JAK2-IGF1 axis in osteoclasts regulates postnatal growth in mice

Osteoclasts are specialized cells of the hematopoietic lineage that are responsible for bone resorption and play a critical role in musculoskeletal disease. JAK2 is a key mediator of cytokine and growth factor signaling; however, its role in osteoclasts in vivo has yet to be investigated. To elucidate the role of JAK2 in osteoclasts, we generated an osteoclast-specific JAK2–KO (Oc-JAK2–KO) mouse using the Cre/Lox-P system. Oc-JAK2–KO mice demonstrated marked postnatal growth restriction; however, this was not associated with significant changes in bone density, microarchitecture, or strength, indicating that the observed phenotype was not due to alterations in canonical osteoclast function. Interestingly, Oc-JAK2–KO mice had reduced osteoclast-specific expression of IGF1, suggesting a role for osteoclast-derived IGF1 in determination of body size. To directly assess the role of osteoclast-derived IGF1, we generated an osteoclast-specific IGF1–KO mouse, which showed a similar growth-restricted phenotype. Lastly, overexpression of circulating IGF1 by human transgene rescued the growth defects in Oc-JAK2–KO mice, in keeping with a causal role of IGF1 in these models. Together, our data show a potentially novel role for Oc-JAK2 and IGF1 in the determination of body size, which is independent of osteoclast resorptive function.


Introduction
Osteoclasts are specialized cells of the hematopoietic lineage and mostly known for their function in bone resorption (1). Osteoclasts play a critical role in the pathogenesis of several diseases, including osteoporosis (2), rheumatoid arthritis (3), periodontal disease (4), myeloma (5), and metastatic cancer (6); thus, there is an increasing need to elucidate the mechanisms of osteoclast function under homeostatic conditions in order to better understand the pathogenesis of osteoclast-related diseases and develop novel therapeutic strategies.
Janus kinase 2 (JAK2) is a ubiquitously expressed tyrosine kinase involved in multiple signal transduction pathways (7). JAK2 is responsible for transmitting extracellular signals from various cytokines and growth factors, including growth hormone, erythropoietin, thrombopoietin, prolactin, IL-6, granulocyte-macrophage CSF (GM-CSF), and IFN-γ. JAK2 is also a clinically relevant gene due to the causal link between activating JAK2 mutations and myeloproliferative disorders (8). In bone, there is some evidence to suggest a role for JAK2 in osteoclast biology. Osteoclasts derived from individuals with myelofibrosis harboring JAK2 activating mutations have impaired osteolytic capacity, implicating JAK2 with osteoclast dysfunction (9). In vitro studies using JAK2 inhibitors have also highlighted a potential role for JAK2 in osteoclasts. Osteoblast/osteoclast cocultures treated with the JAK1/2 inhibitor baricitinib showed decreased osteoclastogenesis due to reduced RANKL expression in osteoblasts (10). In another study, use of the JAK2 inhibitor AG490 in purified osteoclast cultures inhibited osteoclast apoptosis in response to RANKL withdrawal (11). Despite these data, the role of osteoclast JAK2 in vivo has yet to be investigated.
The objective of this study was to elucidate the in vivo function of JAK2 in osteoclasts. To accomplish this, we used the Cre/Lox-P system to generate an osteoclast-specific JAK2-KO (Oc-JAK2-KO) mouse. In this study, we found that Oc-JAK2-KO mice had marked postnatal growth restriction; however, this was not associated with significant changes in bone mineral density (BMD), bone microarchitecture, bone strength, osteoclast number, or osteoclastic gene expression, suggesting that osteoclast JAK2 modulates Osteoclasts are specialized cells of the hematopoietic lineage that are responsible for bone resorption and play a critical role in musculoskeletal disease. JAK2 is a key mediator of cytokine and growth factor signaling; however, its role in osteoclasts in vivo has yet to be investigated. To elucidate the role of JAK2 in osteoclasts, we generated an osteoclast-specific JAK2-KO (Oc-JAK2-KO) mouse using the Cre/Lox-P system. Oc-JAK2-KO mice demonstrated marked postnatal growth restriction; however, this was not associated with significant changes in bone density, microarchitecture, or strength, indicating that the observed phenotype was not due to alterations in canonical osteoclast function. Interestingly, Oc-JAK2-KO mice had reduced osteoclast-specific expression of IGF1, suggesting a role for osteoclast-derived IGF1 in determination of body size. To directly assess the role of osteoclast-derived IGF1, we generated an osteoclast-specific IGF1-KO mouse, which showed a similar growth-restricted phenotype. Lastly, overexpression of circulating IGF1 by human transgene rescued the growth defects in Oc-JAK2-KO mice, in keeping with a causal role of IGF1 in these models. Together, our data show a potentially novel role for Oc-JAK2 and IGF1 in the determination of body size, which is independent of osteoclast resorptive function.
Bone microarchitecture of femurs from 8-week-old and 24-week-old female mice was then assessed by μCT. At the femur neck, there were no changes in trabecular bone measures, including bone volume fraction and trabecular number, thickness, or separation ( Figure 4A). At the distal femur, Ctsk-Cre + Jak2 fl/fl mice had reduced trabecular thickness at 8 and 24 weeks, but there were no changes in trabecular number, separation, or overall bone volume fraction ( Figure 4B). At the femur midpoint, Ctsk-Cre + Jak2 fl/fl mice showed reduction in total area and cortical area; however, the cortical area/total area ratio and cortical thickness were unchanged ( Figure 4C). Overall, the μCT data show some minor changes in bone microarchitecture, which are likely attributable to the differences in body size of Ctsk-Cre + Jak2 fl/fl mice.
Next, we assessed biomechanical strength of femurs and LV using a materials testing system. Threepoint bending at the femur midpoint showed no difference in peak load both before and after adjustment for body size ( Figure 5, A and B). At the femur neck, both male and female Ctsk-Cre + Jak2 fl/fl mice had reduced compressive peak load at 24 weeks of age; however, this was no longer significant after adjustment for body size ( Figure 5, C and D). At the lumbar spine, there was a nonsignificant trend toward reduced compressive peak load in Ctsk-Cre + Jak2 fl/fl mice, which did not persist after adjustment for body size ( Figure 5, E and F). Forelimb grip strength was also measured using a rodent grip strength meter. Male Ctsk-Cre + Jak2 fl/fl mice had reduced grip strength at 8 and 24 weeks, while female Ctsk-Cre + Jak2 fl/fl mice demonstrated reduced grip strength at 24 weeks; however, these differences did not persist after adjustment for body size ( Figure 5, G and H). Overall, the biomechanical strength testing shows reductions in musculoskeletal strength in Ctsk-Cre + Jak2 fl/fl mice that are proportional to body size.
Next, we assessed for effects of JAK2 deficiency on osteoclast parameters. TRAP stain to identify osteoclasts in femur sections showed no alteration in osteoclast numbers or distribution in Ctsk-Cre + Jak2 fl/fl mice ( Figure 6A). Expression of key genes involved in osteoclastogenesis and osteoclast function were quantified in whole bone samples from the distal femur and proximal tibia ( Figure 6B). There were no differences in expression of calcitonin receptor (Calcr), tartrate-resistant acid phosphatase type 5 (Acp5), Ctsk, osteoclast-associated receptor (Oscar), integrin β 3 (Itgb3), matrix metalloproteinase 9 (Mmp9), C-terminal Src kinase (Csk), and nuclear factor of activated t cells 1 (Nfatc1). Growth plates were then evaluated on Safranin O-stained sections of tibia ( Figure 6C). At 2 weeks of age, there were no differences in the size of the growth plate ( Figure 6D); however, at 8 weeks of age, Ctsk-Cre + Jak2 fl/fl mice demonstrated a significant reduction in the size of the zone of hypertrophy (ZH) compared with control mice but not in the size of the total growth plate, zone of proliferation (ZP), or zone of ossification (ZO) ( Figure 6E).
Overall, these data show that deletion of JAK2 in osteoclasts does not have a positive or negative effect on bone density, bone microarchitecture, biomechanical strength, osteoclast number, or expression of osteoclastogenic genes. Histologically, there was a reduction in the size of the ZH in Ctsk-Cre + Jak2 fl/fl mice in association with their attenuation in body growth.
Reduced osteoclast-specific expression of IGF1 in Ctsk-Cre + Jak2 fl/fl mice. Growth defects in osteoclast-specific KO mice often arise due to impaired osteoclast function and an inability to resorb bone at the developing growth plate (14)(15). Despite clear growth defects in Ctsk-Cre + Jak2 fl/fl mice, there was surprisingly no evidence of defects in osteoclast function. Thus, we sought to investigate an alternative mechanism to explain the growth defects observed in Ctsk-Cre + Jak2 fl/fl mice.
Since Oc-JAK2-KO led to overall proportional changes in body size and reductions in the growth plate ZH, we hypothesized that a secreted growth factor might play an important role. JAK2 is a wellestablished regulator of IGF1 expression (16)(17)(18)(19); therefore, we assessed IGF1 levels in Ctsk-Cre + Jak2 fl/fl mice. Serum IGF1 concentrations, measured by ELISA, were equivalent in control and Ctsk-Cre + Jak2 fl/fl mice at 8 weeks of age ( Figure 7A). Quantitative PCR (qPCR), however, showed a decrease in Igf1 mRNA expression in the distal femur/proximal tibia in Ctsk-Cre + Jak2 fl/fl compared with controls, while no changes in Igf1 expression in other tissues -including the liver, testis, and ovary -were present  Figure 7B). To determine if the decreased expression of IGF1 in bone was specifically from osteoclasts, we first performed IHC for IGF1 in femur sections ( Figure 7C). In both control and KO mice, there was diffuse staining for IGF1 in BM cells and chondrocytes, as previously reported (20). In TRAP + multinucleated giant cells, immunostaining for IGF1 was of lower intensity in Ctsk-Cre + Jak2 fl/fl mice compared with controls, while the surrounding BM cells were unaffected. To further confirm a reduction of osteoclast-specific IGF1, we measured Igf1 gene expression in osteoclasts isolated by FACS. We used control and Ctsk-Cre + Jak2 fl/fl mice bred to the mTmG reporter mouse and isolated the GFP + bone cells. Compared with GFPcells, purified GFP + cells had a marked increase in expression of Ctsk ( Figure 7D)  and Acp5 (Figure 7E), consistent with these being mature osteoclasts. Jak2 mRNA was reduced by 62% in GFP + cells in Ctsk-Cre + Jak2 fl/fl mTmG + mice compared with osteoclast Jak2 heterozygous (Ctsk-Cre + Jak2 fl/+ mTmG + ) control mice ( Figure 7F). Similarly, Igf1 mRNA was reduced by approximately 70% in GFP + cells from Ctsk-Cre + Jak2 fl/fl mTmG + mice compared with osteoclast Jak2 heterozygous mice ( Figure 7G). No difference in either Jak2 or Igf1 expression was detected in GFPcells. Overall, these findings are consistent with reduced osteoclast-specific IGF1 expression in Ctsk-Cre + Jak2 fl/fl mice.
Ctsk-Cre + Igf1 fl/fl mice demonstrate postnatal growth restriction. Given our results showing growth defects with reduced osteoclast-derived IGF1 in Ctsk-Cre + Jak2 fl/fl mice, we hypothesized that a causal link exists between osteoclast-derived IGF1 and postnatal growth. To test this hypothesis, we generated osteoclast-specific IGF1-KO mice (Ctsk-Cre + Igf1 fl/fl mice). IHC performed on femur sections confirmed loss of IGF1 specifically in TRAP + multinucleated giant cells in Ctsk-Cre + Igf1 fl/fl mice ( Figure 8A). Similar to Ctsk-Cre + Jak2 fl/fl mice, both male and female Ctsk-Cre + Igf1 fl/fl mice had small but statistically significant reductions in body weight (Figure 8, B and C), body length (Figure 8, D and E), and femur length (Figure 8, F and G) compared with littermate controls at 8 weeks of age, but not at 2 weeks. The extent of reduction in growth parameters was  also similar between the 2 gene KO models. Additionally, histological analysis of the growth plate showed a decrease in the size of the ZH without changes in the ZO ( Figure 8H). Serum IGF1 concentrations were similar in control and Ctsk-Cre + Igf1 fl/fl mice ( Figure 8I). Overall, these findings parallel those of Ctsk-Cre + Jak2 fl/fl mice and support the hypothesis that osteoclast-derived IGF1 regulates postnatal growth. Transgenic overexpression of IGF1 rescues growth defects in Ctsk-Cre + Jak2 fl/fl mice. Since osteoclast expression of IGF1 was found to play a role in body size determination, we sought to determine if the decreased expression of IGF1 in Ctsk-Cre + Jak2 fl/fl mice had a causal role in their attenuated growth. To this end, we used a transgenic IGF1-overexpressing mouse, which expresses human IGF1 ([h]IGF1) under control of the transthyretin (Ttr) promotor to increase circulating IGF1. This model has previously been shown to overcome both endocrine and local deficits in IGF1 in order to restore normal growth (20). Analysis of serum showed equivalent reductions in circulating mouse IGF1 in both control and Ctsk-Cre + Jak2 fl/fl mice carrying the (h)IGF1 transgene ( Figure 9A). Moreover, levels of human IGF1 in circulation were also similar between these mice ( Figure 9B). Mouse IGF1 was likely reduced in the presence of the (h)IGF1 transgene due to feedback inhibition of endogenous IGF1. (h)IGF1 expression did not alter growth parameters in control mice. This is consistent with previous studies showing no effect on body length and femur length in (h)IGF1-expressing mice (21). In Ctsk-Cre + Jak2 fl/fl mice, however, transgenic expression of (h)IGF1 resulted in normalization of body weight ( Figure 9C), body length ( Figure 9D), and femur length ( Figure 9E) at 8 weeks of age. These findings suggest a causal role between reduced IGF1 expression in Ctsk-Cre + Jak2 fl/fl mice and postnatal growth restriction.

Discussion
In this study, we investigated the in vivo role of JAK2 in osteoclasts using Cre/Lox-P-mediated gene deletion. Although Oc-JAK2-KO mice demonstrated postnatal growth restriction, we did not identify any structural or functional alterations in bone, and any changes observed were proportional to the change in body size, suggesting that JAK2 is not essential for canonical bone resorptive function in osteoclasts. We assessed multiple measures of bone homeostasis in both sexes and at multiple time points. Since peak trabecular bone mass is achieved by 8 weeks and then declines (22), our data show that osteoclast JAK2 is dispensable for both normal bone development and in modulating age-related bone loss.
Previous studies have hinted at a potential role for JAK2 in osteoclast biology, since cytokines and growth factors that signal through JAK2 have been reported to modulate osteoclastogenesis (23). For example, growth hormone has been reported to promote osteoclastogenesis from unfractionated bone cells through direct effects on osteoclasts and indirectly through effects on stromal cells (24). IL-6 has been shown to inhibit osteoclastogenesis in BM macrophages and RAW 264.6 cell, which is dependent on STAT3 activation (25). IL-12, which also signals through JAK2-mediated STAT phosphorylation, inhibits osteoclastogenesis indirectly through induction of IFN-γ in stromal and immune cells (26). RANKL, which is essential for osteoclast formation, is also dependent on JAK-STAT signaling (27). RANKL-stimulated osteoclast formation is associated with reduced protein expression of JAK1, which prevents IFN-β inhibition of osteoclast lineage commitment. However, there are no alterations in JAK2 or JAK3 levels upon RANKL stimulation, suggesting that JAK2 is not essential, whereas JAK1 may be essential in the differentiation of osteoclasts. In terms of studies investigating JAK2 specifically, the JAK2 inhibitor AG490 was shown to promote osteoclast survival in vitro upon RANKL withdrawal (11). There is, however, doubt about the specificity of AG490 for JAK2, since it has been reported to also inhibit JAK3, STAT1/3/5, activating protein 1, and mitogen-activated protein kinase pathways; thus, it is unclear whether the effects observed in this study can be attributed specifically to inhibition of JAK2 (28). Overall, the studies linking JAK2-mediated signaling with osteoclast differentiation and survival are limited by the fact that osteoclast biology has only been investigated in vitro and that many of the effects seen may be due to indirect effects on nonosteoclast cells. Based on the results from our current study, we find that JAK2 does not have an appreciable effect on osteoclast formation or resorptive function in vivo. Further investigation, however, may be needed to determine if there is a role of osteoclast JAK2 in the setting of advanced age or sex hormone deficiency.
Despite the lack of bone resorptive effects seen in this study, Oc-JAK2-KO mice displayed significant postnatal growth restriction. Many in vivo gene KO studies have been performed in osteoclasts resulting in growth defects, which often arise in conjunction with impaired osteoclast function in vivo resulting in osteopetrosis (14)(15). This is accompanied by an increase in bone mass and other defects, including doming of the skull, delayed tooth eruption, and radio densities within regions of the BM. In our study, Oc-JAK2-KO mice did not display an increase in bone mass or any stigmata of osteopetrosis; therefore, it is unlikely that growth restriction occurred as a result of impairment in the traditional osteoclast function in bone resorption. We therefore propose a potentially novel role for osteoclasts in determining body growth independently of their canonical function in bone mineral homeostasis. Since JAK2 is a well-established regulator of IGF1 expression, most classically from the liver (16)(17)(18)(19), we hypothesized that JAK2 signaling in osteoclasts might similarly regulate IGF1 expression. The body growth restriction observed in Oc-JAK2-KO mice was associated with reduced expression of IGF1 in bone and osteoclasts. Overexpression of IGF1 abolished these growth defects, suggesting a causal role between JAK2-mediated IGF1 expression in osteoclasts and determination of body size. Additionally, when IGF1 was directly targeted in osteoclasts, this produced almost identical phenotypic findings compared with Oc-JAK2-KO mice. Together, these data support a role for osteoclast-derived IGF1 as a small but significant contributor to postnatal growth.
IGF1 is a ubiquitously expressed hormone and plays an essential role in postnatal growth (29). It is well established that approximately 75% of circulating IGF1 is derived from the liver (30)(31). Our understanding, however, of extrahepatic sources of IGF1 remains unclear, and there is an emerging recognition of the importance of autocrine and paracrine functions of IGF1. For example, targeted deletion of IGF1 in chondrocytes (32), osteoblasts (33), or osteocytes (34) produces unique effects on bone homeostasis and growth, without affecting circulating IGF concentrations. Our study provides evidence that IGF1 in osteoclasts similarly plays an important role in postnatal growth. Given that we saw changes in IGF1 in osteoclasts and bone, but not in circulating IGF1 concentrations, this would support a role for local autocrine or paracrine actions of osteoclast-derived IGF1. It has previously been shown that IGF1 is an important stimulator of chondrocyte hypertrophy (35), and in our study, a reduction in the hypertrophic zone was the only histological change seen in the growth plate. Thus, it is reasonable to suspect that osteoclast-derived IGF1 might act locally on chondrocytes in this zone, since they are located immediately adjacent to the ZO, where numerous active osteoclasts are located. We found the ZH to be reduced in size without associated changes in the overall size of the growth plate. This may be due to inadequate sample size or missing growth plate changes at other critical time