Visualizing 3D neuroskeletal interactions in the skull

Peripheral nerves: an understudied mediator of bone formation

The primary function of the bony skeleton is to provide structural support and protection through its dense, mineralized matrix. Bone has a heterogenous microenvironment containing both bone-forming and non-bone-forming cell types that communicate and respond to environmental stimuli, controlling bone formation during physiologic processes such as development, injury, and aging. Osteocytes, the embedded cells within the bone matrix, have cytoplasmic projections that act as the primary “sensors” within the bone microenvironment. They sense changes in pressure and flow of interstitial fluid and respond by signaling nearby cells to control bone formation and remodeling.  However, peripheral sensory nerves that innervate bone also play roles in sensing environmental stimuli. Previous studies, which date back to 1900, have identified known roles for nerves in the bone remodeling process.¹ In addition to traditional sensing mechanisms, where peripheral nerves send afferent signals to the central nervous system, sensory nerves also have the capacity to secrete signaling factors within the tissues they innervate. Nerve-derived signaling factors, such as calcitonin-gene related peptide and fibroblast growth factors, have been implicated in mediating cell proliferation and differentiation.^2,3 A majority of these studies have been performed in long bones; however, little is known about the role of innervation in the skull, which exhibits unique bone formation mechanisms and has reduced reliance on mechanical integrity. As the field of neuroskeletal signaling continues to investigate these mechanisms, it is increasingly important to develop methods to generate clear and effective 3D visualizations of the innervated bone microenvironment.

In long bones, nerves densely innervate the periosteum and travel into the bone marrow regions through transcortical canals.⁴ Large nerve bundles also travel into the bone marrow through the intercondylar foramen.⁵ While nerves innervate periosteal regions, few nerves are found along endosteal regions of long bones.⁶ Although much is known about the innervation patterns in long bones, craniofacial peripheral nerves, including those in the calvaria, remain vastly understudied.

Our lab is particularly interested in calvarial imaging as our goal is to develop tissue engineered therapies for craniofacial bone injuries. We and others have identified a dense ingrowth of nerves into calvarial defect regions immediately after injury. In long bone, it is proposed that this ingrowth occurs prior to blood vessels and may play roles in encouraging angiogenesis into the defect region.⁷ Therefore, targeting neural ingrowth after injury may function more upstream than angiogenesis and has potential for further improved healing. Before we can develop therapies to target nerves in the calvaria, it is important to understand their distribution and changes with age during homeostasis. Thus, in this study, we investigated peripheral nerve distributions using our QLSM platform throughout the full murine lifespan.

Challenges with 3D skeletal nerve visualization

Traditional methods to visualize cells in bone struggle to reconstruct the dense 3D bone microenvironment. Due to calcified bone tissue, samples are generally decalcified prior to both 2D and 3D imaging methods. Traditional histological sections require additional processing, which has potential to distort the tissue through dehydration or freezing steps. Further, structures such as nerves and blood vessels are difficult to visualize in 2D, due to their tube-like structure. Therefore, our lab has developed Quantitative Light Sheet Microscopy (QLSM), which allows comprehensive 3D imaging of the intact murine calvaria without decalcification, thus preserving the spatial associations.⁸ Our platform enables imaging in three channels with single cellular resolution. In particular, the calvaria presents a unique opportunity for calcified tissue imaging, due to its thin cortical bone regions that surround the inner bone marrow. Thus, we are able to fully clear the calvaria exclusively using refractive index matching with 2,2-thiodiethanol (TDE) (Figure 1A).

Using our QLSM platform, we were able to effectively visualize nerves in all ages studied (Post-natal Day 0 to 80 weeks old) (Video 1). To quantify these nerves, we initially used our previously published Imaris workflow, which utilized threshold-based intensity segmentation. While this method worked for blood vessels, we found difficulty in selecting a threshold that adequately segmented the nerves throughout all ages. In particular, middle-aged and old samples had increasingly high background and were being over-segmented with the threshold that cleanly segmented nerves at young ages. This over-segmentation resulted in inconsistent data where qualitative observations were not matching quantitative results (Figure 1B-D). To combat this challenge, we integrated the Ilastik machine learning-based segmentation into our Imaris workflow. We trained the Ilastik pixel classifier with samples from each age group, allowing the classifier to clearly identify nerve structures and reduce background segmentation. The Ilastik classifier output resulted in image stacks with binary segmentations of either “nerves” or “non-nerves.” Using ImageJ, we were then able to integrate these new binary segmentations into our original Imaris files and continue with our previously defined analysis workflow. Overall, the integration of Ilastik segmentation of calvarial nerves into our standard analysis workflow provides an opportunity to expand this workflow to other neuroskeletal investigations. Given the complexity of quantifying nerves in bone, this machine learning workflow has potential to be applied to both 2D and 3D imaging contexts in the future.

Takeaways and Future Outlook

Thanks to our improved Ilastik-Imaris segmentation workflow, we were able to effectively segment and quantify nerves (Figure 1E) and their interactions with blood vessel phenotypes. We identified key changes in nerve distributions throughout aging that were unique to both the parietal and frontal bones. Due to our full calvaria 3D imaging, these insights were uniquely feasible with our platform. In addition, due to our ability to visualize three channels of signal with QLSM, we identified unique neurovascular associations with Type H blood vessels. These associations were maintained throughout aging, suggesting the importance of neurovascular signaling in bone homeostasis and remodeling. Moving forward, these unique observations provide insights to inspire key questions regarding bone formation and remodeling:

1. How does neuro-skeletal signaling change as the nerve density in the skull increases into adulthood then decreases with aging?
2. Does age-related nerve loss contribute to bone fragility with aging?
3. Does the continued expansion of unmyelinated nerves during postnatal development lead to skeletal maturation?
4. Does nerve-to-vessel signaling control Type H vessel-mediated bone formation?
5. Can nerves be targeted early in healing to encourage angiogenesis and, ultimately, bone formation?

We hope that our QLSM platform and investigation into homeostatic nerve distributions throughout the murine lifespan help to fuel these questions and motivate future investigations that result in effective therapies for bone regeneration. 

References

1. Hassan, M. G., Horenberg, A. L., Coler-Reilly, A., Grayson, W. L. & Scheller, E. L. Role of the peripheral nervous system in skeletal development and regeneration: controversies and clinical implications. Curr. Osteoporos. Rep. 21, 503–518 (2023).
2. Pei, F. et al. Sensory nerve niche regulates mesenchymal stem cell homeostasis via FGF/mTOR/autophagy axis. Nat. Commun. 14, 344 (2023).
3. Wee, N. K. Y. et al. Inhibition of CGRP signaling impairs fracture healing in mice. J. Orthop. Res. 41, 1228–1239 (2023).
4. Fujita, S. et al. Quantitative analysis of sympathetic and nociceptive innervation across bone marrow regions in mice. Exp. Hematol. 112–113, 44-59.e6 (2022).
5. Matsuo, K. et al. Innervation of the tibial epiphysis through the intercondylar foramen. Bone 120, 297–304 (2019).
6. Bataille, C. et al. Different sympathetic pathways control the metabolism of distinct bone envelopes. Bone 50, 1162–1172 (2012).
7. Li, J., Ahmad, T., Spetea, M., Ahmed, M. & Kreicbergs, A. Bone reinnervation after fracture: a study in the rat. J. Bone Miner. Res. 16, 1505–1510 (2001).
8. Rindone, A. N. et al. Quantitative 3D imaging of the cranial microvascular environment at single-cell resolution. Nat. Commun. 12, 6219 (2021).