A quantitative analysis of bone lamellarity and bone collagen linearity induced by distinct dosing and frequencies of teriparatide administration in ovariectomized rats and monkeys

Abstract

The lamellar structure of bone, which endows biomechanical rigidity to support the host organism, is observed in mammals, including humans. It is therefore essential to develop a quantitative analysis to evaluate the lamellarity of bone, which would especially be useful for the pharmacological evaluation of anti-osteoporotic drugs. This study applied a current system for the semi-automatic recognition of fluorescence signals to the analysis of un-decalcified bone sections from rat and monkey specimens treated with teriparatide (TPTD). Our analyses on bone formation pattern and collagen topology indicated that TPTD augmented bone lamellarity and bone collagen linearity, which were possibly associated with the recovery of collagen cross-linking, thus endowing bone rigidity.

Introduction

The most characteristic feature of vertebrates is the bone, which is mainly composed of collagen fibers and hydroxyapatite. The evolution of bone has been associated with the development of a hormonal system that regulates mineral metabolism. The parathyroid hormone (PTH) plays key roles in the homeostasis of calcium phosphate through its actions in the bone and kidneys, and indirectly in the intestine [1–3]. Furthermore, the functional evolution of PTH has been thought to be a key event in appearance of terrestrial vertebrates [4–6].

In skeletal ontogeny, primary bone develops in a de novo manner, where no bone existed previously, in given parts of fetal mesenchymal tissue, which is structurally termed woven bone. Then, mature bone develops in a manner in which previously formed bone (either woven bone or lamellar bone) is replaced by new lamellar bone [7]. It is understood that woven bone forms when rapid bone formation is required, such as in early skeletal development and bone repair. Thus, woven bone appears to be formed as a ‘place-taking’ bone for future mature bone formation. Lamellar bone has a highly organized layered structure with well-oriented thick collagen fibers, whereas woven bone is depolarized hard tissue with randomly oriented and enriched collagen fibrils. Kimmle and Jee estimated that the formation rate of woven bone is six times faster than that of lamellar bone [8]. The lamellar structure of bone is commonly observed in mammals, including humans [9–11], and is proposed to be affected by several factors, including phylogeny, ontogeny, environmental influence and biomechanics [11]. Furthermore, this lamellar structure is incrementally formed. In this formation, new single lamella is deposited on the surface of preformed lamellar bone, which appears to be species-specific and time-dependent [12]. Biochemically, lamellar bone has more enriched non-collagenous proteins, such as bone sialoprotein and osteocalcin, and well-developed calcium phosphate crystals, which are known to be components of mature bone, and it is less susceptible to bone resorption in comparison to those in woven bone [13]. These findings suggest that a slow rate of bone formation is tightly associated with mature bone lamellarity and possibly with maintaining the rigidity of bone such that it is sufficient to sustain the host organism.

Pharmacologically, the exogenous administration of PTH, which promotes bone formation and resorption through its receptor, parathyroid hormone type I receptor depending on whether the administration is intermittent or continuous, has a paradoxical effect in bone [14]. In fact, the intermittent administration of human PTH (1–34) [Teriparatide (TPTD)] and PTH (1–84) are approved as bone anabolic agents for the treatment of osteoporosis [15–18]. Thus, investigation into how different TPTD dosing regimens induce different effects on bone has been a focus of some studies on the pharmacological action of TPTD. To meet clinical requirements, the pharmacological action of TPTD has been evaluated through the parameters of bone formation, bone resorption, bone microarchitecture and bone strength.

Our research group previously reported that the less frequent administration of TPTD (e.g. 3 times/week) more effectively increased the bone mineral density (BMD) and mechanical strength of the lumbar vertebra and femur in an ovariectomized (OVX) rat model in comparison to the frequent administration (e.g. 3 times/day) [19]. This study also revealed that the frequent administration significantly augmented the formation of cortical porosity associated with the increase in bone resorption markers, whereas the less frequent administration was also associated with increased levels of bone formation markers but not with increased levels of bone resorption markers. We also analyzed the effect of the frequency of TPTD administration on rabbit bone with the same total dose but a distinct frequency of administration, such as once-daily or once-weekly. Consistent with the results from rats [18], this work demonstrated that daily but not weekly regimens were associated with the significant development of cortical porosity and marrow fibrosis [20]. Our line of these works therefore demonstrates that the frequency of TPTD administration affects the balance between bone formation and resorption. We also morphologically observed that the low frequency of TPTD administration led to the formation of thicker trabeculae with the maintenance of thick cortices, in comparison to high-frequency administration, suggesting that the frequency of TPTD administration also affects the structural pattern of bone formation, such as bone lamellarity [18,19].

Bone histomorphometry is a unique method that is the gold standard for the evaluation of the rate of bone turnover, which has long been well applied in both basic and clinical bone research [21,22]. However, this morphometric method lacks spatial scores, such as the pattern of bone formation and the lamellarity of bone. Moreover, this method requires a high degree of skill for histological observation and thus shows a degree of subjectivity. This study applied a current semi-automatic recognition system to the non-demineralized histological bone sections with calcian labeling and Villanueva bone staining, which are well-established histomorphometric methods that are used to evaluate the rate of bone turnover. We also subjected these bone sections to collagen imaging by second harmonic generation (SHG) [20,23–27]. To statistically analyze the spatial pattern of bone formation associated with bone collagen, we re-evaluated previously reported rat and cynomolgus monkey bone specimens that were treated with TPTD [19,28,29].

Materials and methods

Animals and experimental design

All experimental protocols were approved by the experimental animal ethics committee of Asahi Kasei Pharma Co., and were conducted in accordance with guidelines concerning the management and handling of experimental animals.

Rats: This study re-examined rat bone specimens that we previously analyzed and reported [19]. Briefly, sexually mature female Sprague-Dawley rats (Charles River, Kanagawa, Japan) of 13 weeks of age were divided into 15 drug regimen groups (Supplementary Fig. 1). Ovariectomy (OVX) or sham surgical operations were conducted under anesthesia, as previously described. At 16 weeks of age, the rats were subcutaneously injected with vehicle saline, or 6 or 30 μg/kg of TPTD acetate (Asahi Kasei Pharma Corporation, Tokyo, Japan), 3 times/week (W3 groups, marked in red); vehicle, 1.2, 6 or 30 μg/kg of TPTD once a day (D1 groups, in orange) or twice a day (D2 groups, in green); and vehicle, 1.2 or 6 μg/kg of TPTD 3 times/day (D3 groups, in blue) for 4 weeks. On the eighth and third days before sacrifice, double-fluorochrome labeling was conducted with two subcutaneous injections of calcein (Dojindo Laboratories, Kumamoto Japan) into each rat at a dose of 10 mg/kg body weight, in order to observe sites of active bone formation.

Monkeys: This study also re-examined the bone specimens of female cynomolgus monkeys (C.V. Universal Fauna, Jakarta Timur, Indonesia), as previously analyzed and reported [28]. Monkeys (age: 12.0 ± 1.5 years, weight: 2.06–3.48 kg) were divided into four groups (n = 19–20): (i) the Sham group; (ii) OVX group; (iii) the TPTD-L group, OVX with low-dose TPTD (1.2 μg/kg, once a week); and (iv) the TPTD-H group, OVX with high-dose TPTD (6.0 μg/kg, once a week) (Supplementary Fig. 2). OVX and sham operations were carried out 1 week before TPTD treatment. TPTD or saline (as a vehicle) was subcutaneously injected once a week, for 18 months (Supplementary Fig. 2a). To determine bone formation rates and sites of active mineralization, double-fluorochrome labeling with intravenous calcein injection (4 mg/kg; Dojindo Laboratories, Kumamoto, Japan) at Days 7 and 21 before sacrifice was conducted for all monkeys.

Bone histomorphometry and selection of specimens for the semi-automatic imaging analysis

Bone histomorphometry was performed on the sagittal sections of second lumbar vertebra from all monkey specimens to evaluate bone mass, structure and bone metabolism (resorption and formation) in cancellous bone using an image analysis system (Histometry RT Camera, System Supply, Nagano, Japan; OsteoMeasure, Decatur, GA, USA). The nomenclature, symbols and units used in this study are those recommended by the Nomenclature Committee of the American Society for Bone and Mineral Research [21,22]. The following three classified parameters were measured: static parameters (trabecular bone volume [bone volume/tissue volume {BV/TV}], trabecular bone surface [BS]/BV, trabecular thickness, trabecular number and trabecular separation), a bone resorption parameter (eroded surface/BS), and bone formation parameters (mineralizing surface/BS, single- and double-labeled surface BS, and bone formation rate/BV).

As a further semi-automatic imaging analysis, we selected three specimens from each of four regimen setting groups (19–20 specimens) that showed histomorphometric scores nearest to the mean value (Supplementary Figs. 2b and c). These selected samples were subjected to SHG imaging.

Differential interference contrast and fluorescence imaging

Differential interference contrast (DIC) imaging and fluorescence imaging were obtained using a microscopy system, ECLIPSE Ni (Nikon, Tokyo, Japan) equipped with a DIC microscope and objectives (Nikon) Plan Apo λ ×10 (numerical aperture [NA] = 0.45), Plan Apo λ ×20 (NA = 0.75) and Plan Apo λ ×40 (NA = 0.95). The fluorescence signals were obtained using filter sets, fluorescein isothiocyanate (FITC) (excitation: 460–500 nm, dichroic mirror (DM): 505 nm, emission: 510–560 nm; Nikon) for calcein and TxRed (excitation: 540–580 nm, DM: 595 nm, emission: 600–660 nm; Nikon) for autofluorescence derived from soft tissue. Tiling fluorescence imaging to acquire the entire, high-contrast view of the tissue sections was carried out using the Plan Apo λ ×10 objective (NA = 0.45). The frame size of a single scan was 1280 × 1024 pixels, with an 8-bit color depth. The fluorescence and DIC images were sequentially acquired, with a pixel size of 0.64 μm. In the case of SHG imaging, image processing was performed using the NIS-Elements AR imaging software program (Nikon, Tokyo, Japan).

Second harmonic generation imaging

SHG imaging of monkey seventh lumbar vertebra was carried out using an upright multi-photon excited microscopy system (A1RMP, Nikon, Tokyo, Japan) at the Imaging Core facility of the Institute of Medical Science, the University of Tokyo, to observe bone collagen that emits SHG signals via nonlinear excitation phenomena. The microscope was equipped with a water-immersion objective lens (Apo LWD 40×, NA: 1.15WIλS, Nikon) and a Chameleon laser oscillator (COHERENT, Inc. CA). The excitation wavelength was 880 nm. The detection wavelength for the acquisition of collagen-derived SHG signals was 446 nm. Three-dimensional (3D) images were acquired by taking 20 optical slices with a step size of 4.0–4.5 µm and a pixel size of 2.49 µm. Image processing, including 3D projection and 3D rendering, were performed using the NIS-Elements AR imaging software program (Nikon).

Semi-quantitative fluorescence morphometry and automatic morphometric recognition analyses using artificial intelligence and deep learning

Quantitative topological analyses of fluorescence signals in un-decalcified bone sections in the current study were established with the application of commercially available imaging analysis tools (NIS-Elements AR and NIS.ai, Nikon, Tokyo, Japan). The following parameters of recognized objects were semi-automatically measured: object number, length, perimeter, outer perimeter, width, object area, object fill-area, orientation and smoothness (hull convex perimeter/perimeter). Examples of the measurement parameters for the analyses of trabecular bone, calcein labeling and collagen fibers are schematically illustrated in Supplementary Fig. 3a, b and c, respectively. Binarization of objects was interactively conducted using the fluorescence intensity threshold, object size and manual histomorphological definition. The recognition of objects was set as follows: trabecular bone (threshold 200, size > 50.0 µm), fibrous collagen (threshold 800, size > 40.0 µm), soft tissue (threshold 10, size > 50.0 µm) and calcein labeling (threshold 50, size > 5.0 µm). The automatic morphometric recognition of osteocytes was carried out by artificial intelligence (AI) and deep learning methods (NIS.ai) with the detection of autofluorescence signals derived from osteocytic lacunae after conducting morphological recognition learning more than 1000 times.

In the measurement of monkey samples, the region of interest was a rectangular form (vertical 2.0 mm; horizontal 2.5 mm). The region was ≥1.0 mm from the cortical bone and growth plate. In the measurement of rat samples, a square form (vertical 1.0 mm; horizontal 1.0 mm) was selected. The region was ≥0.2 mm from the cortical bone.

Statistical analysis

All data are presented as mean ± standard error (SE). The effects of TPTD on bone specimens were evaluated using an analysis of variance. In the rat study, Dunnett’s test was applied to compare the treatment and vehicle control group with each frequency of administration. In the monkey study, Dunnett’s test was applied to compare with OVX group. In both studies, P-values of <0.05 were considered to indicate statistical significance.

Results

By taking full advantage of current histological fluorescence morphometry with a commercially available semi-automatic recognition system, we first analyzed the BV/TV in rat vertebrae in 15 TPTD regimen settings, as we previously reported [19] (Fig. 1, and Supplementary Fig. 1). In this analysis, the BV/TV was measured by measuring the area of autofluorescence derived from soft tissue in histological sections subjected to Villanueva bone staining (Fig. 1a, and Supplementary Fig. 4). The bone area was measured by subtraction of this soft tissue area from the measured total tissue area. This analysis demonstrated that the administration of TPTD augmented the BV/TV in a clear dose-dependent manner at all four frequencies of administration (3 times/week and 1, 2 and 3 times/day) (Fig. 1b and c, and Supplementary Fig. 5). This analysis showed exactly the same pattern as our previous analysis of BMD measured by dual-energy X-ray absorptiometry [19]; thus, this analytical method was well validated.

Fig. 1.

The anabolic effect of TPTD on BV of rat vertebral trabeculae, as quantified by a semi-automatic recognition system. (a) Representative DIC (left panels) and corresponding fluorescence images of rat vertebral trabeculae treated with the indicated regimens of TPTD. In the right panels, fluorescence signals derived from calcein and autofluorescence of soft tissues are shown in green and red, respectively. Scale bars indicate 200 µm. (b, c) Quantitative and statistical analyses of BV/TV by a semi-automatic recognition system are shown. Data are shown as mean. *Indicates P < 0.05 vs. vehicle for each frequency of administration (Dunnett’s test).

We next applied the semi-automatic recognition system to fluorescence morphometry of bone areas labeled by calcein. We set parameters, such as the area, length, outer perimeter and number of calcein-labeled signals using the measurement function provided by the system, and performed comparative scoring (Fig. 2, and Supplementary Fig. 3a and b). Interestingly, in the 3 times/week and 1 time/day groups, the scores of area, length and outer perimeter were increased in a dose-dependent manner, while the number was decreased in a dose-dependent manner (Fig. 2b). These tendencies were less obvious in the 2 times/day and 3 times/day groups. Notably, in the 2 times/day group, the highest dose (30 µg/kg) of TPTD did not augment the scores of the labeled area, length and outer perimeter in comparison to a dose of 6 µg/kg. The calcein-labeled length/bone perimeter showed a similar pattern to the length, while the width of calcein labeling did not differ to a statistically significant extent, except in the 2 times/day group, which showed a dose-dependent increase (Supplementary Fig. 6). In this analysis, a dose-dependent increase in bone formation length by TPTD was only observed at lower administration frequencies of 3 times/week and 1 and 3 (but not 2) times/day (Fig. 2, and Supplementary Figs. 6 and 7) [19,28,29].

Fig. 2.

The quantitative topological analysis of bone formation patterns by morphometry of calcein signals in rat vertebral trabeculae. (a) Representative calcein-derived fluorescence images of rat vertebral trabeculae treated with TPTD at the indicated regimens. These are magnified views of the region of interest framed by white dotted squares in Supplementary Figure 4. Scale bars indicate 200 µm. (b) The topological parameters of calcein signals, such as object area, length, number and outer perimeter were measured and statistically compared. Data are shown as mean. * Indicates P < 0.05 vs. vehicle for each frequency of administration (Dunnett’s test).

To confirm that these parameters in calcein-labeled tissues were increased by the administration of TPTD in another experimental animal treated with TPTD for a longer period (where the pharmacological effect of TPTD appears to be saturated), we re-examined the same experimental set of monkey specimens with four distinct regimen groups (Sham, OVX, TPTD-L [1.2 μg/kg] and TPTD-H [6 µg/kg]) with once-weekly administration for 18 months, as previously reported (Fig. 3, and Supplementary Fig. 2). To subject these monkey specimens to further semi-automatic analyses, we needed to select small numbers of specimens from the four experimental groups for acquisition of whole 3D SHG images of entire vertebral body sections (size, ∼1 × 2 cm; depth, 300–400 μm). For this purpose, we initially conducted conventional bone histomorphometry in all specimens. There were no significant differences in parameters such as BV/TV, BS/BV, trabecular thickness, trabecular number, trabecular separation, single-labeled surface/BS, double-labeled surface/BS, mineralizing surface/BS, bone formation rate/bone and surface (bone formation rate/BV) (Supplementary Fig. 8, Supplementary Table 1a), which was consistent with a previous report [29]. Based on this morphometric analysis, we selected three representative specimens from each of the Sham, OVX, TPTD-L and TPTD-H experimental groups. A bone morphometric analysis of these selected specimens did not show any significant differences (Supplementary Figs. 2 and 9, Supplementary Table 1b), thus indicating that our selection of specimens in this step was not biased. These representative specimens from each experimental group were subjected to further analyses.

Fig. 3.

The quantitative topological analysis of bone formation patterns by morphometry of calcein signals in monkey vertebral trabeculae. (a) Representative calcein-derived fluorescence images of monkey vertebral trabeculae treated with the indicated TPTD regimens. Scale bars indicate 100 µm. (b) Topological parameters of calcein signals, such as object area, length, width and calcein length/bone outer perimeter, were measured and statistically compared. Data are shown as mean ± SE. P-values vs. OVX group did not show statistical significance (Dunnett’s test) (P > 0.05).

In the trabecular bonein the vertebrae of this model animal, the area and length of calcein-derived signals were decreased, albeit not significantly, in the OVX group in comparison to the Sham group (Fig. 3). These parameters in the TPTD-H group were recovered to the level in the Sham group, whereas the average width was the same (Fig. 3).

The combinatorial observation of bone tissues with a DIC microscope and these quantitative analyses of the calcein-labeled tissue area made us notice that the spatial distribution, such as the linearity of the labeled tissue, was associated with lamellar structure formation of trabecular bone in the rat and monkey models (Figs. 1a and 3a). This finding prompted us to take advantage of the application of nonlinear optics, which enabled us to visualize fibers or bundles of collagen in organic tissues by SHG. We conducted large tiling and 3D imaging of whole sagittal sections of monkey vertebral bodies (Fig. 4a and b). Intense SHG-derived fluorescence signal was observed in whole bone tissue, including the inner trabecular bone and outer cortical bone, but not in cartilage or soft tissues, such as bone marrow. Intense fibrous signals that extended multi-directionally in bone tissue were clearly recognizable, while relatively weak and broad signal covers whole bone tissue. Therefore, for morphometrical analyses based on SHG signal, we set up two types of threshold: an intense threshold to detect fibrous collagen and a mild threshold to detect bone tissue through binarization of object size and signal intensity (Fig. 5, and see Materials and Methods). We measured the size, area, length and width of the fibrous SHG signals (Fig. 6) and compared these parameters in the four TPTD administration groups. The average number of fibrous SHG signals/tissue area (N.SHG) in the OVX group was higher than that in the Sham group, but not to a statistically significant extent (Fig. 6). The administration of TPTD tended to reduce N.SHG in a dose-dependent manner. In contrast, the average area and length of the fibrous SHG signal (Area SHG and Length SHG, respectively) were decreased in the OVX group, while these showed a dose-dependent recovery with the administration of higher dose of TPTD: TPTD-H (Fig. 6). In comparison to these parameters, the width of the fibrous SHG signal did not show much difference in these four groups (Fig. 6). The average number of fibrous SHG signals/bone area (N.SHG/bone) and the average area of fibrous SHG signal/bone area (Area SHG/bone) showed tendencies that were consistent with the corresponding parameters of the N.SHG and Area SHG (Fig. 6). The analyses of these parameters demonstrated that the administration to OVX monkeys augmented the length but not the number of bone collagen fibers. Consistently, the smoothness of the fibrous SHG signal (Smo.SHG) tended to be increased by the administration of TPTD in comparison to that in OVX (Fig. 6). The orientation of SHG signals in the TPTD-H group showed an increase in the horizontal direction (Supplementary Fig. 10). Therefore, it was suggested that the administration of TPTD induced the linear and isotropic distribution of collagen fibers, whereas fractioned and anisotropic distribution of collagen was observed in OVX.

Fig. 4.

SHG and DIC tiling images of monkey vertebral trabeculae. (a) Representative SHG (upper panels) and corresponding DIC tilling images were acquired from the Sham group, OVX group, PTH-L (low-dose TPTD) group and PTH-H (high-dose TPTD) group. Scale bars indicate 5.0 mm. (b, c) The mid-dorsal portions framed by dotted boxes in the corresponding images shown in (a) are magnified. Corresponding DIC images are also shown for morphological comparison. Scale bars indicate 100 µm.

Fig. 5.

Binarization of SHG images for the analysis of monkey bone collagen and trabeculae. (a) Raw data of SHG imaging of monkey vertebral trabeculae. (b, c) Corresponding binary images to those of bone collagen fibers and trabecular bone (shown in [a]). The threshold conditions for visualization of bone collagen and trabecular bone are indicated. In the analyses of bone collagen and trabeculae, the object sizes are >40 µm and >50 µm, while the luminance intensities of the objects are >800 and >200, respectively.

Fig. 6.

Quantitative analyses of bone collagen spatial patterns. The topological parameters of SHG signals, such as the number, object area, length, width, number/bone area and smoothness were measured and statistically compared. Data are shown as mean ± SE. * Indicates P < 0.05 vs. OVX group (Dunnett’s test).

We further applied an AI-based recognition system for counting the number of osteocytes that were differentiated from osteoblasts—a number that reflects the pattern and rate of bone formation (Fig. 7). Specimens in the TPTD-H group tended to show greater numbers of osteocytes/bone area, suggesting correlations with length SHG and width SHG.

Fig. 7.

Automatic morphometric recognition analyses of osteocytes using AI and deep learning. (a) Representative views of AI-based morphometric recognition of osteocytes (upper panels) at indicated TPTD administration regimens and corresponding SHG and autofluorescence images (lower panels). (b) The average osteocyte number/bone area was compared (N = 3). (c) The osteocyte number/bone area was plotted for the four administration regimens. Open square, closed triangle, closed diamond and closed circle indicate specimens from the Sham group, OVX group, TPTD-L group and TPTD-H group, respectively. R = 0.36 and 0.45.

Discussion

This study applied a current semi-automatic recognition system to the non-demineralized histological bone sections with calcein labeling and Villanueva bone staining. We have noticed that autofluorescence at a wavelength of ∼560 nm from these bone sections clearly demarcate soft tissues and cells, such as bone marrow, muscle, tendon and osteocytes embedded in bone. We quantified the morphometric parameters of this autofluorescence and calcein-derived fluorescence signal that demonstrated pattern of bone formation.

To confirm the reliability of our approach, we measured the BV/TV in rat vertebrae in 15 TPTD administration groups [19]. The data we obtained were consistent with our previous analysis of BMD measured by dual-energy X-ray absorptiometry [19]. Conventional bone morphometry is performed manually and requires specialized knowledge and skills, and—to some extent—cannot exclude potential subjectivity, especially in the quantification of histological specimens. However, our semi-automatic recognition system enabled the comprehensive measurement and analysis of biometric information more easily and quickly than conventional systems, thus excluding subjectivity. Some clinical attempts to grasp osteoporotic fractures by applying a semi-automatic morphometric method using digital imaging have been reported [30,31]; however, a completely automatic level has not yet been achieved. Therefore, the development of an AI-based recognition system in the near future would be beneficial to biometric informatics.

We next applied this semi-automatic recognition system to the analysis of fluorescence signals from calcein labeling. In conventional histomorphometry, calcein-based parameters reflect the bone formation surface and rate; however, they lack information on spatial patterns of bone formation. Our parameter analysis, which measured the number, area, length, outer perimeters and smoothness of the calcein-labeled area, provided a method for the quantification of spatial patterns of bone formation. This analysis demonstrated the linearity of calcein signaling—possibly due to synchronicity of bone formation—was augmented by the administration of TPTD in a dose-dependent manner. This analysis further indicated that the excess amount and frequency of administration rather negatively affected the synchronicity of bone formation, which would not have been noticed by conventional histomorphometry methods. Interestingly, our calcein length analysis did not demonstrate a dose-dependent increase in bone formation surface especially in the group of 2 times/day (shown in Fig. 2b), while BV/TV obviously showed a dose-dependent increase in BV (shown in Fig. 1b). These findings appeared to be inconsistent, however, can be interpreted as a way that concave trabecular surfaces were getting filled with new bone formed by higher TPTD administration, thus these BSs became flattened, which could cause shorter calcein labeling. In fact, this phenomenon was often observed in our series of previous pharmacological works of TPTD, but failed to be demonstrated in a quantitative way. Therefore, this analysis is available to quantitatively demonstrate the spatial patterns of bone formation.

Bright-field DIC images showed that the increased linearity of calcein labeling observed after TPTD treatment was highly associated with the lamellar structure of bone. It has been reported that the visual appearance of the lamellar structure of bone is formed by the repeated appearance of collagen-rich (dense or thin lamellae) surface and collagen-poor (loose or thick lamellae) surface [32–34], which prompted us to visually quantify the topological characteristics of the bone collagen. We took advantage of advanced nonlinear optics imaging (SHG) using a multi-photon microscopy system, which enabled the visualization of bone collagen fibers so that we could apply our semi-automatic recognition system to measure the spatial pattern of the bone collagen fibers of bone specimens obtained from cynomolgus monkeys of the TPTD regimen set reported by Saito et al. [28].

After a relatively long period of TPTD treatment (18 months), none of the histomorphometric parameters in monkeys showed significant differences, indicating that the increased bone turnover induced by once-weekly TPTD treatment had already plateaued during the long treatment period in these monkey specimens, which was consistent with a previously reported observation in relation to the same monkey specimens [29]. However, a biochemical analysis of collagen cross-link properties of the same monkey specimens demonstrated that low-dose and high-dose TPTD treatments increased the content of enzymatic immature and mature cross-links and decreased advanced glycation end product cross-links, such as pentosidine, in comparison to the OVX group [28]. In our imaging analyses of the same monkey specimens in this study, the average length/bone perimeter (bone formation surface/bone perimeter) in the monkey OVX group tended to decrease in comparison to that in Sham group, while these parameters appeared to recover—in a dose-dependent manner—by the administration of TPTD. Similarly, the average length and area of the collagen fiber–derived SHG signal (Length SHG and Area SHG, respectively) in the OVX group also tended to decrease in comparison to the Sham group, while these parameters recovered—in a dose-dependent manner—with the administration of TPTD. The numbers of collagen fiber–derived SHG signals in the four experimental groups showed an inversed pattern with respect to length and area. Furthermore, the smoothness of the fibrous SHG signal (Smo.SHG) tended to be increased by the administration of TPTD. Although changes in these parameters were not statistically significant possibly due to lack of enough sample size (N = 3/group), they were consistent with the results of a calcein-labeling analysis in rats as mentioned earlier. These analyses, therefore, suggested that the spatial arrangement of collagen fibers in the OVX group was fractioned and anisotropic, whereas TPTD treatment enhanced the linear and isotropic arrangement of collagen fibers and the lamellarity of the trabecular bone microstructure, even after a long treatment period in which the pharmaceutical effects on bone turnover by TPTD was less evident.

In terms of the clinical aspect and pharmacology of TPTD, it is relevant to consider how the enhanced linear collagen arrangement and bone lamellarity contribute to improving the mechanical properties of bone to reduce the fracture risk. Recent studies reported that the connectivity of the collagen network is an essential factor for evaluating the toughness of bone [28,35,36]. The flexibility of physiological enzymatic cross-links of collagen fibers enables the bone matrix to be sufficiently deformed to facilitate energy dissipation [36], whereas the cross-linking by advanced glycation end products that is associated with aging and metabolic disease, such as diabetes, over-stabilize bone collagen, leading to the presentation of brittle bone and a reduction of bone toughness [28,35,36]. Accordingly, smooth linear sliding between collagen molecules is a critical factor for achieving energy dissipation, which can be facilitated by physiological enzymatic cross-links and the linear arrangement of collagen fibers [36]. Therefore, our analyses demonstrated that TPTD treatment enhanced the mechanical properties of bone, not only through the recovery of healthy cross-links but also through the augmentation of the linearity of collagen fibers and the lamellarity of the bone microarchitecture.

Our study demonstrated the changes in properties of collagen fibers and bone formation patterns that occur with the administration of TPTD, focusing on bone lamellarity. In particular, it is significant that the topological character of bone collagen was visualized and quantified by SHG imaging and a semi-automatic morphology recognition system. Our approach here fills the gap between collagen motion—at the sub-nano scale to macro scale—in the bone hierarchical structure [37–39]. Collagen is the main structural protein in bone and contributes to bone toughness by dissipating energy via its deformation [37,40]. At the sub-nano scale to sub-micro scale, the deformation of collagen fibers plays an important role in the energy dissipation, as discussed above [35,37,39]. However, at the microscale, the discussion on bone structures associated with bone toughness moves to structural features of bone, such as trabecularity and density [37,39]. Accordingly, it is relevant to analyze the spatiotemporal arrangement of collagen fibers [34]. This study enabled us to statistically compare the topological characteristics of collagen fibers of bone at the micro scale. In conclusion, an increase in the linearity of collagen fibers, as well as an increase in bone formation surface and area, contributes to the endowment of bone with more toughness by dissipating energy. Thus, collagen linearity is a relevant parameter for estimating the bone toughness pharmacologically imposed by the administration of TPTD.

Concluding remarks

Our study demonstrated the changes in bone formation patterns and collagen fiber properties that occurred with the administration of TPTD. In particular, it is significant that the topological character of bone collagen in conventionally prepared bone sections was visualized and quantified by the application of SHG imaging and a semi-automatic morphology recognition system. Therefore, our approach in this study will highly contribute to pharmaceutical evaluation of anti-osteoporotic drugs, bone biomedicine and the evolutional biology of bone.

Funding

This work was partly supported by Grant-in-Aids for Scientific Research from the Japan Society for the Promotion of Science (JSPS KAKENHI) grant numbers 18H02983 and 18K19649 to T.I. J.W.L. was supported by Grant-in-Aids for Scientific Research from the Japan Society for the Promotion of Science (19K10044) and a grant from Takeda Science Foundation.

Conflict of interest statement

A.T. and R.T.K. are employees of Asahi Kasei Pharma Corporation. K.T. and H.M. are employees of Nikon Corporation. The remaining authors declare no conflicts of interest in association with the present study.

Supplementary data

Supplementary data are available at Microscopy online.