Resilient Calvarial Bone Marrow Supports Retinal Repair in Type 2 Diabetes

Abstract

Using micro-computed tomography, we identified a network of skull channels in the calvarium of type 2 diabetic (T2D) mice that remained structurally intact and numerically stable despite long-standing disease. The retention of calvaria bone marrow structural integrity was associated with preserved hematopoietic capacity under chronic diabetic conditions, which was not observed in the bone marrow of long bones. A distinctive feature of the calvarial bone marrow compartment was its direct exposure to cerebrospinal fluid (CSF), a property not shared by tibial bone marrow. To characterize the biochemical environment of the murine calvarium, we profiled oxysterols in CSF using mass spectrometry. The CSF exhibited elevated levels of neurotrophic and anti-inflammatory oxysterols, including 22-hydroxycholesterol (22-OHC) and 27-hydroxycholesterol (27-OHC). To assess whether this protective oxysterol signature was conserved in humans, we analyzed CSF samples from diabetic and non-diabetic individuals with obesity-associated idiopathic intracranial hypertension (IIH). Human CSF contained 7α-hydroxy-3-oxo-4-cholestenoic acid (7-HOCA), a metabolite of 27-OHC, supporting the conservation of this neuroprotective profile across species. Given the anatomical proximity of the calvarium to the eye, we hypothesized that calvaria bone marrow may serve as a reservoir for immune cells recruited to the injured or infected retina. The calvaria bone marrow was the predominant source of myeloid angiogenic cells (MACs) and neutrophils, mobilizing these cells at levels approximately 20-fold higher than long bones. These findings demonstrate that calvarial bone marrow plays a critical role in retinal immune defense, while maintaining both structural integrity and functional capacity despite chronic T2D.

Keywords: calvarium marrow; ischemia‐reperfusion; long bones; myeloid angiogenic cells; neutrophils; retina; skull; stem/progenitor cells.

FIGURE 1

Ex vivo micro‐CT and immunofluorescence imaging of mouse calvarium. (A, B) Micro‐CT volume renderings of the calvarium from (A) WT and (B) db/db mice, highlighting three regions of interest: frontal, parietal, and occipital. (C) High‐resolution rendering of the frontal region (voxel size: 2 µm), with skull channels outlined in red. (D) Surface mesh of the same region showing bone marrow cavities (blue) and skull channels (red). (E) Immunofluorescence microscopy of skull channels stained with collagen IV, highlighted by yellow dashed lines. n = 3 mice per group.

FIGURE 2

Micro‐CT analysis of femoral bone structure in WT and db/db mice. (A, B) Regions of interest in the metaphysis of femurs from WT and age‐matched db/db mice selected for micro‐CT imaging. (C) Cross‐sectional images of trabecular bone in WT mice (n = 3). (D) Corresponding cross‐sections in db/db mice (n = 3) showing markedly reduced trabecular bone. (E–G) Quantitative analysis of bone parameters: bone volume fraction (E), trabecular thickness (F), and hyaluronic acid content (G) in WT and diabetic mice. n = 3 mice per group.

FIGURE 3

Flow cytometric analysis of hematopoietic stem/progenitor cells (HSPCs) in the calvarial and long bone marrow of db/db and control mice. (A) Representative flow cytometry plots illustrating the gating strategy used to identify Lin⁻/Sca‐1⁻/c‐Kit⁺ (LS‐K) and Lin⁻/Sca‐1⁺/c‐Kit⁺ (LSK) populations. LS‐K and LSK cells were combined and analyzed as HSPCs. (B, C) Bar graphs quantifying total cellularity in the BM of calvaria and long bones, respectively. (D, E) Bar graphs showing the relative proportions of HSPCs in calvarium vs. long bone marrow. (F, G) Representative images of colony‐forming units derived from calvarium and long bone marrow cells. (H, I) Quantification of colony cellularity. At 12 days post‐seeding, single‐cell suspensions from harvested colonies revealed a significant increase in cellularity in colonies derived from db/db calvaria compared to controls, whereas no significant difference was observed in colonies from long bones. n = 3–7 mice per group.

FIGURE 4

Calvarium and long bones demonstrate unique site specific differences in BM composition. (A) Representative images of calvaria and tibia from WT and db/db mice. In db/db mice, the calvarium displays increased red marrow and reduced fat content (black arrows), whereas the femur shows decreased red marrow and increased fat accumulation compared to WT controls (yellow arrows). (B–E) Flow cytometric analysis of hematopoietic and erythroid populations in calvarium and long bone marrow. (B) Representative flow plots and (C) quantification of CD45⁺ cells from calvaria and long bones of WT and db/db mice. (D) Representative flow plots and (E) quantification of erythroid lineage cells, including erythroid precursors (Ery Prec; CD71⁺TER119⁺) and mature erythrocytes (Mat Ery; CD71⁻TER119⁺), from calvaria and long bones of WT and db/db mice. n = 3–7 mice per group.

FIGURE 5

Lipid/Fat content and changes in the vasculature of the calvarium and tibial compartments. (A, B) Lipid accumulation in bone marrow compartments. Representative images (A) and mean fluorescence intensity (B) quantification of Nile Red staining in calvaria and long bones from WT and db/db mice, indicating differences in lipid content between genotypes and marrow sites. (C, D) Vascular structure assessment via Collagen IV staining. (C) Representative images and (D) mean fluorescence intensity quantification of Collagen IV staining in calvaria and long bones of WT and db/db mice, reflecting changes in vascularity across bone marrow niches. n = 3–7 mice per group.

FIGURE 6

High‐resolution/accurate mass LC‐nESI‐MS analysis of oxysterols in CSF and BM compartments of calvaria and long bones. (A–D) Base peak chromatograms showing masses corresponding to oxysterol ions ([M–H₂O]⁺ and [M–2H₂O]⁺) obtained via reverse‐phase liquid chromatography for 22‐hydroxycholesterol (A, C) and 27‐hydroxycholesterol (B, D) in WT and db/db mice. (E) Zoomed view of the 27‐hydroxycholesterol peak in CSF from control and diabetic mice. The inset displays the LTQ‐Orbitrap Velos mass spectrum at the corresponding retention time, highlighting mass‐to‐charge (m/z) values for 27‐hydroxycholesterol ions in WT and db/db CSF samples. n = 3–7 mice per group.

FIGURE 7

In vivo BM cell labelling by photoconverting calvaria and tibia marrow of KIKGR mice for cell tracking. (A) Schematic of the experimental design for tracking bone marrow (BM) cell recruitment to the injured retina. KikGR mice underwent photoconversion of either the tibial or calvarial BM compartments via targeted laser exposure. Retinal ischemia‐reperfusion (I/R) injury was induced immediately after, and tissues were harvested for flow cytometry 6 h post‐injury. (B, C) Representative flow cytometry plots showing proportions of live photoconverted (PE⁺, red) CD45⁺ cells in the tibia, calvarium, and peripheral blood of uninjured control mice before (B) and immediately after (C) photoconversion. (D, E) Quantification of BM‐derived cell recruitment to the retina following I/R injury. Bar graphs show the normalized relative contributions of photoconverted BM cells from the calvarium (D) and tibia (E) to the retina as monocytes, neutrophils, and MACs. (F) Retinal flat mount from a KikGR mouse with calvarium photoconversion followed by I/R injury, showing red photoconverted cells recruited to the retina. (G) Retinal flat mount from a KikGR mouse with tibia photoconversion followed by I/R injury, showing fewer photoconverted cells (yellow/orange), indicating partial conversion and reduced recruitment. (H) Retinal flat mount from the contralateral eye of a KikGR mouse with tibia photoconversion but no I/R injury, showing absence of photoconverted cells confirming injury‐dependent recruitment seen in f and g. n = 3 mice per group; data representative of two independent experiments.