Improvements in Bone Quality by Parathyroid Hormone Treatment are Enhanced in the Nmp4 Knockout Mouse Model

Abstract

Osteoporosis is linked to increased bone fragility. Unlike anti-resorptive therapies, the analogue of parathyroid hormone, PTH (1–34), is an FDA-approved therapeutic for osteoporosis that enhances bone formation. However, as PTH treatment potency declines over time, it is necessary to investigate the mechanisms involved in this attenuation to reinforce its long-term efficacy. This need has led to investigations into the transcription factor nuclear matrix protein 4 (Nmp4), in which PTH treatment of mice globally lacking Nmp4 (Nmp4^−/−) enhanced bone formation. Yet, the changes in the compositional quality of PTH-stimulated bone in Nmp4^−/− mice are unknown, which in turn could impact the efficiency of this approach. To this end, we characterized cortical bone quality in Nmp4^−/− mice and wild-type littermates treated with PTH for 8 weeks, starting at 16 weeks of age, using micro-computed tomography, Raman spectroscopy, X-ray diffraction, biochemical assays, and biomechanical characterization (whole-bone strength, fracture toughness). PTH treatment and Nmp4 ablation increased tissue and marrow area and maximum moment of inertia. Femora from PTH-treated mice exhibited increased stiffness, maximum load, and fracture resistance. Bone in Nmp4^−/− mice with PTH treatment demonstrated lower mineral crystallinity, decreased mineral-to-matrix ratio, lattice spacing, altered levels of advanced glycation end-products, increased levels of osteocalcin, and increased matrix phosphorylation levels. These results suggest that ablation of Nmp4, in concert with PTH treatment, improved bone function by modulating bone structure and matrix composition. Our findings demonstrate the potential utility of targeting Nmp4 to improve PTH potency and bone quality.

Graphical abstract

1. INTRODUCTION

Osteoporosis is a skeletal disorder characterized by declining bone mineral density (BMD) and bone mass (1), clinically defined as a T-score of 2.5 standard deviations below the BMD mean of the young reference population (2). Individuals with osteoporosis are at an increased risk of bone fractures (3), imposing significant social and economic burdens (4,5). Hence, effective therapeutic strategies for the treatment of osteoporosis are urgently needed.

Antiresorptive agents, such as bisphosphonates, which attenuate bone resorption by inhibiting osteoclast activities, are the first-line treatment for osteoporosis prevention (6,7). However, side effects associated with long-term bisphosphonate use, such as excessive suppression of bone turnover, may limit their use (8,9).

The alternative therapeutic strategy to mitigate the potential disadvantages of antiresorptive agents is anabolic therapy, which increases bone formation by acting on the osteoblasts. Anabolic therapies are typically used as the initial therapy for severe cases of osteoporosis and fracture healing (10). The first osteoanabolic agent approved by the Food and Drug Administration (FDA) was parathyroid hormone (PTH) treatment (11). The commercially available PTH treatments include teriparatide (PTH 1–34), a PTH analog, and abaloparatide (PTHrp 1–34), a synthetic analog of human PTH-related protein (12,13). PTH exerts its biological effects by binding to the parathyroid hormone 1 receptor (PTH1R) on osteoblasts, leading to increased numbers of osteoprogenitor cells and osteoblasts (14–18). In addition to an increase in the number of osteoblasts, PTH enhances the production of bone matrix proteins such as type I collagen, serving as a scaffold for bone mineral deposition (19), and osteocalcin (OC), a non-collagenous protein (NCP), playing a pivotal role in bone mineralization (19,20). PTH treatment increases bone mass (21,22). Improves bone microstructure (23–25) and enhances mechanical functionality (24). However, the potency of the osteoanabolic effect declines over time, as the anabolic window exhibited by PTH, during which bone formation rate outpaces bone resorption rate, eventually closes (15,26,27), thereby posing a significant challenge for treating chronic degenerative bone disorders, where long-term, sustained treatments are required.

Several recent studies have investigated various approaches to improve the anabolic potency of PTH analogs to maximize their long-term benefits and enhance their efficacy in treating osteoporosis. These attempts included PTH treatment in combination with other pharmaceutical agents to improve therapeutic outcomes (26). To this end, it was identified that the nuclear matrix protein 4 (Nmp4, Zfp384, ZNF384, Ciz), a transcription factor, suppresses the PTH-induced osteoanabolic activity (23,28,29). Nmp4 represses the gene expression that regulates protein production, secretion, and the unfolded protein response (UPR) (24,30,31). The suppression of UPR by Nmp4 might be responsible for the limited anabolic window with PTH treatment. (31) Mice with a genetic knockout of Nmp4 exhibited heightened osteoanabolism (23,28,29), as assessed by the increased mRNA level of several bone matrix proteins such as type I collagen, OC, and osteopontin (OPN) (24,32). OPN, a bone matrix non-collagenous protein (NCP), regulates mineral size and shape (20). OPN forms high-affinity complexes with OC through calcium bridges (33), facilitating mineralization by interacting with mineral and collagen and impacting bone matrix toughness (34). The effects of PTH treatment have been previously investigated in the Nmp4^−/− mouse model, with particular focus on the transcriptional changes during osteogenic differentiation, and the relevant molecular and cellular activities involved in osteogenic metabolism in Nmp4 ablation (23,24,28,29,31,32,35). However, little is known about bone quality beyond BMD and its influence on fracture resistance in this model.

Building upon this line of investigations and to address the above knowledge gap, our current work extends insights into the role of the combination of Nmp4 ablation and PTH treatment in cortical bone matrix composition and quality. It evaluates whether alterations in matrix composition translate into improved bone functional performance. To this end, we investigated the bone quality in wild-type (WT) and Nmp4^−/− mice following eight weeks of PTH or vehicle treatment. Our analyses for bone quality included skeletal morphology; bone mineralization and extracellular matrix characteristics, including NCPs (OC, OPN) posttranslational modifications (Total fluorescent Advanced Glycation end-products [Total fAGEs], Carboxymethyl-lysine [CML], matrix phosphorylation); and mechanical properties including whole bone strength testing (tissue-level assessment) and fracture toughness testing (material-level assessment).

2. MATERIALS AND METHODS

2.1. Experimental Model

Wild-type (WT) and Nmp4^−/−(KO) mouse femora were obtained from the Indiana University Bioresearch Facility, School of Dentistry. The approach of engineering global Nmp4^−/− animals has been previously described in detail (23,36). Briefly, exons 4–7 of the gene were removed by homologous recombination. The appropriately targeted embryonic stem cell lines from 129SvEv embryonic stem clones were microinjected into C57BL/6J blastocysts, and the resultant chimeric mice were crossbred with C57BL/6J mice (The Jackson Laboratory, Bar Harbor, ME) to generate germline transmission. The mice were housed in two to four mice/cage, under a 12-hour light and dark cycle. Female WT and Nmp4^−/− mice (16 weeks of age) were grouped into four cohorts: (i) WT-Veh, (ii) WT-PTH, (iii) KO-Veh, (iv) KO-PTH (n = 10 per group). Mice were administered human parathyroid hormone (1–34) [hPTH (1–34)] acetate salt (Bachem Americas, Torrance, CA, USA) at 30 μg/kg/day, daily or vehicle control (0.2% BSA/0.1% 1.0 mN HCl in saline, Abbott Laboratories, North Chicago, IL) for eight weeks, a dosage selected based on prior studies (23,28,29). After euthanasia, mice femora and tibiae were dissected and collected, wrapped in gauze with standard saline solution, and stored in Eppendorf tubes at −80°C until testing.

2.2. Experimental Workflow

Two femora, obtained from each mouse, were allocated to separate experimental pipelines. One femur from each mouse was randomly selected and used for whole-bone strength testing. These same femora were subsequently used to quantify total fAGEs. The contralateral femur from each mouse was first analyzed by micro-computed tomography for cortical morphology, followed by fracture toughness testing. After fracture toughness testing, these cortical bone segments were subsequently allocated for compositional analyses, including Raman spectroscopy, matrix protein phosphorylation assays, and X-ray diffraction. Osteocalcin and osteopontin quantification were performed using tibiae rather than femora, as femoral samples had been fully allocated to the aforementioned experiments. Tibiae were obtained from mice with the same genotype and treatment groups; however, these samples were not necessarily derived from the same individual mice used for other analyses, and the number of tibiae used here (n=9) differed from that of femora used in different experiments (n=10).

2.3. Micro-computed Tomography

The diaphyses of the collected femurs were scanned by micro-computed tomography (micro-CT) following the guidelines of Bouxsein et al. (37). Images were collected at 100 kV, 200 μA, and 850 ms integration time at a resolution of 10.019 μm voxel size (Bruker, Skyscan 1276, Switzerland). The cross-sectional geometry of the mid-diaphyseal cortical bone was determined using a volume of interest extending 40 slices from the center of mid-diaphysis to the lateral condyle (BoneJ, version 1.3.3)(38). Cortical thickness (mm), tissue area and marrow area (Tt.Ar and Ma.Ar, mm²), moment of inertia (mm⁴), section modulus (mm³), and tissue mineral density (TMD, mg HA/cm³) were determined from the reconstructed 2D and 3D images.

2.4. Biomechanical Testing

Whole-bone strength testing is a procedure that determines a bone’s load-bearing and deformation capacities until it fails. The femora were subjected to load until failure under wet conditions using displacement-controlled three-point bending (Instron 5543, Norwood, MA) at a rate of 0.025 mm/s. The samples were positioned such that the anterior cortex of the bones was subjected to tension and the posterior to compression. The loading span was approximately 50% of the average measured femoral length. The resulting load-displacement curves were used to determine structural metrics, including stiffness (N/mm), maximum load (N), post-yield deflection (PYD, mm), and work to fracture (W_f, N·mm). A 10% secant line was used to determine the yield point as previously described (39).

Fracture toughness testing characterizes bone’s material-level resistance to crack initiation and growth with a pre-existing stress concentration. With both epiphyses removed, a mid-diaphyseal notch was created on the anterior of each femur through the cortical wall to guide crack initiation (40). Femoral samples were subjected to load until failure under wet conditions using a displacement-controlled three-point bending setting (ELF 3200; TA instrument, New Castle, DE, USA) at a ramp speed of 0.001 mm/s. The sample positioning was consistent with the whole bone strength testing detailed above. The load span was set to 50% of the average diaphysis length. The load-displacement curves were used to calculate material-level fracture toughness metrics (37,41), such as initiation toughness (K_in, MN•m^−3/2), maximum toughness (K_max, MN•m^−3/2), toughening effect (ΔK, MN•m^−3/2), and cracking toughness at maximum load (K_cracking, MN•m⁻²). Initiation toughness is defined as the critical stress intensity factor required to initiate a crack from the notch. Maximum toughness measures stress intensity at the crack tip during propagation at maximum load. The toughening effect is defined as the difference between initiation and maximum toughness and provides a measure of the dissipated energy during crack propagation (42). Cracking toughness at maximum load, a measure of the total energy required to initiate and propagate a crack, is defined as the span-adjusted work to maximum load normalized by the uncracked area (43).

2.5. Raman Spectroscopy

Raman spectroscopy, used to characterize the bone matrix composition, was performed on a Renishaw S2000 Raman spectrometer (Renishaw, Dundee, IL, US) equipped with a 785 nm laser and a 1200 g/cm grating. Three Raman spectra were obtained from the posterior surface of femurs using a 50x objective (10s exposure time, five accumulations per spectrum). Baseline correction and Savitzky-Golay filtering (2^nd order, window size 21) were conducted using a custom MATLAB script (R2021b, MathWorks, Natick, MA, USA). Peak deconvolution and the derived peak intensities and width were also determined in MATLAB. The evaluated metrics include mineral crystallinity (1/FWHM(ν₁PO₄³⁻)), type-B carbonate substitution (carbonate-to-phosphate ratio, ν₁CO₃²⁻ / ν₁PO₄³⁻ (I₁₀₇₀ / I₉₆₀)), mineral-to-matrix ratio (ν₂PO₄³⁻ / Amide III (I₄₃₀ /I₁₂₄₀), carboxymethyl-lysine (CML (I₁₁₅₀ / I₁₄₅₀)), as described previously (44).

2.6. X-ray Diffraction

For each femur sample, a fraction of the bone was cleaned, defatted, powdered, and evaluated using X-ray diffraction (XRD) to characterize the physical traits of the bone mineral. Diffractograms were acquired with the Panalytical X’Pert Diffractometer (Malvern Panalytical, Malvern, United Kingdom). The diffractometer was run at an operating voltage and current of 45 kV and 40 mA, respectively, using a Copper (Cu) Kα radiation of wavelength λ = 1.5406 Å. Diffractograms were captured from 20° to 55° on a 2θ scale with an incremental step size of 0.01° and a step count of 499 seconds. The subsequent analysis of diffractograms was conducted using the HighScore software (Malvern Panalytical, Malvern, United Kingdom). Before peak fitting, the diffractograms were baseline-corrected. The 002 reflection, linked to the c-axis of the bone mineral crystal, was analyzed for mineral lattice spacing (the interplanar spacing between crystal lattice) based on the peak position using Bragg’s law. The mineral crystal size was calculated based on the full width at half maximum (FWHM) of the peak using the Scherrer equation, as previously reported (45–48).

2.7. Measurement of Phosphorylation of Bone Matrix Protein

To quantify bone protein phosphorylation levels, cleaned and powdered femora were used to extract bone extracellular matrix proteins (49–51). The extraction buffer (0.05M EDTA, 4M guanidine chloride, 30mM Tris-HCl, 15% glycerol, and 10 μL/mL of 100X concentrated Halt Protease Inhibitor) was added directly into the tubes containing powdered bone samples. Protein extraction was carried out at 4 °C for 72 hours. The protein extracts were collected and subjected to buffer exchange with Tris-buffered saline (25 mM Tris, 0.15M NaCl, pH 7.2). The buffer exchange was performed for six cycles (20 min each) in a centrifuge set at 14000 rpm and 4°C. Protein extracts, which were removed from undesired salts during buffer exchange, were collected to measure phosphorylation. Quantification of phosphoserine and phosphothreonine (Pierce^™ Phosphoprotein Phosphate Estimation Assay Kit; Thermo Scientific Co; Waltham, MA), which account for the vast majority of phosphorylation sites (52–54), was performed according to the kit protocol, and normalized to the total amount of bone matrix protein content, measured with the kit (Pierce^™ Coomassie Plus (Bradford) Assay Kit; Thermo Scientific Co; Waltham, MA).

2.8. Measurement of Fluorescent Advanced Glycation End-products (fAGEs)

After mechanical testing, a section of cortical bone tissue from each femur underwent demineralization and protein extraction at 4°C for 72 hours, as described above. The obtained hydrolysates were used to quantify total fluorescent advanced glycation end-products (fAGEs), comprising two components, according to previously established protocols (55,56). The first assay measured quinine sulphate fluorescence. In the assay, a quinine stock solution (1 μg Quinine/mL in 0.1 M H₂SO₄) was diluted with sulfuric acid to generate a standard curve. The fluorescence of each sample and standard hydrolysate was quantified using a spectrophotometer (Infinite 200, Tecan Trading AG, Mannedorf, Switzerland) with an excitation wavelength of 360 nm and an emission wavelength of 460 nm. The other component was to determine bone matrix collagen content by measuring hydroxyproline concentration. In the assay, a hydroxyproline stock solution (2000 μg L-hydroxyproline/mL in 0.001 M HCl) was used to dilute with ultrapure water to generate a standard curve. Every sample and standard was incubated at room temperature following the addition of the chloramine-T solution. Subsequently, the 3.15 M perchloric acid solution was added, and the mixture was further incubated at room temperature. Finally, the p-dimethylaminobenzaldehyde (DMAB) solution was added, and the samples were incubated at 60°C. The absorbance of each sample and standard was measured at 570 nm using the same spectrophotometer. The total fAGEs were presented as units of fluorescent quinine per unit of collagen content (Quinine/Collagen [ng/mg]).

2.9. Measurement of Osteocalcin, Osteopontin, and Total Bone Matrix Protein

The levels of two non-collagenous proteins, osteocalcin (OC) and osteopontin (OPN), were extracted from tibiae in mice (n=9 per group). The protein extraction was the same as described above, except for the EDTA concentration in the extraction buffer. Here, 0.5 M EDTA was used in the extraction buffer to achieve a higher yield of total extracted proteins. In contrast to the 0.05 M EDTA extraction method, 0.5 M EDTA extraction yielded protein levels consistent with those reported in previous studies (24,32). Detection and quantitation of OC (Instant ELISA Kit for Osteocalcin (OC); Cloud-Clone Corp; Katy, TX), OPN (Osteopontin (OPN/SPP1) Mouse ELISA Kit; Thermo Scientific Co; Waltham, MA) were performed according to the protocols provided with the kits. Because treatment with PTH and Nmp4 ablation is expected to significantly alter total protein content, OC and OPN levels were normalized to dry bone weight.

2.10. Statistical Analysis

All statistical analyses were performed in Minitab 19 software (State College, PA). Data were tested for normality and equal variance before statistical analysis. When the data were not normally distributed, a Box-Cox transformation with the optimal λ was applied to achieve a normal distribution. The nonparametric Kruskal-Wallis test was used for data that failed the equal-variance assumption. Data that met normality and equal-variance assumptions were analyzed with a two-way ANOVA to determine the effects of genotype (WT and Nmp4^−/−), treatment (vehicle and PTH), and their interaction. Fisher’s LSD (for parametric multiple comparisons) and Dunn’s test (for non-parametric multiple comparisons) were used as post hoc tests to identify group differences. Data for all parameters were displayed either as Box-and-Whisker plots with median and interquartile ranges or in tables. Pearson’s correlation was determined between parameters to identify associations between measurements. Simple and multivariate linear regressions were conducted to assess the predictive power of structural and matrix compositional parameters for mechanical properties. Outliers were identified using a statistical threshold based on the dataset’s distribution. Specifically, each data point outside the range of the mean ± 2 standard deviations was classified as an outlier and excluded from further analysis. Statistically significant correlations and regressions were reported with corresponding Pearson’s R, coefficient of determination, and p-value, respectively. In all analyses, p-values < 0.05 were considered statistically significant.

3. RESULTS

3.1. PTH treatment and Nmp4 ablation led to improved femoral cortical morphology.

Femoral cortical morphologies were evaluated by micro-CT (Table.1). Tissue area was significantly increased in the femora of PTH-treated (p<0.001, Table.1) and Nmp4^−/−(p=0.026, Table.1) groups compared to those of the WT littermates. Marrow area was significantly increased in the femora of PTH-treated (p=0.009, Table 1). Samples of PTH-treated mice showed increased cortical thickness (p=0.002, Table.1) and expanded cortical area (p<0.001, Table.1). Moreover, maximum moment of inertia (I_max) was significantly elevated in the samples of PTH-treated (p<0.001, Table 1) and Nmp4^−/− (p=0.003, Table 1) mice compared to those of the WT littermates.

Table.1.

Femoral Cortical Morphology assessed by micro-CT

Parameters	WT	KO	Genotype	Treatment	Interaction
TMD (mg HA/ mm3)	Veh=861.33±30.43	Veh=849.8±16.43	p=0.644	p=0.829	p=0.303
	PTH=851.7±29.84	PTH=856.11±10.58
Tt.Ar (mm2)	Veh=1.77±0.252	Veh=1.80±0.095	p=0.026	p<0.001	p=0.236
	PTH=1.93±0.089	PTH=1.97±0.06	WT<KO	Veh<PTH
Ma.Ar (mm2)	Veh=0.860±0.137	Veh=0.907±0.068	p=0.177	p=0.009	P=0.010
	PTH=0.936±0.072	PTH=0.908±0.031		Veh<PTH
Ct.Th (mm)	Veh=0.22±0.017	Veh=0.21±0.019	p=0.228	p=0.002	p=0.256
	PTH=0.226±0.017	PTH=0.23±0.013		Veh<PTH
Ct.Ar (mm2)	Veh=0.845±0.063	Veh=0.81±0.063	p=0.104	p<0.001	p=0.120
	PTH=0.916±0.073	PTH=0.938±0.048		Veh<PTH
Imax (mm4)	Veh=0.257±0.114	Veh=0.274±0.067	p=0.003	p<0.001	p=0.385
	PTH=0.309±0.062	PTH=0.306±0.028	WT<KO	Veh<PTH
Imin (mm4)	Veh=0.147±0.032	Veh=0.158±0.022	p=0.164	p<0.001	p=0.921
	PTH=0.181±0.015	PTH=0.192±0.016		Veh<PTH
Zmax (mm3)	Veh=0.267±0.073	Veh=0.248±0.058	p=0.869	p=0.058	p=0.700
	PTH=0.277±0.035	PTH=0.269±0.034
Zmin (mm3)	Veh=0.205±0.039	Veh=0.176±0.048	p=0.089	p= 0.086	p=0.689
	PTH=0.212±0.042	PTH=0.205±0.049

PTH treatment and Nmp4 ablation led to improved femoral cortical morphology, demonstrated by micro-CT. Results are shown as mean ± SD. Statistically significant differences were performed by two-factor ANOVA with replication (α < 0.05), n=10 mice/group. WT=wild type; KO=Nmp4^−/−; Veh=vehicle; PTH=PTH treatment. TMD=tissue mineral density; Tt.Ar=tissue area; Ma.Ar=marrow area; Ct.Th=cortical thickness; Ct.Ar=cortical area; I_max and I_min are the maximum and minimum moment of inertia at the cross section; Z_max and Z_min are the maximum and minimum section modulus at the cross section.

3.2. PTH treatment enhanced tissue-level stiffness, maximum load, and material-level fracture resistance.

In whole-bone strength testing, higher stiffness (p=0.004, Table.2) and maximum load (p<0.001, Table 2) were observed in the femora of PTH-treated mice compared with those of their vehicle-treated littermates. Reduced ultimate stress (p=0.025, Table.2) and elevated post-yield deflection (p=0.006, Table 2) were observed in the femora of Nmp4^−/− mice. From fracture toughness testing, toughening effect was significantly higher in the samples of PTH-treated groups (p=0.05, Fig.1C). This treatment-specific enhancement was magnified within the femora of Nmp4^−/− groups (p=0.034, Fig.1C). Cracking toughness at maximum load was also higher in the femora of PTH-treated groups (p=0.012, Fig.1D), with the femora of KO-PTH mouse group exhibiting the highest toughness relative to those of KO-Veh littermates (p=0.0013, Fig.1D). While neither PTH nor genotype effects were statistically significant in initiation toughness (Fig.1A) and maximum toughness (Fig.1B), subsequent post hoc analyses indicated increased maximum toughness (p=0.031, Fig.1B) in KO-PTH over KO-Veh, and attenuated maximum toughness (p=0.01, Fig.1B) in KO-Veh over WT littermates.

Table.2.

Femoral bone mechanical properties from strength testing

Parameters	WT	KO	Genotype	Treatment	Interaction
Stiffness (N/mm)	Veh=64.35±10.23	Veh=64.02±7.63	p=0.152	p=0.004	p=0.120
	PTH=68.21±4.87	PTH=75.91±8.23		Veh<PTH
Maximum Load (N)	Veh=16.44±0.44	Veh=15.81±1.75	p=0.889	p<0.001	p=0.153
	PTH=17.91±1.02	PTH=18.49±1.21		Veh<PTH
Ultimate Stress (MPa)	Veh=140.27±18.95	Veh=117.64±20.52	p=0.025	p=0.446	p=0.266
	PTH=137.93±19.04	PTH=129.98±16.63	WT>KO
Elastic Modulus (GPa)	Veh=4.14±0.88	Veh=3.44±0.65	p=0.155	p=0.753	p=0.270
	PTH=3.75±0.70	PTH=3.66±1.00
Work-to-fracture (N*mm)	Veh=1.00±0.47	Veh=0.95±0.28	p=0.948	P=0.083	p=0.613
	PTH=1.21±0.48	PTH=1.11±0.32
Post-yield Deflection (mm)	Veh=0.22±0.07	Veh=0.63±0.39	p=0.006	P=0.129	N/A
	PTH=0.49±0.16	PTH=0.55±0.21	WT<KO

PTH treatment enhanced tissue-level stiffness, maximum load, Nmp4 ablation improved post-yield deflection, demonstrated by whole-bone strength testing. Results are shown as mean ± SD. Statistically significant differences were performed by two-factor ANOVA or Kruskal-Wallis test with replication (α < 0.05), n=10 mice/ group, WT=wild type; KO=Nmp4^−/−; Veh=vehicle; PTH=PTH treatment.

Figure 1:

PTH treatment enhanced bone fracture resistance, demonstrated by fracture toughness testing. (A) Initiation toughness. (B) Maximum toughness. (C) Toughening effect. (D) Cracking toughness at maximum load. Results are shown as box plots with median and interquartile range. Statistically significant differences were determined by two-way ANOVA with replication (α < 0.05) to discern genotype, treatment, and interactions effect, with Fisher’s least significant difference (LSD) test (95% confidence interval [CI]) as post hoc tests in pairwise comparisons, or by Kruskal–Wallis (α < 0.05) test to discern genotype and treatment effects, with Dunn’s test (95% CI) as post hoc tests in pairwise comparisons, n=10 mice/group. WT=wild type; KO=Nmp4^−/−; Veh=vehicle; PTH=PTH treatment.

3.3. PTH treatment and Nmp4 ablation functioned synergistically to alter bone mineralization

Analyzed using Raman spectroscopy, both PTH treatment (p<0.001, Fig.2A) and the Nmp4 deletion (p=0.012, Fig.2A) led to significantly lower crystallinity, with the KO-PTH group demonstrating the lowest level of crystallinity across all groups. Mineral-to-matrix ratio (MMR) was significantly lowered with Nmp4 ablation (p=0.005, Fig.2B), and post hoc analyses confirmed that MMR in the KO-PTH group was lower than MMR of the WT-Veh group (p=0.008, Fig.2B) and the WT-PTH group (p=0.008, Fig.2B). Although no genotype and treatment effects were statistically significant, the post hoc test showed that crystal size was significantly smaller in KO-Veh (p=0.023, Fig.2C) compared to the WT-Veh. Both Nmp4 knockout (p<0.001, Fig. 2D) and PTH treatment (p=0.01, Fig. 2D) significantly decreased the interplanar distance, also referred to as the mineral lattice spacing, as measured by XRD. There was no significant difference in type B carbonate substitution between groups (Fig. S1).

Figure 2:

PTH treatment and Nmp4 ablation synergistically altered bone mineralization, demonstrated by Raman spectroscopy and X-ray diffraction. A) Crystallinity. (B) mineral-to-matrix ratio. (C) Mineral crystal size. (D) Mineral lattice spacing. Results are shown as box plots with median and interquartile range. Statistically significant differences were determined by two-way ANOVA with replication (α < 0.05) to discern genotype, treatment, and interactions effect, with Fisher’s least significant difference (LSD) test (95% confidence interval [CI]) as post hoc tests in pairwise comparisons, n=10 mice/group. WT=wild type; KO=Nmp4^−/−

3.4. Nmp4 ablation and PTH treatment altered the amount of OC, OPN, and total protein in the bone matrix.

The level of OC was significantly increased in bones from Nmp4^−/− mouse groups (p<0.001, Fig.3A) compared to bone samples from WT mouse groups. Post hoc analyses verified that the OC level in the KO-PTH group was higher than that in the KO-Veh group (p=0.01, Fig.3A), WT-PTH group (p<0.001, Fig.3A), and WT-Veh group (p=0.005, Fig.3A). The level of OPN was increased in PTH-treated mouse groups (p=0.046, Fig.3 B) but was not different between genotypes. The total bone matrix protein level was increased in bones from Nmp4^−/− mouse groups (p=0.001; Fig. S2), but was not different with or without PTH treatment.

Figure 3:

Nmp4 ablation and PTH treatment altered the levels of bone matrix OC, OPN, demonstrated by ELISA. (A) OC. (B) OPN. Results are shown as box plots with median and interquartile range. Statistically significant differences were determined by two-way ANOVA with replication (α < 0.05) to discern genotype, treatment, and interactions effect, with Fisher’s least significant difference (LSD) test (95% confidence interval [CI]) as post hoc tests in pairwise comparisons, or by Kruskal–Wallis (α < 0.05) test to discern genotype and treatment effects, with Dunn’s test (95% CI) as post hoc tests in pairwise comparisons, n=9 mice/group. WT=wild type; KO=Nmp4−/−; Veh=vehicle; PTH=PTH treatment.; Veh=vehicle; PTH=PTH treatment.

3.5. PTH treatment increased phosphorylation of the bone extracellular matrix protein

Bone matrix protein phosphorylation was significantly higher in the PTH-treated groups (p=0.001, Fig.4) than in the vehicle groups. Post hoc tests further substantiated that the phosphorylation level in the WT-PTH group was higher than that in the WT-Veh group (p=0.023, Fig.4). In addition, the phosphorylation level in the KO-PTH group was higher than that in the KO-Veh group (p=0.0012, Fig.4) and the WT-Veh group (p=0.0027, Fig.4).

Figure 4:

PTH increased phosphorylation of extracellular matrix proteins, demonstrated by phosphoprotein phosphate estimation assay. Results are shown as box plots with median and interquartile range. Statistically significant differences were determined by Kruskal–Wallis (α < 0.05) test to discern genotype and treatment effects, with Dunn’s test (95% CI) as post hoc tests in pairwise comparisons, n=10 mice/group. WT=wild type; KO=Nmp4−/−; Veh=vehicle; PTH=PTH treatment.

3.6. PTH treatment and Nmp4 ablation altered the level of AGEs/AGOEs accumulation in bone matrix

While neither genotype nor treatment showed significant differences for CML, the post hoc tests indicated reduced accumulation of CML in the KO-PTH group compared with the WT-Veh group (p=0.014, Fig.5A). Total fAGEs were significantly lower in femora from Nmp4^−/− groups (p=0.024, Fig.5B), as compared to WT groups, and post hoc tests further substantiated that the total fAGEs level in the WT-Veh group was higher than the KO-Veh group (p=0.006, Fig.5B).

Figure 5:

PTH treatment and Nmp4 ablation altered bone matrix AGEs/AGOEs accumulation, demonstrated by Raman spectroscopy, fluorescence assay, and hydroxyproline assay. (A) Carboxymethyl-lysine. (B) Total fluorescent AGEs. Results are shown as box plots with median and interquartile range. Statistically significant differences were determined by two-way ANOVA with replication (α < 0.05) to discern genotype, treatment, and interactions effect, with Fisher’s least significant difference (LSD) test (95% confidence interval [CI]) as post hoc tests in pairwise comparisons, n=10 mice/group. WT=wild type; KO=Nmp4−/−; Veh=vehicle; PTH=PTH treatment.

3.7. Alterations in structural and compositional parameters following PTH treatment and Nmp4 ablation contributed to functional outcomes.

Based on linear regression analyses (Table 3), greater bone fracture toughness, indicated by toughening effect, was significantly associated with larger mineral crystal size (R² = 0.18, p = 0.01) and reduced crystallinity (R² = 0.15, p = 0.03). While a negative association was observed between the toughening effect and CML levels, the correlation did not reach statistical significance (R² = 0.10, p = 0.07). Increased work-to-fracture from whole bone strength testing was associated with greater tissue area (R² = 0.33, p < 0.001), lower CML levels (R² = 0.28, p = 0.001), and reduced mineral lattice spacing (R² = 0.15, p = 0.02). Increased post-yield deflection was significantly associated with greater tissue area (R² = 0.34, p <0.001), marrow area (R² = 0.32, p <0.001), lower CML levels (R² = 0.27, p = 0.002), and reduced mineral lattice spacing (R² = 0.13, p = 0.04). The levels of bone protein phosphorylation and Total fAGEs did not predict any mechanical outcomes.

Table.3.

Summary of the simple / multivariate linear regression analyses showing linear combinations of three mechanical properties and key predictive variables.

Mechanical Properties	Explanatory Variables	p value	Adjusted R2
toughening effect (fracture toughness test)	mineral crystal size	0.01	0.18
	crystallinity	0.03	0.15
	CML	0.07	0.10
	mineral crystal size + crystallinity	0.005	0.29
	mineral crystal size + crystallinity + CML	0.004	0.37
work-to-fracture (strength test)	Tt.Ar	<0.001	0.33
	CML	0.001	0.28
	mineral lattice spacing	0.02	0.15
	Tr.Ar + CML+mineral lattice spacing	<0.001	0.51
post-yield deflection (strength test)	Tt.Ar	<0.001	0.34
	Ma.Ar	<0.001	0.32
	CML	0.002	0.27
	mineral lattice spacing	0.04	0.13
	Tt.Ar + CML	0.001	0.51
	Tt.Ar + CML + Ma.Ar	<0.001	0.54
	Tt.Ar + CML + Ma.Ar+mineral lattice spacing	<0.001	0.60

Alterations in structural and compositional parameters following PTH treatment and Nmp4 ablation contributed to functional outcomes. The adjusted R² values are highlighted in bold to indicate significant linear regressions. The optimal variable combination for explaining each mechanical property is presented at the bottom of their respective sections. Variables not contributing to predict mechanical properties and models for simple / multivariate linear regressions are not shown.

In multivariate linear regression analyses (Table 3), the explanatory variables of mineral crystal size, crystallinity, and CML levels provided the best model for explaining the toughening effect (R² = 0.37). For work-to-fracture, three variables — tissue area, CML levels, and mineral lattice spacing — offered the most robust explanation (R² = 0.51). Post-yield deflection was best explained by the variable combination of tissue area, marrow area, CML levels, and mineral lattice spacing (R² = 0.60).

4. DISCUSSION

We observed that PTH treatment and Nmp4 ablation independently promoted the tissue area expansion, as demonstrated by the micro-CT results. The concomitant elevated cortical area and marrow area suggested an increase in both bone formation and bone resorption. Evidently, a larger moment of inertia and cortical area increased bone’s bending strength and energy absorption, i.e., work-to-fracture, and deformation before fracture, i.e., post-yield deflection. Notably, tissue mineral density did not differ between the groups, which is most likely due to the young age of the mice (24 weeks old), when age-related bone loss has not yet begun. While this model does not represent an osteoporosis condition, it is still appropriate to investigate how the loss of Nmp4 influences the effect of PTH treatment on bone architecture and quality.

Assessed by Raman Spectroscopy, we observed alterations in mineral quality due to PTH treatment and Nmp4 deletion. PTH treatment significantly lowered crystallinity. This observation aligns with earlier studies (57,58). It can be interpreted as PTH treatment promoting new bone formation, leading to an overall decrease in bone mineral structural and stoichiometric perfection (59,60). Interestingly, the reduction of crystallinity due to PTH treatment was more remarkable in the Nmp4^−/− group. This further implies the enhanced capability to form new mineral crystals with Nmp4 deletion. The newer crystals with a lower degree of perfection might lead to more microstructural irregularities, such as grain boundaries and dislocations, potentially enhancing energy dissipation during crack propagation. Indeed, in this study, reduced mineral crystallinity was predicted to strengthen toughening and work-to-fracture, corroborating the improvement in bone quality.

In the Nmp4^−/− group with PTH treatment, we also observed reduced MMR compared to untreated WT. It has demonstrated an elevation in the mRNA expression of type I collagen in mesenchymal stem progenitor cells derived from Nmp4^−/− mice; hence, this observed increase is linked to the enhanced bone organic matrix content, aligning with our findings of increased total bone matrix protein in Nmp4^−/− mice and explaining the observed reduction in the MMR in our study. Although no significant association with bone mechanical outcomes was observed here, a previous study reported that reduced MMR with PTH treatment was correlated with higher tissue strength (57).

While previous studies on Nmp4 deletion examined the mRNA expression of OC, OPN, and type I collagen in an in vitro model (24,32), as well as the serum concentration of OC in an in vivo animal model (35), our study quantified OC and OPN in bone matrix for the first time in a Nmp4^−/− mouse model with PTH treatment. A prior study demonstrated an enhanced and sustained PTH-induced increase in serum OC in Nmp4^−/− mice over a 7-week treatment period. In contrast, WT mice exhibited a transient surge in serum OC that began to decline in the third week of treatment (28). Consistent with these observations, our findings indicate that PTH treatment and Nmp4 ablation increase bone OC content, with the KO-PTH group showing a significant increase in bone matrix OC. In contrast, the WT-PTH group showed a decreasing trend in OC levels compared to their vehicle-treated counterparts. These findings suggest that Nmp4 ablation enhances the osteoanabolic effects of PTH treatment by sustainably elevating OC levels in bone, potentially extending the anabolic window of this therapy. While Nmp4 ablation did not significantly affect OPN content in bone, we observed that PTH treatment increased OPN levels; this observation is consistent with previous findings on PTH’s influence on OPN (61,62). Earlier studies also showed that serum OPN levels are positively associated with increased bone formation and resorption (63), consistent with the effects of PTH treatment on bone.

Our previous work has shown that OC and OPN are vital for nanoscale energy dissipation (64). Nonetheless, in this study, correlations between fracture toughness and OC and OPN levels were not explored, as the bone samples used to measure OC and OPN were not obtained from the same mice used for fracture toughness testing. Consequently, the link between OC and OPN levels and energy dissipation should be investigated in future work.

Furthermore, we observed that PTH increased the degree of phosphorylation for the bone matrix proteins, specifically at serine and threonine residues. Phosphorylation is a fundamental posttranslational modification (PTM) that influences protein-mineral interactions and bone quality. It should be noted that this work measured bulk levels of global phosphorylation in bone total proteins rather than phosphorylation of specific sites. Reduced levels of phosphorylation of bone extracellular matrix proteins are associated with bone fragility (65,66), as calcium-mediated interactions between phosphorylated bone matrix proteins and hydroxyapatite (HA) increase detachment energy, enabling matrix/mineral interaction and facilitating energy dissipation (66,67). As the capacity of energy dissipation is vital to the bone, the degree of phosphorylation of bone matrix proteins should be one factor contributing to bone quality. However, despite increased phosphorylation levels in the WT-PTH and KO-PTH groups and a concomitant increase in cracking toughness in the KO-PTH group relative to untreated groups, the current study did not observe a significant correlation between phosphorylation levels and bone mechanical properties. Future work should investigate the relationship between PTH treatment and bone matrix phosphorylation to explore this potential avenue by which PTH action in Nmp4^−/− model can improve bone quality.

Interestingly, we observed an altered level of CML, an advanced glycoxidative end-product (AGE), which is present in bone tissue at higher levels than other AGEs (50,68). We also observed an altered level of total fluorescent AGEs (fAGEs), measured in bulk using a biochemical assay. AGEs/AGOEs are widely discussed in studies on the effects of aging and hyperglycemia on bone quality and are regarded as markers of skeletal fragility (50,69–74). The accumulation of AGEs/AGOEs, associated with increased oxidative stress (75), modifies bone matrix proteins (56), negatively impacts skeletal turnover (69), and ultimately reduces bone fracture toughness (45,70,76,77). CML, particularly, has been linked to exacerbated bone fragility by altering energy dissipation (50). It has previously been reported that PTH-treated mice displayed reduced mRNA expression of the receptor for advanced glycation end-products (RAGE) (28). RAGE is a critical player in inflammation response and immunomodulation and contributes to inflammation progression by influencing the formation of reactive oxygen species (ROS) (78,79). Elevated ROS triggered by inflammation promotes the accumulation of AGEs/AGOEs in bone (80–82). In this study, CML content in the KO-PTH group was reduced compared to that of the WT-Veh group.

Furthermore, the accumulation of in-bulk fAGEs in WT samples were lower than that measured in the Nmp4^−/− samples. This decrease in the levels of AGEs/AGOEs (CML and fAGEs) could be explained by the reduced level of RAGE following PTH treatment and Nmp4 ablation (28,83). In agreement with previous studies, we observed negative associations between bone CML levels and mechanical outcomes, including post-yield deflection, work-to-fracture, and toughening effects. These associations confirm the detrimental impact of CML on the organic matrix and suggest a mechanistic link between CML accumulation and reduced energy dissipation in bone. Collectively, PTH treatment helps lower AGE accrual, potentially by preventing glycoxidative damage to the bone organic matrix.

Characterized by XRD, the mineral lattice spacing was significantly reduced under the influence of both Nmp4 deletion and PTH treatment. Intriguingly, this alteration in the mineral also corresponded to the changes in CML level, since a positive correlation between CML and the mineral interplanar spacing, or mineral lattice spacing, was discovered. In turn, the interplanar spacing was shown to predict bone mechanical outcomes, such as work-to-fracture and post-yield deflection. The decreased spacing between mineral crystal lattices suggests higher prestress within the mineral phase, enabling energy dissipation as the crack propagates. One potential explanation for this association lies in the reduced AGEs/AGOEs, which would cause fewer disruptions to collagen fiber bundles (77), which, in turn, are known to mediate the growth of HA crystals (84) and thereby influence the physical characteristics of the mineral, i.e., interplanar spacing. It should be noted that, in this study, crystallinity was defined as the inverse of the full width at half maximum (1/FWHM) of the ν1PO₄³⁻ Raman band, which reflects crystal arrangement and lattice disorder rather than lattice packing density (44). In contrast, mineral lattice spacing by XRD shows the interplanar spacing of the bone mineral crystal lattice and can be affected by ionic substitutions, lattice strain, and mineral matrix interactions (85,86). Consequently, changes in mineral lattice spacing and crystallinity are not necessarily coupled and may vary independently.

Additionally, another parameter assessed by XRD, mineral crystal size, was shown to be a significant predictor of bone toughening effect. Consistent with former findings (45,47), the strong association between mineral crystal size and fracture toughness potentially indicates that the combination of PTH treatment and Nmp4 ablation might restore or optimize mineral crystal size, leading bone to be less prone to internal deformation or sliding under stress, which might assist in resisting microcrack propagation (85). Further investigation of the relationship between CML and the mineral phase may provide mechanistic insights into the role of glycoxidation in bone mineral quality and, subsequently, into the enhancement of bone mechanical integrity with osteoanabolic therapies.

A limitation in our study is that the bone turnover markers were not measured. That being said, increased cortical thickness and cortical area indirectly suggested enhanced bone formation in mice with Nmp4 ablation and PTH treatment. Moreover, the aforementioned study by Childress et al. (28) on the same model reported bone turnover measurements. Consistent with our finding, Nmp4 ablation augmented the PTH-induced increase in bone formation, as shown by serum osteocalcin over the 7-week treatment duration. On the other hand, Nmp4 ablation increased and sustained bone resorption, as measured by serum C-terminal telopeptide (CTX), without a significant interaction with treatment. The finding suggested that the combination effects led to a persistently higher bone resorption rate and a progressive increase in bone formation rate throughout the treatment time. In addition, our study includes no in vitro functional assays of osteogenic capacity, such as calcium nodule formation by bone marrow–derived mesenchymal stem cells (BMMSCs). Assessing whether Nmp4 ablation enhances PTH-stimulated mineralized nodule formation in BMMSCs would provide complementary evidence for alterations in OC and OPN associated with cellular-level mineralization capacity (87). Future work could explore these experiments to co-validate the cellular mechanisms underlying the observed improvements in bone quality.

In conclusion, we established that PTH treatment with Nmp4 ablation significantly improved the quality of the bone matrix, as evidenced by altered bone composition (mineral alterations), elevated OC levels, increased matrix phosphorylation, and reduced AGEs/AGOEs. These parameters of bone composition and structure collectively contributed to increased bone fracture resistance, indicative of improved bone quality. This study extends knowledge of Nmp4 ablation in response to osteoanabolic agents and demonstrates how they influence bone compositional quality and mechanical properties at multiple length scales.

Highlights.

• Nmp4 ablation and PTH treatment both improved cortical bone morphology

• Osteocalcin levels increased with PTH treatment in the Nmp4^−/− group

• Mineral and matrix changes were linked to altered mechanical outcomes

DATA AVAILABILITY

Data will be made available upon reasonable request.