Osteoclast differentiation and modification
In our study, OCs were induced by bone marrow-derived monocyte–macrophage precursor cells (BMMs) stimulated by receptor activator of nuclear factor-κB ligand (RANKL, 50 µg/l) and macrophage colony-stimulating factor (M-CSF, 25 µg/l);38 OCs obtained in vitro were characterized by tartrate-resistant acid phosphatase (TRAP) staining (Fig. 1b), and their decalcification capacity (Supplementary Fig. 1) and bone-targeting ability were confirmed. As expected, their adhesion ability within ectopic calcified tissue and soft tissues was poor (Supplementary Fig. 2), which was also confirmed by previous study5,6. Our results showed that OCs cultured in the presence of TC were characterized by TC cocoons, these cells were named TC-engineered osteoclasts (TC-OCs) in this study. We determined an optimized strategy for treating 2 × 105 OCs/ml with 160 µg/ml of TC, yielding a coating efficiency of approximately 90.7 ± 2.0% and a cell viability of 89.0 ± 4.9% (Fig. 1c–e and Supplementary Fig. 3). As shown in the Fourier transform-infrared (FTIR) spectrum, the intensities of the characteristic peaks of C–N bonds (approximately 1260 cm−1), carbonyl-stretching vibrations (approximately 1680 cm−1) and N–H-stretching vibrations (approximately 3000 cm−1) increased after modification, indicating that the conjugation of amino groups on TC and carboxyl groups on the cell membrane formed more amide bonds (Fig. 1f). Furthermore, the surface-modification process was also detected by Raman spectroscopy, and the performance of the TC peak (1639 cm−1) indicated the successful modification of the cell surface (Supplementary Fig. 4a).
Under a confocal laser-scanning microscope (CLSM), a coated TC (green) layer was tightly associated with each cell membrane (red, labeled by PKH26); naive OCs had no such green layer (Fig. 1g). Multiple nuclei (blue, labeled by Hoechst 33258) were observed in encapsulated cells, confirming the successful chemical modification of TC on the OCs, and the fluorescence intensity of OCs and TC-OCs as determined by fluorescence spectrophotometry was consistent with these results (Supplementary Fig. 4b). The results showed that the TC cocoons were stable on the cells over four days (Fig. 1g and Supplementary Fig. 4c). A quantitative estimation indicated that each cell was modified with approximately 1.35 × 1011 TC molecules (Supplementary Fig. 4d) (see the supplementary file for the detailed calculation process).
The ruffled membranes on the OC cell surface form a specialized region that acidifies calcium minerals in a resorption space39,40. Accordingly, we assessed whether this primary characteristic of OCs was affected by TC modification. As shown by scanning electron microscopy (SEM), no differences were observed between the OCs and TC-OCs in the acid-etched areas on bone tissues (Supplementary Fig. 5a, b). Furthermore, examination by inductively coupled plasma optical emission spectrometry (ICP-OES) showed that the amounts of Ca2+ released from the acid-etched areas into the culture medium in the two groups were the same (Supplementary Fig. 5c). The OCs/TC-OC resorption site on the cortical bone surface was quantitatively and qualitatively confirmed by transmission electron microscopy (TEM) and atomic force microscopy (AFM) (Fig. 1h, i and Supplementary Fig. 5d and Supplementary Fig. 6). The depth of the pit was approximately 20 µm, which confirmed that cells had excellent bone-resorption capacity. These results indicated that the TC-OCs and OCs shared the same calcium-mineral resorption capacity, confirming that the fundamental function of OCs remained after TC modification.
Functional verification of TC-OCs
The Transwell migration assay is a common technique for studying cell migration behavior in vitro41. OCs and TC-OCs (3 × 104/well) were seeded onto permeable filter inserts, which were placed in the wells of culture plates and in direct contact with medium containing ectopic calcified tissue (Fig. 2a, b). Within a period of 12 h, the migration ability of TC-OCs onto the tissues in the lower chamber increased by more than threefold compared with that of naive OCs (numbers: OCs: 80 ± 10; TC-OCs: 355 ± 21) (Fig. 2c–g). The results showed that the cell-migration capacity improved as the TC concentration increased (Supplementary Fig. 7). This improved targeting effect to the ectopic calcified sites was due to the surrounding TC cocoons.
a, b Schematic depicting the coculture assay with a conventional Transwell design for the detection of OC/TC-OC migration to ectopic calcified tissue. c, d Typical images showing migrated OCs/TC-OCs after 12 h of incubation (n = 3 independent samples/group). Bar, 200 µm. e, f Enlargement of c, d. Bar, 100 µm. g Quantitative analysis of migratory OCs/TC-OCs that were manually analyzed using Image J (n = 3 independent samples/group, ****p < 0.0001). h, i Representative images of OCs/TC-OCs with probes attached to calcified tissues (n = 5 independent samples/group). Bar, 100 µm. j, k Representative CLSM images of the probe after treatment with OCs/TC-OCs (n = 3 independent samples/group, tetracycline (green), nucleus (Hoechst 33258, blue)). Scale bar, 20 µm. l A representative force curve of OCs/TC-OCs with calcified tissue. m Quantitative analysis of the force between OC/TC-OC cell membranes and probes attached to calcified tissue. Force curves were calculated from the frequency-shift difference curves using the Sader–Jarvis method (n = 5 independent samples/group, ***p = 0.0006). n, o Schematic of the test design for the detection of OC/TC-OC migration to ectopic calcified tissue. Live/dead staining images of OCs/TC-OCs migrating to calcified tissue. Bar, 300 µm. p Quantitative analysis of migratory OCs/TC-OCs (n = 3 independent samples/group, **p = 0.0019). Data are represented as mean ± SD. Statistical comparisons were made using either unpaired (g, p) or paired (m) t-tests. Source data are provided as a Source Data file (**p < 0.01, ***p < 0.001, ****p < 0.0001).
In addition, the enhanced attractive forces between the TC-OCs and calcified tissue were quantitatively confirmed by AFM. OCs/TC-OCs were separately seeded onto glass plates and cultured with α-MEM (10% FBS) supplemented with a stimulus (25 µg/l M-CSF and 50 µg/l RANKL) for 24 h. Nanomanipulation of calcified tissue-modified AFM probes (Supplementary Fig. 8) on OCs or TC-OCs was performed by writing nanolithography scripts to control the movement of the probe. As expected, the naive OCs could not firmly adhere to the probe owing to the poor adhesion force of only 0.05 ± 0.01 nN (Fig. 2h, j and Supplementary Movie 1). In contrast, the TC-engineered OCs firmly adhered to the probe without detachment (Fig. 2i, k and Supplementary Movie 2), as the adhesion force to the calcified tissue probe was increased by approximately twofold compared with that of the control (0.14 ± 0.019 nN) (Fig. 2l, m), suggesting an enhanced adhesion force of TC-OCs to calcified soft tissues. Moreover, CLSM results further confirmed that the cells attached to the tipless cantilever were multinucleated (Fig. 2j, k). Finally, we performed additional experiments to measure the cell-migration capability and viability using live/dead probes (Fig. 2n, o). As depicted in Fig. 2p, approximately 67 ± 3 cells migrated into calcified tissue in the TC-OC group, but only 30 ± 9 cells migrated into calcified tissue in the native-cell group. Live/dead staining results suggested that compared with naive OCs, TC-OCs had increased adhesion to calcified tissue. This increase in the adhesion force confirmed the efficient targeting effect of TC-OCs to calcified soft tissues, indicating their potential for the biological decalcification of HO.
Ectopic calcification resorption by engineered OCs (in vitro)
Calcium-mineral resorption is achieved by the release of protons from OCs42. Accordingly, the ability of TC-OCs to resorb calcified tissue can be analyzed using a pH-sensing chemical probe such as an AIETM pH probe, which changes color from red to blue with increasing pH43. CLSM results showed that red signals were detected and multinucleated TC-OCs targeted to ectopic calcified tissue (Fig. 3a). To further prove the cells migrated to calcified tissue were mature osteoclasts, the calcified tissues were stained with TRAP, as the standard for osteoclast identification. The results further manifested that the cells that migrated to calcified tissue were TRAP-positive cells (see Supplementary Fig. 9a, b). More cells were observed on ectopic calcified tissue compared with no modification, indicating that cell adhesion and migration capacity was improved due to surface engineering (Supplementary Fig. 9c, d). Importantly, the fluorescence intensity of the red signal in the TC-OC group was substantially improved (approximately doubled) compared with that observed in the OC group, which indicated that engineered OCs have an increased ability to release acid relative to naive OCs (Fig. 3b). The overall appearance of the native OCs and engineered OCs (pink, Fig. 3c) that migrated into ectopic calcified tissue was analyzed by SEM. Consistent with the above results, more TC-engineered cells were observed on ectopic calcified tissue. Furthermore, an analysis of the acid-etched area by SEM further showed that the ability of TC-OCs to resorb calcium in the acid-etched ectopic calcified tissue areas was increased to approximately 2.5-fold (OCs: 21.9 ± 9.4%; TC-OCs: 55.7 ± 3.4%) (Fig. 3d–f).
a CLSM images of OCs/TC-OCs seeded into 48-well plates containing ectopic calcifications and incubated with an AIETM pH probe. Scale bar: 20 µm. b Quantitative intensities of OCs/TC-OCs labeled with the AIETM pH probe (n = 4 independent samples/group, **p = 0.003). c SEM images of OCs and TC-OCs migrating into calcified tissue in vitro (n = 4 independent samples/group). Scale bar: 30 µm. d SEM images of calcified tissue resorption area (n = 4 independent samples/group). Scale bar: 300 µm. e Enlarge of (d). Scale bar: 100 µm. f Statistical analysis shows the changes in the resorption area between the two groups (n = 4 independent samples/group, ***p = 0.0005). g Three-dimensional reconstruction of micro-CT scans of ectopic calcified tissue treated with 0.9% NaCl (blank), OCs, or TC-OCs at 0 and 2 weeks (n = 3 independent samples/group). Scale bar: 2 mm. h Resorption bone-volume (BV) analysis showing the changes in resorption areas among the three groups (n = 3 independent samples/group, ****p < 0.0001). i Ca2+ eluted from the acid-etched areas in three groups, as measured by ICP-OES (n = 3 independent samples/group, **p = 0.0015). Data are represented as mean ± SD. Statistical comparisons were made by using unpaired (b, f) t-tests and ordinary one-way analysis of variance (ANOVA) with multiple-comparison tests (h, i). Source data are provided as a Source Data file (**p < 0.01, ***p < 0.001, ****p < 0.0001).
Microcomputed tomography (micro-CT) reconstruction was performed on ectopic calcified tissue samples to assess decalcification at the histological level (Fig. 3g). The volume of calcification at 0 and 2 weeks was quantitatively estimated by micro-CT. The percentage of bone-volume reduction was used to evaluate the efficiency of bone resorption by engineered osteoclasts. The total ectopic calcification volumes in the OC and TC-OC groups were reduced by 33.7 ± 4.6% and 67.0 ± 1.6%, respectively (Fig. 3h), compared with that observed in the control group (0.7 ± 3.7%). Because of the increased cell migration, the amount of Ca2+ eluted from the acid-etched areas estimated by ICP-OES was used to compare the bone-resorption capability between the native OCs and the modified cells. The amount of Ca2+ eluted from the acid-etched areas into the culture medium in modified OC groups was increased by about three times (Fig. 3i) compared with native OC groups, verifying an improvement in ectopic calcified tissue resorption capacity after surface modification. All these in vitro results provide direct evidence of the advantages of TC-OCs in biological decalcification.
Reversal of heterotopic ossification by engineered OCs in vivo
Given the enhanced ectopic calcified tissue resorption capacity of TC-OCs, three separate models of HO were studied to observe the function of TC-OCs in vivo: (i) tenotomy, (ii) intramuscular model, and (iii) genetic model (Fig. 4a and Supplementary Fig. 10)44. The detailed methods of three separate HO models were provided in the supplementary files. Notably, the engineered OCs with TC molecules on the cell surface can be stored stably in vivo over four days (Supplementary Fig. 11). As depicted in Fig. 4b, we injected 0.9% NaCl (100 µl, normal saline, blank), OCs (100 µl, 106 cells/ml/500 g, control), or TC-OCs (100 µl, 106 cells/ml/500 g) in situ at the Achilles tendon calcification site every four days. Localization to the calcified site was guided by living micro-CT and accomplished with an in situ injection. In the tenotomy model, HO maintained growth in the blank group but showed a volume reduction in the control (OC) group and the TC-OC group based on two- and three-dimensional micro-CT at 30 days after injection (Fig. 4b and Supplementary Fig. 12). Consistent with the micro-CT images, tenotomized rats treated with TC-OCs exhibited reduced HO volumes at 30 days (by approximately 31.8 ± 10.2%) compared with the initial volume (at 0 day) (Fig. 4c). However, in the OC groups, HO volumes were reduced at 30 days (by approximately 18.6 ± 11.2%) compared to the initial volume (at 0 day). It should be noted that there was an increase in bone mass in the blank groups (−19.6 ± 19.1%). TC-OC treatment resulted in the absorption of most of the ectopic bone after 60 days, as shown by analysis and quantitative comparison (Fig. 4c, d). Quantification of the relative BV showed that ectopic bone formation was reduced by more than 70.5 ± 7.6%, compared with the 33.3 ± 7.9% observed in the OC group after 60 days, which indicates that engineered OCs exhibit superior ability to resorb HO. We next confirmed these findings in the intramuscular model, which demonstrated a reduction in total HO volume at 30 days (Fig. 4e, f). In addition, we noted that soft tissue HO was nearly completely abolished in TC-OC treated group at 60 days, consistent with our findings in tenotomy model.
a Three different types of HO animal models: tenotomy model, intramuscular model, and genetic model. b Timeline of the injected OCs/TC-OCs. c, d Three-dimensional reconstructions of micro-CT scans and bone resorption of tenotomy model rats treated with OCs, TC-OCs or 0.9% NaCl (blank) (for 0, 30, or 60 days). Large mineralized HO masses were visible by micro-CT in rats (n = 5 animals per group). Scale bar: 3 mm. Resorption ratio of day 30 **p = 0.0028, ***p = 0.0002. Resorption ratio of day 60 ****p < 0.0001, ****p < 0.0001. e, f Micro-CT scans and bone-resorption ratio of intramuscular model (n = 4 animals). Scale bar: 3 mm. Resorption ratio of day 30 **p = 0.0012, ***p = 0.0003. Resorption ratio of day 60 ***p = 0.0004, ****p < 0.0001. g, h Micro-CT scans of genetic model (n = 5 animals per group). Scale bar: 2 mm. Resorption of genetic model HO was evaluated by measuring the BV resorption ratio of ectopic masses in the blank versus the OCs and TC-OC groups on days 0, 30, and 60. Resorption ratio of day 30: **p = 0.0015, ****p < 0.0001; Resorption ratio of day 60: ****p < 0.0001, ****p < 0.0001. i Ectopic tissues were sectioned and examined by VK, H&E, AB, Masson, and TRAP staining; bar: 500 µm. Data are represented as mean ± SD, all statistical comparisons were made by using an ordinary one-way analysis of variance (ANOVA) with multiple-comparison tests. Source data are provided as a Source Data file (**p < 0.01; ***p < 0.001; ****p < 0.0001).
To strengthen our findings, we next used Mkx−/− knockout mouse that formed HO after being born at eight weeks as genetic model. Again, our concepts were proved using TC-OCs in treated mice, which showed less ectopic bone at 30 days, as shown by micro-CT and quantitative comparison (Fig. 4g, h and Supplementary Fig. 12). After 60 days, TC-OC group developed minimal HO around the calcaneus, and these lesions were substantially smaller than in OCs and blank groups. Taken together, the results were striking and proved the substantially improved efficacy of TC-OCs to reversibly resorb ectopic bone over naive OCs in a different type of HO model.
Bone-resorption capacity of engineered TC-OCs in other calcified tissue, cells were further proved in endochondral bone (cranial bone) and intramembranous bone (tibia bone) in vitro. As depicted in Supplementary Fig. 13 and Supplementary Fig. 14, BMD, BS/TV, and BV/TV were decreased in TC-OC groups compared with OC-treated groups, confirming that TC-engineered OCs promote excellent ability of bone resorption in endochondral and intramembranous bone. These data support the notion that surface-modified osteoclasts have the potential not only for HO-reversing treatment but also for the treatment of other types of ectopic calcification.
Pathological analyses of tendon slices from each group included Von Kossa (VK), hematoxylin and eosin (H&E) staining, Masson’s trichrome (Masson), TRAP staining, and Alcian blue(AB) staining. The VK and H&E staining results suggested that ectopic tissues were prominent in the blank and control groups, but clearly reduced in rats treated with TC-OCs (Fig. 4i). Consistent with these results, the Masson and AB staining results showed that blank and control rats exhibited abundant endochondral bone and cartilage and that oriented collagen fibers were damaged. Furthermore, the ectopic tissues in rats were efficiently diminished and treated with TC-OCs. TRAP staining images and quantitative statistics demonstrated that more TRAP-positive cells were maintained and fewer ectopic calcifications were retained in TC-OC-treated rats than in the other rats. These pathological results demonstrated the effective and precise bone resorption of TC-OCs in the setting of HO.
To further examine the ability of TC-OCs to target and adhere to ectopic calcification in vivo compared with OCs, cells were in situ injected into the Achilles tendons of the model rats44, and these tendons were dissected after 0 h and 2 h. For the specific method of tendon injury used to induce HO, refer to the supplementary file. As shown in Supplementary Fig. 15a, b, and c at high spatial distribution, deep tissues were visualized at various depths with a view up to 100-µm deep by two-photon CLSM. The location and shape of the ectopic calcifications and the density of OCs/TC-OCs in each Achilles tendon were clearly detected under two-photon CLSM at 0 h in the three groups. CLSM images of the OCs/TC-OC distribution in the calcification in tendon were captured at 2 h after injection (Supplementary Fig. 15d, e and f). The number of TC-OCs at the calcification sites was increased compared with that in the OC group (Supplementary Fig. 15g), implying that the engineered OCs have an adhesion and calcification-targeting ability superior to that of naive OCs. Importantly, our results demonstrated that the injected TC-OCs around the HO site were bioactive during the treatment. In summary, in an Achilles’ tenotomy rat model, the injection of TC-OCs around the calcified tendons in situ enabled the visualization of the local sites of calcified mineral resorption following reversible recovery from HO.
To qualitatively and quantitatively evaluate the cell viability of osteoclasts and TC-engineered osteoclasts in vivo, OCs with or without surface modification were used to perform in situ injections for an extended time period. We visualized the cell viability of native and engineered OCs labeled by Cell Trace Far Red DDAO-SE fluorescent tag in vivo based on the in vivo living imaging at 0 day and 4 days (Supplementary Fig. 16). The results showed that labeled TC-OCs showed a stronger staining intensity than OCs. The long-term in vivo circulation of the engineered OCs in ectopic tissue demonstrated that the TC-OCs retain good cell viability and function for more than four days. Moreover, we quantified the cell viability of native and engineered OCs in vivo by flow cytometry at 0 day and four days after injection. Flow cytometry further revealed increased fluorescence signals of Calcein AM in the TC-OC groups (Supplementary Fig. 17a, b, c), and the values were improved from 27.6 ± 7.7% in OC groups to 76.6 ± 10.0% in TC-OC groups. According to the improvement of cell viability with the modification of the cell membrane (Supplementary Fig. 17d), we deduced that the TC-engineered OCs increase their unique function of bone resorption. The results demonstrated that surface engineering is essential for natural OCs to maintain the key functions of bone resorption in vivo for long-term circulation. Long-term circulation in vivo further demonstrated the favorable bone-resorption activity of TC-engineered OCs and the feasibility of HO treatment.
The biocompatibility and activity of the engineered OCs in vivo was further proven by testing cytokine response in tendon lysis. The cytokine response of bone formation and resorption in the area of the cell implants was measured by ELISA (alkaline phosphatase (ALP), bone morphogenetic protein 2 (BMP-2), TRAP, and type-I collagen cross-linked C-telopeptides (CTX)). Although the expression of proteins associated with bone formation (ALP and BMP-2) in the area of the cell implant is lower than that in the control groups (Supplementary Fig. 18), proteins related to bone resorption (CTX and TRAP) were more highly expressed in the TC-OC groups. These results indicated that surface engineering leads to slight reduction of bone-formation marker and presents good biocompatibility and bone-resorption activity. The expression of proteins associated with bone resorption had improved, thus confirming the conceptual potential of this approach to HO-reversing treatment.
At the protein level, the key factors involved in cell differentiation to the HO osteogenic lineage are bone morphogenetic proteins (BMPs) and inflammatory factors45. Blood-based detection revealed no changes in the levels of ALP, BMP-2, interleukin 6 (IL-6), or tumor necrosis factor alpha (TNF-α) in the TC-OC group compared with the control group (0.9% NaCl), indicating that no inflammatory effect and side effects were induced by the “living treatment agents” (Supplementary Fig. 19).
Bone resorption by osteoclasts is normally coupled to bone formation by osteoblasts46. TRAP and CTX were used as standard biomarkers for bone resorption. The serum levels of TRAP and CTX were normal (Supplementary Fig. 20), suggesting that surface engineering did not lead to side effect. Furthermore, no obvious changes were observed in blood chemistry or bone density (Supplementary Fig. 21 and Supplementary Fig. 22) or in trabecular morphology, confirming the biocompatibility nature of TC-OCs in the treatment of HO.
There may be concerns regarding the biosecurity of TC. TC has been widely used as an antibiotic and for bone-targeting purposes36,47. In our experiment, 100 µl of TC-OC solution (160 µg/ml) was used in each injection, and the total amount of TC was only 2 µg, i.e., far less than the threshold levels (100 mg/kg) set by the Steering Committee of the Veterinary International Committee on Harmonization.
Another potential concern regards the phenomenon referred to as the “rebound effect” often observed in HO treatment11. Ectopic calcification tissues grew in control rats; however, only minimal tissue growth was observed in rats treated with TC-OCs, indicating the lack of a rebound effect (Supplementary Fig. 23).
