Zambrano, D. | Search bioRxiv

bioRxiv2026

Computational design of membrane fusion proteins

The fusion of two distinct biological membranes is an evolutionarily conserved process essential to cellular organization and physiology. Membrane fusion is driven by the refolding of fusogenic proteins into low-energy postfusion states that overcome the energetic barrier to bilayer merger. Here we report a computational method for the design of synthetic fusogens inspired by the architecture of the soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) complex. Using machine learning-guided protein design to extensively remodel backbone geometry and sequence, we generated heterodimeric SNARE-like assemblies that efficiently catalyze cell-cell membrane fusion. These minimal two-component fusogens exhibit substantially higher fusion activity than native multisubunit SNARE complexes. Structural and functional analyses identify the key determinants required for fusogenic activity and reveal a modularity that enables control of fusion through chemically induced heterodimerization. In addition to cell-cell fusion, the synthetic fusogens drive fusion between endoplasmic reticulum and mitochondrial membranes from human cells, demonstrating their potential as tools for programmable manipulation of intracellular membranes. Together, these results establish a general framework for the rational design of synthetic fusogens and expand the toolkit for engineering membrane dynamics in living systems.

biochemistry↗

bioRxiv2020

Leveraging spiking deep neural networks tounderstand neural mechanisms underlyingselective attention

Spatial attention enhances sensory processing of goal-relevant information and improves perceptual sensitivity. Yet, the specific neural mechanisms underlying the effects of spatial attention on performance are still contested. Here, we examine different attention mechanisms in spiking deep convolutional neural networks. We directly contrast effects of precision (internal noise suppression) and two different gain modulation mechanisms on performance on a visual search task with complex real-world images. Unlike standard artificial neurons, biological neurons have saturating activation functions, permitting implementation of attentional gain as gain on a neurons input or on its outgoing connection. We show that modulating the connection is most effective in selectively enhancing information processing by redistributing spiking activity, and by introducing additional task-relevant information, as shown by representational similarity analyses. Precision only produced minor attentional effects in performance. Our results, which mirror empirical findings, show that it is possible to adjudicate between attention mechanisms using more biologically realistic models and natural stimuli.

neuroscience↗