Theoretical Approaches in Nervous System Processing

a081460-1-e1429226047518 (1)Adrienne Fairhall PhD., is a Professor in the Department of Physiology and Biophysics at the University of Washington. The Fairhall lab develops theoretical approaches to understanding processing in nervous systems, from single neurons to foraging mosquitos to navigating primates. Using mathematics and statistical methods, the Fairhall lab studies the relationship between neuronal circuitry and functional algorithms of computation.

One of the challenges the Fairhall lab is undertaking is understanding adaptive coding in context sensitive neural systems. As the encoding of a stimulus depends on the temporal, spatial, and semantic context in which it is embedded, many typical features of a system are adjusted according to the signal-to-noise ratio, with this coding range typically adapting to the range of stimuli. Adding to the complexity is the fact that different processes encode information at different timescales. Fast adaptive processes can normalize the response functions to the scale of the stimulus, but there are also slower processes that depend on the history of temporal changes in stimulus statistics.

Recently, the Fairhall lab has come out with a study that advances our understanding of both of these topics. Published in PLOS computational biology this year, they studied the history dependence in insect flight decisions during odor tracking. As many important behaviors require animals to make extended sequences of decisions in response to complex stimuli, they sought to model these sequences in fruit flies and mosquitos by tracking their response to odor plumes. By examining videos and tracking the 3D trajectory of these insects flying in a wind tunnel containing an attractive odor, they were able to ask what features of the encounters with an odor plume could influence flight decisions. Interestingly, although the average response was a reflexive upwind turn towards the stimulus, they found that the strength of the response was modulated by the history of prior plume encounters.


Fig 3. History dependence of crossing-triggered turns in data and models


This history dependence was captured in a model where a simulated tracking agent maximizes information about the position of the plume source. These results suggest that real odor tracking could involve short-term memory processes that occur over multi-encounter timescales that accumulate information about the source location. They did not report, unfortunately, on how to prevent those pesky mosquitos from learning which part of your arm to land on for a bite!

To hear more about the work being done in Dr. Fairhall’s lab be sure to join us at 4 PM, Tuesday 11/27/2018 at the Marilyn G. Farquhar Seminar room in CNCB.

To read the paper, visit:

To learn more about the projects ongoing in the Fairhall lab, swing by:


Joseph Herdy is a first-year PhD student working in Dr. Saket Navlakha laboratory.








Feedback Circuits Regulate Skilled Reaching

jc2Despite the powerful computations our brains can perform and the vivid abstractions of reality it allows us to produce in our minds, our only way of affecting any change in the world around us is through our motor system: the contracting and relaxing of our muscles. One of the most impressive evolutionary accomplishments of humans and other mammals is our ability to very precisely control these movements to achieve fine motor tasks. For example, reaching out a hand, grasping a glass of water, and bringing it to one’s mouth for a drink. While this may require almost no conscious effort, it is a nontrivial feat, requiring the brain to coordinate the movement of many muscles in the hand and arm, and integrate this with tactile and visual information from other bodily systems.

One major theory of how the nervous system accomplishes this smooth and precise pattern of movement is by utilizing internal copy pathways, which was an interest of Dr. Eiman Azim’s while at Columbia University. In this model, descending motor neurons send projections to the cerebellum and provide it with the same set of motor information that is being sent to muscles. This allows the cerebellum to predict whether or not a given set of motor commands will be successful before the action is carried out, and provide immediate corrective feedback to ensure successful movement. If the brain instead were to react solely on sensory feedback provided by vision and touch, information would have to travel all the way from the extremities back to the brain, producing a significant time delay, and forcing the cerebellum to compensate using information that is no longer current.

One good candidate neuronal population for providing an internal copy the cerebellum is the cervical propriospinal neurons (PNs). These neurons have bifurcated axons that extend one branch to the lateral reticular nucleus (LRN), a pre-cerebellar relay, and another branch to the cervical motor neurons that control forelimb movement. In his study, published in Nature in 2014, Dr. Azim identified a major population of the PNs that belong to the V2a interneuron class, which are known to be highly involved in motor control.

To determine what role PNs play in skilled movement, Dr. Azim created a behavioral assay in which mice are presented with a food pellet and tasked with reaching their paw through a small window to retrieve it. Paw position is recorded throughout the test, and is divided into three distinct phases: reaching through the window, the anticipatory phase immediately before grasping, and the grasping phase. Statistics such as the paw position in 3D space, distance to the pellet, and paw velocity are recorded by two nearby cameras.

jcpic1 Figure 2: Reaching kinematics

Dr. Azim first used this assay to determine that specific ablation of PNs by diphtheria toxin resulted in mice moving their paws more slowly, and more frequently changing direction, but only when the animal was reaching out its paw to retrieve the pellet, rather than in the grasping or anticipatory phase. This established an initial role for these neurons in regulating fine motor control. To tease apart what role specifically the PN projection to the cerebellum might play, Dr. Azim optogenetically stimulated PN axons present in the LRN, so that all downstream cerebellar targets would be activated but the direct connections to the motor neurons would be unaffected. Stimulating this region during the pellet retrieval task resulted in a similar perturbation of the movement, with a large increase in the number of reversals in paw direction. This indicates that information sent by PNs through the cerebellar pathway plays a role in regulating forelimb movement, in addition to the connections made directly with motor neurons. Finally, Dr. Azim performed a complementary experiment and severed the connections between the LRN and the cerebellum, and saw that this increased the response latency of motor neurons by 1 to 3 ms as measured by motor neuron field potential response, demonstrating that the cerebellar circuit plays a role in producing rapid compensatory changes in motor output using information from PNs.

Overall, Dr. Azim’s study provided evidence for an internal copy provided by PNs to the cerebellum via the LRN, which in turn provides rapid adjustments to ensure the success of the movement. The finding that ablation of these neurons affects only the reaching phase of pellet retrieval and does not affect grasping supports an emerging idea that different sets of interneurons are responsible for specific sets of movements like reaching or grasping, and that these sets of neurons are recruited modularly in order to perform highly complex motor tasks.

To hear more about the work being done in Dr. Eiman Azim’s lab, please join us at 4:00pm, Tuesday 11/20/2018 at Marilyn G. Farquhar Seminar Room.

To read the paper, visit:

To learn more about the Azim lab, visit:

Understanding Stress Granules in Neurodegenerative Disease

YeoGene-400x400Dr. Gene Yeo, Ph.D., MBA, is an expert in the areas of RNA, genomics, computational biology, and neurodegenerative diseases. Dr. Yeo was the first Junior Fellow at the Crick-Jacobs Center for Theoretical and Computational Biology at the Salk Institute in 2005 and was soon appointed assistant professor of Cellular and Molecular Medicine at UCSD. Dr. Yeo, now a full professor, has been very successful during his time in academia. Of note, his lab was the first to demonstrate the targeting of RNA using CRISPR/Cas9 and furthermore, Dr. Yeo has received the inaugural Early Career Award by the international RNA Society. In addition, Dr. Yeo serves on the Scientific Advisory boards of several biotech companies, is a bioinformatics and business consultant for biotech and pharmaceutical companies, and has co-founded several start-ups.

A primary interest of the Yeo lab is to understand how RNA expression is regulated post-transcriptionally in relation to maintaining cellular homeostasis during development, aging, and disease. To study this, the Yeo lab employs computational and experiment techniques, including genomic data analysis, molecular biology, biochemistry, high-throughput sequencing, and imaging. In their recent study published in Cell, titled “Context-Dependent and Disease-Specific Diversity in Protein Interactions within Stress Granules”, the Yeo lab elucidates the composition and behavior of stress granules during normal and disease states.

Stress granules (SGs) are ribonucleoprotein (RNP) aggregates that transiently assemble in the cytosol upon cellular stress and have been implicated in neurodegenerative diseases as represented in figure 1 below.


Figure 1: Schematic of stress granule formation

To understand the proteome of SGs, the Yeo lab first identified known and previously unknown SG proteins using ascorbate peroxidase (APEX) proximity labeling in combination with mass spectrometry and immunofluorescence. Using these methods, the Yeo lab discovered about 150 previously unknown human SG-related proteins. They next compared the composition of SGs in different cell types and under different cellular stressors and found cell-type specific and context-dependent SG proteins. The Yeo lab then wanted to compare healthy and amyotrophic lateral sclerosis (ALS) SG composition and location in patient specific iPSC-derived motor neurons, the cell type most affected in ALS, and saw altered composition and altered distribution of ALS SGs. Furthermore, they were able to show that when altering these SG proteins, they can modify protein toxicity in Drosophila ALS disease models. A summary schematic of their findings is below in figure 2. These results show that SG homeostasis is altered in ALS and that targeting the proteins involved in SG function may be a therapeutic option for treating ALS. This implicates the importance of understanding SG formation and function in neurodegenerative diseases, especially in diseases where protein aggregation is prominent.


Figure 2: Schematic summary of findings from studying SGs

Please join us Tuesday October 30th at 4pm in the Marilyn G. Farquhar Seminar Room in CNCB to hear Dr. Gene Yeo talk more about his research!


To learn more about stress granules, here is a nice review:

To learn more about the Yeo lab, here is the lab website: 

Figure 1 

Figure 2


Sammi Sison is a first-year neuroscience graduate student, currently working in Dr. Kristin Baldwin’s lab.

Constructing the transcriptomic definition of CNS cell-types


The human central nervous system is a complex entity composed of over 170 billion cells. Deconstructing its complexity requires methods to define and characterize specific cellular populations. Michael Oldham, PhD, neuroscientist and Assistant Professor of Neurological Surgery at UCSF, believes gene expression lies at the root of cellular identity. His team dives into the transcriptome of the central nervous system to find distinct, reproducible signatures of cell types in silico.

The lab’s most recent publication 2018 in Nature Neuroscience, “Variation among intact tissue samples reveals the core transcriptional features of human CNS cell classes,” takes a top-down approach to identifying core signatures of cellular identity in an astonishing 7,221 human CNS samples!

To illustrate their rationale (Fig 1a-b), they first created “synthetic tissue samples.” These samples were composed from single-cell RNA seq data and designed to mimic the cellular heterogeneity found in human CNS samples. The data from each synthetic tissue sample was run through an unsupervised gene coexpression analysis. From this analysis, Oldham’s group (1) identified gene coexpression modules ( and (2) identified modules with the highest abundance of previously published markers of astrocytes, oligodendrocytes, microglia, or neurons (“cell-class modules”).

Fig 1

Fig. 1: Rationale and workflow.

They performed this same unsupervised gene coexpression analysis on data from 62 publicly-available human CNS gene expression data sets representing thousands of tissue samples and billions of cells. Their goal was to identify cell-class modules and use them to define a cell-type specific consensus transcriptomic profile (Fig 1c-g).

In addition to being used as part of a categorizing definition, cell-class modules were also used to retrace cellular composition of the original, intact tissue sample. This is very exciting as in the experimental process of extracting RNA from bulk postmortem CNS tissue, no information about cellular identity is retained. There is no trace to show which transcripts belonged to which cells.

The solution to this problem, the Oldham group posited, is a simple idea: “variation in cellular composition among intact tissue samples will drive covariation of transcripts that are uniquely or predominantly expressed in specific kinds of cells.” Then mathematically, the first principal component (‘module eigengen’ in Fig 1)  of a cell-class module is the relative proportion of the cell class in the sample.

Let’s break down this idea. Cell-class modules should contain transcripts that are (1) highly specific–unique to a specific cell type and (2) highly sensitive–expressed consistently throughout those cells. These genes have a high “expression fidelity.” Mathematically, this is quantified as KME (the cell-class module membership in Fig 1). As they mark cellular identity, expression patterns for these cell-class modules should be predictable and consistent for a given cell-type. Therefore, the majority of variability in the expression of cell-type specific, high fidelity genes across samples should be due to the proportion of cell-type specific cells within a given sample. A sample with 80% neurons will show increased expression of high-fidelity neuronal genes versus a sample with 40% neurons.

They tested this hypothesis using their synthetic tissue samples (as they have known cellular compositions). They found that their module produced accurate predictions of cellular abundance (Supp Fig 1). Truly an exciting finding, as they show a means to define cell-specific transcriptomic signatures WITHOUT the need for physically isolating cellular subpopulations.

And all that just in Figure 1!

Using this workflow, Oldham’s group:

  1. Validated many canonical markers as high fidelity genes for their predicted cell class (Fig 2)
  2. Examined in detail the top 50 expression fidelity genes for each major CNS cell class including expression levels/fidelity, loss-of-function intolerance, PubMed citations, cellular localization, and protein-protein interactions (See Fig 3 below)
  3. Examined less abundant (or more specific) cell classes (such as cholinergic neurons, midbrain dopaminergic neurons, endothelial, ependymal, choroid plexus, and more) and validated many canonical cell markers (Fig 4)
  4. Found differential expression of genes associated with neural disease in different cell populations (e.g., genes associated with most neurodegenerative disorders are enriched in microglia and astrocytes) (Fig 6)
  5. Identified differences in expression fidelity for cell classes in different CNS regions (regional differences greatest for neurons > microglia and astrocytes > oligodendrocytes) (Fig 7)
  6. Identified species-specific transcriptional differences in cell classes (including identifying a putative candidate gene driving astrocyte size differences in human vs. mouse CNS) (Fig 8).

Fig 3

Fig. 3a-d: The core transcriptional identities of human astrocytes, oligodendrocytes, microglia, and neurons include known and novel biomarkers.

Quite a feat!

To learn more about the work from Dr. Michael Oldham’s lab, please join us at 4:00pm, Tuesday 10/22/2018 at Marilyn G. Farquhar Seminar Room.

To read the paper:

To read more about the Oldman group:

Isabel Costantino is a first-year PhD student working in Dr. Jerold Chun’s laboratory.


Piezos: Mechanotransduction in health and disease

I don’t suppose you have given pause recently to ponder the biophysical marvel that is the red blood cell (RBC). And, well, who can blame you really? But then I am, however, confronted with the unhappy task of arguing that you should. This is because the RBC has that annoying habit of surfacing in the unlikeliest of places and then, once noticed, is weirdly found to have been hiding some impressive and seemingly outsized role. Orchestrated, as things forever seem to be, by a cytoskeletal scaffold, the biconcavity of the RBC is, structurally, perhaps its most prominent feature and, functionally, features prominently. It is the RBC’s biconcavity that enables it to maneuver through — or rather be propulsed through — the birth canal-like passages of the vasculature, and it is the RBC’s biconcavity that grants flexibility to volumetrically respond to the osmotic changes happening outside its walls. Much of this volumetric regulation, as it happens, is tuned by something called a piezo.

It is Ardem Patapoutian’s crew at Scripps Research that have been the primary movers pushing the field of piezos. Piezos (which, if you care, our good friend Webster claims is etymologically derived from the Greek piezein, “to press”) are non-selective, mechanosensitive cation channels, meaning they transform the mechanical forces of the outside sensory world into actionable biological signals. Two of them, helpfully christened Piezo1 and Piezo2, have been identified so far, but likelihood suggests more of them remain to be discovered. While Piezo1 enjoys greater expression in non-neuronal tissues (e.g., endothelial, smooth muscle, and red blood cells), Piezo2 is more prominently expressed in sensory neurons and facilitates, for example, proprioception and the detection of light touch. In addition to sensory neural structures, Piezo2 is also expressed in other sites dependent upon mechanosensation (e.g., respiratory structures).

Figure 1

Taking the latter first, Piezo2 has been identified as a structure critical for the mechanotransduction of light touch and proprioceptive information and localizes to nerve terminals, those sites where one could perhaps reasonably expect mechanotransduction to occur. Using green fluorescent Piezo2-GFP mice, fluorescence in skin was observed to concentrate in lanceolate endings and around hair follicles, in Merkel cells, and in Meissner’s corpuscles, all of which are sites and cells involved with touch sensation. Subsequent employment of an ablation-at-will methodology utilized AvCreERT2 mice, mice that are embedded, if you will, with tamoxifen-inducible Cre recombinase in sensory neurons and epidermal Merkel cells that have an Advillin promoter. Mating these AvCreERT2 mice with the Ai9 tdTomato reporter line revealed tdTomato expression in 87% of dorsal root ganglion neurons (DRGs) and 82% co-expression of Piezo2 and tdTomato positivity (Figure 1). Because it has previously been argued elsewhere that silencing Piezo2 in DRGs yields decreased mechanically activated current, Patapoutian’s crew poked cultured DRGs from Piezo2 knockout mice with glass probes to confirm that Piezo2 knockout DRGs do indeed feature fewer rapidly adapting mechanically activated currents. In fact, the DRGs do not even appear to respond more slowly but rather appear to become altogether unresponsive (Figure 2). Upon finding deficient mechanically activated current in DRGs, one might reasonably propose a concordant deficiency in skin sensory fibers, a proposal the group then assessed using an ex vivo skin nerve preparation — using samples drawn, again, from wild-type and Piezo2 knockout mice — and observed, in the knockouts, loss of

Figure 2

mechanosensitivity in 50% of the Aβ-fibers without significant loss in either Aδ-fibers or C-fibers. It is, to say the least, an unfortunate era in which to be a mouse.

The finale of at least this assessment of Piezo2 was a series of behavioral tests, conducted in our by now quite tired and savaged Piezo2 knockout mice. Application of von Frey filaments with varying force to the hind paws of these mice indicated severe deficiency at detecting forces of lower magnitude. Interestingly, however, the ability to detect forces of greater magnitude was retained, suggesting Piezo2 may function within a constricted range of mechanical stimulation. In a separate assay, the so-called cotton swab assay, a cotton swab was gingerly drawn under the mouse’s paws; apparently the Piezo2 knockouts, in contrast to wild-type, couldn’t be bothered to withdraw their paws from its softness. Anyway, to ensure these findings were not merely the byproduct of globally useless sensory reception, application to the Piezo2 knockouts of various forms of thermal stimuli and particularly irritating mechanical forces revealed no differences in response from controls. To thus finish our discussion of Piezo2 perhaps overly briefly, while the argument Patapoutian’s lab has assembled fairly robustly supports the role of Piezo2 for at least light touch mechanotransduction, the equivalency of response observed upon mechanostimulation at higher magnitude would appear to offer an interesting opportunity for one to further define the spectrum of mechanosensation.

Figure 3

To then consider the former second, we turn to Piezo1, the recent structural and functional analyses of which represent impressive steps forward in the study of mechanically activated channels. Using methodology that can perhaps only be described as the very cutting-edge of structural biology, the full structure of the Piezo1 channel was defined in whole rather recently using high-resolution single particle cryo-EM. While the pictorial depiction of Piezo1’s structure is, quite simply, a beauty, we are best served by considering Piezo1’s activation, a feat that involves finely tuned movements of an inner helix, outer helix, C-terminal domain, anchor domain, and latch and beam domains. I know, I know, but stay with me here. The anchor domain in particular is thought to factor prominently in channel gating, most likely via an electrostatic interaction between the E2133 residue and the R2482 residue of the inner helix; these things apparently are somewhat in the vicinity of one another (Figure 3). While this precision tuning may represent nature at its finest, any disruption in the force, so to speak, melts down the whole mechanism. Indeed, the R2482H mutation in human Piezo1 has not just slower inactivation but also has been associated with something called dehydrated hereditary stomatocytosis, a disease of — you didn’t think I forgot about them, did you? — RBCs.

Let us then return, you and I, to the RBC. And while up to this point we have considered piezos in health, here let us consider them, or at least Piezo1, in disease. Also known as hereditary xerocytosis, dehydrated hereditary stomatocytosis is a blood disorder in which RBCs are said to be dehydrated and feature decreased osmotic fragility — a disruption in that critical biconcavity. Many of the mutations in Piezo1, including R2482H, result in slower inactivation and thus increased passage of ions. Consequently considered gain-of-function mutations, these are the mutations that cause dehydrated RBCs. Dehydrated RBCs are in some fashion associated with diminished infectivity by Plasmodium, the agent that causes malaria. Plasmodium is additionally understood to effect a selectivity pressure on the genome. Patapoutian’s group thus assessed the relationship between Piezo1, dehydrated RBCs, and Plasmodium susceptibility first by infecting gain-of-function Piezo1 mice with a GFP-expressing line of a rodent Plasmodium species notably notorious for causing cerebral malaria. In these gain-of-function Piezo1 mutant mice, evaluation of GFP-positive RBCs by flow cytometry showed decreased parasitemia and increased survival. To evaluate for the blood-brain barrier breakdown commonplace in cerebral malaria, they injected Evans blue dye into both wild-type and Piezo1 gain-of-function mice infected with Plasmodium. Here it is the wild-type mice for whom we’re to become soppy-eyed for they all exhibited the blue dye leakage consistent with blood-brain barrier dysfunction. But the Piezo1 mutated mice, on the other hand, exhibited no dye leakage: they were protected from cerebral malaria! The question, of course, is whether any of this has any meaning for humans.

Figure 4

Echinocytes and stomatocytes are denoted by white and yellow arrowheads, respectively.

Figure 5

In pursuit of an answer, Patapoutian’s group assessed whether African populations (whose individuals are expected to more frequently be from areas with endemic malaria) have an increased frequency of gain-of-function Piezo1 mutations. Indeed, one mutation, E756del, was found to have an allelic frequency of 18% and functionally observed to have the slower inactivation time similar to R2456H, the allelic equivalent of which was used to generate the gain-of-function Piezo1 mutant mice. Further characterization suggested this allele was derived (i.e., not ancestral) and under positive selection, almost certainly a selection pressure resulting from its protective effects against Plasmodium. To confirm RBC morphological and infectivity results that had heretofore only been observed in mice, the group lastly evaluated RBC dehydration and infectivity by Plasmodium falciparum in African American E756del carriers. (It is perhaps of value here to recall that P. falciparum is the most grievous and hideous of the whole Plasmodium scourge.) At any rate, scanning electron microscopy of RBCs showed the echinocytes and stomatocytes expected of hereditary xerocytosis (Figure 4), and osmotic fragility testing showed the RBCs of E756del donors to be dehydrated. In vitro infection of RBCs from controls and E756del carriers with P. falciparum revealed decreased parasitemia for carriers (Figure 5). Although a precise mechanism for such a chain of events may yet remain to be elucidated, what we see is that gain-of-function genetic alteration of the Piezo1 mechanosensitive channel confers protection for the most at-risk populations against the most severe malarial parasite, Plasmodium falciparum, by dehydrating the RBC. Seeing as I have by now probably exhausted your attention supply, allow me to rather abruptly conclude by saying that I think some commendation is in order for these individuals — these populations — that have been able to so successfully mooch off that selfish gene.

In order of appearance:

  • Murthy, S. E. et al., Nature Reviews Molecular Cell Biology (2017).
  • Ranade, S. S. et al. Nature (2014).
  • Saotome, K. et al. Nature (2018).
  • Ma, S. et al. Cell (2018).

Jason Adams is a first-year Ph.D. student in the lab of Alysson Muotri.

Understanding Neuronal Cell-Type Diversification From the Perspective of a Worm

From the outside looking in, the brain looks rather homogenous. It has folds and creases, some protruding lobes, but really only a handful of features that make it unique upon gross inspection. Taking a closer look (depending on the species) reveals hundreds to thousands, or even one hundred billion neurons that help make up the brain. Taking an even deeper dive, unfolds a richness of cell types that gives the brain an immense amount of diversity. The mechanisms governing cell-type diversity in the brain is poorly understood, but incredibly important. Understanding the genetic programs that make neurons different may help elucidate what went wrong when neurons (and brain structures) become pathologic.

Spearheading the research that investigates the molecular mechanisms governing neuronal cell-type diversity is Oliver Hobert. Professor Hobert has appointments in biological sciences and molecular biophysics at Columbia University, and has the privilege of being a Howard Hughes Medical Institute investigator. Dr. Hobert’s lab uses Caenorhabditis elegans (C. elegans) to take a “bottom-up” approach to elucidate the genetic programs responsible for cell type diversification in the brain. The “Bottom-up” approach is an attempt to define sequences of DNA (AKA the “gene battery”) that specify anatomical and functional properties of cells, and then dissects the regulatory elements governing the transcription of neighboring genes (AKA “cis-regulatory elements”).  Ultimately, C. elegans provides a genetically tractable model with well described neuroanatomy to test hypotheses regarding the development of different neuronal cell types. Hopefully, the molecular and analytical tools used in C. elegans, can be used to investigate these developmental mechanisms in other species.

Recently, the Hobert lab published a paper in Neuron titled: “Diversification of C. elegans Motor Neuron Identity via Selective Effector Gene Repressor”, which elucidates the mechanisms governing C. elegans motor neuron diversification. In the paper, they point out that C. elegans motor neurons are cholinergic and GABAergic, but can be further subdivided. For example, the cholinergic neurons can be divided into six classes based on their features (Fig 1a). In the end, they sought to uncover the mechanisms governing cholinergic motor neurons (MNs) diversification in the ventral nerve cord (VNC) of C. elegans.

Previous studies showed that 5/6 classes of MNs in the VNC shared the unc-3 transcription factor (Fig 1b). Additional studies showed that the loss of or misexpression of unc-3 lead to the loss of specific MN subtypes. Therefore, the unc-3 transcription factor was not only shared among 5/6 MNs, but also specified MN subtypes, which is a little paradoxical. How can a shared transcription factor also specify MN subtype? The Hobert lab sought to answer this question by testing two models. They hypothesized that either the unc-3 transcription factor requires class specific co-factors to activate class-specific features (co-activator model upper panels of Fig 1d) or unc-3 is capable of activating all features, including class-specific features, but is prevented from doing so via class-specific repressor proteins (repressor model lower panels of Fig 1d). If the activator model were true, loss of unc-3 would lead to loss of class-specific features. If the repressor model were true, loss of unc-3 would lead to ectopic expression of class-specific features.

To test the two models, they screened C. elegans mutants to identify alleles in which MN class-specific effector genes are either misexpressed (supporting repressor model) or lost in specific MN classes (supporting the co-activator model). An example of how the data was collected and analyzed is shown in figure 2. Using the genetic screen, the ot721 allele was found and unc-129 (an effector gene) was ectopically expressed. Figure 2a shows the expression pattern of unc-129 with GFP in both the wild-type and mutant ot721 C. elegans. In the mutant ot721 phenotype there is green protein found in VA and VB MNs, which indicates ectopic expression. Expression patterns for the mutants are shown in fig 2b. Further analysis showed that the mutant ot721 allele corresponded to a previously undescribed zinc finger transcription factor encoding gene bnc-1 (fig 2c). When bnc-1 is expressed in the mutant ot721 C. elegans, the phenotype was rescued.

In the end, they found that MN diversification was a result of class-specific repressor proteins that prevents unc-3 from activating subsets of class-specific effector genes. Furthermore, all the reported repressors are phylogenetically conserved. Therefore, the proposed mechanism for MN diversification in C. elegans may constitute a broadly applicable principle of neuronal identity diversification across species.

To learn more about the tools used to investigate the genetic programs that lead to neuronal diversification join Dr. Hobert and the rest of UCSD neurograduate program at 4pm Tuesday (05/29/18) at the CNCB Seminar Room at UCSD.


figure 1


figure 2_finalElischa Sanders is a first-year Neuroscience Ph.D. student, currently working in Eiman Azim’s lab

Combining Optogenetics and Drug Delivery: Remote-controlled dissection of neural circuits

Michael Bruchas, PhD, is an interdisciplinary scientist at Washington University in St. Louis, with departmental affiliations including Anesthesiology, Neuroscience, Psychiatry, and Biomedical Engineering. His work aims to dissect how G-protein coupled receptor (GPCR) systems function in the contexts of stress, depression, addiction, and pain. He and his collaborators were awarded 3.8 million dollars out of the 2013 White House BRAIN Initiative ( which fostered the development of a novel wireless administration system that manipulates light and drug delivery ( The techniques being innovated in his lab ( allow for greater understanding of GPCR signaling in real time, within intact systems, and with respect to biologically relevant models of behavior.


His paper published in Cell (Jeong et. al 2015, in collaboration with John Rogers’s lab highlights the utility of this novel wireless technology. This technology has since been made available through his company, NeuroLux. An interview with Dr. Bruchas about this remote-controlled drug delivery system can be found here:


Prior to the development of this technology, the existing neural interface technologies metal cannulas connected to external drug supplies for pharmacological infusions and tethered fiber optics for optogenetics. These are not ideal for minimally invasive, untethered studies on freely behaving animals. This paper introduces wireless optofluidic neural probes that combine ultrathin, soft microfluidic drug delivery with cellular-scale inorganic light-emitting diode (μ-ILED) arrays. These probes are orders of magnitude smaller than cannulas and allow wireless, programmed spatiotemporal control of fluid delivery and photostimulation.

The figures below demonstrate the probes and implantation technique.


This is what the animals look like after implantation and recovery from this relatively minimally invasive surgery- NO WIRES!


These implants transmit signal through an antenna tuner and control box, back to any laptop with NeuroLux software.


The paper then demonstrated these devices could modify gene expression (Figure 4), deliver peptide ligands (Figure 5, reproduced below), and provide concurrent photostimulation with antagonist drug delivery (Figure 6, reproduced below) to manipulate mesoaccumbens reward-related behavior in freely moving animals. 

Figure 5. Untethered Delivery of Mu-Opioids (w/ DAMGO) into the Ventral Tegmental Area Causes Stereotypical, Repeated Rotation Behavior Picture5

Figure 6. Wireless Dopamine Receptor D1 Antagonism (w/ SCH23390) in the Nucleus Accumbens Shell (NAcSh) Blocks Photostimulation-Induced Real-Time Preference of Freely Moving Animals


Tl;dr: Michael Bruchas’s group developed neural probes with ultrathin, soft microfluidic channels coupled to μ-ILEDs. These optofluidic probes minimize tissue damage and are suitable for chronic implants, with potential for broad application in biomedicial science, engineering, and medicine. Wireless in vivo fluid delivery of viruses, peptides, and small-molecule agents is possible, and when combined wireless optogenetics, can be invaluable for neural circuit dissection

Emily Ho is a first-year Neurosciences Ph.D. student in the MSTP, currently working in Dr. Pamela Mellon’s lab.

The Genetics of Alzheimer’s Disease

Dr. Rudolph E. Tanzi is a world-renowned scientist and professor of Neurology at Harvard University. Investigating the molecular and genetic basis of neurological disease since the 1980s, he co-discovered three of the first genes that can cause early-onset familial Alzheimer’s disease, including amyloid precursor protein (APP) and presenilin. In 1993, Dr. Tanzi discovered the gene responsible for the neurological disorder known as Wilson’s disease, and over the past 25 years, he has collaborated on studies identifying several other disease genes including those causing neurofibromatosis, amyotrophic lateral sclerosis, and autism.

Dr. Tanzi has published nearly 500 research papers and has received the highest awards in his field, including the Metropolitan Life Foundation Award and Potamkin Prize. He received the 2015 Smithsonian American Ingenuity Award and was named to the 2015 list of TIME100 Most Influential People in the World. He co-authored the popular trade books “Decoding Darkness”, New York Times Bestseller, “Super Brain”, and “Super Genes” He was named by GQ magazine as a Rock Star of Science, and in his spare time, has played keyboards with the band Aerosmith, guitarist, Joe Perry, and singer, Chris Mann.
Most recently, as director of the Alzheimer’s Genome Project, Dr.

Tanzi has used mutations in the same genes he helped identify decades earlier (APP and presenilin) to create a three- dimensional human stem cell-derived neural culture system that recapitulates both AD plaque and tangle pathology (Figure). This revolutionary development made great strides to overcome the limitations of Alzheimer’s disease models to date. Mouse models with familial Alzheimer’s mutations exhibit amyloid accumulation and memory deficits, but fail to recapitulate other features of AD pathology such as neurofibrillary tangle pathology. In contrast, human neurons derived from AD patients have shown elevated levels of both toxic amyloid species and phosphorylated tau, but failed to form both amyloid plaque and neurofibrillary tangle pathology.


Using this system, Dr. Tanzi is able to study the pathogenic mechanisms of Alzheimer’s disease, and test therapeutics for AD including gamma secretase modulators and metal chaperones to lower beta-amyloid and tangle burden in the brain.

A Neural Circuit Linking Location to Surroundings

Most of the time, when you go into into a room you’ve been in before, you have no issue finding your way around. Even if you have never been in the room before, it is not too difficult to orient yourself. The door is on one side of the room, windows on the other, with some furniture placed tastefully throughout. Once you come back into the room a second time, you immediately feel a sense of relative familiarity wash over you. How did you initially learn the layout of this room? Why was your internal map specific to other rooms, without falsely instilling a sense of familiarity in your new environment?

Jeffrey C. Magee became a Professor of Neuroscience at the Baylor College of Medicine and a Howard Hughes Medical Institute investigator in 2017, after a successful decade running a lab at their Janelia Research Campus in Virginia. His research aims to discover how individual neurons and their respective microcircuits process and store information. To achieve this, he uses a wide range of methods, including a variety of optical and electrical recordings and manipulations, in both the hippocampus and the cortex.


Figure 1. CA1 neurons fire rhythmic plateaus at certain locations within an environment.

In 2015, Magee and his collaborators attempt to find how neurons in the CA1 field of the hippocampus compute information they receive from two distinct regions carrying their own distinct information in their paper “Conjunctive input processing drives feature selectivity in hippocampal CA1 neurons,” published in Nature Neuroscience. These two regions are the entorhinal cortex, which uses place cells to process location information, and the CA3 field of the hippocampus, which processes contextual information. While we do know CA1 neurons themselves associate context and location, creating maps of individual environments, we do not know how these neurons actually do it.

In these CA1 neurons, they identified a consistent series of repetitive electrical activity across multiple measures in dendrites at certain locations when a mouse ran laps on a treadmill, which they termed ‘plateaus.’ Interestingly, these plateaus occurred in a regular rhythm (known as ‘theta’) found in the hippocampus during learning and sleep. These rhythms were aligned with input from CA3, which reinforces the importance of CA3 in location-specific processing and in driving the activity of these neurons.

Screen Shot 2018-04-29 at 10.15.25 PM.png

Figure 2. Inputs from entorhinal cortex drive CA1 neuron plateau firing.

However, these CA1 neurons also receive input from the entorhinal cortex. When the entorhinal cortex was inhibited, these plateaus would be shorter and smaller than normal. On the other hand, when the entorhinal cortex was stimulated, plateaus in CA1 dendrites would become longer and larger than usual. CA3 stimulation did not do anything, though, which likely means that CA3 may prime CA1 neurons for firing, but the entorhinal cortex is needed to actually cause these neurons to fire.

Screen Shot 2018-04-29 at 10.13.02 PM

Figure 3. CA1 plateau activity induces location-specific firing in a given environment.

The logical question is: what do these plateaus do, exactly? When plateaus were seen spontaneously in silent neurons, these neurons then began to fire afterwards during each lap when at the location where the plateau occurred. It appeared that these plateaus were key to inducing new place fields for these CA1 neurons when creating new mental maps, combining contextual and location information, as expected.

To check whether these plateaus directly induced place fields, they repeatedly electrically stimulated CA1 in a plateau-like manner at the same location on the track. After these artificial plateaus, the stimulated cells did subsequently respond at the location where stimulation occurred, responding like they did when the plateaus were spontaneous. Interestingly, stimulation had to be similar to a plateau, or else the stimulated neurons would not subsequently respond, showing that plateaus set these fields, instead of activity in general.

In this experiment, Magee and his colleagues were able to show how we map specific locations mentally, matching location and context into a single computation. This occurs via CA3 neurons, which bear contextual information, preparing CA1 dendrites to receive location information from the entorhinal cortex, which creates a plateau of activity. This CA1 neuron will then fire at a specific location in that specific environment, linking the two together. This serves as a good example of how the brain biologically performs a required, fundamental computation many of us normally take for granted.


James R. Howe VI is a first-year Neurosciences Ph.D. student currently rotating in Dr. Cory Root’s lab.

Crucial Neural Circuits Underlying Memory Consolidation

Memory has fascinated human beings for a long time. The French philosopher Rene Descartes described memory as an imprint made in the brain by external experience. Nineteenth-century psychologists had divided memory into distinct steps including acquisition, storage, retrieval. In the past few decades, neuroscientists have gone deeper into the neurobiological basis of memory.

Susumu Tonegawa is the Picower Professor of Biology and Neuroscience at MIT, the director of the RIKEN-MIT Center for Neural Circuit Genetics at the Picower Institute for Learning and Memory, and HHMI Investigator. The main research interest in his lab is to decipher the molecular, cellular and neural circuit mechanisms that underlie learning and memory.

From the 1950s, studies of the famous amnesiac patient Henry Molaison revealed that the hippocampus is essential for the initial formation of episodic memories but not required for long-term memory storage and retrieval. Scientists believe long-term episodic memories are stored in the neocortex, the brain region also responsible for cognitive functions such as attention and planning. How are memories transferred from short- to long-term memory (memory consolidation)?  The standard model proposes that short-term memories are initially formed and stored in the hippocampus only, and then gradually transferred to long-term storage in the neocortex and to long-term storage in the neocortex and disappearing from the hippocampus.

In the recent Science paper from Tonegawa lab, titled ‘Engrams and circuits crucial for systems consolidation of a memory’, the researchers proposed a novel model for the memory consolidation (Figure 1).

New Model

Figure 1: A New Model for Systems Consolidation of Memory

Firstly, they used the activity-dependent cell-labeling approach to label engram cells in mice during fear conditioning — a mild electric shock delivered when the mouse is in a particular chamber, in three brain regions: the hippocampus, the prefrontal cortex, and the basolateral amygdala (Figure 2). Then, they could use light to reactivate these engram cells at different times and see if that reactivation induced a freezing behavior.  The researchers could also determine which engram cells were active when the mice were placed in the chamber where the fear conditioning occurred, using in vivo calcium imaging.

Figure 2

Figure 2: Engram cell labeling in PFC, DG and BLA with H2B-GFP is DOX-dependent.

Just one day after the fear conditioning, the researchers found that memories of the event were being stored in engram cells in both hippocampus and the prefrontal cortex. However, the engram cells in the prefrontal cortex were ‘silent’— they could stimulate freezing behavior when artificially activated by light, but they did not fire during natural memory recall.  Over the next two weeks, the silent engram cells in the prefrontal cortex gradually matured, as reflected by changes in their anatomy and physiological activity, until the cells became necessary for the animals to naturally recall the event (Figure 3).  By the end of the same period, the hippocampal engram cells became silent and were no longer needed for natural recall. However, traces of the memory remained: reactivating those cells with light still provoked the animals to freeze.  While in the basolateral amygdala, once memories were formed, the engram cells remained unchanged throughout the course of the experiment. Those cells, which are necessary to evoke the emotions linked with particular memories, communicate with engram cells in both the hippocampus and the prefrontal cortex.

Figure 3Figure 3: PFC engram cells mature with time

These findings suggest that traditional theories of consolidation may not be accurate, because memories are formed rapidly and simultaneously in the prefrontal cortex and the hippocampus during the day of training. During the consolidation, the prefrontal cortex becomes stronger and the hippocampus becomes weaker.

This study also showed that communication between the prefrontal cortex and the hippocampus is critical, because blocking the circuit connection between these two regions prevented the cortical engram cells from maturing properly. It would be interesting to investigate the mechanism underlying the prefrontal cortex engram maturation process.

Another interesting question in memory is whether and how the hippocampal neurons represent the pure temporal order of episodic events.  For the lecture at CNCB Marilyn G. Farquhar Seminar Room on Tuesday afternoon 4pm, Susumu Tonegawa is going to tell us about a hippocampal code that learns and represents the pure, discrete, temporal order of events simultaneously.  Welcome to join the lecture!

Xiaochun Cai is a first-year neuroscience Ph.D. student, currently rotating in Dr. Xin Jin lab.