The axon initial segment (AIS) constitutes not only the site of action potential initiation, but also a hub for activity-dependent modulation of output generation. Recent studies shedding light on AIS function used predominantly post-hoc approaches since no robust murine in vivo live reporters exist. Here, we introduce a reporter line in which the AIS is intrinsically labeled by an ankyrin-G-GFP fusion protein activated by Cre recombinase, tagging the native Ank3 gene. Using confocal, superresolution, and two-photon microscopy as well as whole-cell patch-clamp recordings in vitro, ex vivo, and in vivo, we confirm that the subcellular scaffold of the AIS and electrophysiological parameters of labeled cells remain unchanged. We further uncover rapid AIS remodeling following increased network activity in this model system, as well as highly reproducible in vivo labeling of AIS over weeks. This novel reporter line allows longitudinal studies of AIS modulation and plasticity in vivo in real-time and thus provides a unique approach to study subcellular plasticity in a broad range of applications.

Decoders for brain-computer interfaces (BCIs) assume constraints on neural activity, chosen to reflect scientific beliefs while yielding tractable computations. Recent scientific advances suggest that the true constraints on neural activity, especially its geometry, may be quite different from those assumed by most decoders. We designed a decoder, MINT, to embrace statistical constraints that are potentially more appropriate. If those constraints are accurate, MINT should outperform standard methods that explicitly make different assumptions. Additionally, MINT should be competitive with expressive machine learning methods that can implicitly learn constraints from data. MINT performed well across tasks, suggesting its assumptions are well-matched to the data. MINT outperformed other interpretable methods in every comparison we made. MINT outperformed expressive machine learning methods in 37 of 42 comparisons. MINT’s computations are simple, scale favorably with increasing neuron counts, and yield interpretable quantities such as data likelihoods. MINT’s performance and simplicity suggest it may be a strong candidate for many BCI applications.