Ingestion of gastrolith mineralized matrix increases bone volume and tissue volume in mouse long bone fracture model

Abstract

Purpose

Fracture healing often requires extended convalescence as the bony fragments consolidate into restored viable tissue for load-bearing. Development of interventions to improve healing remains a priority for orthopaedic research. The goal of this study was to evaluate the ability of a naturally occurring matrix of amorphous calcium carbonate to affect fracture healing in an uninstrumented long bone model.

Methods

Complete transverse fracture was induced in the fibula of mature mice, followed by daily gavage of crushed gastrolith from crayfish at doses of 0 (control), 1 (1 MG), and 5 (5 MG) mg/kg. At Day 17, bones and sera were harvested.

Results

Morphologically, the 1 MG treated group had greater bone volume (BV), and both 1 MG and 5 MG had greater tissue volume (TV) than control (p < 0.05), as determined by μCT; BV/TV and mineral density did not yield a statistical difference. Histologically, regional variations in mineralized matrix were evident in all specimens, indicating a broad continuum of healing within the callus. Among serum proteins, bone-specific alkaline phosphatase, indicative of active mineralization, was greater in 5 MG than control (p < 0.05). Sclerostin, an inhibitor of osteogenesis, was lower in 5 MG than control (p < 0.05), also suggestive of enhanced healing.

Conclusions

An increase in bone volume, tissue volume and cellular signaling for osteogenesis at 17 days following fibula fracture in this mouse model suggests that gastrolith treatment holds potential for improving fracture healing. Further study at subsequent time points is warranted to determine the extent to which the increase in callus size with gastrolith treatment may accelerate restoration of tissue integrity.

1. Introduction

An estimated 6,000,000 fractures occur each year in the US, and 5–10% show delayed or impaired healing.¹ It has been hypothesized that additional dietary calcium during the fracture healing process may help prevent impaired healing and accelerate bone repair. Calcium is fundamental to a wide range of cellular and organ functions, particularly in muscle and bone. Studies of calcium supplementation for treating bone fracture, however, are surprisingly limited and report inconsistent results. An early preclinical study² suggested that excess calcium in the diet does not improve fracture healing. However, the use of modern methods to determine rate of remodeling and bone quality in more recent studies challenges that finding for certain populations.³ For instance, in ovariectomized (OVX) rats, simulating postmenopause, additional calcium in the diet during fracture healing was found to result in improved radiological outcomes, but with no change in biomechanical properties.¹ Conversely, restricting dietary calcium in this model was found to reduce bone mineral density (BMD) at late stages of healing but had little effect before that.⁴ This supports the postulation that calcium is adequately supplied by the bones in early healing, but becomes more dependent on diet later on.

Recently, interest has developed in a dynamic, natural source of calcium and osteologic nutrients found in the gastrolith (“stomach stones”) of some crustaceans. Crayfish, Cherax quadricarinatus in particular, produce a transient matrix of amorphous calcium carbonate (ACC), like that found in coral, which constitutes a ready source of calcium to rapidly renew the molted exoskeleton with a neogenic, mineralized chitinous shell. As such, the gastrolith is inherently unstable, developing in onion-like layers by the fusion of approximately 50-nm spherules,⁵ and is considered a precursor polymorph to crystalline calcium carbonate (CCC) and even calcium phosphate. By limiting crystal growth within the gastrolith compound, ACC increases calcium solubility and bioavailability for osteogenesis,⁶ possibly 100-fold more than CCC. A preclinical radiotracer study indicated this results in about a 30% increase in absorption and retention versus CCC⁶, while a more recent clinical study confirmed that fractional calcium absorption in ACC therapy was about twice that in CCC therapy.⁷

Beyond bioavailability of calcium, research into gastroliths is actively identifying the integrated organic constituents that may promote osteogenic signaling and provide additional scaffolding, especially dual-acting proteins such as GAP65,⁸ Cq-M15,⁹ and (KRMP)-3.¹⁰ These specialized factors appear to possess domains that control both binding to the chitin backbone and precipitation of calcium. In the current study, the objective was to evaluate the ability of unmodified gastrolith matrix to enhance fracture healing in long bones as an oral supplement. A single time point of healing representing the approximate mid-point of fracture repair was selected for proof of concept and to determine optimal parameters for wider analysis.

2. Materials and methods

2.1. Study design

With IACUC approval (No. 358678, Eisenhower Army Medical Center, following guidelines of NIH Publications No. 8023), fibula fracture surgery was performed under isoflurane anaesthesia on the right leg of 12-week old, male CD1 mice from Charles River Laboratories. Fibula fracture has been used in rodents as a model to focus on the biological process of healing, since its limited role in weight-bearing plus restricted symphytic-joint motion proximally and fusion with the tibia distally provides a relatively stable fracture without instrumentation.^11,12 The three experimental groups of n = 5 each all received one-sided fibula fracture. Treatments included: 1) no intervention control (CTRL); 2) 1 mg/kg crushed gastrolith powder from locally-resourced Cherax quadricarinatus crayfish, administered as a suspension by daily gavage (1 MG); and 3) 5 mg/kg gastrolith by same method (5 MG). Based on calcium content, these doses of gastrolith are below those normally prescribed in therapeutic supplemental calcium treatment, as the intention of the study was to determine a threshold for gastrolith ingestion that could influence osteogenesis and fracture healing while minimizing potential risks.

2.2. Surgery

The right leg was shaved from ankle to hip and sterilized with betadine followed by 70% ethanol. Through a sterile drape opening, a 5-mm longitudinal incision was made posterolaterally to expose the calf muscle complex, as previously described.¹³ Blunt dissection with miniature forceps was performed to probe through the muscle planes to mid-diaphysis. Tenotomy scissors were then introduced and a single cut was made to induce a complete transverse fracture, with the incision sealed using a drop of 3M Vetbond^TM glue. After an IP injection of analgesic buprenorphine, the mouse was placed in a heated recovery cage and within 1 h returned in an ambulating state to its original cage with its four cagemates. Buprenorphine injections were continued daily for the first two postoperative days. Healing proceeded for 17 days following fracture, based on earlier experience in this model indicating that 21 days of recovery somewhat exceeds the midpoint of conversion in the callus from cartilage to bone. Voluntary movement was encouraged during healing with enrichment devices and feeding with standard chow and water was allowed ad libitum. On day 17 the mice were euthanized by CO₂ immersion and secondary exsanguination incident to serum collection by cardiac puncture. Tissue and serum were stored at −80 °C prior to analysis.

2.3. Micro-Computed Tomography (μCT)

Bone density scanning was performed on the operated and contralateral fibula of each specimen in a μCT40 device (Scanco Medical AG, Bruettisellen, Switzerland) with isotropic 6 μm resolution, using 55 kV energy, 300 ms integration time, and phantom-based calibration plus verification by ashed bovine cortical bone and chalk references. Automated analysis was made of standard mineral density and morphologic parameters in the callus region within a 1.5 x 1.5 x 2.5-mm-long volume, as well as the distribution of mineral density throughout the callus, referred to as bone density spectrum (BDS),¹⁴ and its integration, which yielded bone mass. Threshold for qualifying as bone was 35% of maximum attentuation.¹⁵ To isolate the healing callus from the original cortical bone, the BDS of the intact contralateral fibula was subtracted from that of the operated leg for each density value as an approximation of the extant original bone, which had undergone only nominal remodeling in the 17-day timecourse of healing. Reconstructed volumes and virtually sectioned surfaces for visualizing callus morphology were produced using manufacturer's software (v6.3-4). In addition, the transverse slice corresponding visually to the widest point of the callus was isolated to represent maximum cross-sectional area.

2.4. Histology

Due to the continuum of variable healing found across the callus, the objective of the histological analysis was to discern qualitatively the non-mineralized from the mineralizing fractions within the woven bone architecture. Selective staining allows direct visualization of the differential rate of osteogenesis throughout the callus as it progresses from an initial amorphous, twisting scaffold of collagenous proteins and supporting compounds into mineralized tissue through secondary healing. Thus, for these non-cellular purposes, the fibula specimens previously stored frozen prior to μCT scanning were decalcified for 10 days using a 2.3:1 mixture of 10% EDTA and 4% PBS, then embedded in paraffin, sectioned, and mounted onto charged slides. Images were made from the conventional stain of safranin-O with a fast-green backstain, along with the less common stain of alcian blue orange G (ABOG), which provides an especially clear delineation of the cartilage matrix as it mineralizes progressively into bone.

2.5. Serum proteins and calcium

Serum was decanted from blood that had been centrifuged at 2500 g for 10 min following cardiac puncture, then aliquoted and stored at −80C for subsequent protein analysis by ELISA kits and determination of free calcium (kit ab102505, Abcam, Cambridge, MA). Four markers of bone turnover and osteogenesis were evaluated. Bone-specific alkaline phosphatase (bALP; kit MBS703336, MyBioSource, San Diego, CA), a dephosphorylating hydrolase enzyme, indicates mineralization activity by osteoblasts. Pyridinoline crosslinks (PyD; kit 8019, Quidel Corp, San Diego, CA) are a byproduct measure of collagen breakdown during bone turnover. Osteoprotegerin (OPG; kit ab100749, Abcam, Cambridge, MA) is a decoy receptor that binds to RANKL to regulate its activity and thus inhibits signaling for osteoclastogenesis from osteoblasts and osteocytes. Sclerostin (SOST (its gene); kit MSST00, R&D Systems, Minneapolis, MN) inhibits osteoblast activity through blockage of the Wnt signaling pathway and is largely produced in osteocytes. Biomarkers were assessed in duplicate for all specimens, following manufacturer's instruction.

2.6. Statistical design

For each of the parameters spanning bone density, bone morphology, histomorphometry, and serum proteins, analysis of variance (ANOVA) was used to detect whether a difference existed among the three experimental groups (SPSS, IBM). Post-hoc pairwise testing in the event of a positive F-test relied on the Newman-Keuls method. For outcomes associated with non-normal distribution of data, the non-parametric equivalents of Kruskal-Wallis and Mann-Whitney tests were employed. Level of significance was set to α = 0.05.

3. Results

All mice survived the 17-day healing period and appeared fully ambulatory within 2 h of surgery. One callus, in the control group, depicted a non-union although its neighboring bone density properties were within the range of the other controls.

3.1. μCT

Regenerative patterns in the microarchitecture of the healing bone yielded varying results for each group with visually greater callus size typically found in the treated groups (Fig. 1). Bone volume (BV) in the callus was statistically greater in the 1 MG treatment group (0.836 ± 0.174 mm³ (mean ± standard deviation); coefficient of variation, CV = 21%; Fig. 2) than in controls (0.532 ± 0.065 mm³, CV = 12%; p = 0.027) and similar to 5 MG (0.837 ± 0.201 mm³, CV = 12%). The apparent augmented size of the repair callus with gastrolith treatment was confirmed statistically by tissue volume (TV), which was greatest in 5 MG (1.306 ± 0.301 mm³, CV = 23%), somewhat less in 1 MG (1.262 ± 0.212 mm³, CV = 17%), and least in control (0.707 ± 0.114 mm³, CV = 16%; p = 0.042, p = 0.044, respectively). Due to TV outpacing BV with gastrolith treatment, the ratio BV/TV trendwise followed the reverse order, with the lowest value in 5 MG (0.636 ± 0.027, CV = 4%), next lowest in 1 MG (0.647 ± 0.062, CV = 10%), and highest value in control (0.769 ± 0.062, CV = 8%).

Fig. 1.

a) Surface reconstruction by scanner software. Controls (top row; 0 mg/kg) in most cases appeared to evince smaller callus volume than did the treatment groups, while also showing in two specimens indications of either incomplete or nonunion. 1 MG specimens (middle row) showed typically greater callus volume than controls and also a specimen with apparent incomplete union. 5 MG specimens (bottom row) produced, on average, more callus volume than 1 MG and represented most consistency in bridging of fracture. b) Sagittal section reconstruction by scanner software. Visualizations confirmed progress of healing while suggesting lowest fractal complexity for controls and, most often, highest for 5 MG.

Fig. 2.

Micro-CT bone density and morphology (normalized to control with standard deviations shown). a) Bone volume (BV) increased with 1 mg/kg gastrolith treatment (differing symbols indicate statistical difference at p < 0.05) versus control. b) Tissue volume (TV) followed a similar pattern, in this case including 5 mg/kg c) BV/TV in treatment groups was trendwise lower than control, indicating that mean increase in TV was greater than mean increase in BV. d) Bone volume density (BVD) was virtually identical among the groups. e) Tissue volume density (TVD), resulting in no significant differences, suggested that the greater change in tissue volume than bone volume defined the statistical outcome. f) Bone mass differences between the treatment groups and controls were considerable but not significant with this effect size.

Within the region qualifying as bone volume, the density of hydroxyapatite, BVD, was very similar among the three experimental groups, and was highly consistent in the two treatment groups (1063 ± 111 μg/mm³ for control, CV = 24%; 1070 ± 25 μg/mm³ for 1 MG, CV = 2%; and 1060 ± 12 μg/mm³ for 5 MG, CV = 1%; Fig. 2). In parallel to the morphological finding of greater TV with gastrolith treatment, mineral density in the tissue volume of the callus (TVD) was trendwise lower in the treatment groups (681 ± 65 μg/mm³ for 1 MG, CV = 9%; 674 ± 32 μg/mm³ for 5 MG, CV = 5%) than in control (829 ± 131 μg/mm³, CV = 16%). Bone mass showed a substantial trend toward an increase with treatment in both groups, but did not achieve significance due to variability in each group (581 ± 108 μg for 1 MG, CV = 19%; 895 ± 203 μg for 1 MG, CV = 23%; 893 ± 223 μg for 5 MG, CV = 25%). A significant difference morphologically also resulted in measurement of widest diameter, where 1 MG (1.261 ± 0.165 mm, CV = 13%) and 5 MG (1.373 ± 0.229 mm, CV = 17%) were greater than control (0.981 ± 0.165 mm, CV = 17%; p = 0.027, 0.016, respectively).

Differences in mineral density characteristics between the groups could be appreciated alternatively by examination of the distribution of densities through the callus volume (Fig. 3). For comparison, the amount of bone of the intact contralateral fibulas was seen to remain low in the lower-density end of the bone density spectrum (BDS) curve to the left, reflective of its intact diaphyseal properties, before elevating as it approached a cortical bone peak close to 180 g/cm³, then decreasing again toward zero as the density abscissa continues. Fig. 3 makes readily evident the contrasting properties between the two treatment groups, which appeared very similar in spectrum, and control, where there was considerably less low-density tissue, slightly more medium-density tissue than the treatment groups beginning about 135 g/cm³, but then somewhat less high density bone as the curve passed through the cortical margin of the others at about 175 g/cm³. The peak for 5 MG and controls occurred at the same density, about 0.15 g/cm³ greater than that of the contralateral peak. By comparison, the maximum in the high-density range for 1 MG was more coincident with the contralateral peak and, given only the nominal representation of total volume in this region, suggested possibly more, if less dense, cortical bone development than in 5 MG and controls.

Fig. 3.

Bone density spectrum. Treatment groups both showed large voxel counts toward the less dense end of the spectrum, indicating a voluminous mix of cartilage and neogenic bone, while controls (0 mg/kg) showed more even distribution, peaking at about 1.35 g/cm³. For reference, contralaterals displayed a nominal presence of low-density tissue, beginning the ascent to greater proportionality for the higher density spectra approximately coincident with the peak of controls. High density peak of 1 MG specimens occurred in sync with that of contralaterals, at about 1.75 g/cm³ 5 MG specimens showed a muted peak at slightly higher density, as did controls to a lesser extent.

3.2. Histology

Mineralization advanced as a continuum of varying degree within the callus of each specimen, resulting in no generalizable difference between groups based on this analysis (Fig. 4). H&E stain highlighted areas rich in proteoglycan content normally associated with hyaline cartilage, which remained concentrated toward the fracture ends. Safranin-O delineated the remodeling cartilaginous core of the trabecular architecture from appositional intramembranous growth occurring at the margins of the woven bone. This distinction was further contrasted in the ABOG stain, which confirmed a wide range of callus maturation throughout the appositional bone on the trabecular surfaces.

Fig. 4.

Histology. a) Specimens of all groups showed varying amounts of a cartilaginous intermediary interposed between fractured fibula ends and a bridging network of woven bone, as indicated by safranin-O staining (yellow arrow; 4x original magnification; F = fibula). b) A functional midpoint of endochondral ossification also was found to be similar among the groups in the trabecular repair network joining the fractured ends. This is shown by H&E staining wherein purplish hematoxylin binds the highly-charged proteoglycans in the cartilaginous core of the trabeculae (yellow arrow) while pinkish eosin binds the amino acid-rich matrix of the appositional bone being deposited on the core. (Green arrow highlights proliferative area of hyaline cartilage; 20x original magnification.) c) Visual differentiation of the cartilaginous core (yellow arrow) as it remodels into mineralized matrix is especially evident with alcian blue orange G (ABOG) stain. This is observed to occur both in the pericellular space of the chondrocytes in the trabeculae, as well as longitudinally along the cartilaginous core (5 MG exemplar, e.g., indicated by green arrow; 40x original magnification). Heterogeneity in the maturation stage of both mineralization processes from one location to another for most specimens resulted in no significant group differences based on these images.

3.3. Serum proteins and calcium

Two serum proteins showed a significant difference among the experimental groups (Fig. 5). bALP, indicative of active mineralization, was statistically higher in 5 MG (40.3 ± 5.6 ng/ml, CV = 14%) than control (32.5 ± 1.3 ng/ml, CV = 4%; p = 0.032) and somewhat higher in 5 MG than 1 MG (35.3 ± 4.2 ng/ml, CV = 12%). Conversely, SOST, related to suppression of osteogenesis, was lower in 5 MG (201 ± 71 pg/ml, CV = 27%) than control (317 ± 35 pg/ml, CV = 11%; p = 0.014) and somewhat lower in 5 MG than 1 MG (283 ± 87 pg/ml, CV = 31%). 1 MG was not different than control in any comparison of serum proteins. For PyD, a biomarker of bone turnover, there were no significant differences between the groups (1.58 ± 0.32 nmol/L, CV = 21% for 1 MG; 1.58 ± 0.10 nmol/L, CV = 6% for 5 MG; 1.51 ± 0.30 nmol/L, CV = 20% for control). For OPG, a downregulator of osteoclastogenesis, the slight trend was downward for 1 MG (323.6 ± 70.7 pg/ml; CV = 22%) and 5 MG (308.8 ± 86.9 pg/ml; CV = 28%) versus control (357.1 ± 66.5 pg/ml; CV = 19%). There were no differences in serum calcium level among the groups.

Fig. 5.

Serum proteins. a) bALP, indicative of osteogenic activity, was significantly higher in 5 MG than in controls (p < 0.05). b, c) PyD and OPG showed no differences among the groups. d) SOST, like bALP, was statistically different for 5 MG versus control (p < 0.05), in this case the higher-dose treatment resulting in significantly lower expression of the osteoblast inhibitor.

4. Discussion

At a time point of fracture healing corresponding to about the mid-point of mineralization in this model, treatment with daily gastrolith ingestion resulted in more tissue volume and, for the lower dose of 1 mg/kg, more bone volume than untreated control. Likewise, osteogenic signaling showed significantly more bALP and significantly less SOST for the higher dose of 5 mg/kg than control, with trendwise similar results for the lower dose. Together with the striking effect on the distribution of bone density, these data support the possibility that dietary gastrolith treatment may enhance bone healing in the repair phase, although its effect specifically on the rate of callus maturation is not yet clear. Whether the larger callus size resulting from gastrolith treatment ultimately helps or hinders recovery and eventual return to native-like bone morphology requires further examination that extends the healing time point beyond that of the current study. As later time points are evaluated that allow enough time for viable consolidation of the callus, the determination of biomechanical properties, in particular, will provide insights into strength and stiffness properties not available in the current study due to the handling risk associated with an unhealed specimen, and moreover because of the limited clinical relevance of testing bones that our earlier experience with this model suggests were likely less than half-strength at 17 days.

In this study, the mouse fibula was chosen for the experimental design due to the ability to create a complete long-bone fracture that does not require instrumentation to stabilize the limb, and unilaterally does not lame the animal. Because the fibula is largely shielded from the load path of the leg in gait, this also allows the effect of the intervention to be mostly isolated from the mechanical influences of physical activity that otherwise can largely define the loading environment and contribute fundamentally to healing of the primary long bones in rodents.

The purported clinical risks of prescribing calcium supplementation, including cardiovascular disease and kidney stones,^16,17 are postulated to be minimized with short-term use of gastrolith expressly to support fracture healing. Calcium supplementation, in fact, was found in one study to reduce risk of fracture by 10% in older people,¹⁸ while also potentiating the beneficial effects of exercise to help prevent fracture in young women.¹⁹ It may also be noted that insufficiently low dietary calcium intake in the general population already is a widely reported health concern,²⁰ and even commonly found in patients following fracture from a fall. In this study, because rodents consume many-fold more calories per unit body weight than do humans, the higher concentration of 5 mg/kg actually represented an increase of only a few percent of calcium intake compared to normal cage diet.

Perhaps most importantly to the field of regenerative medicine, it remains to be determined which components of the gastrolith complex, consisting of many organic and inorganic constituents that help regulate functions like exoskeleton formation and mineral amorphicity in the crayfish, remain bioactive and are most active in promoting osteogenesis in this model. Because this study employed an ingestion route to mimic a nutraceutical application, the inorganic elements may be presumed to have predominated the resulting effect. In another preclinical study, constructs consisting of polyphosphates and similar ACC in a polylactic-glycolic acid scaffold were shown to increase bone regeneration in a critical-size defect and to upregulate carbonic anhydrase and alkaline phosphatase—essential enzymes for initiating bone formation.²¹ Another study described induction of non-enzymatic production of calcium phosphate, the building block of hydroxyapatite.²²

Currently, commercialization of isolated gastrolith-derived compounds for treatment of bone disorders is being actively pursued with limited clinical trials in progress.²³ While patent documents provide some data on characteristics of the materials and performance of the compounds in vivo— most relevantly in distraction osteogenesis—published work in the scientific press appears limited mostly to bioavailability of the calcium, chemical analyses, and other studies with a basic science degree of separation from the immediate impact to bone health or fracture healing concerns in particular.

5. Conclusions

By inducing formation of a relatively large callus, gastrolith treatment resulted in more bone formation than untreated control through enhanced molecular signaling, but with nominally a lower ratio of bone volume to total callus volume, suggesting that the formation of connective tissue scaffolding predominated over mineralization during this healing phase. More study is required to refine these outcomes across complementary time points, and with an inclusive range of parameters that will provide more comprehensive insight into the effect of gastrolith ingestion on the outcome as well as the mechanisms of fracture healing.

Author contributions

Karl Wenger, PI of the study, contributed to the study design, surgery, animal care, experimentation and data analysis, and was primary author of the manuscript.

Steven Zumbrun contributed to the study design and experimentation and edited the manuscript.

Militza Rosas contributed to the surgery, experimentation and data analysis.

Douglas Dickinson contributed to analysis and edited the manuscript.

James McPherson edited the manuscript.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Declaration of competing interest

The authors have no conflicts of interest relevant to this article.