Hairy and enhancer of split 1 is a primary effector of NOTCH2 signaling and induces osteoclast differentiation and function

Notch2tm1.1Ecan mice, which harbor a mutation replicating that found in Hajdu–Cheney syndrome, exhibit marked osteopenia because of increased osteoclast number and bone resorption. Hairy and enhancer of split 1 (HES1) is a Notch target gene and a transcriptional modulator that determines osteoclast cell fate decisions. Transcript levels of Hes1 increase in Notch2tm1.1Ecan bone marrow–derived macrophages (BMMs) as they mature into osteoclasts, suggesting a role in osteoclastogenesis. To determine whether HES1 is responsible for the phenotype of Notch2tm1.1Ecan mice and the skeletal manifestations of Hajdu–Cheney syndrome, Hes1 was inactivated in Ctsk-expressing cells from Notch2tm1.1Ecan mice. Ctsk encodes the protease cathepsin K, which is expressed preferentially by osteoclasts. We found that the osteopenia of Notch2tm1.1Ecan mice was ameliorated, and the enhanced osteoclastogenesis was reversed in the context of the Hes1 inactivation. Microcomputed tomography revealed that the downregulation of Hes1 in Ctsk-expressing cells led to increased bone volume/total volume in female mice. In addition, cultures of BMMs from CtskCre/WT;Hes1Δ/Δ mice displayed a decrease in osteoclast number and size and decreased bone-resorbing capacity. Moreover, activation of HES1 in Ctsk-expressing cells led to osteopenia and enhanced osteoclast number, size, and bone resorptive capacity in BMM cultures. Osteoclast phenotypes and RNA-Seq of cells in which HES1 was activated revealed that HES1 modulates cell–cell fusion and bone-resorbing capacity by supporting sealing zone formation. In conclusion, we demonstrate that HES1 is mechanistically relevant to the skeletal manifestation of Notch2tm1.1Ecan mice and is a novel determinant of osteoclast differentiation and function.

Notch2 tm1.1Ecan mice, which harbor a mutation replicating that found in Hajdu-Cheney syndrome, exhibit marked osteopenia because of increased osteoclast number and bone resorption. Hairy and enhancer of split 1 (HES1) is a Notch target gene and a transcriptional modulator that determines osteoclast cell fate decisions. Transcript levels of Hes1 increase in Notch2 tm1.1Ecan bone marrow-derived macrophages (BMMs) as they mature into osteoclasts, suggesting a role in osteoclastogenesis. To determine whether HES1 is responsible for the phenotype of Notch2 tm1.1Ecan mice and the skeletal manifestations of Hajdu-Cheney syndrome, Hes1 was inactivated in Ctsk-expressing cells from Notch2 tm1.1Ecan mice. Ctsk encodes the protease cathepsin K, which is expressed preferentially by osteoclasts. We found that the osteopenia of Notch2 tm1.1Ecan mice was ameliorated, and the enhanced osteoclastogenesis was reversed in the context of the Hes1 inactivation. Microcomputed tomography revealed that the downregulation of Hes1 in Ctsk-expressing cells led to increased bone volume/ total volume in female mice. In addition, cultures of BMMs from Ctsk Cre/WT ;Hes1 Δ/Δ mice displayed a decrease in osteoclast number and size and decreased bone-resorbing capacity. Moreover, activation of HES1 in Ctsk-expressing cells led to osteopenia and enhanced osteoclast number, size, and bone resorptive capacity in BMM cultures. Osteoclast phenotypes and RNA-Seq of cells in which HES1 was activated revealed that HES1 modulates cell-cell fusion and bone-resorbing capacity by supporting sealing zone formation. In conclusion, we demonstrate that HES1 is mechanistically relevant to the skeletal manifestation of Notch2 tm1.1Ecan mice and is a novel determinant of osteoclast differentiation and function.
Osteoclasts are multinucleated giant cells that are responsible for bone resorption and essential to maintain bone homeostasis. Osteoclasts are derived from the differentiation and fusion of mononuclear cells of the myeloid lineage by the actions of macrophage colony-stimulating factor (M-CSF) and receptor activator of NF-κB ligand (RANKL) (1,2). RANKL triggers downstream signaling to induce the expression of transcription factors required for osteoclastogenesis, such as nuclear factor of activated T cells, cytoplasmic 1 (NFATc1) (3)(4)(5)(6). An imbalance of physiological or pathological conditions causing dysregulation of osteoclast differentiation and function leads to diseases associated with alterations in bone mass (7,8).
Hajdu-Cheney syndrome (HCS) (Online Mendelian Inheritance in Man: 102500) is a rare and devastating disorder characterized by numerous skeletal manifestations, including craniofacial developmental defects, short stature, bone loss with fractures, and acroosteolysis associated with inflammation of the distal phalanges (9)(10)(11)(12). HCS is associated with mutations or short deletions in exon 34 of NOTCH2 upstream of the PEST domain, which is required for the ubiquitination and degradation of NOTCH2 (12)(13)(14)(15)(16). The HCS pathogenic variants lead to the premature termination of a protein product lacking sequences necessary for the proteasomal degradation of the NOTCH2 intracellular domain so that the protein is stable and a gain-of-NOTCH2 function ensues. Autosomal dominant inheritance as well as de novo heterozygous mutations have been reported (12)(13)(14)(15)(16).
Our laboratory created a knock-in mouse model harboring a Notch2 6955C>T mutation reproducing HCS and termed Notch2 tm1.1Ecan (also known as Notch2 Q2319X ) (17,18). The homozygous mutation is associated with craniofacial developmental abnormalities and is lethal, and heterozygous Notch2 tm1.1Ecan mutant mice exhibit profound osteopenia and short limbs, reproducing functional outcomes of the human disease and establishing the first model for the study of HCS (12,17). Notch2 tm1.1Ecan mice have increased bone resorption secondary to a direct effect of the gain-of-NOTCH2 function on osteoclastogenesis as well as the increased expression of RANKL by cells of the osteoblast lineage (17). These are unique functional properties of NOTCH2, which are distinct from those reported for other Notch receptors (19,20). Indeed, NOTCH1 inhibits osteoclastogenesis directly, and NOTCH3 is not expressed in the myeloid lineage; although, by inducing RANKL in cells of the osteoblast lineage, it enhances osteoclastogenesis indirectly (21)(22)(23). Low levels of NOTCH4 are expressed in the myeloid lineage, and it is not known to play a role in osteoblastogenesis (24).
Cultures of bone marrow macrophages (BMMs) revealed that the expression of Hes1, a Notch target gene, is enhanced as cells mature as osteoclasts, and the increased expression is of greater magnitude in cultures from Notch2 tm1.1Ecan mice (17,25). Importantly, other Notch target genes, such as Hes3, Hes5, Hey1, Hey2, and HeyL, are either expressed at very low levels or not detected in BMMs from control or mutant mice. This observation suggests that hairy and enhancer of split 1 (HES1) may be an important regulator of osteoclastogenesis and is in part responsible for the HCS phenotype.
HES1 is a transcriptional modulator that plays a role in the differentiation of embryonic stem and mesenchymal cells (26,27). Although HES1 is considered a transcriptional repressor, transcription factors can function as either positive or negative regulators of transcription in a cell contextdependent manner (28,29). In addition, calcium/calmodulindependent protein kinase 2 can convert HES1 from a repressor to an activator of transcription (30,31). Misexpression of Hes1 in the osteoblast lineage has demonstrated a role as an inhibitor of osteoblast differentiation and function (32). The role of HES1 in osteoclastogenesis is unknown.
The intent of the present study was to determine whether HES1 was mechanistically relevant to the HCS phenotype and to define the function of HES1 in osteoclast differentiation in vitro and in vivo. For this purpose, Hes1 was induced or inactivated in Ctsk-expressing cells of the osteoclast lineage. To determine whether HES1 had a mechanistic role in the skeletal phenotype of HCS, Notch2 tm1.1Ecan mice were studied in the context of the Hes1 inactivation in Ctsk-expressing cells. Skeletal phenotypes were determined by microcomputed tomography (μCT) and histomorphometry and cellular effects by the study of osteoclast differentiation and resorption activity in vitro.
To determine whether the inactivation of Hes1 could reverse the osteopenia of the Notch2 tm1.1Ecan mutation, Ctsk Cre/WT ;He-s1 loxP/loxP mice were crossed with Notch2 tm1.1Ecan ;Hes1 loxP/loxP mice to inactivate Hes1 in the context of the Notch2 mutation. The transcript levels of Hes1 were decreased in bone extracts from 2-month-old male Ctsk Cre/WT ;Hes1 Δ/Δ and Ctsk Cre/WT ; Notch2 tm1.1Ecan ;Hes1 Δ/Δ mice compared with control, whereas Notch2 WT and mutant (Notch2 6955C>T ) mRNA levels were not affected (Fig. S1). Confirming prior observations, Notch2 tm1.1Ecan mice displayed cancellous bone osteopenia associated with decreased connectivity and trabecular number (Fig. 2). The Hes1 inactivation by itself did not alter bone microarchitectural parameters in 2-month-old male mice compared with control sex-matched WT mice. The decreased cancellous bone volume/total volume (BV/TV) observed in Notch2 tm1.1Ecan was significantly increased in the context of the Hes1 inactivation associated with increased connectivity and trabecular number so that the osteopenia of Notch2 tm1.1Ecan mice was ameliorated in Ctsk Cre/WT ;Notch2 tm1.1Ecan ;Hes1 Δ/Δ mice (Fig. 2). Cancellous bone histomorphometry confirmed previous work and demonstrated an increase in osteoclast number and bone resorption, without an effect on osteoblast number and bone formation, in Notch2 tm1.1Ecan mice (17). The increased osteoclast number and eroded surface found in Notch2 tm1.1Ecan mice were decreased 50% in the context of the Hes1 inactivation, so that both parameters were no longer increased in Ctsk Cre/WT ;Notch2 tm1.1Ecan ;Hes1 Δ/Δ male mice compared with Notch2 tm1.1Ecan ;Hes loxP/loxP control mice (Table 1). These results indicate that HES1 is mechanistically relevant to the osteopenia of Notch2 tm1.1Ecan mice although they suggest a minor role of HES1 in the bone architecture of male mice. The Hes1 deletion had only a modest effect on the cortical osteopenic phenotype (not shown) and did not affect the decrease in femoral length observed in Notch2 tm1.1Ecan mice (Fig. S1).

HES1 is a determinant of osteoclastogenesis in vitro
To ascertain the function of HES1 in cells of the osteoclast lineage, BMMs from either Hes1 loxP/loxP or Rosa [STOP]Hes1 mice were isolated. BMMs were cultured in the presence of M-CSF and RANKL for 2 days and transduced with Ad-Cre to delete loxP flanked sequences or Ad-GFP as control. Excision of the STOP cassette in Rosa Hes1 cells resulted in a 20-fold induction of Hes1 mRNA and a 1.7-fold increase in osteoclastogenesis compared with Rosa [STOP]Hes1 cultures transduced with Ad-GFP (Fig. 3). Conversely, deletion of Hes1 resulted in a 50% reduction in Hes1 mRNA levels and a 50% decrease in osteoclast number compared with control cultures transduced with Ad-GFP (Fig. 3). The results demonstrate that HES1 is a determinant of osteoclast differentiation in vitro.

Inactivation of Hes1 in osteoclasts of female mice increases BV in vivo
To confirm a role of HES1 in osteoclastogenesis and bone homeostasis, Hes1 was inactivated in vivo in Ctsk-expressing cells. For this purpose, Ctsk Cre/WT ;Hes1 loxP/loxP mice were crossed with Hes1 loxP/loxP mice to generate Ctsk Cre/WT ;Hes1 Δ/Δ and littermate Hes1 loxP/loxP controls. Ctsk Cre/WT ;Hes1 Δ/Δ appeared healthy, and their weight and femoral length were not different from littermate Hes1 loxP/loxP mice (Fig. S2). Ctsk Cre -mediated recombination was documented in genomic DNA from tibiae of Ctsk Cre/WT ;Hes1 Δ/Δ mice with a consequent decrease in Hes1 mRNA. Confirming the results observed in the context of the Notch2 tm1.1Ecan mutant mice, inactivation of Hes1 in 2-or 4-month-old male mice did not result in an obvious skeletal phenotype, although trabecular number and connectivity were modestly increased ( Table 2). In contrast, 2-month and particularly 4-month-old female mice harboring the inactivation of Hes1 exhibited a significant increase in femoral BV/TV (Table 2 and Fig. 4). Femoral μCT of 4-month-old female Ctsk Cre/WT ;Hes1 Δ/Δ mice revealed an 85% increase in BV/TV associated with an increase in trabecular number and connectivity density and a decrease in structure model index (SMI) compared with controls. Bone histomorphometry of 4-month-old Ctsk Cre/WT ;Hes1 Δ/Δ female mice demonstrated an 50% decrease in osteoclast number and 35% decrease in eroded surface, compared with littermate controls, confirming that HES1 is required for osteoclast differentiation and function in vivo (Table 3 and Fig. 5). Osteoblast number and bone formation were not affected by the Hes1 deletion.

Inactivation of Hes1 decreases osteoclast differentiation in vitro
To confirm that the phenotype of Ctsk Cre/WT ;Hes1 Δ/Δ mice was due to a decrease in osteoclast differentiation, BMMs derived from Ctsk Cre/WT ;Hes1 Δ/Δ and control littermates were cultured in the presence of M-CSF and RANKL. Ctsk Cre/WT ; Hes1 Δ/Δ cultures revealed a 42% decrease in osteoclast number when compared with cells from littermate controls (Fig. 6). The number of osteoclasts with high number of nuclei was decreased in Ctsk Cre/WT ;Hes1 Δ/Δ cultures compared with controls, indicating that the size of osteoclasts was reduced because of a decrease in the fusion capacity of Ctsk Cre/WT ; Hes1 Δ/Δ cells. Mature osteoclasts have a distinct cytoskeletal structure, namely the sealing zone, a circular actin-rich BMMs derived from 2-month-old Notch2 tm1.1Ecan ;Hes1 loxP/loxP and Hes1 loxP/loxP littermate controls were cultured for 2 days with M-CSF at 30 ng/ml and RANKL at 10 ng/ml and transduced with adenoviruses carrying CMV-Cre (Ad-Cre) or adenoviruses carrying GFP (Ad-GFP) as control at MOI 100 and cultured for two additional days in the presence of M-CSF at 30 ng/ml and RANKL at 10 ng/ml until the formation of multinucleated TRAP-positive cells. A, total RNA was extracted, and gene expression was determined by quantitative RT-PCR. Data are expressed as Notch2 6955C>T , Notch2, and Hes1, corrected for Rpl38 copy number. B, representative images of TRAP-stained multinucleated cells are shown. The scale bars in the right corner represent 500 μm. C, TRAP-positive cells with more than three nuclei were considered osteoclasts and counted. Values are means ± SD; n = 4 technical replicates for WT (open circles) and Notch2 tm1.1Ecan (closed circles) cells in the context of Hes1 loxP/loxP (white bar) or Hes1 Δ/Δ (gray bar) deleted alleles. Representative data are shown from two independent experiments. *Significantly different between Notch2 tm1.1Ecan and control, p < 0.05. #Significantly different between Hes1 Δ/Δ and Hes1 loxP/loxP , p < 0.05. BMM, bone marrow-derived macrophage; M-CSF, macrophage colony-stimulating factor; MOI, multiplicity of infection; RANKL, receptor activator of NF-κB ligand; TRAP, tartrate resistant acid phosphatase.  Values are means ± SD; n = 12 for control Hes1 loxP/loxP and n = 6 for Notch2 tm1.1Ecan ;Hes1 loxP/loxP ; n = 10 for Ctsk Cre/WT ;Hes1 Δ/Δ and n = 11 for Ctsk Cre/WT ;Notch2 tm1.1Ecan ;Hes1 Δ/Δ . *Significantly different between Notch2 tm1.1Ecan and control, p < 0.05. #Significantly different between Hes1 Δ/Δ and Hes1 loxP/loxP , p < 0.05. μCT, microcomputed tomography.

Hes1 and osteoclast
structure formed by podosomes in a cluster to create a ring that is tightly adherent to the bone matrix for efficient bone resorption (33). Phalloidin staining of osteoclasts from Ctsk Cre/WT ; Hes1 Δ/Δ mice cultured on bone slices revealed smaller sealing zones than controls and a 30% decrease in the perimeter of the sealing zone ( Fig. 6). Ctsk Cre/WT ;Hes1 Δ/Δ osteoclasts also exhibited a 60% decrease in total bone resorption area, indicating a decrease in osteoclast resorptive activity (Fig. 6).

Induction of HES1 in osteoclasts causes osteopenia
To determine the effect of the HES1 induction on osteoclastogenesis in vivo, homozygous Rosa [STOP]Hes1 mice were crossed with Ctsk Cre/WT mice for the creation of Ctsk Cre/WT ;Rosa Hes1 experimental mice and Rosa [STOP]Hes1 littermate controls. Ctsk Cre/WT ;Rosa Hes1 mice appeared healthy, and their weight was not different from that of littermate controls (Fig. S3). Ctsk Cremediated recombination was demonstrated in genomic DNA from tibiae of Ctsk Cre/WT ;Rosa Hes1 mice, and Hes1 mRNA levels were increased in bone extracts from Ctsk Cre/WT ;Rosa Hes1 mice.
Femoral architecture of 10-week-old male and female Ctsk Cre/WT ;Rosa Hes1 mice revealed a 30% decrease in BV/TV associated with a decrease in connectivity and an increase in SMI in Ctsk Cre/WT ;Rosa Hes1 mice that reached statistical significance in female but not in male mice (Table 4). Bone histomorphometry of 10-week-old female Ctsk Cre/WT ;Rosa Hes1 mice demonstrated a 1.7-fold increase in osteoclast surface and number, and approximately twofold increase in eroded surface, when compared with littermate controls, confirming that HES1 increases osteoclast differentiation and function in vivo (Table 5 and Fig. 7).

Induction of HES1 enhances osteoclast differentiation in vitro
To verify that the phenotype of Ctsk Cre/WT ;Rosa Hes1 mice was due to a direct effect in cells of the osteoclast lineage, BMMs from Ctsk Cre/WT ;Rosa Hes1 and control littermates were cultured in the presence of M-CSF and RANKL. BMMs from Ctsk Cre/WT ;Rosa Hes1 mice exhibited a 4.5-fold increase in osteoclast number in comparison to cells from littermate controls (Fig. 8). In addition, osteoclasts with a high number of nuclei were significantly increased in Ctsk Cre/WT ;Rosa Hes1 cultures compared with controls, indicating that the size of osteoclasts was larger because of highly activated fusion in Ctsk Cre/WT ;Rosa Hes1 cells. Phalloidin staining of osteoclasts from Ctsk Cre/WT ;Rosa Hes1 mice cultured on bone slices confirmed larger cells with sealing zones that were 25% larger than in cells from control littermates (Fig. 8). Accordingly, Ctsk Cre/WT ;Rosa Hes1 osteoclasts exhibited a sixfold increase in total resorption pit area (Fig. 8), indicating enhanced bone resorptive capacity in Ctsk Cre/WT ;Rosa Hes1 osteoclasts.

Mechanisms of HES1 action on osteoclastogenesis
To understand the molecular mechanisms associated with the effect of HES1 on osteoclast differentiation, total RNA

Discussion
The present work uncovers a new function of HES1 on osteoclast differentiation and bone remodeling. The deletion

Hes1 and osteoclast
of Hes1 in Ctsk-expressing cells decreased the osteoclastogenic potential of preosteoclasts, whereas its induction enhanced osteoclastogenesis. Osteoclast phenotypes and RNA-Seq analysis revealed that HES1 regulates cell-cell fusion and the formation of the sealing zone. The gene subsets of fusion markers, integrin signaling, and structural proteins for sealing zone formation were significantly upregulated in osteoclasts overexpressing HES1. These results indicate that HES1 has a direct role in osteoclast differentiation and function. Our study also reveals that the expression of Nfatc1 and that of inhibitors of osteoclastogenesis acting as transcriptional brakes of Nfatc1, such as Irf8, Bcl6, Mafb, and Id1, were regulated by HES1. It is possible that HES1 interacts with transcriptional repressors of osteoclastogenesis in a manner analogous to BLIMP1, although the expression of Blimp1 was not affected by HES1 (47,52). It is probable that HES1 acts as a transcriptional repressor of inhibitors of osteoclastogenesis and as a consequence causes enhanced Nfatc1 expression. Under selected   Hes1 and osteoclast cellular conditions, HES1 can act as a transcriptional activator so that one cannot exclude a direct effect of HES1 on the transcriptional activation of Nfatc1 (30). In accordance with our observations, γ-secretase inhibitors, known to prevent Notch activation, were found to inhibit osteoclast cell fusion and the formation of the podosomal actin belt structure by suppressing HES1/mitogen-activated protein kinase/AKTmediated induction of NFATc1 in vitro (53). However, it is

Hes1 and osteoclast
important to note that γ-secretase inhibitors can target many substrates, and their effect is not specific to Notch signaling (54,55). Although HES1 had a pronounced effect on osteoclast differentiation and function in vitro, this effect was restricted to female mice in vivo. The Hes1 inactivation caused an 85% increase in BV in mature female mice, and the induction of HES1 in Ctsk-expressing cells caused an osteopenic phenotype. The inactivation of Hes1 in male mice did not result in a prominent skeletal phenotype; however, it opposed the osteopenic and resorptive phenotype of Notch2 tm1.1Ecan mice harboring an HCS mutation causing a gain-of-NOTCH2 function. The absence of a phenotype in male mice facilitated the interpretation of the rescue of the Notch2 tm1.1Ecan phenotype by the Hes1 deletion. Since female Hes1-inactivated mice had an increase and Notch2 tm1.1Ecan a decrease in BV, one would expect Notch2 tm1.1Ecan ;Hes1 Δ/Δ female mice to have an intermediate BV. So that an increase in the BV of Notch2 t-m1.1Ecan would not necessarily represent a rescue of the osteopenic phenotype and that HES1 was a mediator of NOTCH2. It is not readily apparent why the Hes1 inactivation caused a phenotype in female but not in male mice, and the observation stresses the importance of examining phenotypes in mice of different sexes independently (56,57). It is not unusual to observe sex-specific phenotypes in genetically engineered mice (58)(59)(60). Possible explanations for the prevalence of a phenotype in female mice include genetic influences, a loss of the inhibitory actions of estrogens on osteoclastogenesis in the context of the Hes1 inactivation as well as the earlier NF-κB-NFATc1 activation and osteoclastogenesis that occurs in female mice (57,61,62).
In previous work, we demonstrated that HES1 is induced as osteoclasts mature, particularly in the context of the Notch2 t-m1.1Ecan mutation (17). A plausible explanation for the modest skeletal phenotype of the Hes1 inactivation in male mice is that under basal conditions HES1 levels are low and play a modest role in skeletal physiology, and only following Notch activation, HES1 plays a significant role in bone homeostasis. This explanation is substantiated by the amelioration of the Notch2 tm1.1Ecan osteopenic phenotype following the Hes1 inactivation. The Notch2 tm1.1Ecan phenotype was not fully reversed, and this is explained by the effects of NOTCH2 enhancing RANKL expression by cells of the osteoblast lineage since these are independent of the induction of HES1 in the myeloid lineage (17,24). Other Notch target genes, such as Hey1, Hey2, and HeyL, are not expressed in cells of the myeloid lineage and as a consequence could not be responsible for the stimulatory effects of NOTCH2 on osteoclastogenesis (19). HES3 and HES5 could compensate for the effects of HES1, but their expression in osteoclasts is low and their role in osteoclastogenesis is unknown (63). Whereas, HES1 mediates direct effects of NOTCH2 on osteoclastogenesis, it is not likely to mediate the effects of NOTCH1, known to inhibit and not enhance osteoclast maturation, or NOTCH3, since this Notch receptor is not expressed in the myeloid lineage and its effects on osteoclastogenesis are indirect (21,23). NOTCH4 is expressed at low levels in the myeloid lineage and not known to play a role in osteoclast differentiation (19,24).
In the present work, we confirm that Notch2 tm1.1Ecan mice are osteopenic because of direct effects of NOTCH2 in cells of the myeloid lineage. The stimulatory effect of NOTCH2 on osteoclastogenesis has been attributed to direct interactions of the NOTCH2 intracellular domain with NF-κB in the context of Nfatc1 regulatory regions and increased Nfatc1 transcription (64). However, recent work from our laboratory has demonstrated that NOTCH2 has NF-κB-independent effects on tumor necrosis factor α (TNFα)-induced osteoclastogenesis, and some of these effects are secondary to the activation of AKT and Il1b expression (25,65). The present work demonstrates that the direct effects of NOTCH2 on osteoclastogenesis are HES1 dependent confirming previous work from this laboratory revealing that the enhancement of the osteolytic actions of TNFα by the Notch2 tm1.1Ecan mutation depend on the induction of HES1 (25).
HES1 is known to inhibit phosphatase and tensin homolog and as a consequence enhance phosphoinositide 3-kinase-  AKT signaling (66). AKT signaling is required for cell-cell fusion during osteoclast differentiation, and inhibitors of AKT lead to a decrease in Dcstamp transcripts and osteoclast size (67). However, the levels of phosphatase and tensin homolog transcripts and the phosphorylation of AKT were not different between Ctsk Cre/WT ;Rosa Hes1 osteoclasts and controls (data not shown). Although RANKL and TNFα share and activate similar downstream molecules, mechanisms triggering osteoclastogenesis are different in part because Nfatc1 and Dcstamp levels are not changed in conditions of proinflammatory cytokine-induced osteoclastogenesis (68,69). Hes1 inactivation decreases Il1b in TNFα-induced Notch2 tm1.1Ecan osteoclasts, and the present work confirms that Il1b and Il1r1 transcripts are increased in HES1-overexpressing osteoclasts (25). IL1β induces pathologically activated osteoclasts bearing a high level of bone-resorbing activity and may be mechanistically relevant to the actions of HES1 in osteoclasts (49). The phenotype of Notch2 tm1.1Ecan mice as well as the osteopenia of humans harboring HCS pathogenic variants is secondary to an increase in bone resorption with no evidence of impaired bone formation (11,17,19). The direct effects of NOTCH2 in the myeloid lineage appear mediated by Notch target gene Hes1. This is further substantiated by the fact that other target genes, such as Hey1, Hey2, and HeyL, are not expressed by cells of the osteoclast lineage; therefore, these cannot mediate the effects of NOTCH2 in this cell population.
A limitation of the present work is the use of a Ctsk Cre mouse model to deliver Cre recombinase since the expression of Ctsk is not exclusive to osteoclasts and Ctsk is also detected in alternate skeletal and nonskeletal cells (61,(70)(71)(72). Although one cannot fully exclude effects outside the osteoclast lineage, it is reasonable to believe that the effects observed in the present work are secondary to the misexpression of Hes1 in osteoclasts since cultures of BMMs from Ctsk Cre/WT ;Rosa Hes1 and Ctsk CreWT ;Hes1 Δ/Δ mice revealed profound effects on osteoclast differentiation. Moreover, the activation and inactivation of Hes1 in BMM cultures using adenoviruses to deliver Cre demonstrated a direct effect of HES1 in the osteoclast lineage.
In conclusion, HES1 plays a critical role in osteoclastogenesis and bone resorption and is mechanistically relevant to the skeletal phenotype of an experimental model of HCS.

Experimental procedures
Genetically modified mice Notch2 tm1.1Ecan mice harboring a 6955C>T substitution in the Notch2 locus have been characterized in previous studies and were backcrossed into a C57BL/6 background for eight The scale bar in the right corner represents 500 μm. B, Hes1 transcript levels were measured by quantitative RT-PCR in total RNA from osteoclasts. Transcript levels are reported as copy number corrected for Rpl38 (left). TRAP-positive cells with more than three nuclei were considered osteoclasts and counted (middle). TRAP-positive cells with differential counting of nuclei/osteoclast are shown (right). C, representative images of Alexa Fluor 594 phalloidin-stained multinucleated cells on bone discs are shown. The scale bar in the right corner represents 100 μm. D, the perimeter of sealing zones was measured in n = 68 osteoclasts from control and in n = 131 osteoclasts from Ctsk Cre/WT ;Rosa Hes1 cultures. E, representative images of toluidine blue-stained resorption pits. The scale bar in the right corner represents 200 μm. F, the total resorption pit area was measured (%). Values are means ± SD; n = 3 biological replicates for control and Ctsk Cre/WT ;Rosa Hes1 . *Significantly different between Ctsk Cre/WT ;Rosa Hes1 and control, p < 0.05. BMM, bone marrow-derived macrophage; M-CSF, macrophage colony-stimulating factor; RANKL, receptor activator of NF-κB ligand; TRAP, tartrate resistant acid phosphatase. and more generations (17,18,73). Hes1 loxP/loxP (Hes1<tm1Imayo) mice, where loxP sequences are knocked into the first intron and downstream of the 3 0 UTR of Hes1 alleles, were obtained from RIKEN (RBRC06047; Wako Saitama) in a C57BL/6 background (74). Rosa [STOP]Hes1 (Gt(ROSA) 26Sor<tm1(Hes1.EGFP)Imayo>) were obtained from RIKEN (RBRC06002) in an ICR background (26). In Rosa [STOP]Hes1 mice, Hes1 coding sequences are cloned into the Rosa26 locus downstream of a Neo-STOP cassette flanked by loxP sequences, so that HES1-IRES-GFP is expressed following the excision of the cassette by Cre recombination. To induce or delete Hes1 in differentiated cells of the osteoclast lineage, mice harboring sequences coding for the Cre recombinase knocked-in into the Ctsk locus (Ctsk Cre ) were used in a C57BL/ 6 background (61,70). Genotyping was conducted in tail DNA extracts by PCR using specific primers from Integrated DNA Technologies (IDT) ( Table S1).

BMM cultures, osteoclast formation, and adenovirus-Cremediated recombination
To obtain BMMs, the marrow from experimental and control sex-matched littermate mice was removed by flushing with a 26-gauge needle, and erythrocytes were lyzed in 150 mM NH 4 Cl, 10 mM KHCO 3 , and 0.1 mM EDTA (pH 7.4), as described previously (73). Cells were centrifuged, and the sediment suspended in α-minimum essential medium (α-MEM) in the presence of 10% fetal bovine serum (FBS; both from Thermo Fisher Scientific) and recombinant human M-CSF at 30 ng/ml. M-CSF complementary DNA (cDNA) and expression vector were obtained from D. Fremont, and M-CSF was purified as previously reported (34). Cells were seeded on uncoated plastic petri dishes at a density of 300,000 cells/cm 2 and cultured for 3 days. For osteoclast formation, cells were collected following treatment with 0.25% trypsin/EDTA for 5 min and seeded on tissue culture plates at a density of 62,500 cells/cm 2 in α-MEM with 10% FBS, M-CSF at 30 ng/ml, Figure 9. Expression of osteoclastogenic genes is increased in Ctsk Cre/WT ;Rosa Hes1 osteoclasts. BMMs derived from 10-week-old Ctsk Cre/WT ;Rosa Hes1 mice and control littermates were cultured for 4 days in the presence of M-CSF at 30 ng/ml and of RANKL at 10 ng/ml. Cells were collected for total RNA and protein extraction. A, RNA was analyzed by RNA-Seq. The bars indicate Log 2 fold changes (p < 0.05) of gene expression between control and Ctsk Cre/WT ;Rosa Hes1 osteoclasts; n = 3 control and Ctsk Cre/WT ;Rosa Hes1 biological replicates. B, Bcl6, Mafb, Nfatc1, Atp6vOd2, Ocstamp, and Acp mRNA levels were measured by quantitative RT-PCR and reported as copy number corrected for Rpl38 mRNA levels. Values are means ± SD; n = 3 control and Ctsk Cre/WT ;Rosa Hes1 biological replicates. C, representative data of protein levels of NFATc1 and HES1. About 40 μg of total protein were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and NFATc1 and HES1 levels were detected using anti-NFATc1 and anti-HES1 antibodies, respectively. β-Actin served as a loading control in the same blot. The band intensity was quantified by Image Lab software (version 5.2.1), and the numerical ratio of NFATc1/β-actin and HES1/β-actin is shown under each blot. Control ratios at the initiation of the culture in the presence of RANKL (day 0) are normalized to 1. *Significantly different between Ctsk Cre/WT ;Rosa Hes1 and control, p < 0.05. BMM, bone marrow-derived macrophage; HES1, hairy and enhancer of split 1; M-CSF, macrophage colony-stimulating factor; NFATc1, nuclear factor of activated T cells, cytoplasmic 1; RANKL, receptor activator of NF-κB ligand. and recombinant murine RANKL at 10 ng/ml. Tnfsf11, encoding RANKL, cDNA expression vector was obtained from M. Glogauer, and glutathione-S-transferase-tagged RANKL was expressed and purified as described (75). Cultures were carried out until multinucleated tartrate resistant acid phosphatase (TRAP)-positive cells were formed. TRAP enzyme histochemistry was conducted using a commercial kit (Sigma-Aldrich), in accordance with the manufacturer's instructions. TRAP-positive cells containing more than three nuclei were considered osteoclasts.
For actin structure staining and bone resorption assay of osteoclasts in vitro, BMMs were seeded at a density of 62,500 cells/cm 2 on bovine cortical bone slices and cultured in α-MEM with 10% FBS, M-CSF at 30 ng/ml, and RANKL at 10 ng/ml. To visualize the sealing zone of osteoclasts on bone slices, cells were fixed with 4% paraformaldehyde for 10 min and permeabilized with 0.3% Triton X-100 for 5 min. To block nonspecific background staining, cells on bone discs were incubated with 2% bovine serum albumin for 1 h and stained with Alexa Fluor 594 Phalloidin (Thermo Fisher Scientific) at a 1:40 dilution for 20 min. The sealing zone was viewed on a Leica fluorescence microscope (model DMI6000B), and collected images were processed using the Leica Application Suite X 1.5.1.1387 (Leica Microsystems). After visualizing the sealing zone, cells were stained for TRAP to assess cellular morphology. To visualize bone resorption pits, bone slices were sonicated to remove osteoclasts and stained with 1% toluidine blue in 1% sodium borate. To assess the ability of osteoclasts to resorb bone, the total resorption area/total bone area was measured on images acquired with an Olympus DP72 camera using cellSens Dimension software, version 1.6 (Olympus Corporation). The total resorption area/total bone area was corrected for the total number of TRAP-positive multinucleated cells (73).
To inactivate or induce Hes1 in osteoclast precursors in vitro, BMMs from homozygous Hes1 loxP/loxP or Rosa [STOP] Hes1 mice were cultured in the presence of M-CSF at 30 ng/ml and RANKL at 10 ng/ml for 2 days, prior to being transduced with Ad-Cre or CMV-GFP (Ad-GFP [Vector Biolabs]) as control, at multiplicity of infection of 100 and cultured with M-CSF and RANKL for two additional days until the formation of multinucleated TRAP-positive cells. To inactivate Hes1 in the context of the Notch2 tm1.1Ecan mutation, Hes1 loxP/loxP alleles were introduced into Notch2 tm1.1Ecan mice to create Notch2 tm1.1Ecan ;Hes1 loxP/loxP mice, and BMMs were cultured and transduced with Ad-Cre or Ad-GFP.

qRT-PCR
Total RNA was extracted from osteoclasts with the RNeasy Mini kit (Qiagen) and homogenized bones with the RNeasy Micro kit (Qiagen), in accordance with the manufacturer's instructions. The integrity of the RNA extracted from bones was assessed by microfluidic electrophoresis on an Experion system (Bio-Rad), and RNA with a quality indicator number equal to or higher than 7.0 was used for subsequent analysis. Equal amounts of RNA were reverse transcribed using the iScript RT-PCR kit (Bio-Rad) and amplified in the presence of specific primers (all from IDT; Table S2) with the SsoAdvanced Universal SYBR Green Supermix (Bio-Rad) at 60 C for 40 cycles. Transcript copy number was estimated by comparison with a serial dilution of cDNA for Acp5 and Notch2 (all from Thermo Fisher Scientific), Hes1 (American Type Culture Collection), Nfatc1 (Addgene; plasmid 11793; created by A. Rao), Bcl6, Mafb, Atp6vOd2, and Ocstamp (all from Dharmacon).
Amplification reactions were conducted in CFX96 qRT-PCR detection systems (Bio-Rad), and fluorescence was monitored at the end of the elongation step during every PCR cycle. Data are expressed as copy number corrected for Rpl38 expression estimated by comparison with a serial dilution of cDNA for Rpl38 (American Type Culture Collection) (78).

Illumina transcriptome library preparation and sequencing
Total RNA was quantified, and purity ratios were determined for each sample using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific). To assess RNA quality, total RNA was analyzed on the Agilent TapeStation 4200 (Agilent Technologies) using the RNA High Sensitivity assay. Ribosomal integrity numbers were recorded for each sample. Only samples with ribosomal integrity number values above 9.0 were used for library preparation.
Total RNA samples were prepared for mRNA-Seq using the Illumina TruSeq Stranded mRNA Sample Preparation kit following the manufacturer's protocol (Illumina). Libraries were validated for length and adapter dimer removal using the Agilent TapeStation 4200 D1000 High Sensitivity assay (Agilent Technologies), and then they were quantified and normalized using the dsDNA High Sensitivity Assay for Qubit 3.0 (Thermo Fisher Scientific). Sample libraries were prepared for Illumina sequencing by denaturing and diluting the libraries per manufacturer's protocol (Illumina). All samples were pooled into one sequencing pool, equally normalized, and run as one sample pool across the Illumina NextSeq 500 using version 2.5 chemistry. Target read depth was achieved for each sample with paired end 75 bp reads. Raw reads were trimmed with Sickle (version 1.33), with a quality threshold of 30 and length threshold of 45, following that the trimmed reads were mapped to Homo Sapiens genome (GRCh38 ensembl release 99) with HISAT2 (version 2.1.0) (79). The resulting SAM files were then converted into BAM format using samtools (version 1.9) (80), and the PCR duplicates were removed using PICARD (http://broadinstitute.github.io/picard/). The counts were generated against the features with HTSeq-count (81). The differential expression of genes between conditions was evaluated using DESeq2 (82). Covariates were introduced in the DESeq2 analysis to increase the accuracy of results, and genes showing less than ten counts across the compared samples were excluded from the analysis. Genes with a false discovery rate <0.05 were considered significant and used in the downstream analysis. The processed RNA-Seq results were further analyzed by using IPA (Qiagen).

Immunoblotting
Cells from control and experimental mice were extracted in buffer containing 25 mM Tris-HCl (pH 7.5), 150 mM NaCl, 5% glycerol, 1 mM EDTA, 0.5% Triton X-100, 1 mM sodium orthovanadate, 10 mM NaF, 1 mM phenyl methyl sulfonyl fluoride, and a protease inhibitor cocktail (all from Sigma-Aldrich). Total cell lysates (40 μg of total protein) were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis in 8 or 12% polyacrylamide gels and transferred to Immobilon-P membranes (Millipore). The blots were probed with anti-HES1 (11988) and β-actin (3700) antibodies from Cell Signaling Technology or anti-NFATc1 antibody (556602) from BD Biosciences. The blots were exposed to anti-rabbit, antirat, or antimouse IgG conjugated to horseradish peroxidase (Sigma-Aldrich) and incubated with a chemiluminescence detection reagent (Bio-Rad). Chemiluminescence was detected by ChemiDoc XSR+ molecular imager (Bio-Rad) with Image Lab software (version 5.2.1), and the amount of protein present in individual bands was quantified (25).
μCT Femoral microarchitecture was determined using a μCT instrument (Scanco μCT 40; Scanco Medical AG), which was calibrated periodically using a phantom provided by the manufacturer (83,84). Femurs were scanned in 70% ethanol at high resolution, energy level of 55 kVp, intensity of 145 μA, and integration time of 200 ms. Evaluation of skeletal microarchitecture was started 1.0 mm proximal from the condyles of the distal femur. A total of 160 consecutive 6 μm thick slices were acquired at an isotropic voxel dimension of 216 μm 3 and selected for analysis. Contours were drawn manually every ten slices a few voxels away from the endocortical boundary to define the region of analysis. The remaining slice contours were iterated automatically. BV/TV, trabecular separation, number and thickness, connectivity density, SMI, and material density were measured in trabecular regions using a Gaussian filter (σ = 0.8) (83,84). For analysis of cortical bone, contours were iterated across 100 slices along the cortical shell of the femoral midshaft, excluding the marrow cavity. Analyses of BV/TV, cortical thickness, periosteal perimeter, endosteal perimeter, total cross-sectional area, and cortical bone area were conducted using a Gaussian filter (σ = 0.8, support = 1).

Bone histomorphometry
Bone histomorphometry was carried out in Ctsk Cre/WT ; Notch2 tm1.1Ecan ;Hes1 Δ/Δ , Ctsk Cre/WT ;Hes1 Δ/Δ , and Ctsk Cre/WT ; Rosa Hes1 mice, and sex-matched controls were injected with calcein 20 mg/kg and demeclocycline 50 mg/kg at a 5 or 7 days of interval and sacrificed 2 days after demeclocycline administration. For static cancellous bone histomorphometry and to assess for the presence of TRAP-positive multinucleated cells, bones were decalcified in 14% EDTA for 14 days and embedded in paraffin, and 7 μm sections were stained for the presence of TRAP and counterstained with hematoxylin and analyzed at a 100× magnification using OsteoMeasureXP software (Osteometrics). Stained sections were used to draw bone tissue and measure trabecular separation, number and thickness, and eroded surface, as well as to count osteoblast and osteoclast number. To assess dynamic parameters of bone histomorphometry, undecalcified femurs were embedded in methyl methacrylate, and 5 μm sections were cut using Microm microtome (Richards-Allan Scientific). Mineralizing surface per bone surface and mineral apposition rate were measured on unstained sections visualized under UV light and a triple diamidino-2-phenylindole/fluorescein/Texas red set long-pass filter, and bone formation rate was calculated (85).

Statistics
Data are expressed as means ± SD and presented as biological replicates except for experiments where BMMs were transduced with adenoviruses or cells were extracted for immunoblotting, and these are presented as technical replicates representative of two or more experiments. Statistical differences were determined by Student's t test or two-way analysis of variance with Tukey analysis for multiple comparisons, respectively.

Data availability
Data not shown will be shared upon request to Ernesto Canalis at canalis@uchc.edu.
Supporting information-This article contains supporting information.
and E. C. writing-original draft; J. Y. and E. C. writing-review and editing; J. Y. visualization; E. C. project administration; E. C. funding acquisition.
Funding and additional information-This work was supported by grants from the National Institute of Arthritis and Musculoskeletal and Skin Diseases, National Institutes of Health grant AR078149 (to E. C.) and AR072987 (to E. C.) and from the National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health DK045227 (to E. C.). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Conflict of interest-The authors declare no conflicts of interest with the contents of this article.