Simons Society of Fellows Retreat 2023

Beatrice Barra
New York University

A behavioral approach to study the neural encoding of sensation intensity in a mouse model

Perception of intensity is crucial across all sensory modalities. The intensity of tactile sensation that we experience during grasping drives the amount of strength that we use to grip objects, while the perceived intensity of light regulates the aperture of pupils to optimize vision. Without intensity information, most sensations would be uninterpretable and most daily tasks would be much harder. However, how the brain encodes intensity is currently unknown. Answering this question would be particularly relevant in the field of olfaction, where the quality of odor perception is tightly intertwined to its intensity. Several studies have attempted to explain the neural encoding of odor intensity by characterizing the neural correlates of odor concentration and formulating different hypotheses on the encoding of odor intensity. However, odor concentrations do not univocally translate to specific levels of perceived intensity: at similar concentrations some odors evoke very strong sensations, while others are merely perceivable. As a result, the neural basis of odor intensity remains unexplained. The major challenge in the study of intensity, as for most perceptual features, consists in obtaining perceptual reports from animal models. Here, we approached this problem by adapting a behavioral paradigm previously developed in a rat animal model, with the aim of finding which odorant concentrations produced percepts that matched in intensity in mice. We found that our behavioral paradigm allows to measure concentrations of different odors that produce intensity-match odor perception at different concentration ranges. We plan to leverage the results of this paradigm to record neural activations with wide field calcium imaging at intensity-matched concentrations as well as concentrations that caused an intensity mismatch, and gain insights on the neural encoding of intensity information in the mouse olfactory system.
 

Paul Chaikin
New York University

Sphere Packing in 1, 2, 3, ∞, dimensions

The densest packing of spheres, although known for millennia to be a Face-Centered Cubic (FCC) crystal with volume fraction φ_FCC ~0.74, has only recently been proven mathematically (2014). An equally ancient problem is “Random Close Packing”, RCP, the densest packing of spheres poured into a jar described in Biblical times (Luke 6:38, KJV) as, “pressed down, and shaken together, and running over”. RCP has escaped a noncontroversial definition although many experiments and simulations agree to a value φ_RCP ~0.64. We have found that a simple model, “Random Organization,” RO, exhibits a dynamical phase transition between absorbing, ‘dead’, and active states that appears to have RCP as its critical endpoint. Invented to understand a reversible to irreversible transition in sheared colloids, RO finds RCP with emergent properties such as, randomness, isotropy, isostaticity, hyperuniformity and jamming, that were previously put in by hand. It also yields an upper critical dimension of 4, ⇒ the critical behavior is mean field, infinite dimensional, for dimensions 4 → ∞.
 

Ashley Chui
New York University

Investigating the substrate network of E3 ligase AMBRA1

Maintaining the proper levels of proteins within a cell is important for homeostasis. The ubiquitin proteasome system is one way that the cell recycles its proteins to ensure proper cellular function. Within this system is a type of enzyme, the E3 ligase, that functions to ubiquitylate numerous substrates, which act as tags for targeted degradation. Furthermore, the cullin-RING ligase complexes make up a majority of the E3 ligases in our cells. One E3 ligase in the cullin-RING ligase family is AMBRA1. Recently, it has been shown that AMBRA1 targets cyclin D for protein degradation, thus acting as a regulator of the cell cycle and cell proliferation. Prior to this, much of what was known of AMBRA1’s functions largely centralized on the premise that AMBRA1 is involved in autophagy, with few discoveries of novel substrates involved in autophagy or otherwise. However, considering AMBRA1’s classic cullin-RING ligase domain architecture, it is unlikely AMBRA1 has such few characterized substrates. To identify novel substrates, we utilize the auxin-inducible degron technology, paired with quantitative mass spectrometry-based strategies. These data reveal several potential AMBRA1 substrates, including proteins involved in regulating the establishment of protein localization to the chromosomes, chaperones to the mitochondria, and receptors for proteins involved in cell attachment and migration. This highlights AMBRA1’s repertoire of substrates involved in various cellular processes beyond autophagy. Moreover, this study features applicable tools for advancing the discovery of E3 ligase-substrate pairs.
 

Benem-Orom Davids
Columbia University

Characterization of ARC Particle Production and Release

A large amount of our genome is made up of integrated retroviral viral elements, known as endogenous retroviruses (ERVs). Due to the accumulation of mutations and/or silencing by cellular defense systems, most ERVs have lost the ability to produce infectious virus. However, some ERVs have been domesticated and utilized as essential proteins for physiological processes. One fascinating example, expressed primarily in the brain, is activity-regulated cytoskeleton-associated protein (ARC), which plays an indispensable role in learning and memory formation. ARC is derived from a vertebrate lineage Ty3 retrotransposons capsid and has been reported to retain aspects of its virus-like functions, such as the ability to self-assemble, encapsidate RNA, and deliver genetic material to neighboring cells, reminiscent of the HIV-1 capsid protein (CA).

Although ARC is widely studied in the context of the brain, the mechanisms underlying particle production, characteristic, and release remain unknown. In this preliminary study, we used various biochemical approaches to investigate ARC particle production and release. We transfected cells with an ARC expression construct and show that the efficiency of ARC release is low relative to HIV. We observed similar velocity sedimentation profiles for ARC and HIV-1 particles. Interestingly, pharmacological inhibition of neutral sphingomyelinase (nSMase), an enzyme important in the first step of the ESCRT-independent exosome pathway, reduced ARC particle release. In summary, ARC particle production varies between cell lines, and the particles released are mostly uniform in size. Additionally, our work suggests ARC particle release may utilize an ESCRT-independent pathway.
 

Aravind Devarakonda
Columbia University

Designer quantum materials in heterostructures and superlattices

Manipulating materials at the atomic scale has driven advances in fundamental physics and paved the way towards technologically useful functionalities; recent examples are exciting achievements at the frontiers of quantum information science. However, the potential promised by these advances is in many cases undercut by fundamental limitations of the underlying materials. Quantum materials hosting complex, emergent phases of matter can potentially overcome these barriers.  

In this talk, I will describe how an interdisciplinary combination of materials synthesis, advanced characterization methods, and nanodevice fabrication can reveal easily accessible and highly-tunable quantum materials with the potential to advance this emergent frontier. I will introduce a new family of bulk van der Waals (vdW) superlattice materials where naturally reduced dimensionality and combined in-plane and out-of-plane periodic modulations support new and unusual electronic phases of matter, for example, exotic superconductivity; I will make connections with contemporary work surrounding artificially assembled moiré materials. I will also discuss emerging avenues which combine natural materials synthesis and artificial assembly of heterostructures to realize a new generation of highly-tunable quantum materials.
 

Robert Fernandez
Columbia University

Mapping homeobox genes in the male C. elegans nervous system to understand their role in neuronal identity and circuit assembly

A central goal in neuroscience is to understand how neural circuits develop, specifically what molecular cues are needed for the proper assembly of individual neurons into neural circuits. Homeobox genes have been shown to regulate the terminally differentiated properties of a mature neuron and delineating neuronal classes in the hermaphrodite C. elegans nervous system. However, the transcriptional programming for regulation of neuronal identity and neural circuit assembly remains little explored in the male C. elegans nervous system. Using available GFP reporter transgenes for homeobox genes and a library of molecular markers, I will test the following concepts in the male: 1) Do homeodomain transcription factors regulate the terminal fate features (neurotransmitters and their receptors, neuropeptides) of male-specific neurons? 2) Does every male-specific neuron express a unique combination of homeobox genes that defines the identity of that neuron and 3) Do homeobox genes expressed in synaptically interconnected neurons regulate neuronal circuit assembly? The C. elegans male nervous system contains five self-isolated neuronal circuits in the male tail making it ideal to test if homeobox genes are expressed in synaptically interconnected neurons. While there is evidence in support of these hypotheses in the hermaphrodite C. elegans nervous system, none of these concepts have been tested in the male C. elegans nervous system, which is a more complex system due to the addition of 91 male-specific neurons. My findings will broaden our understanding of how homeobox genes control the functional properties of a circuit and the assembly of these neurons into functional circuits.
 

Leah Houri-Zeevi
The Rockefeller University

Blood, Love, Memory: genetic programs toward behavioral shifts in mosquitoes

Mosquitoes are widely regarded as the world’s deadliest animals, responsible for more human deaths annually than any other creature. This diverse family of insects includes around 3,600 known species, each exhibiting unique ecologies, behaviors, and vectorial capacities, as well as preferences for hosts to bite. Our research investigates the mechanisms by which behaviors and neural functions are encoded and how they vary across different evolutionary timescales. To this end, we use the incredible diversity of mosquitoes, and particularly the female’s reproductive behaviors, as a model.
 

Patrick Kennedy
Columbia University

Wasps and the meaning of life

In this talk, I will tackle the meaning of life. Everywhere we look, animals and plants seem extremely busy, but what are they all trying to do? Is there a single function – a universal objective – that all organisms are striving to maximize? A controversial answer lies in an elegant theory known as “inclusive fitness.” The story involves angry communists, prison-breaking Russian princes, divine intervention, and – naturally – a lot of wasps. I will provide a whistle-stop tour through the surprising saga of inclusive fitness theory, and argue that social wasps are bringing us closer today to solving some of the outstanding questions in this ambitious but contentious field.
 

Lachlan Lancaster
Columbia University

Star Formation Feedback in Giant Molecular Clouds

Star formation in the universe is an extremely multiscale and mutli-physics process involving complex feedback loops that interact with each other in non-linear ways. It is multiscale in that fully understanding the picture of how stars form means following the movement and collapse of gas in the universe from cosmological scales (millions of light years) down to the scale of an individual star, a difference of 13 orders of magnitude. Over this range of scales, star formation tends to contain nearly a full cross-section of all physical processes: gravity, plasma physics, fluid dynamics, gas and ‘dust’ chemistry, radiative transfer, and how they all interact with one another. Additionally, when stars form they inject energy into the surrounding gas, preventing further star formation, a process called “star formation feedback.” The worst part about trying to understand this complicated process is that, as in all of astronomy, we can’t run controlled experiments! I am drawing a daunting picture, but there is hope! Amongst this vast range of scales one may think of so-called Giant Molecular Clouds (GMCs) as a fundamental building block, since all star formation that we know of proceeds in these, relatively isolated, gravitationally bound clumps of gas. I will explain how we can understand the physical processes relevant to star formation and star formation feedback on the scales of these clouds by running controlled numerical experiments (simulations!) and developing theories to explain these simplified models that can be applied more broadly.
 

Tim Large
Columbia University

Counting in modern geometry

As the mathematicians of antiquity saw it, geometry was the study of two- and three-dimensional Euclidean space through quantifiable measurements such as lengths, areas and angles. Much of contemporary geometry is concerned with the study of higher-dimensional non-Euclidean spaces, where it’s much less clear what quantities one should measure. In this talk I will describe a popular modern take on the subject, albeit one with very classical origins. One can “measure” higher-dimensional spaces by counting the numbers of embedded two-dimensional surfaces in them satisfying particular equations and incidence conditions. These integer- and rational-number-valued counts then give an important window through which we can glimpse the mysteries of certain special geometries. The resulting theory is beautiful and still largely unexplored, with important links to many other parts of math and high-energy physics.
 

Jenny Merritt
Columbia University

The genetic basis of steroid levels in biparental deer mice

Mammalian mothers experience a suite of behavioral and physiological changes that occur during pregnancy, parturition, and lactation. In contrast, fathers do not gestate nor, in most mammalian species, provide care for their offspring. The few mammalian species that exhibit biparental care offer a unique opportunity to study the mechanisms of parental care in both sexes. In mice of the genus Peromyscus, a pair of closely related species have evolved different mating systems. In P. polionotus both parents provide care for the pups, but in P. maniculatus, only mothers provide care. Here, we investigate the endocrine basis of parental care in both sexes. We found that P. polionotus males have 2-fold smaller testes than P. maniculatus and both males and females have a 5-fold larger adrenal cortex. Consistent with these anatomical differences, levels of testosterone, progesterone, and corticosterone have diverged substantially between the species. To understand the genetic basis of steroid hormone levels, we generated 769 F2 hybrids, quantified parental behavior towards pups, measured steroid concentrations, and performed quantitative genetic mapping. We identified the regions of the genome in both virgins and parents that contribute to levels of each steroid hormone. Because we prospectively sampled steroid hormone concentrations, we were able to detect a causal role of hormones on parental behavior. Our mapping showed that four loci contributing to steroid levels overlap with regions of the genome contributing to parental behaviors and defense of offspring, indicating these phenotypes share a genetic basis. The loci contain likely candidate genes, including a glucocorticoid receptor and a prostaglandin receptor. Taken together, our genetic dissection of steroid levels offers many new molecular targets for understanding the control of parental care, and brings us closer to understanding the common mechanisms by which animals contribute to the survival of their offspring. 
 

Amy Lee Norovich
Columbia University

Siamese fighting fish: a novel model for studying sex differences and the role of visual cues in aggression

Aggression is a universal behavior that shapes human and animal societies. In humans and other primates, visual information plays a prominent role in eliciting aggression yet in rodent and fly models commonly used in behavioral neuroscience, aggression is evoked largely by smell and pheromonal cues. How visual cues influence aggressive behavior – and social behavior more generally – is therefore relatively unexplored.

To address this, I have developed the Siamese fighting fish Betta splendens (betta) as a new animal model. Betta exhibit extreme aggression that is evoked solely by visual cues and is markedly more prominent in males than females. How do visual cues drive aggressive behavior? And how do the same cues evoke different behavioral responses in males and females? To answer these questions, I have developed behavioral assays that reliably elicit aggression in adult betta, as well as machine learning-based methods to analyze the behavior. I have used molecular methods to identify brain regions associated with aggression and to determine how these regions intersect with the visual system, and am currently characterizing gene expression in these regions to identify molecular pathways mediating sex differences in behavior. Going forward, we are developing physiological methods to examine neural activity in real time.