Human liver and pancreas innervation: resolving 3D neurohistological challenges and advancing insights
- Review
- Open access (https://www.springernature.com/gp/open-science/about/the-fundamentals-of-open-access-and-open-research)
Published: 10 November 2025
Chih-Yuan Lee1,
Fu-Ting Hsiao2,3,
Chien-Chia Chen1 &
Shiue-Cheng Tang2,3
Journal of Biomedical Science (https://jbiomedsci.biomedcentral.com/) volume32, Article number:97 (2025) Cite this article
Abstract
The human liver and pancreas are central to metabolic regulation, with the autonomic nervous system orchestrating processes that maintain glucose homeostasis and respond to dynamic physiological demands—ranging from acute energy mobilization during stress to postprandial glucose uptake and storage. However, visualizing and examining the intricate three-dimensional (3D) neural networks within clinical liver and pancreas specimens remains challenging, as conventional two-dimensional (2D) histological methods cannot fully resolve the spatial complexity of autonomic innervation in the liver and pancreas. This review identifies and discusses key biological and technical factors—including tissue autofluorescence, autolysis, photobleaching, and steatosis—that compromise the reliability of 3D neurohistological analysis of the human liver and pancreas. We also highlight emerging optical and chemical methodologies that enable high- and super-resolution 3D tissue imaging, improving signal fidelity, preserving structural detail, and supporting consistent, reproducible analysis. Ultimately, these advances aim to facilitate precise mapping of human liver and pancreas innervation, offering deeper insight into the neural regulation of nutrient assimilation, glucose utilization, and metabolic homeostasis in both physiological and pathological contexts.
Background
A dynamic partnership in glycemic regulation
The liver and pancreas are the primary metabolic organs essential for maintaining glucose homeostasis, with their functions tightly regulated by the autonomic nervous system [1,2,3,4,5 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR5) ]. Physiologically, the endocrine pancreas controls blood glucose levels through the opposing actions of its key islet hormones, insulin and glucagon. The two hormones are delivered to hepatocytes via the portal venous system, linking pancreatic hormone secretion to hepatic glucose regulation. Insulin promotes hepatic glycogenesis, while glucagon stimulates glycogenolysis, together ensuring balanced glucose levels in response to hormonal and neural cues. For instance, marked hypoglycemia (blood glucose < 40mg/dl) has been shown to activate the sympathetic nervous system, prompting glucagon release from pancreatic α-cells [6 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR6) ]. This, in turn, stimulates hepatic glycogenolysis, effectively raising blood glucose levels to restore normoglycemia. In contrast, patients with unstable type 1 diabetes may exhibit impaired autonomic function [7 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR7) ], disrupting both pancreatic glucagon secretion and hepatic glycogenolysis. Such disruptions can lead to severe hypoglycemia during insulin therapy—a life-threatening complication in diabetes management [8 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR8) ].
Other physiological states that engage the autonomic nervous system include the “fight-or-flight” and “rest-and-digest” responses, predominantly mediated by the sympathetic and parasympathetic branches, respectively [9 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR9) ]. During acute stress, sympathetic activation triggers catecholamine release from the adrenal medulla, enhancing hepatic glucose production to meet emergency energy demands [10 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR10) ]. Conversely, in the postprandial state, parasympathetic activation and gut hormones stimulate insulin secretion from pancreatic β-cells [11 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR11) , 12 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR12) ], promoting hepatic glycogenesis and lowering blood glucose levels. These intricate neuroendocrine interactions between the nervous system, pancreas, and liver underscore the necessity for detailed anatomical characterization to elucidate hepatic and pancreatic innervation patterns in both health and disease. Such investigations are crucial for advancing our understanding of metabolic regulation and the coordinated control of the two organs. Clinically, a more precise understanding of this innervation could inform new strategies for neuromodulatory therapies aimed at restoring metabolic balance in diabetes, obesity, and fatty liver disease.
The essential role of clinical specimens in translational liver and pancreas research
Experimental animal models, particularly rodents, have been instrumental in elucidating the molecular mechanisms and cellular complexities of hepatic and pancreatic physiology, as well as their associated pathophysiology. However, fundamental differences in lifespan, nutritional requirements, and metabolic profiles between rodents and humans result in notable variations in tissue architecture, vascularization, and innervation. These species-specific differences must be carefully considered when translating experimental findings into clinically relevant contexts.
For instance, studies on hepatic innervation have highlighted substantial interspecies variability in neural distribution [13,14,15,16 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR16) ]. Unlike humans, rodents lack intra-lobular sympathetic nerves in the liver—a structural difference that limits their ability to accurately model certain human-specific neural and pathological features. In human non-alcoholic fatty liver disease (NAFLD), for example, ballooning hepatocytes within the intra-lobular domain are directly exposed to sympathetic axonal contacts (Fig.1 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#Fig1) and Supplementary Video 1), a morphological feature that rodent models fail to replicate adequately.
Human liver with steatosis: panoramic-to-Airyscan super-resolution neurohistology. A Steatotic lobules obtained from the distal resection margin of hepatocellular carcinoma (male, 55years old, right lobe). Sympathetic nerves are labeled with tyrosine hydroxylase (TH, green). The central and portal (P) regions at the top are magnified in panels B (projection) and C (2D image). Bile ducts are labeled with CK19 (magenta), nuclei with DAPI (white), and steatotic areas are visualized using transmitted light signals (gray). B, B’ Projection and 2D images showing lobular innervation and peri-central ballooning hepatocytes. C, C’ Sympathetic nerves and bile ducts in the portal field entering the lobule. D Enlarged view of the central field highlighting sympathetic innervation of ballooning hepatocytes. E, F 3D Airyscan imaging further magnifies the lobular microenvironment of ballooning hepatocytes. Sympathetic nerves in these areas are presented in E’ and F’ (projection views). Curved sympathetic nerve fibers closely follow the contours of ballooning cells in E’, suggesting direct contact and potential modulation. Note that the enlarged volumes of hepatocytes in E and E’ exhibit a local reduction in nerve density (compared with F and F’). Supplementary Video 1 provides a depth-resolved example of 3D Airyscan super-resolution imaging of the steatotic microenvironment
Full size image (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y/figures/1)
Similarly, differences in pancreatic fat deposition underscore the limitations of rodent models. Pancreatic steatosis—a condition characterized by fat infiltration into pancreatic tissue—is commonly observed in adult humans, and its severity is closely associated with metabolic disorders such as obesity and type 2 diabetes (Fig.2 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#Fig2) and Supplementary Fig.1) [17,18,19,20 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR20) ]. Although both humans and mice exhibit peri-lobular fat accumulation in pancreatic steatosis, intra-lobular fatty infiltration is rare in mice. This distinction is critical, as intra-lobular adipocytes in humans appear to remodel and integrate with the pancreatic parenchyma. Consequently, studies on pancreatic innervation in the contexts of obesity, diabetes, and pancreatic neoplasia must account for this remodeled microenvironment; ignoring the role of adipocytes may lead to an oversimplified understanding of disease pathology.
Human pancreas with steatosis (fatty infiltration): panoramic-to-Airyscan super-resolution neurohistology. A, A’ Vibratome section of a steatotic donor pancreas (male, 43years old, tail region). Arrows in A’ indicate infiltrated adipocytes located in both peri-lobular and intra-lobular spaces. B, C Projections of neuro-insular and neurovascular networks. A–C examine the same macroenvironment. Green: neuroendocrine marker PGP9.5; red: endothelial marker CD31. PGP9.5 staining highlights both nerves and islets. Asterisks in B and C indicate a PGP9.5⁺ ganglion (enlarged in the insets), further magnified in D–F. D–F In-depth Airyscan imaging of neurons within the ganglion. Depth-resolved images identify seven neurons: “a” in D, “b” and “c” in E, and “d–g” in F. PGP9.5⁺ neuronal somas are shown, each containing a nucleus with a distinct nucleolus. Green: PGP9.5; red: CD31; white: DAPI. Supplementary Fig.1 provides a depth-resolved example of a ganglion labeled with the glial marker S100B in the human pancreas with steatosis
Full size image (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y/figures/2)
Given the necessity of using human specimens in translational research to avoid misleading results, the following sections examine the challenges and strategies for applying 3D high- and super-resolution neurohistology to the analysis of human liver and pancreas tissues. The goal is to apply 3D imaging techniques capable of resolving structures across multiple scales—from macroscopic (centimeter-scale) views to sub-micrometer precision—enabling accurate mapping and characterization of innervation while minimizing artifacts that could generate false-positive or false-negative results.
False-positive signals: tissue autofluorescence
Autofluorescence presents a major challenge in fluorescence-based immunohistochemistry, particularly when visualizing innervation patterns in human liver tissues. The liver contains abundant endogenous fluorophores—including nicotinamide adenine dinucleotide (NADH) and flavin adenine dinucleotide (FAD), both key components of cellular respiration and metabolic pathways [21,22,23 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR23) ]; bilirubin, a bile pigment [24 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR24) ]; lipofuscin, an aging pigment derived from lipid peroxidation [25 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR25) , 26 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR26) ]; and porphyrins, which participate in heme synthesis [27 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR27) ]. All of these emit fluorescence across a broad spectrum. Their emissions frequently overlap with commonly used fluorescent dyes, complicating signal interpretation, generating false positives, and obscuring specific labeling in fluorescence imaging.
NADH emits blue fluorescence (excitation: 320–380nm, emission: 420–460nm), which can interfere with signals from DAPI, a commonly used nuclear stain. Similarly, the green-yellow fluorescence of FAD (excitation: 450–500nm; emission: 520–550nm) and bilirubin (excitation: 460–490nm; emission: 520–530nm) overlaps with fluorophores such as Alexa Fluor 488, which is commonly used for green fluorescence labeling. Additionally, lipofuscin (excitation: 360–450nm, emission: 500–700nm) and porphyrins (excitation: ~ 400nm, emission: 600–700nm) emit red fluorescence, which can interfere with dyes like Alexa Fluor 594 and Texas Red. This spectral overlap complicates the identification of neuronal markers, especially in liver cells affected by oxidative stress or aging, where lipofuscin and porphyrins accumulate. The resulting background fluorescence lowers the signal-to-noise ratio, making it challenging to accurately visualize nerve fibers, varicosities, and vesicles in fluorescence-based 3D immunohistochemistry.
Beyond endogenous fluorophores, blood-derived components also contribute to autofluorescence in liver biopsy specimens due to the liver’s extensive vascularization [28 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR28) , 29 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR29) ]. Residual blood components exhibit strong intrinsic fluorescence across the visible spectrum, further interfering with imaging. Similarly, pancreatic biopsies are affected by blood-derived autofluorescence [30 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR30) ], which persists even in donor pancreases that have been perfused before dissection [31 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR31) ]. These artifacts vary across clinical specimens, necessitating careful interpretation—particularly when dark regions appear in transmitted light microscopy of optically cleared 3D specimens [31 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR31) , 32 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR32) ].
Strategies for autofluorescence reduction
Several chemical and optical methods have been developed to suppress autofluorescence in fixed specimens. Chemical quenchers—such as TrueBlack and Sudan Black B [33,34,35,36 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR36) ]—are applied post-staining to improve visualization of specific labeling targets. TrueBlack reduces lipofuscin autofluorescence, while Sudan Black B is beneficial for specimens with high lipid content. However, careful optimization is essential because excessive use of quenchers can also reduce the intensity of desired fluorescent signals, leading to false-negative results.
Chemical bleaching offers an additional strategy for suppressing autofluorescence in fixed tissues [37,38,39 (https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-025-01194-y#ref-CR
