Jack Winters (Marquette University)
Abstract. Perhaps missing from the short lists of the great examples of CAS is “living tissue.” Living tissues form a dynamic bridge between organ systems and cell communities, providing 3D structure and functional capacity. As local demands change, cooperative adaptive processes strive to enhance the greater self-good (“fitness”). Growing tissue is complex – just ask tissue bioengineers who’ve strugged to find the optimal mix of cell type diversity plus nurturing biochemical/biomechanical/bioelectrical conditions that enable context-aware, decentralized interconnectivity (including angiogenesis).
In terms of the great interwoven information-material-energy trinity at the pillar of the laws of nature, muscle tissue provides an intimate natural bridge between neural expression and connective tissue structural/functional design. A clever neuro-mechanical strategy enables coordinated actuation of millions of nano-motors distributed within a volume, each accessing “wireless energy” via a tightly-regulated ATP supply. The same basic “winning” myo-design, with a few variants (e.g., ~5 for adult skeletal, 2 for cardiac, ~2 for smooth), enables creation of muscle actuators possessing a remarkable diversity of sizes and functional capacities (e.g., insect wings, elephant legs, beating hearts).
High-throughput microarray mRNA data from a typical skeletal muscle tissue sample finds ~35-40% of genetic content to be “expressed” (e.g., z-scores >-0.1), but to remarkably different degrees (e.g., top 5% with z>1.1, 2% with z>9, with top players changing dramatically with pathology). These transcripts mostly come from ~7 of the ~250 “potential information field” cell types (e.g., 3 muscle types, fibroblasts, endothelial, vascular smooth muscle, satellite). The portfolio includes >40% that can be classified as specifying local “controls apparatus” (e.g., ~10% for each for sensory receptors, network cascades, materials/energy regulators, transcription factors (TFs)). We’ve recently partitioned the myo-collection into 5 transcript-protein families (1 mechanical, 1 energetic, 3 informational), each with 3-6 subfamilies. Unlike our engineered motors, muscle tissues selflessly remodel (up-/down-regulate parts) within all families (mechanical, energetic, informational) based on macro-intuitive yet illusive cyber-rules sampling local “use history” (e.g., self-generated action, micro-trauma, endocrine signals).
Two myo-tissue modeling approaches are briefly presented: i) our “real-time” modeling strategy that extends my work in the 1980s-90s to a “living muscle model” that integrates energetic as well as mechanical-informational dynamic modeling and explicitly ties model parameters to strategic transcript-protein z-scores (e.g., “personalized” for diseased muscle); and ii) our “adaptive-time” modeling strategy based on “trans-genetic soft fuzzy-agent” cyber-nets (TF-nets) that emphasizes TF-circuitry motifs and aims to predict classic resistance and endurance training data, and various muscle “disuse” paradigms.
In closing, it seems that TF-nets – the pillars of living systems - are designed and organized differently from many CAS designs. For example, in neural networks it is connectivity rather than diversity that matters most (“who you know” and “what module/hub you are in”), yet in TF-nets it is “what are your skills” and “who do you work for” that matters, with trans-input connectivity limited to ~10. A typical tissue-specific “portfolio” could be viewed as specialized widgets available to a “workforce” of skilled TF-agents managing cyber-regulation, with no CEO. There are analogies to infrastructure support in cities; perhaps we can learn from some of the CcAS strategies used by robust living tissues such as muscle.