression was sex-dependent.  And the fact that both Or22a/b are expressed in the same neuron, but only Or22a is important is curious.

The most intriguing part, to my naive view, is that last bit, where the Or22a-Gal4/UAS-Or47a neurons still map to the same place as Or22a neurons.  They mention a couple explanations: there could be other guiding odorant receptors; mammals continually regerenate ORNs, while flies don't, so the map could be established before Or22a expression; the mammalian system has more receptors, and is more complex.  This is something I should follow up...

Larva ORs differ from adult ORs

While the previous paper focused on adult flies, this next paper focused on drosophila larva.  Kreher et al performed RT-PCR on larval RNA, and identified 23 Or genes that were expressed.  Thirteen of them were the same genes as expressed in adults, while ten of them had not been detected in adult flies. Then using Or-Gal4 and UAS-GFP, they showed that each Or was expressed in one neuron.

To characterize the odorant receptive field of these newly identified Ors, they used the empty-neuron system developed by Dobritsa, and expressed the Ors in adult flies.  They tested twelve of these Ors, and found that 11 of them responded to odors (pardon the synecdoche of saying Ors respond, it's the ORNs expressing them that respond).  Some of the Ors responded to one odorant, while others responded to a variety.

Some Ors respond to multiple odorants, while others are narrowly tuned (within this set of stimuli).
From Kreher et al, 2005.

They also characterized the temporal dynamics of these responses, and showed that they can be inhibitory, delayed, transient, or sustained.  They also classified the Ors into aliphatic and aromatic groups, and claimed there was a glomerular map, but that seemed fishy to me.

Some responses are inhibitory (top left), transient (top right), or delayed (bottom right).
From Kreher et al, 2005.

The main point of this paper, identifying unique Ors expressed in larva, is fairly straightforward.  What's interesting to me, as a wannabe systems neuroscientist, are the temporal dynamics of the ORN responses.  They had previously characterized similar responses in adults, but the idea of an inhibitory response, is new to me. In the discussion, they guess that the OR is tonically moderately active, and normally is stabilized in an active state by the odor (reviewed here).  Then the inhibitory odors are able to suppress firing be turning off the active state.  Alternatively, the OR might signal to multiple G-proteins, depending on its conformation (they did not consider this possibility, I wonder why).  Since this is five years old, I bet someone has figured this out by now.

And that's where I'm going to leave it for now.  Next post I'm going to cover Keher's 2008 paper (which is pretty awesome), and the paper that made me start this journey, Montague 2011.

Nature vs Science

The goal of any project (after, of course, doing sound research), is publication in a top journal like Nature or Science.  Over the years I've developed some biases about those two journals, like that Science publishes more speculative articles, and Nature has a crush on birdsong.

Of course, developing biases without testing them is bad science, so I decided to go through one year of neuroscience articles (from June 2010 to now; labeled as "neuroscience" by the journal), and see what trends there were.  I made a spreadsheet containing each article, the date published, and a general categorization of the article (these categorizations are rough, especially for some "transdisciplinary" papers, and for papers outside my expertise, like developmental neuroscience).  So what are the findings?

Nature publishes more neuroscience articles. Over the last year, Science published 52 articles tagged "neuroscience," of which twelve were cognitive neuroscience articles.  In comparison, Nature published 73 articles tagged "neuroscience," of which only three I categorized as cognitive.  Without getting into a discussion about the semantics of cognitive science, neuroscience, and psychology, if you work in a non-human system, you may want to try Nature first.  (I am, of course, ignoring the huge issue of how many papers each journal publishes, total across fields, which may also explain this.)

Science publishes more speculative/non-traditional/hard to categorize articles.  For example, "Human Tears Contain a Chemosignal," or "Astrocytes Control Breathing Through pH-Dependent Release of ATP."  In general, I had a much harder time figuring out what to label Science articles.  While I don't doubt the veracity of these articles, if you are truly pushing the envelopes in terms of interdisciplinary work, Science may be a better target.

Regarding subject area, Science skews cognitive, while Nature skews towards systems and translational neuroscience.  As mentioned above, Science published twelve cognitive science articles compared to Nature's three.  In terms of translational neuroscience (ignoring things like addiction models), Science published three translational articles versus Nature's twelve.  And in terms of systems neuroscience, over 1/3rd of all Nature's neuroscience articles were in systems Neuroscience versus 20% of Science.  Of Nature's systems articles, there was a slight bias towards vision (twelve articles).

Finally, what about birdsong in Nature?  In the past year, there's only been one birdsong paper, from the Fee lab.  Since January 2006, there have been eleven birdsong papers in Nature, about two per year.  Of those eleven, though, six came out between December 2007 and December 2008 when I got the impression that Nature birdsong.  So while I was right to think an awful lot of birdsong articles were getting into nature, it was just a coincidence.  And looking through this list reminded me of a cool paper, where they looking at the temporal coding of birdsong by cooling the brain down with a Peltier.

As a systems neuroscientist working in olfaction and taste, the conclusions seem pretty clear.  Try Nature first, unless I've got a good cognitive hook to the data.