This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

Erythropoietin (Epo) plays a dual role as an erythropoiesis-stimulating hormone and a locally produced cytoprotectant in various vertebrate tissues. Splice variants and engineered derivatives of Epo that mediate neuroprotection but do not stimulate erythropoiesis suggest that alternative receptors, different from the 'classical' homodimeric receptor involved in haematopoiesis, mediate neuroprotective Epo functions. Previous studies on grasshoppers demonstrated neuroprotective and neuroregenerative effects of Epo that involved similar transduction pathways as in mammals. To advance the characterization of yet unidentified neuroprotective Epo receptors, we studied the neuroprotective potency of the human non-erythropoietic Epo splice variant EV-3 in primary cultured locust brain neurons. We demonstrate that EV-3, like Epo, protects locust neurons from hypoxia-induced apoptotic death through activation of the Janus kinase/signal transducer and activator of transcription transduction pathway. Using the fluorescent dye FM1-43 to quantify endocytotic activity we show that both Epo and EV-3 increase the number of fluorescently labelled endocytotic vesicles. This reveals that binding of Epo to its neuroprotective receptor induces endocytosis, as it has been described for the mammalian homodimeric Epo-receptor expressed by erythroid progenitors. Reduction in Epo-stimulated endocytotic activity following pre-exposure to EV-3 indicated that both Epo and its splice variant bind to the same receptor on locust neurons. The shared neuroprotective potency of Epo and EV-3 in insect and mammalian neurons, in the absence of erythropoietic effects of EV-3 in mammals, suggests a greater similarity of the unidentified nervous Epo receptors (or receptor complexes) across phyla than between mammalian haematopoietic and neuroprotective receptors. Insects may serve as suitable models to evaluate the specific protective mechanisms mediated by Epo and its variants in non-erythropoietic mammalian tissues.

Visualizing features in deep neural networks (DNNs) can help understanding their computations. Many previous studies aimed to visualize the selectivity of individual units by finding meaningful images that maximize their activation. However, comparably little attention has been paid to visualizing to what image transformations units in DNNs are invariant. Here we propose a method to discover invariances in the responses of hidden layer units of deep neural networks. Our approach is based on simultaneously searching for a batch of images that strongly activate a unit while at the same time being as distinct from each other as possible. We find that even early convolutional layers in VGG-19 exhibit various forms of response invariance: near-perfect phase invariance in some units and invariance to local diffeomorphic transformations in others. At the same time, we uncover representational differences with ResNet-50 in its corresponding layers. We conclude that invariance transformations are a major computational component learned by DNNs and we provide a systematic method to study them.

In genomics, transcriptomics, and related biological fields (collectively known as 'omics'), it is common to work with n p data sets with the dimensionality much larger than the sample size. In recent years, combinations of experimental techniques began to yield multiple sets of features for the same set of biological replicates. One example is Patch-seq, a method combining single-cell RNA sequencing with electrophysiological recordings from the same cells. Here we present a framework based on sparse reduced-rank regression for obtaining an interpretable visualization of the relationship between the transcriptomic and the electrophysiological data. We use an elastic net regularization penalty that yields sparse solutions and allows for an efficient computational implementation. Using several publicly available Patch-seq data sets, we show that sparse reduced-rank regression outperforms both sparse full-rank regression and non-sparse reduced-rank regression in terms of predictive performance, and can outperform existing methods for sparse partial least squares and sparse canonical correlation analysis in terms of out-of-sample correlations. We introduce a 'bibiplot' visualization in order to display the dominant factors determining the relationship between transcriptomic and electrophysiological properties of neurons. We believe that sparse reduced-rank regression can provide a valuable tool for the exploration and visualization of multimodal data sets, including Patch-seq.