Mapping Neuronal Diversity: Functional Validation of Induced Neurons with HD-MEAs and DeePhys

Today I want to put a blip on your radar, turning your attention to a paper by Lin et al., published in Science. What they did in this study is incredible to me, and a real breakthrough in the iPSC field: by combining pro-neural transcription factors (TFs) with hundreds of morphogen patterning conditions, they profiled nearly 700,000 cells across 480 conditions, to set up a framework and define strategies for derivation of a diverse set of iN subtypes. Now, if you are thinking “wow, that is a lot of effort, thank you for your incredible service” well, I would agree with you, but it is not over yet.  

While single-cell transcriptomics identified these subtypes at the molecular level, the key question was: do these cells also act differently? For neurons, function ultimately comes down to one topic: electrophysiology. To address this important question, the team used the MaxTwo HD-MEA System, which enabled them to record subcellular electrophysiological signatures, single-neuron activity, and network-level dynamics from the same (many) cultures over multiple days (see Figure).  

These recording yielded rich functional data regarding their cells, which they turned into detailed insights via DeePhys (Hornauer et al., 2024), an automatic machine learning–driven analysis platform whose developed on MaxWell Biosystems’ recordings. DeePhys automatically extracted and aggregated thousands of electrophysiological features, revealing 11 distinct, single-cell functional phenotypes that - ta-da! - aligned with the morphogen conditions seen at the molecular level! The team also uncovered a clear link between axonal complexity and transcriptional maturity, performing axonal electrical footprint reconstructions which were made possible (and it’s kind of crazy that it’s something that you can just do) by the MaxTwo’s ultra-high spatiotemporal resolution.

What I really like about this study is that you know they were navigating treacherous seas, those of neuronal types and subtypes and the complexity of neuronal development. Nevertheless, they were victorious not by trying to reduce complexity by approximating, but by using the proper tools to explore such complexity in detail. Combining molecular techniques, the high-content and high-resolution of the MaxTwo HD-MEA System, and the advanced/automatic analysis from DeePhys they turned a transcriptomic atlas into a living, functional map of neuronal diversity, where identity and function could be understood side by side. Reads like an epic poem.