Short-term glucocorticoid excess blunts abaloparatide-induced increase in femoral bone mass and strength in mice

Glucocorticoids (GCs), such as prednisolone, are widely used to treat inflammatory diseases. Continuously long-term or high dose treatment with GCs is one of the most common causes of secondary osteoporosis and is associated with sarcopenia and increased risk of debilitating osteoporotic fragility fractures. Abaloparatide (ABL) is a potent parathyroid hormone-related peptide analog, which can increase bone mineral density (aBMD), improve trabecular microarchitecture, and increase bone strength. The present study aimed to investigate whether GC excess blunts the osteoanabolic effect of ABL. Sixty 12–13-week-old female RjOrl:SWISS mice were allocated to the following groups: Baseline, Control, ABL, GC, and GC + ABL. ABL was administered as subcutaneous injections (100 μg/kg), while GC was delivered by subcutaneous implantation of a 60-days slow-release prednisolone-pellet (10 mg). The study lasted four weeks. GC induced a substantial reduction in muscle mass, trabecular mineral apposition rate (MAR) and bone formation rate (BFR/BS), and endocortical MAR compared with Control, but did not alter the trabecular microarchitecture or bone strength. In mice not receiving GC, ABL increased aBMD, bone mineral content (BMC), cortical and trabecular microarchitecture, mineralizing surface (MS/BS), MAR, BFR/BS, and bone strength compared with Control. However, when administered concomitantly with GC, the osteoanabolic effect of ABL on BMC, cortical morphology, and cortical bone strength was blunted. In conclusion, at cortical bone sites, the osteoanabolic effect of ABL is generally blunted by short-term GC excess.

Tissue extraction and femoral length. Immediately after exsanguination, both femora, the right tibia, and L4 were isolated, and soft tissue was carefully removed. The right femur, right tibia, and L4 were stored in Ringer's solution at − 20 °C, while the left femur was immersion-fixed in 0.1 M sodium phosphate-buffered formaldehyde (4% formaldehyde, pH 7.0) for 48 h and then stored in 70% ethanol.
The femoral bone length was determined using a digital sliding caliper and measured from the medial condyle to the top of the femoral head.
Rectus femoris muscle. The right rectus femoris muscle was removed in a standardized manner, and the wet weight was determined immediately using a digital scale (Mettler AT250, Columbus, OH, USA). Then, the muscles were halved at the midpoint, placed with the cut surface on a flat-bed image scanner (Perfection 3200 Photo; Seiko Epson, Nagano, Japan), and scanned to determine the whole muscle cross-sectional area (CSA) 31 .
The halved rectus femoris muscle was then immersion-fixed in 0.1 M sodium phosphate-buffered formaldehyde (4% formaldehyde, pH 7.0) and embedded in plastic-based 2-hydroxyethyl methacrylate (Technovit 7100, Heraeus Kulzer, Wehrheim, Germany). The embedded muscles were cut into 2-μm-thick sections on a microtome (Jung RM2065; Leica Instruments, Nussloch, Germany) and stained with Masson's trichrome to determine muscle cell CSA. Muscle cell CSA was determined using a light microscope (Nikon Eclipse i80, Tokyo, Japan) at a final magnification of × 1132 as previously described 29 .
Dual-energy X-ray absorptiometry (DXA). Areal bone mineral density (aBMD) and bone mineral content (BMC) were determined using a desktop DXA (pDEXA Sabre XL; Norland Stratec, Pforzheim, Germany) as previously described 32 . In brief, the whole femur and tibia were scanned using a pixel size of 0.1 × 0.1 mm 2 and a scan speed of 3.0 mm/s. www.nature.com/scientificreports/ Micro computed tomography (μCT). Cortical and trabecular microarchitectural bone parameters of the distal femoral metaphysis and epiphysis and the vertebral body of L4 were determined using a desktop µCT scanner (Scanco µCT 35; Scanco Medical, Brüttiselen, Switzerland) ( Fig. 1). All scans were performed in highresolution mode (1000 projections/180°) with an isotropic voxel size of 3.5 μm, an X-ray tube voltage of 55 kV, a current of 114 μA, and an integration time of 800 ms. Beam hardening artifacts were reduced using a 0.5 mm aluminum filter. The bone samples were analyzed using separate volumes of interest (VOIs) as previously described 27 (Fig. 1). The VOI at the femoral mid-diaphysis comprised an 820-μm-high region (234 slices) centered on the femoral mid-point, thus containing cortical bone only. The VOI at the distal femoral metaphysis started 200 μm (57 slices) above the point, where the mineralized cartilage from the growth plate fused and ended 1000 μm (229 slices) further above and contained trabecular bone only. The femoral epiphysis was analyzed using a 490-μm-high (140 slices) VOI that started just after the medial and lateral epicondyle have fused to a coherent structure and ended, where the growth plate first appeared, thus containing trabecular bone only. Finally, L4 was analyzed using an approximately 2000-μm-high (571 slices) VOI spanning from the upper to the lower growth plate excluding primary spongiosa containing trabecular bone only. The data were low-pass filtered using a Gaussian filter (σ = 0.8 and support = 1) and segmented with a fixed threshold filter determined with IPL (v. 5.11, Scanco Medical AG, Switzerland). The thresholds used for segmentation were 574 mg HA/cm 3 for femoral bone and 529 mg HA/cm 3 for L4.
The assessment of bone microstructure followed the current guiding principles by Bouxsein et al. 33 . The following microstructural bone parameters were assessed with µCT: Ct.Th: cortical thickness, B.Ar: bone area, M.Ar: marrow area, T: tissue area, Ct.Po: cortical porosity, BV/TV: bone volume/tissue volume, Tb.Th: trabecular  Mechanical testing. Mechanical properties of the femoral mid-diaphysis, femoral neck, and vertebral body of L4 were determined using a material testing machine (Instron model 5566, United Kingdom) as previously described 31,32 . Maximum load to failure was the ultimate load achieved at any point during the compression testing and was determined from the load-displacement data using in-house developed software. The femur was placed on two rounded supporting bars separated by 7.14 mm with the anterior side facing away from the supporting bars for the 3-point bending test. Load was applied with a third rounded bars at the femoral mid-point. Then, the proximal part of the femur was placed in a custom-made fixation device exposing the femoral neck for mechanical testing. L4 was compression tested after the vertebral disc and processi were removed. For all testing procedures, vertical load was applied at a constant deflection rate of 2 mm/min.

Dynamic bone histomorphometry. Cortical bone.
After mechanical testing, a 200-μm-thick middiaphyseal cross-sectional bone slice was sawed from the distal part of the femoral bone and mounted with Pertex on glass slides. Fluorochrome bone labels were analyzed using a brightfield microscope equipped for fluorescence (Nikon Eclipse i80, Tokyo, Japan). Live images were projected to a computer equipped with the stereological software Visiopharm (v. 2020.09, Visiopharm, Hørsholm, Denmark). A 24-arm circular grid radiating outwards from the center of the bone marrow was superimposed on live images to quantify fluorochrome labels. The distance between double labels was measured using a built-in measure tool in the Visiopharm Software.
Mineralizing surface/bone surface (MS/BS), mineral apposition rate (MAR), and bone formation rate (BFR/ BS) was determined as previously described 29 . In the case of missing double labels an imputed value of zero was used to calculate MAR 34 . Tetracycline-covered surfaces (Tetra.S/BS) were used as an indicator of bone resorption throughout the experiment 28,29 . The assessment of fluorescent labels was performed at a final magnification of × 1132.
Trabecular bone. The distal part of the femur was embedded undecalcified in methacrylate and 7-μm-thick longitudinal sections were cut using a microtome (Jung RM2065; Leica Instruments, Nussloch, Germany). The sections were either left unstained for dynamic bone histomorphometry, Masson-Goldner trichome-stained for osteoblast, osteoid, and adipocyte quantification or stained for tartrate-resistant acid phosphatase (TRAP) to identify osteoclasts (Fig. 2). Fluorochrome labels were quantified in a 1000-μm-high region of interest (ROI) starting 300 μm below the most distal part of the growth plate. The fields of view were sampled systematically covering 100% of the ROI with a randomly orientated line grid superimposed on the fields of view. The assessment of fluorescent labels was performed at a final magnification of × 1132.
Bone cells and adipocytes. The Masson-Goldner trichome-stained sections from the distal part of the femur were used to estimate the amount of osteoblast (Ob.S/BS), osteoid-covered surfaces (OB/BS), bone marrow adiposity, and adipocyte mean size. Osteoblasts were defined as cuboidal cells residing at the bone surface and osteoid was defined as the red unmineralized region running along the edge of the bone. Using a point grid, the bone marrow adiposity (adipocyt volume density) was estimated as the total number of points hitting adipocytes within the bone marrow divided by total number of point hitting the bone marrow. Adipocytes were defined as cells containing a single large lipid droplet surrounded by a layer of cytoplasm 35 .
Sections stained for TRAP were used to determine the amount of osteoclast-covered surfaces (Oc.S/BS). Osteoclasts were defined as TRAP-positive cells located at the bone surface.
A similar ROI, final magnification, and superimposed randomly orientated line grid used for trabecular dynamic bone histomorphometry were used to analyze the sections stained with Masson-Goldner trichome and the sections stained for TRAP.
Differences between groups were analyzed using a parametric two-way analysis of variance (ANOVA) followed by a post-hoc Holm-Sidak test. Normality was assessed using the D' Agostino-Pearson test. The mean of all groups was compared to the mean of every other group, except Baseline. The two-way ANOVA is capable of detecting interactions, wherein a significant interaction indicates a difference in the effect of treatment with ABL between mice with a placebo-pellet (Control) or a glucocorticoid-pellet (GC) implanted. Data are presented as means ± SD and statistical significance was considered achieved if the p-value was less than 0.05. An a priori sample size calculation (power = 0.8) on mice showed that a 5% difference in femoral aBMD and a 15% difference in mid-diaphyseal bone strength can be demonstrated between groups with n = 12 animals 27 .

Results
Animals, muscle mass, and bone length. The final body weight was not affected by injections with ABL or insertion of subcutaneous GC-pellets compared with Control ( Table 1).
As expected, treatment with ABL did not affect rectus femoris muscle weight, rectus femoris muscle CSA, or rectus femoris muscle cell CSA (Table 1, Fig. 3a,b). GC alone or in combination with ABL reduced rectus femoris muscle mass (− 28% and − 26%), rectus femoris muscle CSA (− 16% and − 25%), and muscle cell CSA (− 24% and − 24%) compared with Control, respectively. In addition, GC not only resulted in muscle cell atrophy, but also in muscle cell destruction and infiltration of adipocytes into the muscle (Fig. 3c,d).
Neither ABL nor GC affected femoral bone length (Table 1).   www.nature.com/scientificreports/ Dual-energy X-ray absorptiometry. Treatment with ABL increased both femoral and tibial aBMD (+ 12% and + 13%) and BMC (+ 21% and + 21%) compared with Control, respectively (Table 1). GC did not reduce aBMD or BMC compared with Control. However, aBMD and BMC were lower at both femur and tibia in animals treated with both ABL and GC compared to animals treated with ABL only, demonstrating that GC excess blunted the osteoanabolic effect of ABL on femoral and tibial aBMD and BMC. In addition, an interaction between treatment with ABL and GC was found for femoral BMC.
Micro-computed tomography. Cortical bone. At the femoral mid-diaphysis, ABL increased B.Ar (+ 21%) and T.Ar (+ 10%) compared with Control. GC alone decreased B.Ar with 5% compared with Control, as the only observable negative effect (Table 2). An interaction between treatment with ABL and GC was found for B.Ar. Treatment with GC + ABL resulted in lower B.Ar compared with ABL indicating that the effect of ABL on mid-diaphyseal B.Ar was blunted by GC. Neither GC nor ABL had any effects on mid-diaphyseal M.Ar, Ct.Po, or TMD.
Mechanical testing. Treatment with ABL increased femoral mid-diaphyseal bone strength (+ 21%) compared to Control, while GC did not affect mid-diaphyseal bone strength compared to Control. In animals receiv-  www.nature.com/scientificreports/ ing GC alone or in combination with ABL, the mid-diaphyseal bone strength was lower (− 28% and − 26%) than in those treated with ABL alone, respectively (Table 5). An interaction between treatment with ABL and GC was found for mid-diaphyseal bone strength.
At the femoral neck, the bone strength did not differ between any of the groups (Table 5).
Mechanical compression testing revealed that ABL increased vertebral bone strength in both ABL (+ 52%) and GC + ABL (+ 66%) animals compared to Control. In the GC + ABL group, no difference was observed in vertebral compression strength compared with ABL. Moreover, ABL increased bone strength in GC animals (+ 92%) compared with GC animals that was not treated with ABL (Table 5).
These findings indicate that the anabolic effect of ABL on bone strength was blunted by GC at the femoral mid-diaphysis, but not at the vertebral body.
The lower bone turnover in the CG and GC + ABL mice compared to the ABL mice resulted in a more mineralized bone in these animals, which is consistent with the higher TMD in these mice (Table 3).
Cortical bone at the femoral mid-diaphysis. At the periosteal bone surface, ABL increased MS/BS (+ 73%), MAR (+ 128%), and BFR/BS (+ 255%) compared with Control. GC decreased periosteal MS/BS (− 79%), but did not affect either MAR or BFR/BS (Fig. 5). Animals in the GC group treated with ABL did not increase any of the formative bone parameters compared with Control. In addition, an interaction between treatment with ABL and GC was found for BFR/BS and animals in the GC + ABL group had lower MS/BS (− 36%), MAR (− 59%), and BFR/BS (− 69%) compared with animals in the ABL group suggesting that GC blunted the effect of ABL on the bone formative parameters at the periosteal femoral mid-diaphyseal bone surface.
Neither GC nor ABL had any influence on the resorptive indicator Tetra.S/BS at the periosteal surface at the femoral mid-diaphysis (Fig. 4).
At the endocortical bone surface, treatment with ABL did not influence any of the formative bone parameters assessed with dynamic bone histomorphometry in either Control or GC animals. Animals in the GC group had  www.nature.com/scientificreports/ lower MS/BS (− 68%) and MAR (− 69%) compared with Control indicating that GC inhibited bone formation at the endocortical bone surface. An interaction between treatment with ABL and GC was found for Tetra.S/BS. The lower Tetra.S/BS indicated that endocortical bone resorption was elevated in GC + ABL-treated mice compared with Control, ABL, and GC mice (Fig. 5). www.nature.com/scientificreports/ Adipocytes. A non-significant trend towards lower bone marrow adiposity in ABL treated animals compared with Control was observed. GC did not affect bone marrow adiposity, but increased adipocyte size (+ 48%) compared with Control (Fig. 6). ABL prevented the GC-induced increase in adipocyte size.

Discussion
The purpose of the study was to investigate whether the osteoanabolic effect of ABL was affected by short-term GC excess in mice. The main finding of the study was that GC mainly blunted the osteoanabolic effect of ABL in cortical bone. In mice not receiving GC, the potent osteoanabolic effect of ABL materialized as a substantial increase in bone mineral density, cortical and trabecular microarchitecture, and bone strength. These findings are in accordance with the results of previous studies performed in mice 36,37 . In addition, ABL increased the amount of osteoblastcovered surfaces and trended to reduce bone marrow adiposity. This suggests a possible differentiation of bone marrow mesenchymal stem cells towards osteoblasts instead of bone marrow adipocytes 38 , and is consistent with the observed lipolytic effect of PTHrP in mice 39,40 . Similar decrease in bone marrow adiposity has been reported for two other osteoanabolic agents teriparatide and romosozumab in rats 41 .
As expected, treatment with ABL did not influence muscles mass, while muscle wasting is a well-established effect of treatment with GC 42 . In accordance with that, we observed a massive reduction in muscle wet weight, muscle cross-sectional area, and muscle cell cross-sectional area in animals subjected to GC excess.
In addition to muscle wasting, GC has been shown to induce a substantial bone loss 43 . Surprisingly, we found that short-term GC excess had little or no detrimental effect on bone mineral density, trabecular microarchitecture, and bone strength. www.nature.com/scientificreports/ Subcutaneously implanted pellets have been used in various mouse strains such as Swiss Webster 44-51 , CD-1 52,53 , C57BL/6 54-58 , Black Swiss 59 , and FVB/N 60 to explore the skeletal effects of GC, but conflicting results have emerged. Several studies have reported a substantial loss of trabecular bone volume fraction or aBMD in long bones after only 21-28 days using GC doses of 2-15 mg/kg/day 44,[48][49][50]52,54,60 , while others have reported little or no detrimental effect from implanted GC-pellets 45,59,61 . One might speculate whether the large heterogeneity in the outcomes of the different studies reflects differences in strain, GC dose, and study duration. Ersek et al. reported a GC-induced reduction in trabecular bone volume fraction, cortical thickness, and bone strength in CD-1 and C57/BL6 mice and concluded that CD1-mice (that are derived from the same mice as the RjOrl:SWISS mice used in the present study) are particularly susceptible to GC-induced bone loss compared to C57BL/6 mice. They used similarly aged mice, prednisolone pellets with a four times lower dose (2.5 mg/60 days), and a longer study duration (56 days) compared to the present study 53 . However, their findings contrast those of the present study, where GC excess had little or no detrimental effect on trabecular bone. One possible explanation for the discrepancy between the two studies is that GC-induced bone loss in RjOrl:SWISS mice requires a longer time to develop than the 28 days used in the present study or that this mouse strain is less susceptible to GC-induce bone loss. We found that GC in general had limited effect on trabecular bone, but nevertheless resulted in decreased bone resorption and formation indicators. This indicate that the microstructural changes are developing, but have not yet materialized. Furthermore, the lower trabecular bone turnover caused higher trabecular tissue mineral density in GC treated mice, signifying a more mineralized trabecular bone tissue.
In contrast to trabecular bone, GC had a noticeable effect on cortical bone. This is consistent with previous studies in mice using either GC-releasing pellets 53,61 or subcutaneous injections with dexamethasone 62,63 , where the GC-induced bone loss was most pronounced at skeletal sites comprising cortical bone.
The present study clearly demonstrated that the osteoanabolic effect of ABL was blunted by concomitant short-term exposure to GC excess. The GC-induced blunting of ABL was more pronounced at cortical bone sites and less marked at trabecular bone sites, reflecting that GC mainly had detrimental effect on cortical bone. This might be explained by GC predominantly target cortical bone surfaces 61 , thus mainly challenging the osteoanabolic effect of ABL at the cortical bone envelopes, but not at trabecular bone surfaces.
The effect of ABL has previously been studied in botulinum toxin-immobilized rats 64 , ovariectomized rats 65 , rabbits 66 , and monkeys 26 , and intact mice 27,36,67 , but the present study is the first conducted in animals subjected to GC excess. Although no previous study has investigated the effect of ABL and GC in combination, conflicting results have emerged from studies of rodents exposed to GC and treated with teriparatide 20,62,68 . Teriparatide is structurally very similar to ABL 69 , acts through the same receptor on osteoblasts -albeit mainly in different receptor configurations 70 , and has, in general, the same osteoanabolic effect in intact mice 27,36 . One study found that GC-induced bone loss is not prevented by simultaneous treatment with teriparatide in mice 62 , while another found teriparatide effective in increasing bone mass and strength in mice subjected to GC excess by implantation of slow-release prednisolone pellets 68 . A study in rats found that GC severely blunted the anabolic effect of simultaneous treatment with teriparatide 20 .
No clinical study has investigated the osteoanabolic effect of ABL in patients with glucocorticoid-induced osteoporosis. However, one large clinical multi-center study compared the osteoanabolic effect of daily injections with teriparatide (20 μg) or ABL (80 μg) in women with post-menopausal osteoporosis 22 . They found that treatment with ABL improved femoral neck, lumbar spine, and total hip aBMD compared with treatment with placebo or teriparatide. Interestingly, the study also showed treatment with ABL entailed a lower risk of hypercalcemia than treatment with teriparatide.
At skeletal sites mainly composed of trabecular bone such as L4 and the distal femoral metaphysis and epiphysis, treatment with ABL resulted in higher BV/TV and vBMD values in both non-GC and GC exposed animals compared with Control. This suggests that GC does not blunt the osteoanabolic effect of ABL on trabecular bone density at these skeletal sites. Moreover, the increased trabecular bone density was accompanied by higher bone strength, although it was statistically significant for L4 only.
Despite GCs blunting effect on cortical bone, ABL is nevertheless a suitable candidate for counteracting GCinduced bone loss, especially when it is taken into consideration that the anabolic effect of teriparatide is similarly blunted by GC. Moreover, as treatment with ABL has shown a decreased risk of hypercalcemia, ABL provides an alternative to teriparatide and an attractive treatment option for glucocorticoid-induced osteoporosis to selected patients at risk of hypercalcemia. In addition, blunting of the osteoanabolic effect of ABL may be alleviated by increasing the dose, an option not available for teriparatide, since ABL is approved by the U.S. Food and Drug Administration to be administered at four times higher doses than teriparatide 71 .
The present study has some limitations. The assessment of bone marrow adipocytes was performed on Masson-Goldner trichrome-stained sections using cell morphology to identify adipocytes instead of using immunohistochemical staining for lipid-associated proteins like perilipin or adipophilin. However, we have previously conducted an unpublished pilot study comparing bone marrow adiposity assessed by cell morphology on Masson-Goldner trichrome stained sections and immunohistochemical staining for perilipin and found the two methods to provide similar results. In addition, the International Bone Marrow Adiposity Society states that Masson-Goldner trichrome staining can be used to identify mature adipocytes, while immunohistochemical staining for perilipin may be used to more easily discriminate between blood vessels and adipocytes 72 .
ABL was injected five days a week due to staff conveniences, whereas in a clinical setting ABL is administrated seven days a week. It cannot be precluded that this difference may influence the anabolic effect of the ABL treatment regimen. In addition, the study was designed as a prevention study where treatment with ABL commenced when the GC pellets were implanted instead of as an intervention study where a bone loss had been allowed to develop before starting the ABL treatment.
Another limitation of the study is that GC excess was not unequivocally established by the serum prednisolone or corticosterone levels, but only bona fide established by subcutaneous implantation of prednisolone releasing www.nature.com/scientificreports/ pellets and the observable pronounced negative effect on skeletal muscle tissue. A study by Herrmann et al. used mice with the same genetic background and dosage of GC pellets as in the present study, and did not find any increase in serum corticosterone after two and three weeks of pellet implantation, suggesting serum measurements might not establish the presence of GC excess 52 . The present study investigated short-term treatment with GC and lasted four weeks and, surprisingly, only a relatively limited detrimental effect of GC was observed for both cortical and trabecular bone. If the study had lasted longer or the dosage of GC had been higher, a more pronounced bone loss and reduction in bone strength might have occurred.
We used mice to model glucocorticoid-induced bone loss since previous studies have shown subcutaneously inserted prednisolone pellets can induce osteopenia in mice 46,57 . Mice do not show Haversian-like remodeling, but this can be achieved using rabbits, which undergo intracortical remodeling 73 . Moreover, glucocorticoidexcess in rabbits might be combined with estrogen deficiency because of the narrow therapeutic window 74,75 .
In conclusion, the osteoanabolic effect of ABL is generally blunted by short-term glucocorticoid excess in particular at cortical bone sites.