Jonathan Witztum

Emergent Large Language Models (LLMs) show impressive capabilities in performing a wide range of tasks. These models can be harnessed for biophysical use, as well. The main challenge in this endeavor lies in transforming 3D chemical data into 1D language-like data. We developed a method to transform molecular data into language-like data and tokenize it for LLM use in biophysical context. We then trained a model, 3bmGPT, and validated it with a known protein-ligand complex. Using the pre-trained result, the model can assess the chemical properties of targets, detect shared binding properties and structures, and reveal related drugs. 3bmGPT and the synthetic-language to describe binding interactions uncovered novel protein-protein networks influenced by ligands, indicating functionally related yet previously unreported interactions. We provide open access to a fully functional web-based tool utilizing 3bmGPT.

The cellular mechanisms of autism spectrum disorder (ASD) are poorly understood. Cumulative evidence suggests that abnormal synapse function underlies many features of this disease. Astrocytes regulate several key neuronal processes, including the formation of synapses and the modulation of synaptic plasticity. Astrocyte abnormalities have also been identified in the postmortem brain tissue of ASD individuals. However, it remains unclear whether astrocyte pathology plays a mechanistic role in ASD, as opposed to a compensatory response. To address this, we combined stem cell culturing with transplantation techniques to determine disease-specific properties inherent to ASD astrocytes. We demonstrate that ASD astrocytes induce repetitive behavior as well as impair memory and long-term potentiation when transplanted into the healthy mouse brain. These in vivo phenotypes were accompanied by reduced neuronal network activity and spine density caused by ASD astrocytes in hippocampal neurons in vitro. Transplanted ASD astrocytes also exhibit exaggerated Ca²⁺ fluctuations in chimeric brains. Genetic modulation of evoked Ca²⁺ responses in ASD astrocytes modulates behavior and neuronal activity deficits. Thus, this study determines that astrocytes derived from ASD iPSCs are sufficient to induce repetitive behavior as well as cognitive deficit, suggesting a previously unrecognized primary role for astrocytes in ASD.

Why is ketamine an antidepressant?: A better understanding of the mechanisms underlying the action of antidepressants is urgently needed. Moda-Sava et al. explored a possible mode of action for the drug ketamine, which has recently been shown to help patients recover from depression (see the Perspective by Beyeler). Ketamine rescued behavior in mice that was associated with depression-like phenotypes by selectively reversing stress-induced spine loss and restoring coordinated multicellular ensemble activity in prefrontal microcircuits. The initial induction of ketamine's antidepressant effect on mouse behavior occurred independently of effects on spine formation. Instead, synaptogenesis in the prefrontal region played a critical role in nourishing these effects over time. Interventions aimed at enhancing the survival of restored synapses may thus be useful for sustaining the behavioral effects of fast-acting antidepressants. Science , this issue p. eaat8078 ; see also p. 129

Fear can be highly adaptive in promoting survival, yet it can also be detrimental when it persists long after a threat has passed. Flexibility of the fear response may be most advantageous during adolescence when animals are prone to explore novel, potentially threatening environments. Two opposing adolescent fear-related behaviours—diminished extinction of cued fear and suppressed expression of contextual fear—may serve this purpose, but the neural basis underlying these changes is unknown. Using microprisms to image prefrontal cortical spine maturation across development, we identify dynamic BLA-hippocampal-mPFC circuit reorganization associated with these behavioural shifts. Exploiting this sensitive period of neural development, we modified existing behavioural interventions in an age-specific manner to attenuate adolescent fear memories persistently into adulthood. These findings identify novel strategies that leverage dynamic neurodevelopmental changes during adolescence with the potential to extinguish pathological fears implicated in anxiety and stress-related disorders.

Perceptual spaces play a key role in intermediate stages of visual processing, as they constitute representations that support discrimination, categorization, working memory, and other judgments. The classical perceptual space of color has three dimensions; others, such as faces and textures, have very high dimension. Representing such spaces within biological constraints is challenging. To probe how visual cortex does this, we studied responses of macaque single neurons to a well-characterized 10-dimensional domain of visual textures. This model space consists of black-and-white colorings of a checkerboard, and its spatial structure is defined by multipoint correlations within 2x2 blocks (Victor and Conte 2012). The 10 parameters of the space are: a statistic that specifies the balance of white vs. black checks, four pairwise statistics that specify nearest-neighbor correlations, four three-point statistics that specify the prevalence of triangular regions, and a four-point statistic. This space captures the informative local image statistics of natural images (Hermundstad, 2014). Human perceptual sensitivities within the space are well-characterized, and suggest opponent mechanisms that combine via quadratic summation. We recorded from 214 neurons via tetrodes in V1 and V2 of 6 anesthetized macaques. Neurons responsive to one- and two-point statistics were common (>50%) in V1 and V2, with the previously-noted bias towards darks in both regions (Yeh et al., 2009). Responses to three- and four-point statistics were rare (< 10%) in V1, but more common in V2 (~20%). Consistent with an opponent representation, neurons that increased their firing in response to variation in one direction along a coordinate axis tended to decrease their firing in response to variation in the opposite direction. However, typical neurons responded to variations along several coordinate axes, with no clear pattern of "cardinal" directions. Thus, our data suggest a multidimensional opponent representation for local image statistics in the population of V1 and V2 neurons. Meeting abstract presented at VSS 2015

The availability of many complete, annotated proteomes enables the systematic study of the relationships between protein conservation and functionality. We explore this question based solely on the presence or absence of protein homologues (a.k.a. conservation profiles). We study 18 metazoans, from two distinct points of view: the human's and the fly's. Using the GOrilla gene ontology (GO) analysis tool, we explore functional enrichment of the “universal proteins”, those with homologues in all 17 other species, and of the “non-universal proteins”. A large number of GO terms are strongly enriched in both human and fly universal proteins. Most of these functions are known to be essential. A smaller number of GO terms, exhibiting markedly different properties, are enriched in both human and fly non-universal proteins. We further explore the non-universal proteins, whose conservation profiles are consistent with the “tree of life” (TOL consistent), as well as the TOL inconsistent proteins. Finally, we applied Quantum Clustering to the conservation profiles of the TOL consistent proteins. Each cluster is strongly associated with one or a small number of specific monophyletic clades in the tree of life. The proteins in many of these clusters exhibit strong functional enrichment associated with the “life style” of the related clades. Most previous approaches for studying function and conservation are “bottom up”, studying protein families one by one, and separately assessing the conservation of each. By way of contrast, our approach is “top down”. We globally partition the set of all proteins hierarchically, as described above, and then identify protein families enriched within different subdivisions. While supporting previous findings, our approach also provides a tool for discovering novel relations between protein conservation profiles, functionality, and evolutionary history as represented by the tree of life.