Osteoidosis leads to altered differentiation and function of osteoclasts

Abstract In patients with osteomalacia, a defect in bone mineralization leads to changed characteristics of the bone surface. Considering that the properties of the surrounding matrix influence function and differentiation of cells, we aimed to investigate the effect of osteoidosis on differentiation and function of osteoclasts. Based on osteomalacic bone biopsies, a model for osteoidosis in vitro (OIV) was established. Peripheral blood mononuclear cells were differentiated to osteoclasts on mineralized surfaces (MS) as internal control and on OIV. We observed a significantly reduced number of osteoclasts and surface resorption on OIV. Atomic force microscopy revealed a significant effect of the altered degree of mineralization on surface mechanics and an unmasking of collagen fibres on the surface. Indeed, coating of MS with RGD peptides mimicked the resorption phenotype observed in OIV, suggesting that the altered differentiation of osteoclasts on OIV might be associated with an interaction of the cells with amino acid sequences of unmasked extracellular matrix proteins containing RGD sequences. Transcriptome analysis uncovered a strong significant up‐regulation of transmembrane glycoprotein TROP2 in osteoclastic cultures on OIV. TROP2 expression on OIV was also confirmed on the protein level and found on the bone surface of patients with osteomalacia. Taken together, our results show a direct influence of the mineralization state of the extracellular matrix surface on differentiation and function of osteoclasts on this surface which may be important for the pathophysiology of osteomalacia and other bone disorders with changed ratio of osteoid to bone.


| INTRODUC TI ON
Normal bone function requires bone remodelling that is a lifelong and complex process involving bone formation and bone resorption. During this, the osteoclast plays a pivotal role due to its unique ability of bone resorption. Osteoclasts belong to the mononuclear phagocyte system and originate from HSCs through differentiation of CD14+ monocytes. 1,2 During differentiation, osteoclastogenesis requires two essential factors: macrophage colony-stimulating factor (M-CSF) and receptor activation of NF-kB ligand (RANKL). M-CSF mainly promotes proliferation and survival of osteoclast precursors. RANKL is known to function as the primary factor driving differentiation of osteoclast precursors by controlling gene expression by activating its receptor RANK.
Secondary, osteoclast function highly depends on bone-cell interaction and its cytoskeleton organization. Bone-cell interaction is mediated by integrin avß3 which recognizes the RGD sequence that is present in various bone matrix proteins. Interaction of integrin avß3 with the bone matrix induces cytoskeleton organization that leads to polarization of the osteoclast and establishment of the typical resorptive compartment. 3 Any imbalance in the regulation of bone remodelling can result in a metabolic bone disease like osteoporosis, osteopetrosis, osteosclerosis, pycnodysostosis or osteomalacia. [4][5][6][7][8][9] Vitamin D deficiency, in particular, is an increasingly important global health problem across all ages. It leads to changes in the metabolism of calcium that eventually result in decreased mineralization of bone. 10 Based on histological findings, accumulation of unmineralized bone tissue (osteoid) is referred to as osteoidosis. According to Parfitt, the resulting condition with a ratio of osteoid volume to bone volume >10% is called osteomalacia. 11 In a cross-sectional study of a normal German population, osteomalacia occurred in more than 25%, independent of the subjects age. 12 The degree of mineralization largely determines the stiffness of a tissue. Thus, osteoidosis leads to a decreased elastic modulus (Young's modulus) of the affected bone surface. 13 Therefore, it can be supposed that the matrix mechanics for the bone cells which are attached to the bone surface are modified by this disorder. For a variety of cells, it has already been shown that the matrix properties can influence cellular functions, including cell proliferation, locomotion, adhesion, spreading, morphology, striation and differentiation. [14][15][16][17][18][19][20] For haematopoietic stem cells (HSCs) in particular, Lee-Thedieck et al have shown that their migration and adhesion behaviour are dependent on the elastic modulus of the substrate. Similar effects are reported by Holst et al who described an altered proliferation behaviour of HSCs caused by an altered Young's modulus of the surrounding substrate. 21,22 If the differentiation of HSCs can be influenced by the substrate's elasticity, we have been suggested that the differentiation and function of osteoclasts might also be influenced by changes of the bone surface. Therefore, we investigated the effect of matrix mineralization on the differentiation and function of osteoclasts.

| Morphometric analysis of osteomalacia
After fixation in buffered formalin, the tissues were embedded undecalcified in methyl-methacrylate, cut into sections of 5 µm thickness using a k-microtome (Jung) and stained with Goldner trichrome staining. The retrospective analysis of patient's records was performed in compliance with the Hamburg Hospital Law, Germany (HmbKHG, Version: 17th of April, 1991; Second chapter: Patients data protection, Paragraph 12).

| In vitro model of mineralized tissue and OIV
Dentin is widely used as a model for mineralized tissue. It was provided by German customs in accordance with the international laws for the protection of species. Discs of 10.0 × 10.0 × 0.7 mm were prepared by using a diamond saw (PSI Grünewald GmbH & Co. KG).
Physiologically mineralized dentin discs (Mineralized surface, MS) were served as control. To generate an unmineralized matrix and imitate an osteoidosis (OIV), the surface of dentin discs was demineralized by immersion in one molar hydrochloric acid (Carl Roth GmbH & Co. KG) for 1 minute. After demineralization, the dentin was washed thoroughly with distilled water and stored at room temperature (RT) until its use in cell culture.

| Preparation of RGD peptide-coated dentin
Dentin discs were prepared as described above. The cyclic RGD peptide c(RGDfK) (RGD, Novabiochem) was solubilized in MilliQ water (Merck MilliPore) to give a concentration of 100 µmol/L. Each disc was covered with 500 µL RGD solution and incubated for 10 minutes under ambient conditions. Finally, discs were cleaned with 100% isopropanol for 5 minutes and dried for 3 hours inside the laminar flow hood.

| Atomic force microscopy
Atomic force microscopy (AFM) measurements were carried out as previously described. 23 Measurements were carried out on a NanoWizard AFM (JPK Instruments) combined with an Axiovert

| TRAP staining and evaluation of osteoclast numbers
After 28 days of cultivation, cells on dentin discs were fixed with 37% buffered formaldehyde and incubated with TRAP staining solution (Sigma Aldrich). Then, cells were analysed with the Eclipse TS 100 inverted optical microscope combined with a digital camera and the NIS-Elements BR software (Nikon GmbH).
To determine the number of osteoclasts, cells positive for TRAP and with three or more nuclei were counted. The experiment was repeated for three times, each time including six individual replicates.

| Infinite focus microscopy
The three-dimensional resorption activity of osteoclasts on MS, OIV and RGD-coated dentin was quantified by infinite focus microscopy (IFM) as described by us previously. 26 Briefly, cells were detached from the surface after 28 days of cultivation. For three-dimensional quantification of the resorbed volume, three-dimensional surface data sets were generated with voxel sizes of 800 nm × 800 nm × 100 nm using IFM (InfiniteFocus Alicona). To minimize the effect of surface roughness on measured resorption, a roughness exclusion plane was defined as the mean of all depth values of the topographic surface of the untreated material minus two times standard deviation.
Only pits which were deeper than this roughness exclusion plane were included in the analysis. Pseudocolour visualization as well as computation of resorbed volume, resorbed area and depth was performed with the built-in software in combination the freeware image analysis software (UTHSCSA Image Tool V 3.0). Measurements were repeated with three different samples for each surface topography.

| mRNA microarray analysis
Microarray analysis was performed after 28 days of cultivation using the Human Genome U133 Plus 2.0 Arrays (Affymetrix). Procedures for cDNA synthesis, labelling and hybridization were carried out according to the manufacturer's protocol (Affymetrix). In brief, 50 ng of total RNA was used for first-strand cDNA synthesis with an HPLC-purified T7-(dT)24 primer. Synthesis of biotin-labelled cRNA and clean up was carried out using the IVT Express Kit (Affymetrix). Microarrays were scanned with the Affymetrix GeneChip Scanner 7G, and the signals were processed using GCOS (v.1.4; Affymetrix).

| Quantitative realtime PCR analysis
At day 28, RNA isolation was carried out using the High Pure RNA Cell Isolation Kit (Roche Diagnostics GmbH) following the manufacturer's instructions. RNA was then transcribed to cDNA in the iQ5 qPCR cycler by the use of 5× iScript RT Supermix (Bio-Rad Laboratories).
For qPCR analysis, 2 ng of resulting cDNA of each probe was used.
The performed PCR protocol was as follows: one step of 2 minutes at 98°C, 40 primer-specific cycles of 10 seconds at 95°C and 30 seconds at 60°C, a melt curve was generated with 80 cycles of 6 seconds and a starting temperature of 65°C with 0.5°C increments.
TGG-5′, served as reference. In order to quantify the relative change of gene expression, the ΔΔCt-method was used.

| TROP2 immunohistochemistry staining
Cells on MS and OIV were fixed with 3.7% formaldehyde.
Afterwards, cells were incubated for 1 hour at RT with PBS containing 10% BSA, followed by incubation of the primary antibody

| Statistical analysis
All experiments were repeated for three times to allow statistical analysis, except for the mRNA microarray, which was performed once with a pooled sample. Statistical analysis and the generation of graphics were performed using Sigma Plot software. For normally distributed data, oneway ANOVA testing was performed, followed by Tukey post hoc testing, if there were more than two groups. For non-normally distributed data, Mann-Whitney rank sum test was performed. The probability of a type I error was set to 5% (α = .05). Error bars represent SD.

| Model for osteoidosis in vitro
While we could not detect an osteoid seam in the bone biopsy of a healthy patient ( Figure 1A), histomorphometric examination of bone biopsies from patients with osteomalacia showed an osteoid seam of 57.8 ± 6.9 µm ( Figure 1B). In our dentin model (Osteoidosis in vitro, OIV), we generated an unmineralized seam that was in the same order of magnitude ( Figure 1D; 56.5 ± 1.0 µm). In the control specimens (Mineralized surface, MS), no zone of unmineralized matrix was detectable ( Figure 1C).

| Surface analysis
To the exposure of the collagen network due to demineralization. As expected, the distance between the collagen molecules was found to be 63.6 ± 3.6 nm.

| Effects on the number of osteoclasts
In comparison to MS ( Figure 3A), there was a significantly lower number of differentiated osteoclasts after 28 days on OIV ( Figure 3B); MS: 30.08 ± 8.60 osteoclasts/field of view vs OIV: 9.59 ± 1.88 osteoclasts/field of view, ( Figure 3C, P = .03).

| Resorption activity on RGD peptide coating
As collagen is known to directly influence cell behaviour through its high level on RGD peptide, we coated MS with a very thin layer of RGD peptides, without detectable changes of the surface mechanics.

| Analysis of gene expression
To find an explanation for the significantly lower number of osteoclasts as well as the decreased resorption area on OIV, whole genome microarray analysis was performed for both groups after 28 days of cultivation ( Figure 6A). We identified the tumour-associated calcium signal transducer 2 (TROP2) to be the most strongly up-regulated gene on OIV compared with MS which was confirmed by qPCR analysis ( Figure 6B). On OIV, TROP2 showed a significant 15-fold up-regulation ( Figure 6B, MS = 0.12 ± 0.06 vs OIV = 1.85 ± 2.15, P = .008).

| TROP2 + cells on osteoidosis in vitro
TROP2 staining of cells on MS and OIV confirmed our qPCR results on a protein level. While we could not detect TROP2 in osteoclast cultures on MS ( Figure 6C), there was a strong expression in the cells on OIV ( Figure 6D).

| D ISCUSS I ON
We show here that a change of matrix mineralization, which can be found in several metabolic bone diseases, alters the function and differentiation of osteoclasts in vitro. This is most likely due to an exposition of collagen fibrils resulting in an RGD-mediated pathway.
In our model system, we were able to generate a unmineralized seam that was of the same order of magnitude as that in bone biopsies from osteomalacia patients and consistent with Parfitt's definition of osteomalacia referring to a mean osteoid thickness above an absolute value of 12.5 µm. 11 Although the same cell metabolic activity indicating a similar number of progenitor cells on OIV and MS was found at day 1 It has so far not been detected in bone marrow or peripheral blood. 32  The RGD sequence of the exposed collagen fibrils that were identified by their typical cross-striation 36  While the inhibitory effect of RGD-containing proteins such as echistatin has been observed before, it was suggested that the occupation of the vitronection receptor prevented any attachment of osteoclasts to the RGD peptide of the extracellular matrix and thus cell spreading and resorption. [39][40][41] In our experiments, the collagen-containing RGD sequences were strongly attached to the surface, suggesting a direct effect on the precursor cells which challenges the original concept of competitive inhibition.
Another possible influence on cell differentiation could be mediated by surface mechanics. The strongly decreased Young's modulus of OIV may therefore be in part responsible for the expression of a gene that is typically found in much softer tissues. However, when we analysed the expression of typical mechanotransduction genes in our samples, we could find no significant changes between MS and OIV ( Figure S3), making this hypothesis unlikely.
A limitation of our model system is that we used dentin instead of bone. However, despite their different composition and structure, we have shown previously that dentin and human bone are indistinguishable in terms of their osteoclastic resorption activity in vitro. 42   In conclusion, we show that a change of matrix mineralization alters the differentiation and function of human osteoclasts in vitro.
The reduced number of osteoclasts and resorbed surface area was associated with an up-regulation of TROP2 expression on an osteoidosis model as well as in patients with osteomalacia, which has not been described so far. This is possibly due to binding of precursor cells to unmasked extracellular matrix proteins containing RGD sequences, resulting in an RGD peptide-mediated pathway. Based on our results, we suppose that the different surface characteristics of the extracellular matrix may be important for the pathophysiology of several metabolic bone diseases with a defect in matrix mineralization such as osteomalacia, rickets, chronic kidney disease-mineral bone disorder or hypophosphatasia.

ACK N OWLED G EM ENTS
In the past 5 years, AF Schilling has received institutional support from Evonik Industries and consulting fees from Amgen. There are no royalties to disclose.

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.

AUTH O R CO NTR I B UTI O N S
LG, CP, TW, MK, UH, HCS and AFS designed the study and performed the research.
LG, CP, TW and AFS analysed the data and wrote the paper. KK, TW, TS, JZ, MMM and HGM contributed to the establishment of the osteoidosis model. All authors approved the submitted and final version of the paper.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.