Pain in fibrous dysplasia: identifying nociceptive mechanisms in a preclinical model

Abstract

Pain is a common symptom of fibrous dysplasia (FD), a rare mosaic disorder characterized by fibro-osseous lesions in the bone. Despite the prevalence of pain in FD patients, there is little knowledge about the nociceptive mechanisms and few efficacious treatments. As such, understanding FD pain is essential for patient care. The overall aim of this study was to identify nocifensive behaviors and potential underlying mechanisms in a transgenic mouse model of FD, previously shown to display high face and translational validity. Significant nocifensive behaviors were observed in FD mice (male and female), compared to control mice in the burrowing, grid hanging, home cage activity, and wheel running assays. These changes corresponded to lesion development, as visualized by X-ray imaging. Behavioral deficits improved when analgesics were administered, indicating a nociceptive origin. Tibias and femurs from FD mice demonstrated characteristic FD lesions and the presence of mono- and multi-nucleated CD68⁺ cells, calcitonin gene-related peptide sensory nerve fibers, and vascularization. Lumbar dorsal root ganglia from male FD mice displayed increased staining for activating transcription factor-3 and tyrosine hydroxylase neurons. No difference was observed in the spinal cords between the FD and control groups for glial cell presence and neuropeptide expression. Bone marrow stromal cells were obtained from FD and control mice and cultured in vitro. FD cells developed an increased concentration of inflammatory cytokines (IL-6, tumor necrosis factor-alpha), chemokines (monocyte chemoattractant protein, keratinocyte chemoattractant/human growth-regulated oncogene), and nerve growth factor as compared to controls. Taken together, this study demonstrated for the first time that nociceptive mechanisms such as axonal growth in FD lesions, nerve injury, and inflammation may contribute to FD pain, and it provides a foundation for conducting further studies of pain- and disease-modifying therapeutics for FD patients.

Introduction

Fibrous dysplasia (FD) is a rare bone disorder where fibro-osseous tissue replaces normal bone in a mosaic distribution. This tissue is structurally unstable, leading to deformity, fractures, and pain (both fracture and non-fracture pain), all of which diminish the patients’ quality of life.^1–3 The cause of FD is well described; a post-zygotic substitution mutation occurs in the GNAS gene, which encodes the G-protein (guanine nucleotide-binding protein) alpha subunit (Gα_S) and other isoforms. Here, the arginine amino acid in exon 8 is replaced by a cysteine or histidine amino acid (R201C or R201H, respectively). Rarely, FD will be caused by a mutation of codon 227 in the same gene.¹ These mutations alter the GTPase autoregulatory domain of the Gα_S subunit, which remains constitutively active, acting on adenylyl cyclase to produce excess cyclic adenosine monophosphate (cAMP) and activate downstream signaling mechanisms.^1,4 McCune-Albright syndrome (MAS) is caused when the same genetic mutation affects endocrine organs, sometimes with skin macules and bone lesions, whereas FD only encompasses changes in the bone.¹

Pain is a common feature of FD, but the underlying mechanisms are unknown, making adequate treatment a challenge.⁵ Previous studies have attempted to characterize pain through standardized questionnaires indicating that pain could have both neuropathic-like and nociceptive components;^2,5,6 however, there are yet no studies on FD tissue to support these findings. A recent clinical study has demonstrated that reported pain scores do not correlate with disease burden in FD patients, but pain intensity was more often correlated with reports of neuropathic-like pain qualities.⁶ Opioids and non-steroidal anti-inflammatory drugs are only effective in approximately half of patients.³ Bisphosphonates and anti-receptor activator of nuclear factor kappa-Β ligand treatments have been reported to reduce pain in FD patients, likely through a disease-modifying effect, but the effects are variable between patients and often transient; ultimately, the anti-nociceptive effect of these drugs has not been experimentally assessed.^3,7–9 Understanding the underlying mechanisms of pain in FD may inform more effective pain treatment options and individualized pain treatment modalities.

In general, pain is classified as neuropathic (ie, pain caused by a lesion or disease of the somatosensory nervous system), nociceptive (ie, pain that arises from actual or threatened damage to non-neural tissue and is due to the activation of nociceptors), or nociplastic (ie, pain that arises from altered nociception despite no clear evidence of actual or threatened tissue damage).¹⁰ The bone micro-environment poses a unique challenge to investigating and treating pain. The bone is highly innervated, and disruption of nerves within the different bone compartments (eg, fracture, inflammation, abnormal cell proliferation, etc) can lead to the development of debilitating pain.¹¹ However, whether and how FD lesions are innervated is currently unknown. FD lesions are areas of increased bone remodeling and osteoclastogenesis^1,12,13 and as such, there may be pathological changes in innervation, vascularization, acidosis, or FD-specific factors that contribute to nociception.

The aim of this study was to identify nocifensive behaviors and potential underlying nociceptive mechanisms in a translationally relevant transgenic mouse model of FD.¹⁴

Materials and methods

The detailed methodology is presented in Supplementary Methods.

Animals

A transgenic mouse model of FD was used in this study.^12,15 Paired related homeobox 1 (Prrx1)-Cre-expressing mice¹⁶ were bred with Tet-Gα_s^R201C mice (with a TetO-lacZ gene) that express a human cDNA-derived Gα_s^R201C transgene, responsible for FD.¹⁵ The mice established with these 3 gene regions (Prrx1-Cre, Tet-Gα_s^R201C, TetO-lacZ) are termed “Gα_s^R201C mice.” Gα_s^R201C mice were bred with “Linker mice” (B6.Cg-Gt(ROSA)26Sor^{tm1(rtTA, EGFP)Nagy}/J; Strain #005670; The Jackson Laboratory). After genotyping, the mice were categorized according to the genes they possessed for further experiments (Figure S1), while mice without any of those genotypes were euthanized. All mice possess the Prrx1-Cre and Linker genes, but the FD mice also possess the Gα_s^R201C transgene. Figure S1 illustrates how the site-specific and inducible model functions.

Experiments were conducted in accordance with the Danish Animal Experiments Inspectorate (licenses 2021-15-0201-00872 and 2016-15-0202-00045). Gα_s^R201C mice were obtained from the National Institute of Health Skeletal Disorders and Mineral Homeostasis Section and rederived by Charles River. Linker mice were purchased from The Jackson Laboratory.

Detailed information about housing and study endpoints can be found in the Supplementary Methods.

Fibrous dysplasia induction

Doxycycline (doxycycline hyclate; Sigma-Aldrich) was administered in drinking water (protected from light and changed twice per week) of control and FD mice at approximately 13-18 wk old. Mice were dosed with 0.1 g/L in the behavioral studies and 0.08 g/L in the intervention studies.

Study design

The study designs are shown in Figure S2. The first behavioral study was conducted to assess the development of nocifensive behavior in male and female mice using burrowing, grid hanging, home cage activity, and wheel running.

Details and justification of the study design can be found in the Supplementary Methods.

Behavior

Grid hanging

The grid hanging methodology was adapted from Falk et al.¹⁷ and the similar inverted screen test.¹⁸ Mice were placed on a metal grid, filmed for 2.5 min, and the videos were analyzed to determine the latency to first fall.

Burrowing

Burrowing was conducted as previously described.¹⁹ Mice were placed in individual boxes with a burrowing tube filled with 500 g of sand and left to burrow for 2 hr. The total sand burrowed was assessed.

Cage activity and wheel running

Mice were housed in Digital Ventilated Cages (DVCs) by Tecniplast with electromagnetic wheels (GYM500, Tecniplast) where distance is automatically tracked. The cage activity system and metrics have been described in a previous publication by Ianello.²⁰

Details about behavior training and assessment can be found in the Supplementary Methods.

Analgesic treatment

Ibuprofen 30 mg/kg (Sigma-Aldrich) or saline was administered intraperitoneally 30 min prior to burrowing. Ibuprofen was used as it does not reduce burrowing behavior in naïve mice.²¹ Morphine 10 mg/kg (Abcur) or saline was administered subcutaneously 45 min prior to grid hanging. Morphine was used as a strong, known analgesic. Both FD and control mice received treatment or vehicle.

Imaging

X-ray imaging

X-ray imaging was conducted weekly using Lumina XR IVIS (In Vivo Imaging System) apparatus (Caliper Life Sciences, Teralfene). Mice were sedated and maintained using 2.5% isoflurane (1000 mg/g isoflurane, Attane vet, ScanVet) for 10-20 min. X-ray images were obtained of the entire skeleton and the hind limbs.

Micro-computed tomography imaging and analysis

Post-euthanasia and prior to decalcification, hind limbs were imaged using computerized microtomography (micro-CT; Skyscan 1272, Bruker) according to the guidelines for bone structure analysis in rodents.²²

Details of micro-CT acquisition and analysis can be found in the Supplementary Methods.

Tissue collection and embedding

Serum was taken from half of the FD and control mice after X-ray imaging. Remaining mice were euthanized by intracardial perfusion similar to previous literature at 15-20 wk old.²³

Details about euthanasia and tissue collection can be found in the Supplementary Methods.

Immunohistochemistry

The tibia, femur, spinal cord, and dorsal root ganglia (DRGs) underwent sectioning, immunohistochemical staining, imaging, and quantitative analysis. All details about the methods and materials can be found in the Supplementary Methods.

Cell culture and in vitro assays

Bone marrow stromal cells were extracted from 6 female mice (3 FD mice, 3 control mice), approximately 8 wk old according to a previously established protocol,²⁴ and with the gene profile described in Figure S1. After isolation and primary culture, cells were stimulated with normal media or doxycycline-supplemented media. Conditioned media and cell lysate were collected for further analysis.

Details on cell isolation, culture, and stimulation can be found in the Supplementary Methods.

cAMP, Meso Scale Discovery Multiplex, and nerve growth factor ELISA assays

cAMP concentration assay (Cisbio GS Dynamic Kit, PerkinElmer), Meso Scale Discovery Multiplex assay (U-PLEX; MSD), and nerve growth factor (NGF) ELISA (NGF Rapid ELISA kit: mouse; Avantor) were conducted according to manufacturer directions. The details of the chosen markers and analysis can be found in the Supplementary Methods.

Statistics

All statistical analysis was conducted using GraphPad Prism 9 (Version 9.0.4). Statistical tests are defined for each experiment in the results, with statistical significance set at p < .05.

Results

FD mice develop nocifensive behavior

Female and male FD mice chronically treated with doxycycline, but not doxycycline-treated control mice, developed FD-like lesions, as evidenced by X-ray (Figure S3), μCT, and histology (detailed below, Figure 2A and B). FD mice burrowed significantly less sand in the burrowing test (female BL vs D14: p = .0003, male BL vs D6: p = .0048; Figure 1A and B; 2 control and 3 FD mice excluded due to low BL burrowing), and their performance was impaired in grid hanging (female BL vs D11: p = .0106, female BL vs D18: p < .0001, male BL vs D7 and D14: p < .0001; Figure 1G-I), home cage activity (female and male BL vs week 2: p < .0001; Figure 1C and E) over time compared to control mice, and wheel running (female BL vs week 2: p < .0001, male BL vs week 2: p = .0005; Figure 1D and F).

Figure 2.

Histological hematoxylin and eosin staining and micro-CT in FD mice compared to control mice. (A, B) H&E histological staining of bone tissue. (C-J) Representative micro-CT images of the distal femur (C-F) and proximal tibia and fibula (G-J) of FD (D, F, H, J) and control (C, E, G, I) mice, post-euthanasia. (K-N) micro-CT analysis of BMD (K, L) and BV/TV (M, N) comparing FD mice and control mice. Unpaired Student t-test used to compare groups and graphs are individual values with bars representing median ± IQR. ^*p < .05. ^**p < .01. ^***p < .001. ^****p < .0001. Abbreviations: BV/TV, bone volume fraction; FD, fibrous dysplasia.

Figure 1.

Behavior changes in FD and control mice. (A) Burrowing behavior in female mice over time with ibuprofen (30 mg/kg) and vehicle treatment. (B) Burrowing behavior in male mice. (C) AUC of home cage activity in female mice over time. (D) AUC of wheel running distance in females over time. (E) AUC of home cage activity in male mice over time. (F) AUC of wheel running distance in males over time. (G) Grid hanging behavior in female mice before treatment. (H) Grid hanging behavior in female mice after morphine (10 mg/kg) and vehicle treatment at D11. (I) Grid hanging behavior in male mice. Except wheel running and home cage activity where n represents a cage, n represents 1 mouse; female mice were housed 2 per cage and males housed 1 per cage. Statistics: A-H, two-way ANOVA with Tukey’s multiple comparison correction; I, one-way ANOVA. Graphs are individual values with bars representing median ± IQR. ^*p < .05. ^**p < .01. ^***p < .001. ^****p < .0001. Abbreviations: AUC, area under the curve; BL, baseline; D, day; FD, fibrous dysplasia.

To test whether these behavioral deficits were pain related, analgesic intervention was applied to the grid hanging test and the burrowing test (Figure S2). Morphine was administered prior to grid hanging as it is a strong analgesic that is effective for most nociceptive conditions. However, burrowing is highly sensitive to analgesic administration, with most analgesics having a sedative effect. According to a previous systematic review, ibuprofen and celecoxib do not have a sedative effect on burrowing behavior and therefore ibuprofen was selected as an intervention for the burrowing test.²¹ Due to a more rapid disease progression in the male mice, female mice were used for intervention assessments. Female FD mice treated with ibuprofen maintained their burrowing behavior over time, similar in magnitude compared to baseline and the control mice.

In contrast, FD mice treated with vehicle demonstrated reduced burrowing behavior over time (Figure 1A; FD-ibuprofen BL vs D14: p = .5355, FD-vehicle BL vs D14: p = .0003). Likewise, morphine significantly improved the grid hanging time in FD mice as compared to vehicle-treated FD mice on day 11 (FD-morphine vs FD-vehicle: p = .0316; Figure 1H). There was no significant difference between the groups on day 18 (FD-morphine vs FD-vehicle: p = .1115).

Both male and female FD mice demonstrated reduced home cage activity and wheel running over time (Figure 1C-F) as FD-lesions developed and in concert with the reduced burrowing and grid hanging behavior (Figure S4). Male control mice did not display significant changes in home cage activity over time (male baseline vs week 2: p = .1955). Female control mice were significantly less active in the home cage at 1 wk on doxycycline (BL vs week 1: p < .0001), but there was no change between baseline and week 2 (p = .8601) on doxycycline. In contrast to the FD mice, wheel running improved in control mice over time (female BL vs week 2: p = .0096, male BL vs week 2: p = .1955; Figure 1D).

The diurnal behavioral patterns (Figure S4) during the dark phase were similar to those observed in other studies, where C57Bl/6 mice demonstrated 2-3 activity peaks during the 12-hr dark period for both home cage activity and wheel running.²⁵ This suggests that the data obtained by the DVCs is comparable to that obtained by previous studies using other home cage systems for behavioral assessment.

Mice develop FD-like bone lesions that correspond to nocifensive behavior

Hematoxylin and eosin staining of the proximal tibia and distal femur confirmed the presence of fibrous tissue lesions in the FD mice, but not in the control mice (Figure 2A and B). End-stage micro-CT (Figure 2C-J) further demonstrated that both male and female FD, but not control mice, developed FD pathology. Based on micro-CT analysis, 2 male mice and no female mice were excluded, as they did not develop FD. Micro-CT analysis revealed that male FD mice had significant pathological changes compared to the control mice (Table 1) for all measured parameters of the cortical bone (femur [Figure 2C-F] and tibia [Figure 2G-J]) and most of the trabecular bone parameters, suggesting extensive bone resorption. The trabecular thickness in the tibia and femur and the trabecular spacing in the femur were the only non-significant parameters.

Table 1.

Micro-CT analysis and comparison between FD mice and control mice in both female and male mice.

	Female			Male
	FD	Control	p-value	FD	Control	p-value
Trabecular bone
Proximal tibia
BMD (g/cm3)	0.099 ± 0.047	0.223 ± 0.214	.0232	0.069 ± 0.055	0.218 ± 0.029	<.0001
BV/TV (%)	6.550 ± 3.880	10.90 ± 3.130	.0040	4.672 ± 3.679	16.79 ± 2.42	<.0001
Tb.Th (mm)	0.050 ± 0.022	0.050 ± 0.002	.4598	0.053 ± 0.016	0.047 ± 0.002	.1364
Tb.N (mm−1)	1.290 ± 0.247	2.160 ± 0.591	<.0001	0.938 ± 0.765	3.560 ± 0.396	<.0001
Tb.Sp (mm)	0.290 ± 0.033	0.263 ± 0.043	.0487	0.334 ± 0.137	0.178 ± 0.011	.0030
Distal femur
BMD (g/cm3)	0.096 ± 0.020	0.115 ± 0.032	.0409	0.132 ± 0.034	0.187 ± 0.024	.0009
BV/TV (%)	6.420 ± 1.480	8.0 ± 2.660	.0374	10.10 ± 3.120	14.20 ± 2.250	.0039
Tb.Th (mm)	0.044 ± 0.006	0.048 ± 0.004	.0466	0.046 ± 0.005	0.046 ± 0.002	.4579
Tb.N (mm−1)	1.460 ± 0.268	1.670 ± 0.515	.0965	2.160 ± 0.509	3.10 ± 0.452	.0006
Tb.Sp (mm)	0.254 ± 0.023	0.210 ± 0.031	.2547	0.217 ± 0.041	0.193 ± 0.012	.0646
Cortical bone
Proximal tibia
BMD (g/cm3)	1.130 ± 0.038	1.120 ± 0.028	.267	1.010 ± 0.051	1.10 ± 0.017	.0002
Ct.Th (mm)	0.182 ± 0.008	0.184 ± 0.008	.368	0.149 ± 0.031	0.183 ± 0.009	.0045
Ct.Ar (mm−1)	0.357 ± 0.252	0.660 ± 0.052	.0006	0.042 ± 0.025	0.737 ± 0.058	<.0001
Distal femur
BMD (g/cm3)	1.218 ± 0.042	1.213 ± 0.029	.362	1.090 ± 0.660	1.180 ± 0.034	.0025
Ct.Th (mm)	0.198 ± 0.0	0.023 ± 0.0	.073	0.147 ± 0.023	0.184 ± 0.012	.0005
Ct.Ar (mm−1)	0.696 ± 0.0	0.721 ± 0.0	.119	0.263 ± 0.271	0.765 ± 0.060	<.0001

Italic values represent significant differences (p < .05) between the FD and control group based on the unpaired Student t-test between the 2 values; no correction for multiple comparisons. Abbreviations: BV/TV, bone volume fraction; Ct.Ar, cortical area; Ct.Th, cortical thickness; FD, fibrous dysplasia; Tb.N, trabecular number; Tb.Sp, trabecular spacing; Tb.Th, trabecular thickness.

The female FD mice also exhibited significant pathological changes; however, fewer parameters were significantly different than the male mice. In the cortical bone, only the cortical area of the tibia was significantly reduced in the female FD mice, while all other cortical bone parameters were not different. The female FD mice had significant deterioration in different trabecular bone parameters, except for the trabecular thickness of the tibia and the trabecular spacing and trabeculae number in the femur. The micro-CT analysis also revealed that at the experimental end stage, male mice developed greater trabecular bone loss than females. The difference in trabecular BMD in both the femur (Figure 2K) and tibia (Figure 2L) is greater in males than in females. Similarly, there is a greater difference in the BV/TV in males than females in the femur (Figure 2M) and tibia (Figure 2N). This is in accordance with the observations of behavioral differences between the male and female mice.

Similarly, X-ray analysis conducted throughout the study showed faster progression in male FD mice, where pathological changes were evident in the calcaneus and tibia at D7; at this time, the female mice only showed pathological changes in the calcaneus. At D14, male FD mice demonstrated pathological changes and deformity in the calcaneus, tibia, fibula, and femur, whereas female FD mice demonstrated pathological changes in the calcaneus, tibia, and fibula (Figure S3). Finally, analysis of the FD lesions in the femur showed that males—as compared to females—developed larger lesions that were present in over half the bone and CD68⁺ (marker of cells of the monocyte/macrophage lineage) clusters within the lesion were also larger (Table 2; Figure S5).

Table 2.

Percentage of FD lesions per section with nerve fibers and lesion sizes in the tibia (Figure S5).

Sex	Group (age, wk)	CGRP +	TH +	PGP 9.5	NF200	CD68 +  staining in FD lesions
		Percentage of FD lesions per section with nerve fibers				Width of FD lesion (mm)	Percentage of bone with FD lesion	Diameter of CD68+ cell cluster (μm)
Female	Control (18-20)	0	0	0	0	N/A	N/A	N/A
	Control (14-16)	0	0	0	0	N/A	N/A	N/A
	Control (13-15)	0	0	0	0	N/A	N/A	N/A
	Control (13-15)	0	0	0	0	N/A	N/A	N/A
	Control (15-17)	0	0	0	0	N/A	N/A	N/A
	Control (14-16)	0	0	0	0	N/A	N/A	N/A
	FD (18-20)	66.60	24.99	52.73	6.25	1	25	400
	FD (14-16)	12.50	32.50	28.13	6.66	2	35	300
	FD (18-20)	0	33.27	75	19.43	a	a	a
	FD (18-20)	36.83	22.13	48.60	11.30	1	35	200
	FD (14-16)	35	27.77	44.17	8.33	0.8	40	300
	FD (13-15)	36.60	13.07	59.48	2.38	1	30	500
	Average control	0	0	0	0	N/A	N/A	N/A
	Average FD	31.25 ± 9.40	25.62 ± 3.05	51.35 ± 6.39	9.05 ± 2.39	1.16	33.0	340
Male	Control (13-15)	0	0	0	0	N/A	N/A	N/A
	Control (13-15)	0	0	0	0	N/A	N/A	N/A
	Control (14-16)	0	0	0	0	N/A	N/A	N/A
	Control (14-16)	0	0	0	0	N/A	N/A	N/A
	Control (13-15)	0	0	0	0	N/A	N/A	N/A
	Control (13-15)	0	0	0	0	N/A	N/A	N/A
	FD (18-20)	15.27	3.57	35.04	0	1	35	500
	FD (18-20)	8.30	0	31.70	3.46	2	75	500
	FD (14-16)	33.33	32.62	49.48	0	a	80	a
	FD (13-15)	24.97	2.77	39.23	0	a	80	a
	FD (14-16)	28.89	22.07	48.44	8.48	1.5	30	500
	FD (13-15)	9.38	20.83	56.21	33.33	0.7	40	200
	Average control	0	0	0	0	N/A	N/A	N/A
	Average FD	20.02 ± 4.29	13.64 ± 5.44	43.35 ± 3.87	7.54 ± 5.33	1.35	56.7	425

Abbreviations: CD68⁺, cluster of differentiation 68; CGRP, calcitonin gene-related peptide; FD, fibrous dysplasia; mm, millimeter; μm, micrometer; NF200, neurofilament 200 kDa; PGP 9.5, protein gene product 9.5; TH, tyrosine hydroxylase; N/A, not applicable.

^aTissue analysis incomplete due to poor tissue integrity.

Corresponding to the progression of the FD lesions, nocifensive behavior was observed in male FD mice 1 wk earlier than in female FD mice in all pain assessments (Figure 1). Furthermore, nocifensive behavior tended to become progressively worse as FD-lesions became more apparent in X-ray analysis in both sexes. Comparison of the X-ray (Figure S3), micro-CT results (Figure 2C-J; Table 1), and the FD lesion size (Table 2) with the nocifensive behavior patterns (Figure 1) suggests that more advanced FD development corresponds with increased nocifensive behavior.

Vascularization, innervation, and cellular infiltration of FD lesions

Confocal imaging of bones from FD mice revealed pathological changes that may contribute to pain. Confocal images were acquired from healthy control samples (Figure 3D, F, H, J, L, N) to illustrate the differences between healthy bone and FD lesions, and the nerve fibers in FD lesions were quantified (Table 2). CD68⁺ cells—both mononuclear (eg, macrophage/monocyte/pre-osteoclast) and multinuclear (osteoclast)—were present within the FD lesions in the bones analyzed (Figure 3A-E; Figure S6). Use of PGP 9.5, a pan-neuronal marker, revealed a high density of nerve fibers in the tibia of 51.35 ± 6.39% lesions in female mice and 43.35 ± 3.87% lesions in male mice (Figure 3F-G). CGRP⁺ nerve fibers (Figure 3H-I) were present in most FD mice, with 31.25 ± 9.40% of lesions in female mice and 20.02 ± 4.29% of lesions in male mice demonstrating positive staining. TH⁺ nerve fibers (Figure 3L-M; Table 2) were present in most mice, but a low proportion of lesions had positive staining, while the myelinated nerve fiber marker NF200 was present in a low proportion of FD lesions (Figure 3J-K; Table 2). Finally, vascularization (endomucin) was observed in all FD lesions (Figure 3N and O). None of the unaffected bones (bone that was not an FD lesion) within the FD mice had aberrant nerve fibers or vascularization (Figure S6).

Figure 3.

Representative images of IHC staining of bone innervation and vascularity in FD mice (A, B, C, E, G, I, K, M, O) and control mice (D, F, H, J, L, N). (A-E) CD68⁺ and DAPI staining demonstrating cells from the monocyte lineage. Asterisk demonstrates mononuclear cells (eg, monocytes, macrophages, pre-osteoclasts) and arrowhead demonstrates multinuclear cells (eg, osteoclasts); (F, G) PGP 9.5 and DAPI staining; (H, I) CGRP⁺ and DAPI staining; (J, K) NF200 and DAPI staining; (L, M) TH and DAPI staining; (N, O) Endomucin and DAPI staining. Mouse details: (A, B, C, E, I) female FD mouse, 14-16 wk old; (D, H, J, L, N) female control mouse, 18-20 wk old; (F) female control mouse, 14-16 wk old; (G, K) female FD mouse, 13-15 wk old; (M, O) male FD mouse, 14-16 wk old. Abbreviations: ♀, female; ♂, male; CD68⁺, cluster of differentiation 68; CGRP⁺, calcitonin gene-related peptide; DAPI, 4′,6-diamidino-2-phenylindole; FD, fibrous dysplasia; IHC, immunohistochemistry; NF200, neurofilament 200 kDa; PGP 9.5, protein gene product 9.5; TH, tyrosine hydroxylase.

Indication of peripheral nerve damage present in male FD mice

Increased expression of ATF3 in DRG cell bodies is a widely accepted sign of peripheral nerve damage, indicating a potential neuropathic state (Figure 4A-D).²⁶ DRGs from male FD mice demonstrated significantly more cell bodies with positive ATF3 staining compared to control mice (female FD vs control: p = .1492, male FD vs control: p = .0009; Figure 4E) and increased presence of TH⁺ nerve fibers (female FD vs control: p = .3456, male FD vs control: p = .0279; Figure 4F-J).²⁷ It is unlikely that the observed changes were due to Gα_s^R201C expression in the neurons. DRGs from FD mice with the LacZ reporter gene were cultured in vitro. Cells were stained for beta-galactosidase expression after doxycycline induction and no staining was observed (Figure S7E and F).

Figure 4.

Protein expression change in DRGs from FD and control mice. Representative IHC images of DRGs from control (A, C, F, H) and FD (B, D, G, I) mice, both female (A, B, F, G) and male (C, D, H, I). (A-D) ATF3 staining of DRGs; (F-I) TH staining of DRGs; (E) graph demonstrating proportion of DRG cell bodies with ATF3 staining in female and male mice; (J) graph demonstrating proportion of TH-positive axons in DRGs from female and male mice. Unpaired Student t-test used to compare groups and graphs are individual values with bars representing median ± IQR. ^*p < .05. ^***p < .001. Abbreviations: ATF3, activating transcription factor 3; DRG, dorsal root ganglion; FD, fibrous dysplasia; TH, tyrosine hydroxylase.

No indication of neurochemical or glial cell changes in the spinal cord dorsal horn of FD mice

Changes in neuropeptide expression, microglia, and astrocytes in the dorsal horn of the spinal cord can provide information about pain type and the development of changes in the central nervous system in response to pathological changes.²⁸ There were no significant differences between FD and control mice (male and female) in the spinal cord for neuropeptide expression (Figure 5A-F; CTCF) or in the presence of astrocytes (Figure 5G-H; CTCF) and microglia (Figure 5I-L; cells/area and Figure S8).

Figure 5.

Neuropeptide expression and glial cell presence in spinal cords from FD and control mice (female and male). Neuropeptide and GFAP were measured using CTCF and microglia (Iba1⁺ and Pp38) were measured using cells/area. Unpaired Student t-test used to compare groups and graphs are individual values with bars representing median ± IQR. No significant differences observed. Abbreviations: CTCF, corrected total cell fluorescence; FD, fibrous dysplasia; GFAP, glial fibrillary acidic protein; Iba1⁺, ionized calcium-binding adaptor molecule 1; Pp38, phospho-p38 mitogen-activated protein kinase.

In vitro induction of FD cells generates factors associated with painful conditions

Having established the development of nocifensive behavior in FD mice when induced with doxycycline and the efficacy of an anti-inflammatory analgesic to reverse this behavior, we wanted to determine whether BMSCs derived from FD mice (FD cells; Figure S1) express inflammatory factors that have been associated with painful conditions. The expression of the transgenes was confirmed by staining for LacZ expression after doxycycline induction (Figure S7A-D).

FD cells stimulated with doxycycline in vitro produced significantly more cAMP than control cells (BMSCs extracted from control mice [Figure S1; p < .0001]) stimulated with doxycycline, as well as both cell types that were treated with vehicle, thus confirming the in vitro induction of the Gα_s^R201C gene mutation (Figure 6A). Further analysis of the conditioned media showed significantly upregulated NGF expression in doxycycline-stimulated FD cells compared to treatment and cell control conditions (Figure 6B; p < .0001). A multiplex assay identified 5 secreted factors that were significantly upregulated in FD cells stimulated with doxycycline: IL-6 (p < .0001), vascular endothelial growth factor (VEGF) (p < .0001), KC/GRO (p < .0001), TNFα (p < .0001), and MCP-1 (p < .0001; Figure 6C-G). However, the other factors analyzed were either below the level of quantification (IFNγ, IL-1β, IL2, IL4, IL5, Eotaxin), or not significantly different between cell types (Figure 6H-K). TNFα stimulation was employed to establish that the cultured cells could express and secrete the chosen factors and to determine if there were differences between the FD and control cells. These results showed that the FD and control cells were capable of secreting all 15 factors when stimulated with TNF-α and there was no significant difference between the 2 cell types (Figure S9). Serum taken from end-stage mice (FD and control) of the intervention study (study 2; Figure S2) did not reveal a significant difference for any of the assessed markers (Figure S10).

Figure 6.

In vitro cAMP assay (A), NGF ELISA (B), and multiplex assay (C-K). FD cells and control cells were activated with doxycycline (10 μg/mL) or vehicle (phosphate buffered saline) for 24 hr with IBMX. Intracellular cAMP was assessed using cAMP Gs dynamic kit and protein factors were assessed using MSD U-PLEX assay. Two-way ANOVA with Tukey’s multiple comparison test. Graphs are individual values with bars representing median ± IQR. ^****p < .0001. Abbreviations: cAMP, cyclic adenosine monophosphate; FD, fibrous dysplasia; IBMX, 3-isobutyl-1-methylxanthine; NGF, nerve growth factor.

Discussion

This study demonstrates for the first time that a nociceptive phenotype develops in a mouse model of FD. Nocifensive behavior was observed in FD mice when they were induced with doxycycline and it corresponded to FD progression. Additionally, a fully developed pathology was coupled with worse nocifensive behavior in both sexes.

The results of this study suggest that this mouse model is suitable for investigating FD-related pain and testing new treatments, but further research would benefit from investigating monostotic models and non-painful models of FD to fully gauge the mechanisms contributing to FD-related pain. In so doing, this would yield further information about pain in this highly variable and complex disease.^3,6 One challenge of this model and assessing nocifensive behavior is the bilateral development of FD, which prevents the implementation of commonly used behavioral assessments such as gait analysis or weight bearing. As such, it was necessary to employ spontaneous behavioral tests that rely on welfare and functional assessments to determine the development of nocifensive behavior and subsequent relief from analgesics.⁷ Morphine treatment had a minor effect on grid hanging behavior. This behavior is typically utilized for muscle strength and mice may have difficulty gripping the grid. However, partial reversal was achieved in the morphine-treated FD mice, demonstrating that nociceptive development contributed to grid hanging deficiency, but this test and morphine treatment may not be suitable to assess nociception and analgesic treatment in this model. While morphine can affect behavioral tests, the morphine- and vehicle-treated control mice did not demonstrate significant differences. However, a robust change was observed in burrowing behavior when FD was induced, which was reversed with ibuprofen treatment. This strongly supports that the behavioral deficits observed in the FD mice were pain-related, and further suggests an inflammatory component to the pain. An unexpected finding was the faster FD development in male mice than in female mice. Females tend to make up a larger cohort in clinical studies of FD pain, but sex does not seem to play a role in FD-associated pain patients.³ The variability between the male and female development is an interesting phenomenon that we do not yet understand. Sex differences in terms of hormones, metabolism, and bone development may explain the differences observed. However, as this is an inducible model, factors such as doxycycline intake may also lead to the variable outcomes. Further studies on this model are necessary to determine whether the changes are due to a sexually dimorphic trait or model design, and if it is the former, is this translational to FD patients?

This study suggests different mechanisms that may contribute to nociception development in FD: (1) Inflammation might play a role given the cellular expression of inflammatory chemokines (MCP-1, KC/GRO) and cytokines (IL-6, TNF-α),^29–32 confirmed in a similar, separate study³³ together with the efficacy of ibuprofen, the presence of macrophages and/or monocytes, and blood vessels (with increased expression of VEGF in vitro).^34–36 However, the upregulated expression of these factors in vitro and not in the serum would suggest that these changes are local, rather than systemic. The significant accumulation of cAMP by FD cells in vitro demonstrated that the FD cells produce the known causative mechanism of FD. FD cells were not purified for in vitro culture based on previous literature demonstrating that FD cells cultured without non-FD cells (cells without the Gα_s^R201C mutation) cannot survive.³⁷ Therefore, the inflammatory factors produced in vitro, may be by FD cells, non-FD cells, or both; nevertheless, these pro-inflammatory factors have been produced when FD is induced. The paracrine effect of FD cells on non-FD cells has yet to be established and whether this effect can be therapeutically manipulated. As such, this might suggest that the nociceptive influence of the FD lesion is localized and the best option for future FD pain treatments would be to select a therapeutic target within the lesion. The expression of NGF, vascular endothelial growth factor (VEGF), cytokines, and chemokines by FD cells themselves suggests that they may directly contribute to nociception at early stages before extensive skeletal changes have occurred. (2) Infiltration of sensory nerve fibers (CGRP⁺) and TH⁺ neurons within the FD lesion and NGF upregulation in vitro suggest increased peripheral sensitivity due to FD development.^38–40 Furthermore, MCP-1 was upregulated in FD cells and previous research has demonstrated that MCP-1 can enhance transient receptor potential vanilloid 1 density, which may contribute to nociception.³² (3) MCP-1 may also contribute to the mononuclear CD68⁺ cells in the FD lesion, due to its chemokine activity.⁴¹ Nociception development in this model may also be due to the presence of monocytes, macrophages, or osteoclasts that are associated with other painful bone disorders.⁴² Osteoclasts secrete protons as one of several mechanisms to degrade and resorb bone. This high presence of osteoclasts may generate an acidic micro-environment that contributes to the activation of acid-sensing ion channels that can be found in sensory nerve fibers.^5,43 (4) ATF3 (a marker of peripheral nerve damage) and TH were upregulated in the DRGs of male mice that exhibited more advanced FD as compared to female FD mice, suggesting nerve damage in more advanced stages of FD.^26,27 In previous studies, patients have reported features of neuropathic-like pain, which is commonly associated with greater pain severity and reduced quality of life.^5,6 As seen in other painful bone disorders, a range of mechanisms may contribute to patients’ pain experience.^11,44 The complicated and individualistic nature of FD, as well as the difficulty of treating FD pain^3,5,6 can suggest that different nociceptive mechanisms might variably contribute to pain in FD patients.

Analysis of neuropeptides and glial cells in the spinal cord revealed no differences between the FD and control mice (males and females). Honore et al.²⁸ previously demonstrated that different murine models of painful conditions generated unique neurochemical and cellular changes⁴⁵ in the spinal cord within a similar time frame of 14 d. The neuropathic and inflammatory models presented by Honore et al.²⁸ do not represent complex disease states with development stages and as such, such changes may occur at late stages. Within the spinal cord in this study, neuropeptide expression and glial cell proliferation do not seem to play a role in FD pain. However, we cannot yet rule out the possibility of central mechanisms contributing to FD pain. In vitro, in vivo, and clinical electrophysiology has been performed to assess central excitability in neuropathic pain models⁴⁶ and microglia polarization and altered activity can contribute to neuroinflammation in the central nervous system.⁴⁷ So, even though no changes in microglial or neuropeptide expression were found in the dorsal horn of this model, further studies are necessary to rule out central involvement in patients with FD pain.

Taken together, this study provides an important foundation for more targeted studies investigating these different mechanisms and assessing how they may contribute to pain individually and in combination.

A limitation of the study is the lack of similar information from human tissue samples. FD is a rare disease and tissue samples are challenging to obtain. However, it would be highly beneficial to conduct histological and molecular assessments on clinical samples of FD lesions similar to those conducted here to determine the correlation strength. This would improve the validity and translatability of this model in studying FD pain and novel analgesic strategies. Another limitation is that FD is a lifelong condition, which is challenging to assess in this model that develops somewhat rapidly. A reduced and sustained doxycycline dose may allow for long-term in vivo assessment of nociception development. Lastly, male mice developed FD and the associated nociception rapidly, which made it challenging to perform an analgesic intervention study before the pathological changes were too extensive to relieve with standard analgesic doses. Our data is supported by a new article on Biorxiv by Palmisano et al.,⁴⁸ which uses an EF1α-Gsα^R201C transgenic mouse model. In this study, they investigated the innervation of FD lesions in another mouse model of FD. In the mouse model, the causative FD mutation is expressed in all cells (due to the EF1α promoter) and FD develops (due to the Gsα^R201C mutation) for over a year; they recorded behavioral deficiencies and importantly, sensory nerve fibers in the FD lesions. Overall, the results from these 2 different FD models are complementary. The confirmation of the development of a nociceptive phenotype establishes our model as a tool to test novel analgesics to treat pain in FD patients, particularly novel antibody treatments targeting factors contributing to pain. Both TNF-α and NGF were upregulated in vitro in doxycycline-stimulated FD cells and analgesic treatments have been developed specifically targeting these 2 factors individually.^49,50 Both drugs have adverse side effects; however, a proof-of-concept study applying these analgesics could be beneficial in further elucidating the underlying nociceptive mechanisms of FD and also testing novel analgesics for FD treatment. The cAMP secondary messenger pathway is fundamental to cell processes, including pain and inflammatory pathways.⁴ Further studies on cellular mechanisms may elucidate cAMP-directed pathways contributing to FD nociception.

In conclusion, we established that this site-specific and inducible model developed FD with a corresponding nociceptive phenotype. It can be further utilized to study novel and effective analgesics for FD patients, which are sorely needed.

Author contributions

Chelsea Hopkins (Conceptualization, Methodology, Formal analysis, Investigation, Resources, Data curation, Writing—original draft, Writing—review & editing, Visualization, Project Administration, Funding acquisition), Luis Fernandez de Castro (Methodology, Investigation, Resources, Writing—review & editing), Julie Benthin (Formal analysis, Investigation, Data curation, Writing—review & editing, Visualization), Marta Diaz-delCastillo (Methodology, Investigation, Writing—review & editing), Pravallika Manjappa (Methodology, Investigation, Writing—review & editing), Alison Boyce (Resources, Writing—review & editing), Ruth Elena Martinez Mendoza (Investigation, Writing—review & editing, Visualization), Juan Antonio Vazquez Mora (Investigation, Writing—review & editing, Visualization), Giovanni Emmanuel Lopez-Delgado (Investigation, Writing—review & editing, Visualization), Lizeth Yazmin Ponce Gomez (Investigation, Writing—review & editing, Visualization), Khaled Elhady Mohamed (Methodology, Investigation, Resources, Writing—review & editing), John E. Linley (Resources, Writing—review & editing), Michael T. Collins (Resources, Writing—review & editing), Juan Miguel Jimenez-Andrade (Methodology, Investigation, Resources, Writing—review & editing, Visualization, Funding acquisition), and Anne-Marie Heegaard (Conceptualization, Formal analysis, Resources, Data curation, Writing—review & editing, Supervision, Project administration, Funding acquisition).

Funding

C.H., P.M., K.E.M., J.E.L., and A.-M.H. received funding for this project from the European Union's Horizon 2020 research and innovation program under the Marie Sklodowska-Curie grant agreement No. 814244. C.H., J.B., J.M.J.-A., and A.-M.H. received funding for this project from a research grant from the University of Pennsylvania Orphan Disease Center in partnership with the FD/MAS Alliance. The Core Facility for Integrated Microscopy, Faculty of Health and Medical Sciences, University of Copenhagen was used to obtain microscope images. In vivo imaging data were collected at the Center for Advanced Bioimaging (CAB) Denmark, University of Copenhagen. A.-M.H. and J.M.J.-A. received funding from the Lundbeck Foundation under the grant agreement R392-2022-216.

Conflicts of interest

None declared.

Data availability

Data is available upon request. Please contact the corresponding author of this article.