Osteocyte- and late osteoblast-derived NOTUM reduces cortical bone mass in mice

Osteoporosis is a common skeletal disease, with increased risk of fractures. Currently available osteoporosis treatments reduce the risk of vertebral fractures, mainly dependent on trabecular bone, whereas the effect on nonvertebral fractures, mainly dependent on cortical bone, is less pronounced. WNT signaling is a crucial regulator of bone homeostasis, and the activity of WNTs is inhibited by NOTUM, a secreted WNT lipase. We previously demonstrated that conditional inactivation of NOTUM in all osteoblast lineage cells increases the cortical but not the trabecular bone mass. The aim of the present study was to determine if NOTUM increasing cortical bone is derived from osteoblast precursors/early osteoblasts or from osteocytes/late osteoblasts. First, we demonstrated Notum mRNA expression in Dmp1- expressing osteocytes and late osteoblasts in cortical bone using in situ hybridization. We then developed a mouse model with inactivation of NOTUM in Dmp1 -expressing osteocytes and late osteoblasts ( Dmp1-creNotum ﬂ ox/ ﬂ ox mice). We observed that the Dmp1-creNotum ﬂ ox/ ﬂ ox mice displayed a substantial reduction of Notum mRNA in cortical bone, resulting in increased cortical bone mass and decreased cortical porosity in femur but no change in trabecular bone volume fraction in femur or in the lumbar vertebrae L5 in Dmp1-creNotum ﬂ ox/ ﬂ ox mice as compared with control mice. In conclusion, osteocytes and late osteoblasts are the principal source of NOTUM in cortical bone, and NOTUM derived from osteocytes/late osteoblasts reduces cortical bone mass. These ﬁ ndings demonstrate that inhibition of osteocyte/ late osteoblast-derived NOTUM might be an interesting pharmacological target to increase cortical bone mass and reduce nonvertebral fracture risk. NEW & NOTEWORTHY NOTUM produced by osteoblasts is known to regulate cortical bone mass. Our new ﬁ ndings show that NOTUM speci ﬁ cally derived by DMP1 -expressing osteocytes and late osteoblasts regulates cortical bone mass and not trabecular bone mass. P compared with Notum ox/ using Student s t test. A mixed model two-way ANOVA to assess the of genotype ( Dmp1-creNotum ﬂ ox/ ox and Notum ﬂ ox/ ox as as their ﬁ cant


INTRODUCTION
Osteoporosis is a common skeletal disease, leading to a reduction in bone density and quality and an increased fracture risk. One in two elderly women and one in four elderly men will at some point suffer from an osteoporotic fracture and the prevention of fractures is an important public health goal (1,2).
Approximately 80% of the human skeleton is comprised of cortical bone, and the cortical bone mass and quality are crucial for the overall bone strength. Most of the osteoporotic fractures occur at nonvertebral bone sites, mainly dependent on cortical bone (3,4). Currently used antiresorptive drugs substantially reduce the risk of vertebral fractures, mainly dependent on decreased trabecular bone mass, whereas the effect on nonvertebral fractures is less pronounced, suggesting that trabecular and cortical bone might respond differently to signals involved in the regulation of skeletal homeostasis (5). Although cortical bone mass and quality are major determinants of bone strength and fracture risk in humans (1,6,7), only a limited number of studies have focused on the cellular and molecular mechanisms specifically regulating cortical bone mass (1,6,7).
WNT signaling is a crucial regulator of bone homeostasis, and WNT ligands are known to affect bone mass by targeting the trabecular and/or cortical bone compartments (5,8). WNT16 is the major known cortical bone-specific WNT (5), while WNT10b protects against age-dependent trabecular bone loss (9,10). The activity of WNTs is inhibited by NOTUM, a secreted WNT lipase highly expressed by the osteoblast lineage cells in cortical bone (11)(12)(13)(14). NOTUM acts as a WNT inhibitor by deacetylating WNT proteins, making them unable to bind to their receptor Frizzled (12).
We have previously reported that osteoblast-derived NOTUM reduces cortical bone mass in mice and that the NOTUM locus is associated with bone mineral density in humans (11,13). The specific effect of NOTUM on cortical bone was demonstrated by mice with heterozygous global lifelong deletion of Notum, mice with lifelong inactivation of Notum in both early and late osteoblast-lineage cells (Runx2cre mouse model), and mice with a globally induced inactivation of Notum in adulthood (CAGG-cre-ER mouse model), all of which experienced increased cortical bone mass and cortical bone quality with no effect on the trabecular bone mass (11,13). In the mouse model with an inducible inactivation of Notum in adulthood, we observed that the increased cortical bone mass was caused by a rapid increase in cortical bone formation. In contrast, for the Notum inactivated mouse models with increased cortical bone mass as a result of lifelong inactivation of Notum, adult bone turnover was essentially unchanged as a new steady state had been reached (13).
It is unknown if NOTUM with an impact on cortical bone is derived from osteoblast precursors/early osteoblasts or from osteocytes/late osteoblasts. In this study, we hypothesized that osteocytes/late osteoblasts are the principal source of NOTUM with an impact on cortical bone. We, therefore, determined Notum expression in Dmp1-expressing osteocytes and late osteoblasts, and we evaluated the skeletal phenotype of a novel mouse model with Notum inactivated specifically in Dmp1-expressing osteocytes and late osteoblasts.
Male and female mice included in the experiment were euthanized at 20 wk of age. All animal experiments included in this study were approved by the Ethics Committee in Gothenburg, and the care of the animals was according to their guidelines. The mice were housed in a standard animal housing facility with a 12-h dark-light period. Food and water were available ad libitum. Before termination, the mice were given an intraperitoneal injection with Ketalar (Pfizer, New York, NY) and Dexdomitor (Orion Pharma, Esbo, Finland) before they were bled and euthanized with cervical dislocation.

Assessment of Bone Parameters
Dual-energy X-ray absorptiometry.

Peripheral quantitative computed tomography.
Peripheral quantitative computed tomography (pQCT) scans were performed using the peripheral quantitative computed tomography XCT Research M (v.4,5B; Norland Stratec, Pforzheim, Germany), using a voxel size of 70 mm, as previously described (18). The cortical bone parameters were analyzed in the mid-diaphyseal region of the femur, and the threshold was set to 710 mg/cm 3 (19).

High-resolution microcomputed tomography.
High-resolution microcomputed tomography (mCT) was used to analyze the distal femur and vertebra L5 (Skyscan, 1172; Bruker MicroCT, Aartselaar, Belgium), as previously described (5). Briefly, the cortical region was analyzed in the distal part of the femur, starting 5.3 mm from the growth plate and continuing 131 mm in proximal direction. The trabecular region in femur was analyzed $655 mm in the proximal direction from the distal growth plate and continued 131 mm in the proximal direction. For vertebra L5, the trabecular bone was analyzed 235 mm from the lower end of the pedicles and continued for $229 mm. The data were analyzed using the CTAn software (Bruker MicroCT).

Dynamic histomorphometry.
For the dynamic histomorphometric analyses of the middiaphyseal region of femur, the mice were injected intraperitoneally with the fluorochromes calcein and alizarin (Merck GmbH, Darmstadt, Germany) 9 and 2 days before they were euthanized, respectively (5). After dissection, the femurs were fixated in 4% formaldehyde and dehydrated in EtOH. The femurs were defatted in xylene and embedded in methyl methacrylate. The sections used had a thickness of 50-80 mm and were obtained from a standardized site of the femoral shaft. The analyses of dynamic cortical bone parameters were performed without staining. All parameters were analyzed using OsteoMeasure7 histomorphometry system (OsteoMetrics, Atlanta, GA) and following the guidelines of the American Society for Bone and Mineral Research (20).

Mechanical strength.
The left humerus was frozen at À20 C directly after dissection. Three-point bending of humerus was performed with a loading speed of 0.155 mm/s and a span length of 4.5 mm, using an Instron 3366 (Instron, Norwood, MA). The biomechanical parameters were based on the recorded load deformation curves and calculated from Bluehill 2 software v2.6 (Instron) with custom-made Excel (Microsoft, Redmond, WA) macros (5).

Real-Time Quantitative PCR
Total mRNA was prepared from bone using Trizol reagent (15596018; Thermo Fischer Scientific) and RNeasy Mini Kit (74106; Qiagen). The mRNA were reversed transcribed to cDNA (4368814; Thermo Fischer Scientific), and real-time PCR analyses were performed using the StepOnePlus Real-Time PCR System (Thermo Fischer Scientific) and the following Assay-on-Demand primer and probe sets: Notum, (encoding Dickkopf-1), Mm00438422_m1; Axin 2, Mm-00443610_m1; and cMyc Mm00487804_m1. The expressions of each gene were normalized to 18S ribosomal subunit (4310893E; Thermo Fischer Scientific). The 2 ÀDDCt method was used to calculate the relative gene expression.

Serum Analyses
The ELISA RatLaps Kit (AC-06F1, Immunodiagnostic Systems, East Boldon, UK) was used to measure serum levels of COOH-terminal type I collagen (CTX) fragments to assess bone resorption. To assess bone formation, serum levels of procollagen type I NH 2 -terminal propeptide (P1NP) were measured using a Rat/Mouse EIA Kit (AC-33F1, Immunodiagnostic Systems).

In Situ Hybridization
Femur from 5-mo-old wild-type mice were fixed in 4% formaldehyde in phosphate-buffered saline, followed by demineralization in 15% EDTA with 0.4% PFA for 3 wk. The decalcified femurs were dehydrated and processed for paraffin embedding. Six-micrometer-thick longitudinal consecutive dewaxed sections were used for in situ hybridization using RNAscope 2.5 HD Brown technology 322300 (Advanced Cell Diagnostics, Bio-Techne Ltd., Abingdon, UK). The following probes were used: Notum (428981, target sequence 406-1623, Advanced Cell Diagnostics) and Dmp1 (441171, target sequence 689-1543, Advanced Cell Diagnostics). A modified RNAscope manual protocol from Advanced Cell Diagnostics was used (21). Briefly, after pretreatment, the sections were hybridized with the probes overnight at 40 C, followed by six steps of amplification according to the manufacturer's instructions. To further amplify the signal, the DAB staining step was omitted and the signal was amplified with Digoxigenin-labeled Tyramide signal amplification (NEL 748001KT, Akoya Biosciences, Marlborough, MA), detected by alkaline phosphatase-conjugated Anti-Digoxigenin Fab fragments (11093274910, Sigma-Aldrich, St Louis, MO), and visualized with Liquid Permanent Red (K0640, DAKO, Carpinteria, CA). Finally, the sections were counterstained with Mayer's hematoxylin (MHS1, Sigma-Aldrich) and mounted with Aquatex (1085620050, Sigma-Aldrich).

Statistical Analyses
All values are presented as means ± SE. The statistical differences between Dmp1-creNotum flox/flox mice and Notum flox/flox mice were calculated using a mixed model two-way ANOVA evaluating the effects of genotype and sex as well as their interaction, using GraphPad Prism version 8.3.0 for Windows. Two-tailed Student's t tests were used when only one sex was evaluated. A difference was considered significant when P < 0.05.

Osteocytes and Late Osteoblasts Express Notum
We have previously demonstrated that Notum expression is high in cortical bone (11,13). To determine if Notum is expressed in Dmp1-expressing osteocytes and/or late osteoblasts in cortical bone, chromogenic in situ hybridization was performed. Notum mRNA was detected in Dmp1-expressing osteocytes and late osteoblasts (Fig. 1A), suggesting that Dmp1-expressing cells are the principal source of Notum in cortical bone.

Inactivation of Notum in Dmp1-Expressing Cells Leads to Increased Cortical Bone Mass and Reduced Cortical Porosity
To directly determine if Dmp1-expressing osteocytes and late osteoblasts are the principal source of Notum in cortical bone and to evaluate the skeletal effect of osteocyte/late osteoblast-derived Notum, we generated a conditional Notum-inactivated mouse model. To achieve inactivation of Notum specifically in osteocytes/late osteoblasts, mice with exons 2-8 of Notum flanked by LoxP sites were mated with mice expressing the Cre recombinase driven by the Dmp1 promoter (Fig. 1B), hereafter called Dmp1-creNotum flox/flox mice. Notum flox/flox littermates were used as controls.
The Dmp1-creNotum flox/flox mice were born according to Mendel's law of inheritance, and they appeared healthy compared with control littermates, with no difference in body weight (Fig. 1D) or femur length (Fig. 1E). Furthermore, the relative tissue weights of several tissues including liver, kidney, spleen, gonadal fat, retroperitoneal fat, seminal vesicles, and uterus were normal in the Dmp1-creNotum flox/flox mice ( Table 1).
As we previously demonstrated that mice with inactivation of Notum in both early and late osteoblast-lineage cells, using Runx2-creNotum flox/flox mice (13), have increased cortical bone mass in the diaphyseal region of femur, we first performed a DXA at 12 wk of age to evaluate the phenotype of the Dmp1-creNotum flox/flox mice. We observed an increase in total body bone mineral density, when compared with Notum flox/flox mice ( þ 5.4 ± 1.7%, P = 0.004; Supplemental Fig. S1; all Supplemental material is available at https://doi.org/ 10.6084/m9.figshare.13816040). After 20 wk, we screened the cortical bone parameters in the diaphyseal region of femur in Dmp1-creNotum flox/flox mice and Notum flox/flox littermates using pQCT. Two-way ANOVA analyses, evaluating both male and female mice, revealed that Dmp1-creNotum flox/flox mice displayed significantly increased cortical bone thickness ( þ 6.1 ± 1.5%, P < 0.001), cortical volumetric BMD ( þ 2.5 ± 0.6%, P < 0.001), and cortical bone mineral content ( þ 9.5 ± 2.1%, P < 0.0001) in the  Values are given as means ± SE (Notum flox/flox : females n = 10, males n = 8; Dmp1-creNotum flox/flox : females n = 7, males n = 5, 20-wk-old mice) in mg/g body wt. A mixed model two-way ANOVA was used to evaluate the effect of genotype, sex, and their interaction. Twotailed Student's t tests were used when only one sex was evaluated. A difference was considered significant when P < 0.05. diaphyseal region of femur compared with Notum flox/flox littermates, with no significant sex interaction for any of these effects on cortical bone (Fig. 2, A-D). Analyses of the periosteal and endocortical circumferences in Dmp1-creNotum flox/flox mice did not reveal any significant effects (Fig. 2, D and E). However, a nonsignificant tendency of increased periosteal circumference ( þ 1.1 ± 1.1%, P = 0.2) was observed in Dmp1-creNotum flox/flox mice compared with Notum flox/flox mice. The increased cortical bone mass in Dmp1-creNotum flox/flox mice was confirmed by mCT (Fig. 2F), revealing increased cortical bone area ( þ 7.6 ± 2.3%, P = 0.003) in the femur diaphysis compared with Notum flox/flox littermates (Fig. 2G). mCT analysis also revealed lower cortical porosity (À5.7 ± 1.9%, P = 0.006) in Dmp1-creNotum flox/flox mice compared with Notum flox/flox littermates, indicating that inactivation of Notum results not only in increased cortical bone mass but also in a higher quality of the cortical bone (Fig. 2H). No significant sex interaction was observed for the effect of Notum inactivation on cortical bone area or on cortical porosity (Fig.  2, G and H). The cross-sectional bone marrow area was not affected in Dmp1-creNotum flox/flox mice compared with Notum flox/flox littermates (Fig. 2I). The increased amount of cortical bone in the diaphyseal region of femur in Dmp1-creNotum flox/flox mice was also confirmed by histomorphometry analyses ( Table 2). In contrast to the clear effects on cortical bone parameters, Dmp1-creNotum flox/flox mice had unchanged trabecular bone volume fraction in the distal metaphyseal region of femur and in the lumbar vertebra L5 as measured using mCT (Fig. 2, J and K). Thus, similar to the mice with inactivation of Notum in both early and late osteoblast-lineage cells, using Runx2-creNotum flox/flox mice (13), Notum inactivation specifically in osteocytes/late osteoblasts, using Dmp1-creNotum flox/flox mice, results in increased cortical bone mass without any effect on trabecular bone volume fraction.
Three-point bending analyses of the mechanical strength of humerus revealed a nonsignificant tendency of increased maximal load at failure ( þ 7.7 ± 5.0%, P = 0.1) in Dmp1-creNotum flox/flox mice compared with Notum flox/flox mice while no effect was observed on toughness or stiffness (Table 3).

Dmp1-creNotum flox/flox Mice
We have previously demonstrated that an acute inducible Notum inactivation in adulthood results in an increased cortical bone thickness via a rapid increase in cortical bone formation, while the increased cortical bone thickness in adult Runx2-creNotum flox/flox mice with lifelong Notum inactivation is not associated with a sustained adult increased bone formation, most likely as a new steady state of bone turnover has been reached (13). To determine if the increased cortical bone mass in the Dmp1-creNotum flox/flox mice with lifelong Notum inactivation in osteocytes/late osteoblasts is associated with an increased cortical bone formation in adulthood, dynamic histomorphometry was performed in the mid-diaphyseal cortical region of femur in 20wk-old mice. Neither the periosteal nor the endocortical bone formation rate was affected in the adult Dmp1- The results refer to 20-wk-old mice. All values are given as means ± SE. A mixed model two-way ANOVA was used to assess the effects of genotype (Dmp1-creNotum flox/flox and Notum flox/flox ), sex, as well as their interaction. A difference was considered significant when P < 0.05. Values are given as means ± SE (Notum flox/flox : females n = 10, males n = 8; Dmp1-creNotum flox/flox : females n = 7, males n = 5, 20-wk-old mice). A mixed model two-way ANOVA was used to evaluate the effect of genotype, sex, and their interaction. A difference was considered significant when P < 0.05. creNotum flox/flox mice compared with Notum flox/flox mice ( Table 2). Furthermore, gene expression analyses of cortical bone revealed unchanged mRNA levels both of genes reflecting bone formation and of genes reflecting bone resorption (Supplemental Table S1). Analyzes of the expression of osteocyte and WNT signaling genes showed no differences in mRNA levels between Dmp1-creNotum flox/flox and Notum flox/flox mice (Supplemental Table S1). In addition, no significant effect of lifelong Notum inactivation in Dmp1-creNotum flox/flox mice was observed on the adult serum levels of the bone resorption marker COOH-terminal telopeptide (CTX) or the bone formation marker Procollagen type I NH 2 -terminal propeptide (P1NP; Supplemental Table S2).
These results indicate that a new steady state of bone remodeling has been reached in the adult Dmp1-creNotum flox/flox mice with lifelong inactivation of Notum in osteocytes/late osteoblasts.

DISCUSSION
Cortical bone is a major determinant of fracture risk at nonvertebral bone sites, but the knowledge of the mechanisms for regulation of cortical bone mass is limited (22). We have previously demonstrated that the WNT lipase NOTUM is a major regulator of cortical bone mass (11,13). However, the exact origin of NOTUM with an impact on cortical bone was unknown. We, herein, demonstrate Notum expression in Dmp1-expressing osteocytes/late osteoblasts in cortical bone. Furthermore, in functional studies we demonstrate that Dmp1-expressing osteocytes/late osteoblasts are the principal source of Notum in cortical bone and that Notum derived from osteocytes/late osteoblasts reduces cortical bone mass. There is medical need for new safe treatment strategies that increase cortical bone mass and thereby reduce nonvertebral fracture risk. The present findings, together with our recent reports (11,13), indicate that NOTUM inhibition may be a promising drug target to increase cortical bone mass and reduce nonvertebral fracture risk.
We have previously demonstrated that Notum is highly expressed in cortical bone and that relatively high Notum expression is observed in cultured osteoblasts but not osteoclasts (11,13). In the present study, we extended those findings by showing that Notum is expressed in Dmp1expresssing osteocytes/late osteoblasts using the in situ hybridization technique RNAscope. We, therefore, hypothesized that osteocytes/late osteoblasts are the principal source of NOTUM with an impact on cortical bone. To directly determine if Dmp1-expressing osteocytes and late osteoblasts are the principal source of NOTUM in cortical bone and to evaluate the skeletal effect of osteocytes/late osteoblasts-derived NOTUM, we generated a conditional Notum-inactivated mouse model. We observed that the Dmp1-creNotum flox/flox mice displayed a substantial reduction of Notum mRNA in cortical bone, demonstrating that the major part of NOTUM in cortical bone is indeed derived from osteocytes/late osteoblasts. In contrast, no difference in Notum mRNA levels was observed in the liver, confirming the specificity of the Notum inactivation in the Dmp1-creNotum flox/flox mouse model.
The Dmp1-creNotum flox/flox mice appeared healthy, with no difference in body weight or femur length, demonstrating that the observed increased cortical bone mass is specific and not only the result of an increased overall growth of the mice. This finding correlates with our previous finding using mice with inactivation of Notum in both early and late osteoblast-lineage cells, using Runx2-creNotum flox/flox mice that also had increased cortical bone mass but normal overall body size (13).
The reduction of Notum mRNA in cortical bone and the skeletal phenotype observed in the Dmp1-creNotum flox/flox mice, including increased cortical bone mass but unchanged trabecular bone mass, are identical to what we have previously reported for mice with lifelong inactivation of Notum in both early and late osteoblast lineage cells and mice with global inducible inactivation of Notum in adulthood (13). When comparing the effects of Notum inactivation in the Dmp1-creNotum flox/flox mice, in the present study, and the effects of Notum inactivation in the Runx2-creNotum flox/flox mice from our previous report (13), we observed that Notum mRNA in cortical bone was substantially reduced (Dmp1-creNotum flox/flox mice: À77.6 ± 9.3%; Runx2-creNotum flox/flox mice: À87.9 ± 11.4%) while cortical bone area was significantly increased (Dmp1-creNotum flox/flox mice: þ 7.6 ± 2.3%; Runx2-creNotum flox/flox mice: þ 13.5 ± 2.5%) in both mouse models (13). In addition, the trabecular bone volume fraction in the vertebra or femur was not significantly affected in any of the two mouse models (13). Thus the two mouse models display a similar specific cortical bone phenotype, strongly suggesting that NOTUM in osteocytes/late osteoblasts but not in osteoblast precursors/early osteoblasts is crucial for the physiological effect of NOTUM on cortical bone.
These functional data demonstrate that NOTUM derived from osteocytes/late osteoblasts reduces cortical bone mass. Deletion of b-catenin in osteocytes has been reported to reduce bone mass in mice, demonstrating that WNT/b-catenin signaling in osteocytes is needed for normal bone homeostasis (23). It is, therefore, possible that NOTUM inhibits b-catenin-dependent WNT signaling in osteocytes of cortical bone. Values are given as means ± SE (Notum flox/flox : females n = 10, males n = 8; Dmp1-creNotum flox/flox : females n = 7, males n = 5, 20-wk-old mice). A mixed model two-way ANOVA was used to evaluate the effect of genotype, sex, and their interaction. A difference was considered significant when P < 0.05.
As the inactivation of Notum in the Dmp1-creNotum flox/flox mice is lifelong, one may argue that the observed effects on the skeleton may be the result of early developmental effects in this mouse model. However, the identical cortical bone phenotype observed in our previous study using global inducible Notum inactivation in adulthood strongly argues against major confounding developmental effects in the Dmp1-creNotum flox/flox mice (13).
We have previously demonstrated that an acute inducible Notum inactivation in adulthood results in an increased cortical bone thickness via a rapid increase in adult cortical bone formation while the increased cortical bone thickness in the adult Runx2-creNotum flox/flox mice with lifelong Notum inactivation is not associated with a sustained adult increased bone formation, most likely as a new steady state of bone turnover had been reached (13). Consistent with these previous findings, the increased cortical bone mass in the adult Dmp1-creNotum flox/flox mice with lifelong Notum inactivation in osteocytes/late osteoblasts in the present study was not associated with any significant changes in parameters reflecting bone formation, bone resorption, or WNT signaling, indicating that a new steady state of bone remodeling has been reached in the adult Dmp1-creNotum flox/flox mice.
A limitation of the present study is the lack of further cellular and molecular mechanisms by which NOTUM exerts its effect besides those which we have previously reported (13).
Another limitation of the present study is the relative low numbers of mice, resulting in low power in some of the analyses, including the bone strength measurements using three-point bending.
In conclusion, Dmp1-expressing osteocytes/late osteoblasts are the principal source of NOTUM in cortical bone and NOTUM derived from osteocytes/late osteoblasts reduces cortical bone mass. These findings support the concept that inhibition of osteocytes/late osteoblasts-derived NOTUM might be an interesting pharmacological target to increase cortical bone mass and reduce nonvertebral fracture risk.