Leveraging dendritic properties to advance machine learning and neuro-inspired computing

The brain is a remarkably capable and efficient system. It can process and store huge amounts of noisy and unstructured information, using minimal energy. In contrast, current artificial intelligence (AI) systems require vast resources for training while still struggling to compete in tasks that are trivial for biological agents. Thus, brain-inspired engineering has emerged as a promising new avenue for designing sustainable, next-generation AI systems. Here, we describe how dendritic mechanisms of biological neurons have inspired innovative solutions for significant AI problems, including credit assignment in multi-layer networks, catastrophic forgetting, and high-power consumption. These findings provide exciting alternatives to existing architectures, showing how dendritic research can pave the way for building more powerful and energy efficient artificial learning systems.

Dendrites and efficiency: Optimizing performance and resource utilization

The brain is a highly efficient system that has evolved to optimize performance under limited resources. In this review, we highlight recent theoretical and experimental studies that support the view that dendrites make information processing and storage in the brain more efficient. This is achieved through the dynamic modulation of integration versus segregation of inputs and activity within a neuron. We argue that under conditions of limited energy and space, dendrites help biological networks to implement complex functions such as processing natural stimuli on behavioral timescales, performing the inference process on those stimuli in a context-specific manner, and storing the information in overlapping populations of neurons. A global picture starts to emerge, in which dendrites help the brain achieve efficiency through a combination of optimization strategies that balance the tradeoff between performance and resource utilization.

Leveraging dendritic properties to advance machine learning and neuro-inspired computing

The brain is a remarkably capable and efficient system. It can process and store huge amounts of noisy and unstructured information using minimal energy. In contrast, current artificial intelligence (AI) systems require vast resources for training while still struggling to compete in tasks that are trivial for biological agents. Thus, brain-inspired engineering has emerged as a promising new avenue for designing sustainable, next-generation AI systems. Here, we describe how dendritic mechanisms of biological neurons have inspired innovative solutions for significant AI problems, including credit assignment in multilayer networks, catastrophic forgetting, and high energy consumption. These findings provide exciting alternatives to existing architectures, showing how dendritic research can pave the way for building more powerful and energy-efficient artificial learning systems.

Dendrites and Efficiency: Optimizing Performance and Resource Utilization

The brain is a highly efficient system evolved to achieve high performance with limited resources. We propose that dendrites make information processing and storage in the brain more efficient through the segregation of inputs and their conditional integration via nonlinear events, the compartmentalization of activity and plasticity and the binding of information through synapse clustering. In real-world scenarios with limited energy and space, dendrites help biological networks process natural stimuli on behavioral timescales, perform the inference process on those stimuli in a context-specific manner, and store the information in overlapping populations of neurons. A global picture starts to emerge, in which dendrites help the brain achieve efficiency through a combination of optimization strategies balancing the tradeoff between performance and resource utilization.

RIPK1 and TRADD Regulate TNF-Induced Signaling and Ripoptosome Formation

TNF is a proinflammatory cytokine that is critical for the coordination of tissue homeostasis. RIPK1 and TRADD are the main participants in the transduction of TNF signaling. However, data on the cell fate-controlling functions of both molecules are quite controversial. Here, we address the functions of RIPK1 and TRADD in TNF signaling by generating RIPK1- or TRADD-deficient human cell lines. We demonstrate that RIPK1 is relevant for TNF-induced apoptosis and necroptosis in conditions with depleted IAPs. In addition, TRADD is dispensable for necroptosis but required for apoptosis. We reveal a new possible function of TRADD as a negative regulator of NIK stabilization and subsequent ripoptosome formation. Furthermore, we show that RIPK1 and TRADD do not appear to be essential for the activation of MAPK signaling. Moreover, partially repressing NF-κB activation in both RIPK1 and TRADD KO cells does not result in sensitization to TNF alone due to the absence of NIK stabilization. Importantly, we demonstrate that RIPK1 is essential for preventing TRADD from undergoing TNF-induced ubiquitination and degradation. Taken together, our findings provide further insights into the specific functions of RIPK1 and TRADD in the regulation of TNF-dependent signaling, which controls the balance between cell death and survival.

TNF Is Partially Required for Cell-Death-Triggered Skin Inflammation upon Acute Loss of cFLIP

cFLIP is required for epidermal integrity and skin inflammation silencing via protection from TNF-induced keratinocyte apoptosis. Here, we generated and analyzed cFLIP epidermal KO mice with additional TNF deficiency. Intriguingly, the ablation of TNF rescued the pathological phenotype of epidermal cFLIP KO from characteristic weight loss and increased mortality. Moreover, the lack of TNF in these animals strongly reduced and delayed the epidermal hyperkeratosis and the increased apoptosis in keratinocytes. Our data demonstrate that TNF signaling in cFLIP-deficient keratinocytes is the critical factor for the regulation of skin inflammation via modulated cytokine and chemokine expression and, thus, the attraction of immune cells. Our data suggest that autocrine TNF loop activation upon cFLIP deletion is dispensable for T cells, but is critical for neutrophil attraction. Our findings provide evidence for a negative regulatory role of cFLIP for TNF-dependent apoptosis and partially for epidermal inflammation. However, alternative signaling pathways may contribute to the development of the dramatic skin disease upon cFLIP deletion. Our data warrant future studies of the regulatory mechanism controlling the development of skin disease upon cFLIP deficiency and the role of cFLIP/TNF in a number of inflammatory skin diseases, including toxic epidermal necrolysis (TEN).

Emergence of Coordinated Activity in the Developing Entorhinal-Hippocampal Network

Correlated activity in the entorhinal–hippocampal neuronal networks, supported by oscillatory and intermittent population activity patterns is critical for learning and memory. However, when and how correlated activity emerges in these networks during development remains largely unknown. Here, we found that during the first postnatal week in non-anaesthetized head-restrained rats, activity in the superficial layers of the medial entorhinal cortex (MEC) and hippocampus was highly correlated, with intermittent population bursts in the MEC followed by early sharp waves (eSPWs) in the hippocampus. Neurons in the superficial MEC layers fired before neurons in the dentate gyrus, CA3 and CA1. eSPW current-source density profiles indicated that perforant/temporoammonic entorhinal inputs and intrinsic hippocampal connections are co-activated during entorhinal–hippocampal activity bursts. Finally, a majority of the entorhinal–hippocampal bursts were triggered by spontaneous myoclonic body movements, characteristic of the neonatal period. Thus, during the neonatal period, activity in the entorhinal cortex (EC) and hippocampus is highly synchronous, with the EC leading hippocampal activation. We propose that such correlated activity is embedded into a large-scale bottom-up circuit that processes somatosensory feedback resulting from neonatal movements, and that it is likely to instruct the development of connections between neocortex and hippocampus.

A20 Promotes Ripoptosome Formation and TNF-Induced Apoptosis via cIAPs Regulation and NIK Stabilization in Keratinocytes

The ubiquitin-editing protein A20 (TNFAIP3) is a known key player in the regulation of immune responses in many organs. Genome-wide associated studies (GWASs) have linked A20 with a number of inflammatory and autoimmune disorders, including psoriasis. Here, we identified a previously unrecognized role of A20 as a pro-apoptotic factor in TNF-induced cell death in keratinocytes. This function of A20 is mediated via the NF-κB-dependent alteration of cIAP1/2 expression. The changes in cIAP1/2 protein levels promote NIK stabilization and subsequent activation of noncanonical NF-κB signaling. Upregulation of TRAF1 expression triggered by the noncanonical NF-κB signaling further enhances the NIK stabilization in an autocrine manner. Finally, stabilized NIK promotes the formation of the ripoptosome and the execution of cell death. Thus, our data demonstrate that A20 controls the execution of TNF-induced cell death on multiple levels in keratinocytes. This signaling mechanism might have important implications for the development of new therapeutic strategies for the treatment of A20-associated skin diseases.

The Impact of Movement on Disease Dynamics in a Multi-city Compartmental Model Including Residency Patch

The impact of population dispersal between two cities on the spread of a disease is investigated analytically. A general SIRS model is presented that tracks the place of residence of individuals, allowing for different movement rates of local residents and visitors in a city. Provided the basic reproduction number is greater than one, we demonstrate in our model that increasing the travel volumes of some infected groups may result in the extinction of a disease, even though the disease cannot be eliminated in each city when the cities are isolated.

cFLIP regulates skin homeostasis and protects against TNF-induced keratinocyte apoptosis

FADD, caspase-8, and cFLIP regulate the outcome of cell death signaling. Mice that constitutively lack these molecules die at an early embryonic age, whereas tissue-specific constitutive deletion of FADD or caspase-8 results in inflammatory skin disease caused by increased necroptosis. The function of cFLIP in the skin in vivo is unknown. In contrast to tissue-specific caspase-8 knockout, we show that mice constitutively lacking cFLIP in the epidermis die around embryonic days 10 and 11. When cFLIP expression was abrogated in adult skin of cFLIPfl/fl-K14CreERtam mice, severe inflammation of the skin with concomitant caspase activation and apoptotic, but not necroptotic, cell death developed. Apoptosis was dependent of autocrine tumor necrosis factor production triggered by loss of cFLIP. In addition, epidermal cFLIP protein was lost in patients with severe drug reactions associated with epidermal apoptosis. Our data demonstrate the importance of cFLIP for the integrity of the epidermis and for silencing of spontaneous skin inflammation.

Cell death in the skin: how to study its quality and quantity?

The characterization of the quality and quantity of cell death has gained substantial interest over the past decades. More recently necroptosis as a programmed form of necrosis has been identified as an important additional form of cell death with relevance in the skin. Understanding how to assay cell death in specific is of critical importance for cancer research and treatment. Here we describe six different methods that can be used to assay cell viability and to study the quality or quantity of cultured human keratinocytes in vitro. These methods include crystal violet assay, hypodiploidy analysis, caspase-8 cleavage, release of HMGB1, annexin V/propidium iodide co-staining, and Hoechst/SYTOX green co-staining.

The impact of CFNS-causing EFNB1 mutations on ephrin-B1 function

Background

Mutations of EFNB1 cause the X-linked malformation syndrome craniofrontonasal syndrome (CFNS). CFNS is characterized by an unusual phenotypic pattern of inheritance, because it affects heterozygous females more severely than hemizygous males. This sex-dependent inheritance has been explained by random X-inactivation in heterozygous females and the consequences of cellular interference of wild type and mutant EFNB1-expressing cell populations. EFNB1 encodes the transmembrane protein ephrin-B1, that forms bi-directional signalling complexes with Eph receptor tyrosine kinases expressed on complementary cells. Here, we studied the effects of patient-derived EFNB1 mutations predicted to give rise to truncated ephrin-B1 protein or to disturb Eph/ephrin-B1 reverse ephrin-B1 signalling. Five mutations are investigated in this work: nonsense mutation c.196C > T/p.R66X, frameshift mutation c.614_615delCT, splice-site mutation c.406 + 2T > C and two missense mutations p.P54L and p.T111I. Both missense mutations are located in the extracellular ephrin domain involved in Eph-ephrin-B1 recognition and higher order complex formation.

Methods

Nonsense mutation c.196C > T/p.R66X, frameshift mutation c.614_615delCT and splice-site mutation c.406+2T > C were detected in the primary patient fibroblasts by direct sequencing of the DNA and were further analysed by RT-PCR and Western blot analyses.

The impact of missense mutations p.P54L and p.T111I on cell behaviour and reverse ephrin-B1 cell signalling was analysed in a cell culture model using NIH 3T3 fibroblasts. These cells were transfected with the constructs generated by in vitro site-directed mutagenesis. Investigation of missense mutations was performed using the Western blot analysis and time-lapse microscopy.

Results and Discussion

Nonsense mutation c.196C > T/p.R66X and frameshift mutation c.614_615delCT escape nonsense-mediated RNA decay (NMD), splice-site mutation c.406+2T > C results in either retention of intron 2 or activation of a cryptic splice site in exon 2. However, c.614_615delCT and c.406+2T > C mutations were found to be not compatible with production of a soluble ephrin-B1 protein. Protein expression of the p.R66X mutation was predicted unlikely but has not been investigated.

Ectopic expression of p.P54L ephrin-B1 resists Eph-receptor mediated cell cluster formation in tissue culture and intracellular ephrin-B1 Tyr324 and Tyr329 phosphorylation. Cells expressing p.T111I protein show similar responses as wild type expressing cells, however, phosphorylation of Tyr324 and Tyr329 is reduced.

Conclusions

Pathogenic mechanisms in CFNS manifestation include impaired ephrin-B1 signalling combined with cellular interference.