Distinct Responses of Modeling- and Remodeling-Based Bone Formation to the Discontinuation of Intermittent Parathyroid Hormone Treatment in Ovariectomized Rats

Abstract

Anabolic agents, such as intermittent PTH, exert their treatment efficacy through activation of two distinct bone formation processes, namely remodeling-based bone formation (RBF, bone formation coupled with prior bone resorption) and modeling-based bone formation (MBF, bone formation without prior activation of bone resorption). However, if not followed by an antiresorptive agent, treatment benefit was quickly lost upon withdrawal from anabolic agents. By using in vivo micro computed tomography imaging and multiplex cryohistology with sequential immunofluorescence staining, we investigated the temporal response of newly formed bone tissue from MBF and RBF and the pre-existing bone tissue to withdrawal from PTH treatment and the associated cellular activity in an OVX rat model. We first demonstrated continued mineral apposition at both RBF and MBF sites after PTH discontinuation, resulting in an extended anabolic effect after 1-week withdrawal from PTH. It was further discovered that MBF sites had a greater contribution than RBF sites to the extended anabolic effect upon early withdrawal from PTH, evidenced by a higher percent of ALP+ surfaces and far greater bone formation activity at MBF vs. RBF sites. Furthermore, significant bone loss occurred after 3 weeks of discontinuation from PTH, resulting from marked loss of newly formed bone tissue from RBF and pre-existing bone tissue prior to treatment. In contrast, MBF surfaces had a delayed increase of TRAP activity following PTH discontinuation. As a result, newly formed bone tissue from MBF had greater resistance to PTH discontinuation-induced bone loss than those from RBF and pre-existing bone. Understanding various responses of two distinct bone formation types and pre-existing bone to anabolic treatment discontinuation is critical to inform the design of follow-up treatment or cyclic treatment strategies to maximize treatment benefit of anabolic agents.

1. Introduction

Postmenopausal osteoporosis is a life-long, chronic disease associated with imbalanced bone remodeling with osteoclast resorption outpacing osteoblast bone formation, leading to compromised bone strength and increased risk of fragility fractures. Currently, there are two major classes of therapeutic strategies to treat postmenopausal osteoporosis: antiresorptive agents, which suppress osteoclast-mediated bone resorption, and anabolic agents, which promote osteoblasts for new bone formation. Recent studies have demonstrated that anabolic agents lead to a more rapid and greater reduction of fracture risk amongst postmenopausal women with a very high risk of fracture [1, 2]. However, despite the potent effect of anabolic agents on stimulating new bone formation, the gains in bone mineral density (BMD) were quickly lost upon treatment withdrawal if not followed by an antiresorptive agent [3–6]. In a 2-year study with a 1-year treatment of parathyroid hormone (PTH) 1–84 followed by a 1-year treatment of alendronate (a form of bisphosphonate) or placebo in postmenopausal women, women who were randomly assigned to receive placebo following PTH treatment had substantial loss in bone density, especially at the trabecular bone compartment [3]. Similarly, in a 3-year study with a 2-year treatment of romosozumab (a humanized anti-sclerostin antibody) followed by a 1-year treatment of denosumab (a RANK ligand inhibitor) or placebo, BMD returned toward pretreatment levels in women who transitioned to 1-year placebo [6]. However, the mechanisms behind this rapid withdrawal effect were not fully understood. It is unclear whether the substantial bone loss following treatment discontinuation was mainly from the newly formed bone tissue in response to treatments or the pre-existing bone tissue prior to treatments.

It has recently been reported that anabolic agents exert their treatment efficacy through activation of two distinct bone formation processes [7–12], namely remodeling-based bone formation (RBF) and modeling-based bone formation (MBF) [13–15]. RBF, a bone formation that is coupled with prior osteoclast resorption, plays critical roles in maintaining bone density and healthy bone quality. In contrast, MBF, a de novo bone formation without prior activation of bone resorption, occurs less often in the adult skeleton [8, 9, 16–18]. Intermittent injections of PTH 1–34, teriparatide, was the first FDA-approved anabolic treatment for postmenopausal osteoporosis. Lindsay et al. reported that clinical treatment by PTH 1–34 rapidly exerts its anabolic effect in 1 month through the activation of RBF and MBF where MBF may constitute more than 30% of bone gain induced by PTH in postmenopausal women [8]. In our previous investigation, we established an adult rat tibia model for studying MBF and RBF by using a novel cryohistology imaging platform, coupled with sequential multi-color fluorochrome injections [19]. Unlike the adult human skeleton, which is dominated by RBF, we first demonstrated that MBF and RBF have similar contributions to trabecular bone homeostasis in the tibia of male, intact female, and ovariectomized (OVX) female rats, making rat an ideal model for studying the cellular mechanisms of MBF and RBF. Furthermore, we demonstrated robust activation of both MBF and RBF by PTH treatment in rats, irrespective of their sex and estrogen status. Moreover, by tracking the sequence of multi-color fluorochrome labels, we discovered that MBF is activated more rapidly but attenuated faster than RBF in response to PTH, suggesting different contributions from MBF and RBF to the anabolic effect of PTH [19]. We speculated that osteoblasts recruited to MBF and RBF sites may come from different osteoprogenitor pools, resulting in the distinct activation and attenuation of RBF and MBF. To date, there have been no data regarding the retention of bone tissue formed from MBF, RBF, or pre-exist prior to treatment in response to anabolic treatment withdrawal. Given the distinct cellular mechanisms involved with the activation of MBF and RBF, we further hypothesize that MBF, RBF, and pre-existing bone surfaces may respond differently at both the cellular and tissue level to PTH treatment discontinuation.

In our previous study, we have demonstrated significant trabecular bone loss and bone microarchitecture deterioration after discontinuation from PTH treatment in an OVX rat model [20]. Interestingly and unexpectedly, we discovered an extended anabolic period (EAP) upon 1-week withdrawal from 3-week PTH treatment, where continued mineral deposition occurred at more than 2/3 of active bone formation sites. Despite the EAP, rapid bone loss started at the 2nd week after treatment discontinuation, leading to significant bone loss and bone microarchitecture deterioration. In the current study, by using a similar OVX rat model and treatment strategy, we first quantified and compared the bone tissue dynamics between RBF, MBF, and pre-existing bone in response to the discontinuation of PTH treatment using a cryohistology and polarizing imaging platform that we previously established [19]. Subsequently, cellular activity at RBF, MBF, and pre-existing bone surfaces was investigated using multiple-round, sequential immunofluorescence staining on a single cryosection. We hypothesize that MBF, RBF, and pre-existing bone surfaces are associated with distinct cellular activity, leading to different extents of resistance to PTH withdrawal-induced bone loss.

2. Materials and Methods:

2.1. Animal experiment design and treatment plans

All experiments were approved by the University of Pennsylvania’s Institutional Animal Care and Use Committee. A total of 30 female Sprague Dawley rats were used in two sets of animal experiments. These rats were bred and raised in our colony at the University of Pennsylvania. All rats were fed a standard diet (LabDiet 5001 Rodent Diet) and housed in standard conditions of three rats per cage. All rats underwent bilateral OVX surgery at the age of 16 weeks followed by osteopenia development for 6 weeks before being used for experiments at the age of 22 weeks. Following OVX surgery, rats were separated to one rat per cage for one week for recovery and were otherwise housed in standard conditions with groups of three rats per cage throughout the study. For the in vivo micro computed tomography (μCT) imaging study, rats (n=12) were randomly allocated to treatment groups to minimize variation in baseline bone volume fraction between groups. For the histology study, rats (n=18) were grouped according to weight before being randomly assigned to treatments.

We first performed an in vivo μCT experiment to confirm the treatment dose (20 μg/kg/day) and treatment duration (3-week PTH treatment and 3-week withdrawal period) in OVX rats. Six rats received subcutaneous injections of PTH (Human Recombinant PTH1–34, Bachem, Bubendorf, Switzerland) at 20 μg/kg/day, 5 times a week, for 3 weeks. Similarly, age-matched, vehicle (VEH)-treated rats (n=6) received subcutaneous saline injections following the same injection schedule as rats in the PTH group. Weekly in vivo μCT scans were performed for both groups for 6 weeks (from the age of 22 to 28 weeks, 3-week PTH on and 3-week PTH off).

Next, a dynamic bone histomorphometry study was performed with 18 OVX rats following the same treatment dose and schedule. All rats received subcutaneous PTH injections for 3 weeks. At the same time, a series of multicolor fluorochrome label injections was administered to each rat, including subcutaneous injections of calcein (green, G, 15mg/kg), an intraperitoneal injection of alizarin complexone (red, R, 30mg/kg), and an intraperitoneal injection of tetracycline (yellow, Y, 30mg/kg) in the order of G-R-Y-G at days −2, 5, 12, and 19 (initiation of PTH treatment on day 0). By the end of the 3-week PTH treatment (day 20), rats were randomly assigned to one of the three groups: 0-week, 1-week, or 3-week PTH withdrawal groups (n=6/group), followed by euthanasia at days 21 (0-week PTH withdrawal group), 28 (1-week PTH withdrawal group), and 42 (3-week PTH withdrawal group, n=6), respectively (Figure 1A). Experiments were performed unblinded.

Figure 1.

(A) Experimental plan of treatment, fluorochrome injections, and euthanasia. (B-G) Representative images of 4 rounds of multiplexed histology imaging shown by (B) dark field and fluorescent light microscopy, (C) TRAP channel, (D) ALP channel, (E) polarizing microscopy, and (F) all channels overlaid. (G) Enlarged image of the region highlighted in (F). (H-M) Representative (H-J) RBF and (K-M) MBF sites shown by (H&K) dark field and fluorescent light microscopy and (I&L) polarizing microscopy, respectively. It should be noted that images in 1I and 1L represented images taken at one fixed polarizer angle. The collagen organization of all bone packets was examined by a polarizer rotation of 180° and the observed results were shown as schematics in 1J and 1M. (H-J) RBF sites were identified by (I) scalloped cement lines with (J) interrupted collagen fiber alignment while (K-M) MBF sites were identified by (L) smooth cement lines and (M) uniform collagen fiber alignment.

2.2. In vivo μCT Scans, image registration, and bone microstructural analysis

In vivo μCT scans of the proximal tibia were performed weekly using the Scanco vivaCT40 (Scanco Medical AG, Brüttisellen, Switzerland) at a 10.5 μm voxel size with a 200 ms integration time, 145 μA current, and 55 kVp energy. During the scan, rats were anesthetized (4/2% isoflurane) and the right tibia was fixed into a customized holder to minimize motion artifacts [21]. A 4.2-mm-thick segment of the proximal tibia (located 0.3 mm below the proximal growth plate) was scanned, with an average scan time of 20 minutes per rat.

To aid in the identification of a consistent volume of interest (VOI), a mutual-information-based, landmark-initialized, rigid registration procedure based on open source software (National Library of Medicine Insight Segmentation and Registration Toolkit) [21, 22] was used to register and align the follow-up scans to the baseline scan, as described in our previously published study [21]. As a result, a 1.5-mm-thick VOI of trabecular bone, located 2.6 mm distal to the growth plate, was identified in μCT images of each sequential scan. Subsequently, the registered VOIs of trabecular bone were Gaussian filtered (sigma=1.2, support=2) and thresholded using a global threshold (corresponding to 544 mgHA/cm³) and trabecular bone microstructure parameters, including bone volume fraction (BV/TV), trabecular thickness (Tb.Th), and structure model index (SMI), were evaluated for all registered VOIs [23].

2.3. Multiplexed cryohistology and sequential immunofluorescence staining

For rats used in the histomorphometry study, the right tibia was harvested after euthanasia and fixed in 4% paraformaldehyde (PFA) at 4 °C for 48 hours. Afterwards, samples were transferred to a solution of 20% sucrose and 2% polyvinylpyrrolidone (PVP) for 48 hours, followed by cryoembedding in Tissue-Tek O.C.T. Compound (Sakura Finetek USA Inc, Torrance, CA) using liquid nitrogen. 8μm-thick, mineralized coronal sections were collected using cryofilm IIC tape (Section-Lab Co. Ltd., Hiroshima, Japan). Sections were attached to glass microscope slides with 1% chitosan adhesive and allowed to adhere for 48 hours. Slides containing the firmly attached sections were submerged in a 1x phosphate buffered saline (PBS) solution for 15 min to rehydrate the sections and mounted with 50% glycerol for multiplexed histology.

Four rounds of histology imaging were performed on every cryosection (Figure 1B–E). First, dark field and fluorescent images were taken with the Axio Scan Z1 (Zeiss, Oberkochen, Germany) to capture fluorochrome labels (Figure 1B). After the first round of imaging, the cover slip was removed by submerging the slide in 1x PBS. In the second round of imaging, TRAP staining was performed using ELF 97 Phosphatase Substrate (Life Technologies, Carlsbad, CA) as previously described [24] before imaging with the Axio Scan Z1 (Figure 1C). After removing the cover slip, sections were subjected to the third round of imaging where Alkaline Phosphatase (ALP) staining was performed using VECTOR Blue Alkaline Phosphatase Substrate Kit (Vector Laboratories, Burlingame, CA). After ALP-stained sections were imaged with the Axio Scan Z1 (Figure 1D), the cover slip was removed and the slide was then submerged into Formical-2000™ decalcifier (StatLab, McKinney, TX) for 1 hour to completely decalcify the tissue section. Then, decalcified sections were remounted and observed under a polarizing microscope (Leica, Wetzlar, Germany) to visualize the cement line and collagen fiber alignment (Figure 1E). All the images were co-localized by using Adobe Photoshop (Adobe Inc., San Jose, CA, USA) for further analysis (Figure 1F&G).

2.4. Identification and histomorphometric analyses of RBF and MBF sites

First, all bone formation sites (including both RBF and MBF sites) were identified and numbered based on the first round of darkfield and fluorescent images. Bone formation sites were identified as sites if they had at least two fluorochrome labels. Next, each identified bone formation site was examined under a polarizing microscope with a 20X objective (Figure 1I&L). Due to the anisotropic nature of the bone matrix, a thorough examination of collagen alignment at each formation site required a polarizer rotation of 180°. Of note, the images from the polarized light microscope in Figure 1I&L only represent images taken at a single, fixed polarizer angle. Based on the organization of the cement line and surrounding collagen fibers, a bone formation site was identified as RBF if it satisfied one of the two conditions: (1) the underlying cement line was scalloped and/or (2) collagen structure underneath the cement line was interrupted (Figure 1H–J) [19]. In contrast, a bone formation site was identified as MBF if it satisfied both conditions: (1) the underlying cement line was smooth and (2) collagen fiber alignment underneath the cement line was uniform (Figure 1K–M) [19]. In previous studies of human bone biopsies, a third category of overflow MBF (oMBF), where new bone formation with a smooth underlying cement line extends beyond the boundary of the RBF, has been reported, which accounts for 17% of all formation sites in PTH treated patients [9, 10]. However, in adult rat bone, we found less than 1% of all formation sites were oMBF [19]. Because of its rare presence, we categorized oMBF as RBF sites in this study.

Dynamic bone histomorphometry analysis was then performed for RBF and MBF sites, respectively. First, new bone formation during the 3-week PTH treatment was quantified based on the 0-week withdrawal group. Mineralizing surface (MS/BS, %) was calculated as the percent of bone surface that displays the last fluorochrome label reflecting active minearlization over the total bone surface. Mineral apposition rate (MAR, μm/day) was measured as the distance between the first and last fluorochrome labels, divided by the time interval between the two label injections. Bone formation rate (BFR/BS, μm³/μm²/day or μm/day) was defined as the amount of newly formed bone tissue in unit time per unit bone surface and calculated as MS/BS multiplied by MAR. Second, new bone formation during the 1-week EAP following PTH withdrawal was analyzed. Mineralizing surface was defined as the surface of newly formed bone tissue beyond the last fluorochrome label and the MAR was measured as the distance between the last injected fluorochrome label and the newly formed bone tissue surface, divided by the withdrawal duration (7 days).

Next, bone area divided by total area (B.Ar/T.Ar) was quantified for MBF, RBF, and pre-existing bone tissue for all three groups. Bone tissue that was formed during both the 3-week treatment and after treatment withdrawal was included. Therefore, B.Ar of MBF and RBF sites was defined as the bone area between the first deposited fluorochrome label and the bone surface that is adjacent to the bone marrow. Lastly, the B.Ar/T.Ar of newly formed tissue after treatment withdrawal (bone area between the last injected fluorochrome label and the bone surface adjacent to bone marrow over the total area) was quantified for the 1-week and 3-week withdrawal groups. Additionally, the percent of MS that continued forming new bone during the EAP was calculated at both MBF and RBF sites.

The above measurements were defined and expressed according to recommendations from the ASBMR Histomorphometry Nomenclature Committee [25] and were measured using OsteoMeasure™ (OsteoMetrics, Atlanta, GA). All histomorphometric analyses were performed by a single reader (WW).

2.5. Evaluation of TRAP and ALP activity on MBF, RBF, and pre-existing bone surfaces

Subsequently, measurements of TRAP & ALP activity were performed. We discovered four types of bone surfaces with different cellular activity (Figure 2), including: 1) TRAP+ surfaces, indicating ongoing osteoclast resorption activity; 2) ALP+ surfaces, indicating ongoing osteoblast formation activity; 3) Mixed surfaces (surfaces covered by mixed TRAP and ALP signals), indicating a transition between osteoblast formation and osteoclast resorption; and 4) Quiescent surfaces, indicating no active osteoblast or osteoclast activity. The images of TRAP & ALP staining were co-localized with fluorescent and polarizing images to facilitate the quantification of the percentage of different types of cellular activity at MBF, RBF and pre-existing bone surfaces, respectively (Figure 2).

Figure 2.

Representative images of (1^st row) ALP+, (2^nd row) TRAP+, (3^rd row) Mixed TRAP+ and ALP+, and (4^th row) quiescent surfaces shown by (A) fluorescent channel for TRAP, (B) fluorescent channel for ALP, (C) TRAP and ALP overlaid, (D) dark field and fluorescent channels for fluorochrome labels, (E) polarizing microscopy, and (F) all channels overlaid.

2.6. Statistical analysis

All statistical analyses were performed using NCSS Statistical System (NCSS, LLC, Kaysville, UT). Results were presented as boxplots with median and interquartile range (IQR; 25^th to 75^th percentile), whiskers indicating maximum and minimum values, and all data points plotted. For longitudinal μCT image-based measurements, a mixed model was used to compare treatment groups over time. All comparisons were adjusted for baseline measures. In the presence of a significant interaction effect between time and treatment, individual comparisons were evaluated using Bonferroni post hoc corrections. A one-way analysis of variance (ANOVA) test with Bonferroni post-hoc test was used to compare the BV/TV of MBF, RBF and pre-existing bone between different time points. A paired students’ t-test was used to compare histomorphometric variables of continuous mineral apposition during the EAP between RBF and MBF sites. A two-way ANOVA was used to compare the effects of bone formation type and withdrawal duration on BV/TV of continuous mineral apposition during the withdrawal period. In the presence of a statistically significant interaction effect, individual comparisons were evaluated using Bonferroni post hoc corrections. For measurements of TRAP & ALP signals, a one-way ANOVA with Bonferroni post-hoc test was used to compare cell activity between different types of bone surfaces or between different withdrawal durations. For all analyses, p≤0.05 was considered to indicate statistical significance.

3. Results:

3.1. Significant activation of MBF and RBF in response to PTH treatment

Results from registered in vivo μCT images (Figure 3A) suggested that 3-week PTH treatment led to a 40% and 29% increase in tibial BV/TV and Tb.Th, and a 17% decrease in SMI (indicating an increase in plate-likeness of trabecular microarchitecture), respectively (Figure 3B–D). Consistent with our previously published study [19], bone formation sites with distinct multicolor labels were prevalent in all PTH-treated groups (Fig 4A–C). Dynamic histomorphometry was further performed for the 0-week PTH withdrawal group to assess the activation of MBF and RBF in response to 3-week PTH treatment. We confirmed our previous finding [19] that once a bone formation site was activated by PTH, mineral deposition at the site continued throughout the entire treatment duration, as demonstrated by the presence of the calcein green label injected 2 days before euthanasia as the last label on all bone formation sites (Fig 4A–C). Moreover, there was robust activation of both MBF and RBF by PTH treatment, with a MS/BS of 27% for MBF and 11% for RBF (Figure 4F). As a reference, our previous study reported that without anabolic treatment, RBF and MBF activity was minimal (2–5% MS/BS) and at similar levels in vehicle-treated adult rats [19]. Moreover, consistent with our previous findings, there was greater activation of MBF than RBF in response to PTH: MS/BS, MAR, and BFR/BS of MBF sites were 141%, 21%, and 192% greater than those of RBF sites, respectively (Figure 4F–H).

Figure 3.

μCT image analysis of tibial trabecular bone in response to VEH and PTH treatment and PTH discontinuation in OVX rats. (A) Representative 3D renderings of the proximal tibia of a VEH- and PTH-treated rat at weeks 0, 3 (after 3-week treatment), and 6 (after 3-week withdrawal). (B-D) Changes in (B) BV/TV, (C) Tb.Th, and (D) SMI of tibial trabecular bone in PTH and VEH groups. + indicates a significant difference from week 0 in the PTH group (p<0.05); * indicates a significant difference from week 0 in the VEH group (p<0.05); # indicates a significant difference between the PTH and VEH groups (p<0.05).

Figure 4.

(A-C) Representative images of secondary spongiosa in the proximal tibia of OVX rats after (A) 0-week, (B) 1-week, and (C) 3-week PTH withdrawal, respectively. (D-H) Histomorphometric analysis of new bone formation during 3-week PTH treatment. (D) A representative image of a bone formation site where the entire site (highlighted by diagonal patterning in E) was analyzed for dynamic histomorphometry to derive (F) MS/BS, (G) MAR, and (H) BFR/BS. (I-J) Histomorphometric analysis of new bone tissue formed during the 1-week EAP upon PTH withdrawal. (I) A representative image of the extended anabolic effect on a bone formation site where only the tissue formed during the EAP (highlighted by diagonal patterning in J) was analyzed for dynamic histomorphometry to derive (K) MS/BS, (L) MAR, and (M) BFR/BS.

3.2. MBF has a greater contribution to the EAP upon early withdrawal from PTH than RBF

One week after withdrawal from PTH, BV/TV, Tb.Th, and SMI continued to show trends toward bone microarchitecture improvement, resulting in 50% and 37% increase in BV/TV and Tb.Th, and a 21% decrease in SMI by the end of week 4 as compared to week 0, respectively (Figure 3B–D). Consistently, continued mineral apposition on new bone formation surfaces upon 1-week PTH withdrawal was discovered in the histology sections. As illustrated in Figure 4I&4J, new bone formation during the EAP was identified as bone tissue beyond the last green label that was injected at the end of the 3-week treatment. This continuous mineral apposition was found on 84% of all MBF surfaces, in contrast to only 31% of RBF surfaces, demonstrating a greater contribution from MBF to the induction of the EAP. Overall, MS/BS of the continuous mineral apposition during the EAP on MBF surfaces was 5.0-fold greater than that of RBF surfaces (Figure 4K). However, MAR did not differ (Figure 4L). This yielded a 5.5-fold greater BFR/BS on the surface of MBF vs. RBF (Figure 4M).

3.3. MBF has a greater resistant to PTH discontinuation-induced bone loss as compared to RBF and pre-existing bone

Following the 1-week EAP, a decline in BV/TV and Tb.Th and an increase in SMI occurred from the second week (week 5) upon withdrawal from PTH (Figure 3B–D). As a result, the treatment benefit on BV/TV diminished as compared to week 0 while Tb.Th was still 34% greater and SMI 16% lower than in week 0, respectively.

There was no difference in B.Ar/T.Ar of RBF between 0- and 1-week PTH withdrawal groups. However, B.Ar/T.Ar of RBF in the 3-week PTH withdrawal group was 81% lower than that of the 0-week withdrawal group, suggesting that the majority of bone tissue formed at RBF sites was resorbed after 3-week discontinuation from PTH. In contrast, B.Ar/T.Ar of MBF did not differ between 0-, 1-, and 3-week PTH withdrawal groups (Figure 5A), suggesting that bone tissue formed at MBF sites was less susceptible to PTH discontinuation-induced bone loss. Similar to RBF, B.Ar/T.Ar of pre-existing bone was 34% lower in the 3-week vs. 0-week withdrawal group while no difference was found between 0- and 1-week withdrawal groups.

Figure 5.

Histomorphometric analysis of B.Ar/T.Ar of RBF, MBF, and pre-existing bone in response to PTH withdrawal in OVX rats. (A) B.Ar/T.Ar of MBF, RBF and pre-existing bone in response to PTH withdrawal in OVX rats. (B) A representative image of a bone formation site in the 3-week PTH withdrawal group where a part of the new bone formation induced by PTH treatment and the EAP has been resorbed. The remaining tissue that was formed during the EAP (highlighted by diagonal patterning in C) was analyzed for dynamic bone histomorphometry to derive (D) B.Ar/T.Ar.

The new bone tissue generated during the EAP was further examined and quantified. In the 3-week PTH withdrawal group, resorption was found on the surface of newly formed bone tissue during the EAP, indicated by formation sites with scalloped surface and partially missing fluorochrome labels (Figure 5BC). As a result, B.Ar/T.Ar of the tissue formed during the EAP at RBF and MBF was 95% and 56% lower in the 3-week vs. 1-week withdrawal group, respectively (Figure 5D).

3.4. Greater osteoblast activity and lower osteoclast activity on MBF vs. RBF surfaces following PTH discontinuation

In the 0-week PTH withdrawal group (Figure 6 A&B), osteoblast activity dominated bone formation sites, with 98% of both RBF and MBF surfaces being ALP+ while the remaining 2% of surfaces had mixed ALP and TRAP signals. On the other hand, pre-existing bone surfaces were either quiescent (56%) or TRAP+ (44%, Figure 6 A&B).

Figure 6.

Analysis of TRAP and ALP activity in OVX rats after (A-B) 0-week, (C-D) 1-week, and (E-F) 3-week PTH withdrawal. (A, C, and E) Representative images showing TRAP and ALP staining overlaid with dark field and fluorescent light microscopy. (B, D, and F) Percentage of different types of surfaces on MBF, RBF and pre-existing bone shown by pie plot. a, b, and c indicate significant difference between MBF and RBF, between MBF and pre-existing bone, and between RBF and pre-existing bone, respectively (p<0.05 by one-way ANOVA with Bonferroni corrections).

After 1-week of PTH withdrawal (Figure 6 C&D), MBF sites were still dominated by ALP signal (82%), while mixed signal and TRAP+ only signal comprised 16% and 2% of MBF surfaces, respectively. In contrast, RBF surfaces that were ALP+ dropped to 34%, while 41% of RBF surfaces became TRAP+ and 25% were covered by mixed ALP and TRAP signals. Similar to the 0-week withdrawal group, pre-existing bone surfaces were either quiescent (60%) or TRAP+ (40%, Figure 6C&D).

By the end of the 3-week withdrawal from PTH (Figure 6 E&F), 96% of RBF surfaces became TRAP+ and 4% of RBF surfaces had mixed signals. For MBF surfaces, 72% became TRAP+ while ALP+, mixed, and quiescent surface were 3%, 13%, and 12%, respectively. While 53% of pre-existing bone surfaces were quiescent and 45% were dominated by TRAP signals, interestingly, 1% of pre-existing bone had mixed TRAP and ALP signals and another 1% was ALP+ (Figure 6 E&F).

4. Discussion:

In this study, we investigated the temporal response of different types of bone tissue to PTH treatment withdrawal and the associated cellular activity in an OVX rat model. We first demonstrated continued mineral apposition at both RBF and MBF sites after PTH discontinuation, resulting in an extended anabolic effect upon 1-week withdrawal from PTH. It was further discovered that MBF sites had a greater contribution than RBF sites to the extended anabolic effect upon early withdrawal from PTH, evidenced by a higher percent of ALP+ surfaces and far greater bone formation activity at MBF vs. RBF sites. Furthermore, significant bone loss occurred after 3 weeks of discontinuation from PTH, resulting from marked loss of newly formed bone tissue from RBF and bone tissue that existed prior to treatment. In contrast, compared to RBF, MBF surfaces had a delayed increase of TRAP activity following PTH discontinuation. As a result, newly formed bone tissue from MBF had greater resistance to PTH discontinuation-induced bone loss than those from RBF.

A recent study by Tseng et al. discovered an EAP upon the first week of withdrawal from PTH in estrogen-deficient rats [20]. In the current study, we confirmed the extended anabolic effect of PTH treatment at a lower dose (20μg/kg/day, as compared to 40μg/kg/day in Tseng et al.) during a 1-week EAP upon withdrawal and elucidated the contributions of MBF and RBF to the extended anabolic effect. First, we found that after 3-week PTH treatment, bone surfaces of MBF and RBF sites were dominated by ALP signal, indicating active osteoblast activity. This is consistent with our dynamic histomorphometry results from the current and previous study that once a formation site is activated by PTH treatment, mineral deposition continues throughout the entire treatment duration [19]. Furthermore, the extent to which mineral deposition continues was drastically different between MBF and RBF after 1-week PTH discontinuation. Dynamic histomorphometry analysis indicated that 84% of MBF surfaces had continuous mineral apposition during the 1-week EAP and consistently, immunofluorescence results suggested that 82% of MBF surfaces were ALP+. In contrast, only about 1/3 of RBF surfaces had mineral deposition or were ALP+ after 1-week PTH discontinuation. Intriguingly, our previous study showed significant differences between the activation, development, and attenuation of MBF and RBF in response to PTH treatment [19]. The distinct response of MBF and RBF to early withdrawal from PTH further supports our working hypothesis that osteoblasts elicited at MBF sites may originate from different osteoprogenitor pools than those at RBF sites, resulting in different dynamic responses to anabolic treatments and discontinuation. Nevertheless, the detailed cellular mechanisms behind the responses of RBF and MBF require further investigation.

After 3-week PTH discontinuation, bone tissue formed from RBF and those that existed before treatment significantly declined while no change was detected for bone tissue formed from MBF. The resorption of tissue formed at RBF sites may have started as early as 1-week after withdrawal as 1/3 of RBF surfaces were TRAP+ and another 1/3 had mixed TRAP and ALP activity. By the 3-week withdrawal, most RBF surfaces were dominated by TRAP activity. In contrast, less than 20% of MBF surfaces had TRAP or mixed TRAP and ALP activity following 1-week PTH discontinuation, consistent with a delayed loss of bone tissue formed from MBF. Nevertheless, after 3-week PTH discontinuation, 70% of MBF surfaces became TRAP+. Although there was no significant loss of bone tissue formed from MBF during the 3-week PTH treatment, more than half of bone tissue formed at MBF surfaces during the 1-week EAP was resorbed after 3-week PTH discontinuation. It is expected that a significant portion of bone tissue formed from MBF during the 3-week PTH treatment would be resorbed after a longer duration of PTH discontinuation. Interestingly, 12% of MBF surfaces became quiescent after 3-week PTH discontinuation, suggesting that at least a portion of newly formed bone tissue induced by PTH would be able to endure long-term PTH discontinuation. Although the benefit of net bone gain may be lost following PTH discontinuation, it is expected that overall bone tissue quality would be improved, as the bone tissue heterogeneity would be increased by the newly formed MBF tissue that endured PTH discontinuation.

Compared to bone tissue formed by MBF and RBF, pre-existing bone is most susceptible to osteoclast resorption. TRAP activity on pre-existing bone stayed stable at around 40–45% by the end of 3-week PTH treatment and during PTH discontinuation. Clinically, it has been recommended that anti-resorptive treatment should be administered upon discontinuation of PTH to prevent bone loss. Meanwhile, the concept of administering PTH cyclically with or without an anti-resorptive was proposed and tested in both clinical studies [26–28] and in a rat model [20]. It was demonstrated clinically that the same cumulative dose of PTH administered cyclically for 4 years led to a comparable increase in BMD as standard daily PTH administration over 2 years [27]. In a rat model, it was reported that a cyclic administration regimen with repeated cycles of on and off PTH treatment was able to leverage the EAP upon early withdrawal from PTH to alleviate PTH withdrawal-induced bone loss, leading to improved bone mass, bone microarchitecture, and whole-bone mechanical properties [20]. The findings from the current study further indicate that pre-existing bone and tissue formed by RBF would be preferentially resorbed during the PTH off cycle while most tissue formed by MBF would remain. Therefore, an important treatment benefit of a cyclic treatment regimen would be improvement of bone tissue quality by replacement of the older, pre-existing bone tissue with newly formed bone tissue from MBF. Moreover, a cyclic administration of PTH with an intervening anti-resorptive agent would be highly desired as it may increase the mineral deposition and number of MBF sites and improve the retention of bone tissue at RBF sites and pre-existing bone surfaces. These potential treatment benefits will be further assessed in our future studies.

To better interpret our results, it is important to recognize both the similarities and discrepancies between the mature human skeleton and the rat skeleton. In clinical practice, the recommended duration of PTH treatment is 18–24 months [29], which corresponds to approximately 3–4 weeks in a rat’s life span [30]. A series of studies investigated the action of MBF and RBF in postmenopausal osteoporotic women at 1 [8], 3 [9], 6, and 24 [10] months of PTH treatment. Iliac crest biopsies showed that in the control subjects, all formation was RBF, whereas in the PTH-treated subjects, both MBF and RBF were activated and MBF accounts for more than 30% of trabecular bone gain by PTH [8] during the first month of treatment and continues to be elevated at month 3 and 6 [9, 10]. However, both RBF and MBF were reduced by 24 months of PTH treatment, as compared to 6 months of treatment [10]. The main findings of these studies are consistent with our previous findings in rats [19], suggesting that early anabolic action of PTH is achieved by stimulating both MBF and RBF and emphasize a greater capacity of PTH to stimulate bone formation early in the course of treatment. However, it should be noted that MBF is intrinsically rare in the adult human skeleton, yet MBF still accounts for more than 30% of trabecular bone gain during the early action of PTH. In contrast, similar amounts of MBF and RBF were found in the adult rat tibia regardless of sex and estrogen status and MBF accounts for 75% of new bone formation in response to PTH treatment in rats [19]. Given the lower amount of MBF-derived new bone tissue in humans than in rats, the extended anabolic effect would be less significant and the activation of osteoclasts would be faster following PTH withdrawal. Therefore, we expect the PTH withdrawal effect to be more severe in humans as compared to rats. This information should be taken into consideration when designing follow-up therapies or cyclic PTH treatment regimens for clinical practice.

This study has several important strengths. To our knowledge, this is the first study that tracked the response of RBF sites, MBF sites, and pre-existing bone to the discontinuation of PTH treatment. It resulted in important information about the consequences of anabolic treatment discontinuation on the two distinct bone formation types and pre-existing bone, which supports strategies of cyclic administration of anabolic agents with and without an intervening anti-resorptive agent to improve anabolic treatment benefit. The current study used a novel imaging strategy based on cryosections, which not only allows for the reliable identification of MBF and RBF, but also enables the investigation of spatial-temporal coupling between bone cell activity and MBF/RBF through sequential immunofluorescence staining on a single cryosection. This technical innovation helped us to elucidate the cellular activity on MBF, RBF, and pre-existing bone surfaces which explained the distinct response of these three types of bone surfaces to PTH discontinuation.

However, this study is not without limitations. Due to the two-dimensional (2D) nature of cryosections, some formation sites could be incorrectly interpretated as MBF based on the observed section plane while a missed scalloped cement line may be present outside of the section plane. This type of RBF site was referred to as overflow MBF (oMBF), where new bone formation with a smooth underlying cement line extends beyond the boundary of the RBF [9, 10]. To overcome the limitations from 2D sections, we developed a new 3D imaging strategy using confocal microscopy coupled with second harmonic generation (SHG) to image thick plastic sections [19], which allows 3D identification of MBF and RBF sites. By comparing the identification of MBF/RBF between 2D and 3D imaging strategies in rat bone, we observed that only 4.1% of formation sites that were incorrectly identified as MBF on a 2D image had scalloped cement lines present on adjacent images from a 3D image stack [19]. Because of the uncommon occurrence of oMBF in adult rat bone (less than 1% of all formation sites in rats [19] vs. 17% in human [9, 10] based on 2D sections), we do not expect a significant overestimation of MBF in the current study. However, it is possible to overestimate the pre-existing bone surfaces, especially in the 3-week withdrawal group. If an entire labeled bone region of a MBF or RBF site was completely resorbed, this location would have been identified as a pre-existing bone surface.

Another limitation of cryosections is the poorly maintained cell morphology, making it difficult to reliably associate TRAP or ALP signals with specific cells. Therefore, we quantified TRAP or ALP signals as percentage of positive TRAP or ALP bone surfaces instead of positive cells. Moreover, previous studies have demonstrated that reversal cells could take up TRAP from neighboring osteoclasts [31]and that these cells could also express ALP [32]. On mixed surfaces, we have observed that some surfaces expressed strong TRAP signals and weak ALP signals. Meanwhile, we’ve also seen some mixed surfaces with strong ALP signals and weak TRAP signals. While most TRAP+ surfaces were eroded surfaces, not all mixed surfaces showed signs of erosion. This is because mixed surfaces could be at any stage of transition from resorption to formation. Such surfaces would not have characteristics of eroded surfaces if the mixed bone surface were at the beginning of the reversal phase.

We used fluorochrome injections to label the active mineralizing bone surface in order to capture bone formation sites that were active during PTH treatment. However, some of these formation sites (those that start with a G label) could have been active even before PTH treatment was initiated. In our previous study, we compared the bone formation activity between VEH-treated and PTH-treated rats that followed the same schedule of fluorochrome injections [19] and found that the majority of bone formation sites in PTH-treated rats were activated after initiation of PTH treatment. VEH-treated OVX rats have low levels of bone formation activity (4.0% and 2.1% MS/BS for MBF and RBF over 3-week VEH treatment, respectively) while in contrast PTH-treated OVX rats had far greater bone formation activity (21.5% and 9.3% MS/BS for MBF and RBF over 3-week PTH treatment, respectively) [19]. These data suggested that bone formation events captured in our current study are dominated by those activated after PTH treatment was initiated.

Lastly, the current study focused on the trabecular bone in the proximal tibia of the adult rat, partially because trabecular bone is more susceptible to bone loss upon PTH discontinuation [5]. Meanwhile, bone formation at the periosteal and endosteal cortex of the rat tibia metaphysis is confounded by the continuous endochondral ossification due to longitudinal growth in rats [33, 34]. Future studies regarding the responses of MBF and RBF in the rat lumbar vertebrae to anabolic treatment and discontinuation will be of great interest as the longitudinal growth of the rat lumbar vertebrae is minimal at the age of 6 months and becomes undetectable beyond the age of 9 months [33].

In summary, this study demonstrated an extended anabolic effect upon early withdrawal from PTH treatment in OVX rats, which was mainly a result of continuous osteoblast activity on MBF surfaces. Moreover, MBF sites were associated with a delayed activation of osteoclast resorption upon PTH withdrawal while in contrast, pre-existing bone and RBF surfaces immediately responded to PTH withdrawal with rapid activation of osteoclast resorption. This study also showcased our cryohistology imaging platform that utilizes cellular activity related staining to enable tracking of both dynamic responses and cellular activity of RBF and MBF in a single section. This innovative imaging platform and the newly-established rat model will enable our future investigations to advance fundamental understanding of cellular mechanisms associated with MBF and RBF and their contributions to different osteoporosis treatment strategies.