Committed osteoclast precursors colonize the bone and improve the phenotype of a mouse model of autosomal recessive osteopetrosis

Abstract

Osteopetrosis is a genetic disease characterized by defective osteoclasts. Autosomal recessive osteopetrosis is fatal within the first years of life. Hematopoietic stem cell transplantation (HSCT) cures fewer than 50% of cases but often leaves severe neurologic damages and other dysfunctions. Osteoclast appearance after HSCT is a slow process, during which disease progression continues. We hypothesize that a support osteoclast precursor therapy may contribute to improve the osteopetrotic phenotype. To this end, we established a procedure to obtain the best yield of osteoclast precursors from human peripheral blood or mouse bone marrow mononuclear cells. These cells were injected in vivo in animal models, testing different cell injection protocols, as well as in association with CD117+ stem cells. Injected cells showed the ability to form multinucleated osteoclasts and to improve the phenotype of oc/oc osteopetrotic mice. In the best working protocol, animals presented with longer survival, improved weight and longitudinal growth, increased tibial length, tooth eruption, decreased bone volume, reduced bone marrow fibrosis, and improved hematopoiesis compared with sham-treated mice. These results provide first-hand information on the feasibility of a support osteoclast precursor therapy in osteopetrosis. © 2010 American Society for Bone and Mineral Research

Introduction

Bone health is guaranteed by the continuous and coordinated action of bone-depositing cells, the osteoblasts, and bone-resorbing cells, the osteoclasts.1 When this balanced activity is disrupted, bone pathologies appear, among which is osteopetrosis.2 Osteopetrosis is a heterogeneous family of rare genetic bone diseases caused by defective osteoclast function.3–8 The most severe form (incidence 1:250.000) is autosomal recessive osteopetrosis (ARO), leading to fatal outcome within the first 2 to 3 years of life.9–11 ARO is characterized by an extensive sclerosis of the whole skeleton and reduced medullar cavities. This sclerosis leads to fragility and ease to fracture, as well as to failure of bone marrow hematopoiesis with consequent hepatosplenomegaly, pancytopenia, infections, and neurologic damage such as blindness and deafness, generally owing to nerve compression syndromes.9–11 ARO is treated clinically with certain drugs without clear-cut efficacy.9 So far, the only useful treatment is hematopoietic stem cell transplantation (HSCT),12 which, however, is ineffective in some forms of osteopetrosis and cannot cure previous irreversible damage.13 Furthermore, patients who survive transplant complications often keep worsening, especially in terms of neurologic failures.13, 14 The variability of HSCT outcome is not well understood, but it is probably complicated by a latency period for osteoclast appearance, during which hematologic syndromes are restored, but the bone phenotype and consequent neurologic symptoms continue to progress.13, 14 To circumvent this inconvenience, our hypothesis is that inoculation of osteoclast progenitors may give patient cells readily resorbing bone. Therefore, we examined for feasibility of a cell-support approach that may open the way for a new protocol for the treatment of severe forms of ARO.

Materials and Methods

Cell cultures

Human adult peripheral blood was obtained from healthy donors with their informed consent. Mouse bone marrow mononuclear cells were flushed out from long bones of 15- to 20-day-old mice. The mononuclear cell fraction was isolated from both sources by centrifugation on Histopaque 1077 (Sigma-Aldrich) and cultured for 2 weeks in Dulbecco's modified Minimal Essential Medium (DMEM) or in StemPro34 medium (Invitrogen) containing 10% fetal bovine serum (FBS), 50 U/mL penicillin, 50 ng/mL streptomycin, 2 mM L-glutamine, 25 ng/mL macrophage colony-stimulating factor (M-CSF), and 30 ng/mL soluble receptor activator of NF-κB ligand (RANKL) (Fig. 1, standard treatment). Sets of cultures were pretreated for 1 week with 20 ng/mL granulocyte-macrophage (GM)–CSF (Fig. 1, single pretreatment), preceded or not by treatment for 1 week with 50 ng/mL stem cell factor (SCF), 20 ng/mL interleukin 3 (IL-3), and 20 ng/mL IL-6 (Fig. 1, double pretreatment). All cytokines were recombinant human (Peprotech). Cultures were refreshed twice weekly. At the end of treatment, cells were fixed in 4% paraformaldehyde and stained histochemically for the osteoclast-specific marker tartrate resistant acid phosphatase (TRAcP). Osteoclast formation was assessed by light microscopy, whereas fluorescence microscopy was used to visualize nuclei stained with Hoechst 3342. Cells also were cultured on bone slices for a further week and then removed by sonication prior to metachromatic staining of the slices with 1% toluidine blue. The formation of resorption pits was quantified by the pit index method.15 Cells were cryopreserved in FBS + dimethylsulfoxide (9:1) after treatment for 1 day with M-CSF/RANKL.

Figure 1. In vitro osteoclastogenesis protocols. The diagram shows the three culture protocols used in the study. Standard M-CSF/RANKL treatment was shared by all protocols. Single and double pretreatments with the indicated cytokines were tested both in DMEM and in StemPro34 media. Cytokine concentrations are in the Materials and Methods section.

Immunomagnetic sorting

Immunomagnetic cell sorting was performed with midi-MACS kit (Miltenyi Biotec) to isolate CD117 + cells.

Animals

Balb-c nu/nu mice were purchased from Charles River. oc heterozygous mice, kindly donated by Dr. Uwe Kornak (Charité University Hospital, Berlin, Germany), were mated in our animal facility, and newborn mice were genotyped. Animal procedures conformed to the Italian regulation for animal care (DL 116/92) and were approved by the internal ethical board of our institution. Before any invasive manipulation, animals were anesthetized with an ip injection of a mixture of ketamine (25 mg/mL) and xylazine (5 mg/mL).

Balb-c nu/nu mice were injected once with vehicle (PBS) or with 3 × 10⁵ human peripheral blood mononuclear cells. Injections were performed in the left ventricle through the second left intercostal space. Mice were sacrificed after 1, 3, 4, 5, and 6 weeks from injections. Groups of mice also were treated with human recombinant parathyroid hormone 1–34 (hrPTH), 80 µg/kg of body weight, injected ip daily for 5 days/week until sacrifice.

Newborn oc/oc mice were injected with PBS or with the cells every 3 days until sacrifice, starting on the second day of life. Injections were performed intrahepatically for the first 8 days of life and then intraperitoneally. Numbers and types of cells injected are listed in Table 1.

Table 1. In Vivo Treatment Protocols of oc/oc Mice

Treatment	No. of injected cells/mice	No. of treated mice	Outcomea
a Arbitrary evaluation based on growth rate, survival, and tooth eruption (+ to ++++: poor to good).
Phosphate-buffered saline (PBS)	None	12	–
Osteoclast precursors	60,000	4	+
Osteoclast precursors	90,000	4	+
Osteoclast precursors	120,000	3	+
Osteoclast precursors + CD117+	300,000 (20:1)	4	++
Osteoclast precursors + CD117+	500,000 (20:1)	3	++
Osteoclast precursors + CD117+	500,000 (10:1)	3	++
Osteoclast precursors + CD117+	120,000/gr (20:1)	4	+++
Osteoclast precursors + CD117+	120,000/g (20:1) (from birth to 12 days of age) 240,000/g (20:1) (from 13 days of age to sacrifice)	3	++++
CD117+	6000/g (from birth to 12 days of age) 12,000/g (from 13 days of age to sacrifice)	3	+

Polymerase chain reaction (PCR)

DNA was extracted from whole-limb long bones of sham- and human cell–injected Balb-c nu/nu mice, and detection of human ALU sequences was performed by PCR (45 cycles, 60.7°C) using the primer pairs (forward) 5'-CGAGGCGGGTGGATCATGAGGT-3' and (reverse) 5'-TCTGTCGCCCAGGCCGGACT-3'. Positive control was DNA extracted from the human cells prior to injection.

Histology, in situ hybridization, and immunohistochemistry

At sacrifice, tibias were excided, cleaned from soft tissues, fixed, decalcified, and embedded in paraffin by standard procedure. Sections were subjected to hematoxylin and eosin (H&E) and trichrome Masson's staining or to histochemical TRAcP detection to evidence osteoclasts. In situ hybridization was performed with an In Situ Hybridization Detection Kit (Maxim Biotech, Inc.). Immunohistochemistry was performed to detect erythrocytes with an anti-Ter119 antibody (Biolegend). Histomorphometric analysis was performed with the IAS2000 software (Delta Sistemi). Images were captured by a Zeiss AXIOSKOP 2 PLUS microscope connected with a AXIOCAM HRc and acquired with AXIOVISION 3.06.38 software in .JPG files. Merge pictures were obtained by the Adobe Photoshop CS 8.0.1 software using overlapping function on the .JPG files acquired.

Results

Osteoclast yield and resorption ability

Peripheral blood or bone marrow osteoclast yield in standard culture conditions is insufficient for therapeutic applications. Therefore, in order to improve the yield and the bone-resorbing performance of osteoclasts, we assessed the osteoclastogenic potential of mononuclear progenitors in various culture settings using human cells, which, we believe, have a high translational relevance for future applications in patients. Then we transferred the best working protocol to mouse cells to experimentally test their therapeutic potential in a mouse model of ARO.

To this end, we treated human peripheral blood mononuclear cells for 2 weeks with standard human recombinant osteoclastogenic cytokines M-CSF and RANKL16, 17 and compared the osteoclast yield with the very same treatment preceded by a single pretreatment for 1 week with GM-CSF or a double pretreatment for 1 week with SCF, IL-3, and IL-6 followed by 1 week with GM-CSF before standard osteoclastogenic induction with M-CSF and RANKL (see Fig. 1). Moreover, we tested the effect of StemPro34, a culture medium alternative to DMEM, developed to support the growth of human HSC. With our conditions, the best working osteoclastogenic protocol was the double pretreatment in StemPro34 medium, which resulted in the appearance of TRAcP-positive osteoclast formation after 3 days of exposure to M-CSF/RANKL, remarkably earlier than in the other cultures (Fig. 2A). Using the double-pretreatment protocol, after 1 week of exposure to M-CSF and RANKL, the rate of osteoclastogenesis was higher than for other cultures (not shown), improving further after 2 weeks (see Fig. 2B). The double pretreatment, along with the use of StemPro34, not only improved the yield of multinucleated cells versus standard conditions (see Fig. 2B, i) but also increased the number of nuclei per osteoclast (see Fig. 2B, ii) and the resorption ability, with a pit index about fourfold greater than that of cells cultured in DMEM with the standard treatment (see Fig. 2B, iii). Similar results were obtained using mouse bone marrow primary cell cultures (not shown).

Figure 2. Human osteoclast cultures. (A) 10⁶ peripheral blood mononuclear cells were seeded in 96-well plates in (i) DMEM or (ii–iv) StemPro34 and treated with recombinant human cytokines as indicated. At the end of incubation, cells were fixed and stained for TRAcP activity. M-CSF and RANKL were administered to all cultures for 2 weeks (i, ii) alone or (iii, iv) after administration of the indicated cytokines. Concentrations and times of treatments are in the text. In (iv), TRAcP-positive osteoclasts were already observed after 3 days of treatment with M-CSF/RANKL (inset). Original magnification × 20. (B) Osteoclast cultures were stained for TRAcP activity and subsequently with Hoechst 33342 reagent to evidence the nuclei. (i) TRAcP-positive cells were then counted along with (ii) the number of nuclei per osteoclast. (iii) Cultures performed on bone slices were continued for a further week. After sonication of bone slices and staining with toluidine blue, bone resorption was measured by the pit index method.15 (C) Cells were frozen at –80°C in FBS + dimethylsulfoxide (9:1) after double pretreatment with cytokines followed by 1 day of treatment with M-CSF/RANKL to trigger osteoclastogenic (prefusion) commitment. Evaluation of their (i) osteoclastogenic potential and (ii) bone resorption after thawing showed no difference compared with freshly isolated cells in cultures in which equal numbers of viable cells were plated. y axes represent the mean ± SD of at least four independent cell preparations. y-axis legends are on top of each graph. ^ap < .01 versus DMEM + M-CSF/RANKL; ^bp < .03 versus StemPro34 + M-CSF/RANKL.

For translational purposes, the source of osteoclasts must be readily available, and cryopreservation procedures are used to maintain viable cells until use. We therefore tested osteoclast commitment ability after various methods of osteoclast precursor cryopreservation. The best working protocol was preservation at –80°C in FBS:dimethylsulfoxide (9:1). This condition guaranteed about 65% survival after thawing, assessed by the trypan blue exclusion test and by cytofluorimetric analysis of propidium iodine–positive cells. Osteoclastogenesis and bone resorption potential were indistinguishable from those of freshly isolated cells (see Fig. 2C).

Engraftment of human osteoclast precursors in immunecompromised mice

Next, we assessed the in vivo impact of our cells in animal models, first testing engraftment of human cells in 4-week-old immunocompromised Balb-c nu/nu mice,18 which could have a significant translational relevance. Mice were injected once in the left ventricle with 3 × 10⁵ human peripheral blood mononuclear cells precultured in StemPro34 with the double pretreatment with cytokines plus M-CSF and RANKL for 1 day to trigger osteoclastogenic pathways. Animals were sacrificed at various times after cell injection, as indicated in Fig. 3A, and assessed for human cell engraftment by PCR for the human ALU sequence on extracted DNA from whole-limb long bones. We noted ALU PCR amplification from the first to the fourth week from cell inoculation, with no further amplification afterward, indicating transient engraftment of human cells (see Fig. 3A). In situ hybridization in the tibias showed TRAcP-positive multinucleated cells containing ALU-positive nuclei (see Fig. 3B), which were estimated to be 0.82% ± 0.052% of total osteoclast count. This percent is likely to be underestimated because ALU-positive osteoclasts were identified in 60% of ALU-positive mice as assessed by PCR, suggesting that in our hands this latter method is more sensitive than in situ hybridization. Notably, ALU-negative polymorphic nuclei also were apparent inside the osteoclasts carrying ALU-positive nuclei. They were morphologically distinguishable from the latter and similar to ALU-negative nuclei of surrounding cells (see Fig. 3B, iv) presumably belonging to the host. This observation might suggest fusion of injected ALU-positive human precursors with resident mouse ALU-negative osteoclastic cells. Groups of mice treated with 80 µg/kg hrPTH(1–34) showed no further improvement of osteoclast engraftment (not shown).

Figure 3. Assessment of engraftment of human cells in Balb/c nu/nu mice. Four-week-old Balb/c nu/nu mice were deeply anesthetized and injected once in the left ventricle with 3 × 10⁵ human osteoclast precursors per mouse or with PBS as control. (A) The bone marrow of right-side tibias was flushed off at the indicated times, and the cells recovered were subjected to DNA extraction and detection of human ALU sequences by PCR, as described in the Materials and Methods section. Positive control was DNA extracted from the human cells prior to injection. (B) Left-side tibias, cleaned from soft tissues, were fixed in 4% paraformaldehyde and processed for paraffin embedding and in situ hybridization to evidence human ALU sequences in bone marrow cells. Sections 7 µm thick, obtained from a mouse sacrificed at the third week from cell injection, were processed for TRAcP staining and then assessed for (i) ALU positivity (arrows), followed by (ii) nuclear staining with Hoechst 3342. (iii) Merge picture shows coincidence of ALU positivity with the nuclei (large arrows). Small arrows indicate ALU-negative polymorphic nuclei inside the osteoclast (delimited by the white line). Host mononuclear cells containing ALU-negative polymorphic nuclei are also visible outside the osteoclast. Original magnification ×63. (iv) Negative control: osteoclast of a tibia from a PBS-injected mouse processed by in situ hybridization for human ALU.

Experimental cell therapy in oc/oc osteopetrotic mice

To evaluate any beneficial effect on the disease, we injected newborn osteopetrotic oc/oc mice (intrahepatically for the first 8 days of life and then intraperitoneally)19 with mouse bone marrow mononuclear cells isolated from syngeneic wild-type littermates and subjected to the very same treatment as human cells, testing different cell injection protocols, also in association with untreated, immunomagnetically isolated CD117+ stem cells (see Table 1). Treatments were started on the second day of life and repeated every 3 days for the lifespan of the animals. We then monitored disease progression. Results from the best working protocol (see Table 1, last row) are shown in Fig. 4, in which improvement of survival, increase of weight and longitudinal growth (see Fig. 4A), and tooth eruption (see Fig. 4B) are represented.

Figure 4. Treatment of osteopetrotic mice. Genetically identified oc/oc mice were injected with PBS (control) or with the cells, as indicated in Table 1, last row. Injections were started on the second day after birth and were repeated every 3 days for the lifespan of the animals. Animals were monitored for (A) biometrics parameters, including (i) survival, (ii) weight, (iii) longitudinal growth, (iv) tibial length (mean ± SD, n = 3/group, ^ap < .03), and (B) tooth eruption. y-axis legends are on top of each graph.

Histologic analysis revealed an improved bone phenotype in cell-injected mice. Tibias of cell-treated oc/oc mice were bigger and had more medullar cellularity than those of PBS-treated oc/oc mice (Fig. 5A). They also showed reduced fibrosis (see Fig. 5B) and increased hematopoiesis, assessed by detection of the Ter119 antigen of red blood cells (see Fig. 5C). Quantification of erythropoiesis showed an increased number of Ter119-positive cells in oc/oc mice injected with osteoclast precursors (red blood cells/field, 29.78 ± 4.84) compared with oc/oc mice treated with PBS (red blood cells/field, 16.78 ± 3.34, mean ± SD, n = 3, p < .01). Notably, some polarized multinucleated cells delimiting resorption lacunae were observed only in cell-injected oc/oc mice (see Fig. 5D), indicating that functional osteoclasts presumably arise from donor cells. This phenotypic improvement also was confirmed by bone histomorphometry, which showed a reduction of bone volume/total volume in cell-injected compared with PBS-injected oc/oc mice (see Fig. 5E).

Figure 5. Histologic analysis of paraffin-embedded tibial sections. oc/oc mice were injected with PBS or cells as described in Fig. 4. At their death, which occurred between the 30th and 33rd day, tibias were dissected and embedded in paraffin for histologic examination. (A) H&E staining of 7-µm-thick tibial sections of mice treated with PBS or cells as indicated (original magnification: i, iv, ×2.5; ii, iii, ×20). (B) H&E staining showing (i) more fibrotic, less hematopoietic bone marrow in a PBS-injected mouse compared with a (ii) cell-injected mouse. Original magnification ×40. Note in (ii) two apparently resorbing osteoclasts (arrows). (C) Erythropoiesis is shown by immunohistochemistry for the Ter119 antigen of red blood cells in a (i) PBS-injected mouse compared with a (ii) cell-injected mouse. Original magnification ×20. (D) High magnification (×63) of (i) a nonresorbing osteoclast in an oc/oc mouse treated with PBS and (ii) an apparently polarized resorbing osteoclast delimiting a resorption lacuna (arrow) in an oc/oc mouse treated with the cells (Masson's staining). (iii) An apparently resorbing osteoclast in a cell-injected mouse stained with H&E. Arrows indicate typical areas of resorption. Original magnification ×100. (E) Histomorphometric analysis of tibias showing the bone volume/total volume (BV/TV) parameter (mean ± SD, n = 3, ^ap < .05). y-axis legend is on top of the graph.

Discussion

ARO presents with severe symptoms generally a few weeks after birth. Patients have a miserable quality of life and need frequent blood transfusions to correct the severe hematologic deficiency.9, 10 The disease is currently treated with HSCT, and the age of the patient at transplantation is a critical parameter for outcome, with better results observed in earlier treatments.13 Even when HSCT is successful, skeletal improvement can be radiologically apparent as early as 4 months,14 and data in the literature establishing the exact time of appearance of donor osteoclasts are not available. During this latency period, hematologic syndromes are restored, but the bone phenotype and consequent neurologic symptoms continue to progress.13, 14

This outcome potentially may be improved with a support-cell therapy, and in this study, we examined a novel and more efficient protocol for in vitro preparation of cells capable of originating mature osteoclasts that proved to be suitable for in vivo transient engraftment. These cells also maintain their differentiation and resorbing ability after cryopreservation, a result that provides the basis for creating cell banks in which high numbers of readily available injectable osteoclast precursors may be stocked.

Using human-derived osteoclast precursors, we demonstrated that they can colonize the bone and form mature osteoclasts in vivo. Their engraftment was low perhaps owing to the limited survival of human cells in the mouse background because there are differences between these two species in terms of response to cytokines and other factors. However, this result has a high translational relevance because we could set up an efficient method to obtain human osteoclast precursors suitable for experimental cell therapy.

The oc/oc mice are affected by a disease similar to the most frequent form of human ARO. They harbor a mutation in the ATP6i gene encoding the α3 subunit of the vacuolar proton pump H⁺-ATPase, murine counterpart of the human TCIRG1 gene.³ Our results indicate that the outcome of this murine form of ARO could be mitigated when our wild-type committed osteoclast precursors were supplied. Many symptoms of ARO were improved in animals treated with our best working cell injection protocol. For instance, a feature of the oc/oc phenotype is lack of incisors owing to the fact that the teeth are embedded in the alveolar bone and require bone resorption for their eruption. This feature was used in our study as an immediate clue to the reappearance of bone resorption. Indeed, in most treated oc/oc animals, less primitive and better formed teeth were noted. Moreover, the survival of animals was improved, and weight and longitudinal growth also were better than in untreated mice. The improvement in the bone phenotype and the appearance of erythropoiesis also were confirmed histologically and histomorphometrically. These data suggest that some bone remodeling has occurred in the cell-injected mice, providing evidence of the feasibility of an osteoclast precursor support therapy in osteopetrosis. It is interesting to note that best treatment protocols were those in which 1/20 of CD117+ stem cells also were injected, revealing the potential transient contribution of this stem population to the donor osteoclast population.

In conclusion, we have identified a cell-support approach to improve the phenotype in an animal model of osteopetrosis. In a translational perspective, although our cell therapy cannot fully cure the disease, it may represent a substantial method to transiently provide the patient with preosteoclasts for an early onset of bone resorption in the latency period after HSCT. It also could provide a new tool to manage the disease in patients in whom HSCT engraftment fails or in whom microenvironmental changes, such as those due to mutations of the osteoclastogenic cytokine RANKL,7 make the HSCT ineffective. Of course, no permanent engraftment can occur with our approach; however, it may pave the way for new strategies for which optimization will be necessary before translation into benefits for patients.

Disclosures

The authors state that they have no conflicts of interest.

Author contributions

A Cappariello: Conception and design, collection and assembly of data, data analysis and interpretation, manuscript writing. AC Berardi: Conception and design, final approval of manuscript. B Peruzzi: Collection and assembly of data, data analysis and interpretation. A Del Fattore: Collection and assembly of data, data analysis and interpretation. A Ugazio: Conception and design, financial support, final approval of manuscript. GF Bottazzo: Conception and design, financial support, final approval of manuscript. A Teti: Conception and design, financial support, data analysis and interpretation, manuscript writing, final approval of manuscript.

Acknowledgements

We thank Dr. Rita Di Massimo for her assistance in the editing of this manuscript.

This work was supported by the Telethon Grant N.GGP06019 to AT and by grants from the Ministry of Health “Rare Diseases” and from E-rare (project OSTEOPETR) to AU, GFB, and ACB. Dr. Alfredo Cappariello is supported by the Fellowship Research Programme, Ospedale Pediatrico Bambino Gesù, Rome, Italy.