Eric H. Chang ^1,2,3*

• ¹ Feinstein Institutes for Medical Research, Northwell Health, Manhasset, United States
• ² Donald and Barbara Zucker School of Medicine, Hofstra University, Hempstead, New York, United States
• ³ Elmezzi Graduate School of Molecular Medicine, Manhasset, New York, United States

Neuroscientists have traditionally taken a reductionist approach to understanding the immense complexity of the nervous system. As is the case in other fields of biology, the method of reducing nervous systems into their constitutive parts has proven useful for understanding neural circuits and how they function. As a result, modern neuroscience has thrived on cataloging and scrutinizing individual components of intricate neural systems. Yet, substantial gaps persist in comprehending how these disparate parts coalesce and interact to generate higher-order functions such as behavior, memory, and even consciousness. Bridging these gaps requires a concerted effort to integrate knowledge across subfields in neuroscience, and more broadly, across biology. By taking a systems biology approach to understanding nervous system complexity, we can attempt to build links between molecules, genes, synapses, and behavior.The struggle between understanding individual parts and the whole has been a part of neuroscience since its origin as a scientific discipline. Over a century ago, the field was shaped by the opposing theories of two leading neuroanatomists, Santiago Ramón y Cajal and Camillo Golgi. On the one hand, Golgi's reticular doctrine posited that the nervous system was an interconnected nerve network ("a large syncytium") that was seamless and continuous [Glickstein, 2012] In contrast, Cajal proposed the neuron doctrine which stated that individual nerve cells were the basic structural and functional units of the nervous system [Cajal, 1888]. The structural evidence from the microscopes and stains available to scientists at the time supported Cajal's neuron doctrine. In fact, it was Golgi's "black reaction" (known as a Golgi stain) that produced the most convincing structural evidence that neurons were structurally separated elements [Glickstein 2006]. The introduction of the electron microscope in the 1940s definitively demonstrated that neurons were not continuous but were instead distinct entities separated by synapses with extracellular space in between them. Eventually, Santiago Ramón y Cajal would widely be considered to be the father of modern neuroscience and his neuron doctrine has served as a foundation for the field of neuroscience [Yuste, 2015].Perhaps because of these foundational principles, many of the workhorse techniques and methods of modern neuroscience have thus far been catered to the investigation of individual components that make up neural circuits. For example, Golgi stains and patch-clamp electrophysiological recordings highlight individual neurons. This conceptual focus on individual neurons has obscured, to some extent, our ability to integrate data on how individual function enables higher order processes [Yuste 2015]. As a result, what is missing in the field are general theories of nervous system function that explain how individual neurons contribute to neural circuits that then give rise to behavior, cognition, and other emergent properties of nervous systems. This section of Integrative Systems Neuroscience seeks to address some of these knowledge gaps with multidisciplinary and multiscale analyses of nervous system function. Integrative systems neuroscience is a field that aims to connect disparate components and levels of analyses to gain a better understanding of systems neuroscience. It represents the union of systems biology and neuroscience to integrate results across multiple different levels of analysis and scales, in both the spatial and temporal domains. Using a systems biology approach, we seek to understand complex biological systems by considering them as a whole rather than just the sum of their individual parts. This is accomplished by integrating data from various biological levels and through different analysis approaches to construct comprehensive models or simulations of how different biological components of the nervous system interact.One particular challenge for integrating information in nervous systems is that the constitutive components operate over many orders of magnitude, at least six in the spatial domain [Figure 1] and over nine in the temporal domain (e.g. milliseconds to years). As a result, we are often left with gaps in knowledge between several scales and domains that precludes a more general understanding of function or behavior. The extreme complexity of nervous systems also represents a huge challenge to integrating across levels and scales. In fact, it was the high complexity of nerve tissue that made neural structures so difficult to characterize in the late 1800s, even though the cell theory was already developed and generally accepted for other biological systems.One approach to addressing potential knowledge gaps within a space is to start with the data and go toward abstraction, such that you start from first principles and then you ascend. But oftentimes nervous systems are so complicated and dynamic that this cannot be accomplished in any logical manner. For example, if we have structural biology data that includes angstrom-level resolution to resolve the crystalography of individual proteins (e.g. ion channels), then how can we map this to single neuron structure and function? How do we map that microscopic information to brain-wide circuits? If we instead approach the problem from the top-down (e.g. circuits to proteins), we still encounter large gaps in our understanding before we reach the level of large neural circuits.These knowledge gaps are where a systems biology approach may be able to leverage large amounts of quantitative information to draw insights about neuroscience. As systems biology seeks to understand how biology is organized at multiple levels, this approach can help us make connections between the many levels and types of neuroscientific data. The advent of omics technologies such as genomics, transcriptomics, proteomics, and metabolomics (known as the "Big Four"; Dai and Shen, 2022) has transformed the landscape of biological research. In the 1990s, high-throughput DNA sequencing and mass spectroscopy introduced a new generation of quantitative datasets that could systematically capture genetic and/or molecular changes with high accuracy.In neuroscience, these omics technologies allow scientists to analyze and understand the brain at unprecedented levels of detail, enabling the exploration of genetic and molecular factors that underlie specific brain functions and dysfunctions. Genomics, for example, has provided insights into the genetic basis of neurodevelopmental disorders like autism and schizophrenia. By identifying specific genes and genetic variations associated with these conditions, researchers can develop targeted therapies and interventions. Transcriptomics, on the other hand, focuses on the study of gene expression patterns in the brain. This approach has been instrumental in understanding the molecular mechanisms involved in memory formation and cognitive processes. By comparing gene expression profiles between healthy and diseased brains, scientists can identify potential therapeutic targets for conditions like Alzheimer's disease. Spatial transcriptomics is a newer technique that combines single-cell RNA sequencing (scRNA-seq) information with spatial information so genomic data can be spatially resolved in intact tissue [Jung and Kim, 2023]. This technique has recently been used to reveal specific neuronal cell types and circuits within the prefrontal cortex that regulate chronic pain [Bhattacherjee 2023], and many more studies using spatial transcriptomics will provide potential bridges between individual genes and physiological functions.While the omics provide incredibly rich datasets, they do not necessarily provide us with a rich understanding of the interactions between the bits of data points. One aspect of these large-scale omics approaches is that they are hypothesis-free, which can be advantageous to avoid scientific biases. However, this unbiased approach could be a double-edged sword in that it may be difficult to interpret or understand the functional significance of some of these very large datasets. No matter what one thinks of big data approaches to solving biological problems, the power of these techniques is undeniable to identify potential biologicallyrelevant targets, whether the goal is to understand basic function or to treat a disease. By integrating data from a variety of omic technologies, neuroscientists can create sophisticated models that connect genetic and molecular information with the brain's structural and functional properties. This integrative approach has the potential to transform our understanding of brain health and disease. One specific omics approach that is particularly relevant to neuroscience is connectomics, which aims to comprehensively map the synaptic connections between individual neurons of a piece of neural tissue or within an entire nervous system of an organism. Connectomics displays the anatomic hard wiring that underlies information processing, and as such, provides an important ground truth for computational models and simulations.More sophisticated anatomical tract-tracing and improvements in neuroimaging will undoubtedly improve these models and the underlying hard wiring data, hopefully leading to an improved understanding of brain function. As certain psychiatric illnesses are thought to be manifestations of circuit dysfunction, improved validation of indirect measures [Chang et al., 2017] should improve connectomes [Haber et al., 2021]. Returning to the structure and function relationship, recent work using a multimodal and high-resolution approach in the mouse neocortex has illuminated important principals of how synaptic size relates to synaptic strength [Holler et al., 2021].The Human Connectome Project, for instance, utilizes advanced neuroimaging techniques to map the intricate connections within the brain. By capturing the structural and functional connectivity patterns between different brain regions, scientists can create detailed models of the brain's network architecture. This approach has led to groundbreaking discoveries about the brain's role in cognition, emotion, and neurological disorders.So far, the nematode roundworm Caenorhabdhitis elegans genome and connectome have been fully mapped, yet we do not have a full understanding of nematode physiological functions and behaviors. Structure-function models have been constructed of the network architecture of the nematode brain, with possible scalable features that may be universal across different phyla [Brittin et al., 2021].On a counterpoint to connectomics, some neuroscientists have posited that mapping the worm (or any other) connectome does not improve our understanding of how the worm's nervous system produces behavior. This is, at least in part, due to the fact that synaptic connections are not all equal in weight and that there are important differences between structurally similar synaptic connections that will not be visible through connectomics. In addition to contributing variables such as synaptic weights, there are a number of other factors that contribute to neural signaling but would not be captured by connectomic-based synaptic wiring diagrams. A recent study showed that "wireless" signaling through neuropeptides also takes place and have important effects on neural circuit function [Ripoll-Sanchez et al., 2023].For this reason and others, there are many who believe that these connectome efforts are mostly futile in advancing our understanding of brains. While connectomics has its critics, some organizational principles have emerged from mapping certain task-specific circuits, for example in Octopus vulgaris (Bidel et al., 2023). While many people may instinctively assume that the main goal of neuroscience is to understand the human brain, there is in fact much to learn from much simpler nervous systems. The sheer diversity of nervous systems found within biology tells us that there are many different neural solutions to the various environments in which different organisms thrive [Figure 2]. For example, many of the fundamental principles of learning and memory were discovered and demonstrated in Aplysia. Studies in bats, barn owls, and zebra finches were fundamental to our understanding of computational maps [Laurent, 2020]. Similarly, studies that focus on neural systems outside of the CNS will bring an improved understanding of physiology and function on the whole-organism level. Concerted efforts to study the peripheral and autonomic nervous systems in certain species are already underway [SPARC], and lab groups are beginning to reveal important organizing principles within the periphery [Huerta et al., 2023;Prescott and Liberles, 2022] The comparative approach is an important, but often neglected sub-field in neuroscience. Neuroethology capitlizes on this comparative approach that unveils general principles. For example, the original action potential work from Alan Hodgkin and Andrew Huxley [Nobel Prize 1963] was performed in the squid giant axon and revealed the general basic unit of functional communication used by neurons. Incredibly, action potentials in the squid are the same as those in a grasshopper, mouse, or human. Similarly, early work on the visual system was conducted on photoreceptors from one of the oldest animals on earth, the horsehoe crab Limulus Polyphemus. A recent connectome study revealed principles that underlie brain maturation in the roundworm Caenorhabdhitis elegans, principles that may also be relevant for neurodevelopment in higher order specieis [Witvliet et al., 2021].While preclinical research tends to be dominated by rodent brains, and the human brain remains the ultimate challenge, there is a widerange of relevant model systems that provide useful insights into disparate aspect of nervous system function. Recent work has shown that rats have the cognitive substrates for mental navigation and spatial imagination, which were mental capacities previously reserved for higher species [Lai Science 2023]. This indicates that while rodent brains have nowhere near the capacities of primate brains, they are capable of fairly high-level mental representations. Just as the prior decades of neuroscience discoveries were driven by technological innovations [Yuste, 2015], there is similarly a new era of advances tied to developments in modern computing. Technological advancements have been the driving force behind the progress in systems biology and integrative systems neuroscience. High-throughput sequencing, advanced neuroimaging techniques, and powerful computational tools have enabled the collection and analysis of vast amounts of data, propelling our understanding of the brain to new heights. Moreover, artificial intelligence and machine learning algorithms have played a pivotal role in making sense of the enormous datasets generated by omics approaches. These tools can identify complex patterns and relationships within the data, leading to new insights and discoveries that would be impossible to discern through traditional methods.Rapid advances in computing are producing impressively comprehensive datasets that are frankly unprecedented. A recent study mapped a single cubic millimiter of human cortical tissue at electron microscopic resolution (nanometer scale) that contained 57,000 cells and 150 million synapses that were individually segmented and labelled using machine-learning algorithms in 2D space to produce high resolution 3D rendering [Shapson-Coe 2024]. The resulting dataset was 1.4 petabytes in size and this was only a single cubic millimeter of human brain tissue. For comparison, a mouse brain has an estimated volume of 500 mm 3 while a human brain is more than 1200 mm 3 . We have now reached a point in modern neuroscience when we can test, with large sets of empirical data [Randi et al., 2023;Urai et al., 2022], many of the assumptions and speculations of our intellectual neuroscience forefathers. I believe that in the next decades, neuroscientists will leverage the many advances in computing to integrate the immense datasets that have become available in the past decade. For example, there is now a spatial atlas of the entire mouse CNS at molecular resolution [Shi et al., 2023], and it is now possible to conduct disease-relevant 3D spatial proteomics on whole mammalian specimens, such as intact mice [Bhatia et al., 2022]. As tissue labeling and clearing techniques improve alongside advances in volumetric microscopy, it will soon become possible to image larger intact samples, including intact human organs [Erturk, 2024;Mai et al., 2022]. These developments will enable additional studies to explore the ground truth of many noninvasive imaging measures used widely in neuroscience, particularly to examine the human brain. The Integrative Systems Neuroscience section in Frontiers in Systems Biology aims to be a journal destination for work that integrates findings across disciplines to better understand nervous system function. In the spirit of a multidisciplinary approach and a diversity of viewpoints, we aim to include data from different animal and non-animal models. The integration of multiple leading-edge techniques that I have discussed here, along with other advanced neuroscience techniques for manipulating and interrogating neural circuits including optogenetics, chemogenetics, and neuronal activity tracking will open new avenues to establish causal interactions between the micro and macro scales of neural systems.With access to modern datasets of unprecedented scale and detail, the Grand Challenge for this field lies in the ability to use these empirical datasets to test hypotheses about how nervous systems function. As high-throughput technologies have given us massive data troves of molecular, genetic, and neural data, the challenge is to extract useful knowledge and understanding from these to build foundational models about how nervous systems function.