Ncf1 affects osteoclast formation but is not critical for postmenopausal bone loss

Background Increased reactive oxygen species and estrogen deficiency contribute to the pathophysiology of postmenopausal osteoporosis. Reactive oxygen species contribute to bone degradation and is necessary for RANKL-induced osteoclast differentiation. In postmenopausal bone loss, reactive oxygen species can also activate immune cells to further enhance bone resorption. Here, we investigated the role of reactive oxygen species in ovariectomy-induced osteoporosis in mice deficient in Ncf1, a subunit for the NADPH oxidase 2 and a well-known regulator of the immune system. Methods B10.Q wild-type (WT) mice and mice with a spontaneous point mutation in the Ncf1-gene (Ncf1*/*) were ovariectomized (ovx) or sham-operated. After 4 weeks, osteoclasts were generated ex vivo, and bone mineral density was measured using peripheral quantitative computed tomography. Lymphocyte populations, macrophages, pre-osteoclasts and intracellular reactive oxygen species were analyzed by flow cytometry. Results After ovx, Ncf1*/*-mice formed fewer osteoclasts ex vivo compared to WT mice. However, trabecular bone mineral density decreased similarly in both genotypes after ovx. Ncf1*/*-mice had a larger population of pre-osteoclasts, whereas lymphocytes were activated to the same extent in both genotypes. Conclusion Ncf1*/*-mice develop fewer osteoclasts after ovx than WT mice. However, irrespective of genotype, bone mineral density decreases after ovx, indicating that a compensatory mechanism retains bone degradation after ovx.


Background
Bone remodeling occurs throughout life, shifting to favor bone resorption in postmenopausal women. Osteoclasts (OCL) resorb bone, and the presence of reactive oxygen species (ROS) at the interface between OCL and bone suggests a role for ROS in bone resorption [1][2][3][4]. ROS are produced by membrane-localized nicotinamide adenine dinucleotide phosphate (NADPH) oxidases (NOX); [5], where NOX 1, 2, and 4 are active in OCL and their myeloid precursor cells [1,4]. The activity of NOX 2 is regulated by the neutrophil cytosolic factor 1 (Ncf1, alias p47 phox ) subunit [6]. ROS from NOX 2 kills bacteria in professional phagocytes (such as macrophages and neutrophils with highly efficient phagocytosis), and damages cells and tissues during inflammation. However, reduced production of ROS in mice lacking Ncf1 promotes T-cell dependent autoimmune diseases [6]. Further, Tcells activated by antigen presenting cells (APCs) induce bone degradation in postmenopausal osteoporosis. After ovariectomy (ovx), the excess tumor necrosis factor alpha (TNFα) from T-cells activates OCL and results in reduced bone density [7][8][9]. Other mechanisms underlying postmenopausal bone loss include increased receptor activator of nuclear factor κ B ligand (RANKL) signaling [10] and increased OCL formation [11].
To investigate the role of ROS and Ncf1 in postmenopausal bone loss, we employed a postmenopausal boneloss model by ovx or sham operating mice with a spontaneous mutation in Ncf1. We investigated osteoclastogenesis ex vivo, used peripheral quantitative computed tomography (pQCT) for measuring bone mineral density (BMD), and flow cytometry was used to investigate pre-OCL and immune cells, including immune cell activation. We found that bone mineral density decreased to the same extent in both Ncf1-deficient mice and wild type (WT) mice, however, the OCL formation and preosteoclast populations were differentially affected in the two genotypes.

Animal procedures
The regional ethical review board in Gothenburg approved this study. The point mutation in Ncf1 [12] has been backcrossed onto the B10Q background (WT) [6] and kept in breeding. WT mice and mice derived from the same background but with a homozygous mutation in the Ncf1-gene (Ncf1*/*) were housed in a temperature-controlled room with a 06.00-18.00 h light cycle and consumed a soy-free diet (R70, Lantmännen, Stockholm, Sweden) and tap water ad libitum.

Ovariectomy and treatment
At 7 to 10 weeks of age, female mice were ovariectomized or sham-operated, under isoflurane (Baxter Medical Ab, Kista, Sweden) anesthetization by inhalation. Carprofen (Orion Pharma AB, Animal Health, Sollentuna, Sweden) was administered subcutaneously as postoperative analgesic.

Osteoclast differentiation
Bone-marrow derived macrophages (BMM) were cultured and stimulated with macrophage colony stimulating factor (M-CSF) and RANKL to induce osteoclast differentiation [13]. Bone marrow was flushed with minimum essential medium Eagle's alpha modification (αMEM; Gibco, Grand Island, NY, USA) supplemented with 10 % heat-inactivated fetal bovine serum (FBS; Sigma-Aldrich®, St Louis, USA). Cells were incubated for 2 days at 37°C in a humidified CO 2 chamber on a culture dish (Corning®, Hickory, NC, USA) with αMEMmedium containing 30 ng/ml M-CSF (R&D system Europe LTD, Abingdon, UK). After two days, the adhering cells (BMMs) were detached, after which 5000 cells in 5 μl αMEM/10 % FBS were seeded at the center of 96well plates and left to adhere for 10 min. Subsequently, 200 μl medium was added containing either 30 ng/ml of M-CSF (controls), or 30 ng/ml M-CSF + 2 ng/ml of RANKL. After 4 days, the cells were fixed and stained for tartrate-resistant acid phosphatase (TRAP). TRAPpositive cells with ≥3 nuclei were considered osteoclasts. The amounts of TRAP-positive multinucleated osteoclasts were calculated by standardized counting of the midsection of a 96-well by Osteomeasure 3.2.1.0 software (Osteometrics, Inc., Decatur, GA, USA) [14]. The total OCL covered area in the well was calculated by tracing the OCL boundaries. Area per individual OCL was calculated by dividing OCL covered area with number of OCL. Three mice per group were used and assayed in quadruplicate.

Assessing BMD
After being kept for 2 days in 4 % PFA and subsequently stored in 70 % ethanol, femurs were scanned using peripheral quantitative computed tomography (pQCT) with Stratec pQCT XCT Research M software (version 5.4B; Norland, Fort Atkinson, WI, USA), at a resolution of 70 μm, as described previously [15]. Cortical thickness was determined by one mid-diaphyseal scan at a distance from the distal growth plate corresponding to 36 % of total femur length. This mid-diaphyseal region of mouse femur only contains cortical bone. Trabecular BMD was determined by one metaphyseal scan at a distance from the distal growth plate corresponding to 3 % of the femur length. The trabecular bone region was set as 45 % of the inner total cross-sectional area.

Single cell preparation and immunophenotyping by flow cytometry
Single cell suspensions for flow cytometry were prepared from bone marrow as previously described [16]. Erythrocytes were lysed using Tris-buffered 0.83 % NH 4 Cl. Cells were then stained with fluorochrome-conjugated antibodies to detect different cell populations (Table 1). Fluorescence-minus-one stained samples were used as controls. Data was acquired on a BD FACS Canto II and analyzed using Flow Jo 8.8.6 (Three Star Inc, Ashland, USA). Macrophages were defined as CD11b + M-CSFR + cells; and pre-osteoclasts defined as CD11b + M-CSFR + RANK + cells. Median Fluorescence Intensity, MFI, was calculated of RANK expression on CD11b + M-CSFR + cells.

Estimation of ROS production
Dihydrorhodamine (DHR) 123 dye was used as a free radical sensor (final concentration of 0.25 μg/ml, Life Technologies, Carlsbad, CA, USA), which forms the fluorescent product rhodamine 123 after oxidation. To quantify the respiratory burst activity, bone marrow cells were incubated at 37°C for 15 min with phorbol myristate acetate (PMA, 0.5 μg/ml; Sigma-Aldrich). Cells were washed with cold PBS, acquired by a BD FACS Canto II (BD Biosciences, San Jose, CA, USA), and analyzed using Flow Jo software (FlowJo, LLC, Ashland, OR, USA).

Statistical analysis
Statistical evaluations were performed using IBM SPSS Statistics for Macintosh, Version 22.0.0.0, released 2013 (IBM Corp, Armonk, NY, USA). As bone mineral density measurements are sensitive to sibling controls, statistical analysis was only used to compare mice of the same genotypes. Student's ttest was used for comparison of two groups. Whenever Levene's test revealed unequal variances between the groups, Welch's t-test was used instead. Logarithmic transformations were used when appropriate to ensure normal distribution of data. P values <0.05 were considered statistically significant. GraphPad Prism version 6.0c was used for graphical representations. Reported values are mean ± SD unless otherwise stated.

RANK-expressing pre-OCL increase after ovx in Ncf1*/ *-mice
The two factors M-CSF and RANKL are important in the formation of OCL [17]. The cytokine M-CSF influences hematopoietic stem cells in the bone marrow to differentiate into the family of macrophages. OCLs forms from this M-CSF-dependent precursor, shared with macrophages. When OCL precursors are incubated with RANKL, these cells can further differentiate through the RANK receptor and fuse into multinucleated OCL. To evaluate the importance of Ncf1 in the formation of macrophages and pre-osteoclasts, we analyzed cells from the bone marrow using flow cytometry. The frequency of macrophages expressing the M-CSF receptor did not differ in either genotype (Fig. 3a). Neither did the frequency of cells expressing F4/80 (results not shown). However, the population of pre-osteoclastic macrophages expressing RANK expanded in Ncf1*/*-mice after ovx, but not in WT mice (Fig. 3b). There was no difference in MFI expression of RANK (Fig. 3d).

Ovx leads to lymphocyte activation in both genotypes
According to previous studies, lymphocytes are involved in the activation of OCL after ovx [7]. Ovx triggers APCs to activate T-cells in order to stimulate OCL activity. Neither the general dendritic cell (DC) population (CD11c + cells) nor the MHC class II + DCs were affected in either genotype 2 or 4 weeks after ovx ( Table 2). The CD80 + DCs expanded in the Ncf1*/*-mice 4 weeks after ovx ( Table 2). The general CD4 + T-cell population decreased after ovx (Table 2), whereas the CD69 + CD4 + cells increased in both genotypes 2 weeks after ovx ( Table 2). Similar results were found for the CD69 + CD8 + cell population (Table 2).

Discussion
Bone resorption and bone formation shifts with advancing age, leading to loss of bone mass and strength. In postmenopausal women, both age and loss of estrogens contribute to activating osteoclasts, inflammatory cells, and induce ROS production in the bone marrow. In our study, we found that ovx increases OCL formation ex vivo in WT mice, accompanied by reduced cortical thickness and trabecular BMD. In Ncf1*/* mice, no increase in OCL formation was seen after ovx, indicating that the Ncf1*/* should retain bone density after ovx. Surprisingly, the Ncf1*/* mice loose BMD similar to WT mice after ovx. We investigated several possible mechanisms for this found discrepancy. First, we investigated whether the difference was present already on pre-OCL in the bone marrow (defined as CD11b + M-CSFR + RANK + macrophages) destined to become OCL upon stimulation. The crucial role of Ncf1 for macrophage activation was previously shown where Ncf1 deleted specifically in macrophages regulated chronic inflammation [18][19][20]. In our study, Ncf1*/* mice showed an expanded population of pre-osteoclastic macrophages expressing RANK after ovx (Fig. 3b), indicating that they are target cells also in this model. One compensatory mechanism could be to increase OCL sensitivity and upregulate RANK. We could however not find any differences in the intensity of RANK expression using FACS (Fig. 3d). In addition, Ncf1*/* did not affect OCL area. This suggests that the OCL retain their efficiency regardless of numbers, and implies that it is the total bone area covered by the OCL that is important. Previous studies have implicated that NOX 1, 2 and 4 are important in bone loss. The OCL has been shown to switch NOX-complex in order to produce the necessary levels of ROS, possibly as a compensatory mechanism to ensure an adequate bone resorption process [21]. NOX 4 was shown to be more active in bone resorption in young mice [22], whereas the NOX 2 complex was shown to be important for bone loss in elderly mice (more than 2 years of age [23]. Those studies verified the importance of the two NOX complexes for bone resorption at different ages, and suggest that NOX 4 may compensate for lost ROS production in Ncf1*/* mice.
The bone remodeling process was comparably regulated by estrogen treatment in this study, as treating ovx mice with E2 resulted in similar increases in cortical thickness and trabecular BMD in both Ncf1*/* and WTmice. However, interactions between E2 and Ncf1 were possibly concealed in this study due to the high dose and duration of the estrogen treatment. In previous studies of postmenopausal bone loss, activated T cells and APCs were shown to regulate the bone resorption process [7]. Since Ncf1 regulates auto-reactive T cells through APCs in autoimmunity, we further investigated activated T cells and DCs. We found similar changes at 2 weeks in both WT and Ncf1*/* mice, whereas at 4 weeks after ovx, the CD80 + DCs population was expanded in the Ncf1*/*-mice. This indicates a possible rate difference, and that a target mechanism lies further up in the hierarchy, such as on the pre-OCL or macropages, or on soluble factors secreted from the activated APCs. TNFα is produced in abundance after ovx, and can induce osteoclastogenesis by itself. A possible discrepancy between OCL formation after stimulation with RANKL, TNFα and other cytokines in the Ncf1*/*-mice will be subject for further studies. Both the increased RANK-expression on macrophages and lingering increased CD80 + DCs in Ncf1*/*-mice could contribute to an altered bone degradation process, such as we discerned of the cortical bone in Ncf1*/*-mice. The lack of statistically significant difference in cortical thickness of Ncf1*/*-mice could either be due to a too small number of animals used in each group, or it is possible that other more sensitive BMD measurement techniques such as microCT could detect finer details and differences in the rate dependent bone degradation processes, thus revealing additional aspects regarding effects of Ncf1 on bone. The model we used in this study is a global Ncf1-deficiency model, reducing Ncf1 function in all cells. Therefore, possible mechanisms regulated by Ncf1 includes effects on osteoblasts, altered regulation of the cross-talk between osteoblasts and OCL, compensatory ROS production from another NOX or regulating another mechanism altogether, all of which are subjects for future studies.

Conclusion
Here, using a model of postmenopausal bone loss, we report that Ncf1, an essential subunit of the ROSproducing NOX 2 complex, contributes to OCL formation possibly by regulating pre-OCL. The NOX complexes are active in many physiological and pathophysiological processes. Increasing our understanding of their function will help in producing new targets for alleviating disease.  Table 2 Lymphocyte frequencies (mean %) after ovx in WT and Ncf1-deficient mice (n = 6-9 mice per group, t-test comparing ovx vs. sham for each genotype) Leukocyte population Sub population WT Sham (%, ±SD) WT ovx (%, ±SD) p-value Ncf1*/* Sham (%, ±SD) Ncf1*/* ovx (%, ±SD) p-value