Sustained Morphine Delivery Suppresses Bone Formation and Alters Metabolic and Circulating miRNA Profiles in Male C57BL/6J Mice

Opioid use is detrimental to bone health, causing both indirect and direct effects on bone turnover. Although the mechanisms of these effects are not entirely clear, recent studies have linked chronic opioid use to alterations in circulating miRNAs. Here, we developed a model of opioid‐induced bone loss to understand bone turnover and identify candidate miRNA‐mediated regulatory mechanisms. We evaluated the effects of sustained morphine treatment on male and female C57BL/6J mice by treating with vehicle (0.9% saline) or morphine (17 mg/kg) using subcutaneous osmotic minipumps for 25 days. Morphine‐treated mice had higher energy expenditure and respiratory quotient, indicating a shift toward carbohydrate metabolism. Micro‐computed tomography (μCT) analysis indicated a sex difference in the bone outcome, where male mice treated with morphine had reduced trabecular bone volume fraction (Tb.BV/TV) (15%) and trabecular bone mineral density (BMD) (14%) in the distal femur compared with vehicle. Conversely, bone microarchitecture was not changed in females after morphine treatment. Histomorphometric analysis demonstrated that in males, morphine reduced bone formation rate compared with vehicle, but osteoclast parameters were not different. Furthermore, morphine reduced bone formation marker gene expression in the tibia of males (Bglap and Dmp1). Circulating miRNA profile changes were evident in males, with 14 differentially expressed miRNAs associated with morphine treatment compared with two differentially expressed miRNAs in females. In males, target analysis indicated hypoxia‐inducible factor (HIF) signaling pathway was targeted by miR‐223‐3p and fatty acid metabolism by miR‐484, ‐223‐3p, and ‐328‐3p. Consequently, expression of miR‐223‐3p targets, including Igf1r and Stat3, was lower in morphine‐treated bone. In summary, we have established a model where morphine leads to a lower trabecular bone formation in males and identified potential mediating miRNAs. Understanding the sex‐specific mechanisms of bone loss from opioids will be important for improving management of the adverse effects of opioids on the skeleton. © 2022 American Society for Bone and Mineral Research (ASBMR).


Introduction
O pioid use disorder has become a critical US public health concern with a staggering number of overdose deaths across the country. In addition to risk of overdose and death, there are endocrine side effects related to opioid use, and among them is increased fracture risk. (1) Recently, Emeny and colleagues have estimated that opioid prescription was associated with three-to fourfold increase in fracture risk in a random sample of Medicare patients. (2) Among preclinical studies, there is an observed sex difference in the effect of opioids in bone, with male animals being negatively affected. (3,4) However, it should be noted that these animal models mimic diverse clinical conditions (ovariectomy versus cancer) and have not examined mechanisms of bone loss in otherwise healthy animal models.
Currently, only in vitro studies suggest opioid-induced bone changes may be due to an impairment of bone formation activity, and comprehensive in vivo analysis have not been performed. There is evidence that μ-opioid receptor (MOR) is expressed by human osteoblast-like cell line MG-63 (5) and by human bone marrow-derived mesenchymal stem cells. (6) In vitro studies suggest that opioids could directly act on bone formation by either modulating mesenchymal stem cell (MSC) fate (6) or reducing mature osteoblast activity and osteocalcin synthesis. (5) Despite the indices that opioids are detrimental to bone, there are limited alternative therapies for pain caused by bone fracture. (7) Therefore, there is still a need to better understand the mechanisms behind opioid-induced bone alterations.
Previous studies indicate morphine tolerance-associated chronic opioid use causes dysregulation of the central and peripheral expression of microRNAs (miRNAs), small noncoding functional RNAs that modulate gene expression and various biological processes. (8) In the clinical settings, Toyama and colleagues reported that patients using hydromorphone or oxycodone exhibited an upregulation of circulating miRNAs such as let-7 family and miR-339-3p, which were associated with the suppression of MOR activity. (9) Moreover, morphine tolerance altered the expression of miR-93 in a bone cancer pain mouse model and affected the downstream target Smad5, (10) which is important for bone homeostasis. (11) However, the role of these opioid-induced miRNA changes on bone remodeling has not been systematically investigated. Therefore, our aim was to develop a mouse model to evaluate the impact of sustained morphine exposure on bone turnover of both male and female mice and to identify candidate miRNA-mediated regulatory mechanisms that could affect bone. We initially hypothesized that chronic morphine exposure would uncouple bone turnover by reducing osteoblast and increasing osteoclast function. However, we observed no effect of morphine on osteoclasts. Briefly, we identified a sex difference in the effects of morphine on bone outcomes and circulating miRNAs. Trabecular bone loss occurred in males treated with morphine, as a consequence of impaired osteoblast function, but no changes in bone microarchitecture were observed in females. Among the enriched KEGG pathways identified, HIF signaling and fatty acid metabolism pathways were predicted to be affected by the set of miRNAs upregulated by morphine treatment in males. Our findings provide novel insight into the morphine-induced disruption of various metabolic parameters in both male and female mice, but the bone phenotype was observed exclusively in morphine-treated male mice.

Mice
Male and female C57BL/6J mice (stock #000664) were purchased from the Jackson Laboratory (Bar Harbor, ME, USA) at 6 weeks of age. Mice were placed in a barrier animal facility at MaineHealth Institute for Research (MHIR) on 14-hour light and 10-hour dark cycle at 22 C (standard room temperature). Mice were housed in groups of 3 or 4 per cage and they were given water and regular chow (Teklad global 18% protein diet, #2918, Envigo, Indianapolis, IN, USA) ad libitum. All mice acclimated to the MHIR animal facility for 2 weeks before the beginning of the study (day 0). All animal protocols in this study were approved by the Institutional Animal Care and Use Committee (IACUC) of MHIR, an Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC)-accredited facility.

Morphine delivery
Morphine sulfate salt pentahydrate powder was purchased from Sigma-Aldrich (St. Louis, MO, USA; M8777). Weight of morphine powder (in grams) used in each experiment as well as the volume of morphine solution (in μM) was logged and discarded in a locked pharmaceutical waste container. We used Alzet osmotic minipumps (Durect, Cupertino, CA, USA; model 2004: delivery rate = 0.25 μL/hr and total capacity 220 μL) to mimic a chronic exposure of morphine. The osmotic minipumps were filled under sterile conditions 40 hours before the implantation with either sterile vehicle solution (0.9% saline) or morphine solution. Morphine was administered at a dose of 17 mg/kg, based on average body weight, which was previously shown to cause bone loss but limit sedation. (3) All osmotic minipumps were weighed before and after being filled to ensure the entire pump was filled and air bubbles were not present, then placed individually into sterile 0.9% saline solution at 37 C until the implantation day.
At day 0 (baseline, 8 weeks of age), mice were randomly assigned to groups and osmotic minipumps were implanted subcutaneously in all mice. Surgeries were performed in cohorts such that male and female mice were started a week apart. The entire procedure was performed in a sterile surgical field with sterile tools that were cleaned between mice. Mice were anesthetized with 2% to 3% isoflurane and kept on 2% oxygen during the whole procedure. After shaving and sterilizing with betadine, a cutaneous dorsal incision was made perpendicular to the spine, approximately 0.5 inches caudal to the base of the neck. A hemostat was used to clear space for the osmotic minipump caudal to the incision, such that the inserted minipump was located closer to the hind quarters and would not interfere with healing of the incision. The incision was closed using wound clips, and all mice received 1 mg/kg sc meloxicam (Patterson Veterinary, Loveland, CO, USA) b.i.d. for 1 day after the surgery. Mice were weighed before and after the osmotic minipump implantation, and the latter was considered the baseline weight measure. Mice were examined for signs of pain or distress twice a day during 4 consecutive days after the surgery and the wound was examined for signs of inflammation. We euthanized one male from the vehicle group before the endpoint of the study because it exhibited impaired wound healing and signs of distress. We finished this 25-day experiment with a total of 12 male mice in the vehicle group, 14 male mice in the morphine group, 11 female mice in the vehicle group, and 11 female mice in the morphine group. Euthanasia was performed after 25 days of morphine treatment (12 weeks of age) using isoflurane anesthesia followed by decapitation (except where noted below).

Metabolic cage system
A total of 8 male and female mice from each treatment group (vehicle and morphine groups) were placed individually into metabolic cages for 5 days twice during the entire experiment (total of 10 days: 5 days on week 2 and 5 days on week 4) to assess metabolic and behavioral changes using the Promethion metabolic cage system (Sable Systems International, North Las Vegas, NV, USA), located in the Physiology Core at MMCRI. We are limited to 16 cages, therefore the number for metabolic cage data is only 8 per group (male and female mice were tested during separate weeks). Data acquisition and instrument control were performed using Meta Screen version 1.7.2.3, and the raw data obtained were processed with ExpeData version 1.5.4 (Sable Systems International) using an analysis script detailing all aspects of data transformation. (12) Summary 24-hour metabolic and behavioral assessment are presented. On occasion, technical problems occur with specific components of the cages (ie, water bottle weight measurements absent but activity data normal). Inaccurate data were excluded when such problems were identified, which resulted in a smaller number for some measurements.
Circulating morphine and morphine metabolite measurement Serum concentrations of morphine and morphine metabolites, morphine-3-glucuronide (M-3-G) and morphine-6-glucuronide (M-6-G), were determined by liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis, based on the method of Clavijo and colleagues. (13) Morphine, M-3-G, and M-6-G were extracted from serum via protein precipitation with acetonitrile. Separation was accomplished using a Phenomenex Synergi Hydro-RP analytical column (2.0 Â 150 mm, 4 μm). Mobile phase consisted of 0.1% formic acid in purified water (A) and 0.1% formic acid in acetonitrile (B). The flow rate was 0.5 mL/min and heated to 30 C. Gradient elution was employed, with initial conditions 97% A and 3% B. Solvent composition was held at the initial conditions for 1.5 minutes and then was ramped over the next 2.0 minutes to 25% B. Morphine and its metabolites were detected via an Agilent (Waldbronn, Germany) 6460 triple quadrupole mass spectrometer operated in positive ion MRM mode. The following transitions were monitored: morphine (286.1 ! 152.0) M-3-G and M-6-G (462.5 ! 286.1). (13) Circulating morphine and morphine metabolites levels were determined in an aliquot of serum after sustained treatment of 25 days (first experiment) and 12 days (second experiment). Dual-energy X-ray absorptiometry (DXA) All mice were weighed before DXA measurement. Areal bone mineral content (aBMC), areal bone mineral density (aBMD), lean mass, and fat mass measurements were performed on each mouse at baseline by a PIXImus dual-energy X-ray densitometer (GE Lunar, GE Healthcare, Madison, WI, USA). The PIXImus was calibrated daily with a mouse phantom provided by the manufacturer. Mice were placed ventral side down with each limb and tail positioned away from the body. Full-body scans were obtained, and the head was excluded from analysis because of concentrated mineral content in the skull and teeth. X-ray absorptiometry data were processed and analyzed with Lunar PIXImus 2 (version 2.1) software. (12) DXA measurements were repeated on mice that underwent metabolic cage testing at weeks 2 and 4.

Micro-computed tomography (μCT)
A high-resolution desktop micro-tomographic imaging system (μCT40, Scanco Medical AG, Brüttisellen, Switzerland) was used to assess trabecular bone architecture in the distal femoral metaphysis and L 5 vertebral body and cortical bone morphology of the femoral mid-diaphysis of male and female, vehicle and morphine-treated mice, ex vivo, after 25 days of treatment. Final femoral sample numbers include, in males, n = 12 vehicle and n = 14 morphine; and in females, n = 11 vehicle and n = 11 morphine. Final L 5 vertebral sample numbers include, in males, n = 10 vehicle and n = 14 morphine; and in females, n = 11 vehicle and n = 10 morphine. L 5 samples that were damaged during bone collection were excluded from the analysis (two male vehicle samples and one female morphine sample). Scans were acquired using a 10 μm 3 isotropic voxel size, 70 kVP, 114 μA, 200 ms integration time, and were subjected to Gaussian filtration and segmentation. Image acquisition and analysis protocols adhered to guidelines for the assessment of rodent bones by μCT. (14) In the femur, trabecular bone microarchitecture was evaluated in a 1500 μm (150 transverse slices) region beginning 200 μm superior to the peak of the growth plate and extending proximally. In the L 5 vertebral body, trabecular bone was evaluated in a region beginning 100 μm inferior to the cranial endplate and extending to 100 μm superior to the caudal endplate. The trabecular bone regions were identified by manually contouring the endocortical region of the bone. Thresholds of 335 mgHA/cm 3 and 385 mgHA/cm 3 were used to segment bone from soft tissue in the femur and L 5 vertebrae, respectively. The following architectural parameters were measured using the Scanco Trabecular Bone Morphometry evaluation script: trabecular bone volume fraction (Tb.BV/TV, %), trabecular bone mineral density (Tb.BMD, mgHA/cm 3 ), specific bone surface (BS/BV, mm 2 /mm 3 ), trabecular thickness (Tb.Th, mm), trabecular number (Tb.N, mm 1 ), trabecular separation (Tb.Sp, mm), and connectivity density (Conn.D, 1/mm 3 ). Cortical bone was assessed in 50 transverse μCT slices (500 μm long region) at the femoral mid-diaphysis, and the region included the entire outermost edge of the cortex. Cortical bone was segmented using a fixed threshold of 708 mgHA/cm 3 . The following variables were computed: total cross-sectional area (bone + medullary area) (Tt.Ar, mm 2 ), cortical bone area (Ct.Ar, mm 2 ), medullary area (Ma.Ar, mm 2 ), bone area fraction (Ct.Ar/Tt.Ar, %), cortical tissue mineral density (Ct. TMD, mgHA/cm 3 ), and cortical thickness (Ct.Th, mm). Cortical bone images were taken of the mouse using the median total area value within each group.

Histomorphometric bone analysis
Bone histomorphometric analysis was performed on the left femur of male treated with vehicle solution and with morphine solution for 25 days. Calcein solution (20 mg/kg; Sigma) and Alizarin solution (40 mg/kg) were injected at 8 days and 2 days before animal euthanasia, respectively. The femur was dissected and formalin-fixed (10%) for 48 hours before transferred to 70% ethanol solution. Fixed, nondecalcified samples were dehydrated using graded ethanol solutions and subsequently infiltrated and embedded in methylmethacrylate. Longitudinal sections (5 μM) were cut using a microtome (RM2255, Leica, Wetzlar, Germany) and stained with Goldner's Trichrome for measurements of bone microarchitecture and cellular parameters. Dynamic bone parameters were evaluated on unstained sections by measuring the extent and the distance between double labels using the Osteomeasure analyzing system (OsteoMetrics Inc., Decatur, GA, USA). Measurements were made in the same position for each sample at 3600 μm 2 area in the femur (200-250 μm below growth plate). Quantification of bone parameters was done in a blinded manner. The structural, dynamic, and cellular parameters were evaluated using standardized guidelines. (15)

Real-time PCR
Tibias were collected from 12-week-old male and female mice from vehicle and morphine groups for RNA extraction under liquid nitrogen conditions (n = 7-14). For the opioid receptor expression analysis, we used a set of nontreated 8-week-old C57BL6/J male and female mice. Total RNA was prepared using the standard TRIzol (Sigma-Aldrich) method for tissues. cDNA was generated using the High-Capacity cDNA Reverse Transcriptase Kit (Applied Biosystems, Carlsbad, CA, USA) according to the manufacturer's instructions. mRNA expression analysis was carried out in duplicate using an iQ SYBR Green Supermix with a Bio-Rad Laboratories (Hercules, CA, USA) CFX Connect Real-time System thermal cycler and detection system. TATA binding protein 1 (Tbp1) was used as an internal standard control gene for all quantification. (16) Primers used were from Integrated DNA Technologies (IDT, Coralville, IA, USA), (17)(18)(19) Primer Design (Southampton, UK), or Qiagen (Germantown, MD, USA). All primer sequences (if provided) are listed in Table S1. Total RNA integrity of the samples from the 25-day treatment experiment was evaluated by running a denaturing 1% agarose gel stained with ethidium bromide (EtBr). The number of samples included in the statistical analysis varied from 7 to 14 per group, depending on the quality of the RNA. For the opioid receptor expression analysis, the number of samples included varied from 6 to 8.

Serum turnover markers
Serum P1NP and CTx-1 were measured as previously described and according to the manufacturer instructions. (20) microRNA array analysis-12-day experiment A second cohort of male and female mice were implanted subcutaneously with osmotic minipumps filled with vehicle or morphine solution as described earlier. DXA analyses also were performed during this experiment. Mice were euthanized by day 12 using isoflurane. For each animal, whole blood was collected at death in a 1.5 mL tube using cardiac puncture, allowed to clot for 30 minutes at room temperature, and then placed on ice. All blood samples were centrifuged for 10 minutes at 500 relative centrifugal force (RCF). At least 200 μL serum was removed from the top translucent phase and stored at À80 C (n = 3-5 male and n = 3-4 female). All mice were 10 weeks of age at the endpoint. Total RNA was isolated from 200 μL mouse serum following a standardized protocol (21) using the miRNeasy Serum Plasma kit with ce-miR-39 spike-in (Qiagen), QIAcube (Qiagen) automation, and eluted with 14 μL of nuclease-free water.
Global circulating miRNA screen Exactly 8 μL of isolated RNA was prepared for Affymetrix Gene-Chip miRNA v4 microarrays (Thermo Fisher Scientific, Waltham, MA, USA), allowed to hybridize for 42 hours, and data processed as described. (22) The R script for this normalization is included as a supplement. Differential expression significance was assessed by one-way ANOVA. The data have been deposited into the GEO repository (GSE197198).

Statistical analysis
GraphPad (La Jolla, CA, USA) Prism 9 XML Project software was used to perform statistical tests. Data are presented as mean AE standard deviation (SD). Outliers were defined as data points >3 standard deviations from the mean and were excluded from analysis. Student's t test or two-way ANOVA was performed and Holm-Sidak post hoc multiple comparison test was performed where appropriate (after a significant interaction effect). α ≤ 0.05 was considered statistically significant. Heat-map values and principal component analysis were generated in Spotfire v2 (Tibco, Palo Alto, CA, USA). All data were imported into Adobe Illustrator CC for final figure creation. Volcano plot analysis was performed to identify miRNA with large-fold changes (threshold of ≥2-fold change) that were also statistically significant (p < 0.01). Diana miRPath v3.0 software was used to search for predicted affected experimentally validated (Tarbase V.7) and nonexperimentally validated (Target Scan) enriched KEGG pathways (23) at p values <0.05, with the following settings: pathways union, false discovery rate (FDR) correction box checked, and conservative stats box unchecked. For this, we evaluated each miRNA separately to identify the potential KEGG pathways affected by each one of them.

Male mice exhibited reduced femoral trabecular bone after sustained morphine delivery
We first wanted to investigate whether sustained morphine delivery caused changes in bone. Areal bone mineral density was evaluated at three time points during the study (baseline, day 7, and day 21). After 7 days of the osmotic minipump implantation, male and female mice exhibited a drop in aBMD, regardless of treatment groups. However, sustained morphine delivery did not worsen this initial aBMD loss, which may have been caused by the surgical procedure (Fig. S1C).
Morphine treatment affected femoral trabecular bone in male mice, which exhibited reduced Tb.BV/TV (15%, p = 0.035) and Tb.BMD (14%, p = 0.015) compared with the vehicle group ( Fig. 1A-C). We observed no changes related to morphine treatment in any other bone parameters evaluated, such as Tb.BS/BV, Conn.D, Tb.N, Tb.Th, and Tb.Sp (p > 0.05) ( Fig. 1D-F). Unlike males, however, sustained morphine delivery did not alter any of the trabecular bone parameters evaluated in female mice, which were  similar between treatment groups (p > 0.05) ( Fig. 1A-H). Because of these differences, circulating levels of morphine and metabolites were measured to ensure that the drug solution was still being delivered by the osmotic minipumps after 25 days. We detected morphine and morphine-3-glucuronide (M-3-G) in the serum of morphine-treated male and female mice at the endpoint (p < 0.0001), whereas morphine-6-glucuronide (M-6-G) levels were undetectable (Table S2). We also found that morphine levels were similar between sexes, whereas females have a higher circulating level of M-3-G compared with males ( Fig. S1A, B).
On the other hand, unlike the effects found in the trabecular compartment, our μCT findings revealed that morphine treatment had no effect on cortical bone (Fig. 2), which is consistent with aBMD data that usually reflects alterations in cortical microarchitecture. We observed that there were no changes in Tt.Ar, Ma.Ar, and Ct.Ar after morphine exposure in either in males or females (p > 0.05) ( Fig. 2B-D). Both morphinetreated male and female mice also exhibited similar Ct.Ar/Tt. Ar, Ct.Th, and Ct.TMD compared with vehicle groups (p > 0.05) ( Fig. 2E-G).
In contrast to femoral trabecular bone results, morphine treatment had no significant impact on L 5 vertebral body microarchitectural parameters (Fig. 3A-H), although vertebral BV/TV and BMD tended to be lower in male mice treated with morphine compared with vehicle.

Morphine caused reduced osteoblast mineralization activity
We next performed histomorphometric analysis in the femur to better understand morphine-induced changes in bone turnover in male mice. This analysis also confirmed that chronic morphine was associated with decreased femoral BV/TV and Tb.Th in males (p < 0.0001) ( Table 1). Morphine-treated male mice also exhibited reduced bone area (B.Ar) compared with vehicle group (p < 0.0001). Unexpectedly, we found that the number of osteoblasts (Ob.S/BS) and the number of osteoclasts (Oc.S/BS) per bone surface were similar between vehicle and morphine groups after 25 days of treatment (p > 0.05). However, morphine exposure significantly decreased BFR/BS (p = 0.033) ( Table 1). Moreover, we also observed a trend toward reduced MAR (p = 0.086), MS/BS (p = 0.055), and decreased osteoclast activity (eroded surface, ES/BS) (p = 0.077) in male mice. However, these latter parameters did not reach statistical significance (Table 1). In addition, we found reduced N.Ot but not N.Ot corrected by bone area (Table 1).
We also evaluated the expression of osteoblast/osteocytes and osteoclast markers in the whole tibia from male and female mice (Fig. 4A, B). As expected, the expression data corroborated our histomorphometric results. We found a reduced expression of Bglap, Dmp1, and Fgf23 in the whole tibia of morphine-treated male compared with vehicle-treated male mice (p < 0.001), suggesting that morphine has a major impact in the function of mature osteoblast lineage cells (Fig. 4A). Our findings also demonstrated a reduction in the expression of Ctsk (p < 0.05) (Fig. 4A). On the other hand, we found that sustained morphine treatment had no effect on Rankl/Opg system. Tnfrsf11b (Opg) expression, Tnfs11 (Rankl) expression, and Rankl/Opg ratio were similar between treatment groups (p > 0.05) (Fig. 4A). Despite the absence of altered bone microarchitecture in females, we also examined gene marker expression. We found significant decreases in Runx2 and Dmp1 in females treated with morphine (Fig. 4B). Serum P1NP was also significantly suppressed in females but not males (Fig. 4C, D), whereas CTx-1 only tended to be lower in male mice (Fig. 4C). Collectively, these findings suggest altered bone formation is the predominant mediator of bone loss in mice treated with morphine and that female mice may not be completely protected.
We next wanted to confirm the magnitude of the opioid receptors (MOR, DOR, and KOR) expression in whole tibia of male and female mice because others have shown MOR is expressed by osteoblast-like cells. (5) However, expression of the opioid receptors was exceptionally low to absent, and they were significantly lower compared with brain expression of these receptors (p < 0.0001) (Fig. Fig. S2), which suggests that morphine likely has an indirect effect on the bone.

Morphine treatment outcomes in body weight and composition
Differences in body weight and composition between male and female were observed, as expected (Table S3). At the baseline, treatment groups started with similar body weight and body composition evaluated by two-dimensional DXA analysis. Conversely, after 21 days of sustained morphine exposure, we found a significant reduction in the % of fat mass and adiposity index (Table S3). However, we observed no sex by treatment interaction effects, suggesting that morphine influenced body weight and composition similarly in both sexes.

Morphine treatment had no major effect on motor activity
We performed a 2-way ANOVA analysis to better understand the effects of sustained morphine exposure in metabolic and motor activity over time (from weeks 2 to 4). Overall, we found that sustained morphine exposure had no major impact in motor activity in morphine-treated mice. In males, we observed that sustained morphine treatment led to reduced X beam breaks within the cage (p < 0.05) (Table S4), but this was not enough to influence the distance walked or time spent walking in the metabolic cages. Furthermore, morphine did not influence time or speed on running wheels in either sex of mice. Over time, there was an increase in Z beam breaks movements and wheel speed (p < 0.05), with no differences between treatment groups, which was likely related to surgery recovery. In females, morphine caused a reduction in the percentage of time spent staying still compared with the vehicle group (p < 0.05) (Table S4). From weeks 2 to 4, in general, females became less active, which was observed by a reduction in X and Y beam movements, cage walking meters, and percentage of time spent walking and an increase in the percentage of time sleeping (p < 0.05). However, this behavior was similar between groups with no statistical difference (Table S4). Moreover, there was no treatment by time point interactions effect for any of the motor activity parameters assessed (p > 0.05) (Table S4). Our findings suggest that sustained morphine exposure had only a minimal impact on motor activity.
Sustained morphine treatment led to changes in energy expenditure and fuel utilization In males, sustained morphine exposure increased energy expenditure (EE), CO 2 expelled, respiratory quotient (RQ), resting respiratory quotient (RRQ), and active respiratory quotient (ARQ)    (p < 0.05) (Table S5). Resting energy expenditure (REE) was also increased in morphine-treated male mice compared with vehicle group (p = 0.0001), although there was a reduction of REE values over time (from weeks 2 to 4) (p = 0.002) (Table S5). However, we found no treatment by time point interactions effect for any of the metabolic parameters assessed (Table S5). In addition, while an increase in water consumption over time was observed, food intake of male mice was not significantly affected by sustained morphine treatment, time of morphine exposure, or interaction effect between variables (Table S5). In females, we also observed an increase in EE, CO 2 expelled, O 2 consumed, REE, active energy expenditure (AEE), and ARQ related to chronic morphine treatment. We found that CO 2 expelled and AEE also increased over time (from weeks 2 to 4), but there were no treatment by time point interaction effects for these parameters (Table S5). Because we found a significant treatment by time point interaction for RQ data in females, we performed Holm-Sidak post hoc multiple comparison test, which demonstrated that RQ was higher in morphine-treated female mice compared with vehicle mice at both time points (p < 0.001), but from weeks 2 to 4, there was no significant changes in RQ related to morphine treatment (p = 0.074) (Table S5). Together, these findings seem to indicate that a greater length of continuous morphine exposure (4 weeks) had no further effect on energy expenditure and fuel utilization, and changes in metabolism appear to begin at an early stage of morphine treatment (2 weeks).

H G F
Sustained morphine exposure induced sex-difference changes in circulating miRNA profile To further understand the potential molecular mechanisms that might be involved in the sustained morphine-induced bone phenotype, serum was collected from male and female C57BL/6J mice treated with morphine for 12 days to perform miRNA array analysis (Fig. 5). Circulating levels of morphine and its metabolites (M-3-G and M-6-G) were also measured to confirm that the osmotic minipumps were delivering the drug (Table S2). At this short-term experiment, after 12 days, females exhibited a trend toward a higher concentration of morphine (p = 0.051) and M-3-G levels (p = 0.066) compared with males (Fig. S1E). Interestingly, we found a sex difference in the miRNA profile associated with morphine exposure. Of 1908 mature miR-NAs evaluated, morphine induced changes in 260 miRNAs. Principal component analysis (PCA) demonstrated that morphine treatment seems to induce males to have an miRNA profile close to female mice (Fig. S1F). The heat map illustrates differences in miRNA expression between vehicle-and morphine-treated male mice after 12 days of morphine treatment, which overall had a suppression of miRNA from morphine (Fig. 5A). We also identified that in females only 2 miRNAs were significantly and differentially expressed miRNAs, whereas in males there were 14 differentially expressed miRNAs with morphine that reached a threshold of ≥2-fold change and p < 0.01 (Fig. 5B).
Fatty acid metabolism and HIF signaling pathways are targeted by sustained morphine treatment Of those 14 differentially expressed miRNAs in morphine-treated male mice, 4 miRNAs were upregulated and 10 miRNAs were downregulated (p < 0.05) ( Table 2). {TBL 2} Many of these miR-NAs have a previously known connection to bone (summarized in Table 2).
In female mice, we found the expression of miR-1982-5p and -3090-5p were upregulated after morphine exposure. There were no related experimentally validated enriched pathways found (Tarbase V.7). However, when searching for the non-experimentally validated enriched pathways (Target Scan), we found that miR-3090-5p was mostly associated with fatty acid metabolism (Table S6).
To determine if the miRNA changes could be responsible for bone effects in male mice, we tested if any of the downregulated genes (Fig. 4A) were targets of the upregulated miRNAs ( Table 2) and found that none were experimentally validated or predicted targets of the upregulated genes from male mice. We next examined expression of a subset of targets of the upregulated miRNAs from male mice that comprised the enriched HIF1α and fatty acid metabolism pathways. Some targets of miR-223-3p, including insulin-like growth factor I receptor (Igf1r), eukaryotic translation initiation factor 4E family member 2 (Eif4e2), and signal transducer and activator of transcription (Stat3), were significantly suppressed by morphine treatment (Fig. 6). Similarly, expression of the miR-484 target Oxsm, which encodes the protein 3-oxoacyl-ACP synthase, was also significantly lower in the morphinetreated mouse bone (Fig. 6). Further studies will be required to determine if these and/or other miRNA/target gene changes mediate morphine effects on bone.

Discussion
Sustained opioid use has previously been associated with deleterious effects on the skeleton (3,40) and a higher fracture risk. (2) In this study, our aim was to develop a mouse model of opioid-induced bone loss to study the impact of chronic morphine exposure on bone turnover and to identify potential miRNA-mediated regulatory mechanisms that contribute to the effects of morphine on bone tissue. To our knowledge, we are the first to evaluate the effects of sustained morphine treatment on the skeleton, body composition, metabolic and motor activity, and circulating miRNA in male and female C57BL/6J mice. Our initial hypothesis was that chronic morphine exposure would lead to bone loss by uncoupling bone turnover through suppression of bone formation and increasing bone resorption. However, the resulting data did not indicate changes in osteoclast activity during morphine treatment. Rather, the present study found that sustained morphine exposure for 25 days produced a reduction in osteoblast activity and decreased expression of bone-forming genes, which led to a lower trabecular bone in male mice. Moreover, morphine treatment modulated the expression of miRNAs, more so in males than in females, and some of the targets of elevated circulating miRNAs were significantly downregulated in bone.
Sex differences in bone outcomes between males and females are noticeable among studies investigating the effects of opioid on bone in humans. (40,41) Opioid-induced bone loss has been consistently reported in men, (40,42,43) whereas the   effects of opioids on BMD in women are less clear. (41,44) Some studies show that long-term opioid treatment (>5 years) has no significant effect on BMD changes over time in women. (41,44) However, opioid treatment appears to be more harmful to bone compared with other analgesics evaluated (acetaminophen and nonsteroidal anti-inflammatory drugs) (44) and remains associated with increased fracture risk in women. (2,45) Despite the similar circulating levels of morphine, in males and females, our findings revealed a clear lack of morphine effects on the skeleton of females compared with males, which corroborates with the existing data. (3,4) An apparent explanation for this is that opioid use is described to cause reduction in circulating testosterone levels. (40) However, bone phenotype observed in morphinetreated male mice is not as dramatic as that of orchiectomized animal model. The reduced testosterone levels obtained in an orchiectomized model leads not only to a dramatic decrease in trabecular bone volume but also impaired cortical thickness. (46) Additionally, bone loss observed in the orchiectomized animal model is associated with an increased osteoclast activity. (46) Although microarchitectural parameters were unchanged in females, we did find reduced gene expression of bone formation markers and reduced serum P1NP, suggesting a longer treatment, higher dose, or differential treatment context (ie, with cancer) of morphine may alter bone parameters.
We found that osteoblast cell number was not affected by chronic morphine exposure, but morphine had a deleterious effect on bone formation rate, also appearing to have a trend toward a reduced mineralized surface and mineral apposition rate in the femur region. These data were supported by reduced gene expression of osteoblast/osteocyte markers (Bglap, Dmp1, Fgf23) in males exposed to the drug. This is consistent with the very limited evidence demonstrating that opioids have a direct impact on osteoblasts. (5) Ultimately, whether direct effects are important in our findings are unclear, since opioid receptor expression in bone is very low to absent. In contrast, we did not observe any effects of opioids on bone resorption. We found no changes in the numbers of osteoclasts with chronic morphine exposure, different from what we were expecting. In fact, our data suggest that chronic morphine treatment may have a suppressive effect on bone resorption function with reduced expression of Ctsk in the whole tibia. Alonso-Pérez and colleagues (47) described that morphine can signal through a toll-like receptor 4 (TLR4)/myeloid differentiation protein 2 (MD-2) complex, which is expressed by osteoblasts, and indirectly modulates osteoclast functionality. (47) However, our data have not demonstrated a clear change in osteoclast activity. Alternately, the source of Ctsk could be osteocytes or periosteal stem cells and reduction in Ctsk in morphine-treated mice could be a reflection of the reduction in other osteocyte markers. (48,49) Future studies examining osteoblast maturity and osteocyte morphology in vivo, and whether circulating FGF-23 and phosphate metabolism are altered with morphine treatment, may be warranted given the decreases in late osteoblast and osteocyte markers.
Our focus on the miRNAs as one of the potential mechanisms contributing to morphine effects in the skeleton also revealed a sex difference in miRNA profile related to morphine  Journal of Bone and Mineral Research treatment, with more prominent morphine-induced changes in miRNA profile in males than in females. Notably, in contrast to previous data, we have not found the expression of any miRNA (eg, let-7) known to be associated with MOR activity. (8) The experimentally validated enriched KEGG pathways, targeted by the miRNAs, were not directly associated with morphine or opioid-related pathways. However, miR-484 and -328-3p were predicted to affect non-experimentally validated enriched KEGG pathways related to morphine addiction. Moreover, the expression of these two miRNAs (miR-328-3p and -484) is also associated with BMD elsewhere. (24) Namely, Gautvik and colleagues (24) found a complex relationship between these miR-NAs and BMD, which exhibited different effect depending on the skeletal site. Wang and colleagues (50) described that miR-484 is involved in mitochondrial fission in cardiomyocytes and adrenocortical cancer cells, which is controlled by FOXO3A and FIS1, demonstrating that miR-484 has a role in controlling bioenergetic homeostasis. Directly related to bone mineral density (BMD) in femur but negatively associated with that of iliac in postmenopausal women with normal to osteoporotic bone.
-miR-182-5p À3.670 0.002 When overexpressed in human bone marrow mesenchymal stem cells, miR-182-5p inhibits chondrogenesis, decreases expression levels of SOX9 and COL2A1, and increases expression levels of COL1A1 and COL10A1. miR-182-5p upregulation was found in patients with fibrous dysplasia, demonstrating a potential involvement on the dysregulation of gene expression in bone. On the other hand, downregulation of miR-182-5p seems to promote osteoblast proliferation and differentiation in osteoporotic rats through Rap1/ MAPK signaling pathway activation by upregulating ADCY6.
Kelch et al. (37) and Chen et al. (38) À2.263 0.005 None known. - In regard to potential bone-related pathways, miR-223-3p was found to affect HIF-1 signaling pathway, which is described to regulate osteocyte-mediated osteoclastic differentiation by promoting RANKL expression through the activation of JAK2/STAT3 pathway. (39) Indeed, others have reported that miR-223 can modulate the expression of other bone markers such as Ctsk , (51) Runx2, Bglap, Alpl, and Spp1. (52) In addition, this miRNA is involved in the suppression of cell proliferation by targeting Igf1r, which is related to skeletal response to mechanical loading. (53) These findings, in addition to our finding of reduced expression of Stat3 and Igf1r in bone, suggest that miR-223 may be involved in morphine-induced bone loss in males. Previous evidence also demonstrated a link between miR-223-3p and adipose tissue. Macartney-Coxson and colleagues observed downregulation of miR-223-3p expression in the omentum and subcutaneous adipose tissue after weight loss induced by a gastric bypass. (54) Whether miR-223-3p contributed to morphine-induced metabolic phenotypes in our mice is unknown. We found fatty acid metabolism pathways in morphine-treated male mice was targeted by the upregulated miRNAs. In fact, miRNA data indicated that sustained morphine treatment affected fatty acid metabolism pathways in both sexes, which is consistent with the changes in energy expenditure and respiratory quotient during metabolic assessment. However, our data also suggest that the morphineinduced changes in fat metabolism may occur through distinct mechanisms between sexes.
It is recognized that metabolic disturbances are associated with disruption of bone anabolic pathways (eg, Wnt signaling, parathyroid hormone signaling, insulin, and peroxisome proliferator-activated receptor γ), resulting in impairment of osteoblast function and uncoupling of bone turnover. (55) Skeletal cell fate is regulated by the nutritional environment, and lipids are an essential energy source to bone cells. (56) Osteoblasts also require fatty acid oxidation for normal bone acquisition. (57) Van Gastel and colleagues (55) also demonstrated that lipid scarcity leads skeletal progenitor cell into chondrogenic over osteogenic lineage differentiation. Although chondrocytes are highly glycolytic, energy production in osteoblasts mainly relies on a higher rate of fatty acid oxidation, not glucose oxidation. (55) In face of this, we speculate that the changes in circulating miRNA profile and in substrate utilization observed in males after sustained morphine delivery might be contributing to the poor metabolic control of bone mineralization process causing reduction in bone mass (58) ; however, additional studies would be necessary to confirm this mechanism. Among the downregulated miRNAs identified, there is evidence connecting miR-351-5p, (32) -203-3p, (33,34) -200c-3p, (35,36) and -125b-5p (37) to osteogenesis (Table 2). Interestingly, we found that miR-125-5p has affected the greatest number of pathways in males exposed to morphine. There is evidence in the literature that miR-125-5p is associated with bone but in conditions different from our study. miR-125b-5p expression is described to be sex-dependent (37) and is downregulated in osteoporotic postmenopausal women compared with non-osteoporotic ones. (38) Moreover, miR-125b-5p (59) and miR-200b-3p (60) are miRNAs described to be involved in steroidogenesis. It is reported that miR-125b-5p expression is decreased in polycystic ovary syndrome (PCOS) women, (59) whereas miR-200b-3p targets steroidogenic pathway enzyme (eg, CYP19A1) that produces estradiol. (60) These findings suggest that miR-125b-5p and -200b-3p might unveil potential mechanisms of how morphine may impact steroid production. Here, we have not explored circulating steroid hormone levels and their receptors in our mouse model to understand to what extent the changes in this sex hormone's levels influenced our bone phenotype, and we believe that such aspect should be considered moving forward. On the other hand, we have identified systemic metabolic alterations associated with morphine treatment that may have affected bone homeostasis.
Finally, future work should examine the nervous system and other potential sources of circulating miRNA changes. Since the nervous system is the major target of opioids, it is possible that the overwhelming suppression miRNAs in males comes from altered secretion by neurons. Although efficacy of opioids for pain is linked to suppression of neural transmission, opioids may also stimulate (directly or indirectly) sympathetic nervous system pathways, which could also indirectly modulate miRNA or catecholamine release to influence bone and metabolism. In summary, we found that sustained morphine delivery leads to reduced osteoblast functionality and lower bone density in male mice, but bone microarchitecture was preserved in female mice after chronic morphine exposure. We also observe that morphine influences substrate utilization and fatty acid metabolism and that these are associated with the upregulation of miR-484, -223-3p, and -328-3p expression in males, which is a potential link between morphine and altered metabolic control of bone mineralization process. Our novel findings have set a precedence for future investigations into how morphine-induced metabolic changes influence bone formation could lead to clinical mitigation strategies for preventing the adverse effects of opioids on bone health.

Disclosures
All authors state that they have no conflicts of interest.

Acknowledgments
This work was supported by the Northern New England Clinical and Translational Research (NNE-CTR) Network Pilot Project Program under the National Institutes of Health (NIH)/National Institute of General Medical Sciences (NIGMS) award number U54GM115516, as well as National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS) and the NIGMS under award numbers K01AR067858, P20GM121301, and R01AR076349 to KJM, 5R01AR039588-29 to JBL, and T32GM132006. This work utilized services of the Maine Medical Center Research Institute (MMCRI) Molecular Phenotyping Core, which is supported by NIH/NIGMS P30GM106391, the Physiology Core, which is supported by NIH/NIGMS P30GM106391 and P20GM121301, and the Mouse Transgenic and In Vivo Imaging Core, which is supported by NIH/NIGMS P30GM103392. All cores also received support from the NNE-CTR Network NIH/NIGMS U54GM115516. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Authors' roles: ALC was responsible for experimental design, data acquisition, data analysis, interpretation, and drafting of the manuscript. DJB was responsible for data acquisition, data  analysis, interpretation, and drafting and critical revision of the manuscript. DB was responsible for data acquisition, interpretation, and drafting and critical revision of the manuscript. ALL was responsible for data acquisition, interpretation, and drafting and critical revision of the manuscript. BM was responsible for data acquisition, interpretation, and drafting and critical revision of the manuscript. KLH was responsible for data interpretation and critical revision of the manuscript. MLB was responsible for data interpretation and critical revision of the manuscript. JBL was responsible for data interpretation and critical revision of the manuscript. TK was responsible for data interpretation and critical revision of the manuscript. NHF was responsible for data acquisition, data analysis, data interpretation, and drafting and critical revision of the manuscript. KJM was responsible for experimental design, data acquisition, data analysis, interpretation, and drafting and critical revision of the manuscript. All authors had final approval of the manuscript.

Author Contributions
Daniel